A refined pharmacophore model for HIV-1 integrase inhibitors: Optimization of potency in the 1H-benzylindole series

Article abstract of DOI:10.1016/j.bmcl.2008.03.089

We report herein the development of a new three-dimensional pharmacophore model for HIV-1 integrase inhibitors which led to the discovery of some 4-[1-(4-fluorobenzyl)-1H-indol-3-yl]-2-hydroxy-4-oxobut-2-enoic acids that are able to specifically inhibit the strand transfer step of integration at nanomolar concentration. The synthesis of the new designed molecules is also described.

Full text of DOI:10.1016/j.bmcl.2008.03.089

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 2891–2895
A refined pharmacophore model for HIV-1 integrase inhibitors:
Optimization of potency in the 1H-benzylindole series
Laura De Luca,a Maria Letizia Barreca,b Stefania Ferro,a Nunzio Iraci,a
Martine Michiels,c Frauke Christ,c Zeger Debyser,c Myriam Witvrouwc
and Alba Chimirria,*
a
`
Dipartimento Farmaco-Chimico, Universita di Messina, Viale Annunziata, 98168 Messina, Italy
bDipartimento di Chimica e Tecnologia del Farmaco, Via del Liceo 1, 06123 Perugia, Italy
cMolecular Medicine, Katholieke Universiteit Leuven and IRC KULAK, Kapucijnenvoer 33, B-3000 Leuven, Flanders, Belgium
Received 22 January 2008; accepted 29 March 2008
Available online 8 April 2008
Abstract—We report herein the development of a new three-dimensional pharmacophore model for HIV-1 integrase inhibitors
which led to the discovery of some 4-[1-(4-fluorobenzyl)-1H-indol-3-yl]-2-hydroxy-4-oxobut-2-enoic acids that are able to specifi-
cally inhibit the strand transfer step of integration at nanomolar concentration. The synthesis of the new designed molecules is also
described.
Ó 2008 Elsevier Ltd. All rights reserved.
Integrase (IN), one of the three enzymes expressed by
HIV, catalyses the insertion of viral DNA into the gen-
ome of the host cell through a multistep process. In the
‘30-processing’ step, IN removes a dinucleotide from
each 30-end of viral cDNA, while in the ‘strand transfer’
reaction, the two newly processed 30-viral DNA ends are
inserted into the host cell DNA.1 Moreover, it has been
suggested that the integration mechanism involves diva-
lent metal ions (Mg2+ or Mn2+) in the enzyme catalytic
site. The HIV-1 IN crystal structures available to date
DKA IN inhibitors characterized by the presence of a
1H-benzylindole skeleton.6 The biological results
showed that all derivatives selectively inhibited the ST
step and highlighted the validity of our molecular mod-
elling approach. Recent clinical trials have shown two
new INSTIs, 1 (GS-9137) and 2 (MK-0518) (Chart 1),
to be highly effective, and both compounds appeared
well tolerated with respect to toxicity compared with
placebo controls.7 More recently, the FDA approved
MK-0518 (raltegravir) as the first drug (Isentress) that
show only a single binding site for Mg2+ or Mn2+
,
is active against IN enzyme.
8
whereas both biochemical and structural studies offer a
plausible two-metal model for the catalytic centre of
IN.2 Integrase strand transfer inhibitors (INSTIs) repre-
sent the major leads in the development of anti-HIV-1
IN drugs.3 These molecules act by sequestering the me-
tal ions bound in the active site of the enzyme.2,4 In par-
ticular, b-diketo acids3 (DKAs) and their derivatives5
were the first INSTIs to display potent antiviral activity
via inhibition of IN activity. We previously generated a
three-dimensional (3D) pharmacophore model for
DKA-like derivatives acting as INSTIs, which enabled
us to rationally design and synthesize new potential
The aim of this work was to create a new pharmaco-
phore model using structural and biological data of
compounds 1 and 2 in order to explain their putative
binding requirements and to use the information
obtained to optimize our INIs.
Our previous 3D pharmacophore model for IN inhibitor
DKA-like derivatives, developed using the Catalyst
HypoGen module, contained four features: one hydro-
phobic aromatic region (Y0), two hydrogen-bond accep-
tors (A10–A20), and one hydrogen-bond donor (D0).6
The superimposition of the new compounds 1 and 2
on our HypoGen pharmacophore model showed that
these compounds fitted with all the chemical features
of our hypothesis, thus confirming the reliability of the
developed model (Fig. 1).
Keywords: Integrase; HIV; DKAs; Pharmacophore model; Synthesis of
indole derivatives.
*
Corresponding author. E-mail: chimirri@pharma.unime.it
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.03.089

2892
L. De Luca et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2891–2895
O
OH
OH
O
N
H
N
N
O
N
O
O
F
N
N
O
O
HN
Cl
F
OH
2 (MK-0518)
1 (GS 9137)
H
N
F
O
N
OH
N
O
O
OH
OH
O
3 (S-1360)
4
Chart 1. Chemical structures of integrase strand transfer inhibitors.
Ten hypotheses, with a ranking score ranging from
37.6397 to 35.8411, were developed using the algorithm
Catalyst/HipHop.10 The results are presented in Table 1.
The first three ranked pharmacophore hypotheses
(Hypo1–3) had the same ranking score and contained
the same chemical features in a similar spatial arrange-
ment: four hydrogen-bond acceptors (A1–A4), two
hydrophobic aliphatic regions (Z1 and Z2) and one aro-
matic feature (Y) (Fig. 3) highlighting important struc-
tural similarities between compounds 1 and 2. Since
Hypo1–3 had therefore identical statistical relevance
and composition, Hypo1 was selected for further studies
(see Fig. 3).
Figure 1. INSTIs 1 (A) and 2 (B) mapped into HypoGen pharmaco-
phore model (A01–A02, hydrogen-bond acceptor, green; Y0, hydro-
phobic aromatic, cyan; D0, hydrogen-bond donor, magenta).
The superimposition of this automated common-fea-
tures pharmacophore model and our previously devel-
oped hypothesis showed the same three-dimensional
orientation of the four features Y and A1–A3 compared
with features Y0 and A01–D0–A02 (Fig. S1 in supplemen-
tary data). In particular, both features A1–A3 and A01–
D0–A02 seem able to coordinate the two magnesium ions
required for IN activity. The other three features (A4,
However, the superimposition of compounds 1 and 2
with respect to other first-generation INSTIs such as 3
(S-1360),9 and 42 (Chart 1) highlighted some additional
structural characteristics which were not explored by
our pharmacophore (Fig. 2) implying an explanation
for their higher activity.
On this basis we used derivatives 1 and 2 as templates to
generate a new 3D pharmacophore model.
Table 1. Summary of the common feature hypothesis run
Numbera
Compositionb
Ranking scorec
Hypo1
Hypo2
Hypo3
Hypo4
Hypo5
Hypo6
Hypo7
Hypo8
Hypo9
Hypo10
YZZAAAA
YZZAAAA
YZZAAAA
ZZHAAAA
YZHAAAA
YZHAAAA
YZHAAAA
YZHAAAA
ZZHAAAA
ZZHAAAA
37.6397
37.6397
37.4411
36.0397
36.0397
36.0397
36.0397
36.0397
36.0397
35.8411
a Numbers for the hypothesis are consistent with the numeration
obtained by the hypothesis generation.
b Y, hydrophobic aromatic; Z, hydrophobic aliphatic; A, hydrogen
bond acceptor; H, hydrophobic group.
c The higher the ranking score, the less likely it is that the molecules in
the TS fit the hypothesis by a chance correlation. Best hypotheses
have the highest ranks.
Figure 2. HypoGen pharmacophore model aligned to compounds 1
(green), 2 (blue), 3 (yellow) and 4 (magenta) (A01–A02, hydrogen-bond
acceptor, green; Y0, hydrophobic aromatic, cyan; D0, hydrogen-bond
donor, magenta). Part not explored by pharmacophore is circled.

L. De Luca et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2891–2895
2893
Figure 3. The top scoring HipHop pharmacophore Hypo1 is mapped to IN inhibitors 1 (A), 2 (B) 31 (C) (A1–A4, hydrogen-bond acceptor, green;
Z1–Z2, hydrophobic aliphatic, blue; Y, hydrophobic aromatic, cyan).
Z1 and Z2) of the new model represent a novelty for
DKA derivatives and could explain the higher anti-IN
activity of compounds 1 and 2 which could be particu-
larly useful for the optimization of already known INS-
TIs. In order to confirm the reliability of this
pharmacophore model, we used some INSTIs reported
by Sato et al., in which some structural modifications
allowed the transformation of quinolone antibiotics into
potent IN inhibitors such as compound 1.11 Indeed, on
matching these compounds in the HipHop model, a
good correlation was observed between activity data
and BestFit values (supplementary data, Table S1).
These observations were interpreted by us as an indica-
tion that substituents able to occupy the new features
could be exploited to modify the intrinsic potency of
our 1H-benzylindole derivatives. This idea was sup-
ported by biological results obtained for some of our
very potent INSTIs.
tem, proved to be more potent (IC50 against
ST = 0.27 lM) than its demethoxy analogue, 29 (IC50
against ST = 2.33 lM).
Interestingly, the overlapping of compound 31 onto our
new hypothesis (Fig. 3C) showed that the methoxy
group in position 5 was able to fit the hydrophobic
sphere Z1 thus explaining its major ST activity with
respect to the demethoxy derivative.
Taking into account these findings, we focused our
efforts on exploring the importance of the two hydro-
phobic areas (Z1 and Z2) by inserting methoxy groups
in different positions of the benzene-fused ring.
Following our modelling studies, the designed molecules
were synthesized and tested.
The synthesis of the new benzyl-1H-indole derivatives
24, 26–28, 30 and 32–34 was achieved in a multistep
reaction, as previously reported by us.6,12
In fact, compound 31 (Scheme 1), which is characterized
by the presence of a methoxy group on the indole sys-
F
H
N
H
N
N
AcCl
ArCH2Br
RR'
RR'
RR'
Me
Me
O
O
5-10
11-16
17-22
diethyl
oxalate
F
F
N
NaOH
COOH
N
O
RR'
RR'
5, 11, 17, 23, 29 R=R'=H
6, 12, 18, 24, 30 R=4OMe, R'=H
7, 13, 19, 25, 31 R=5OMe, R'=H
8, 14, 20, 26, 32 R=6OMe, R'=H
9, 15, 21, 27, 33 R=7OMe, R'=H
COOEt
O
HO
HO
29-34
23-28
10, 16, 22, 28, 34 R=4OMe, R'=7OMe
Scheme 1.

2894
L. De Luca et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2891–2895
Table 2. Inhibition of HIV-1 integrase enzymatic activity, replication of HIV-1 (IIIB) and cytotoxicity in MT-4 cells
Compound
IN enzymatic activity
ST
Activity in MT-4 cells
Over-all
IC50 (lM)a
HIV-1
EC50 (lM)c
Cytotoxicity
CC50 (lM)d
SIe
IC50 (lM)b
24
25
86.08 12.58
4.60 1.00
1.96 1.31
4.76 2.74
1.24 0.06
0.86 0.18
0.08 0.003
0.30 0.05
0.19 0.05
0.27 0.19
0.06 0.01
0.009 0.0002
0.004 0.003
92.30 25.16
7.55 0.05
2.34 2.14
2.64 1.86
0.66 0.01
2.33 0.83
0.14 0.11
0.27 0.05
0.22 0.21
0.19 0.14
0.03 0.01
0.007 0.0005
0.015 0.002
3.17 1.71
11.5 8.7
>100
74.0 51.3
126.0 37.5
100
24
11
<1
1
26
27
>100
>100
81.1
28
29
4.09 1.29
8.35 0.53
0.59 0.38
5.63 1.87
8.39 2.60
7.04 2.17
2.13 0.05
0.013 0.0005
0.0008 0.00009
20
8
64.06 9.74
41.1 16.7
63.6 43.3
66.0 14.8
82.9 76.1
213.3 50.3
>18
30
31
70
12
8
32
33
12
34
MK-518
GS9137
100
>1387
>2650
2.12 0.36
All data represent average results SD.
a Concentration required to inhibit the in vitro overall integrase activity by 50%.
b Concentration required to inhibit the in vitro strand transfer step by 50%.
c Effective concentration required to reduce HIV-1-induced cytopathic effect by 50% in MT-4 cells.
d Cytotoxic concentration to reduce MT-4 cell viability by 50%.
e Selectivity index: ratio CC50/EC50
.
The appropriate 1H-indole was first 3-acetylated by the
reaction with acetyl chloride and a slight excess of diet-
hylaluminium chloride. With microwave assisted synthe-
sis the reactions were carried out in only 10 min, instead
of 2 h using the conventional method.
oxy-substituted DKAs exhibited higher potency than
the corresponding unsubstituted derivative 29. In partic-
ular, derivative 30 characterized by the presence of one
methoxy group at position 4 proved to be the most
potent compound in the enzyme assays at nanomolar
concentration (Table 2). Interestingly, our idea to in-
crease the IN binding affinity by occupying both the
two hydrophobic areas Z1 and Z2 was confirmed by
the biological results of compounds 28 and 34 in which
we held the 4-methoxy group constant and introduced
an additional methoxy moiety at the 7-position of the
indole nucleus, obtaining disubstituted derivatives. As
expected, the 4,7-dimethoxysubstituted DKA, 34,
showed a best-fit value which was better than monosub-
stituted analogues (30) (supplementary data, Table S2).
The obtained derivatives were then N-alkylated by treat-
ment with 4-fluorobenzyl bromide and a small amount
of sodium hydride to give intermediates 17–22.
Under the ‘classical’ method, reactions were performed
in 30 min; whereas applying the microwave irradiation
to this step of our synthetic strategy took only 5 min,
thus confirming that focused microwave irradiation is a
highly effective technique for accelerating organic reac-
tions, even at low temperatures. The coupling with
diethyl oxalate in the presence of a catalytic amount of
sodium methoxide and using microwave irradiation, eas-
ily provided derivatives 23–28. The final aryl hydroxyke-
toacids (29–34) could be readily produced from the
hydrolysis of the corresponding arylhydroxyketoesters
(23–28) in basic medium (Scheme 1). Both analytical
and spectral data of all synthesized compounds are in full
agreement with the proposed structures. All synthesized
compounds were tested for their inhibitory effect on IN
enzymatic activity and against HIV-1 replication.
This behaviour could be explained by considering
that derivative 34 (Fig. 4) was able to occupy both
hydrophobic areas Z1 and Z2. In agreement with the
To determine the susceptibility of the HIV-1 integrase
enzyme towards synthesized compounds we used en-
zyme-linked immunosorbent assays. However, their
inhibitory effect on the HIV-induced cytopathic effect
(CPE) in human lymphocyte MT-4 cell culture was
determined by the MT-4/MTT-assay13 (supplementary
data for biological assay methods).
The biological results showed that the DKAs (30–34)
were more active than the corresponding esters (24–28)
in both assays (Table 2). Moreover, all mono meth-
Figure 4. Alignment of compound 34 into pharmacophore Hypo1 by
HipHop (A1–A4, hydrogen-bond acceptor, green; Z1–Z2, hydropho-
bic aliphatic, blue; Y, hydrophobic aromatic, cyan).

L. De Luca et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2891–2895
2895
pharmacophore mapping values, the introduction of the
References and notes
second methoxy group provided a potent nanomolar
INSTI (34) and improved the potency of the most active
monosubstituted derivative 30.
1. Pommier, Y.; Johnson, A. A.; Marchand, C. Nat. Rev.
Drug Disc. 2005, 4, 236.
2. Grobler, J. A.; Stillmock, K.; Hu, B.; Witmer, M.;
Felock, P.; Espeseth, A. S.; Wolfe, A.; Egbertson, M.;
Bourgeois, M.; Melamed, J.; Wai, J. S.; Young, S.;
Vacca, J.; Hazuda, D. J. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 6661.
3. Hazuda, D. J.; Felock, P.; Witmer, M.; Wolfe, A.;
Stillmock, K.; Grobler, J. A.; Espeseth, A.; Gabryelski,
L.; Schleif, W.; Blau, C.; Miller, M. D. Science 2000, 287,
646.
4. Marchand, C.; Johnson, A. A.; Karki, R. G.; Pais, G. C.;
Zhang, X.; Cowansage, K.; Patel, T. A.; Nicklaus, M. C.;
Burke, T. R., Jr.; Pommier, Y. Mol. Pharmacol. 2003, 64,
600.
5. Johnson, A. A. M. C. A. A.; Marchand, C.; Pommier, Y.
Curr. Top. Med. Chem. 2004, 4, 1059.
Finally, when the synthesized compounds were tested
against HIV-1 replication in cell cultures all DKAs
showed an EC50 lower than 10 lM and proved to be bet-
ter than corresponding esters.
In summary we have developed a new pharmacophore
model and identified a series of novel and potent IN
inhibitors based on the benzylindole template.
These results suggest that this new model could be used
to improve the potency of known INSTIs by suitable
modification. More detailed SAR examination and the
exploration of other features suggested by the new phar-
macophore model are still ongoing and will be reported
in due course.
6. Barreca, M. L.; Ferro, S.; Rao, A.; De Luca, L.; Zappala,
M.; Monforte, A. M.; Debyser, Z.; Witvrouw, M.;
Chimirri, A. J. Med. Chem. 2005, 48, 7084.
7. Ramacharan, J.; Skalka, A. M. Future Med. 2006, 1,
717.
8. www.fda.gov/bbs/topics/NEWS/2007/NEW01726.html.
9. Billich, A. Curr. Opin. Investig. Drugs 2003, 4, 206.
10. Catalyst, v. A. I. S. D., CA, 2006.
11. Sato, M.; Motomura, T.; Aramaki, H.; Matsuda, T.;
Yamashita, M.; Ito, Y.; Kawakami, H.; Matsuzaki, Y.;
Watanabe, W.; Yamataka, K.; Ikeda, S.; Kodama, E.;
Matsuoka, M.; Shinkai, H. J. Med. Chem. 2006, 49,
1506.
Acknowledgments
We gratefully acknowledge expert technical assistance at
K.U. Leuven Campus Kortrijk by Linda Desender. This
work was supported by the European Commission
(LSHB-CT-2003-503480) (TRIoH project).
12. Ferro, S.; Barreca, M. L.; De Luca, L.; Rao, A.;
Monforte, A. M.; Debyser, Z.; Witvrouw, M.; Chimirri,
A. Arch. Pharm. (Weinheim) 2007, 340, 292.
13. Pauwels, R.; Balzarini, J.; Baba, M.; Snoeck, R.; Schols,
D.; Herdewijn, P.; Desmyter, J.; De Clercq, E. J. Virol.
Methods 1988, 20, 309.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.bmcl.
2008.03.089.