Structure-based design of novel HCV NS5B thumb pocket 2 allosteric inhibitors with submicromolar gt1 replicon potency: Discovery of a quinazolinone chemotype

Article abstract of DOI:10.1016/j.bmcl.2013.05.037

We describe the structure-based design of a novel lead chemotype that binds to thumb pocket 2 of HCV NS5B polymerase and inhibits cell-based gt1 subgenomic reporter replicons at sub-micromolar concentrations (EC₅₀ <200 nM). This new class of po

Full text of DOI:10.1016/j.bmcl.2013.05.037

Bioorganic & Medicinal Chemistry Letters 23 (2013) 4132–4140
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Structure-based design of novel HCV NS5B thumb pocket 2 allosteric
inhibitors with submicromolar gt1 replicon potency: Discovery of a
quinazolinone chemotype
a
b
a
a
a
Pierre L. Beaulieu a, , René Coulombe , Jianmin Duan , Gulrez Fazal , Cédrickx Godbout , Oliver Hucke ,
Araz Jakalian a, Marc-André Joly a, Olivier Lepage a, Montse Llinàs-Brunet a, Julie Naud a, Martin Poirier a,
Nathalie Rioux b, Bounkham Thavonekham a, George Kukolj b, Timothy A. Stammers a
⇑
a Medicinal Chemistry, Boehringer Ingelheim (Canada) Ltd, Research and Development, 2100 Cunard Street, Laval, Quebec H7S 2G5, Canada
b Biology Department, Boehringer Ingelheim (Canada) Ltd, Research and Development, 2100 Cunard Street, Laval, Quebec H7S 2G5, Canada
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 11 March 2013
Revised 2 May 2013
Accepted 13 May 2013
Available online 23 May 2013
We describe the structure-based design of a novel lead chemotype that binds to thumb pocket 2 of HCV
NS5B polymerase and inhibits cell-based gt1 subgenomic reporter replicons at sub-micromolar concen-
trations (EC50 <200 nM). This new class of potent thumb pocket 2 inhibitors features a 1H-quinazolin-4-
one scaffold derived from hybridization of a previously reported, low affinity thiazolone chemotype with
our recently described anthranilic acid series. Guided by X-ray structural information, a key NS5B–ligand
interaction involving the carboxylate group of anthranilic acid based inhibitors was replaced by a neutral
two-point hydrogen bonding interaction between the quinazolinone scaffold and the protein backbone.
The in vitro ADME and in vivo rat PK profile of representative analogs are also presented and provide
areas for future optimization of this new class of HCV polymerase inhibitors.
Keywords:
HCV NS5B polymerase
Thumb pocket 2
Non-nucleoside inhibitor
Allosteric inhibition
Structure-based drug design
Ó 2013 Elsevier Ltd. All rights reserved.
Inhibitors of the virally-encoded NS5B RNA-dependent RNA
polymerase of the hepatitis C virus have acquired a prominent po-
sition in the development of more effective and better-tolerated
interferon-free therapies for the treatment of chronic HCV infec-
tion. The addition of 1st generation NS3/4A protease inhibitors to
inhibitor of gt1–6 replicons (EC50 = 11À98 nM).5 The IFN-free com-
bination of faldaprevir + BI 207127 + RBV demonstrated high effi-
cacy and good tolerability in treatment-naïve patients with 16–
28 weeks of treatment achieving overall SVR rates up to 69% (up
to 85% in gt1b) and is currently being evaluated in phase 3 clinical
trials for the treatment of gt1b HCV chronic infection.6
the pegylated-interferon-a/ribavirin (PEG-IFN/RBV) standard of
care (SoC) has increased sustained virological response (SVR) to
68–75% and shortened duration of treatment in genotype (gt) 1 pa-
tients.1 However, interferon-free regimens consisting of multiple
combinations of Direct Acting Antivirals (DAAs) with complemen-
tary mechanisms of action and RBV, have shown similar SVR rates
with increased tolerability compared to the newly introduced SoC
(IFN + RBV + PI).2 More recently, triple DAA combinations have
shown potential to increase SVR rates to >90% in the gt1 popula-
tion.3 A key element of these new promising interferon-free thera-
pies is the use of nucleoside (tide) or allosteric non-nucleoside
NS5B inhibitors in combination with NS3/4A or NS5A inhibitors.
Over the last years, we have been investigating the use of
thumb pocket 1 NS5B allosteric inhibitors in combination with
the protease inhibitor faldaprevir.4 BI 207127 (Fig. 1) is a potent
To augment our efforts in the development of IFN-free regimens
based on the combination of faldaprevir with non-nucleoside NS5B
thumb pocket 1 inhibitors we recently reported the discovery of
anthranilic acid derivatives (e.g., compounds 1 and 2, Figs 1 and
2) that bind to ‘thumb pocket 2’, a spatially distinct allosteric site
on the polymerase situated at the base of the thumb domain, some
30 Å away from the enzyme active site.7,8 Similarly to previously
reported clinical candidates that bind to this allosteric pocket
(e.g., filibuvir and lomibuvir; Fig. 1)9 anthranilic acid-based inhib-
itors are structurally distinct from indole-based thumb pocket 1
compounds (e.g., BI 207127) and were shown to exhibit additive
potency and distinct resistance profiles with the later. Anthranilic
acid amides such as 1 possessed comparatively promising antiviral
potencies in both gt1a and gt1b replicons (RT-PCR EC50 ꢀ150 nM)
and optimization of these acid derivatives was actively pursued
toward the identification of more potent analogs with suitable
qualifications for preclinical development. In a parallel effort we
also pursued structure-based design aiming to increase chemical
⇑
Corresponding author. Tel.: +49 (6132) 77-3092.
E-mail address: resgeneral.lav@boehringer-ingelheim.com (P.L. Beaulieu).
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.05.037

P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
4133
O
O
O
N
N
N
N
N
N
H
Br
N
S
COOH
3
NH
O
I482
BI 207127
O
gt1b/1a RT-PCR EC50 = 11 / 23 nM
F
Y477
S
OH
COOH
N
S476
N
N
OH
N
L497
N
O
O
L419
O
N
Lomibuvir
gt1b/1a EC50 = 23 / 15 nM
Filibuvir
gt1b/1a EC50 = 52 / 33 nM
CF
3
1 (Thumb Pocket 2)
COOH
O
Figure 3. X-ray structure of thumb pocket 2 thiazolone 3 (PDB 2HWI) in complex
gt1b Δ21NS5B IC50 = 80 nM
N
with NS5B polymerase.
N
gt1b/1a RT-PCR EC50 = 160 / 140 nM
gt1b Luciferase reporter EC50 = 95 nM
O
Figure 1. Structure of NS5B thumb pocket 1 and 2 clinical candidates and thumb
pocket 2 lead.
O
O
O
F
N
F
F
I482
O
2
Y477
S476
L497
L419
Figure 2. X-ray structure of
anthranilic acid derivative 2 and NS5B polymerase (PDB 4JJS).
a complex depicting key interactions between
Figure 4. Overlap of NS5B complexes with compounds 2 and 3 showing conser-
vation of key inhibitor–protein interactions.
diversity by replacing the carboxylic acid function of 1 by non-ion-
izable surrogates in order to identify inhibitors with distinct
ADME-PK profiles and potential for addressing possible liabilities
often associated with carboxylic acid containing molecules such
as reduced cell permeability and the formation of reactive metab-
olites (e.g., acylglucuronides). The discovery of a novel NS5B thumb
pocket 2 chemotype based on a neutral 1H-quinazolin-4-one scaf-
fold is described herein.
contacts include the lipophilic 4-methylcyclohexyl and trifluoro-
methyl groups of the inhibitor that bind to hydrophobic pockets
on the enzyme. The N-isopropyl group points toward solvent and
serves to orient the cyclohexyl amide moiety towards a well-de-
fined binding pocket. The remainder of the molecule (i.e., the ben-
zene ring scaffold) lies flat between the two walls of an extended
protein channel, resting on the side chain of L419 and enclosed
on each side by L497 and I482 that in part constitute the opposing
walls of the pocket.
The X-ray crystal structure of anthranilic acid derivative 2 (
D21
NS5B IC50 = 0.20 M/gt1b luciferase replicon EC50 = 0.64 M), a
l
l
Our design of non-acidic inhibitors had to consider the critical
interactions mediated by the carboxylic acid group of 1 or 2 and
the proximal protein backbone residues. Not surprisingly, isosteric
replacements for the COOH group were not well tolerated, presum-
ably because they require significant protein conformational adap-
tation to accommodate larger groups (unpublished results).
Several years ago, Valeant Pharmaceuticals reported a series of
close analog of 1, in complex with NS5B polymerase is shown in
Figure 2.7b,10
A key inhibitor–protein interaction that is characteristic of
anthranilic acid inhibitors such as 2 is the two-pronged hydrogen
bond highlighted in Figure 2 between the carboxyl group of the
inhibitor and the backbone NHs of Y477 and S476. Other notable

4134
P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
ble potency to thiazolone 3 (within twofold) providing encourage-
O
CF3
O
N
ment for further optimization. Previous studies in the anthranilic
acid (unpublished results) and the thiophene carboxylic acid class
of thumb pocket 2 inhibitors,8a exposed very strict structural
requirements for binding of substituents in the well-defined
right-hand-side hydrophobic pocket. Aliphatic groups such as
those present in 4 and 12–14 or unsubstituted analog 11 proved
unsuitable for improving potency. However, replacement of the
cyclohexyl moiety by an N-benzyl substituent (15) was tolerated
and addition of a para-methyl group on the aromatic ring (16) pro-
vided submicromolar biochemical potency, suggesting that the
neutral quinazolinone core had indeed potential as an anthranilic
acid replacement. We are unable to ascertain whether the ob-
served potency differences are due in part to changes in strength
of the H-bonds between a strongly solvated carboxylic acid func-
tion with the backbone protein NHs compared to the quinazoli-
none isosteres. However, requirements for binding in the right-
hand side lipophilic pocket (e.g., exit trajectory of the substituent
from the rigid core) were likely altered with the new scaffold, per-
haps due to a modified angle of approach, and a specific SAR
needed to be reconstructed at the N1-position of the quinazolinone
template. To this end, a library of approximately 40 additional 1-
benzylquinazolin-4-one derivatives bearing a variety of groups (al-
kyl, halo, alkoxy) and substitution patterns on the phenyl ring were
evaluated. Some SAR trends became apparent upon examination of
the selected results presented in Table 1. First, substitution in the
para-position of the phenyl ring is limited by size and appears to
be optimal with lipophilic small groups such as Me, Et or halogens
O
N
O
3
O
O
S
N
NH
O
1
O
F
Anthranilic acid
Thiazolone
SBDD
CF3
O
N
O
N
N
4
1H-Quinazolin-4-one hybrid
Scheme 1. Structure-based drug design (SBDD) of the thiazolone-anthranilic acid
1H-quinazolin-4-one hybrid. Groups that serve equivalent functions are color
coded.
low affinity thumb pocket 2 NS5B inhibitors featuring a thiazolone
scaffold (e.g., compound 3, Fig. 3, IC50 = 3 l
M).11
Based on the published crystal structure of 3 (PDB: 2HWI), this
class of compound appeared to be involved in similar interactions
as anthranilic acid based inhibitors within the thumb pocket 2 allo-
steric site of the protein (i.e., hydrogen bonds between the C@O
and N@C groups of the thiazolone ring to protein backbone Y477
and S476 NHs and hydrophobic interactions of the 4-fluorophenyl
and ethyl groups of the inhibitor with lipophilic enzyme pockets).
An overlay of the complexes of compounds 2 and 3 is shown in Fig-
ure 4 and reveals extensive overlap between the two structures.
The carboxyl group of 2 is perfectly superimposed on a portion
of the thiazolone ring system making similar H-bond interactions
with the protein backbone. The shift observed between the ben-
zene ring scaffold of 1 and the thiazolone ring of 3 suggested that
bicyclic systems may be accommodated within the binding pocket,
while maintaining the other key elements necessary for intrinsic
potency (e.g., a good overlap is seen for the sterically demanding
trans-4-methylcyclohexyl moiety of 2 and the benzyl group of 3).
Based on these observations, a 1H-quinazolin-4-one derivative 4
was designed from hybridization of the two structures as described
in Scheme 1.
(16, 18–20) with IC50 = 0.45–1.2 lM. Polar substituents (e.g., meth-
oxy, 17) resulted in a significant loss in potency. Second, substitu-
tion meta- to the attachment point was detrimental to potency in
all cases (e.g., 22 and 23). Finally, ortho-substitution appeared to
be beneficial and combinations of para- and ortho-substituents
led to the most interesting compounds (e.g., 25–27, 29 and 30) pro-
viding sub-micromolar inhibition of the polymerase (IC50 = 0.1–
0.5 lM) and the cell-based replicon (EC50 = 0.4–0.6 lM). Inhibitor
30 exhibited twofold improved intrinsic potency in the enzymatic
assay and was only threefold less potent in cell culture compared
to reference anthranilic acid derivative 1. This result demonstrated
that through optimization of interactions in the right-hand-side
hydrophobic pocket, quinazolinone analogs are capable of provid-
ing potency levels comparable to anthranilic acid derivatives, de-
spite the absence of the ionized carboxylic acid moiety.
Quinazolinone derivatives 4 and 11–30 were prepared by the
general route depicted in Scheme 2 starting from 2-nitro-5-
hydroxybenzoic acid 5. This acid was converted to its primary
amide 6 and the nitro group reduced to provide aniline 7. The tri-
fluoromethylpyridine ether was assembled by SNAr displacement
on 2-fluoropyridine 8 to give ether derivative 9. The N-alkyl substi-
tuent was introduced through reductive amination of anilines 9
using a variety of aldehydes and ring-closure to quinazolinone final
products (4, 11–30) was performed with trimethylorthoformate
under acidic conditions.12
O
O
O
b
HO
HO
a
HO
NH2
NH2
NH2
NO2
OH
NO2
7
6
5
CF3
CF3
O
F
c
O
+
NH2
RCHO
+
An initial set of quinazolinone analogs 4 and 11–16 (Table 1)
maintaining the trifluoromethylpyridine ether substituent of 1 at
C-6 of the bicycle was prepared and tested in previously described
N
N
NH2
8
9
D
21 NS5B polymerase and cell-based replicon assays.5a,13–15
CF3
O
CF3
O
N
gt1b
O
d
O
e
As mentioned previously, the trans-4-methylcyclohexyl moiety of
1 was also maintained in the initial design since replacement by
a benzylic moiety (as present in the less potent compound, 3)
was not tolerated in the anthranilic series. Direct replacement of
the carboxylic acid function of 1 with the neutral quinazolinone
acylimide moiety resulted in an inhibitor (4) that was ꢀ85-fold less
NH2
N
R
N
N
NH
10
4, 11 - 30
R
Scheme 2. Synthesis of quinazolinone derivatives: (a) (i) SOCl2, THF, 0 °C, (ii)
NH4OH, 95%. (b) 10% Pd/C, H2 (1 atm), MeOH, RT, 76%; (c) K2CO3 DMSO, 80 °C, 42%;
(d) aldehyde, NaBH3CN, AcOH, 50 °C; (e) trimethylorthoformate, TFA, 26–55% (two-
steps).
potent than 1 (IC50 = 6.9
lM) and did not inhibit in the cell-based
replicon (EC50 > 10 M). However, quinazolinone 4 had compara-
l

P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
4135
Table 1
Quinazolinone N1 SAR
CF3
O
O
N
N
N
R
g/mL) pH 2.0/6.8c
LogDd
a
b
Entry
4
R
Gt1b IC50
6.9
(l
M)
Gt1b EC50
>10
(
lM)
Solubility (
l
11
12
H
Me
4.4
24
>10
>10
11/4.7
2.5
13
14
15
4.3
2.4
1.6
13
8.4
1.0
86/3.3
4.7/0.5
3.3
3.7
16
17
18
19
0.45
48
2.5
>10
3.1
3.1
MeO
1.1
1.1
Cl
F
20
21
1.2
2.0
4.5/0.6
>7.5
>10
F3C
F
22
23
4.0
2.9
F
F
>7.5
>10
F
24
25
26
27
0.80
0.40
0.31
0.26
1.1
F
Cl
0.45
0.49
0.60
Cl
Br
F
3.6 / 0.7
14 / 3.8
F
3.8
F
F
28
29
0.20
0.51
0.10
1.2
F
0.51
0.41
19 / 4.3
108 / 26
F
F
F
F
30
3.3
F
a
Gt1b
D
21NS5B (nP2).
b
c
HCV replicon luciferase reporter assay (nP2).
Solubilities were measured on amorphous solids by the 24 h shaking flask method and are reported for pH 2.0 and pH 6.8.
Measured at pH 7.4.
d

4136
P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
Corroboration that this novel class of NS5B inhibitors binds to
O
N
N
the thumb pocket 2 allosteric site and rationalization of the SAR
assembled so far was provided by a crystal structure of compound
29 in complex with NS5B (Fig. 5). The expected conservation of all
key ligand–protein interactions between the two chemotypes is
apparent from the superposition of 29 with compound 2, shown
in Figure 6. The phenyl ring of the quinazolinone core also main-
tains the stacking with the side chain of L419.
O
N
F
F
F
F
I482
29
F
Y477
Quinazolinones exhibited moderate to satisfactory metabolic
stability in the presence of human liver microsomes (HLM), with
half-lives extending into the 2 h range for the best compounds
(data not shown). Despite the loss of the ionizable carboxylic acid
moiety, overall lipophilicity remained in a moderate range with
measured LogD7.4 = 3.3–3.8 for N-substituted benzyl analogs and
moderate solubility was observed at pH 2.0 and 6.8 for some ana-
logs. Interestingly, trifluorobenzyl derivative 30 exhibited the best
solubility in this set despite the presence of six fluorine atoms in
the molecule. SAR at the N1-position of the quinazolinone scaffold
is consistent with that substituent fitting in a rather small but
well-defined hydrophobic pocket, as was the case for anthranilic
acid derivatives (Fig. 2).7b
V494
S476
L497
L419
Figure 5. Crystal structure of inhibitor 29 complexed with NS5B polymerase
showing conservation of all key inhibitor–protein interactions relative to anthra-
nilic acid analogs (PDB 4JJU).
Having validated our hybrid design, we next turned to improv-
ing potency by optimizing key interactions of the inhibitors within
the protein binding channel. In particular, we investigated the
introduction of functionality on the phenyl ring of the scaffold to
modulate electronically the strength of hydrogen bonds to back-
bone Y477 and S476 NHs and influence stacking of the scaffold it-
self with L419 as shown in Figure 5. To this end, we investigated
substitutions at the C-7 position of the 1H-quinazolin-4-one ring
system using inhibitor 29 as a reference point. The synthesis of
these analogs is depicted in Scheme 3 and begins with commer-
cially available difluorobenzoic acid derivative 31.
Regioselective displacement of the more activated 3-fluoro
group with benzyl alcohol provided ether 32 that was sequentially
O
O
a
b
F
F
O
F
OH
NO2
OH
NO2
31
33
32
O
HO
F
F
O
O
O
F
F
F
c
HO
F
O
NH
F
+
+
F
NH2
F
N
34
F
F
F
Figure 6. Superposition of anthranilic acid and quinazolinone inhibitors 2 and 29,
showing overlap of key structural features and stacking over L419.
F
F
F
F
O
O
O
O
O
NH2
e
d
N
N
F
NH
F
F
NH
F
Based on the literature and our own in-house experience, the
binding of non-nucleoside NS5B inhibitors to three of the four allo-
steric sites (i.e., thumb pocket 1 and 2 and palm site 1) exploits
similar ligand pharmacophoric elements including hydrogen bonds
to the protein backbone (often implicating a carboxylic acid group
or amide) as well as two hydrophobic binders.9a In order to verify
that 1H-quinazolin-4-one inhibitors still bound in the intended
thumb pocket 2 site, compound 16 was profiled in a panel of single
amino acid site-specific mutants of HCV NS5B. As expected for a
thumb pocket 2 inhibitor, no potency shifts were observed for
the thumb pocket 1 (P495S) or palm site 1 (M414T) mutants.9a
Interestingly, compound 16 experienced a greater shift against
the L419 M (>20-fold) thumb pocket 2 mutants than the usual
M423T variant (threefold) that has been reported for other chem-
otypes that bind to this allosteric site.9a Similar profiles were seen
for more optimal analogs (vide infra).
36
35
F
F
F
F
F
F
F
F
O
N
O
O
X
O
N
N
g
f
N
N
F
F
N
F
37
F
F
X = OMe (38) or NMe2 (39)
Scheme 3. Synthesis of C-7 substituted 1H-quinazolin-4-ones: (a) BnOH, NaH,
DMF, 84%; (b) (i) MeOH, H2SO4, reflux, (ii) 10% Pd/C, H2 (1 atm), MeOH, RT, 75%; (c)
2,4-difluorobenzaldehyde, NaBH3CN, AcOH, 50 °C, 84%, (d) Cs2CO3, MeCN, 80 °C,
43%; (e) (i) 10 N NaOH, DMSO, (ii) NH4CO3, HATU, THF; (f) trimethylorthoformate,
TFA, 97% (three-steps); (g) X = OMe: MeOH, 10 N NaOH, 32%; X = Me2N: dimeth-
ylamine, THF, Et3N, 11%.

P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
4137
Table 2
C-7 quinazolinone SAR
CF3
O
N
F
O
R
N
N
F
g/mL) pH 2.0/6.8c
LogDd
a
b
Entry
R
Gt1b IC50
(l
M)
Gt1b EC50
(lM)
Solubility (
l
29
37
38
39
H
F
OMe
Me2N
0.51
0.24
2.4
0.51
0.65
0.85
>10
19/4.3
22/18
3.9
25
a
Gt1b
D
21NS5B (nP2).
b
c
HCV replicon luciferase reporter assay (nP2).
Solubilities were measured on amorphous solids by the 24 h shaking flask method and are reported for pH 2.0 and pH 6.8.
Measured at pH 7.4.
d
esterified and hydrogenolyzed to phenol 34. SNAr reaction with 3-
trifluoromethyl-2-fluoropyridine as previously described gave
ether 35 that was converted to amide 36 prior to quinazolinone
ring-formation under the usual conditions to yield 37. Analogs 38
and 39 were obtained from 37 by displacement of the fluorine
atom with MeOH or dimethylamine under basic conditions. As de-
picted in Table 2, the effect on potency of an electron withdrawing
fluorine in the para-position of the quinazolinone carbonyl group
was negligible (analog 37). On the other hand, bulkier substitution
at this position (e.g., MeO or Me2N; compounds 38 and 39) had a
strong negative impact on intrinsic potency (5- to 50-fold)
although 38 retained good replicon cell-based activity. The loss
in enzymatic potency may be the result of steric clashes with res-
idue L497 which is situated in close proximity to the scaffold (see
Fig 5), and/or unfavorable conformational effects between C-6 and
C-7 substituents altering the ability of the ether linkage to properly
position the trifluoromethylpyridyl group.
The next phase of this prospecting study focused on SAR of the
O-pyridyl heterocyclic ring system at C-6 using the 2,4-difluorob-
enzyl of 29 or 2,4,6-trifluorobenzyl group of 30 as reference points
(Table 3). The synthesis of these analogs is shown in Scheme 4 and
begins with the reductive amination of 2,4-difluoro or 2,4,6-triflu-
orobenzaldehyde with phenol 7 followed by quinazolinone ring-
formation with trimethylorthoformate to provide intermediate
40. Phenol 40 was converted to a variety of ether derivatives
(42–50) under SNAr conditions using activated halo-heterocycles.
Methoxy analog 41 was prepared by direct alkylation of phenol
40 with iodomethane.
The O-heterocycle C-6 group is an important contributor to
potency in this class of inhibitors as deduced from the 10-fold loss
in potency for methoxy analog 41 relative to reference compound
29 (Table 3). While it is apparent from X-ray crystal structures that
the trifluoromethyl group provides potency through binding in a
lipophilic pocket, the role of the nitrogen atom was probed with
isomeric pyridine analogs 42 and 43. Interestingly, positioning of
the nitrogen atom ortho- to the CF3 group (42) resulted in a four
to fivefold improvement in intrinsic potency, while meta-isomer
43 was more or less equipotent. Both analogs exhibited improved
aqueous solubility compared to 29. The improved intrinsic potency
of 42 translated into improved cell culture activity, providing one
of the best analogs thus far in this class (EC50 = 140 nM). The
trifluoromethyl group was more effective for interacting with the
enzyme sub-pocket than other groups (e.g., Br, compound 44). An-
other increase in potency was realized with a methoxy group ortho
to the CF3 substituent of 29 that resulted in a 2-fold improvement
in both enzymatic and replicon potencies (analog 45). The
beneficial findings encountered in the 2,4-difluorobenzyl series
were transposable into the 2,4,6-trifluorobenzyl analogs (com-
pounds 30, 46–50) resulting in additional compounds displaying
excellent potency (e.g., analog 47: gt1b EC50 = 100 nM). Interest-
ingly, removal of the nitrogen atom from this ring was also well
tolerated with 2-trifluomethylphenoxy analog 48 providing
excellent inhibition of the polymerase (IC50 = 75 nM) but
somewhat offset by its reduced cellular potency (EC50 = 380 nM),
high lipophilicity (LogD7.4 = 4.2) and low solubility. The trifluoro-
methyl group however remains an essential component to main-
tain the potency of this chemotype in an acceptable range (cf.,
compound 49).
Analog 45, a representative member of this new class of NS5B
inhibitors, was found to be HCV-specific and did not inhibit RNA-
dependent RNA polymerase from poliovirus (IC50 >100
mammalian RNA polymerase II isolated from calf thymus (IC50
>210
M).13 Furthermore, in light of the high prevalence of gt1a
lM) or a
l
HCV in North America, it was also important to assess the ability
of this class to inhibit this particular virus sub-type. Compounds
45 and 50 were tested in replicon assays in which HCV RNA was
quantified by RT-PCR (gt1b/1a).14a In these assays, comparable
gt1b/1a EC50 values of 0.76–0.83 lM and 0.81 (gt1a) were obtained
for the two compounds. Similar to anthranilic acids 1 and 2, the
quinazolinone thumb pocket 2 inhibitors inhibit both gt1 replicon
sub-types 1a and 1b with comparable efficiency. Consistent with
other thumb pocket 2 ligands however, compounds from this class
(e.g., 45) demonstrated significantly lower potency when tested
against NS5B genotypes 2b and 3a polymerases (>30- to 60-fold).
Analog 45 exhibited comparable potency against gt4a enzyme.9c,16
Having identified inhibitors displaying good potency in the cell-
based replicon assay (EC50 <200 nM) we wished to assess prospects
for acceptable ADME-PK profiles in this class. A set of representa-
tive quinazolinone analogs were evaluated for in vitro metabolic
stability, Caco-2 permeability, CYP450 inhibition and oral absorp-
tion in rats. The data is presented in Table 4 and is compared to
representatives from the anthranilic acid series (compounds 1
and 2). Analogs from both series had comparable metabolic stabil-
ity in human liver microsomes (HLM; mostly in the t1/2 = 1–2 h
range except for 47 which was significantly more stable). In con-
trast, compounds from both series were noticeably less stable in
rat microsome incubations (RLM; t1/2 < 1 h). Caco-2 permeability
was excellent in the apical ? basolateral direction and no signifi-
cant efflux was noted for analog 45 (results not shown). Though
some quinazolinone analogs showed varying levels of CYP2C19

4138
P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
Table 3
Quinazolinone left-hand side Sar
O
O
R
N
N
X
F
F
l
g/mL) pH 2.0/6.8c
LogDd
a
b
Entry
R
X
Gt1b IC50
64
(
lM)
Gt1b EC50
>10
(
lM)
Solubility (
>900/130
41
MeO
H
1.8
N
29
H
0.51
0.51
19/4.3
CF3
CF3
CF3
42
43
44
H
H
H
0.12
0.49
0.86
0.14
1.1
410/105
700/31
N
3.0
N
N
2.7
Br
N
45
30
H
F
0.23
0.10
0.25
0.28
0.4/0.6
108/26
3.5
3.3
CF3
OMe
N
CF3
CF3
CF3
CF3
N
46
47
F
F
0.11
0.06
0.23
0.10
27/4.6
34/2.7
3.0
2.9
N
48
49
F
F
0.075
0.30
0.38
0.40
0.4/<0.1
34/4.6
4.2
3.2
N
50
F
0.10
0.17
6.1/5.4
3.3
CF3
OMe
a
b
c
Gt1b
D21NS5B (nP2).
HCV replicon luciferase reporter assay (nP2).
Solubilities were measured on amorphous solids by the 24 h shaking flask method and are reported for pH 2.0 and pH 6.8.
Measured at pH 7.4.
d
inhibition (e.g., IC50 = 0.3 and 1.2
exhibited relatively moderate inhibition across CYP450 isozymes
(IC50 = 11, 20, >30 and 18 M for 2C9, 2C19, 2D6 and 3A4, respec-
lM for 45 and 47) compound 50
O
N
F
O
l
O
HO
a
N
tively). Unfortunately, the quinazolinones in Table 4 exhibited re-
duced exposure compared to anthranilic acid derivatives, when
administered orally to rats, either as mixtures of four compounds
(screening mode) or as individual compounds.17 The poor exposure
(and bioavailability) in rats is likely due to a combination of low
metabolic stability of compounds in this species (as reflected by
the lower RLM stabilities compared to anthranilic acids), resulting
in high in vivo clearances, and the lower solubility of the com-
pounds. The cross-species metabolic trends were similar in liver
microsome and hepatocyte preparations where compound 50
had in vitro clearance values of =6.6, 38, 4.7 and 13 ml/min/kg in
human, rat, dog and monkey microsomes and 2.2, 33, 4.2 and
4.1 ml/min/kg in hepatocytes, respectively. The low stability of
quinazolinones in rat liver microsomes or hepatocytes appears to
HO
X
F
NH2
+
X
NH2
40
7
F
F
CF3
CF3
O
b
Cl
O
+
X
X
R
X
N
X
X
X
X
N
F
R
42 - 50
X = H or F
F
Scheme 4. (a) (i) Aldehyde, NaBH3CN, AcOH, (ii) trimethylorthoformate, TFA, 40%,
(b) K2CO3, DMSO, 80 °C, 13–86%.

P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
4139
Table 4
In vitro ADME and oral rat PK parameters for representative compounds.
c
d
e
Entries
HLM/RLM t1/2
AB Caco-2b (cm/s)
C1/2 h
M)
Cmax
M)
AUCd
M.h)
t1/
Vss
(L/
CLe
(mL/min/
kg)
%F
[liver] (liver/plasma ratio)f
M)
(min)a
Â10À6
(
l
(
l
(
l
(l
e
2
(h)
kg)
1
2
69/60
119/92
93/60
19
17
7.1
8.4
13
12
11
11
7.9
1.0/0.2
14.3
21.7
16
3.4
6.8
26
27
29
30
45
47
50
0.3/0.3
0.4/0.2
0.2/0.1
0.2/0.1
6.6 (23)
1.8 (9)
0.8 (11)
1.1 (15)
45/15
110/31
123/38
80/26
>300/22
84/15
0.03
0.07
0.20
0.45
0.5
0.6
3.2
0.7
50
19
7
5
0.1/0.03
<0.01
a
b
c
Human and rat liver microsome half-life at 2
lM starting concentration.
Apical to Basolateral permeability at 10 M using a pH6.8–7.4 gradient.
l
A mixture of four compounds was administered orally as a suspension to rats at 2 mg/kg each (0.5% Methylcellulose and 0.3% Tween-80 + 1% N-methylpyrrolidone) and
individual plasma compound concentrations were determined at 1 and 2 h timepoints.17
d
Single compound PK dosed at 5 mg/kg as an oral suspension in 0.5% methylcellulose and 0.3% Tween-80 + 1% N-methylpyrrolidone (NMP).
Bolus injection (2 mg/kg) prepared in 70% PEG-400:30% water.
Liver compound concentration and ratio of compound concentration in liver/plasma at 2 h post a single oral dose.
e
f
Biotechnol. 2011, 29, 993; (c) Berman, K.; Kwo, P. Y. Clin. Liver Dis. 2009, 13, 429;
(d) Asselah, T.; Marcellin, P. Liver Int. 2011, 31, 68.
2. (a) Soriano, V. ACS Med. Chem. Lett. 2012, 3, 440; (b) Gane, D. Liver Int. 2011, 31,
62; (c) Pockros, P. J. Drugs 2012, 72, 1825.
be specific to that species. Importantly, compounds exhibited good
distribution to the liver target organ as reflected by compound li-
ver/plasma concentration ratios (9- to 23-fold). Since this chemo-
type features several aromatic rings and LogD values in the 3–4
range, a few analogs were tested in a plasma protein-free func-
tional hERG inhibition assay. Compounds 16, 27 and 49 inhibited
3. (a) For recent reports on all-oral triple DAAs combinations see: Kowdley, K. V.;
Lawitz, E.; Poordad, F.; Cohen, D. E.; Nelson, D. R.; Zeuzem, S.; Everson, G. T.;
Kwo, P. Y.; Foster, G. R.; Sulkowski, M. S.; Xie, W.; Pilot-Matias, T.; Liossis, G.;
Larsen, L. M.; Khatri, A.; Podsadecki, T. J.; Bernstein, B. Presented at the 63rd
Annual Meeting of the American Association for the Study of the Liver Diseases
(AASLD) November 9–13th, Boston, MA, Late-Breaker Abstract LB-1; Hepatology
2012, 56.; (b) Everson, G. T.; Sims, K. D.; Rodriguez-Torres, M.; Hezode, C.;
Lawitz, E.; Bourliere, M.; Loustaud-Ratti, V.; Rustgi, V. K.; Schwartz, H.; Tatum, H.
A.; Marcellin, P.; Pol, S.; Thuluvath, P. J.; Eley, T.; Wang, X.; Huang, S.-P.; McPhee,
F.; Wind-Rotolo, M.; Chung, E.; Pasquinelli, C.; Grasela, D. M.; Gardiner, D. F.
Presented at the 63rd Annual Meeting of the American Association for the Study
of the Liver Diseases (AASLD) November 9–13th, Boston, MA, Late-Breaker
Abstract LB-3; Hepatology 2012, 56.; (c) Pelosi, L. A.; Voss, S.; Liu, M.; Gao, M.;
Lemm, J. A. Antimicrob. Agents Chemother 2012, 56, 5230.
4. (a) White, P. W.; Llinàs-Brunet, M.; Amad, M.; Bethell, R. C.; Bolger, G.;
Cordingley, M. G.; Duan, J.; Garneau, M.; Lagacé, L.; Thibeault, D.; Kukolj, G.
Antimicrob. Agents Chemother. 2010, 54, 4611; (b) Zeuzem, S.; Asselah, T.;
Angus, P.; Zarski, J.-P.; Larrey, D.; Müllhaupt, B.; Gane, E.; Schuchmann, M.;
Lohse, A.; Pol, S.; Bronowicki, J.-P.; Roberts, S.; Arasteh, K.; Zoulim, F.; Heim, M.;
Stern, J. O.; Kukolj, G.; Nehmiz, G.; Haefner, C.; Boecher, W. O. Gastroenterology
2011, 141, 2047; (c) Zeuzem, S.; Gane, E.; Roberts, S.; Mensa, F. J. Gastroenterol.
Hepatol. 2012, 27, 182.
5. (a) Beaulieu, P. L.; Anderson, P. C.; Brochu, C.; Bös, M.; Cordingley, M. G.; Duan,
J.; Garneau, M.; Lagacé, L.; Marquis, M.; McKercher, G.; Poupart, M.-A.;
Rancourt, J.; Stammers, T.; Tsantrizos, Y. S.; Kukolj, G. J. Hepatol. 2012, 56,
S321; (b) Larrey, D. G.; Benhamou, Y.; Lohse, A. W.; Trepo, C.; Moelleken, C.;
Bronowicki, J.-P.; Heim, M. H.; Arastéh, K.; Zarski, J.-P.; Boulière, M.; Wiest, R.;
Calleja, J. L.; Enríquez, J.; Erhardt, A.; Wedemeyer, H.; Gerlach, T.; Nehmiz, G.;
Boecher, W. O.; Berger, F. M.; Steffgen J. Hepatol. 2009, 50, 1044A; (c) Larrey, D.;
Lohse, A. W.; de Ledinghen, V.; Trepo, C.; Gerlach, T.; Zarski, J.-P.; Tran, A.;
Mathurin, P.; Thimme, R.; Arastéh, K.; Trautwein, C.; Cerny, A.; Dikopoulos, N.;
Schuchmann, M.; Heim, M. H.; Gerken, G.; Stern, J. O.; Wu, K.; Abdallah, N.;
Girlich, B.; Scherer, J.; Berger, F.; Marquis, M.; Kukolj, G.; Böcher, W.; Steffgen, J.
J. Hepatol. 2012, 57, 39; (d) Zeuzem, S.; Asselah, T.; Angus, P.; Zarski, J.-P.;
Larrey, D.; Müllhaupt, B.; Gane, E.; Schuchmann, M.; Lohse, A.; Pol, S.;
Bronowicki, J.-P.; Roberts, S.; Arastéh, K.; Zoulim, F.; Heim, M.; Stern, J. O.;
Kukolj, G.; Nehmiz, G.; Haefner, C.; Böcher, W. O. Gastroenterology 2011, 141,
2047.
the activity of this channel with IC50 = 17, 14 and 3.7 lM respec-
tively. Compound 50 was highly bound to human and rat plasma
proteins (99.5%), a property that could reduce free compound frac-
tion and modulate potency as well as in vivo properties such as
distribution and metabolism, and compound exposure to off-tar-
gets such as the hERG channel and CYP450 enzymes.
In summary, a structure-guided design strategy based on
hybridization of thiazolone and anthranilic acid NS5B thumb pock-
et 2 inhibitors led to the discovery of a specific and novel 1H-qui-
nazolin-4-one lead chemotype. Prospecting SAR studies
demonstrated the potential for generating potent inhibitors of
gt1a and 1b replicons (gt1b EC50 <200 nM). Examination of struc-
tural data from these early compounds (Fig. 5) reveals additional
opportunities for further potency improvements. Elaboration of
molecules toward the left-hand side of the binding pocket (shown
as the orange-delineated channel in Fig. 5) guided by molecular
dynamics simulations has led to the identification of additional
interactions with protein residues and further improvements in
potency (e.g., extensions into the hydrophobic pocket from the tri-
fluoromethylpyridine ring toward the backbone carbonyl of V494
and the end of the pocket). These results,18 as well as optimization
studies addressing other aspects requiring improvement such as
the ADME-PK and off-target profile of the lead compounds de-
scribed in this work will be reported separately.
Acknowledgments
6. (a) Zeuzem, S.; Soriano, V.; Asselah, T.; Bronowicki, J.-P.; Lohse, A.; Müllhaupt,
B.; Schuchmann, M.; Bourlière, M.; Buti, M.; Roberts, S.; Gane, E.; Stern, J. O.;
Kukolj, G.; Dai, L.; Böcker, W. O.; Mensa, F. J. J. Hepatol. 2012, 56, S45; (b)
Soriano, V.; Gane, E.; Angus, P.; Stickel, F.; Bronowicki, J.-P.; Roberts, S.; Manns,
M.; Zeuzem, S.; Dai, L.; Boecher, W.; Stern, J.; Mensa, F. J. Hepatol. 2012, 56,
S559.
7. (a) Stammers, A. T.; Coulombe, R.; Rancourt, J.; Thavonekham, B.; Fazal, G.;
Goulet, S.; Jakalian, A.; Wernic, D.; Tsantrizos, Y.; Poupart, M.-A.; Bös, M.;
McKercher, G.; Thauvette, L.; Kukolj, G.; Beaulieu, P. L. Bioorg. Med. Chem. Lett.
2013, 23, 2585; (b) Stammers, A. T. et al., submitted for publication.
8. For reviews on NS5B inhibitors see for example: (a) Beaulieu, P. L. Expert Opin.
Ther. Pat. 2009, 19, 145; (b) Watkins, W. J.; Ray, A. S.; Chong, L. S. Curr. Opin.
Drug Discov. Devel. 2010, 13, 441; (c) Sofia, M. J.; Chang, W.; Furman, P. A.;
Mosley, R. T.; Ross, B. S. J. Med. Chem. 2012, 55, 2532.
We thank the following colleagues from Boehringer Ingelheim
(Canada) Ltd. that participated in generating the data presented
in this Letter: Nathalie Dansereau, Lisette Lagacée, Martin Marquis,
Ginette McKercher, Charles Pellerin and Louise Thauvette for EC50
and IC50 determinations. Gordon Bolger, Josie De Marte, Michel
Garneau, Francine Liard, Hélène Montpetit and Christine Zouki
for ADME-PK profiling. Norman Aubry, Sylvain Bordeleau, Colette
Boucher and Serge Valois for analytical support.
References and notes
9. (a) Shi, S. T.; Herlihy, K. J.; Graham, J. P.; Nonomiya, J.; Rahavendran, S. V.; Skor,
H.; Irvine, R.; Binford, S.; Tatlock, J.; Li, H.; Gonzalez, J.; Linton, A.; Patick, A. K.;
Lewis, C. Antimicrob. Agents Chemother. 2009, 53, 2544; (b) Yi, G.; Deval, J.; Fan,
1. (a) Gentile, I.; Carleo, M. A.; Borgia, F.; Castaldo, G.; Borgia, G. Expert Opin.
Investig. Drugs 2010, 19, 151; (b) Kwong, A. D.; Kauffman, R. S.; Mueller, P. Nat.

4140
P. L. Beaulieu et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4132–4140
B.; Cai, H.; Soulard, C.; Ranjith-Kumar, C. T.; Smith, D. B.; Blatt, L.; Beigelman, L.;
17. All rat PK studies were performed at Boehringer Ingelheim (Canada) Ltd. PK
studies in dogs and monkeys were performed at LAB Pre-Clinical Research
International Inc, Laval, QC. All protocols involving animal experimentation
were reviewed and approved by the respective Animal Care and Use
Committee of each test facility. In-life procedures were in compliance with
the Guide for the Care and Use of Laboratory Animals from the Canadian
Council of Animal Care. All chemicals used were reagent grade or better.
Animals were fasted overnight prior to dosing. IV dose administration
(2 mg/kg) was performed using a 70% PEG400:30% water dosing solution.
Kao, C. C. Antimicrob. Agents Chemother. 2012, 56, 830.
10. The coordinates for NS5B polymerase in complex with compounds 2 and 29
have been deposited in the Protein Data Bank as entries 4JJS and 4JJU,
respectively.
11. Yan, S.; Larson, G.; Wu, J. Z.; Appleby, T.; Ding, Y.; Hamatake, R.; Hong, Z.; Yao,
N. Bioorg. Med. Chem. Lett. 2007, 17, 63.
12. Inhibitors had >95% homogeneity as determined by reversed-phase HPLC analysis
and had 1H NMR and mass-spectral data consistent with their structures.
13. IC50 determinations using a
D
21 NS5B construct were performed in duplicates
Oral dose administration (5 mg/kg) was performed using a suspension
as described elsewhere: (a) McKercher, G.; Beaulieu, P. L.; Lamarre, D.;
LaPlante, S.; Lefebvre, S.; Pellerin, C.; Thauvette, L.; Kukolj, G. Nucleic Acids Res.
2004, 32, 422.
containing 0.3% Tween-80 and 0.5% methylcellulose, with 1% NMP, as
additional solubilizer (N = 3 per dose and per route). In the cassette screen
experiments, each ‘cassette’ containing 4 compounds at 2 mg/kg for each
compound was dosed to two rats. Blood samples collected from all time
points were placed on ice, and then centrifuged at 4 °C. The plasma was
separated and stored frozen at approximately À20 °C until analysis. Plasma
samples were extracted by solid phase extraction. Samples were analyzed
by HPLC using either a UV diode array detector between 200 and 400 nm
with quantitative determination made by peak height at the wavelength
representing the best signal to noise ratio or a LC/MS systems using the
appropriate retention time and m/z+. Calibration standards were prepared
in blank plasma. The calibration curve was linear to cover the time-
concentration curve with an r2 values >0.99, and a limit of quantification
(LOQ) at 7 ng/mL. The temporal profiles of drug concentrations in plasma
were analyzed by non-compartmental methods using WinNonlin (version
3.1; Scientific Consulting, Inc, Cary, NC).
14. (a) EC50 determinations using the cell-based 1b luciferase reporter assay or
replicon assays using RT-PCR for RNA quantification were performed in
duplicates as described elsewhere: Beaulieu, P. L.; Fazal, G.; Goulet, S.; Kukolj,
G.; Poirier, M.; Tsantrizos, Y.; Jolicoeur, E.; Gillard, J.; Poupart, M.-A.; Rancourt,
J. WO Patent WO 03/010141, 2003.; (b) Llinàs-Brunet, M.; Bailey, M. D.;
Goudreau, N.; Bhardwaj, P. K.; Bordeleau, J.; Bös, M.; Bousquet, Y.; Cordingely,
M. G.; Duan, J.; Forgione, P.; Garneau, M.; Ghiro, E.; Gorys, V.; Goulet, S.;
Halmos, T.; Kawai, S. H.; Naud, J.; Poupart, M.-A.; White, P. W. J. Med. Chem.
2010, 53, 6466; (c) Vaillancourt, F. H.; Pilote, L.; Cartier, M.; Lippens, J.; Liuzzi,
M.; Bethell, R. C.; Cordingley, M. G.; Kukolj, G. Virology 2009, 387, 5.
15. In a MTT assay using Huh-7 cells, the compounds were not significantly
cytotoxic at the highest concentration tested (TC50 >10 lM).
16. See for example: Fenaux, M.; Tian, Y.; Matles, M.; Mabery, E.; Zhang, J.; Eng, S.;
Murray, B.; Mwangi, J.; Lazerwith, S.; Lew, W.; Canales, E.; Liu, Q.; Byun, D.;
Doerffler, E.; Ye, H.; Clarke, M.; Mertzman, M.; Morganelli, P.; Zhang, J.; Leavitt,
S.; Appleby, T.; Hashash, A.; Bidgood, A.; Krawczyk, S.; Watkins, W. J. Hepatol.
2011, 54, S476.
18. Hucke, O.; Coulombe, R.; Bonneau, P.; Bertrand-Laperle, M.; Brochu, C.; Gillard,
J.; Joly, M. - A.; Landry, S.; Lepage, O.; Llinàs-Brunet, M.; Pesant, M.; Poirier,
Martin.; Poirier, Maude; McKercher, G.; Marquis, M.; Kukolj, G.; Beaulieu, P. L.;
Stammers, T. A. Med. Chem. accepted for publication.