Synthesis, anti-HIV-1 activity, and modeling studies of N-3 Boc TSAO compound

Article abstract of DOI:10.1016/j.bmcl.2008.03.010

The synthesis and the biological evaluation of the anti-HIV-1 activity of TSAO-Boc³T (8) are described. The computational analysis showed that the N-3 Boc group promotes new interactions in the binding site of the enzyme leading to a good inhib

Full text of DOI:10.1016/j.bmcl.2008.03.010

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 2277–2281
Synthesis, anti-HIV-1 activity, and modeling studies
of N-3 Boc TSAO compound
a
a
b
c
´
Cyrille Tomassi, Albert Nguyen Van Nhien, Jose Marco-Contelles, Jan Balzarini,
Christophe Pannecouque,c Erik De Clercq,c Elena Sorianod and Denis Postela,*
aLaboratoire des Glucides (UMR6219), Universite´ de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens, France
bLaboratorio de Radicales Libres (IQOG, CSIC), C/Juan de la Cierva, 3, 28006 Madrid, Spain
cRega Institute for Medical Research, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium
dLaboratorio de Resonancia Magnetica (IIB, CSIC), C/Arturo Duperier 4, 28029 Madrid, Spain
Received 4 February 2008; revised 3 March 2008; accepted 3 March 2008
Available online 7 March 2008
Abstract—The synthesis and the biological evaluation of the anti-HIV-1 activity of TSAO-Boc3T (8) are described. The computa-
tional analysis showed that the N-3 Boc group promotes new interactions in the binding site of the enzyme leading to a good inhib-
itory activity.
Ó 2008 Elsevier Ltd. All rights reserved.
Non-nucleoside reverse transcriptase inhibitors (NNR-
TIs) represent a particular group of compounds which
bind to a hydrophobic pocket near to, but not at the
HIV-1 polymerase active site of p66, resulting in an
inactive conformation of the enzyme.1–3 TSAO com-
pounds represent a peculiar family of HIV-1 RT inhib-
itors since they are the first non-peptide molecules that
interact with amino acids at both HIV-1 RT subunits
(p66 and p51), at the dimer interface, and thus can inter-
fere with enzyme dimerization.4 Since 1992, TSAO
compounds have been subjected to extensive structure–
activity studies (SAR) to assess the structural require-
ments for their optimal interaction with HIV-1 RT, a
prerequisite for antiviral activity. The ribo configured
sugar plays an essential role in the interaction as well
as the presence of both a 30-spiro-500-(400-amino-100,200-
oxathiole- 200,200-dioxide) moiety and OTBDMS groups
at positions 20 and 50 of the sugar moiety. Modifications
either at positions 20 or 50,5 on the 30-spiro moiety,6 or
on the base part,7 have been performed. Like for NRTIs
and NNRTIs, TSAOs provoke resistance of HIV-1 RT
due to the Glu-138 ! Lys mutation, suggesting that
they may interact with the Glu-138 residue located in
the b7–b8 loop in p51.8 The phenomenon of resistance
and eagerness for complete elucidation of the precise
binding mode of this drug class to this RT has prompted
the exploration of further structural modifications of
TSAO analogs for SAR study.
Camarasa et al. reported the synthesis and biological eval-
uation of several compounds bearing a variety of polar,
lipophilic, or aromatic groups linked through flexible
polymethylene linkers. The TSAO derivative with an
N-methylcarboxamide group at the N-3 position was
found to be 5- to 6-fold more active than the TSAO-T.9
Recently, we reported the first synthesis of ATSAOs in
which the 30-spiro-oxathiazole ring is substituted with
a spiro-isothiazole moiety (Fig. 1).10a The evaluation
O
O
4
4
3
3
5
5
R
R
N
N
6
6
2
2
5'
5'
N1
N1
TBDMSO
H2N
TBDMSO
H2N
O
O
4'
O
4'
O
1'
1'
3' 2'
3' 2'
4"
4"
5"
R2
3"
N 2"
R1
OTBDMS
OTBDMS
O1''
O
S
S
O
O
1'' O
2"
ATSAO-T derivatives
R= H, Boc, CH3
R1= H, CH3
TSAO-T
, R= H
, R= CH3
TSAO-m3T
Keywords: NNRTI; Aminonitrile; CSIC reaction; TSAO; ATSAO;
Computational study.
R2= H, Ph
*
Corresponding author. Tel.: +33 22 82 75 70; fax: +33 22 82 75 68;
e-mail: denis.postel@sc.u-picardie.fr
Figure 1. Aza analogs of TSAO.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.03.010

2278
C. Tomassi et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2277–2281
of the inhibitory effects on HIV-1 replication demon-
strated that these ATSAO derivatives showed the same
HIV-1 specific activity spectrum (EC50: 0.13–0.53 lM)
but significantly, showed no inhibitory activity against
HIV-2 at subtoxic concentrations. A computational
study revealed that the ATSAO compounds adopt a
similar conformation at the p66–p51 interface as that
of TSAO analogs.11
tion of a Boc group at the N-3 position improves the
antiviral activity over that of the lead TSAO-T com-
pound (EC50 = 0.06 0.010 lM). We hypothesized that
this might probably be due to the establishment of addi-
tional interactions with the dimer interface near the b7–
b8 loop. Moreover, N-3 carbamoylation decreased the
cytotoxic concentration compared with TSAO-T
(CC50 = 26.2 1.12 lM vs 16 1.0 lM for TSAO-T).
No antiviral activity was observed against HIV-2. This
result is in accordance with our preliminary results ob-
tained in the ATSAO series.
AzaTSAO-T derivatives bearing a substituted dihydro-
isothiazole dioxide ring with a phenyl group at 500 posi-
tion have also been reported. Their biological evaluation
revealed that the phenyl group leads to a dramatic de-
crease in the inhibitory effect. Continuing our work on
this area, and in order to gain a deeper insight into the
SAR of (A)TSAO derivatives, herein we report the syn-
thesis of TSAO-Boc3T (8) as well as its biological evalu-
ation and modeling study of the most active TSAO and
ATSAO compounds (with a Boc functionality at N-3).
In order to further elucidate the binding mode induced
by this novel pharmacophore, a modeling study was
undertaken. A recent computational study focused on
the role of the heteroatom at the spirocycle revealed that
its endocyclic moiety is mostly exposed to the solvent,
and it was argued that additional electrostatic stabiliza-
tion by the polar solvent might stabilize the complex
formed with the oxathiole or unmethylated aza deriva-
tives.11 It was postulated, on the other hand, that the
presence of a substituent on the endocyclic nitrogen
atom constrains the ring pucker to a narrow region of
the pseudorotational cycle. This leads to a fairly dis-
torted disposition of the base and substituents of fura-
nose, mainly the 50-O-TBDMS group. Because of the
well-known rigorous geometric requirements for the
functional groups at C-50,5a,12 we speculated that this
enhanced conformational rigidity may impede the com-
pound’s efficient adaptation to the enzyme binding site,
leading to a loss of affinity.
In the previous work,10a we demonstrated that ATSAO-
Boc3T, which has an unsubstituted isothiazolic ring,
proved to be only 2- to 7-fold less active than TSAO-T
against HIV-1 replication in MT-4 and CEM cells. The
inhibitory effects for ATSAO-Boc3T and the unsubstitut-
ed N-3 analog (EC50: 0.13 vs 0.53 lM) suggested that it
was the Boc group at the N-3 position that increases the
activity. In spite of the large number of compounds
tested,4 it is surprising that no TSAO analog equivalent
to our ATSAO-Boc3T has been synthesized so far.
In order to establish the role of a Boc group at N-3 on
the biological activity and to further elaborate SAR
studies, here we describe the synthesis, modeling studies,
and the anti-HIV-1 activity of compound TSAO-Boc3T
(8).
According to the proposed model, the N-3 substituent
would be mostly exposed to the solvent and run parallel
to the subunit interface. Thus additional interactions with
other interface residues may help destabilize the RT dimer
by disrupting key p51/p66 interface interactions.8 In fact,
it has been demonstrated that modifications at the N-3 po-
sition are well tolerated,13 and in some cases are even
better than the prototype TSAO-T at disrupting the in-
ter-subunit interactions. Thus, they are more potent
inhibitors of the enzyme’s DNA polymerase activity,
and also exhibit potent antiviral activity.14
Following the synthetic pathway described by Camar-
asa et al., carbamoylation of the TSAO-T was unsuc-
cessful; consequently, the Boc group was introduced
before the Carbanion mediated Sulfonate Intramolecular
Cyclization (CSIC)-reaction step.10b,c Accordingly, com-
pound 4, obtained from precursor 1, was reacted under
the usual conditions to give the Boc-derivative 5, which
was submitted to the CSIC protocol to provide product
6 in 60% yield. Subsequent deprotection (NH3/MeOH)
and silylation afforded the target compound 8 in 50%
yield.
We have now conducted a molecular modeling analysis
(see Supplementary Data) on the effect of a Boc substi-
tuent on N-3 to identify the structural features that are
important for the improved ability, relative to the lead
TSAO compound, to inhibit HIV-1 RT dimerization.
The inhibitory activity of compound 8 against HIV-1
(IIIB) and HIV-2 (ROD) was evaluated in CEM cell
cultures. In a previous work,10a we demonstrated that
the ATSAO derivatives lacking a substituent (methyl)
on the endocyclic nitrogen atom of the spiro moiety
were potent and selective inhibitors of HIV-1 in
CEM cell cultures, as none of them showed inhibitory
activity against HIV-2 replication, at subtoxic
concentrations.
Docking studies15 (see Supplementary Data) were per-
formed on the proposed binding site using the com-
pounds shown in Figure 2. The number of clusters
found for each compound, the proportion of highly
populated clusters, and the corresponding binding and
docking energies of each complex were calculated (see
Supplementary Data) (Scheme 1).
Our data indicate that the Boc-derivatives can pose in
the presumed binding site mainly in two different modes,
each differing in orientation of the Boc group at N-3:
complexes a and b. Figure 3 shows both conformations
for compound 8. As can be seen, the Boc carbonyl oxy-
However, for compound 8, which has an oxathiole spiro
ring, the concentration required to protect CEM cells
against the cytopathogenicity of HIV by 50% (EC50:
0.023 0.010 lM) clearly demonstrates that introduc-

C. Tomassi et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2277–2281
2279
O
We have monitored the root-mean square deviation
(rmsd) profiles for the moving part of the system, and
the ligand to get insights into the changes that are expe-
rienced upon binding. For both complexes a and b, the
ligand remained docked in the binding site, conserving
the initial conformation, but the backbone torsion
about C40–C50, evolving to the expected ap conforma-
tion8 upon relaxing. In addition, for complex a, the li-
O
R
R
N
N
N
N
TBDMSO
H2N
O
TBDMSO
H2N
O
O
O
OTBDMS
OTBDMS
O
NH
S
S
O
O
O
O
˚
A: TSAO-T: R=H
B: ATSAO: R= H
C: ATSAO-Boc3T: R= Boc
gand showed an rmsd > 1.7 A due to additional
conformational adjustments on the Boc moiety, leading
8: TSAO-Boc3T : R= Boc
to an average value of rmsd higher than for complex b
˚
Figure 2.
(ꢀ1.2 A). The rmsd profiles determined for the subset
of residues that form the mobile region remained stable
˚
˚
along the simulations at ꢀ1.8 A (complex a) and ꢀ1.6 A
(complex b), thus suggesting the occurrence of soft local
distortions in the protein as a consequence of the bind-
ing of the drug. The main rearrangements are induced
by the tert-butyl moiety, since it forces the side chains
of residues on the palm subdomain of p61 (Thr-A165
and Glu-A169) away. This effect is enhanced in complex
a, where initially there were more intense steric contacts.
BzO
O
BzO
NC
BzO
NC
O
O
O
O
O
OAc
OAc
b
a
O
O
OMs
O
H3C
O
S
O
3
1
2
O
O
N
H
Boc
N
N
N
BzO
NC
BzO
NC
O
O
O
d
O
The analysis of the intermolecular hydrogen bonds
formed upon complexation has revealed significant
properties. The sulfone group hydrogen bonds to one
of the anticipated lysines (Lys-A101) throughout the tra-
jectory (occupied 66%), with a mean distance separating
the Nf atom of the residue and the closest sulfone oxy-
e
c
OAc
OAc
O
O
H3C
O
H3C
O
S
S
O
O
5
4
O
O
˚
gen of 2.9 0.16 A. The distance for Lys-A103
˚
(3.6 0.11 A) and the occupancy (18%) of the pertinent
Boc
Boc
N
N
N
N
R1O
H2N
O
H-bond are far from a true hydrogen bond during most
of the simulation, although this may be considered as an
electrostatic interaction which allows an optimal accom-
modation of the sulfone group.11 The spiro amino group
establishes two stable hydrogen bonds with carboxylate
oxygen atoms of Glu-B138, as can be deduced from the
TBDMSO
H2N
O
O
O
g
OR2
OTBDMS
O
O
O
S
S
O
O
O
6 R1=Bz ; R2=Ac
8
f
7 R1=R2= H
˚
distances (2.8 0.15 and 3.0 0.18 A, respectively) and
occupancy along the simulation (78% and 59%). These
values are very similar for both complexes a and b, with
absolute differences on occupancy and distance <3% and
Scheme 1. Reagents and condition: (a) i—NaCN, NaHCO3, Et2O–
H2O; ii—MsCl, py (78%); (b) i—TFA–H2O; ii—Ac2O, py (95%); (c)
silylated thymine, TMSOTf (77%); (d) (Boc)2O, CH2Cl2/py (4/1; v/v)
(85%); (e) Cs2CO3, CH3CN (60%); (f) NH3/MeOH (70%); (g)
TBDMSCl, CH3CN (50%).
˚
<0.1 A, respectively.
The hydrogen bond formed by carbonyl oxygen on Boc
shows a contrasting behavior. For complex a, the hydro-
gen bond between Thr-B139 and the carbonyl was pro-
gressively weakened during the MD simulation, owing
to the tert-butyl group of the Boc substituent, which ham-
pers an effective H-bond with the secondary hydroxy
group at the amino acid. This results in a long average dis-
tance (3.5 0.68) and low occupancy (29%) of the
H-bond, pointing to a weak interaction. For the complex b,
the carbonyl oxygen is engaged in a H-bond with Lys-
A172 all over the simulation (2.7 0.19; occupancy
87%). The amino acid side chain can freely rotate since
the terminal amino group is not initially hydrogen-
bonded to other residues, and the new H-bond is sterically
unhindered, so it appears stable along the simulation.
gen may hydrogen bond to the hydroxy group of the
side chain of Thr-B139 (complexes of type a) or with
Lys-A172 (complexes of type b). Such stabilizing inter-
actions might be responsible for the improved activity
seen for the Boc-derivatives relative to unsubstituted
ones. At first sight, the tert-butyl group seems to be
unhampered, well accommodated in both conforma-
tions, which provides an ambiguous identity of the rele-
vant amino acid involved in this stabilizing interaction.
To take into account protein flexibility, the behavior of
the predicted complexes was studied in a dynamic con-
text (see Supplementary Data). The lowest energy
docked structure for each conformation was selected,
and for each binding mode, a molecular dynamics
(MD) simulation.16 The whole system was partitioned
into a frozen and a mobile region, where positional re-
straints on both the ligand and the protein were progres-
sively reduced and finally removed.
The energetic analysis reveals that both drug conforma-
tions give rise to favorable enzyme–ligand complexes
(ꢁ10.4 and ꢁ15.1 kcal molꢁ1, for complex a and b,
respectively). The greater stability of the binding mode b
is mainly due to the electrostatic component computed

2280
C. Tomassi et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2277–2281
Figure 3. Schematic representation of the two binding models from docking simulations: a, left; b, right. The Ca trace of the enzyme is displayed as a
ribbon, colored according to the two subunits p66 (green) and p51 (yellow). The side chains of some relevant residues (Lys-A101, Lys-A103, Lys-
A172, Glu-B138, Thr-B139) are shown as sticks, with carbon atoms colored according to their subunit. Inhibitor 8 is displayed as sticks with carbon
atoms in cyan.
by the MM force field, which results in a stabilization by
acts as a potential novel pharmacophore for the oxo
and aza TSAO derivatives. Further modifications in
N-3 are in progress with a view to obtaining compounds
with increased biological activity, and the results will be
reported in due course.
7.2 kcal molꢁ1 of complex b relative to a.
Although the base part of TSAO derivatives is mostly
exposed to the solvent and run parallel to the p51/p66
subunits interface, the substitution at N-3 may induce
local distortions in the enzyme. The results summarized
above suggest that the binding site of HIV-1 RT is not
flexible enough as to easily accommodate ligands bear-
ing bulky substituents, and appears as visibly sensitive
to the drug conformation. For complex a, the tert-butyl
unit interferes with the side chains of residues Thr-A165
and Glu-A169 on the palm subdomain of p61, thus
forcing the drug to reorient inside the binding pocket
(torsion of Boc substituent by 52° from the B3LYP/6-
31G(d) optimized rotamer), which further inhibits the
formation of an efficient hydrogen bond between the
hindered carbonyl oxygen and the hydroxy group of
Thr-B139. For the alternative ligand conformation
(complex b), such an effect is less marked, since these
unfavorable steric contacts are softer, which induces a
minor displacement of the inhibitor and conformational
torsion (28° from the B3LYP/6-31G(d) optimized rot-
amer). Moreover, the free and unhampered Nf of Lys-
A172 may form a stable H-bond with the carbonyl
group, and is preserved along the simulation.
Acknowledgments
`
C.T. thanks the Ministere Franc¸ais de la Recherche
(France) for a fellowship. D.P. thanks the Conseil
´
`
Regional de Picardie and the Ministere Franc¸ais de la
Recherche for financial support. E.S. thanks MEC
(Spain) for a Juan de la Cierva contract.
Supplementary data
Docking results and computational methodology for the
molecular modeling can be found in the online version.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2008.03.010.
References and notes
1. Restle, T.; Muller, B.; Godoy, R. S. J. Biol. Chem. 1990,
265, 8986.
2. Wang, J.; Smerdon, S. J.; Jager, J.; Kohlstaedt, L. A.;
Rice, P. A.; Friedman, J. M.; Steitz, T. A. Proc. Natl.
Acad. Sci. U.S.A. 1994, 91, 7242.
3. (a) Pandey, P. K.; Kaushik, N.; Talele, T. T.; Yadav, P. N.
S.; Pandey, V. N. Biochemistry 2001, 40, 9505; (b) Pandey,
P. K.; Kaushik, N.; Singh, K.; Sharma, B.; Upadhyay, A.
K.; Kumar, S.; Harris, D. BMC Biochem. 2002, 3, 18.
In summary, these results point to a better accommoda-
tion for conformation b,17 and suggest that while a car-
bonyl group at N-3 position of the base is favorable for
establishing binding interactions with the enzyme, the
attachment of associated bulky substituents restricts
the proper adjustment of the ligand inside the binding
site. However, it might be speculated that compounds
bearing an unhindered carbonyl group at the position
N-3 could improve activity or resistance profiles by opti-
mizing the H-bond interaction with binding pocket and
thus reducing the unfavorable steric hindrance.
´
4. (a) Camarasa, M. J.; Perez-Perez, M. J.; San-Felix, A.;
Balzarini, J.; De Clercq, E. J. Med. Chem. 1992, 35, 2721;
´
´
´
(b) Perez-Perez, M. J.; San-Felix, A.; Camarasa, M. J.;
Balzarini, J.; De Clercq, E. Tetrahedron Lett. 1992, 33,
´
´
´
3029; (c) Camarasa, M. J.; San-Felix, A.; Velazquez, S.;
Perez-Perez, M. J.; Gago, F.; Balzarini, J. Curr. Top. Med.
Chem. 2004, 4, 945.
´
The work described herein reinforces our previous data
which demonstrates that a Boc group in N-3 position
´
´

C. Tomassi et al. / Bioorg. Med. Chem. Lett. 18 (2008) 2277–2281
2281
´
5. (a) Chamorro, C.; Perez-Perez, M. J.; Rodrıguez-Barrios,
F.; Gago, F.; De Clercq, E.; Balzarini, J.; San-Felix, A.;
´
´
´
2003, 3713; (c) Domınguez, L.; Nguyen Van Nhien, A.;
Tomassi, C.; Len, C.; Postel, D.; Marco-Contelles, J. J.
Org. Chem. 2004, 69, 843.
´
Camarasa, M. J. Antiviral Res. 2001, 50, 207; (b) Cam-
´
arasa, M. J.; Perez-Perez, M. J.; Velazquez, S.; San-Felix,
A.; Alvarez, R.; Ingate, S.; Jimeno, M. L.; De Clercq, E.;
´
´
´
11. Soriano, E.; Marco-Contelles, J.; Tomassi, C.; Nguyen Van
Nhien, A.; Postel, D. J. Chem. Inf. Model. 2006, 46, 1666.
´
´
12. (a) Perez-Perez, M. J.; San-Felix, A.; Balzarini, J.; De
Clercq, E.; Camarasa, M. J. J. Med. Chem. 1992, 35, 2988;
´
´
Balzarini, J. In Antiinfectives. Recent Advances in Chem-
istry and Structure–Activity Relationships; Bentley, P. H.,
O’Hanlon, P. J., Eds.; The Royal Society of Chemistry,
1997; p 259, Ch. 18.
´
´
(b) Ingate, S.; Perez-Perez, M. J.; De Clercq, E.; Balzarini,
J.; Camarasa, M. J. Antiviral Res. 1995, 27, 281.
´
6. Alvarez, R.; Jimeno, M. L.; Gago, F.; Balzarini, J.; Perez-
Perez, M. J.; Camarasa, M. J. Antiviral Chem. Chemother.
1998, 9, 333.
13. Chamorro, C.; De Clercq, E.; Balzarini, J.; Camarasa, M.
´
J.; San-Felix, A. Antiviral Chem. Chemother. 2000, 11, 61.
´
´
14. Sluis-Cremer, N.; Hamamouch, N.; San Felix, A.; Velaz-
quez, S.; Balzarini, J.; Camarasa, M. J. J. Med. Chem.
2006, 49, 4834.
´
7. (a) Velazquez, S.; San-Felix, A.; Perez-Perez, M. J.;
Balzarini, J.; De Clercq, E.; Camarasa, M. J. J. Med.
´
´
´
´
´
Chem. 1993, 36, 3230; (b) San-Felix, A.; Velazquez, S.;
Perez-Perez, M. J.; Balzarini, J.; De Clercq, E.; Camarasa,
15. AutoDock, Version 3.05; Morris, G. M.; Goodsell, D. S.;
Huey, R.; Hart, W. E.; Halliday, S.; Belew, R.; Olson, A.
J. The Scripps Research Institute, La Jolla, CA 92037-
1000, 2000.
´
´
M. J. J. Med. Chem. 1994, 37, 453.
´
´
´
´
8. Rodrıguez-Barrios, F.; Perez, C.; Lobaton, E.; Velazquez,
´
´
´
S.; Chamorro, C.; San-Felix, A.; Perez-Perez, M. J.;
Camarasa, M. J.; Pelemans, H.; Balzarini, J.; Gago, F.
J. Med. Chem. 2001, 44, 1853.
16. AMBER 8, Case, D. A.; Darden, T. A.; Cheatham, T. E.;
Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Merz, K.
M.; Wang, B.; Pearlman, D. A.; Crowley, M.; Brozell, S.;
Tsui, V.; Gohlke, H.; Mongan, J.; Hornak, V.; Cui, G.;
Beroza, P.; Schafmeister, C.; Caldwell, J. W.; Ross, W. S.;
Kollman, P. A. University of California, San Francisco,
2004.
17. It should be noted that both drug conformers show a
similar energy in the free state, being conformer a
0.11 kcal molꢁ1 more stable than b (B3LYP/6-31G*).
´
9. Bonache, M. C.; Chamorro, C.; Velazquez, S.; De Clercq,
E.; Balzarini, J.; Rodrıguez-Barrios, F.; Gago, F.; Cam-
´
´
arasa, M. J.; San-Felix, A. J. Med. Chem. 2005, 48, 6653.
10. (a) Nguyen Van Nhien, A.; Tomassi, C.; Len, C.; Marco-
Contelles, J.; Balzarini, J.; Pannecouque, C.; De Clercq,
E.; Postel, D. J. Med. Chem. 2005, 48, 4276; (b) Postel, D.;
Nguyen Van Nhien, A.; Marco, J. L. Eur. J. Org. Chem.