Design, synthesis, and evaluation of 1-(N-benzylamino)-2-phenyl-3-(1H-1,2,4-triazol-1-yl)propan-2-ols as antifungal agents

Article abstract of DOI:10.1016/j.bmcl.2008.02.027

A series of 1-(N-benzylamino)-2-phenyl-3-(1H-1,2,4-triazol-1-yl)propan-2-ols 6a-c, 7a-c, 8a, and 9a were prepared in five steps and evaluated for their antifungal activity. The most active compound 7b was docked into a home-made 3D model of the targeted enzyme confirming the importance of Tyr118, His377, and Ser378 residues in its binding mode.

Full text of DOI:10.1016/j.bmcl.2008.02.027

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1820–1824
Design, synthesis, and evaluation of 1-(N-benzylamino)-2-phenyl-3-
(1H-1,2,4-triazol-1-yl)propan-2-ols as antifungal agents
a
a
b
a
´
´
Francis Giraud, Cedric Loge, Fabrice Pagniez, Damien Crepin,
Patrice Le Papeb and Marc Le Borgnea,*
aUniversite´ de Nantes, Nantes Atlantique Universite´s, De´partement de Pharmacochimie, Biomole´cules et Cibles The´rapeutiques,
BioCiT-EA 1155, UFR de Sciences Pharmaceutiques, 1 rue Gaston Veil, Nantes, F-44035 Cedex 1, France
bUniversite´ de Nantes, Nantes Atlantique Universite´s, De´partement de Parasitologie et Mycologie Me´dicale,
Biomole´cules et Cibles The´rapeutiques, BioCiT-EA 1155, UFR de Sciences Pharmaceutiques,
1 rue Gaston Veil, Nantes, F-44035 Cedex 1, France
Received 17 November 2007; revised 8 February 2008; accepted 9 February 2008
Available online 14 February 2008
Abstract—A series of 1-(N-benzylamino)-2-phenyl-3-(1H-1,2,4-triazol-1-yl)propan-2-ols 6a–c, 7a–c, 8a, and 9a were prepared in five
steps and evaluated for their antifungal activity. The most active compound 7b was docked into a home-made 3D model of the tar-
geted enzyme confirming the importance of Tyr118, His377, and Ser378 residues in its binding mode.
Ó 2008 Elsevier Ltd. All rights reserved.
Invasive fungal infections are frequently observed in im-
mune-compromised patients suffering from AIDS or
subjected to invasive surgery, anti-cancer therapy or
graft receivers. Several treatments have been developed
to reduce the impact of fungal diseases such as azoles
(fluconazole, itraconazole, voriconazole, posaconazole),
amphotericin B, 5-fluorocytosine, and caspofungin.
Each molecule is targeting diverse biological pathways
which are essential for the fungi. Unfortunately, massive
use of those compounds as a curative or prophylactic
approach has favored the emergence of resistance show-
ing the need of the discovery of new antifungal
compounds.
Recent works were published by the group of Sheng
et al.3 who mentioned a promising approach for the
design CYP51 inhibitors from C. albicans. They studied
a large library of derivatives from the azole family and
suggested a pharmacophoric model. They noticed the
importance of Tyr118 and Ser378 key residues in the sta-
bilization of the inhibitors within the channel 2 which is
oriented to the FG loop.
Thus to verify their hypotheses and to confirm our own
observations, we decided to build chlorinated analogues
stacking
Our group is involved in the design and synthesis of
azole antifungals for several years now.1,2 We decided
to use molecular modeling tools with the aim to ratio-
nalize our previous results and to suggest original struc-
tures to the synthesis. A homology model would help us
to identify potential inhibitors of the lanosterol 14a-
demethylase (CYP51) of Candida albicans and
Aspergillus fumigatus.
N
Tyr118
N
OH
N
N
R
*
Ser378
Cl
Y
H-bonds
Cl
His377
Y : NO2, CN, CF3, NH2
R : H or CH3
Keywords: Homology model; Docking; H-bond acceptor; Triazole;
Oxirane; Azide; Microwave irradiation; Candida albicans; CYP51 in-
hibitors; Antifungal agents.
*
Scheme 1. General structure of compounds 6a–c, 7a–c, 8a and 9a and
targeted interactions.
Corresponding author. Tel.: +33 2 40 41 11 14; fax: +33 2 40 41 28
76; e-mail: marc.le-borgne@univ-nantes.fr
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.02.027

F. Giraud et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1820–1824
1821
of their benzylamine series (Scheme 1). A new synthetic
strategy was performed for the design of our 1-(N-ben-
zylamino)-2-phenyl-3-(1H-1,2,4-triazol-1-yl)propan-2-ols.
Introduction of adequate substituents in position Y such
as nitro, nitrile or trifluoromethyl groups would create
potential H-bond interactions with key amino acids.
On the other hand, the aromatic ring would generate
the p–p stacking with the Tyr118. We chose to compare
the biological activity of the nitro compounds with their
analogues bearing an amino group in Y.
mediate 4 was isolated in a 95% yield.6 Reduction of the
azide moiety to amine is described in several papers.
Fringuelli et al.7 developed a copper-catalyzed prepara-
tion of amino-alcohols, Kempf et al.6 and Shiozaki
et al.8 showed that the use of ammonium chloride could
reduce azides into amines. Other systems like the couple
zinc powder/ammonium formiate in methanol can in-
duce in situ generation of hydrogen to reduce the azide
moiety.9 Moreover the use of triphenylphosphine
according to Staudinger conditions could lead to the
formation of the amino-alcohol.10 Usually most of these
techniques need an aqueous work-up. We noticed that
amino-alcohol 3 was partially soluble in water, thus
we used an easier procedure based on a hydrogenated
reduction and we were able to isolate 1-amino-2-(2,4-
dichlorophenyl)-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (5)
using a simple filtration of the reaction medium in quan-
titative yield.11
We herein describe the synthesis of benzylamine
derivatives and their antifungal activities on C. albicans
and A. fumigatus strains. Additionally SAR studies will
be discussed with the help of molecular modeling.
Our laboratory already synthesized several families of
azole compounds and the use of microwave irradiation
provided us with an efficient procedure to access to the
key intermediate 3 (Scheme 2).4,5 So in the first step,
1H-1,2,4-triazole was alkylated with 2,20,40-trichloroace-
tophenone (1) in the presence of potassium carbonate as
a base, affording the 2-(1H-1,2,4-triazol-1-yl)-20,40-
dichloroacetophenone (2) in a 88% yield. Then, Cor-
ey–Chaykovsky reaction was performed using the reac-
tion couple sodium hydroxide/trimethylsulfoxonium
iodide (TMSOI) and led to the desired oxirane 3 in
excellent yield.
The following steps of the synthesis are described in
Scheme 3. The main reaction is a nucleophilic substitu-
tion of amino-alcohol 5 with appropriate substituted
benzyl bromides in the presence of the Hunig base.
¨
Experimental protocol was optimized to afford the
mono-N-substituted derivatives. Two equivalents of
amino-alcohol 5 was placed in the presence of the base
in a diluted medium and a solution of one equivalent
of bromide was slowly added via a syringe. Depending
on the volume added, several hours are needed for the
complete addition.12 We isolated three original deriva-
tives 6a–c in moderate to good yields. Furthermore, a
methyl group was introduced on the amine spacer using
formaldehyde and reductive amination conditions.13
This led to the preparation of three additional deriva-
tives 7a–c in satisfactory yields.14 Amino compounds
8a and 9a were obtained from their nitro analogues 6a
and 7a using reducing conditions.15
Whereas the group of Sheng used the nucleophilicity of
N-methylbenzylamines to open the oxirane ring,3 we
developed an original synthetic route to prepare our
compounds (Schemes 2 and 3).
In the third step, the ring of epoxide 3 is opened by the
use of sodium azide in the presence of ammonium chlo-
ride in methanol and after one night under reflux, inter-
Compounds 6a–c, 7a–c, 8a, and 9a were screened for
their antifungal activity on C. albicans CA98001 and
A. fumigatus AF98003 strains. Inhibition growth was
measured according to the protocol described in a previ-
ous journal.16 Fluconazole and itraconazole were used
as positive controls. The minimum inhibitory concentra-
tion (MIC80) values (in ng mLꢀ1) are summarized in
Table 1.
N
N
N
N
O
N
N
Cl
O
O
a
b
*
Cl
Cl
Cl
Cl
Cl
Cl
1
2 (88%)
3 (97%)
On the C. albicans strain, our compounds showed a
high level of activity with MIC values 6- to 500-fold
lower than that of fluconazole. The 4-{N-[2-(2,4-
dichlorophenyl)-2-hydroxy-3-(1H-1,2,4-triazol-1-yl)pro-
pyl]-N-methyl}aminomethylbenzonitrile (7b) showed a
MIC80 value of 0.37 ng mLꢀ1 and was the most ac-
tive compound of the series whereas compound 8a
N
N
N
OH
N
N
N
OH
c
d
N3
NH2
*
*
Cl
Cl
(Y = NH2, linker N–H, MIC80 = 18,830.0 ng mLꢀ1
)
was inactive. A simple comparison of the two sub-
series (linker N–CH3 or N–H) showed that, excepted
for compounds 6c and 7c, the methyl group would
be favorable for the activity. For example, compound
9a bearing an amino group in position Y and a N–
Cl
4 (95%)
Cl
5 (quantitative)
Scheme 2. General synthetic route for the preparation of intermediate
5. Reagents and conditions: (a) 1H-1,2,4-triazole, K2CO3, CH3CN,
MW, 50W, 85 °C, 50 min; (b) NaOHaq, TMSOI, toluene, MW, 10W,
80 °C, 50 min; (c) NaN3, NH4Cl, MeOH, reflux, 15 h; (d) H2 (5 bars),
Pd/C 5% (10% weight), EtOH, rt, 16 h.
CH3 linker showed a MIC80 value of 29.0 ng mLꢀ1
.
It was almost 1000-fold more active than its analogue
8a. Comparison of 7b (MIC80 = 0.37 ng mLꢀ1) which

1822
F. Giraud et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1820–1824
N
N
N
N
OH
N
OH
N
Br
N
N
N
OH
Y
NH2
N
H
N
Me
*
*
*
Cl
Cl
Cl
Y
Y
a
b
NO2 (76%)
CN (63%)
CF3 (66%)
7a
7b
7c
Cl
Cl
Cl
NO2 (56%)
CN (56%)
CF3 (32%)
6a
6b
6c
5
Y : NO2
Y : NO2
c
c
N
N
N
OH
N
OH
N
N
N
H
N
Me
*
*
Cl
Cl
NH2
NH2
Cl
Cl
8a (98%)
9a (47%)
Scheme 3. Final steps for the preparation of targeted compounds 6a–c, 7a–c, 8a, and 9a. Reagents and conditions: (a) benzyl bromide (0.50 equiv), (i-
Pr)2NEt (0.55 equiv), CH3CN, rt, 16–48 h; (b) HCHO, NaBH3CN, AcOH/MeOH 2% v/v, rt, 16 h; (c) H2 (5 bars), Pd/C 5% (10% weight), EtOH, rt,
16 h.
Table 1. In vitro antifungal activity of benzylamine derivatives 6a–c, 7a–c, 8a, and 9a
Compound
R
Y
MIC80 values (ng mLꢀ1
)
Candida albicans CA98001
Aspergillus fumigatus AF98003
6a
6b
H
NO2
CN
6.0 1.3
2.8 0.4
27,020.0 840.00
4020.0 800.00
24,040.0 2230.00
3060.0 120.00
1960.0 170.00
2410.0 40.00
21,130.0 1380.00
2320.0 120.00
—
H
6c
8a
H
CF3
NH2
NO2
CN
24.0 2.0
18,830.0 2750.00
0.6 1.3
H
7a
7b
CH3
CH3
CH3
CH3
0.37 0.16
30.0 3.0
29.0 2.0
190.0 6.0
—
7c
9a
CF3
NH2
Fluconazole
Itraconazole
420.0 40.0
was 10-fold more active than its analogue 6b which
has a nitrile group and a non-substituted spacer
(MIC80 = 2.8 ng mLꢀ1) led to the same conclusion.
The presence of a methyl group would play a major
role in the orientation of the inhibitors within the ac-
tive site.
(Scheme 4).18 These studies helped us to understand
the structure–activity relationship of this compound
within the active site. The nitrile group should share
two H-bonds with the key amino acids His377 and
Ser378 whereas the N-methyl group of the linker would
be oriented to a hydrophobic pocket (Tyr118, Leu121,
Phe126, and Phe228).
In addition the most suitable groups in position Y are H-
bond acceptor entities. Indeed compounds 6a (Y =
These results are in accordance with the conclusions of
the group of Sheng and additionally we identified a
new residue which seems to be essential for the binding
of inhibitors: His377.
NO2, MIC80 = 6.0 ng mLꢀ1), 6b (Y = CN, MIC80
=
2.8 ng mLꢀ1), and 6c (Y = CF3, MIC80 = 24.0 ng mLꢀ1
are almost 1000-fold more active than their amine
analogue 8a.
)
Furthermore, biological results showed the emergence of
activity of our compounds on the Aspergillus strain with
In our precedent series, (S)-isomers were the most active
compounds.5,17 Based on this observation, we realized
the docking of the most active compound 7b under this
configuration in our model of CYP51-C. albicans
MIC80 values ranging from 1960.0 to 2410.0 ng mLꢀ1
.
The N-methyl group seems to be also important here.
Indeed, 2-(2,4-dichlorophenyl)-1-[N-methyl-N-(4-nitro-

F. Giraud et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1820–1824
1823
chloride (1.95 g, 36.48 mmol) was added and the solution
was refluxed for 15 h. Mixture was diluted with water and
product was extracted with dichloromethane. Organic
layers were combined, dried over anhydrous Na2SO4 and
evaporated to get the right product (95% yield, yellow
1
powder) without further purification. mp: 116–117 °C; H
2
NMR (DMSO-d6): d 3.76 (d, 1H, J = 13.1 Hz), 4.13 (d,
1H, 2J = 13.1 Hz), 4.71 (d, 1H, 2J = 14.3 Hz), 4.88 (d, 1H,
2J = 14.3 Hz), 6.51 (s, 1H, OH), 7.41 (dd, 1H, 3J = 8.5 Hz,
4J = 2.3 Hz), 7.60 (d, 1H, 3J = 8.5 Hz), 7.63 (d, 1H,
4J = 2.3 Hz), 7.82 (s, 1H), 8.36 (s, 1H). IR (KBr, cmꢀ1):
783 (m C–Cl), 1268 (m C–N), 1473, 1515 (m C@C), 1587 (m
C@N), 2098 (m C–N3), 3111 (m O–H).
7. Fringuelli, F.; Pizzo, F.; Rucci, M.; Vaccaro, L. J. Org.
Chem. 2003, 68, 7041.
8. Shiozaki, M.; Arai, M.; Kobayashi, Y.; Kasuya, A.;
Miyamoto, S.; Furukawa, Y.; Takayama, T.; Haruyama,
H. J. Org. Chem. 1994, 59, 4450.
9. Kamal, A.; Srinivasa, K.; Reddy, B.; Prasad, R.; Hari
Babu, A.; Venkata Ramana, A. Tetrahedron Lett. 2004,
45, 6517.
10. Mangelinckx, S.; Boeykens, M.; Vliegen, M.; Van der
Eycken, J.; De Kimpe, N. Tetrahedron Lett. 2005, 46, 525.
11. Reitz, A. B.; Tuman, R. W.; Marchione, C. S.; Jordan,
A. D.; Bowden, C. R.; Maryanoff, B. E. J. Med. Chem.
1989, 32, 2110, Synthesis of 1-amino-2-(2,4-dichloro-
Scheme 4. Docking of compound (S)-7b in the active site of CYP51-
Candida albicans. Hip377 is the protonated form of histidine residue.
benzyl)amino]-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (7a)
was the most active compound on the A. fumigatus strain
phenyl)-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (5). To
a
solution of 1-azido-2-(2,4-dichlorophenyl)-3-(1H-1,2,4-
triazol-1-yl)-propan-2-ol (4) (14.8 g, 47.26 mmol) in
120 mL of ethanol was added 5% active coal-supported
palladium (1.48 g). The solution was stirred overnight
at room temperature under a hydrogen atmosphere (5
bars) and filtered off through celite. The residue was
washed with ethyl acetate and evaporated to get the
desired product 5 (quantitative yield, white powder)
without further purification. mp: 215–216 °C; 1H NMR
(DMSO-d6): d 3.10 (d, 1H, 2J = 13.4 Hz), 3.29 (d, 1H,
2J = 13.4 Hz), 4.65 (d, 1H, 2J = 14.3 Hz), 4.91 (d, 1H,
2J = 14.3 Hz), 7.35 (dd, 1H, 3J = 8.5 Hz, 4J = 2.1 Hz),
7.52–7.58 (m, 2H), 7.77 (s, 1H), 8.33 (s, 1H). IR (KBr,
cmꢀ1): 806 (m C–Cl), 1272 (m C–N), 1464, 1508 (m
C@C), 1586 (m C@N), 1620 (m N–H), 3100–3450 (m O–
H and m NH2).
and showed the lowest MIC80 value of 1960.0 ng mLꢀ1
.
All these compounds could bind to the active site by H-
bond interactions.
To verify our conclusions, an HPLC separation of com-
pounds 6a–c, 7a–c, 8a, and 9a should be undertaken to
check if the most active isomers are in the (S)-configura-
tion as previously observed. Moreover, a homology
model of the CYP51-A. fumigatus enzyme should be
build according to the same procedure. More insightful
observations of the active site would give us some key
informations for further design of broad-spectrum anti-
fungal agents.
12. Synthesis of 2-(2,4-dichlorophenyl)-1-(4-nitrobenzylami-
no)-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (6a). To a solu-
References and notes
tion
of
1-amino-2-(2,4-dichlorophenyl)-3-(1H-1,2,4-
triazol-1-yl)propan-2-ol (5) (613 mg, 2.13 mmol) in
24 mL of acetonitrile was added N,N0-diisopropylethyl-
amine (0.17 mL, 1.17 mmol) under argon at room
1. Na, Y. M.; Le Borgne, M.; Pagniez, F.; Le Baut, G.; Le
Pape, P. Eur. J. Med. Chem. 2003, 38, 75.
2. Pagniez, F.; Le Pape, P.; Na, Y. M.; Lebouvier, N.; Le
Borgne, M.; Le Baut, G. Abstracts of paper, 10th Congress
of the European Confederation of Medical Mycology,
Wroclaw, Poland, June 17–20, 2004; P023.
3. Sheng, C.; Zhang, W.; Ji, H.; Zhang, M.; Song, Y.; Xu,
H.; Zhu, J.; Miao, Z.; Jiang, Q.; Yao, J.; Zhou, Y.; Zhu, J.;
Lu, J. J. Med. Chem. 2006, 49, 2512.
4. Lebouvier, N.; Giraud, F.; Corbin, T.; Na, Y. M.; Le
Baut, G.; Marchand, P.; Le Borgne, M. Tetrahedron Lett.
2006, 47, 6479.
5. Lebouvier, N.; Pagniez, F.; Duflos, M.; Le Pape, P.; Na,
Y. M.; Le Baut, G.; Le Borgne, M. Bioorg. Med. Chem.
Lett. 2007, 17, 3686.
temperature.
A
solution of 4-nitrobenzyl bromide
(230 mg, 1.06 mmol) in 10 mL of acetonitrile was slowly
added to the mixture in 40 min and the solution was
stirred for 24 h at room temperature. Solvent was
removed under reduced pressure and residue was
partitioned between dichloromethane and water. Prod-
uct was extracted with dichloromethane and organic
layers were dried over anhydrous Na2SO4 and concen-
trated in vacuo. The residue was purified on silica gel
column chromatography (ethanol/dichloromethane 1:10)
and compound 6a was obtained in a 56% yield as a
yellow oil. 1H NMR (DMSO-d6):
d 3.02 (d, 1H,
2J = 12.5 Hz), 3.29 (d, 1H, 2J = 12.5 Hz), 3.82 (s, 2H),
4.68 (d, 1H, 2J = 14.1 Hz), 4.90 (d, 1H, 2J = 14.1 Hz),
5.95 (s, 1H, OH), 7.34 (dd, 1H, 3J = 8.8 Hz,
4J = 2.4 Hz), 7.53–7.59 (m, 4H), 7.75 (s, 1H), 8.19 (d,
2H, 3J = 8.8 Hz), 8.32 (s, 1H). IR (NaCl, cmꢀ1): 805 (m
C–Cl), 1266 (m C–N), 1348 (m NO2 sym), 1512 (m NO2
asym), 1589 (m C@C and m C@N), 2927 (m C–Haliph.),
3318 (m O–H). MS m/z 423.1 (M+H).
6. Kempf, D. J.; De Lara, E.; Stein, H. H.; Cohen, J.;
Platnner, J. J. J. Med. Chem. 1987, 30, 1978, Synthesis
of 1-azido-2-(2,4-dichlorophenyl)-3-(1H-1,2,4-triazol-1-
yl)propan-2-ol (4). To a solution of 2-(2,4-dichloro-
phenyl)-3-(1H-1,2,4-triazol-1-yl)-1,2-epoxypropane
(3)
(4.36 g, 16.14 mmol) in 50 mL of methanol was added
sodium azide (3.15 g, 48.42 mmol). Then ammonium

1824
F. Giraud et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1820–1824
13. Bolognesi, M. L.; Andrisano, V.; Bartolini, M.; Banzi, R.;
Melchiorre, C. J. Med. Chem. 2005, 48, 24.
2J = 14.0 Hz), 4.69 (d, 1H, 2J = 14.1 Hz), 4.86 (d, 1H,
3
2J = 14.1 Hz), 5.95 (s, 1H, OH), 7.31 (d, 2H, J = 8.8 Hz),
14. Synthesis of 2-(2,4-dichlorophenyl)-1-[N-methyl-N-(4-nitro-
benzyl)amino]-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (7a).
To a solution of 2-(2,4-dichlorophenyl)-1-(4-nitrobenzyla-
mino)-3-(1H-1,2,4-triazol-1-yl)propan-2-ol (6a) (116 mg,
0.27 mmol) in 2 mL of methanol and 2% of acetic acid
was added formaldehyde (30% weight solution, 22.5 lL,
0.27 mmol) under argon at room temperature. Then
sodium cyanoborohydride (21 mg, 0.39 mmol) was added
and the mixture stirred for one night at room temperature.
It was then diluted with water and the product was
extracted with dichloromethane. Organic layers were dried
over anhydrous Na2SO4 and concentrated in vacuo. The
residue was purified on silica gel column chromatography
(ethanol/dichloromethane 2:98) and compound 7a was
obtained in a 43% yield as a yellow oil. 1H NMR (DMSO-
d6): d 2.10 (s, 3H, CH3), 2.85 (d, 1H, 2J = 13.7 Hz), 3.47 (d,
1H, 2J = 13.7 Hz), 3.54 (d, 1H, 2J = 14.0 Hz), 3.76 (d, 1H,
7.34 (dd, 1H, 3J = 8.2 Hz, 4J = 2.1 Hz), 7.58 (d, 1H,
3
4J = 2.1 Hz), 7.60 (d, 1H, J = 8.2 Hz), 7.78 (s, 1H), 8.10
3
(d, 2H, J = 8.8 Hz), 8.33 (s, 1H). IR (NaCl, cmꢀ1): 1274
(m C–N), 1347 (m NO2 sym), 1519 (m NO2 asym), 1464,
1602 (m C@C and m C@N), 2927 (m C–Haliph.), 3370 (m O–
H). MS m/z 436.2 (M+).
15. Compounds 8a and 9a were prepared from 6a and 7a
according to the same protocol as described for compound
5 but the reaction time was reduced to 1.5 h. Products
were purified on silica gel column chromatography (eth-
anol/dichloromethane 1:10).
16. Pagniez, F.; Le Pape, P. J. Mycol. Med. 2001, 11,
73.
´
17. Lebouvier, N. Ph.D. Thesis, Universite de Nantes, Nantes
´
18. Giraud, F. Ph.D. Thesis, Universite de Nantes, Nantes
Atlantique Universites, October 2004.
´
´
Atlantique Universites, October 2007.