Structure-activity relationship (SAR) studies of quinoxalines as novel HCV NS5B RNA-dependent RNA polymerase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2006.12.103

From chemical compound library screening using an HCV NS5B RNA-dependent RNA polymerase enzymatic assay, we identified a substituted quinoxaline hit with an IC₅₀ of 5.5 μM. A series of substituted quinoxaline amide derivatives were synthesized

Full text of DOI:10.1016/j.bmcl.2006.12.103

Bioorganic & Medicinal Chemistry Letters 17 (2007) 1663–1666
Structure–activity relationship (SAR) studies of quinoxalines as
novel HCV NS5B RNA-dependent RNA polymerase inhibitorsq
Frank Rong,* Suetying Chow, Shunqi Yan, Gary Larson, Zhi Hong and Jim Wu
Drug Discovery, Valeant Pharmaceutical Research & Development, 3300 Hyland Avenue, Costa Mesa, CA 92626, USA
Received 22 November 2006; accepted 22 December 2006
Available online 8 January 2007
Abstract—From chemical compound library screening using an HCV NS5B RNA-dependent RNA polymerase enzymatic assay, we
identified a substituted quinoxaline hit with an IC50 of 5.5 lM. A series of substituted quinoxaline amide derivatives were synthe-
sized based on the hit’s pharmacophore, and a good structure–activity relationship was observed. Computer modeling analysis was
employed to help comprehend the SAR.
Ó 2007 Elsevier Ltd. All rights reserved.
Chronic hepatitis C virus (HCV) infection is the
leading cause of liver diseases and hepatocarcinoma.1
An estimation suggests that 170 million people
worldwide are infected with the virus with 4 million
of them residing in the United States.2 The HCV viral
genome exhibits a high degree of heterogeneity, which
presents a formidable hurdle to develop effective
prophylactic vaccines. Current standard therapies con-
sist of combinations of pegylated IFN-a and ribavirin,
and are inadequate to cure the viral infection in a
significant proportion of patients, especially those
infected with genotype 1 HCV.3 Progresses have been
made in the discovery of direct anti-HCV inhibitors,
especially the HCV NS3 protease inhibitors. However,
they are presently still in various stages of clinical
development. Thus, it is important to continue the
effort to discover and develop novel and potent anti-
HCV therapeutics.
NS5B RdRp constitutes the key component of the
viral replication machinery. It plays a central role in
the viral RNA synthesis, and has been proven to be
a validated drug target.6 We have devised an HCV
NS5B RdRp enzymatic assay to screen our small
molecular compound library for NS5B inhibitors.
From the high throughput screening, an interesting
hit 2 was identified. Although this compound is not
particularly potent with an IC50 of 5.5 lM, its
quinoxaline 6-carboxylate pharmacophore bears good
resemblance to a known benzoimidazole 5-carboxylate
NS5B RdRp inhibitor 1 reported by Boehringer
Ingelheim.7 The hit shares the same structural feature
of a bicyclic aromatic core substituted with two single
cyclic groups at the left side and a carboxylate group
at the right. Since 1 has been successfully optimized to
achieve nanomolar potency, we attempted to optimize
compound 2 through the similar SAR studies. In
this communication, we report the chemical synthesis
and structure–activity relationship (SAR) of these
quinoxaline compounds.
HCV is a small positive-sense single-stranded RNA
virus that belongs to the Flaviviridae family.4 Its
RNA genome encodes a few enzymes essential for
viral replication, including the NS3 protease and
NS5B RNA-dependent RNA polymerase (RdRp).5
Naturally they have become popular drug targets.
O
F
5
O
N
N
OH
2
6
3
N
N
OH
O
2
1
Keywords: HCV; NS5B polymerase; Quinoxaline; Inhibitor.
q
This paper is dedicated to Professor Albert H. Soloway of The Ohio
State University at the occasion of his 82th birthday.
F
1, IC50 = 1.6 uM
(Boehringer Ingelheim)
*
Hit: 2, IC50 = 5.5 uM
Corresponding author. Tel.: +1 714 729 5548; fax: +1 714 641
7222; e-mail: frong@ardeabiosciences.com
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.12.103

1664
F. Rong et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1663–1666
Table 2. IC50 values of compounds 8–29
O
O
Ar
Ar
O
O
H2N
H2N
Ar
Ar
N
N
O
R1
a
OH
OH
O
Ar
Ar
N
N
N
H
X
A
B
C
b
N
H
O
R1
Compound
Ar
R1
X
IC50 (lM)
O
8
8a
2-Furyl–
Ph–
OCH3
OH
H
>100
33
Ar
Ar
N
N
R1 = -OMe
R1 = -OH
N
H
H
c
9
9a
OCH3
OH
H
>100
15
H
X
HN
10
4-F–Ph–
4-Me–Ph–
3-OMe–Ph–
2-Pyridyl–
2-Furyl–
Ph–
OCH3
OH
H
>100
1.9
10a
11
H
Scheme 1. Reagents and conditions: (a) NaOAc, AcOH, 118 °C, 4 h;
(b) amino acid methyl ester, TBTU, DIPEA, DMF, 0 °C–rt, 24 h; (c)
NaOH, DMF, H2O, rt, 5–16 h.
OCH3
OH
H
>100
ND
>100
>100
>100
>100
26
11a
12
ND
H
OCH3
OH
12a
13
H
OCH3
OH
H
Compounds were generally prepared according to
Scheme 1. Basically, condensation of a-diketones (A)
with 3,4-diaminobenzoic acid (B) in the presence of
sodium acetate in acetic acid afforded 2,3-biaryl
substituted quinoxaline-6-carboxylic acid derivatives
(C) with a typical yield of 80–100%.8 The compounds
(C) were precipitated out of the solution upon addition
of cold water and were isolated by vacuum filtration.
Coupling of C with L-tryptophan methyl or ethyl ester
hydrochloride using TBTU and N,N-diisopropyl-ethyla-
mine followed by saponification at the room tempera-
ture provided the desired products.7 The resulting
compounds were separated by reverse phase HPLC,
and their corresponding structures were confirmed by
1H and 13C NMR and high resolution MS analysis. In
some cases, the structures were assigned using 2-D
NMR. These resultant compounds were evaluated for
inhibitory activity against HCV NS5B RdRp using a
published procedure.9 Their IC50s were determined,
which are summarized in Tables 1–4.
13a
14
H
OCH3
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
14a
15
16
72
OCH3
OH
15a
16
11
5.5
4-F–Ph–
4-Me–Ph
3-OMe–Ph–
OCH3
OH
16a
17
1.3
49
OCH3
OH
17a
18
6.1
>100
15
OCH3
OH
18a
by 8- and 14-fold, respectively. This highlights the
sensitivity of the size and electronic property of the
substituents on the phenyl ring. The addition of –OMe
at the meta position of phenyl ring in compound 6
rendered a total loss of activity with an IC50 > 100 lM.
Replacement of the phenyl ring with a furyl or 2-pyridyl
group in compounds 3 and 7 also led to diminishing
activity.
We first set out to explore SAR of the aromatic
substituents of the initial hit 2 by synthesizing a small
focused library of compounds 2–7 (Table 1). Compound
2 with bis 4-F-phenyl substitutions was the most
potent among them with an IC50 of 5.5 lM. By
comparison, compounds with unsubstituted phenyl (4)
or 4-Me-phenyl (5) reduced the inhibitory potency
It was reported that the modification of the carboxylate
group in 1 by coupling it with L-tryptophan derivatives
afforded compounds with nanomolar inhibitory poten-
cy.7 We applied the same strategy to the hit 2 and made
a series of compounds. As illustrated in Table 2, the
most potent compound in this series, 16a, had an IC50
of 1.3 lM, which is 4-fold lower than that of 2. Interest-
ingly, compound 16a carries 4-F-phenyl substituents as
2. Other compounds with different Ar groups, such
as 8a–15a, 17a, 18a, all exhibited weaker activities
(Table 2). It is also worth to note that the ester deriva-
tive 16, in comparison to 16a, is 4-fold weaker, which
applies to all the ester derivatives of the carboxylic acid
counterparts listed in Table 2 (8–18). This result suggests
the potential importance of –COOH group in forming
NS5B-inhibitor interactions.
Table 1. Initial SAR studies on 2
O
3
2
Ar
Ar
N
N
OH
Compound
Ar
IC50 (lM)
2
3
4
5
6
7
4-F–Ph–
2-Furyl–
Ph–
5.5
17
The improved potency of 16a over the initial hit 2
prompted us to further probe a more diversified set of
amino-acid derivatives. As the 4-F-phenyl had been
shown to be the optimized among the substitutions
evaluated, we kept it unchanged in the new series of
79
40
4-Me–Ph
3-OMe–Ph
2-Pyridyl–
>100
>100

F. Rong et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1663–1666
1665
Table 3. SAR of 2, 3-difluorophenyl quinoxalines
Table 4. SAR of 2,3-asymmetric quinoxalines
F
W
O
OH
OH
O
OH
N
N
R2
N
H
O
2
NH
NH
1
O
CO
H
2
W
CO
H
F
2
R1
R2
N
N
O
N
H
3
4
Compound R2
Y
IC50 (lM)
H
N
O
Y
CO
H
2
O
5
10
10a
OMe >100
OH 1.9
Compound
R1
R2
W
1
IC50 (lM)
25
Ph–
Cyclohexyl–
Ph–
Cyclohexyl–
Ph–
3.7
1.8
5.0
0.6
3.4
1.6
4.1
4.6
1.3
4.1
4.6
1.8
1.6
5.1
1.3
N
H
25a
26
Cyclohexyl–
Ph–
2
1
2
O
Y
26a
27
Cyclohexyl–
4-F–Ph–
Cyclohexyl–
4-F–Ph–
Cyclohexyl–
Ph–
Cyclohexyl–
4-F–Ph–
27a
28
16
16a
OMe 5.5
OH 1.3
Cyclohexyl–
4-F–Ph–
OH
28a
29
N
H
Cyclohexyl–
Cyclohexyl–
4-F–Ph–
3
3
30
4-F–Ph–
Cyclohexyl–
Ph–
30a
31
Cyclohexyl–
Ph–
4
5
O
Y
31a
32
Cyclohexyl–
4-F–Ph–
Cyclohexyl–
Cyclohexyl–
4-F–Ph–
O
19
19a
OMe 100
OH 1.7
32a
O
Y
N
H
compounds. As illustrated in Table 3, many of the
compounds from this series showed similar activity
as 16a. Compound 23a is the most potent one, show-
ing an IC50 of 0.69 lM, which is a 2-fold improvement
over 16a. A closely related analog (24a) carrying a
cyclopentyl ring is slightly weaker than 23a. Other
L-tryptophan containing carboxylic acid derivatives,
10a–21a, are almost equally potent as 16a. Similar to
the results in Table 2, all of the ester derivatives (10,
16, and 19–24) are either significantly weaker or inac-
tive (IC50 > 100 lM) in the NS5B RdRp assay, under-
lining the importance of a carboxylic acid group for
inhibitory activity.
O
Y
O
20
20a
OMe 53
OH 2.2
O
Y
N
H
O
Y
21
21a
OMe 8.6
OH 3.2
O
CONH2
N
H
All the compounds discussed above have symmetric
bis Ar groups on the quinoxaline ring. From an early
publication of inhibitor 1, it seems that compounds with
asymmetric substituents, especially the presence of a
cyclohexyl group, possess optimal activity.7 This postu-
lation prompted us to synthesize a focused set of
asymmetric quinoxalines with R1 and R2 being cyclo-
hexyl, 4-F-phenyl or phenyl. Different amino acids were
used to form the quinoxaline amide derivatives. As dem-
onstrated in Table 4, all of the compounds possessed
low micromolar potency with 26a being the most potent
(0.6 lM). In contrary to scaffold 1, it appears that cyclo-
hexyl in the R1 position yielded more potency than in
the R2 position. This is the trend for the pairs of com-
pounds 25a/25, 26a/26, 27a/27, and 32a/32. The rest
two pairs of compounds in Table 4 have very similar
IC50 values. Unlike in scaffold 1, a cyclohexyl group in
the quinoxaline scaffold does not significantly enhance
IC50 values.
O
22
22a
OEt >100
OH 17
Y
H
N
Y
23
23a
OEt 37
OH
0.69
O
O
H
N
Y
24
24a
OEt >100
OH
1.2
O
O

1666
F. Rong et al. / Bioorg. Med. Chem. Lett. 17 (2007) 1663–1666
NS5B, any extended groups stemming from the
coupling of –COOH of 2 with an amino-acid derivative
would extend into the surface of NS5B without any
twell-defined interactions. This determines the insensi-
tivity of the amino acid derivatives to the potency and
explains the flat SAR in compounds 10, 16a–24a
(Table 3), and 25a–32a (Table 4).
In summary, we identified an interesting substituted qui-
noxaline hit from high throughput screening using an
NS5B enzymatic assay. Based on the hit’s pharmaco-
phore and published similar inhibitors, many substitut-
ed quinoxaline amide derivatives were synthesized. A
good structure–activity relationship has been observed
for these compounds. One of the most potent com-
pounds has an IC50 of 0.6 lM in the enzymatic assay.
Further optimization would yield more potent com-
pounds and explore NS5B RdRp as a potential drug tar-
get for anti-HCV drug discovery.
Figure 1. Binding mode of 2 in the allosteric site of NS5B (PDB code:
2BRL) predicted by GLIDE docking.
Acknowledgments
To further understand SAR and binding mode of qui-
noxaline derivatives to HCV NS5B RdRp, we per-
formed a molecular modeling study using GLIDE
docking software.10 X-ray structure of a related analog
of 1 was published and available in PDB database
(PDB code: 2BRL).11 We assumed the hit 2 binds to
the same allosteric site as compound 1. The extreme
precision mode of GLIDE was employed for the dock-
ing, and up to 10 poses were saved for analysis. All of
the saved poses were similar and therefore, the top
scored pose is depicted in Figure 1.
The authors thank Drs. Stanley Lang, Nanhua Yao,
Haoyun An, and Weidong Zhong for helpful discussion
and suggestions, Dr. Xiaogang Han for 2-D and 13C
NMR analysis, and Dr. David Li for performing high
resolution mass analysis.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2006.12.103.
The resultant binding mode displays numerous interac-
tions between 2 and NS5B. These interactions are
remarkably comparable with those for the analog of 1
reported in the literature.11 Notably, the –COOH group
of 2 engages in the ionic-ionic type H-bonding interac-
tions with the basic guanidine side chain of Arg 503 of
NS5B. The 4-F-phenyl group at the position 2 of qui-
noxaline extends into a narrow hydrophobic pocket de-
fined by two a-helixes, and the same group specifically
makes hydrophobic contacts with the pocket defined
by the side chains of Leu 392, Ala 396, Ala 395, Thr
399, Leu 425, and Phe 429. The other 4-F-phenyl at
the 3-position of quinoxaline forms hydrophobic inter-
actions with side chains of Leu 392, Ala 393, and Leu
492. In addition, the quinoxaline ring makes hydropho-
bic interactions with Ile 424, Gly 493, Val 494, and Pro
495 in the binding site.
References and notes
1. Poynard, T.; Bedossa, P.; Opolon, P. Lancet 1997, 349,
825.
2. World Health Organization: Hepatitis C–Global Preva-
lence (update). Wkly Epidemiol. Rec. 2000, 75, 18.
3. Alberti, A.; Benvegnu, L. J. Hepatol. 2003, 38(Suppl. 1),
S104.
4. Choo, Q. L.; Kuo, G.; Weiner, A. J.; Overby, L. R.;
Bradley, D. W.; Houghton, M. Science 1989, 244, 359.
5. Ali, N.; Tardif, K. D.; Siddiqui, A. J. Virol. 2002, 76,
12001.
6. Behrens, S. E.; Tomei, L.; De Francesco, R. EMBO J.
1996, 15, 12.
7. Beaulieu, P. L.; Bos, M.; Bousquet, Y.; DeRoy, P.; Fazal,
G.; Gauthier, J.; Gillard, J.; Goulet, S.; McKercher, G.;
Poupart, M. A.; Valois, S.; Kukolj, G. Bioorg. Med.
Chem. Lett. 2004, 14, 967.
8. Nobuo, W.; Isao, M.; Kazuya, K.; Rina, S. Collect. Czech.
Chem. Commun. 1990, 22, 99.
9. Yan, S.; Appleby, T.; Larson, G.; Wu, J. Z.; Hamatake,
R.; Hong, Z.; Yao, N. Bioorg. Med. Chem. Lett. 2006, 16,
5888.
The predicted binding mode agrees well with the
observed SAR. For example, when –COOH group was
converted into –COOMe, the only ionic H-bond interac-
tion between an inhibitor and NS5B disappears upon
docking. This drastically reduced the inhibitory potency,
resulting in higher IC50 values for ester derivatives of
8–18 (Table 2). The hydrophobic nature of the Ar
binding sites in NS5B prefers a hydrophobic substituent,
and therefore, a hydrophilic group such as 2-pyridyl (7)
would be detrimental to the activity (Table 1). More
interestingly, according to the predicted pose of 2 in
10. Glide www.schrodinger.com, NY.
11. Di Marco, S.; Volpari, C.; Tomei, L.; Altamura, S.;
Harper, S.; Narjes, F.; Koch, U.; Rowley, M.; De
Francesco, R.; Migliaccio, G.; Garfi, A. J. Biol. Chem.
2005, 280, 29765.