Discovery of a novel HCV helicase inhibitor by a de novo drug design approach

Article abstract of DOI:10.1016/j.bmcl.2009.04.074

Herein we report a successful application of a computer-aided design approach to identify a novel HCV helicase inhibitor. A de novo drug design methodology was used to generate an initial set of structures that could potentially bind to a putative binding

Full text of DOI:10.1016/j.bmcl.2009.04.074

Bioorganic & Medicinal Chemistry Letters 19 (2009) 2935–2937
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a novel HCV helicase inhibitor by a de novo drug design approach
a
a
a
a
a
Sahar Kandil , Sonia Biondaro , Dimitrios Vlachakis , Anna-Claire Cummins , Antonio Coluccia ,
b
c
c
a,
*
Colin Berry , Pieter Leyssen , Johan Neyts , Andrea Brancale
The Welsh School of Pharmacy, Cardiff University, Cardiff, United Kingdom
Cardiff School of Biosciences, Cardiff University, Cardiff, United Kingdom
Rega Institute for Medical Research, KULeuven, Leuven, Belgium
a
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
Herein we report a successful application of a computer-aided design approach to identify a novel HCV
helicase inhibitor. A de novo drug design methodology was used to generate an initial set of structures
that could potentially bind to a putative binding site. Further structure refinement was carried out
through docking a series of focused virtual libraries. The most promising compound was synthesised
and it exhibited a submicromolar inhibition of the HCV helicase.
Received 9 March 2009
Revised 14 April 2009
Accepted 17 April 2009
Available online 22 April 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
HCV
Helicase inhibitor
De novo
Molecular modelling
HCV infection is the second most common chronic viral infec-
tion in the world with global prevalence. An estimated 180 million
people are chronically infected with this hepatitis C virus (HCV)
and thus at increased risk of developing serious life threatening li-
ver diseases including cirrhosis that may progress to hepatocellular
molecule; we focused therefore on the area around Arg393. This
residue is conserved in all HCV isolates and is also known to play
a key role in activity of the enzyme since mutation R393A has a
detrimental effect on the unwinding activity of the enzyme.11 Tar-
geting this specific amino acid represented a considerable chal-
lenge: after removing the nucleic acid from the structure it
became apparent that the area around Arg393 was completely ex-
posed to the solvent, making it more difficult for a small molecule
to bind tightly to it without being replaced by water or the nucleic
acid itself. However, on further inspection it was observed that the
sulfur atom of Cys431 (a residue not involved directly in the nu-
cleic acid binding and that is situated ꢀ10 Å away from Arg393)
had interacted to form an adduct with a molecule of mercap-
toethanol present in the crystallisation water. This observation
may suggest that Cys431 is accessible to a small molecule and that
it is sufficiently reactive to establish a covalent bond with an
appropriate compound. These observations led us to formulate
the hypothesis that a molecule that would (i) interact with
Arg393 and (ii) at the same time form a covalent bond with
Cys431 through a chemically reactive group may have the poten-
tial to inhibit helicase activity. Such a covalent binder would be
very difficult to displace by water or the incoming nucleic acid.
In the initial stages of inhibitor design, the de novo software
1
,2
carcinoma.
There is no prophylactic or therapeutic vaccine available against
HCV and the development of such a vaccine is not expected to oc-
cur soon. Current standard therapy, that is, the combination of
pegylated interferon with ribavirin is only effective in about 60%
3
of the patients and is associated with important side-effects. Sev-
eral drugs that target various stages of the replication cycle of HCV
are currently in preclinical or clinical development; these include
NS3 protease inhibitors, nucleoside and non-nucleoside polymer-
ase inhibitors and cyclophylin binding compounds.4
,5
So far the NS3 helicase has not been extensively explored as a
6–9
target for inhibition of HCV replication.
We devised a rational approach for the design of selective inhib-
itors of the HCV NS3 helicase. Several crystal structures of the HCV
NS3 have been reported, that is, as individual domains (protease
and helicase) as well as the full length protein. The structure re-
1
0
ported by Kim was co-crystallised with a strand of DNA bound
to the helicase domain (PDB 1A1V). It was decided to use this
structure as the starting point for the de novo design of novel po-
tential inhibitors that could compete for the nucleic acid binding
site. The first step was to define a potential binding site for a small
1
2
package LigBuilder was used. Programs of this type normally re-
quire the user to define an initial ‘seed’ in a binding site, after
which the computer builds a series of molecules by adding to the
growing structures the most suitable fragments taken from a given
1
3
library. The major drawback of this approach is that very often
the proposed structures are highly complex and not synthetically
*
Corresponding author. Tel.: +44 29 20874485.
E-mail address: brancalea@cf.ac.uk (A. Brancale).
0
960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.04.074

2
936
S. Kandil et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2935–2937
accessible. Indeed this is the problem we faced when we examined
the results obtained using the fragment library available in the
software (Fig. 1, Structure 1).
It is evident that the software attempted to fill the complete
space in the large pocket generating a series of structures with
complex chirality. To overcome this problem, the fragment library
was reduced to include only residues that, when combined, were
less likely to generate chiral centres. Also, a virtual wall of dummy
atoms (atoms with no physical property) was created to reduce the
site size and to force the software in designing a molecule that
would indeed be interacting with the desired residues. The com-
pounds thus generated were considerably simpler. Among them
structure 2 was able to interact with Arg393 and also with another
arginine located on the opposite side of the putative binding site,
namely Arg481. Furthermore it was located close to Cys431 and,
although it did not posses any functional group able to react with
the amino acid, this represented an appropriate basic scaffold for
further functionalization. At this point we employed a different ap-
proach to optimise our in silico design process. A series of virtual
Figure 2. Proposed binding of compound 4 in the HCV helicase.
It also presented a side chain that might increase the interaction
points with the enzyme. A molecular dynamic simulation was then
performed on the ligand/protein complex to evaluate whether the
14
libraries were generated, varying the linker between the two aro-
matic rings whilst replacing one of the carboxylic acid groups with
a Michael’s acceptor, which are known to react efficiently with nat-
ural thiols. These libraries were then docked1 into the binding site
and scored based on their ability to interact with the two arginines
and to place the vinyl ketone in close proximity to Cys431. Our best
structure (Fig. 1, Structure 3) presented the reactive position of the
Michael’s acceptor ꢀ5 Å away from the sulfur atom of the cysteine.
1
6
binding of this compound was stable. The 1 ns simulation re-
vealed a potentially relatively stable interaction; however a slow
drifting of the compound away from Cys431 (6–7 Å) was also ob-
served. This was probably due to the steric hindrance of the aro-
matic ring. Therefore the replacement of the benzene with a
smaller heterocycle was investigated. In addition, the linker side
chain did not provide any apparent advantage in the MD simula-
tion and was, therefore, removed in subsequent molecules. Finally,
a pyrrolo derivative with a slightly longer linker (Fig. 1, Structure
4) was identified as a potential inhibitor. Compound 4 presented
all the key interactions (Fig. 2): hydrogen bond between the ester
moiety and Arg393; hydrogen bond between the carbonyl group of
the vinyl ketone and Arg481; distance between the sulfur atom of
Cys431 and the reactive position of the Michael’s acceptor of <4 Å,
which remained stable in a MD simulation. This binding conforma-
tion is further stabilised by a hydrogen bond between the pyrrole
NH and the carbonyl group of Val432. Encouraged by these results,
compound 4 was synthesised and evaluated in the helicase assay.
The synthesis of this molecule is depicted in Scheme 1. Formy-
lation of methyl pyrrole-2-carboxylate was carried out using DMF
5
NH2
HO
OH
H
H
N
H
H
N
HO
S
HO
H
O
N
1
3
and POCl mixture to yield 2 isomers, the 5- and 4- formyl deriva-
HO
1
7
tives (6 and 7, respectively) in a 2:1 ratio. The two isomers were
separated and were used in an aldolic condensation reaction with
acetone in water and presence of pyrrolidine 30 mol % to afford 8
and 9, respectively. This was followed by saponification reaction
to yield the free carboxylic acid derivatives 10 and 11, respectively.
The later compounds were coupled with methyl 4-(aminomethyl)
O
O
2
OH
O
1
8
benzoate in the presence of 1-ethyl 3-(3-dimethyl aminopropyl)
OH
carbodiimide hydrochloride (EDCI) and dimethyl aminopyridine
HN
19
(
DMAP) according to the parallel synthesis protocol of Boger to
O
get compounds 4 and 12, respectively.
Compounds 4 and 12 were evaluated in a strand-displacement
enzymatic assay based on the method of Hicham against purified,
O
2
0
recombinant HCV helicase. Compound 4 showed an IC50 of
.26 M, while the regioisomer 12 did not show any activity at a
concentration as high as 100 M. This is consistent with the model
3
0
l
l
prediction, where the 2–4 substituted pyrrole does not dock with
the Michael acceptor placed in proximity to Cys431, thus making
the formation of the covalent bond less likely. To establish the
importance of the cysteine residue to the binding of 4 we have per-
formed the enzymatic assay in presence of the thioreactive agent
NEM (N-ethylmaleimide) and we have observed the loss of inhibi-
tory activity of the compound when added to the enzyme pre-incu-
bated with NEM. Not surprisingly, the enzyme preserved its
helicase activity in these conditions, probably because NEM, as
O
H
N
O
N
H
OMe
4
O
Figure 1. Evolution in the design of a novel HCV helicase inhibitor.

S. Kandil et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2935–2937
2937
H
N
pocket on the enzyme. These results are based on an in silico mod-
el. It is our aim and priority to obtain a crystal structure of the com-
plex to validate our approach. This in turn may allow us to
rationally optimise these compounds and to obtain a novel class
of potent anti-HCV inhibitors.
H
N
COOMe
H
N
COOMe
COOMe
H
i
ii
O
O
5
6,7
8,9
iii
Acknowledgements
H
N
COOH
COOMe
The authors are grateful to Dr. Klaus Klumpp and Roche for pro-
viding us the clone of the HCV helicase. This work is supported by
an FWO grant to Johan Neyts. The authors would also like to
acknowledge the Embassy of the Arab Republic of Egypt for the
award of a PhD scholarship to Sahar Kandil and the Istituto Pasteur
10,11
O
NH2
iv
-
Fondazione Cenci Bolognetti for supporting Antonio Coluccia.
O
H
N
References and notes
O
O
N
H
OMe
OMe
1. Brass, V.; Moradpour, D.; Blum, H. E. Int. J. Med. Sci. 2006, 3, 29.
2. Webster, D. P.; Klenerman, P.; Collier, J.; Jeffery, K. J. Lancet Infect. Dis. 2009, 9,
4
O
O
108.
3.
4.
5.
Hayashi, N.; Takehara, T. J. Gastroenterol. 2006, 41, 17.
Lam, A. M.; Frick, D. N. J. Virol. 2006, 80, 404.
Soriano, V.; Peters, M. G.; Zeuzem, S. Clin. Infect. Dis. 2009, 48, 313.
O
N
H
6. Diana, G. D.; Bailey, T. R.; Nitz, T. J. WO9736554, 1997.
7
8
.
.
Janetka, J. W.; Ledford, B. E.; Mullican, M. D. WO0024725, 2000.
Maga, G.; Gemma, S.; Fattorusso, C.; Locatelli, G. A.; Butini, S.; Persico, M.;
Kukreja, G.; Romano, M. P.; Chiasserini, L.; Savini, L.; Novellino, E.; Nacci, V.;
Spadari, S.; Campiani, G. Biochemistry 2005, 44, 9637.
NH
1
2
9.
Zhang, N.; Chen, H. M.; Koch, V.; Schmitz, H.; Liao, C. L.; Bretner, M.; Bhadti, S.
V.; Fattom, A.; Naso, R. B.; Hosmane, R. S.; Borowski, P. J. Med. Chem. 2003, 46,
4149.
Scheme 1. Reagents and conditions: (i) DMF, POCl
3
reflux 15 m; (ii) acetone,
pyrrolidine 30%, H O; (iii) KOH, MeOH reflux 15 h; (iv) EDCI, DMAP, rt, on.
2
10. Kim, J. L.; Morgenstern, K. A.; Griffith, J. P.; Dwyer, M. D.; Thomson, T. A.;
Murcko, M. A.; Lin, C.; Caron, P. R. Structure 1993, 6, 89.
1
1
1
1. Lam, A.; Keeney, D.; Frick, D. J. Biol. Chem. 2003, 278, 44514.
2. Wang, R.; Gao, Y.; Lai, L. J. Mol. Modell. 2000, 6, 498.
3. Schneider, G.; Fechner, U. Nat. Rev. Drug Discovery 2005, 4, 649.
mercaptoethanol, is not big enough to reach the nucleic acid bind-
ing site. To further highlight the role of the Michael acceptor in the
biological activity of compound 4, the mono-substitute pyrrolo
analogue with the vinyl ketone moiety missing was also prepared
directly from 5 using the same methodology described above and it
did not shown any inhibition of the HCV helicase in our assay. It
should also be noted that in the original crystal structure used,
two other cysteine residues beside Cys431 appeared to have re-
acted with mercaptoethanol (Cys279 and Cys499), but these are
placed on the enzyme surface well away from the nucleic acid
binding site (>20 Å), thus, making their involvement in the activity
of these compounds less probable. The active molecule was also
14. Molecular Operating Environment (MOE) was used for the preparation and the
visualization of the simulations. Chemical Computing Group, Inc. Montreal,
Quebec, Canada. http://www.chemcomp.com.
1
5. FlexX, BioSolveIT GmbH, Sankt Augustin, Germany. http://www.biosolveit.de.
16. Molecular dynamics simulations were performed using GROMACS 3.1
www.gromacs.org), using a triclinic water box and NPT conditions.
(
1
1
7. David, M.; Sam, H.; Senge, M. O.; Smith, K. M. J. Org. Chem. 1993, 53, 7245.
8. Goodyer, C.; Chinje, E.; Jaffar, M.; Stratford, I.; Threadgill, M. Bioorg. Med. Chem.
2003, 19, 4189.
1
2
2
9. Boger, D.; Dechantsreiter, M.; Ishii, T.; Fink, E.; Hedrick, M. J. Am. Chem. Soc.
2000, 122, 6382.
0. Hicham Alaoui-Ismaili, M.; Gervais, C.; Brunette, S.; Gouin, G.; Hamel, M.;
Rando, R. F.; Bedard, J. Antiviral Res. 2000, 46, 181.
1. Paeshuyse, J.; Vliegen, I.; Coelmont, L.; Leyssen, P.; Tabarrini, O.; Herdewijn, P.;
Mittendorfer, H.; Easmon, J.; Cecchetti, V.; Bartenschlager, R.; Puerstinger, G.;
Neyts, J. Antimicrob. Agents Chemother. 2008, 52, 3433.
evaluated for a potential inhibitory effect on HCV subgenomic rep-
licon replication as reported earlier.21 Compound 4 proved how-
ever rather cytostatic to the hepatoma cells (EC50
3
l
g/ml; CC50
22. General procedure for preparing 4 and 12: To a solution of the appropriate
pyrrole-2-carboxylic acid derivative (0.001 mol) in dry DCM, 1-ethyl 3-(3-
dimethylaminopropyl) carbodiimide hydrochloride (EDCl) (0.002 mol) and
dimethyl aminopyridine (DMAP) (0.002 mol) were added. Methyl 4-
10 lg/ml), which was expected, given the fact that the vinyl ketone
group is a known toxicophore. It is worth noting that the ester ana-
logues were prepared and evaluated for biological activity. It is
possible to speculate from the model that the free acid might have
a better interaction with Arg393. However, it should also be taken
into consideration that the increased polarity of the latter com-
pounds would reduce the cellular permeation, thus making them
less attractive for the development of future derivatives. Indeed,
the free acid analogue is also currently being prepared also with
the aim of obtaining a co-crystallised ligand/protein complex,
which might give us the definitive proof that compound 4 binds
to the helicase as predicted by the model.
(
aminomethyl)benzoate (0.001 mol) in dry THF was stirred for 3 min under
atmosphere then was added to the above mentioned DCM mixture. The
resulting mixture was stirred at room temperature for 48 h under N . The
N
2
2
resulting mixture was evaporated under reduced pressure. The crude product
was chromatographed using ethyl acetate as eluent to afford1 4, 12;
Characterization of compound 4. Off-white solid, mp 208–209 °C. H NMR
(
DMSO-d
6
): d 12.09 (br s, 1H), 8.90 (t, J = 6.05 Hz, 1H), 7.94 (d, J = 8.3 Hz, 2H),
7.44 (m, 3H), 6.93 (dd, J = 2.05, 3.7 Hz, 1H), 6.75 (d, J = 16.3 Hz, 1H), 6.69 (dd,
13
J = 2.2, 3.6 Hz, 1H), 4.54 (d, J = 5.95 Hz, 2H), 3.84 (s, 3H), 2.24 (s, 3H). C NMR
(DMSO-d ): d 27.17 (CH ). d 41.78 (CH ). d 52.05 (CH ). d 128.08, 129.79,
31.14, 145.36, 160.08, 166.08, 197.45 (7C, quaternary C). d 111.99, 113.76,
23.88, 127.34, 129.26, 132.79 (6C, CH). Anal. Calcd for C18 : C, 66.25;
6
3
2
3
1
1
18 2 4
H N O
H, 5.56; N, 8.58. Found: C, 66.01; H, 5.68; N, 8.33.
In conclusion, we have used a de novo approach to design a no-
vel HCV helicase inhibitor22 that could target a newly identified