Antifungal agents. 10. New derivatives of 1-[(aryl)[4-aryl-1H-pyrrol-3-yl]methyl]-1H-imidazole, synthesis, anti-Candida activity, and quantitative structure-analysis relationship studies

Article abstract of DOI:10.1021/jm011087h

The synthesis, anti-Candida activity, and quantitative structure-activity relationship (QSAR) studies of a series of 2,4-dichlorobenzylimidazole derivatives having a phenylpyrrole moiety (related to the antibiotic pyrrolnitrin) in the α-position are repor

Full text of DOI:10.1021/jm011087h

2720
J . Med. Chem. 2002, 45, 2720-2732
An tifu n ga l Agen ts. 10. New Der iva tives of
1-[(Ar yl)[4-a r yl-1H-p yr r ol-3-yl]m eth yl]-1H-im id a zole, Syn th esis, An ti-Ca n d id a
Activity, a n d Qu a n tita tive Str u ctu r e-An a lysis Rela tion sh ip Stu d ies
Andrea Tafi,† Roberta Costi,‡ Maurizio Botta,† Roberto Di Santo,‡ Federico Corelli,† Silvio Massa,†
Andrea Ciacci,† Fabrizio Manetti,† and Marino Artico*,‡
Dipartimento Farmaco Chimico Tecnologico, Universita` di Siena, Via Aldo Moro, S. Miniato, I-53100 Siena, Italy,
“Istituto Pasteur-Fondazione Cenci Bolognetti”, Dipartimento di Studi Farmaceutici, Facolta` di Farmacia,
Universita` di Roma “La Sapienza”, P.le Aldo Moro 5, I-00185 Roma, Italy
Received November 2, 2001
The synthesis, anti-Candida activity, and quantitative structure-activity relationship (QSAR)
studies of a series of 2,4-dichlorobenzylimidazole derivatives having a phenylpyrrole moiety
(related to the antibiotic pyrrolnitrin) in the R-position are reported. A number of substituents
on the phenyl ring, ranging from hydrophobic (tert-butyl, phenyl, or 1-pyrrolyl moiety) to basic
(NH2), polar (CF3, CN, SCH3, NO2), or hydrogen bond donors and acceptor (OH) groups, were
chosen to better understand the interaction of these compounds with cytochrome P450 14-R-
lanosterol demethylase (P45014DM). Finally, the triazole counterpart of one of the imidazole
compounds was synthesized and tested to investigate influence of the heterocyclic ring on
biological activity. The in vitro antifungal activities of the newly synthesized azoles 10p -v,x-
c′ were tested against Candida albicans and Candida spp. at pH 7.2 and pH 5.6. A CoMFA
model, previously derived for a series of antifungal agents belonging to chemically diverse
families related to bifonazole, was applied to the new products. Because the results produced
by this approach were not encouraging, Catalyst software was chosen to perform a new 3D-
QSAR study. Catalyst was preferred this time because of the possibility of considering each
compound as a collection of energetically reasonable conformations and of considering
alternative stereoisomers. The pharmacophore model developed by Catalyst, named HYPO1,
showed good performances in predicting the biological activity data, although it did not exhibit
an unequivocal preference for one enantiomeric series of inhibitors relative to the other. One
aromatic nitrogen with a lone pair in the ring plane (mapped by all of the considered compounds)
and three aromatic ring features were recognized to have pharmacophoric relevance, whereas
neither hydrogen bond acceptor nor hydrophobic features were found. These findings confirmed
that the key interaction of azole antifungals with the demethylase enzyme is the coordination
bond to the iron ion of the porphyrin system, while interactions with amino acids localized in
proximity of heme could modulate the biological activity of diverse antifungal agents. In
conclusion, HYPO1 conveys important information in an intuitive manner and can provide
predictive capability for evaluating new compounds.
Ch a r t 1. Natural Sterols
In tr od u ction
Azole antifungal agents are useful drugs and are
widely used for the treatment of topic or inner mycoses,
in particular AIDS-related mycotic pathologies.1-4 It is
well-known that azole derivatives block ergosterol bio-
synthesis, causing its depletion and accumulation of
lanosterol and some other 14-methylsterols5-7 (Chart
1). Such sterols alter membrane fluidity with con-
comitant reduction in the activity of membrane-associ-
ated enzymes, increased permeability, and inhibition of
cell growth and replication.8-9
The first step of ergosterol biosynthesis is the de-
methylation of lanosterol performed by 14-R-lanosterol
demethylase (P45014DM, CYP51), a member of the
enzyme cytochrome P450 dependent superfamily,10,11
which catalyzes the removal of 14-R-methyl group of
lanosterol. Crossover inhibition of CYP51 in different
species is assumed to cause undesirable side effects and
is one of the reasons for the search of better, more
selective agents.
The problem of specificity is commonly addressed
empirically by applying computer simulation techniques
such as receptor fitting and receptor mapping. In the
first approach, the development of structural models for
the active site of enzymes (based on primary sequence
analyses and modeling by homology) is exploited. The
structures of a number of P450-dependent enzymes
* To whom correspondence should be addressed. Phone and fax:
+39-6-4462731. E-mail: marino.artico@uniroma1.it.
† Universita` di Siena.
‡ Universita` di Roma “La Sapienza.
10.1021/jm011087h CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/29/2002

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2721
Ch a r t 2. Azole Antifungal Agents Used in Clinical Practice
belonging to prokaryotic microorganisms, such as
P450cam from Pseudomonas putida,12 P450eryF from
Saccaropolyspora erythrea,13 and P450ter from Pseudo-
monas sp.,14 have been elucidated, while only the crystal
structure of a CYP51 cytochrome, a soluble orthologue
from Mycobacterium tuberculosis (MTCYP51), is avail-
able until now.15 Despite a limited sequence identity
(generally 10-30%), all cytochromes P450 show very
similar secondary and tertiary structures; consequently,
homology modeling studies have been performed in the
past few years, starting from 3D structures of the above-
mentioned prokaryotic enzymes.16-18
A pure, indirect approach focusing mainly on struc-
tural and physicochemical properties of active and
inactive compounds (receptor mapping) has been applied
in recent years19,20 to identify structure-activity rela-
tionships helpful for designing new, more specific drug
candidates and eventually for understanding the en-
zyme active-site topography. Both of the above-men-
tioned theoretical approaches have reached similar
conclusions concerning structure-activity relationships,
indicating that the binding of azole antifungal agents
involves a principal interaction between the azole ring
and the CYP51 heme moiety. There are additional and
mainly lypophilic (π-π stacking and steric) interactions
within the heme environment. Supplementary interac-
tions with residues of the substrate channel access have
also been invoked, even though they are not relevant
for activity, in the case of inhibitors such as keto-
conazole.16 In the case of fluconazole interacting with
MTCYP51, this overall scenario has only just been
confirmed; the inhibitor, in fact, has been found to
coordinate with the iron ion and to show a few further
nonbonded contacts with residues of the active-site
chamber.15
azoles27,34 showed the highest activity in either in vitro
or in vivo tests and compounds 10a and 10o were found
to be effective against experimental cutaneous candi-
dosis (C. albicans A170 strain) produced in albino
female rabbits with topical efficacy higher or comparable
to that of bifonazole, used as a reference drug.34 In this
research, we synthesized and tested a number of
derivatives substituted on the aryl moiety of the ben-
zylimidazole group with a halogen, methyl, amino, or
nitro group. In particular, the 2,4-dichloro compound
proved to lead to the most active compounds.34
Now we describe the synthesis, anti-Candida activity,
and 3D quantitative structure-activity relationship
(QSAR) studies of a series of 2,4-dichlorobenzylimidazole
derivatives 10p -c′ linked in the R-position to a phenyl-
pyrrole moiety related to the antibiotic pyrrolnitrin. A
number of groups, ranging from hydrophobic (tert-butyl,
phenyl, or 1-pyrrolyl moiety) to basic (NH2), polar (CF3,
CN, SCH3, NO2), or hydrogen-bonding groups (OH),
were chosen as substituents on the phenyl ring to get
an insight on the influence of the substitution pattern
on biological activity. Finally, the triazole counterpart
of 10o (namely, compounds 10b′ and 10c′) was synthe-
sized and tested (Chart 3). A pharmacophore model was
developed in order to explain the structure-activity
relationships of the new derivatives 10p -c′ and of the
antifungals 4 and 7a -10o. Catalyst software was
preferred to perform the 3D-QSAR study mainly be-
cause of the opportunity it gave to consider each
compound as a collection of energetically reasonable
conformations and to consider alternative stereoisomers.
Ch em istr y
Compounds 10p -c′ were synthesized as depicted in
Schemes 1-3. Azolyl derivatives 10p -s,y-c′ were ob-
tained by the annulation of toluene-4-sulfonylmethyl-
isocyanide (TosMIC) with the properly substituted chal-
cones 11a -d ,g,h , in the presence of sodium hydride as
a catalyst, to give 4-aryl-3-aroyl pyrroles 12a -d ,i,j.
Compounds 11a -d ,g,h , employed as starting material,
were synthesized by condensation of the 2,4-dichloro-
acetophenone with substituted benzaldehydes in the
presence of aqueous sodium hydroxide.
Since the identification of clotrimazole (1)21 in 1972,
a number of antifungal azole agents were studied and
are now used in clinical practice. Miconazole (2) and
related derivatives, together with ketoconazole (3)22,23
and bifonazole (4),24 belong to the class of imidazoles;
fluconazole (5) and itraconazole (6) are the most impor-
tant drugs of the triazole family1-4 (Chart 2).
Our decennial interest in antifungal drugs,19,25-36 led
us to the discovery of potent antimycotic compounds
belonging to azolyl diaryl ethane 726,32,35 or azolyl diaryl
methane groups 8, 9, and 10a -o.19,27-30,33,34 In the latter
class, 1-[(aryl)[4-aryl-1H-pyrrol-3-yl]methyl]-1H-imid-
Alkylation of the known pyrrole 12j34 with methyl or
ethyl iodide in alkaline medium (K2CO3) gave N-
alkylpyrrolyl methanones 12k or 12l, respectively, and

2722 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
Ch a r t 3. Newly Synthesized Derivatives 10p -c and
Azoles 7-9 and 10a -o Used in the QSAR Studies
Sch em e 1a
a
Reagents: (a) tosylmethylisocyanide (TosMIC), NaH, DMSO,
Et2O; (b) alkyl iodide, K2CO3, DMF; (c) NaBH4, THF, H2O; (d) 1,1′-
carbonyldiimidazole (CDI), MeCN; (e) 1,1′-carbonylditriazole,
MeCN.
Sch em e 2a
sodium borohydride reduction of ketones 12a -d ,i-l in
a refluxing tetrahydrofuran/water mixture afforded the
corresponding carbinols 13a -d ,h -k . The latter com-
pounds were treated with 1,1′-carbonyldiimidazole to
give imidazoles 10p -s,y-a ′, while reaction of derivative
13i with 1,1′-carbonylditriazole, led to a mixture of 1,2,4-
and 1,3,4-triazolyl isomers 10b′ and 10c′, which were
separated by column chromatography (Scheme 1).
a
Reagents: (a) dihydropyran, PPTS; (b) 1-(2,4-dichlorophenyl)-
The synthesis of hydroxy derivative 10x is depicted
in Scheme 2. The required basic reaction conditions of
the two first steps forced us to a preliminary protection
of the hydroxy group. For this, 4-hydroxybenzaldehyde
was first treated with 2,3-dihydro-2H-pyran in the
presence of pyridinium p-toluenesulfonate (PPTS) as a
catalyst to afford 4-(tetrahydro-2-pyranyloxy)benzalde-
hyde, which was then condensed with (2,4-dichloro-
phenyl)ethanone in alkaline medium (barium hydrox-
ide), leading to the chalcone 11f. The TosMIC reaction,
sodium borohydride reduction, and CDI treatment in
turn led to protected imidazole derivatives 10w , which
was then treated with p-toluenesulfonic acid to afford
imidazole 10x.
ethanone, Ba(OH)2, MeOH; (c) tosylmethylisocyanide (TosMIC),
NaH, DMSO, Et2O; (d) NaBH4, THF, H2O; (e) 1,1′-carbonyldiimid-
azole (CDI) MeCN; (f) PTSA, MeOH.
The synthetic pathway for derivatives 10t-v is
reported in Scheme 3. Sodium borohydride reduction of
methanone 12e, in turn prepared by reaction of TosMIC
with nitrochalcone 11e, gave the corresponding carbinol
13e, which was then reacted with CDI to furnish
imidazole 10t. Catalytic hydrogenation of the latter
compound with hydrogen in the presence of 5% pal-
ladium on charcoal provided the amino derivative 10u ,
while tin chloride reduction in acidic medium of 12e led
to amino methanone 12f, which was treated with 2,5-

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2723
Sch em e 3a
a
Reagents: (a) tosylmethylisocyanide (TosMIC), NaH, DMSO, Et2O; (b) NaBH4, THF, H2O; (c) 1,1′-carbonyldiimidazole (CDI), MeCN;
(f) PTSA, MeOH; (d) SnCl2, HCl, EtOH; (e) H2, Pd/C, AcOEt; (f) 2,5-dimethoxytetrahydrofuran, AcOH.
Ta ble 1. IR and 1H NMR Spectra of Azole Derivatives 10p -c′
1H NMR
solvent
compd
IR, cm-1
3110 (NH)
1H NMR, δ
10p
CDCl3
1.30 (9H, s, CH3), 6.28 (1H, m, pyrrole CR-H), 6.57-7.47 (12H, m, pyrrole
CR-H, imidazole H, benzene H and CH), 8.96 (1H, bs, NH)
10q
10r
3100 (NH)
3110 (NH)
CDCl3
6.31 (1H, m, pyrrole CR-H), 6.77-6.99 (4H, m, pyrrole CR-H, benzene H and
CH), 7.10-7.60 (14H, m, imidazole H and benzene H), 8.73 (1H, bs, NH)
6.28 (1H, m, pyrrole CR-H), 8.79 (1H, d, J o ) 8.4 Hz, benzene H), 6.94 (3H, m,
imidazole C4-H and C5-H and benzene H), 7.11 (1H, s, CH), 7.23 (1H,
m, pyrrole CR-H), 7.34-7.45 (3H, m, benzene H), 7.58-7.63 (4H, m,
imidazole C2-H and benzene H), 11.33 (1H, bs, NH)
DMSO-d6
10s
10t
3100 (NH), 2210 (CN)
3100 (NH)
DMSO-d6
DMSO-d6
6.26 (1H, m, pyrrole CR-H), 6.77 (1H, d, J o ) 8.4 Hz, benzene H), 6.96 (2H, m,
benzene H), 7.10 (1H, s, CH), 7.28-7.44 (4H, m, imidazole C4-H and
C5-H and benzene H), 7.63-7.72 (4H, m, pyrrole CR-H, imidazole C2-H
and benzene H), 11.37 (1H, bs, NH)
6.28 (1H, d, J o1 ) 8.4 Hz, benzene H), 6.94-7.11 (3H, m, imidazole C4-H
and C5-H and CH), 7.37-7.48 (4H, m, pyrrole CR-H and benzene H),
7.65 (1H, s, imidazole C2-H), 8.10 (2H, d, J o2 ) 8.8 Hz, benzene H near
nitro group), 11.46 (1H, bs, NH)
10u
10v
3410, 3320, 3110 (NH2, NH)
3110 (NH)
CDCl3
1.77 (2H, bs, NH2), 6.25 (1H, m, pyrrole CR-H), 6.59-7.44 (12H, m, pyrrole
CR-H, imidazole H, benzene H and CH), 8.44 (1H, bs, NH)
DMSO-d6
6.24 (3H, m, pyrrole CR-H), 6.78-7.50 (13H, m, pyrrole CR-H, imidazole
C4-H and C5-H, benzene H and CH), 7.62 (1H, m, imidazole C2-H),
11.19 (1H, bs, NH)
10x
3300, 3110 (OH, NH)
3100 (NH)
CDCl3
6.21 (1H, m, pyrrole CR-H), 6.66-6.83 (7H, m, imidazole C4-H and C5-H,
benzene H and CH), 6.92 (1H, m, pyrrole CR-H), 7.11-7.38 (3H, m,
benzene H), 7.53 (1H, m, imidazole C2-H), 8.50 (1H, bs, NH)
3.38 (3H, s, CH3), 6.25 (1H, m, pyrrole CR-H), 6.77-7.63 (12H, m, pyrrole
CR-H, imidazole H, benzene H and CH), 11.12 (1H, bs, NH)
3.66 (3H, s, CH3), 6.11 (1H, d, J 2,5 ) 2.2 Hz, pyrrole CR-H), 6.69-7.42 (11H, m,
pyrrole CR-H, imidazole C4-H and C5-H, benzene H and CH), 7.48 (1H,
m, imidazole C2-H)
10y
10z
DMSO-d6
CDCl3
10a ′
10b′
10c′
CDCl3
CDCl3
CDCl3
1.42 (3H, t, CH3), 3.87 (2H, q, CH2), 6.12 (1H, m, pyrrole CR-H), 6.66-7.44
(12H, m, pyrrole CR-H, imidazole H, benzene H and CH)
3110 (NH)
3110 (NH)
6.11-7.40 (12H, m, pyrrole CR-H, imidazole C3-H, benzene H and CH),
7.93 (1H, m, imidazole C5-H), 8.23 (1H, bs, NH)
6.40 (1H, m, pyrrole CR-H), 6.77-7.41 (9H, pyrrole CR-H and benzene H),
7.97 (1H, d, J 2,5 ) 4.0 Hz, imidazole C2-H and C5-H), 8.89 (1H, bs, NH)
dimethoxytetrahydrofuran in boiling acetic acid, accord-
ing to the Clauson-Kaas procedure, to give pyrrole
derivative 12g. Finally, sodium borohydride reduction
of 12g afforded the corresponding carbinol 13f, which
on treatment with CDI led to imidazole 10v. The IR and
NMR data of azoles 10p -c′ and the chemical and
physical properties of compounds 10p -c′-13 are re-
ported in Tables 1 and 2, respectively.
Resu lts a n d Discu ssion
Micr obiology. In Vitr o An ti-Ca n d id a Activities.
The in vitro antifungal activities of the newly synthe-

2724 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
Ta ble 2. Chemical and Physical Data of Derivatives 10p -c′ and 11-13a
recrystb
solvent
reaction
time (h)
chromatogr.c
system
compd
R
R′
X
Y
mp (°C)
yield (%)
10p
10q
10r
10s
10t
tert-Bu
Ph
CF3
H
H
H
H
H
H
H
H
H
H
CH3
C2H5
H
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
N
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
N
198-200
233-235
235-236
208-210
219-221
183-185
232-234
204-205
120-122
227-229
155-157
142-144
130-132
106-108
oil
a
b
c
b
d
d
b
c
c
b
a
a
c
c
20
43
100
95
77
81
80
37
71
68
44
62
24
54
100
62
100
100
43
18
1.5
18
2.5
1.5
15
1
0.5
2
0.5
0.5
1
3.5
3.5
1.5
1
0.5
0.5
0.5
1
A
A
CN
NO2
NH2
pyrrol-1-yl
OTHP
OH
SCH3
Cl
Cl
Cl
Cl
tert-Bu
Ph
CF3
A
10u
10v
10w
10xd
10y
10z
10a ′
10b′
10c′
11a
11b
11c
11d
11eg
11f
11g
11h a
12a
12b
12c
12d
12e
12f
A
A
A
A
B
B
e
f
H
CH
117-119
44-46
165-167
150-152
oil
b
e
b
b
CN
NO2
OTHP
SCH3
Cl
tert-Bu
Ph
52
74
C
D
115-117
b
1
H
H
H
H
H
H
H
H
H
H
CH3
C2H5
H
H
H
H
H
220-222
262-264
228-230
194-196
203-205
255-257
242-243
240-242
231-233
a
f
27
84
77
44
59
45
70
44
74
0.5
1
CF3
CN
b
b
b
b
b
f
0.5
0.5
0.5
120
0.5
1
NO2
NH2
pyrrol-1-yl
OTHP
SCH3
Cl
Cl
Cl
tert-Bu
Ph
CF3
D
D
A
12g
12h
12i
f
0.5
12ja
12k
12l
oil
97
48
53
18
16
30
6.5
8
16
17
17
24
12
155-157
a
e
a
e
b
c
c
a
g
D
13a
13b
13c
13d
13e
13f
13g
13h
13ia
13j
65-67
130 dec
66-68
97
100
100
100
100
100
94
CN
177-179
162-164
168-170
103-105
149-150
NO2
pyrrol-1-yl
OTHP
SCH3
Cl
H
H
H
H
CH3
C2H5
Cl
Cl
oil
124-126
94
93
1.5
0.5
13k
e
a
b
Data of known derivatives 11h , 12j, 13i are reported in ref 34. Recrystallization solvents: (a) benzene/cyclohexane; (b) ethanol; (c)
benzene; (d) iso-propanol; (e) cyclohexane; (f) N,N′-dimethylformamide/water; (g) toluene. c Chromatographic system silica gel: (A) aluminum
d
oxide/ethyl acetate; (B) silica gel/ethyl acetate: (C) silica gel, n-hexane/ethyl ether 1:1; (D) aluminum oxide/chloroform. Nitrate salt, mp
87-89 °C (ethanol/ethyl ether). e Nitrate salt, mp 224-226 °C (iso-propanol/iso-propyl ether). f Nitrate salt, mp 226-227 °C (ethanol/
g
ethyl ether). Reference 44.
sized azoles 10p -v,x-c′ against 40 strains of Candida
albicans and against 3 strains of Candida spp. (two
Candida glabrata, one Candida krusei) at pH 7.2 and
5.8 are reported in Tables 3 and 4, respectively. Data
refer to MIC mean values, MIC50, MIC90, % R (percent-
age of resistant species at 256 µg/mL), and the MIC
range. With few exceptions, MIC50 and MIC values
parallel the related MIC90 data, which were chosen for
a preliminary comparison between the new derivatives
and reference drugs (miconazole, ketoconazole, bifon-
azole, and 10o34) and for QSAR studies. Antifungal
experiments were carried out at pH values reproducing
in the growth medium the systemic physiological condi-
tions (pH 7.2) and the mildly acidic conditions of the
skin in topical application (pH 5.8).
The most potent derivative at both pH 5.8 and 7.2
was imidazole 10a ′, which showed MIC90 ) 2 µg/mL (pH
) 5.8) comparable to the values of miconazole and
ketoconazole and was 2-fold more potent than bifonazole
(MIC90 ) 4 µg/mL).
A number of azole derivatives (10r , 10v, 10y, 10z,
10a ′) showed good MIC values ranging from 3.2 to 9.14
µg/mL (pH ) 7.2) and proved to be as potent as
miconazole, ketoconazole, and bifonazole (MIC ) 1.16-
9.26 µg/mL). The high potency of inhibition of the fungal
growth shown by derivatives 10z and 10a ′ at either pH
5.8 or pH 7.2 supported the importance of N-substitu-
tion of the pyrrole ring. The chlorine atom on the phenyl
group linked to the pyrrole moiety could be replaced
with pyrrol-1-yl, CF3, SCH3, and CN groups without
decreasing the antifungal activity. In contrast, the
introduction of OH, NH2, or tert-butyl substituents led
to inactive derivatives 10p , 10u , and 10x and replace-
The majority of the newly synthesized azole deriva-
tives inhibited the growth of all strains of C. albicans
used for antifungal assays. % R > 0 was observed for
compounds 10p , 10u , 10x, 10b′, and 10c′.

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2725
Ta ble 3. In Vitro Antifungal Activities of Imidazole Derivatives against C. albicans and Candida spp. (Two C. glabrata and One C.
krusei) at pH 7.2
fungi (no. of tested strains)
C. albicans (40)
Candida spp. (3)
MIC
MIC50
MIC90
range
MIC
MIC50
MIC90
range
compd
10p
10q
10r
10s
10t
10u
10v
10x
10y
10z
10a ′
10b′
10c′
10o
% R
(µg/mL)
(µg/mL)
(µg/mL)
(µg/mL)
% R
(µg/mL)
(µg/mL)
(µg/mL)
(µg/mL)
80
0
0
0
0
54
0
0
0
0
164.6
14.9
7.63
13.2
14.7
85
>256
>256
128 to >256
2-128
100
0
0
0
0
67
0
0
0
0
>256
8
>256
256
32
>256
8
4
8
8
8
8
32
8
90.6
17.3
32.4
56
256
14.13
256
66.6
6.1
13.3
8-256
4-32
4-64
16
16
32
2-64
64
1-32
2-256
128
>256
32
256
128
32
8-128
256 to >256
2-32
>256
>256
32-128
1-32
>256
4.4
2
16
4
2
4
4
77.8
9.14
5.3
3.2
128
164.6
2.14
1.16
2.44
9.26
256
8
16
2
>256
>256
4
16-256
4-64
256
64
16
256-256
8-128
4-16
1-32
0
1
>256
>256
2
0.5-16
128 to >256
128 to >256
0.5-32
<0.25-16
0.5-32
2-32
0
16
16
8-16
97
77
0
0
0
100
100
0
0
0
>256
>256
8
>256
>256
8
>256
>256
5.01
0.83
4.16
8.6
0.5-8
0.5-1
0.5-8
2-16
miconazole
bifonazole
ketoconazole
0.5
2
8
2
4
32
1
4
8
1
8
16
0
0
Ta ble 4. In Vitro Antifungal Activities of Imidazole Derivatives against C. albicans and Candida spp. (Two C. glabrata and One C.
krusei) at pH 5.8
fungi (no. of tested strains)
C. albicans (40)
Candida spp. (3)
MIC
MIC50
MIC90
range
MIC
MIC50
MIC90
range
compd
10p
10q
10r
10s
10t
10u
10v
10x
10y
10z
10a ′
10b′
10c′
10o
% R
(µg/mL)
(µg/mL)
(µg/mL)
(µg/mL)
% R
(µg/mL)
(µg/mL)
(µg/mL)
(µg/mL)
97
0
0
0
0
80
0
26
0
0
0
68
94
0
0
0
256
10.05
4.8
41.3
15.03
219.4
3.95
150
>256
>256
256 to >256
4-64
100
0
0
0
0
67
0
67
0
0
0
100
100
0
0
0
>256
16
>256
128
16
256
256
>256
16
>256
8
4
8
8
8
8
32
8
53.3
16
58
133.3
256
9.3
128
9.3
16-128
8-16
1-16
16
64
128
>256
8
4-64
8-256
2-128
16-256
256 to >256
4-16
>256
>256
128 to >256
0.25-64
32-256
1-32
2
32
4
4
128
8
2
2
>256
>256
>256
>256
128 to >256
4-16
7
8
2
4
16
64
8
2.54
2.36
209.4
128
0.5
1
>256
>256
1
0.5-4
4.5
5.33
1-64
0.25-32
128 to >256
128 to >256
1-8
4-8
>256
>256
8
>256
>256
64
8
32
128
>256
>256
3.6
2
2
4
2
6.0
0.156
4
4-64
miconazole
bifonazole
ketoconazole
0.57
5.85
6.35
0.5
2
8
<0.25-8
1-8
<0.25-32
0.5
4
32
<0.25-0.25
2-32
0
0
6.53
<0.25-16
ment of imidazole with triazole abated antimicotic
activity of 10o (see derivatives 10b′ and 10c′).
antimycotic activities of imidazole derivative 10o and
the newly synthesized azoles 10p -v,x-c′ were evalu-
ated, as in the case of the compounds used to develop
COMFA96, by means of the minimal inhibitory concen-
tration (MIC). The results produced by this preliminary
QSAR approach were not encouraging, since the activi-
ties of the newly synthesized molecules were mispre-
dicted by COMFA96, as clearly indicated by the low
value computed for the predictive r2 (0.52). The poor
predictive power of COMFA96, however, when applied
to 10o-v,x-c′ was not completely unexpected. It was
noted, in fact, that the new compounds represent a sort
of extrapolation with respect to those used to derive
COMFA96, since structural modifications have been
introduced with respect to previous compounds 10,
which were not fairly sampled in that model (see Chart
3). Because of these unsatisfactory results, we resorted
to another QSAR approach. Catalyst software37 was
preferred this time because of its peculiarities that
allowed us to consider each compound as a collection of
energetically reasonable conformations. Moreover, al-
In general, the activity against Candida albicans was
also confirmed against Candida spp. and the trends of
microbiological profiles of the test derivatives were very
similar. The most potent azole at both pH 5.8 and 7.2
was compound 10a ′, which showed the same activity of
miconazole (MIC90 ) 8 µg/mL at pH 5.8) and was 4- and
16-fold more potent than bifonazole and ketoconazole,
respectively.
Molecu la r Mod elin g Stu d ies. A CoMFA model was
derived on a series of antifungal agents (4, 7a -10n )
belonging to chemically diverse families related to
bifonazole (COMFA96 in the following).19 Considering
the structural similarities between the training and test
sets of COMFA96 (listed in Chart 3 and Table 5) and
antifungals 10o-v,x-c′, the opportunity was taken to
apply that model to perform the 3D-QSAR study of the
new products and of 10o, using them as an external test
set that could further assess the predictive power of
COMFA96. This experiment was allowed because the

2726 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
Ta ble 5. Number of Conformers, Enantiomer Used in the Fit
Operation, and Experimental and Calculated Anti-Candida
Activity [MIC90(cpd)/MIC90(bifonazole)] for All the Compounds
Used in This Study
The general paradigm considered for the QSAR study
of compounds 10o-v,x-c′ with Catalyst is the fol-
lowing. The pharmacophoric model was developed tak-
ing into consideration only compounds used to derive
COMFA96 and derivatives 10o-v,x-c′ constituted
again an external test set. In fact, a general pharma-
cophoric model, able to explain the structure-activity
relationships of azole antifungal agents and endowed
with predictive power, should not be strictly dependent
on the choice of the training set (as COMFA96 appar-
ently was). On the other hand, our choice gave us the
opportunity to perform a comparison between the
performances of two well-established 3D-QSAR meth-
odologies (CoMFA and Catalyst), updating at the same
time the previous CoMFA model. In summary, the
chosen procedure allowed the QSAR study of compounds
10o-v,x-c′ as well as the evaluation of the predictive
character and robustness of the new approach.
MIC90 (cpd)/MIC90 (bifonazole)
compd conformers enantiomer
experimental
predicted
4
7a
7b
7c
37
30
98
26
45
R
S
R
S
1.0
4.0
3.6
2.1
58
1.1
33
3.5
31
7d
S
29
7e
31
S
63
29
7f
61
S
63
30
7g
56
S
66
32
7h
7i
91
89
S
S
123
64
29
30
7j
7k
39
93
S
S
7.2
32
31
31
7l
7m
7n
7o
7p
107
28
44
208
205
66
S
S
8.0
32
7.3
21
62
32
29
30
26
30
12
30
1R,2S
1S,2S
1R,2S
S
To use the Catalyst methodology at its best, the
training set was redesigned: 19 among 56 compounds
were selected,19 including the most active compound 8c,
whose biological activity values spanned 3 orders of
magnitude (see Experimental Section).
7q
7r
56
S
63
30
7s
30
S
16
31
7t
56
S
29
30
7u
7v
8a
8b
8c
8d
8e
8f
8g
8h
9a
9b
9c
9d
9e
9f
51
60
32
16
32
59
15
20
36
20
94
76
81
99
80
44
114
131
70
61
106
49
24
60
63
186
171
37
50
152
35
1S,2S
1R,2R
S
R
S
63
30
15
To build the pharmacophore hypotheses, the following
chemical features were considered: aromatic nitrogen
with a lone pair in the ring plane (AWLP), hydrogen
bond acceptor (HBAlip), hydrophobic aliphatic (HydAl),
hydrophobic aromatic (HydAr), and ring aromatic (RA).39
Because of the molecules’ rigidity and structural com-
plexity, the hypothesis generator was constrained to
report hypotheses with at least three features and to
include AWLP in each pharmacophore to satisfy the key
interaction between azole inhibitors and the enzyme.
Azole compounds, in fact, inhibit the binding of the
natural substrate lanosterol to CYP51 by coordination
of the ring nitrogen atom (N-3 of imidazole and N-4 of
triazole) to the sixth coordination position of the iron
ion of the enzyme protoporphyrin system, as also
demonstrated by a recent literature report.15
10.5
0.25
2.2
0.13
0.70
1.0
0.50
0.50
2.5
4.1
1.0
3.0
1.8
0.50
1.9
1.8
4.0
2.0
0.89
1.0
0.46
0.90
1.0
41
0.11
1.6
0.15
0.81
0.87
0.66
0.094
1.1
4.0
1.4
5.7
4.9
3.4
1.3
2.4
3.0
8.8
1.6
1.5
1.1
0.91
1.4
29
S
S
R
R
R
R
S
S
R
R
S
R
R
R
S
R
R
R
R
R
R
R
S
R
R
R
S
S
S
S
R
S
S
S
S
9g
9h
9i
9j
9k
The 18.7 cost range over the 10 best hypotheses
generated by Catalyst indicated the existence of a
homogeneous set of hypotheses due to a strong signal
generated by the chosen training set. Composition (in
terms of chemical features necessary for activity),
ranking score, and statistical parameters associated
with these pharmacophore hypotheses are reported in
Table 6. Because the highest scoring hypothesis is
usually the most likely to yield relevant information
about the pharmacophore elements of a set of com-
pounds and because of the fact that all the better
hypotheses had identical composition, hypothesis 1,
characterized by the best statistical parameters, was
chosen to represent the “pharmacophore model” and was
selected for further evaluation (HYPO1 in the following).
10a
10b
10c
10d
10e
10f
10j
10k
10l
10m
10o
10p
10q
10r
10s
10t
10u
10v
10x
10y
10z
10a ′
10b′
10c′
22
40
9.0
20
6.0
1.0
29
3.8
4.0
42
4.0
0.80
47
5.6
0.70
6.3
1.5
52
0.72
26
3.0
3.0
2.9
49
8.2
1.6
4.7
4.0
1.5
11
0.90
9.6
0.98
8.1
6.9
3.5
13
88
81
167
162
203
132
94
102
96
147
91
In Figure 1, HYPO1 is shown, with the S enantiomer
of 8c (the most active compound) superimposed, as a
four-feature hypothesis characterized by one aromatic
nitrogen with a lone pair (AWLP) and three aromatic
rings (RA1, RA2, and RA3), two of which are close to
each other and are coplanar (RA2 and RA3). Activity of
8c was correctly estimated by HYPO1 because of the
fact that a complete mapping of this molecule onto the
pharmacophore was possible. In detail, because of the
coplanarity of the phenyl and pyrrole rings of the bulky
R
R
R
S
162
89
73
30
30
S
49
ternative stereoisomers were automatically generated
and considered by the software, since the chirality of
the asymmetric center was not specified (in agreement
with the synthetic pathways).38

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2727
Ta ble 6. Composition (Features), Cost (Bits), and Statistical Parameters (rms and Correlation) Associated with the 10 Best
Hypotheses (Pharmacophore Models)
hypothesis
feature 1a
feature 2a
feature 3a
feature 4a
costb
rms
correlation
1
2
3
4
5
6
7
8
9
AWLP
AWLP
AWLP
AWLP
AWLP
AWLP
AWLP
AWLP
AWLP
AWLP
RA1
RA1
RA1
RA1
RA1
RA1
RA1
RA1
RA1
RA1
RA2
RA2
RA2
RA2
RA2
RA2
RA2
RA2
RA2
RA2
RA3
RA3
RA3
RA3
RA3
RA3
RA3
RA3
RA3
RA3
74.12
74.18
75.49
75.85
77.01
77.76
84.20
85.11
90.72
92.86
1.08
1.09
1.18
1.17
1.23
1.18
1.50
1.54
1.73
1.80
0.944
0.943
0.932
0.934
0.927
0.936
0.887
0.880
0.845
0.832
10
a
b
AWLP: aromatic nitrogen with lone pair feature. RA: ring aromatic feature. Expressed in bits.
F igu r e 1. Compounds (S)-8c (in yellow), the most active of
the whole set, and (R)-4 (in red), the reference compound,
mapped to the pharmacophore hypothesis HYPO1. Pharma-
cophore features are color-coded: red for an aromatic nitrogen
with lone pair (AWLP) and orange for aromatic ring (RA1,
RA2, and RA3).
F igu r e 2. Regression line for HYPO1 (the pharmacophore
model). Experimental versus estimated (or predicted) anti-
Candida activity is reported for each member of the training
set, of the test set, and for the new derivatives 10o-c′.
and the null hypothesis was 80 bits, corresponding to
an about 90% chance of true correlation in the data.
Although one chiral center is present in all the
molecules of the training set, HYPO1 did not exhibit a
univocal preference for one enantiomeric series of
inhibitors relative to the other. On the contrary, as
reported in Table 5, each time a compound was super-
imposed on HYPO1, either the R or S enantiomer gave
the best mapping according to an apparently random
pattern. In the case of the reference antifungal drug 4
(bifonazole), whose best fitting onto HYPO1 is depicted
in Figure 1, the R enantiomer showed the highest
correlation between experimental and calculated activ-
ity values (1.1 estimated vs 1.0 experimental), with only
three chemical features of the hypothesis, out of four,
matched by the chemical groups of the ligand. The
biphenyl group (bulky substituent) was found to cor-
respond to RA1, the phenyl ring to RA2, and the N-3
nitrogen atom of the imidazole ring to the AWLP
feature. Notably, (S)-4 showed a conformation whose
orientation on HYPO1 was similar to that depicted in
Figure 1 for 8c. This mapping gave a slightly worse
activity correlation (1.2 estimated vs 1.0 experimental)
with respect to the R enantiomer because of a worse
superimposition of the chemical groups of the ligand
with the pharmacophore. Notably, in a recent study by
some of us devoted to the homochiral synthesis and
4-(pyrrol-1-yl)phenyl moiety, both RA2 and RA3 were
mapped. Furthermore, RA1 was mapped by 8c’s 4-
methylphenyl ring and, finally, the N-3 nitrogen atom
of the imidazole ring corresponded to the AWLP fea-
ture of HYPO1. The strong inhibitory activity of 8c,
[MIC90(8c)/MIC90(4) ) 0.13] suggests that it possesses
many or all of the molecular features important for
antifungal activity. Taking into account that the phar-
macophore model correctly estimated the inhibitory
properties of this compound (MIC90 ) 0.15 relative to
bifonazole), one may conclude that HYPO1 possibly
accounts for relevant interactions between inhibitors
and enzyme.
All the calculated activities from HYPO1, and the
number of generated conformers for each studied mol-
ecule, are listed in Table 5, while the regression line of
experimental versus estimated anti-Candida activities
for the training set, exhibiting a correlation coefficient
r ) 0.94 and a root-mean-square deviation rms ) 1.08,
is shown in Figure 2. Comparison between the esti-
mated activities of the compounds in the training set
and their experimental values was, in the worst case,
inferior to a 5-fold difference and, in most cases, inferior
to a 2-fold difference, indicating the reliability of the
model. Moreover, the cost difference between HYPO1

2728 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
biological evaluation of bifonazole,40 no differential
activity between the two enantiomers of this compound
was observed, and the present computational analysis
confirms that unexpected result.
From an analysis of the fitting mode of the training
set derivatives onto the best hypothesis, it resulted that
all the ligands mapped the AWLP feature of the model,
in accordance to their generally accepted mechanism of
action. In general (except in the case of 8c), compounds
belonging to both classes 8 and 9 mapped three features
missing RA3 (4, 8h , and 9a ) or RA2 (9f and 9j).
Compounds belonging to class 7, as well as those of class
10, showed the following mapping on HYPO1: 7b, 7v,
10a , 10c, 10e, 10j, and 10k matched AWLP, RA1, and
RA2; 7g, 7h , 7j, 7q, 7t, and 10l matched only two
features, missing two of the three RA’s. Consequently,
the three-dimensional structural properties required for
an ideal CYP51A1 inhibitor can be summarized as
follows: (i) an aromatic nitrogen with an accessible lone
pair is strictly required (feature interacting with the
iron atom of the enzyme protoporphyrin system); (ii) a
diarylmethyl moiety on the azole N-1, which is preferred
over a 1,2-diarylethyl; (iii) one of the two aryls on the
methyl group could be better a phenyl ring with a para
lypophilic substituent; (iiii) the second aryl group should
be made up of two coplanar aromatic rings as in the
case of 8c.
F igu r e 3. Compound 10p (in yellow) and 10z (in red) mapped
to HYPO1. Pharmacophore features are color-coded: red for
an aromatic nitrogen with lone pair (AWLP) and orange for
aromatic ring (RA1, RA2, and RA3).
of the substituted pyrrole moiety to the inhibitory
activity, as in the case of molecules in classes 9 and 10.
In particular, compounds in which the pyrrole ring has
an aromatic substituent in position 4 had no conforma-
tions able to map RA3.
After construction and before its application to the
QSAR analysis of compounds 10o-v,x-c′, the Cat-
alyst pharmacophore was used to predict the anti-
Candida activity values of the inhibitors used to derive
COMFA96 not already included in the training set (see
Experimental Section). These predictions, when com-
pared with the corresponding observed values (see Table
5), generated a regression line that exhibited a correla-
tion coefficient r ) 0.84 and a root-mean-square devia-
tion rms ) 1.43 (see Figure 2). The activities of all
compounds in the test set were predicted by the model
to be within 0.88 log units of their actual inhibitory
activities (excluding 7a and 7c, which showed residuals
of 0.92 and 1.17 log units, respectively). In particular,
the activity of (S)-8a (most active compound in the test
set) against Candida albicans was predicted by the
model to be 0.11, in good agreement with the experi-
mentally determined value of 0.25 (residual 0.36 log
units). This result accounted for the mapping of (S)-8a
with all the features of the pharmacophore: the N-3 of
the imidazole ring and the phenyl group mapped AWLP
and RA1, respectively, while RA2 and RA3 were rep-
resented respectively by the phenyl ring and by the
pyrrole ring of the bulky aromatic moiety. It is interest-
ing to note that the R enantiomer was found to map
only three features (namely, AWLP, RA1, and RA2) and
consequently was predicted to be much less active (1.3
vs 0.25).
HYPO1 was than applied to perform the QSAR study
of the compounds 10o-v,x-c′. Accordingly, their anti-
Candida activity values were predicted (see Table 5 and
Figure 2) and the predicted-versus-experimental regres-
sion line was analyzed. The cited derivatives exhibited
a correlation coefficient r ) 0.78 and a root-mean-square
deviation rms ) 1.54 that were judged to be satisfactory.
The activities of all compounds in this external test set
were predicted by the model to be within 0.73 log units
of their actual values (excluding 10p , which showed a
residual of 1.00 log unit). None of the new compounds
fulfilled all four pharmacophoric features, and univocal
preference for one enantiomeric series relative to the
other was not exhibited (see Table 5) even though S
enantiomers of the more active inhibitors were consid-
ered by the software. Compounds 10o, 10p , 10q, 10r ,
10t, 10x, and 10v, notwithstanding their different
activity values, were superimposed by the program on
HYPO1 in a similar manner. The N-3 nitrogen atom of
the imidazole and the 2,4-Cl2-phenyl substituent were
placed on the AWLP and RA2 features, respectively,
while the pyrrole group of the bulky aromatic substitu-
ent mapped RA1. The activities of 10o, 10q, 10r , 10t,
10x, and 10v were predicted to be in good agreement
with the experimental values (1.6 vs 0.80, 4.0 vs 5.6,
1.5 vs 0.70, 0.90 vs 1.5, 8.1 vs 26, and 0.98 vs 0.72,
respectively), while in the case of ligand 10p , a value
of 4.7 was calculated versus an actual activity of 47. The
poor agreement obtained in the case of 10p , however,
could be reasonably rationalized by assigning the low
experimental activity shown by this derivative to steric
HYPO1 accounted for the trend of the biological data
associated with class 8 and class 7 (derivatives homolo-
gous to the 8’s with a methylene spacer) compounds in
the test set. The markedly decreased inhibitory activity
of compounds 7 was ascribed as being due to the
unmatching of pharmacophoric features; 7n and 7p
missed the mapping of RA3, while all other compounds
were completely unable to match both RA2 and RA3
features. The model also accounted for the contribution
hindrance problems (not well understood by Catalyst41
)
of the R group (tert-butyl).
Inhibitors 10s, 10u , 10y, 10z, and 10a ′ were charac-
terized by an orientation into the pharmacophore model
very similar to that of 8c and by the molecular frag-

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2729
ments fulfilling all but one of the features of HYPO1
(Figure 3). In fact, the N-3 nitrogen atom of the
imidazole ring mapped AWLP, and the 2,4-Cl2-phenyl
substituent mapped RA1 and the pyrrole RA2. Predicted
activities of these compounds, as shown in Table 5, were
in reasonable agreement with the experimentally de-
termined values. In the case of 10u , the derivative
showed the second worst connection between experi-
mental and calculated activities (9.6 vs 52); the devia-
tion found might be due to the presence of a NH2 group
(R1), which, decreasing the lypophilic characteristics of
the molecule, could prevent the crossing of the mem-
brane of the fungi cell.
compounds 8a , 8c, and 8g were able to match all four
features of HYPO1 because of the structural character-
istics peculiar to these three compounds (possessing a
completely planar phenylpyrrolyl aromatic moiety).
Many compounds, showing intermediate activities, could
match three features, and finally the less active inhibi-
tors were found to match two features only, since only
conformations poorly fitting the enzyme were found in
the considered energy window.
One aromatic nitrogen with lone pair (mapped by all
of the considered compounds) plus three aromatic ring
features were recognized by HYPO1 to have pharma-
cophoric relevance, whereas no hydrogen bond acceptor,
hydrophobic aliphatic features, and hydrophobic aro-
matic features were found. These findings confirmed
that the key interaction of azole antifungines with the
demethylase enzyme is the coordination bond with the
iron ion of the porphyrine system. On the other hand,
the detection of aromatic interactions only, together
with the missing of hydrogen bonds, could be interpreted
as due to the poor structural diversity of the studied
compounds. On the contrary, because the features
constituting HYPO1 were sufficient to reproduce in a
satisfactory manner the activities of all the compounds
in the training and test sets and to predict those of 10o-
v,x-c′ (about 65 compounds), we believe this pharma-
cophore might reflect the structural characteristics of
the cavity surrounding the prostethic group in the
involved cytochrome. Accordingly, and as a rule, aro-
matic interactions with amino acids localized in proxim-
ity of heme (described by the features RA1, RA2, and
RA3) could be responsible of the different activities of
diverse antifungal agents, while hydrogen bonds or
similar hydrophilic interactions with farther residues
(not pointed out in this study) could fortuitously increase
the affinity of an inhibitor with respect to another one.
Recently, a paper concerning the crystal structure of
cytochrome P450 14R-sterol demethylase from Myco-
bacterium tuberculosis in complex with azole inhibitors
has appeared in the literature, which indirectly supports
our results. In that paper it is reported that “azole
resistance in fungi develops in the protein region
involved in orchestrating passage of the enzyme through
the different conformational stages along the catalytic
cycle rather than in residues directly contacting inhibi-
tors” (fluconazole in that case).15 Moreover, the active-
site chamber dome in the case of fluconazole is reported
to show prevalently phenylalanine residues that con-
stitute the ceiling above the porphyrin plane (F78, M79,
F83, and F255) and interact with the inhibitor. Finally,
in that paper, none of the nonbonded interactions
detected between the inhibitors and the enzyme is
reported to play a major role in determining the binding.
All these results seem to be highly compatible with
HYPO1.
As expected, and because of their low activity, triazole
compounds 10b′ and 10c′ matched only two features of
HYPO1, AWLP, and RA1 (with the 2,4-Cl2-phenyl
substituent), so the importance of the nature of the azole
ring for the activity was demonstrated.
A further analysis of HYPO1 was devoted to under-
stand the role played by the pyrrole ring substituents
(R and R1) on the biological activity of the compounds.
The effect of these groups seemed to be related to the
influence they exert on the overall 3D shape of the
molecules. In the homologous series 10o, 10z, and 10a ′,
for instance, the substitution of the hydrogen atom (10o)
at the 1-position of the pyrrole ring with an alkyl group
(R1 ) Me or Et) forced compounds 10z and 10a ′ to
switch the enantiomer and the orientation mapped onto
HYPO1. Particularly, in the case of 10a ′, Catalyst was
unable to find a conformation (in the energetic window
of 20 kcal/mol)42 that could give a fit similar to that of
10o. As a consequence, the match of 10a ′ onto HYPO1
was slightly worse, as demonstrated by the predicted
activity value of 15. The R substituent seemed to be
responsible for the fine-tuning of the activity of the
compounds 10q, 10r , 10s, 10t, 10v, and 10y, since small
differences between the shape of putative bioactive
conformers produce slight variations in the super-
position onto HYPO1.
Con clu sion s
As a summary of this QSAR study, the pharmaco-
phore model developed by Catalyst and named HYPO1
showed good performance in predicting the biological
activity data of a set of inhibitors belonging to the
1-[(aryl)[4-aryl-1H-pyrrol-3-yl]methyl]-1H-imidazole class.
HYPO1 did not exhibit an unequivocal preference for
one enantiomeric series of inhibitors relative to the
other. On the contrary, an apparently random pattern
was found. Notably, this result, even though surprising
and unexpected, appears to be in agreement with recent
experimental findings.40
The pharmacophore model postulates that there is an
essential three-dimensional arrangement of functional
groups that each molecule must possess to strongly
inhibit the enzyme. The modulation of the activities of
the compounds considered in this study resulted in a
dependence mainly on the number of pharmacophoric
features mapped by each one of them. Hence, our model
provides both crucial information about how well the
chemical features of a subject molecule overlap with the
hypothesis and the ability of molecules to adjust their
conformations in order to fit the enzyme with energeti-
cally reasonable conformations; only the most active
In conclusion, HYPO1 conveys important information
in an intuitive manner and can provide predictive
capability for evaluating new compounds.
Exp er im en ta l Section
1. Ch em istr y. Melting points were determined on a Bu¨chi
530 melting point apparatus and are uncorrected. Infrared (IR)
spectra (Nujol mulls) were recorded on a Perkin-Elmer 297
1
spectrophotometer. H NMR spectra of imidazole derivatives
were recorded at 200 MHz on a Bruker AC 200 spectrometer

2730 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
using tetramethylsilane (Me4Si) as the internal reference
standard. Column chromatographies were performed on alu-
mina (Merck; 70-230 mesh) or silica gel (Merck; 70-230
mesh) column. All compounds were routinely checked by TLC
by using aluminum-baked silica gel plates (Fluka DC-Alufolien
Kieselgel 60 F254) and aluminum-baked aluminum oxide plates
(Fluka DC-Alufolien Kieselgel 60 F254). Developed plates were
visualized by UV light. Solvents were reagent grade and, when
necessary, were purified and dried by standard methods.
Concentration of solutions after reactions and extractions
involved the use of a rotary evaporator (Bu¨chi) operating at a
reduced pressure (ca. 20 Torr). Organic solutions were dried
over anhydrous sodium sulfate. Analytical results agreed to
within (0.40% of the theoretical values. All compounds were
analyzed for C, H, N, and, when present, S, Cl, and F.
(2,4-Dich lor op h en yl)-[4-[4-(1H-p yr r ol-1-yl)p h en yl]-1H-
p yr r ol-3-yl]m eth a n on e (12g). A solution of amine 12f (1.0
g, 3.0 mmol) and 2,5-dimethoxytetrahydrofuran (410 mg, 3.1
mmol) in glacial acetic acid (20 mL) was refluxed for 0.5 h.
After the mixture was cooled, the solvent was removed and
the residue was chromatographed on aluminum oxide (ethyl
acetate as eluent) to give pure 12g (810 mg, 70% yield).
r-(2,4-Dich lor op h en yl)-4-(4-t r iflu or om et h ylp h en yl)-
1H-p yr r ole-3-m eth a n ol (13c). A mixture of 12c (3.0 g, 7.8
mmol) and sodium borohydride (880 mg, 23 mmol) in a
tetrahydrofuran/water mixture (200:0.7 mL/mL) was refluxed
for 8 h. After cooling, the suspension was treated with water
(200 mL) and concentrated under reduced pressure. The
residue was extracted with ethyl acetate (3 × 100 mL), and
the organic extracts were collected, washed with brine, and
dried. Evaporation of the solvent gave 3.0 g (100% yield) of
13c, which was used in the following reaction without further
purification.
1-[(2,4-Dich lor op h en yl)[4-(4-t r iflu or om et h ylp h en yl)-
1H-p yr r ol-3-yl]m eth yl]-1H-im id a zole (10r ). A solution of
13c (3.1 g, 8.0 mmol) and 1,1′-carbonyldiimidazole (5.5 g, 34
mmol) in anhydrous acetonitrile (130 mL) was stirred at room
temperature for 18 h. The solvent was removed, and the
residue was dissolved in ethyl acetate (150 mL). The organic
solution was washed with brine (3 × 100 mL) and the solvent
was evaporated to afford the pure imidazole derivative 10r
(3.9 g, 100% yield).
1-[[4-(4-Am in oph en yl)-1H-pyr r ol-3-yl](2,4-dich lor oph en -
yl)m eth yl]-1H-im id a zole (10u ). The nitro derivative 10t
(600 mg, 1.5 mmol), dissolved in ethyl acetate (150 mL), was
hydrogenated in the presence of 10% palladium on charcoal
(240 mg) as a catalyst for 15 h in a Parr apparatus at room
temperature at an initial pressure of 4 atm. The catalyst was
then filtered off and the filtrate was evaporated to give 450
mg of 10u (81% yield).
2. Syn th eses. Specific examples presented below illustrate
the general synthetic procedures.
1,3-Dia r yl-2-p r op en on es 11. These compounds were pre-
pared according to reported procedures.43 Chemical and physi-
cal data of newly synthesized derivatives 11a -d ,f,g are re-
ported in Table 2 together with those of the known propenones
11e44 and 11h .34 11f was obtained as reported below starting
from 4-(tetrahydropyran-2-yloxy)benzaldehyde, which was
synthesized as reported elsewhere.45
1-(2,4-Dich lor op h en yl)-3-(4-tetr a h yd r op yr a n -2-yloxy-
p h en yl)eth a n on e (11f). A mixture of 4-(tetrahydropyran-2-
yloxy)benzaldehyde (6.9 g, 33.5 mmol) and 1-(2,4-dichloro-
phenyl)ethanone (6.3 g, 33.5 mmol) in methanol (180 mL) was
treated with barium hydroxide octahydrate (15.8 g, 50 mmol)
and heated at 50 °C for 1 h. Then the solvent was removed
under reduced pressure and the residue was treated with
water (200 mL) and was extracted with ethyl acetate (3 × 150
mL). The organic extracts were collected, washed with brine
(2 × 150 mL), dried, and stripped of the solvent. The residue
was chromatographed on a silica gel column (n-hexane/ethyl
ether 1:1) to afford pure 11f (6.6 g, 52% yield).
1-[(2,4-Dich lor op h en yl)[4-(4-h yd r oxyp h en yl)-1H-p yr -
r ol-3-yl]m eth yl]-1H-im id a zole (10x). A suspension of tetra-
hydropyranyl derivative 10w (290 mg, 6.2 mmol) in methanol
(50 mL) was treated with a catalytic amount of p-toluene-
sulfonic acid, and the mixture was stirred at room temperature
for 2 h. After this time the solvent was removed and the
residue was dissolved in ethyl acetate (50 mL). The organic
phase was washed with 1 N sodium carbonate (3 × 50 mL),
then was washed with brine (3 × 50 mL), and dried, and the
solvent was evaporated under reduced pressure. The residue
was chromatographed on aluminum oxide (ethyl acetate as
eluent) to furnish pure 10x (170 mg, 71% yield).
[4-(4-Bip h en ylyl)-1H-p yr r ol-3-yl]-(2,4-d ich lor op h en yl)-
m eth a n on e (12b). A solution of 11b (5.0 g, 14.3 mmol) and
toluene-4-sulfonylmethylisocyanide (2.8 g, 14.3 mmol) in a
mixture of anhydrous dimethyl sulfoxide/ethyl ether (45:90
mL) was added dropwise into a well-stirred suspension of
sodium hydride (60% in white oil; 2.6 g, 65 mmol) in dry ethyl
ether (75 mL). After the addition, the mixture was stirred at
room temperature for 40 min and then treated with water (300
mL). The formed precipitate was filtered off, and the filtrate
was washed with cold ethanol and then with light petroleum
ether to afford 4.7 g (84% yield) of pure 12b.
3. Mycology. An ti-Ca n d id a in Vitr o Assa ys. Derivatives
10p -v,x-c′ were tested for antimycotic activities against C.
albicans and Candida spp. The antifungal potency was evalu-
ated by means of the minimal inhibitory concentration (MIC)
using the serial dilution test in a liquid nutrient medium. MIC
was defined as the lowest concentration of test substance at
which there was no visible growth in comparison with a blank
experiment after the preset incubation time. For the prepara-
tion of the dilution series, 5 mg of substance was dissolved in
DMSO (1 mL) and the solution was treated on shaking with
distilled water (9 mL). Further progressive double dilutions
with test medium furnished the required concentrations in the
range from 0.25 to 256 µg/mL. In some cases dissolution was
completed by addition of a few drops of diluted hydrochloric
acid. Blanks were prepared in the test medium with the above-
reported quantities of water and DMSO, without adding test
substance. All tested microorganisms were preliminarily in-
cubated at 37 °C for 18 h on Sabouraud (BBL) dextrose broth
and then added to media containing the antimycotic agent.
Antimicrobial tests were performed on Sabouraud broth (Difco)
using inocula of 103/mL of fungi. Readings of MICs were taken
at 24 h of growth at 37 °C. MIC values were calculated by the
expression MIC ) (∑MICi)/st, where MICi is the minimal
inhibitory concentration values of all strains at the used
concentration Ci and st is the total number of strains. MIC50
and MIC90 refer to MIC for 50% and 90% of strains, respec-
tively. Strains with MIC greater than 256 µg/mL are regarded
[4-(4-Chlorophenyl)-1-ethyl-1H-pyrrol-3-yl]-(2,4-dichloro-
p h en yl)m eth a n on e (12l). A well-stirred suspension of 12j
(4.5 g, 12.7 mmol), iodoethane (14.2 g, 91 mmol), and anhy-
drous potassium carbonate (3.5 g, 25.4 mmol) in dry N,N-
dimethylformamide (20 mL) was heated at 90 °C for 16 h. After
cooling, the mixture was treated with water (100 mL) and the
formed precipitate was extracted with ethyl acetate (3 × 70
mL). The collected organic extracts were washed with brine
(3 × 200 mL) and dried and the solvent was evaporated to
afford a crude product that was chromatographed on an
aluminum oxide column (chloroform as eluent) to give pure
12l (2.3 g, 48% yield).
[4-(4-Am in op h en yl)-1H-p yr r ol-3-yl]-(2,4-d ich lor op h en -
yl)m eth a n on e (12f). The nitro derivative 12e (3.0 g, 8.3
mmol) was added portionwise to a well-stirred solution of
tin(II) chloride dihydrate (6.6 g, 29.4 mmol) in a concentrated
hydrochloric acid/ethanol mixture (7:60 mL). The resulting
suspension was stirred at room temperature for 120 h. After
this time, the acidic solution was diluted with water (150 mL)
and then was treated with 1 N NaOH until pH 8 was attained.
The inorganic precipitate was filtered off, and the filtrate was
concentrated under reduced pressure. The residue was ex-
tracted with ethyl acetate (3 × 100 mL), and the collected
organic extracts were washed with brine (3 × 200 mL), dried,
and stripped of the solvent. The residue was chromatographed
on aluminum oxide to afford pure 12f (1.2 g, 45% yield).

Antifungal Agents
J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13 2731
as resistant (R) and are expressed as percentage by the
equation R(%) ) (Nt - Ns)/Nt × 100, where Nt is the total
number of tested strains and Ns is the number of sensitive
strains. Experiments were carried out at both pH 7.2 and 5.8
employing two different lots of C. albicans and Candida spp.
freshly isolated from hospitalized patients (strains were
identified using standard methods). The specimens used were
40 strains of C. albicans and 3 strains of Candida spp. (two
C. glabrata and one C. krusei).
4.2. P h a r m a cop h or e Descr ip tion . As a result of the
pharmacophore generation, 61.4 bits as the total fixed cost
(ideal hypothesis) and 153.8 bits as the cost of the null
hypothesis were found. The cost range over the 10 best
generated hypotheses was 18.7 bits (Table 6), suggesting that
there was a homogeneous set of hypotheses and that the signal
generated by this training set was strong. Moreover, the fact
that the hypotheses were much closer to the fixed cost (i.e.,
the cost of hypothesis 1 is 74.1) meant that the signal could
be interpreted.
4. Molecu la r Mod elin g. 4.1. P h a r m a cop h or e Gen er a -
tion . The training set for the pharmacophore development was
chosen according to the Catalyst guidelines.41 In particular,
to enhance the probability of finding a significant model, two
rules were applied: (i) inclusion of the most active compound
(8c) in the set because of the fact that Catalyst pays particular
attention to this molecule when it generates the chemical
features space; (ii) maximization of the kinds and relative
positions (substitution pattern) of the chemical features shared
by the molecules because the program recognizes the molecules
as collections of chemical features and not as assemblies of
atoms or bonds. Accordingly, 4, 8c, 8h , 7b, 7g, 7h , 7j, 7q, 7t,
7v, 9a , 9f, 9j, 10a , 10c, 10e, 10j, 10k , and 10l were selected
as the training set. Their biological activity (MIC90) values
span 3 orders of magnitude, with 8c (the most active deriva-
tive) being 1024 times more active than 7h (the less active
compound). The activity data were expressed, after having con-
verted all the µg/mL values to µmol/mL units, as [MIC90(cpd)/
MIC90(bifonazole)] in the range between 0.13 and 123 and are
reported in the fourth column of Table 5.
4.3. P h a r m a cop h or e Va lid a tion . F isch er Test. Despite
the quite high range found between the costs for an ideal
versus null hypotheses (92 bits) and the good difference
between the costs of HYPO1 and of null hypothesis (79 bits)
ensuring a 90% chance of true correlation,41 special care was
taken to test the model for chance correlation. The pharma-
cophore model corresponding to HYPO1 was evaluated for
statistical significance using a randomization trial procedure
derived from the Fischer method.47 Experimental activities of
the training set were scrambled 19 times to obtain spread-
sheets with randomized activity data. A total of 19 hypothesis
generation experiments were performed using the scrambled
training sets, and among the 190 resulting hypotheses (10 for
each computational run), none was found with lower cost than
hypothesis 1. Thus, there was at least a 95% chance that
hypothesis 1 represented true correlation in the data.
Ack n ow led gm en t. Thanks are due to Ministero
della Sanita`, Istituto Superiore di Sanita`, Progetto AIDS
1999 (Grant No. 40C.8) and to Italian CNR and MURST
(40%) for partial support. Microbiological tests were
performed by “Istituto di Microbiologia” of Rome Uni-
versity “La Sapienza”. Maurizio Botta thanks the Italian
Research National Council (CNR target project on
“Biotechnology”) and the Merck Research Laboratories
for the 2001 Academic Development Program (ADP)
Chemistry Award. Elemental analyses were performed
by Dott M. Zancato, University of Padova, Italy.
Compounds 10g, 10h , 10i, and 10n , bearing a naphthyl as
the R group and used in COMFA96, were excluded this time
by both training and test sets because their low activities had
been already assigned as being due to excluded volume
problems,19 and this peculiarity is reported to be not well
understood by Catalyst.41 In other words, these molecules
might have the naphthyl functionality that bumps into a
portion of the enzyme, preventing close contacts of the binding
functions. This assumption is strengthened by the observation
that derivatives 10i and 10n , both possessing a methyl group
that generates further steric hindrance, show the lowest
activities in the subset of naphthyl compounds. From another
viewpoint, because Catalyst pays particular attention to the
most active compounds in the training set, excluding such
inactive compounds may not generally affect the generation
of the chemical feature space relevant to the experiment.
Refer en ces
(1) D’Arcy, P. F.; Scott, E. M. Antifungal Agents. Prog. Drug Res.
1987, 22, 94-147.
(2) Kerridge, D. Antifungal Drugs. Drugs Today 1988, 24, 705-
715.
(3) Koltin, Y. Target for Antifungal Drug Discovery. Annu. Rep.
Med. Chem. 1990, 25, 141-148.
Because no experimental data on the biologically relevant
conformations of the selected compounds (for example, atomic
coordinates derived from X-ray crystallographic studies of their
complexes with the CYP51A1) are available, we resorted to a
molecular mechanics approach to build the conformational
populations to be used for pharmacophore generation. By
means of the Catalyst 2D-3D sketcher, a local minimum-
energy conformer of each compound of the training set was
built and submitted to conformational search using the “best”
procedure with the aim of collecting a representative set of
conformers chosen within a range of energetically reasonable
conformations (CHARMm force field).42,46 In particular, all the
conformers of each compound, within a range of 20 kcal/mol
with respect to the global minimum, were employed to build
a set of pharmacophore hypotheses comprising all/any of the
following chemical features: aromatic nitrogen with lone pair
(AWLP), hydrogen bond acceptors (HBAlip), hydrophobic
aliphatics (HydAl), hydrophobic aromatics (HydAr), and ring
aromatics (RA). Because of the molecules’ rigidity and struc-
tural complexity, the hypothesis generator was constrained to
report hypotheses with at least three features and to include
an aromatic nitrogen with lone pair group in each pharma-
cophore to satisfy the key interaction between azole inhibitors
and the enzyme. Azole compounds, in fact, inhibit the binding
of the natural substrate lanosterol to CYP51A1 by coordination
of the ring nitrogen atom (N-3 of imidazole and N-4 of triazole)
to the sixth coordination position of the iron atom of the
enzyme protoporphyrin system, as also demonstrated by a
recent literature report.15
(4) Fromtling, R. A. Imidazoles as Medically Important Antifungal
Agents: An Overview. Drugs Today 1986, 20, 235-349.
(5) Vanden Bossche, H.; Bellens, D.; Cools, W.; Gorrens, P. M.;
Verhoeven, H.; Willemsens, G.; De Coster, R.; Beerens, D.;
Haelterman, C.; Coene, M. C.; Lauwers, W.; Le J eune, L.
Cytochrome P450: Target for Itraconazole. Drug Dev. Res. 1986,
8, 287-298.
(6) Hichcock, A.; Brown, B.; Evans, E. G. V.; Adams, D. J . Cyto-
chrome P-450-Dependent 14R-Demethylation of Lanosterol in
Candida albicans. Biochem. J . 1989, 260, 549-556.
(7) Asai, K.; Tsuchimori, N.; Okonogi, K.; Perfect, J . R.; Gotoh, O.;
Yoshida, Y. Formation of Azole-Resistant Candida albicans by
Mutation of Sterol 14-Demethylase P450. Antimicrob. Agents
Chemother. 1999, 43, 1163-1169.
(8) Adams, J . L.; Metcalf, B. W. Therapeutic Consequences of the
Inhibition of Sterol Metabolism. In Comprehensive Medicinal
Chemistry; Hansch, C., Sammes, P. G., Taylor, J . B., Eds.;
Pergamon Press: Oxford, England, 1990; Vol. 2, pp 333-364.
(9) Yeagle, P. L.; Martin, R. B.; Lala, A. K.; Lin, H.; Block, K.
Differential Effects of Cholesterol and Lanosterol on Artificial
Membranes. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 4924-4926.
(10) Dawson, J . H.; Sono, M. Cytochrome P-450 and Chloroperoxi-
dase: Thiolate-Ligated Heme Enzymes. Spectroscopic Determi-
nation of Their Active Site Structures and Mechanistic Impli-
cations of Thiolate Ligation. Chem. Rev. 1987, 87, 1255-1276.
(11) Ruckpaul, K.; J anig, G. R.; Pfeil, D.; Honeck, H.; Blanck, J .;
Smettan, G.; Rein, H.; Ristau, O.; J ung, C. Cytochrome P-450-
Structure and Function. Pharmazie 1978, 33, 307-331.
(12) Poulos, T. L.; Finzel, B. C.; Hovvard, A. J . High-Resolution
Crystal Structure of Cytochrome P450cam. J . Mol. Biol. 1987,
195, 687-700.
(13) Cupp-Vickery, J . R.; Poulos, T. L. Structure of Cytochrome
P450eryF Involved in Erythromycin Biosynthesis. Nat. Struct.
Biol. 1995, 2, 144-153.

2732 J ournal of Medicinal Chemistry, 2002, Vol. 45, No. 13
Tafi et al.
(14) Hasemann, C. A.; Ravichandran, K. G.; Peterson, J . A.; Deisen-
hofer, J . Crystal Structure and Refinement of Cytochrome
P450terp at 2.3 Å Resolution. J . Mol. Biol. 1994, 236, 1169-
1185.
(30) Massa, S.; Di Santo, R.; Costi, R.; Simonetti, G.; Retico, A.;
Apuzzo, G.; Artico, M. Antifungal Agents. III. Naphthyl and
Thienyl Derivatives of 1H-Imidazol-1-yl-4-phenyl-1H-pyrrol-3-
ylmethane. Farmaco 1993, 48, 725-736.
(15) Podust, L. M.; Poulos, T. L.; Waterman, M. R. Crystal Structure
of Cytochrome P450 14R-sterol Demethylase (CYP51) from
Mycobacterium tuberculosis in Complex with Azole Inhibitors.
Proc. Natl. Acad. Sci. U.S.A. 2001, 98 (6), 3068-3073.
(16) J i, H.; Zhang, W.; Zhou, Y.; Zhang, M.; Zhu, J .; Song, Y.; Lu, J .;
Zhu, J . A Three-Dimensional Model of Lanosterol 14R-Demethyl-
ase of Candida albicans and Its Interaction with Azole Anti-
fungals. J . Med. Chem. 2000, 43, 2493-2505 and references
therein.
(17) Lewis, D. F. V.; Wiesman, A.; Tarbit, M. H. Molecular Modelling
of Lanosterol 14R-demethylase (CYP51) from Saccharomices
cerevisiae via Homology with CYP102, a Unique Bacterial
Cytochrome P450 Isoform: Quantitative Structure-Activity
Relationships (QSARs) within Two Related Series of Antifungal
Azole Derivatives. J . Enzyme Inhib. 1999, 14, 175-192.
(18) Tsukuda, T.; Shiratori, Y.; Watanabe, M.; Ontsuka, H.; Hattori,
K.; Shirai, M.; Shimma, N. Modeling, Synthesis and Biological
Activity of Novel Antifungal Agents (1). Bioorg. Med. Chem. Lett.
1998, 8, 1819-1824.
(19) Tafi, A.; Anastassopoulou, J .; Theophanides, T.; Botta, M.;
Corelli, F.; Massa, S.; Artico, M.; Costi, R.; Di Santo, R.; Ragno,
R. Molecular Modeling of Azole Antifungal Agents Active Against
Candida albicans. 1. A Comparative Molecular Field Analysis
Study. J . Med. Chem. 1996, 39, 1227-1235.
(20) Talele, T. T.; Kulkarni, V. M. Three-dimensional Quantitative
Structure-Activity Relationships (QSAR) and Receptor Mapping
of Cytochrome P-45014RDM Inhibiting Azole Antifungal Agents.
J . Chem. Inf. Comput. Sci. 1999, 39 (2), 204-210.
(21) Buchel, K. H.; Draber, W.; Regel, E.; Plempel, M. Synthesen und
Eigenschften von Clotrimazol und Weiteren Antimykotischen
1-Triphenylmethylimidazolen (Syntheses and Properties of Clo-
trimazole and Further Antimycotic 1-(Triphenylmethyl)imid-
azoles). Arzneim.-Forsch. 1972, 22, 1260-1272.
(22) Heeres, J .; Backx, L. J . J .; Mostmans, J . H.; Van Cutsem, J .
Antimycotic Imidazoles. 4. Synthesis and Antifungal Activity of
Ketoconazole, a New Potent Orally Active Broad Spectrum
Antifungal Agent. J . Med. Chem. 1979, 22, 1003-1005.
(23) Heel, R. C.; Brodgen, R. N.; Carmine, A.; Morley, P. A.; Speight,
T. M.; Avery, G. S. Ketoconazole: A Review of Its Therapeutic
Efficacy in Superficial and Systemic Fungal Infections. Drugs
1982, 23, 1-36.
(24) Plempel, M.; Regel, E.; Buchel, K. H. Antimycotic Efficacy of
Bifonazole in Vitro and in Vivo. Arzneim.-Forsch. 1983, 33, 517-
524.
(25) Massa, S.; Di Santo, R.; Mai, A.; Botta, M.; Artico, M.; Panico,
S.; Simonetti, G. Research on Antibacterial and Antifungal
Agents. XIII. Synthesis and Antimicrobial Activity of 1-Aryl-
methyl-4-aryl-1H-pyrrole-3-carboxylic acids. Farmaco 1990, 45,
833-846.
(26) Massa, S.; Di Santo, R.; Retico, A.; Artico, M.; Simonetti, N.;
Panico, S.; Fabrizi, G.; Lamba, D. Antifungal Agents. 1. Syn-
thesis and Antifungal Activities of Estrogen-Like Imidazole and
Triazole Derivatives. Eur. J . Med. Chem. 1992, 27, 495-502.
(27) Massa, S.; Di Santo, R.; Artico, M.; Costi, R.; Di Filippo, C.;
Simonetti, G.; Retico, A.; Artico, M. Researches on Antibacterial
and Antifungal Agents. XV. 3-Aryl-4-[R-(1H-imidazol-1-yl)ben-
zyl]pyrroles with Potent Antifungal Activity. Eur. Bull. Drug
Res. 1992, 1, 12-17.
(31) Massa, S.; Di Santo, R.; Costi, R.; Mai, A.; Artico, M.; Retico, A.;
Apuzzo, G.; Artico, M.; Simonetti, G. Antifungal Agents. 4.
Synthesis of neo-Isopyrrolnitrin and other 3-Aryl-4-nitro-1H-
pyrroles via TosMIC. Med. Chem. Res. 1993, 3, 192-199.
(32) Artico, M.; Massa, S.; Di Santo, R.; Costi, R.; Retico, A.; Apuzzo,
G.; Simonetti, N. Antifungal Agents. 5. Chloro and Amino
Derivatives of 1,2-Diaryl-1-(1H-imidazol-1-yl)ethane with Potent
Antifungal Activities. Eur. J . Med. Chem. 1993, 28, 715-720.
(33) Di Santo, R.; Massa, S.; Costi, R.; Simonetti, G.; Retico, A.;
Apuzzo, G.; Troccoli, F. Antifungal Agents. VIII. Synthesis and
Antifungal Activities of Bipyrryl Analogues of Bifonazole. Far-
maco 1994, 49, 229-236.
(34) Artico, M.; Di Santo, R.; Costi, R.; Massa, S.; Retico, A.; Artico,
M.; Apuzzo, G.; Simonetti, G.; Strippoli, V. Antifungal Agents.
9. 3-Aryl-4-[R-(1H-imidazol-1-yl)arylmethyl]pyrroles: A New
Class of Potent Anti-Candida Agents. J . Med. Chem. 1995, 38,
4223-4233.
(35) Di Santo, R.; Costi, R.; Artico, M.; Massa, S.; Musiu, C.; Scintu,
F.; Putzolu, M.; La Colla, P. Antifungal estrogen-like imidazoles.
Synthesis and antifungal activities of thienyl and 1H-pyrrolyl
derivatives of 1-aryl-2-(1H-imidazol-1-yl)ethane. Eur. J . Med.
Chem. 1997, 32, 143-149.
(36) Di Santo, R.; Costi, R.; Artico, M.; Massa, S.; Musiu, C.; Milia,
C.; Putzolu, M.; La Colla, P. N-(1-Naphthylmethyl)-N-(1-alkyl-
4-aryl-1H-pyrrol-3-ylmethyl)methylamines Related to Naftifine.
Synthesis and Antifungal Activity. Med. Chem. Res. 1997, 7, 98-
108.
(37) Catalyst, version 4.6; Molecular Simulation Inc., 9685 Scranton
Road, San Diego, CA 92121-3752.
(38) Kaminski, J . J .; Rane, D. F.; Snow, M. E.; Weber, L.; Rothofsky,
M. L.; Anderson, S. D.; Lin, S. L. Identification of novel farnesyl
protein transferase inhibitors using three-dimensional database
searching methods. J . Med. Chem. 1997, 40, 4103-4112.
(39) Barnum, D.; Greene, J .; Smellie, A.; Sprague, P. Identification
of common functional configurations among molecules. J . Chem.
Inf. Comput. Sci. 1996, 36, 563-571.
(40) Botta, M.; Corelli, F.; Gasparrini, F.; Messina, F.; Mugnaini, C.
Chiral azole derivatives. 4.1. Enantiomers of bifonazole and
related antifungal agents: synthesis, configuration assignment,
and biological evaluation. J . Org. Chem. 2000, 65 (15), 4736-
4739.
(41) Catalyst
guidelines, http://www.msi.com/support/catalyst/
hypogen.html.
(42) Smellie, A.; Teig, S.; Towbin, P. Poling: promoting conforma-
tional variation. J . Comput. Chem. 1995, 16, 171-187.
(43) Davey, W.; Gwilt, J . R. Chalcones and Related Compounds. Part
I. Preparation of Nitro-, Amino-, and Halogeno-chalcones. J .
Chem. Soc. 1957, 1008-1014.
(44) Rekhter, M. A.; Grushetskaya, G. N.; Panasenko, A. A.; Krimer,
M. Z. New Derivatives of N-Vinyltriazoles-1,3-diaryl-3(1H-1,2,4-
triazol-1yl)-2-propen-1-ones. Khim. Geterotsikl. Soedin. 1995, 7,
910-914; Chem. Abstr. 1996, 124, 202126k.
(45) Severi, F.; Costantino, L.; Benvenuti, S.; Vampa, G.; Mucci, A.
Synthesis and Description of Chalcone-like Compounds, Inhibi-
tion of Aldose Reductase. Med. Chem. Res. 1996, 6, 128-136.
(46) Hahn, M. Three-dimensional shape-based searching of confor-
mationally flexible compounds. J . Chem. Inf. Comput. Sci. 1997,
37, 80-86.
(28) Massa, S.; Di Santo, R.; Artico, M.; Costi, R.; Apuzzo, G.;
Simonetti, G.; Artico, M. Novel In Vitro Highly Active Antifungal
Agents with Pyrrole and Imidazole Moieties. Med. Chem. Res.
1992, 2, 148-153.
(29) Di Santo, R.; Massa, S.; Artico, M. Synthesis of Biologically
Active Azoles via TosMIC. Farmaco 1993, 48, 209-229.
(47) Fischer, R. In The design of experiments; Hafner: New York,
1996.
J M011087H