Discovery of a Hepatitis C Virus NS5B Replicase Palm Site Allosteric Inhibitor (BMS-929075) Advanced to Phase 1 Clinical Studies

Article abstract of DOI:10.1021/acs.jmedchem.7b00328

The hepatitis C virus (HCV) NS5B replicase is a prime target for the development of direct-acting antiviral drugs for the treatment of chronic HCV infection. Inspired by the overlay of bound structures of three structurally distinct NS5B palm site allosteric inhibitors, the high-throughput screening hit anthranilic acid 4, the known benzofuran analogue 5, and the benzothiadiazine derivative 6, an optimization process utilizing the simple benzofuran template 7 as a starting point for a fragment growing approach was pursued. A delicate balance of molecular properties achieved via disciplined lipophilicity changes was essential to achieve both high affinity binding and a stringent targeted absorption, distribution, metabolism, and excretion profile. These efforts led to the discovery of BMS-929075 (37), which maintained ligand efficiency relative to early leads, demonstrated efficacy in a triple combination regimen in HCV replicon cells, and exhibited consistently high oral bioavailability and pharmacokinetic parameters across preclinical animal species. The human PK properties from the Phase I clinical studies of 37 were better than anticipated and suggest promising potential for QD administration.

Full text of DOI:10.1021/acs.jmedchem.7b00328

Subscriber access provided by ORTA DOGU TEKNIK UNIVERSITESI KUTUPHANESI
Article
Discovery of an HCV NS5B Replicase Palm Site Allosteric
Inhibitor (BMS-929075) Advanced to Phase 1 Clinical Studies
Kap-Sun Yeung, Brett R. Beno, Kyle Parcella, John A. Bender, Katherine A. Grant-Young, Andrew Nickel,
Prashantha Gunaga, Prakash Anjanappa, Rajesh Onkardas Bora, Kumaravel Selvakumar, Karen Rigat,
Ying-Kai Wang, Mengping Liu, Julie A Lemm, Kathy Mosure, Steven Sheriff, Changhong Wan, Mark Witmer,
Kevin Kish, Umesh M Hanumegowda, Xiaoliang Zhuo, Yue-Zhong Shu, Dawn Parker, Roy Haskell, Alicia
Ng, Qi Gao, Elizabeth Colston, Joseph Raybon, Dennis M. Grasela,, Kenneth S. Santone, Min Gao, Nicholas
A. Meanwell, Michael W. Sinz, Matthew G. Soars, Jay O. Knipe, Susan B. Roberts, and John F Kadow
J. Med. Chem., Just Accepted Manuscript • Publication Date (Web): 21 Apr 2017
Downloaded from http://pubs.acs.org on April 21, 2017
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 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Knipe, Jay; Bristol-Myers Squibb
Roberts, Susan ; Bristol-Myers Squibb
Kadow, John; Bristol-Myers Squibb
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Discovery of an HCV NS5B Replicase Palm Site
Allosteric Inhibitor (BMS-929075) Advanced to
Phase 1 Clinical Studies
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
†
†
†
†
†
Kap-Sun Yeung,* Brett R. Beno, Kyle Parcella, John A. Bender, Katherine A. Grant-Young,
†
‡
‡
‡
Andrew Nickel, Prashantha Gunaga, Prakash Anjanappa, Rajesh Onkardas Bora, Kumaravel
‡
†
†
†
†
†
Selvakumar, Karen Rigat, Ying-Kai Wang, Mengping Liu, Julie Lemm, Kathy Mosure, Steven
§
§
§
§
†
†
Sheriff, Changhong Wan, Mark Witmer, Kevin Kish, Umesh Hanumegowda, Xiaoliang Zhuo,
§
†
†
§
#
§
Yue-Zhong Shu, Dawn Parker, Roy Haskell, Alicia Ng, Qi Gao, Elizabeth Colston, Joseph
§
§
†
†
†
Raybon, Dennis M. Grasela, Kenneth Santone, Min Gao, Nicholas A. Meanwell, Michael
†
†
†
†
†
Sinz, Matthew G. Soars, Jay O. Knipe, Susan B. Roberts, John F. Kadow
†Bristol-Myers Squibb Research and Development, P.O. Box 5100, 5 Research Parkway,
Wallingford, Connecticut 06492, USA
‡Department of Discovery Chemistry, Biocon Bristol-Myers Squibb Research and Development
Center, Biocon Park, Jigani Link Road, Bommasandra IV, Bangalore 560099, India
§Bristol-Myers Squibb Research and Development, P.O. Box 4000, Princeton, New Jersey 08543,
USA
#Bristol-Myers Squibb Research and Development, 1 Squibb Drive, New Brunswick, New Jersey
0
8901, USA
ACS Paragon Plus Environment
1

Page 3 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ABSTRACT: The hepatitis C virus NS5B replicase is a prime target for the development of direct-
acting antiviral drugs for the treatment of chronic HCV infection. Inspired by the overlay of bound
structures of three structurally distinct NS5B palm site allosteric inhibitors, the high throughput
screening hit anthranilic acid 4, the known benzofuran analog 5 and the benzothiadiazine
derivative 6, an optimization process utilizing the simple benzofuran template 7 as a starting point
for a fragment growing approach was pursued. A delicate balance of molecular properties
achieved via disciplined lipophilicity changes was essential to achieve both high affinity binding
and a stringent targeted ADME profile. These efforts led to the discovery of BMS-929075 (37)
which maintained ligand efficiency relative to early leads, demonstrated efficacy in a triple
combination regimen in HCV replicon cells, and exhibited consistently high oral bioavailability
and pharmacokinetic parameters across preclinical animal species. The human PK properties from
the Phase I clinical studies of 37 were better than anticipated and suggested promising potential
for QD administration.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
INTRODUCTION
Hepatitis C virus (HCV) infection is a major medical problem on a global scale. Between
1
3
30 and 150 million people are chronically infected with HCV worldwide, with an estimated 2.7-
.9 million infections in the United States alone.1,2 Patients with HCV-related chronic liver
diseases are at a very high risk of developing liver cirrhosis and progressing to hepatocellular
carcinoma. Chronic HCV infection is therefore the leading cause of liver transplantation in the
US, and is responsible for approximately 700,000 deaths per year worldwide.1,2
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3
HCV was identified in 1989. It is a single stranded positive sense RNA virus belonging to
the Hepacivirus genus within the Flaviviridae family. HCV has been classified into 6 major
4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
genotypes with multiple subtypes, and a seventh genotype was identified recently. Genotypes 1a
and 1b predominate in the Americas, Europe and Asia, followed by genotypes 2 and 3, while
genotypes 4 and 5 are more prevalent in Africa. The HCV genome encodes a single polyprotein
of ~3000 amino acids. Over the past 25 years, enormous efforts have been invested in
understanding the HCV replication cycle and the structures and functions of the viral proteins with
5
the goal of identifying viable therapies for treating HCV infection. In particular, three of the HCV
non-structural proteins, the NS3/4A serine protease, the NS5A replication cofactor and the NS5B
6
replicase, have been extensively exploited as drug targets. This intensive research has culminated
in the approval of several HCV-targeted direct-acting antiviral agents that are used in combination
in the clinic to replace pegylated-interferon/ribavirin-based therapy.7 These advanced all oral
treatment options offer a cure for chronic HCV infection with increased sustained virological
response rates, shorter treatment durations, improved safety profiles and expanded genotype
coverage.8,9
The HCV NS5B replicase is an RNA dependent RNA polymerase that performs the
synthesis of HCV genomic RNA within the HCV replication complex machinery, and is therefore
5
an indispensable enzyme for HCV replication. From the initial positive-sense RNA template, a
negative-sense RNA intermediate is synthesized, which in turn is used to generate progeny
positive-sense RNA for further translation and packaging into new virions. Early on, three X-ray
crystallographic structures of constructs of HCV genotype 1b (gt-1b) BK strain NS5B with the C-
ACS Paragon Plus Environment
3

Page 5 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
terminal membrane anchor region truncated were determined.10,11,12 The structures revealed that
the catalytic domain adopts a right-hand configuration with fingers, palm and thumb domains
typical of polymerase enzymes (Figure 1a). The highly conserved active site that harbors the GDD
aspartic acid motif, Gly317-Asp318-Asp319, is located within the palm domain. The pivotal role
of the NS5B replicase in the HCV replication cycle and the absence of a mammalian counterpart
together with the structural information that enables structure-based drug design have made NS5B
an excellent anti-HCV drug target. The discovery and development of HCV NS5B replicase active
site-directed nucleoside and nucleotide inhibitors and their prodrugs, as well as non-nucleoside
inhibitors that bind in the three allosteric sites (thumb-I, thumb-II and palm site) have been
comprehensively reviewed.13,14,15
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. (a) Structure of the HCV NS5B protein.10 The finger loops are colored cyan, and the
fingers, palm, thumb, and C-terminal domains are depicted in green, blue, red, and white,
respectively. The “primer grip” and active site GDD motif are shown in yellow and orange,
respectively, and the location of the palm site is denoted with a magenta sphere. (b) A close-up
view of the HCV NS5B active site and palm binding pocket highlighting key residues and
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
structural features. α-Helices are labeled according to the nomenclature in ref. 12. Image
generated with The PyMol Molecular Graphics System (v. 1.8, Schrödinger, LLC).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The palm binding pocket is near the active site and is surrounded by the Tyr448 β-hairpin
loop from the thumb domain, the Ile363-Asn369 primer grip loop and two α-helices (αI, Pro197-
Ala210 and αP, Ser407-Tyr410) from the palm domain. The side chain of residue 316 on the β-
sheet preceding the catalytic Gly317-Asp318-Asp319 loop extends into the binding site, while
movement of the Arg200 side chain accommodates inhibitor binding (Figure 1b). The palm site
is the largest of the three allosteric binding sites and is deeply recessed and hydrophobic in nature.
The palm site has been divided into site I and site II in some publications. However, unlike thumb
site I and site II, the two palm pockets are overlapping and contiguous, thus a single palm site
naming is adopted in this report.13,14,15 Several different structural classes of palm site inhibitors
have been reported to be in various stages of clinical development.13,14,15 These include nesbuvir
1
6
(
HCV-796, 1) (Figure 2), which established clinical proof of concept in HCV-infected patients
for benzofuran-based inhibitors that bind to the allosteric palm site. A boronic acid derivative 2
17
18
19
(
GSK-5852) and a tetracyclic analog 3 (MK-8876) have also been reported recently.
ACS Paragon Plus Environment
5

Page 7 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
O
NH
F
HO
O
N
O S
O
1
Nesbuvir)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(HCV-796,
O
O
O
NH
NH
N
N
F
F
F
O
O
N
N
HO
O S
O
O S
O
2
F
3
B
(
GSK-5852)
(MK-8876)
OH
Figure 2. Structures of benzofuran NS5B palm site inhibitors 1, 2 and 3.
MEDICINAL CHEMISTRY STRATEGY
A high throughput screening campaign conducted against HCV gt-1b NS5B enzyme
identified an anthranilic acid hit compound 4 (Figure 3). An X-ray co-crystal structure of 4 in
complex with the ∆C18 truncated gt-1b Con1 strain NS5B protein revealed that 4 bound in the
palm site of the enzyme.20 Interestingly, the bound structure of 4 partially overlapped with that of
21
benzofuran analog 5, as observed in an NS5B/5 co-crystal structure, with its benzoic acid moiety
projecting from the direction of the benzofuran C5 position of 5 (Figure 4a). Another NS5B
22
complex structure with 6 in the benzothiadiazine series of NS5B inhibitors was determined and
showed that the bound structure of 6 intersected with both of those of 4 and 5 with its thiadiazine
ring placed in the location of the benzoic acid of 4 (Figure 4a). Overall the three-dimensional
structures of the protein molecules in the three complexes were very similar with pairwise
R.M.S.D. values derived from protein structural alignments varying from 0.3 to 0.7 Å. The protein
molecules were also structurally similar in the vicinity of the bound ligands with variations
observed in a few flexible loops. The overlapping structures created an opportunity to merge the
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
three chemotypes by growing a benzofuran inhibitor into available binding regions via substitution
at the benzofuran C5 position. The primary objective of the project was to identify an NS5B
replicase palm site inhibitor clinical candidate that could be utilized in combination with NS5A
and NS3/4A inhibitors. Moreover, included in this goal was to achieve activity towards the pre-
existing C316N mutant which is present in about 40% of gt-1b HCV-infected patients. Benzofuran
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
was reported to exhibit 26-fold resistance in a BB7 gt-1bC316N cell-based replicon assay
compared to wild type.23 This was considered to be a deficiency in the profile of 1 which would
substantially limit its clinical application.
O
NH
HO
O
O
O
O
HN
F
Cl
O
HN
O S
O
4
5
IC50 1b = 8.4 M
EC50 = Inactive
µ
O
O
NH
NH2
O
O
O
S
O
OH
N
N
F
O
F
N
H
7
O
6
IC50 1b = 21 M
µ
EC50 1b / 1a = 14 / >11 M
µ
LE = 0.27; MW = 327
HAC = 24; cLogP = 4.18
Figure 3. Structures of NS5B palm site inhibitors 4, 5, 6 and 7, and the HCV inhibitory potency
towards gt-1b NS5B enzyme (IC50) and replicon (EC50) for 4 and 7 (the IC50 value for 4 was the
average of n ≥2 independent experiments; values for 7 were from a single experiment performed
in duplicate).
ACS Paragon Plus Environment
7

Page 9 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 4. (a) Overlay of the bound structures of anthranilic acid 4 (magenta), benzofuran 5
(orange) and benzothiadiazine 6 (green) in the palm site of the wild-type (WT) gt-1b NS5B protein.
(
b) The X-ray structure of 5 bound to the gt-1b NS5B WT protein illustrating key interactions of
with palm site residues. Images generated with The PyMol Molecular Graphics System (v. 1.8,
Schrödinger, LLC).
5
As exhibited in the NS5B/5 complex co-crystal structure, the NH and the carbonyl of the
methyl amide from the inhibitor engage in two key hydrogen bonding interactions with the side
chains of Ser365 in the primer grip loop and Arg200 respectively (Figure 4b), while the 4-
fluorophenyl substituent forms tight van der Waals contacts with the surrounding hydrophobic
residues in the binding site. These two moieties together serve to anchor the core benzofuran ring
to the palm site. The initial chemistry strategy was therefore to utilize the structurally simpler C6-
unsubstituted benzofuran analog 7 (Figure 3), which showed weak but detectable activity in both
the in vitro enzymatic assay and cell-based HCV replicon assay, as a template to explore the
structure activity relationship (SAR) at the benzofuran C5 position in the palm binding site. The
C5 oxygen atom provided a convenient handle for library synthesis, and reagent selection was
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
facilitated by iterative molecular modeling based on NS5B/inhibitor complex structures. Although
the molecular weight (MW= 327) and lipophilicity (cLogP = 4.18) of 7 are somewhat outside of
2
4
the boundaries of those associated with a good fragment, its ligand efficiency (LE) value of 0.27
calculated from (1.37/HAC) × pIC50, IC50 determined from the gt-1b NS5B enzymatic assay and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
HAC = heavy atom count)25,26 was considered a suitable starting point for a fragment growing
approach.
RESULTS AND DISCUSSIONS
An initial benzofuran C5 aryl library obtained from Suzuki cross-coupling reaction
between the C5-triflate derivative of compound 7 and commercially available aryl boronic acids
or esters quickly established the importance of a carboxylate anion group for enhancing inhibitory
activity against the NS5B enzyme, as summarized in Table 1. Both the meta-benzoic acid 8 and
the para analog 9 were equally active with >20-fold improvements in IC50 values against gt-1b as
compared to the starting benzofuran 7. However, the ortho analog 11 was essentially inactive,
reflecting the importance of both the hydrogen bonding distance and angle for interaction in the
binding site.27 The desired position of the acid group for inhibitory activity was recapitulated in
the thiophene-2-carboxylic acid 10 in which the 5-membered thiophene ring with its longer C-S
bond and smaller C-S-C bond angle served as a suitable isostere for the phenyl ring of 8 and 9.28
The loss in activity of ester 12 and benzyl alcohol 13 suggested that an oxygen atom with strong
29
hydrogen bond accepting potential was necessary for potency against the enzyme. These SAR
results were consistent with a strong hydrogen bond between the carboxylate of the inhibitor and
backbone N-H (O-N distance 2.8 Ả) of Tyr448, as observed in the co-crystal structure of NS5B/8
complex that was obtained subsequently. The existence of a binding pocket adjacent to the
ACS Paragon Plus Environment
9

Page 11 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
carboxylate group, as revealed in the complex structure, provided an opportunity to further
functionalize the acid moiety via amide coupling and potentially form additional interactions with
the enzyme. Such modification could also convert these polar anionic compounds to neutral
species that were anticipated to show enhanced cell permeability for increasing potency in the cell
based replicon assay, which was far from satisfactory in the initial acidic lead compounds 8 and 9
(gt-1b EC50 = 12 and 4.5 µM, respectively; and gt-1a EC50 >20 µM).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 1. HCV Inhibitory Potency towards NS5B Enzyme (IC50) for Compounds 8 to 13
HN
O
Ar
F
O
gt-1b NS5B
Enzyme IC50 (µM)
Compd
8
Ar
HO
1.1
O
O
HO
HO
0.65a
9
0.36a
>50
1
0
S
O
11
HO
MeO
O
1
1
2
3
>50
>50
O
HO
ACS Paragon Plus Environment
1
0

Journal of Medicinal Chemistry
Page 12 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Data were obtained from a single experiment performed in duplicate; average value of n ≥2
independent experiments.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
As shown in Table 2, a notable improvement in the activity against gt-1b and gt-1a replicon
cells was obtained starting from the methyl amide 15. This was attributed to a combination of
increased inhibition of the NS5B enzyme and improved cell penetration, reflected by the ~2-fold
ratio between the gt-1b enzyme IC50 and replicon EC50 values. A key discovery from this amide
library SAR exercise was that successive addition of a methyl group to the α-carbon of the alkyl
amide series (from 15 to 18) as well as the benzyl amide analogs (from 21 to 23) enhanced the
potency both towards the enzyme and in replicon cells. Pleasingly, potent activity in the
nanomolar range in the replicon was achieved with these compounds (gt-1b EC50 = 18 nM for 18
and 10 nM for 23). N-Methylation of the secondary amide NH decreased the potency by 3-fold in
the enzymatic assay and this decline translated into the replicon assay as exemplified by 19, the
N-methyl analog of 18. This was subsequently rationalized by the fact that the amide NH engaged
in a hydrogen bonding with the hydroxyl group of the side chain from Tyr452 via a water bridge
(Figure 5a). The change in the amide conformation from a secondary to a tertiary amide could
also affect the favorable interaction. In general, compounds from the meta-amide series provided
better potency profiles than the corresponding para-amides which were more than 10-fold less
active as represented by 20, the para-isomer of 18. Thus, subsequent SAR studies were
concentrated on the meta-amide series. The Me to Ph matched molecular pair transformation on
the amide moiety improved the potency against the gt-1b NS5B enzyme by up to 9-fold (compare
1
6 and 21, 17 and 22, 18 and 23). This phenomenon was attributed to a favorable π-stacking
interaction between the amide phenyl ring and the phenyl ring from the Phe551 side chain.
ACS Paragon Plus Environment
1
1

Page 13 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
30
However, the 1.4 unit LogP increase in lipophilicity associated with a Me → Ph transform in
this molecular context might also play a role. Although the activity in the replicon cells,
particularly in gt-1a, was not as sensitive to the introduction of an amide phenyl ring, when the
gem-dimethyl group of 23 was cyclized into a cyclopropyl ring, single digit nanomolar potency
against both the NS5B enzyme and gt-1b replicon cells was realized, as represented by 24. The
cyclopropyl ring in 24 presumably fills the volume enclosed by Met414, Ile447 and Tyr452 more
effectively, resulting in improved van der Waals contacts. Encouragingly, the potency against the
gt-1a replicon cells was also driven below 50 nM in the phenylcyclopropyl analog 24, which
became the new lead structure for further optimization. Compared to the starting template 7, the
ligand efficiency was improved (LE = 0.31 for 18 and 0.30 for 24) even though the heavy atom
count was increased by 38% in 18 and as much as 58% in 24.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
During the course of SAR studies, multiple inhibitor/NS5B complex co-crystal structures
were determined. In these structures, remarkable conformational flexibility of the NS5B C-
terminal region and Tyr448 loop that comprise part of the palm site was observed. Due to this
plasticity, the size and shape of the binding pocket vary significantly in response to different sizes
and shapes of benzofuran inhibitor C5 substituents, as illustrated in Figure 5b. Thus, the NS5B
palm site offers an interesting example of an inducible binding pocket.31 This conformational
flexibility suggested that diverse amide structures could be accommodated, providing an
opportunity for further optimization focused on potency enhancement and/or modulation of
physicochemical properties for an improved ADME (absorption, distribution, metabolism,
excretion) profile as noted below and to be disclosed in future publications.
ACS Paragon Plus Environment
1
2

Journal of Medicinal Chemistry
Page 14 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 2. HCV Inhibitory Potency towards NS5B Enzyme (IC50) and Replicon (EC50) for
Compounds 14 to 24
HN
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R
F
O
O
gt-1b NS5B
Replicon EC50 (µM)
gt-1b
Compd
R
gt-1aa
Enzyme IC50 (µM)
2
5.7a
2.2
0.78a
>3.7
a
ND
ND
1.0
1
4
NH
NH
0.59a
0.33a
15
NH
1
6
NH
NH
1
7
0.19
0.041
0.12
0.098
0.018
0.076
0.63
0.24
1.4
1
1
8
9
NMe
NH
3.6a
0.98a
2
0
ND
(
para-isomer)
Ph
NH
0.091a
0.026a
21
0.067
0.054
1.8
Ph
NH
2
2
2
2
3
4
0.60
Ph
Ph
NH
NH
0.0080
0.0059
0.010
0.16
0.0036
0.036
a
Data were average values of n ≥2 independent experiments; data from a single experiment
performed in duplicate; ND = Not determined.
ACS Paragon Plus Environment
1
3

Page 15 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. (a) The X-ray crystal structure of 18 bound to gt-1b NS5B WT protein showing direct
and water-bridged hydrogen bonds between the protein and ligand. Hydrogen bonds are indicated
with blue dashes. (b) Overlay of co-crystal structures of NS5B complexed with 8 (green) and 18
(orange) highlighting the Y448 loop, C-terminal region, and F551 which adopt different
conformations depending on the identity of the benzofuran C5 substituent. (c) The hydrophobic
cavity proximal to the 2’-position of the inhibitor amide phenyl ring and formed by the side chains
from F193, C316, C366 and Y448 as illustrated in the close-up view of that region in the NS5B/18
complex structure in (a). Images generated with The PyMol Molecular Graphics System (v. 1.8,
Schrödinger, LLC).
Subsequent optimization performed on inhibitor 24 was again guided by observations from
the NS5B/inhibitor complex co-crystal structures. Ligand elaboration to fill a small hydrophobic
cavity proximal to the 2’-position of the amide phenyl ring and formed by the side chains from
Phe193, Cys316, Cys366, and Tyr448 appeared to be feasible (Figure 5c). Indeed, substitution at
the 2’-position with small groups of increasing size (25 to 28, Table 3) not only maintained the
single digit nanomolar potency against gt-1b in the replicon assay but also led to a stepwise
increase in the activity against both gt-1a replicon and gt-1bC316N mutant replicon. The 2’-
methyl analog 28 emerged from these SAR studies with a more balanced inhibitory potency
ACS Paragon Plus Environment
1
4

Journal of Medicinal Chemistry
Page 16 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
profile, and exhibited a favorable cytotoxicity window (CC50 >25 µM in the replicon Huh-7 host
cell line). However, a major drawback to the in vitro profile of this series of potent inhibitors was
their low metabolic stability in liver microsomes. For example, only 68% and 39% of lead
compound 28 remained un-metabolized in human and rat liver microsomes (HLM and RLM),
respectively, after a 10-minute incubation in a high throughput metabolism screen (Table 4). Met-
ID studies on the earlier compound 23 identified the electron-rich gem-dimethyl phenyl moiety as
the major metabolic soft spot, thus suggesting the introduction of a heteroatom to the phenyl ring
to increase polarity and potentially block metabolic sites could be a viable approach for solving
this problem. Incorporation of a nitrogen atom into the phenyl ring to increase polarity of the
molecule would be desirable in view of the already high lipophilicity of the 2’-methyl lead
compound 28 (cLogP = 6.28).30 It was determined that the 2-pyridyl analog 29 with a reduced
cLogP value of 5.44 provided excellent stability in both HLM and RLM while maintaining the
potency profile of 28 (Table 4) compared to the 3- and 4-pyridyl analogs 30 and 31, respectively.
The microsomal stability of 29 in RLM (t1/2 = 61 min) together with its high membrane
permeability (999 nm/s, pH 7.4) as measured in a parallel artificial membrane permeability assay
(PAMPA) translated into good plasma exposure following oral administration of the compound to
rats in a 6-hour pharmacokinetic screen. A high concentration of 29 (15.7 µM) was also obtained
in rat liver at the 6 hour time point (Figure 6), indicating favorable exposure level in the target
organ could be achieved with this compound series. The clearance estimated from the screen was
a modest 14 mL/min/kg. These encouraging data established the direction for further optimization.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 3. Inhibitory Potency towards HCV Replicon (EC50) for Compounds 24 to 28
ACS Paragon Plus Environment
1
5

Page 17 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
'
R
HN
O
H
N
F
O
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Replicon EC50 (nM)
Compd
R
gt-1b
3.6
gt-1a
36
16
6.2
7.9
2.1
gt-1bC316N
2
2
4
5
H
F
Cl
OMe
Me
ND
150
63
32
35
3.6
2.5
1.8
1.3
26
2
2
7
8
Data were average values of n ≥2 independent experiments; ND = Not determined.
Table 4. Inhibitory Potency towards HCV Replicon (EC50) and In Vitro Profiling for
Compounds 28 to 31
HN
O
N
2
H
N
F
O
O
H / R LMa
Replicon EC50 (nM)
CYP3A4
IC50 (µM) IC50 (µM)
CYP2C8
cLogPb
cpK
a
b
Compd
N-Ar
gt-1b gt-1a gt-1bC316N %remain
t1/2 (min)
2
8
Phenyl
1.3
1.9
2.1
2.8
17
35
55
68 /39
89 / 82
0.7 / 0.6
11 /11
11 / ND
108 / 61
ND
6.28
5.44
5.05
5.05
--
2.1
4.1
1.5
1.0
29
2-pyridyl
3-pyridyl
4-pyridyl
3.95
4.77
5.01
3
3
0
1
2.2
193
63
0.063
0.073
0.35
0.026
0.83
8.3
ND
a
Data were average values of n ≥2 independent experiments; H / R LM denotes human and rat
b
30
liver microsomes; Calculated values ; ND = Not determined.
HN
O
H
N
N
F
O
O
2
9
ACS Paragon Plus Environment
1
6

Journal of Medicinal Chemistry
Page 18 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 6. 6-Hour rat pharmacokinetic screen of 2-pyridyl analog 29 (IV, 2 mg/kg; PO, 6 mg/kg;
9:1 PEG400/EtOH; n = 2); PO exposure at the 6-hr time point: plasma = 1.64 µM and liver = 15.7
µM. (The open and filled circles and diamonds indicate IV data obtained from two individual rats;
the open and filled squares and triangles indicate PO data obtained from two individual rats.)
During these SAR studies, it was noted that the 3- and 4-pyridyl analogs 30 and 31
appeared to be almost instantly metabolized in liver microsomes and were significantly less stable
than the 2-pyridyl lead 29 and the parent phenyl analog 28, although both 30 and 31 are less
lipophilic than 29 and 28 by about 0.4 to 1.5 calculated LogP units (Table 4). Quantum chemical
calculations performed at the MP2/6-31+G** level on the N-(1-(pyridin-2-
yl)cyclopropyl)benzamide model system predict a 0.8 kcal/mol preference for a conformation in
which the plane of the pyridine ring bisects the cyclopropyl C2-C3 bond (Figure 7).32 This is in
contrast to the corresponding phenyl model system which prefers a perpendicular conformation.33
It is possible that shielding of the pyridyl nitrogen lone pair in 29 by the cyclopropyl methylene
ACS Paragon Plus Environment
1
7

Page 19 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
moiety is partially responsible for the reduced basicity of 29 relative to 30 and 31 where the pyridyl
nitrogen atoms are more accessible. Diminished steric encumbrance and higher basicity together
contribute to the greater susceptibility of the pyridine rings in 30 and 31 towards potential N-
oxidation by microsomal CYP enzymes, and potent inhibition of CYPs by 30 and 31 via
coordination to the heme iron at the nanomolar level in an in vitro assay using recombinant CYPs
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(Table 4). These findings are in line with the general trend observed in the change in lipophilicity
and CYP3A4 inhibition when a mono-substituted phenyl ring is replaced by the three pyridine
regioisomers.34
Figure 7. Relative energies of aryl-cyclopropyl bond rotamers for model systems (left) and
bisected and perpendicular conformers of the 2-pyridyl model (right) calculated at the MP2/-6-
3
1+G** level of theory.32
Although the 2-pyridyl lead compound 29 exhibited promising replicon potency,
microsomal stability in HLM and RLM, membrane permeability and oral exposure, other in vitro
parameters still required optimization. For example, 29 was not as metabolically stable in dog and
ACS Paragon Plus Environment
1
8

Journal of Medicinal Chemistry
Page 20 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
monkey liver microsomes (t1/2 = 14 and 20 min, respectively), possessed low micromolar
+
inhibition of CYP3A4 and CYP2C8 enzymes, hERG inhibition in a Tl ion flux assay with an IC50
of 5.8 µM and activity in a pregnane X receptor (PXR) transactivation assay with an EC50 of 1.6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
µM. Controlling the lipophilicity of the molecule (cLogP for 29 = 5.44, shake flask LogDpH6.5
=
5
.42) to avoid these off-target liabilities while preserving the good qualities of 29 was desirable at
this stage of the optimization process. However, this represented a challenge since the palm site
is deeply recessed and highly hydrophobic, and a trend of decreasing potency particularly against
gt-1a and the gt-1bC316N mutant was noted in molecules with reduced lipophilicity. A subtle
balance of molecular properties was thus needed to achieve both potent antiviral activity and
favorable ADME properties. It was anticipated that the conformational flexibility of the NS5B C-
terminal region, in which the cyclopropyl amide portion of the molecule resides as described
above, would allow productive structural modifications of the inhibitor that improved potency
while permitting optimization of molecular properties. Thus, further introduction of heteroatoms
into the amide region by surveying 5- and 6-membered heterocycles was pursued. One avenue
that proved to be effective was the addition of a second nitrogen atom into the pyridine ring of 29,
and walking the nitrogen atom around the ring to deliver the four isomeric diazine analogs 32 to
3
5 (Table 5). Among them, only the more lipophilic 2-pyrimidinyl analog 32 (cLogP = 4.83,
shake flask LogDpH6.5 = 4.76) maintained the activity profile of the 2-pyridyl parent compound 29.
The replicon activity against gt-1a and the gt-1bC316N mutant deteriorated with increasing
polarity in the other three isomers. Other analogs possessing 5-membered heterocyclic rings (e.g.
imidazole, pyrazole, triazole, tetrazole, oxazole, isoxazole) were also inferior primarily due to
unfavorable replicon potency against the gt-1bC316N mutant (EC50 values ranged from 100 nM
to >1 µM) occasionally accompanied by an undesirable CYP inhibition profile (IC50 < 1 µM). The
ACS Paragon Plus Environment
1
9

Page 21 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
oxadiazole analog 36, which interestingly has the same cLogP value as the 2-pyrimidyl derivative
3
2, was the only heterocyclic analog surveyed in this series that exhibited a potency profile
comparable to 32. However, 36 lacked stability in rat and monkey liver microsomes (t1/2 = 10 and
5 min, respectively). New lead compound 32 not only maintained good stability in HLM and
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
RLM and high membrane permeability compared to parent compound 29, but also improved upon
the metabolic stability in dog and monkey liver microsomes, off-target activity against CYP
enzymes and PXR transactivation (Table 6). In a 24-hour pharmacokinetic study in rats,
compound 32 exhibited low clearance and a high oral bioavailability. Although the concentration
of 32 in the liver at the 6-hour time point (8.7 µM) following oral dosing was about half of that
obtained with 29, distribution into the liver was still favored for the neutral molecule (C24
liver/plasma ratio = 2.4). These in vitro and in vivo properties, together with a modest 4-fold shift
in potency in the presence of 40% human serum and favorable cytotoxicity (CC50 = 20 µM in Huh-
7
cells) exhibited by 32, set the stage for fine tuning towards a viable development candidate.
Table 5. Inhibitory Potency towards HCV Replicon (EC50) and cLogP Values for
Compounds 29 and 32 to 36
HN
O
H
Ar
N
F
O
O
Replicon EC50 (nM)
gt-1a
cLogPb
Compd
Ar
gt-1b
gt-1bC316N
5.44
1.9a
2.8a
55a
2
9
N
ACS Paragon Plus Environment
2
0

Journal of Medicinal Chemistry
Page 22 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
N
4.83
4.22
1.0a
2.6
2.3a
51a
3
3
2
3
N
N
15
308
N
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
N
N
3
3
3
4
5
6
4.77
4.49
4.83
1.9
2.9
13
17
109
139
46
N
N
O
0.81a
1.2a
N
a
Data were obtained from a single experiment performed in duplicate; average value of n ≥2
b
30
independent experiments; Calculated values ; ND = Not determined.
Table 6. In Vitro Profiling Data and Pharmacokinetic Parameters in Rats (0 to 24 hour) for
3
2
H / R / Mk / D LMa
CYP3A4
CYP2C8
hERG Flux
IC50 (µM)
PXR-TA
EC50 (µM)
PAMPA
pH7.4 (nm/s) IC50 (µM) IC50 (µM)
hERG
Patch Clamp
t
1/2 (min)
>120 / 86 / 51 / 46
769
11.4
3.9
4.8
96% Inh @10 µM
>50
IV
PO
%F
CL
IV t1/2
(hr)
Vss
(L/kg)
Cmax Tmax
AUC
(µM*hr)
C
6
Liver
C
24 Liver
(µM)
C
24
mL/min/kg
(µM)
(hr)
(µM)
Liver/Plasma
5.5
5.3
6.1
6.7
4.7
48.7
8.7
0.614
2.4
>100
a
IV, 2 mg/kg; PO, 6 mg/kg; 9:1 PEG400/EtOH; n = 3; H / R / Mk / D LM denotes human, rat,
monkey and dog liver microsomes respectively.
At this point, improving inhibitory potency against the gt-1bC316N mutant to an EC50
value of less than 50 nM without increasing lipophilicity, while minimizing hERG activity (Tl+
flux IC50 = 4.8 µM; whole cell patch clamp: 96% inhibition @10 µM for 32) in this class of neutral
ACS Paragon Plus Environment
2
1

Page 23 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
molecules was challenging. Improving the gt-1bC316N inhibitory potency and resolving the
hERG off-target liability represented a dilemma since modulation of the two parameters involved
opposing lipophilicity requirements. One approach that had been contemplated to increase gt-
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
bC316N inhibition was to engage the N316 residue via a direct interaction with the inhibitor. It
was observed from crystal structures of NS5B WT/inhibitor complexes that the C-H bond at the
benzofuran 4-position points directly towards the side chain of the C316 residue with
approximately 3.0 Å separating the benzofuran C4-H and the C316 sulfur atom. Thus, introduction
of a dipole at the benzofuran 4-position could potentially capture a dipole-dipole or a hydrogen
bonding interaction with the N316 side chain. Installation of a fluorine atom that can fit into the
binding space (C-F bond length ~1.35 Å) as a dipole probe at the 4-position of the benzofuran ring
35
of 32 provided compound 37 (BMS-929075), which exhibited a 2.8-fold improvement in the
activity against the gt-1bC316N mutant (EC50 = 18 nM, Table 7). Attempts to obtain a co-crystal
structure of 37 bound to gt-1bC316N mutant protein to provide direct structural evidence for
interaction between the 4-F and the N316 side chain were not successful. However, a model of a
4
-F-benzofuran analog bound to the gt-1bC316N mutant protein based on a co-crystal structure of
a gt-1bC316N/inhibitor complex (supporting information) that was subsequently determined,
revealed that the negatively polarized fluorine atom at the 4-position of the benzofuran ring can
form a favorable electrostatic interaction with the positively polarized carbonyl carbon in the
primary amide group of the N316 side chain (Figure 8). Both the angle (82°) and distance (2.8 Å)
are favorable for a C-F⋅⋅⋅C=O interaction and consistent with those values reported by Banner,
Müller and Diederich.36,37 It is conceivable that this atypical molecular interaction contributed to
the enhancement, though modest, in the potency against the gt-1bC316N mutant.
ACS Paragon Plus Environment
2
2

Journal of Medicinal Chemistry
Page 24 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Although 37 is 98.2% bound in human serum, it exhibited only a modest 5.5-fold right
shift in the replicon EC50 against gt-1a and gt-1b in the presence of 40% human serum. At the
onset of the project, the targeted antiviral profile was for coverage of both WT gt-1 virus and the
pre-existing gt-1bC316N mutant. Further evaluation of 37 in hybrid replicons representing the
NS5B replicase of other HCV genotypes that became available in the later stages of the project
revealed that the compound retained high inhibitory potency against genotypes 3 to 6 but was less
potent towards gt-2 (Table 7). Combination studies of 37 with the approved NS5A inhibitor
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
38
39
40
daclatasvir and the NS3/4A protease inhibitor asunaprevir in a colony elimination experiment
revealed that the triple combination eradicated HCV gt-1b replicons more effectively than the dual
combination of daclatasvir/asunaprevir or monotherapy with any of the single inhibitors at similar
total concentrations of ≥10 times the replicon EC50 value, suggesting that 37 can be a very effective
agent for combination therapy.
Table 7. Inhibitory Potency towards HCV Replicons (EC50) Representing Genotypes 1 to 6
for Compound 37
HN
N
F
O
H
N
N
F
O
O
37
(BMS-929075)
Replicon EC50 (nM)
1a
9
1b
4
1bC316N
18
2a
2b
3a
22
4a
5a
2a
6a
5
481
115
3
a
Data were average values of n ≥2 independent experiments; HCV NS5B Enzyme IC50
.
ACS Paragon Plus Environment
2
3

Page 25 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. A computer model of a 4-fluorobenzofuran analog in complex with N316 mutant
enzyme based on a co-crystal structure of gt-1bC316N/inhibitor complex. Image generated with
The PyMol Molecular Graphics System (v. 1.8, Schrödinger, LLC).
In the optimization from the simple template 7 to the initial meta-amide lead compound 18
and eventually to the clinical candidate 37, ligand efficiency was maintained if not improved
despite a significant increase in HAC (38% and 67%, Figure 9). This is in contrast to the generally
observed trend that LE often decreases as ligand size increases.41,42 The conservation of LE as
HAC increased during this lead optimization process is due to the fact that each additional
functional moiety growing out from template 7 was optimally deployed based on SAR analysis
coupled with insights from molecular modeling. Each part of 37 engages the protein in the palm
binding pocket as observed in the gt-1b NS5B/37 complex co-crystal structure (Figure 10a),
maximizing the enthalpic gain in binding: the 4-fluorophenyl ring makes tight van der Waals
contacts; the methyl amide forms two strong hydrogen bonds, one as an acceptor with Arg200 and
one as a donor to Ser365; the 4-fluorine atom on the benzofuran ring is in close contact with the
ACS Paragon Plus Environment
2
4

Journal of Medicinal Chemistry
Page 26 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
thiol of the Cys316 side chain in a hydrogen bonding-type interaction; the benzofuran ring and the
meta-amide phenyl ring form favorable electrostatic interactions with Arg200 and the phenyl ring
of the Tyr448; the 2’-methyl group fills a hydrophobic hole; the cyclopropyl ring projects into a
hydrophobic pocket, and the pyrimidine ring forms an edge-to-face interaction with the phenyl
ring of the Phe551 side chain. Based on higher resolution X-ray structures of similar compounds
bound to gt-1b NS5B WT protein, it is likely that the secondary meta-amide forms two water-
bridged hydrogen bonds, one to Arg200 and one to the hydroxyl group of the Tyr452 side chain.
The small molecule crystal structure of the unsolvated neat form of 37 exhibits two conformers
per asymmetric unit (Figure 10b). The relative three-dimensional arrangements of the two
conformers are largely the same, with subtle differences in the orientation of the C2-methyl amide
and the 4-fluorophenyl ring, and are close to the NS5B-bound conformation of 37. The bound and
unbound molecular conformations of 37 differ in the positioning of pyrimidinylcyclopropyl moiety
by a 180° rotation about the cyclopropane-amide C1-N bond. However, in both the bound and
unbound forms, the pyrimidinylcyclopropyl group adopts a bisected conformation which is
predicted to be the only energy minimum for rotation about the cyclopropane-pyrimidine bond
based on MP2/6-31+G** calculations on a model system (Figure 7).32 These observations on the
solid state structures of 37 suggest that the entropic cost of binding of 37 to NS5B can be
substantially reduced. The extended structure of 37 as displayed in both the bound and unbound
states suggests the solvent accessible surface area of 37 can be maximized for binding. Here,
maximization of enthalpic gain and accessible surface area combined with reduction of entropic
cost as ligand size increases are beneficial design approaches for improving ligand efficiency in
this fragment growing optimization process.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment
2
5

Page 27 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
HN
HN
O
HN
O
N
F
O
H
H
O
N
N
N
F
F
F
O
O
O
O
O
7
18
LE = 0.31
HAC = 33
37
LE = 0.28
LE = 0.26
HAC = 24
inc.)
HAC = 40
inc.)
(
38%
(67%
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 9. Ligand efficiency of HCV NS5B inhibitors 7, 18 and 37 with the % increase in HAC
indicated.
Figure 10. (a) gt-1b NS5B/37 complex co-crystal structure with key interacting residues
highlighted (see text). The omit Fo-Fc electron density is contoured at 3 R.M.S.D. in magenta
mesh and hydrogen bonding, π-cation, and π-stacking interactions are denoted with blue, red, and
cyan dashes, respectively. Image generated with The PyMOL Molecular Graphics System (v. 1.8,
Schrödinger, LLC).
ACS Paragon Plus Environment
2
6

Journal of Medicinal Chemistry
Page 28 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 10. (b) Small molecule crystal structure of the unsolvated neat form of 37, showing the two
conformers per asymmetric unit. Image generated with The PyMOL Molecular Graphics System
(v. 1.8, Schrödinger, LLC).
The incorporation of the 4-F atom on the benzofuran ring also served to mitigate hERG
inhibition (Table 8), despite an increase in the lipophilicity of the neutral molecule (cLogP = 4.98,
shake flask LogDpH6.5 = 5.09 for 37) compared to that of the 4-H analog 32. Compound 37
exhibited an IC50 of 24 µM in the hERG ion flux assay which was consistent with an IC50 of 18
µM in a whole cell patch clamp electrophysiology assessment. This favorably compared to the
96% inhibition @ 10 µM observed with 32. The lipophilicity value of 37 is also above the upper
limit (cLogP = 4 / LogD = 3.3) estimated for a neutral molecule to ensure a hERG inhibition IC50
4
3
value of >10 µM. Further analysis of the benzofuran series suggested that the small structural
change from benzofuran to 4-F-benzofuran constituted a matched pair transformation for
attenuation of hERG inhibition although accompanied by a 0.2 to 0.3 unit increase in measured
LogD or cLogP value.44 As shown in Figure 11, the 4-F atom reduces the negative electrostatic
potential in the benzofuran ring relative to the 4-unsubstituted benzofuran ring. This could
potentially impact the electrostatic complementarity with the hERG channel pore binding pocket,
ACS Paragon Plus Environment
2
7

Page 29 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
thus disrupting the affinity of the inhibitor to the hERG channel.45 Compound 37 also exhibited
no significant inhibition of the cardiac sodium Nav1.5 and L-type calcium Cav1.2 channels.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 8. In Vitro Profiling Data for Compound 37
H / R / Mk / D LMb
H / R / Mk / Db
protein binding (% bound)
LogDpH6.5
shake flask)
PAMPA
pH7.4 (nm/s) A to B (nm/s)
Caco Pc
cLogPa
(
t
1/2 (min)
4.98
5.09
>120 / >120 / >120 / >120
972
>100
98.2 / 99.7 / 97.2 / 98.1
hERG
Flux
IC50 (µM)
hERG
Patch Clamp
IC50 (µM)
Primary human
HepG2
CYP3A4
IC50 (µM)
CYP2C8
IC50 (µM)
PXR-TA
EC50 (µM)
Huh-7
CC50 (µM)
hepatocytes
CC50 (µM)
CC50 (µM)
>
13
6.7
30
24
18
>50
60
>12.5
>50
a
b
Calculated values ; H / R / Mk / D denotes human, rat, monkey and dog respectively and LM =
liver microsomes.
ACS Paragon Plus Environment
2
8