Structure-based optimization of azole antifungal agents by CoMFA, CoMSIA, and molecular docking

Article abstract of DOI:10.1021/jm051211n

In a continuing effort to develop highly potent azole antifungal agents, the three-dimensional quantitative structure-activity relationship methods, CoMFA and CoMSIA, were applied using a set of novel azole antifungal compounds. The binding mode of the compounds at the active site of lanosterol 14α-demethylase was further explored using the flexible docking method. Various hydrophobic, van der Waals, π-π stacking, and hydrogen bonding interactions were observed between the azoles and the enzyme. Based on results from the molecular modeling, a receptor-based pharmacophore model was established to guide the rational optimization of the azole antifungal agents. Thus, a total of 57 novel azoles were designed and synthesized by a three-step optimization process. In vitro antifungal assay revealed that the antifungal activities of these novel azoles were greatly improved, which confirmed the reliability of the model from molecular modeling.

Full text of DOI:10.1021/jm051211n

2512
J. Med. Chem. 2006, 49, 2512-2525
Structure-Based Optimization of Azole Antifungal Agents by CoMFA, CoMSIA, and Molecular
Docking
Chunquan Sheng, Wannian Zhang,* Haitao Ji, Min Zhang, Yunlong Song, Hui Xu, Jie Zhu, Zhenyuan Miao, Qingfen Jiang,
Jianzhong Yao, Youjun Zhou, Ju¨ Zhu, and Jiaguo Lu¨
School of Pharmacy, Second Military Medical UniVersity, 325 Guohe Road, Shanghai 200433, People’s Republic of China
ReceiVed December 1, 2005
In a continuing effort to develop highly potent azole antifungal agents, the three-dimensional quantitative
structure-activity relationship methods, CoMFA and CoMSIA, were applied using a set of novel azole
antifungal compounds. The binding mode of the compounds at the active site of lanosterol 14R-demethylase
was further explored using the flexible docking method. Various hydrophobic, van der Waals, π-π stacking,
and hydrogen bonding interactions were observed between the azoles and the enzyme. Based on results
from the molecular modeling, a receptor-based pharmacophore model was established to guide the rational
optimization of the azole antifungal agents. Thus, a total of 57 novel azoles were designed and synthesized
by a three-step optimization process. In vitro antifungal assay revealed that the antifungal activities of these
novel azoles were greatly improved, which confirmed the reliability of the model from molecular modeling.
Introduction
that catalyzes the oxidative removal of the 14R-methyl group
of lanosterol to give ∆14,15-desaturated intermediates in ergos-
terol biosynthesis.15 In yeast and fungi, CYP51 is the key
enzyme in sterol biosynthesis. Selective inhibition of CYP51
would cause depletion of ergosterol and accumulation of
lanosterol and other 14-methyl sterols resulting in the growth
inhibition of fungal cells.16 Due to the importance of CYP51
in antifungal drug studies, it is of great importance to know
the three-dimensional (3D) structures of CYP51s, particularly,
the CYP51s from pathogenic fungi. However, eukaryotic
CYP51s are membrane-associated proteins, and solving their
crystal structures remains a challenge. In our previous studies,17
we have constructed a 3D model of CYP51 from Candida
albicans (CACYP51) through homology modeling on the basis
of the crystallographic coordinates of four prokaryotic P450s:
P450BM3,18 P450cam,19 P450terp,20 and P450eryF.21 The
pharmacophoric conformation of azole antifungal agents was
searched by the active analogue approach and then docked into
the active site of the enzyme. Subsequently, the active site of
the modeled structure was investigated by MCSS functional
group mapping and LUDI calculations. Several non-azole lead
compounds were designed and synthesized by structure-based
de novo design approach.22 The designed lead compounds
exhibited a strong inhibitory effect on CYP51 of C. albicans,
but they only showed marginal antifungal activity in vitro as
compared to fluconazole. The low structural identity between
the prokaryotic P450s and the eukaryotic CYP51s limits the
development of highly potent inhibitors and led us to build a
more accurate model of CYP51 from pathogenic fungi.
During the past two decades, the life threatening infections
caused by pathogenic fungi have become increasingly common,
especially in individuals with immunocompromised hosts, such
as patients undergoing anticancer chemotherapy or organ
transplants and patients with AIDS.1 Clinically, candidosis,
aspergillosis and cryptococcosis are three major fungal infections
in immunocompromised individuals.2,3 Moreover, dermatomy-
coses, such as toenails and tinea pedis, are among the most
widespread human superficial and cutaneous fungal infections.
However, the current antifungal therapy suffers from drug
related toxicity, severe drug resistance, nonoptimal pharmaco-
kinetics, and serious drug-drug interactions. Therefore, there
is an emergent need to develop novel antifungal drugs with
higher efficiency, broader spectrum, and lower toxicity.
The common antifungal agents currently used in clinic are
azoles (such as fluconazole, ketoconazole, and itraconazole),4
polyenes (such as amphotericin B5 and nystatin6), echinocandins
(such as caspofungin and micafungin),7 and allylamines (such
as naftifine and terbinafine).8 Among those, azoles have
fungistatic, orally active, and broad-spectrum activities against
most yeasts and filamentous fungi. They are widely used in
antifungal chemotherapy. Fluconazole shows good antifungal
activity with relatively low toxicity and is preferred as first-
line antifungal therapy.9 However, fluconazole is not effective
against invasive aspergillosis and has suffered severe drug
resistance.10;11 Itraconazole is an improvement of fluconazole
in terms of having a broader antifungal spectrum and better
toleration. However, its use is hampered by variable oral
absorption and low bioavailability. This situation has led to an
ongoing search for new azoles, and several novel azoles have
been developed with improved profiles. The second generation
of azoles (Chart 1), such as voriconazole,12 posaconazole,13 and
ravuconazole,14 are marketed or currently in the late stages of
clinical trials. They are noted for their broad antifungal spectrum,
low toxicity, and improved pharmacodynamic profiles.
In 2001, Podust et al. reported the crystal structure of a
prokaryotic sterol 14R-demethylase from Mycobacterium tu-
berculosis (MTCYP51), in complex with two azole inhibitors
(4-phenylimidazole and fluconazole), which is the only crystal
structure for the CYP51 family.23 More recently, estriol bound
and ligand free structures of MTCYP51 were also reported.24
Although this crystal structure of CYP51 was not from fungi,
it can provide a more accurate structural template for modeling
the 3D structures of fungal CYP51 than the prokaryotic P450s.
For this purpose, we rebuilt the 3D model of CYP51 from C.
albicans and Aspergillus fumigatus on the basis of the crystal
Azole antifungals act by competitive inhibition of the
lanosterol 14R-demethylase (P45014DM, CYP51), the enzyme
* To whom correspondence should be addressed. Phone: 86-21-
25074460. Fax: 86-21-25074460. E-mail: zhangwnk@hotmail.com.
10.1021/jm051211n CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/22/2006

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2513
Chart 1
coordinates of MTCYP51. The binding mode of the substrate
with fungal CYP51 was also investigated.25 In a continuing
effort to find more potent azole antifungal agents and design
lead compounds with higher affinity, we have designed a series
of novel azoles with substituted piperazinyl side chains. These
new azoles exhibited better antifungal activity in vitro than
fluconazole and itraconazole.26;27 These compounds can be used
as a reliable data set to perform computational simulations to
gain insights into the structural and chemical features required
for the antifungal activity.
Chemistry
The synthesis of all compounds in the data set (1-40) has
been published previously.26,27 All the target compounds
synthesized that possess a chiral center on the C2 atom are
racemates. As a key intermediate of our designed triazole
antifungals, the oxirane compound 44 was synthesized by an
improved procedure by this labratory.41 The preparation of
trizoles with benzylamine side chains (compounds 45-57) is
outlined in Scheme 1. The title compounds 45-57 were
prepared by ring-open reaction of epoxide 44 with various
N-methyl benzylamines, which were obtained by the reported
methods.42 This reaction proceeded better in the protic solvent
(such as ethanol) than in nonprotic solvents. The triazoles with
amide-benzylamine side chains (compounds 58-67) were
synthesized starting from compound 57, using conventional
methods as depicted in Scheme 2. The good yields were
obtained when these reactions were performed in dichlo-
romethane in the presence of a base at room temperature.
Triethylamine is a better base than either potassium carbonate
or sodium hydroxide. The triazoles with triazolone-benzylamine
side chains (compounds 76-109) were synthesized via eight
steps (Scheme 3). The amino group of compound 68 was first
protected by (BOC)2O. Then the nitro group on the phenyl ring
was reduced to an amino group in the presence of Raney Ni
and hydrazine hydrate. In the presence of a base at room
temperature, the aniline 70 was converted to phenylcarbamate
71 by reacting with phenyl chloroformate. Using pyridine as
the base and ethyl acetate as the solvent results in a better yield
than the combination of triethylamine (base) and dichlo-
romethane (solvent). Phenylcarbamate 71 was treated with
hydrazine hydrate to give semicarbazide 72, which was subse-
quently reacted with formamidine acetate in the presence of
DMF at 80 °C to give the triazolone 73. Triazolone 73 was
treated with substituted benzyl chlorides and substituted 2-chloro-
1-phenyl-ethanones in the presence of potassium carbonate and
DMF at 80 °C to afford various protected triazolone-benzyl-
Although extensive efforts have been put to the structural
modification of current azole antifungals,28,29 most of them were
undertaken only by traditional methods, which was time-
consuming and less effective. The structural information of the
target enzyme can be utilized to accelerate the discovery of novel
antifungal agents. Several three-dimensional quantitative struc-
ture-activity relationship (3D-QSAR) studies for different azole
antifungals have been reported.28,30-33 However, most of these
models did not incorporate the structural information of the
receptor, and the robustness of the models remained to be
confirmed by further drug design and synthesis. Also, a number
of docking analysis studies of current azole antifungal agents
have been performed,34-38 but these papers paid more attention
to the mechanism of drug resistance and the selectivity between
candidosis and aspergillosis. In the present study, we applied
comparative molecular field analysis (CoMFA)39 and compara-
tive molecular similarity indices analysis (CoMSIA)40 based on
a set of novel azole antifungals from our laboratory to build
robust 3D-QSAR models. The binding mode of these com-
pounds was further clarified by molecular docking. On the basis
of the results from 3D-QSAR and docking, a pharmacophore
model for the rational optimization of azole antifungal agents
was constructed. Novel antifugal azoles were then designed and
synthesized according to the developed pharmacophore model,
which exhibited excellent antifungal activity with broad anti-
fungal spectrum.

2514 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
Scheme 1
Scheme 2
amines, which were then deprotected by 10% HCl in ethyl
acetate to give triazolone-benzylamine side chains. The target
compounds 76-109 were obtained by treating epoxide 44 with
various triazolone-benzylamine side chains in the presence of
triethylamine and ethanol at 80 °C with moderate to high yields.
the racemates are suitable for the 3D-QSAR studies. Second,
one enantiomer is active while the other is completely inactive.
Thus, the activities of the racemates should be doubled when
we build 3D-QSAR models on single enantiomers. Third, both
enantiomers have activities but one enantiomer is more active.
In this case, the activities of the enantiomers are difficult to be
determined when we only have data from the racemates.
However, the use of the doubled values of the biological data
should be an alternative way. In the present studies, the 3D-
QSAR calculations were repeated on both R and S isomers of
the compounds with their MIC80 values doubled. However, only
minor changes in the statistical parameters were observed, and
the original MIC80 values for the racemates were used to build
the final CoMFA and CoMSIA models.
A correlation coefficient of r2 ) 0.940 and 0.952 and a cross-
validated coefficient of q2 ) 0.718 and 0.655 were obtained
for the best CoMFA and CoMSIA model, respectively (Table
2). The final CoMFA and CoMSIA models were satisfactory
from the viewpoint of the statistical significance and the
predictive ability of the training set and test set (see Table 1,
Figures 3 and 4). However, the ultimate test for the usefulness
of a 3D-QSAR model in the drug design process is to predict
the activity of new compounds that are not included in the data
set. Therefore, we designed and synthesized a test set comprising
57 novel triazole antifungal compounds on the basis of the
results from 3D-QSAR and molecular docking studies, which
is discussed below.
Results and Discussions
CoMFA and CoMSIA Models. A data set of our newly
synthesized triazole antifungal agents was used to perform the
3D-QSAR studies. The antifungal activities of the compounds
were all determined utilizing the same experimental protocol
in our laboratory, covered a range of 3 orders of magnitude,
and uniformly distributed. During the optimization of 3D-QSAR
models, variations of parameters, such as grid spacing and
attenuation factor, were considered. However, the best results
were obtained from the parameters with the default values. It
has been discussed that the five different descriptor fields are
not totally independent of each other and such dependencies of
individual fields usually decrease the statistical significance of
the models.43,44 An evaluation, which of the five CoMSIA fields
are actually needed for a predictive model, was performed by
calculating all possible combinations of fields, and the much
faster SAMPLS45 algorithm was used (Figure 2). The first five
models, using a single CoMSIA field, indicated that electrostatic
field and hydrophobic field were more important than the other
three fields. The combination of electrostatic and hydrophobic
field gave the best two-field model (q2
) 0.652). The
SAMPLS
addition of steric field resulted in the slight increase of q2
CoMFA and CoMSIA Contour Maps. To visualize the
information content of the derived 3D-QSAR models, CoMFA
and CoMSIA contour maps were generated. The field energies
at each lattice point were calculated as the scalar results of the
coefficient and the standard deviation associated with a particular
column of the data table (“stdev*coeff”), which was always
plotted as the percentage of the contribution to the CoMFA or
CoMSIA equation. Because most structural modifications of
azole antifungal agents were aimed to change the side chains
attached to C3, the contour plots around the substitutions on
SAMPLS
value, and the combination of steic, electrostatic, and hydro-
phobic fields was proved to be the best model (q2
0.655).
)
SAMPLS
As shown in Figure 1 and Table 1, compounds in the data
set were used as racemates for antifungal testing. CoMFA and
CoMSIA studies performed on both R and S isomers of the
compunds were also considered. In most cases, there are three
possibilities when the enantiomers affect the bioactvity. First,
the two isomers have the same activity, and the activities from

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2515
Scheme 3
the piperazinyl side chain could reflect the properties of the
corresponding region in the active site of CACYP51 and could
guild the rational optimization of the azoles in the data set.
Because CoMFA and CoMSIA contour maps of steric and
electrostatic fields revealed similar results, only CoMFA contour
maps are shown. In Figure 5, the CoMFA contour map of steric
field is displayed. The green contours represent the regions of
high steric tolerance, while yellow contours represent regions
of unfavorable steric effects. The sterically favored regions are
around the ortho-position and para-position of the phenyl ring,
which indicates that compounds with larger substitutions are
essential for high antifungal activity. For example, the substitu-
tions on the para-position of the phenyl group of compounds
27, 34, and 22 are methyl, ethyl, and tert-butyl, respectively,
and the bulkier substitutions lead to more active compounds.
There is a small sterically disfavored region around the meta-
position of the phenyl ring, and this result can be understood
in considering the higher antifungal activity of compound 24
than that of compound 36. Negatively charged (electron-rich)
favorable red regions are found at ortho- and para-positions of
the phenyl (Figure 6), indicating that an electron-withdrawing
substituent enhances the activity. These regions support the
observation that compound 1, 2, and 3 with electron-rich
substitution at the ortho-position are among the compounds with
highest antifungal activity. A blue region around the linker
between phenyl and piperazinyl represents an area where
positive charge is favored (Figure 6). This result is in good
agreement with the fact that the benzoyl analogues showed lower
Figure 1. Structure of the compounds in the training set and test set
(asterisk indicates compounds in test set).
antifungal activity than the other compounds and the negative
charge of the carbonyl is disfavored for the antifungal activity.
Furthermore, CoMSIA reveals other contour maps that are
not shown by CoMFA. Cyan regions in Figure 7 indicate areas
where hydrophobic substitutions are preferred, and hydrophobic
disfavored regions are not shown with contribution level of 20%.

2516 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
Table 1. Experimental and Calculated Antifungal Activity (-log MIC80) of Compounds in the Training Set and the Test Set for the Final CoMFA and
CoMSIA Models
CoMFA
calcd
CoMSIA
calcd
CoMFA
calcd
CoMSIA
calcd
compd
obsd
compd
obsd
1
2
3
4
5
6
7
8
1.470
1.456
1.490
1.495
0.863
-0.398
0.807
0.906
0.267
-0.349
-0.335
-0.025
-1.229
-0.355
-0.258
0.564
-1.559
-1.558
-0.369
-1.243
1.371
1.500
1.389
1.429
0.806
-0.928
0.753
1.117
-0.152
-0.087
-0.655
0.136
-1.137
-0.437
-1.172
0.620
-1.016
-1.532
-0.705
-1.226
1.540
1.519
1.396
1.466
0.856
-0.414
0.812
0.847
-0.069
-0.042
-0.347
-0.048
-1.318
-0.384
-0.774
0.549
-1.368
-1.379
-0.766
-1.363
21
22
23a
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39a
40a
-0.015
1.469
1.447
0.855
0.237
0.596
-0.369
0.534
-1.543
0.912
0.912
0.596
1.213
0.247
0.253
0.635
0.936
1.480
-0.666
0.545
0.291
1.584
1.534
0.740
0.706
0.694
-0.316
0.421
-1.328
0.862
0.638
0.512
1.362
-1.101
0.289
0.701
0.890
1.390
-0.502
0.428
-0.090
1.328
1.878
0.917
0.633
0.714
-0.111
0.274
-1.544
1.050
0.666
0.596
1.334
-0.055
0.367
0.860
1.016
1.313
-0.624
0.387
9
10
11
12
13a
14
15
16
17
18a
19
20
Table 2. Statistical Parameters for the Final CoMFA and CoMSIA Models
fraction
model
q2
r2
standard error
F
n
steric
electrosatic
hydrophobic
0.405
CoMFA
CoMSIA
0.718
0.655
0.940
0.952
0.234
0.204
90.838
50.621
5
5
0.334
0.208
0.666
0.387
Figure 2. Results of 31 possible CoMSIA field combinations (s )
steric, e ) electrostatic, h ) hydrophobic, d/a ) H-bond donor/acceptor)
Figure 3. Experimental and calculated pMIC80 for the CoMFA analysis
of the training set and test set.
using q2
.
SAMPLS
groups are favored, which is further supported by the docking
analysis below. For the substitutions on the ortho-position, steric,
hydrophobic, and electro-rich groups (e.g. halogens) were
favored. The substitutions on the meta-position of the phenyl
are not important for the antifungal activity, and it was suitable
to leave this position unsubstituted.
Binding Mode of the Azoles. Because 3D-QSAR is only a
kind of ligand-based approach, the robustness of the derived
models is greatly affected by the diversity of the molecules in
the data set. Thus, the information got from the 3D-QSAR
analysis is relatively limited. Therefore, the molecular docking
was used to further clarify the binding mode of the compounds
and provide straightforward information for further structural
optimization. Our homologous model of CACYP51 built from
the crystal structure of MTCYP51 was used in the present
docking study.41 To evaluate the effectiveness of the Affinity
The hydrophobic favored regions are around the ortho-position
and para-position of the phenyl ring, which is similar to the
steric favored regions. In fact, some steric groups are always
hydrophobic. A comparison of compound 25 and 30 (also
compound 26 and 33) shows that the addition of a halogen on
the ortho-position of benzyl results in the increase of the
antifungal activity, which may be due to the increase in
hydrophobicity. Moreover, all the compounds with hydrophobic
tert-butyl substitution at the para-position of the phenyl ring
(compounds 5, 21, and 22) showed high antifungal activity.
Analysis of CoMFA and CoMSIA contour plots revealed that
the substitutions on the para-position of the phenyl ring are more
important for the antifungal activity than those on the ortho-
position and meta-position. For the substitutions on the para-
position, the steric and hydrophobic groups or the electro-rich

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2517
Figure 6. Contour plots of CoMFA electrostatic field. Positive charge
favored areas (contribution level of 80%) are represented by red
polyhedra, and positive charge disfavored areas (contribution level of
20%) are represented by blue polyhedra. Compound 9 is shown in ball-
and-stick mode.
Figure 4. Experimental and calculated pMIC80 for the CoMSIA
analysis of the training set and test set.
Figure 5. Contour plots of CoMFA steric field. Sterically favored areas
(contribution level of 80%) are represented by green polyhedra, and
sterically disfavored areas (contribution level of 20%) are represented
by yellow polyhedra. Compound 9 is shown in ball-and-stick mode.
Figure 7. Contour plot of CoMSIA hydrophobic field. Cyan polyhedra
indicates hydrophobic substituents favored regions with the contribution
level of 80%, while magenta polyhedra indicating hydrophobic sub-
stituents disfavored regions is not shown with the contribution level of
20%. Compound 9 is shown in ball-and-stick mode.
method46 in the flexible docking of azole antifungal agents, the
crystal structure of MTCYP51 in complex with fluconazole
(PDB entry 1EA1) was chosen as a test case.23 Using the flexible
ligand docking procedure in the Affinity module within the
InsightII software package, the observed crystallographic con-
formation can be reproduced with the RMSD value of 1.981
Å, indicating that this docking method can be used as a powerful
tool to clarify the binding mode of the azole antifungal agents.
The calculated interaction energies and LUDI scores47,48 of each
compound in the data set and fluconazole are given in Table 3.
The calculated interaction energies of the compounds in the data
set correlated well with the experimentally determined antifungal
activity (r2 ) 0.808), indicating the interaction energy may be
a useful parameter in optimizing the azole antifungal com-
pounds.
The docked conformation of fluconazole into the active site
of CACYP51 was similar to the bioactive conformation of
fluconazole into the crystal complex of MTCYP51 (Figure 8),
with the RMSD value of 0.4192 Å. The binding mode of
fluconazole with CACYP51 was similar to that of our previous
studies.17 Fluconazole binds to the active site of CACYP51
through the formation of a coordination bond with iron of the
heme group. The diflurophenyl group of fluconazole is located
in the hydrophobic binding cleft lined with A114, F126, L139,
M140, F145, I304, and M306, and the 17-side chain of the
substrate was also found to be located in the same pocket.25
Several residues lined with L121, T122, F228, T311, P375,
L376, and M508 were observed to form indirect nonbonding
interactions with another triazolyl ring attached to C3.
The development of novel triazole antifungal agents has been
focused on the modification of the side chain attached to C3,
and it is of great importance to understand the properties of the
corresponding part in the active site of fungal CYP51. Therefore,
the azoles in the data set can be used as probes to investigate
the important residues in the active site of CACYP51 and to
guide the design of more potent antifungal agents. Both R and
S isomers of the compounds were considered in our docking
studies, and they interacted with CACYP51 through a similar
binding mode. However, R isomers showed lower interaction
energy with CACYP51 than the S isomers, which indicated that
the R isomers might have better antifungal activity than the S

2518 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
Table 3. Calculated Interaction Energies (kcal/mol) for the Complexes of Triazole Antifungal Compounds with the Active Site of CACYP51, Ludi
Scores, and Experimentally Determined Antifungal Activity (MIC80, µmol/L)
a
compd
MIC80
Evdw
Eelect
Etotal
Ludi score 1b
Ludi score 2c
1
2
3
4
5
6
7
8
0.0339
0.0350
0.0324
0.0320
0.1372
2.4973
0.156
0.1242
0.5412
2.2325
2.1649
1.0582
16.9326
2.2651
18.1208
0.2732
36.2417
36.1606
2.3394
17.4871
1.0339
0.034
0.0357
0.1395
0.5794
0.2538
2.3391
0.2923
34.8986
0.1224
0.1224
0.2535
0.0612
0.5662
0.5581
0.2317
0.1158
0.0331
4.6354
0.2850
0.8162
-76.031
-70.017
-73.302
-70.050
-73.704
-70.636
-68.549
-80.832
-73.543
-64.854
-56.585
-66.842
-69.903
-64.459
-73.031
-73.342
-64.745
-64.392
-64.716
-66.4795
-70.305
-85.038
-72.062
-73.894
-72.371
-73.373
-76.951
-78.007
-69.934
-78.975
-72.547
-69.422
-80.855
-80.565
-70.449
-74.093
-74.624
-82.108
-72.017
-72.014
-54.864
-6.011
-7.884
-4.734
-5.985
-4.7834
4.792
-7.121
-4.485
0.230
-0.271
-10.642
-10.392
3.297
-82.642
-77.901
-77.676
-76.035
-78.487
-65.844
-75.670
-85.317
-73.313
-65.125
-67.227
-77.234
-66.606
-66.906
-63.241
-75.235
-70.334
-52.563
-67.871
-58.804
-77.655
-81.050
-75.544
-79.477
-73.166
-77.413
-78.527
-72.811
-63.093
-81.859
-78.714
-76.213
-85.423
-76.204
-75.547
-77.045
-76.731
-84.950
-73.363
-75.380
-58.063
545
548
534
560
521
415
465
610
468
468
550
416
465
488
452
533
447
389
503
522
506
625
489
520
474
543
528
530
368
528
539
409
507
554
516
540
561
596
497
458
336
905
925
1017
1055
941
778
972
1150
793
786
809
785
847
790
908
900
790
707
864
872
873
1024
965
946
868
961
914
946
695
963
929
750
957
1061
944
911
969
1028
967
723
791
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
-2.447
9.790
-1.893
-5.589
11.829
-3.155
7.675
-7.351
3.989
-3.482
-5.583
-0.795
-4.040
-1.576
5.196
6.841
-3.064
-6.617
-6.792
-4.618
4.362
-5.098
-2.952
-2.107
-2.842
-1.346
-3.366
-3.199
fluconazole
a Etotal ) Evdw + Eelect. bThe Ludi score 1 was calculated by the method in ref 10. cThe Ludi score 2 was calculated by the method in ref 11. The difference
between the LUDI score 1 and score 2 is that an additional term evaluating the contribution of binding of aromatic-aromatic interaction is added to LUDI
score 2.
isomers. In the following discussion, all the docked conforma-
tions refer to the R configuration of the compounds. The docking
results revealed that the general conformation of the compounds
in the active site was similar to that of fluconazole. The distance
between the N4 atom of triazolyl group and the iron of the heme
group varies from 2.30 to ∼2.37 Å, which is suitable to form
a coordination bond. The difluorophenyl group of the com-
pounds is also located in the hydrophobic pocket lined with
A114, F126, L139, M140, F145, I304, and M306. The
hydrophobic interaction between them could stabilize the
compounds in the active site. The hydroxyl group attached to
C2 was important for the antifungal activity, but no interaction
was observed between this hydroxyl group and the active site
of CACYP51. In the crystal structure of MTCYP51 complexed
with fluconazole, the active site is filled with water molecules
and fluconazole interacts with at least three water molecules23
which bridge the interactions with the active site. From the
docking results, it is possible that the water molecules in the
active site mediate the interaction between the hydroxyl group
and the vicinal H310, which is a highly conserved residue in
the CYP51 family and has been found to form a hydrogen bond
with the 3-OH of the substrate in our previous studies.25
Crystal structures of MTCYP51 and other P450 proteins
revealed that there were two putative substrate access channels
(channel 1 and channel 2).23 Channel 1 is parallel to the plane
of the heme group and is formed by the bending the helix I
away from the heme, which released the BC loop in an open
conformation. Channel 2 is perpendicular to the plane of heme
group and its entrance is surrounded by helix A′ and FG loop.
Channel 1 is apparent in the crystal structure of MTCYP51 and
P450 2C5, and channel 2 is based on the crystal structure of
P450BM3. In our model of CACYP51, the ligand access channel
shared similar topological features to that of MTCYP51, with
BC loop having an open conformation and FG loop having a
closed conformation.25 Flexible docking results indicated that
the piperazinyl side chains of the compounds had preferable
orientation into channel 2, which is consistent with a recent
docking result of posaconazole with CACYP51.37
Moreover,
the resulting binding model can well explain the structure-
antifungal activity relationship of the compounds in the data
set. The piperazinyl group of the side chain interacted with the
surrounding hydrophobic residues such as A117, L376, and
I379. The phenyl group of the side chain interacts with the
phenol group of Y118 through the formation of π-π face-to-
edge interaction. This result is consistent with the CoMFA
electrostatic contour map in that electron-rich (or electron-
withdrawing) groups are favored in the ortho- and para-position
of the phenyl group. Furthermore, a docking model of 1,4-

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2519
Figure 8. Docking of fluconazole into the active site of CACYP51
and residues within 6 Å from the ligand are shown. The docked
conformation of fluconaole (blue) and its conformation in the crystal
structure of MTCYP51 (black) are superimposed.
Figure 9. Docking of compound 2 into the active site of CACYP51
and the residues within 6 Å from the ligand are shown. The hydrogen
bonds are indicated with dotted lines.
benzothiazine and 1,4-benzixazine imidazole derivatives with
CACYP51 also found Y118 could form π-π interaction with
the inhibitors.49 Y118 is a highly conserved residue through the
CYP51 family, and site-directed mutagenesis of corresponding
Y76 of MTCYP51 has revealed its importance in maintaining
proper orientation of the heme.50 Therefore, Y118 may be an
important residue for the design of CACYP51 inhibitors.
Various substituent groups on the phenyl group could form
different interactions with the active site of CACYP51, which
resulted in the different antifungal activity. Docking results of
compounds 1, 2, and 3 showed that all the three compounds
could form a hydrogen bond with the side chain of S378 by
the substitutions on the para-position of the phenyl group (Figure
9). In our previous studies, S378 was used as a hydrogen bond
site to design non-azole inhibitors, and the results supported
our hypothesis.22 Therefore, we suggest that S378 is an
important residue in the design of CACYP51 inhibitors.
Compounds with steric and hydrophobic substitutions on the
para-position of the phenyl (e.g. compound 5 and 22) also show
excellent antifungal activity, which is attributed to the hydro-
phobic and van der Waals interactions with L121, F233, and
M508. From the docking results, there remained a relatively
large space near the para-position of the phenyl group, and this
area is at the end of ligand access channel 2 (FG loop) lined
with hydrophobic residues such as L87, L88, L121, P230, I231,
F233, V234, and M508 (Figure 10). Therefore, the side chain
of azoles in the data set could be extended to form stronger
hydrophobic and van der Waals interactions with the active site
of CACYP51. Compared to compounds 23, 25, and 26,
compounds 30, 31, and 33 show higher antifungal activity.
It is because the addition of a halogen on the phenyl can form
a stronger hydrophobic interaction with the active site of
CACYP51. When the benzyl groups of compounds 22-40
are changed into benzoyl groups (compounds 9-21), their
antifungal activity is decreased dramatically. From the dock-
ing results, we found that a hydrogen bond was formed be-
tween the carbonyl group and the backbone of F380, which
resulted in the moving of phenyl group away from Y118 and
Figure 10. Docking of compound 22 into the active site of CACYP51
and the residues within 6 Å from the ligand are shown.
the π-π interaction between them was lost. Correspondingly,
the interaction energies, especially LUDI score 2,48 were
decreased.
Design, Synthesis, and Antifungal Activities of Novel
Azoles Based on 3D-QSAR and Docking. 3D-QSAR and
molecular docking are two complementary methods, and the
combination of them is especially useful when the high-
resolution ligand-receptor structure is unavailable. In the
present study, the crystal structure of CACYP51 has not been
solved until now, and the combination of 3D-QSAR and
molecular docking studies can provide precise information to
guide the drug design. Based on the findings from 3D-QSAR
and docking studies, a pharmacophore model was generated to

2520 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
Chart 2
Chart 3
guide the further rational optimization of the compounds in data
set (Chart 2). In this model, we mainly focused on the
optimization of the side chain attached to C3. First, a linker is
necessary to restrict the side chain in proper conformation and
adjust the physicochemical properties of the whole molecule.
Second, a phenyl group is a suitable attachment to the linker
because it can form a π-π interaction with Y118. Third, a
hydrogen bond acceptor is favored at the para-position of the
phenyl, which could form hydrogen bonding interaction with
S378. Last, a steric and hydrophobic group is necessary because
it could form hydrophobic and van der Waals interactions with
the corresponding residues in the ligand access channel 2 (FG
Loop). Therefore, a three-step optimization of the azole anti-
fungal agents was performed on the basis of the present
pharmacophore model (Chart 3). During this process, both
predicted activities from 3D-QSAR and interaction energies
from molecular docking were considered in determining the
compounds for synthesis.
surrounding A117, L376, and I379. A total of 13 compounds
with benzylamine side chains (compounds 45-57) were syn-
thesized. In vitro antifungal activity assay indicated that their
antifungal activities were higher than those of corresponding
compounds with piperazinyl groups (Tables 4 and 5). Moreover,
these compounds showed a broader antifungal spectrum with
good activity against both systemic pathogenic fungi (e.g. C.
albicans, Cryptococcus neoformans, A. fumigatus) and der-
matophytes, (e.g. Trichophyton rubrum and Trichophyton men-
tagrophytes). Especially, compound 53 showed excellent activity
against most of the tested pathogenic fungi, and it was a good
candidate for developing novel azole antifungal agents with
broad spectrum.
Second, we extended the side chain of the benzylamine
analogues on the basis of the pharmacophore model. Amide
was selected as the hydrogen bond acceptor to be attached to
the phenyl group, which was designed to form a hydrogen bond
with S378. Then, we chose substituted phenyl groups as steric
and hydrophobic groups to attach the amide group in order to
form additional hydrophobic and van der Waals interaction with
the residues in the ligand access channel 2 (FG loop). A total
of 10 amide-benzylamine compounds (compounds 58-67) were
synthesized and tested for in vitro antifungal activity. Most of
them showed improved antifungal activity in comparison with
the benzylamine compounds (Table 6), which was attributed to
the additional hydrogen bonding and hydrophobic interaction
with the target enzyme. Most of the compounds showed better
activity against C. albicans and Crypt. neoformans than flu-
First, we attempted to change the linker between the C3 and
the phenyl group. Docking results revealed that the piperazinyl
group was not the best choice and a smaller linker would be
more favorable for the π-π interaction of the phenyl group
with Y118. After testing various linkers based on 3D-QSAR
predictive values and the interaction energy with CACYP51,
we chose the -N-CH3 group for synthesis because the
improved flexibility of the molecule allows the phenyl group
to more easily lock its proper position, and the methyl group
could form additional van der Waals interactions with the

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2521
a
Table 4. Antifungal in Vitro Activities of the Benzylamine Compounds (MIC80, µg·mL-1
)
compd
C. alb.
C. par.
C. tro.
C. neo.
A. fum.
F. ped.
T. rub.
S. sch.
45
46
47
48
49
50
51
52
53
54
55
56
57
Flu
Ket
Ter
0.032
2
0.032
0.032
0.5
0.032
0.032
1
0.032
0.016
0.031
0.032
0.0625
0.25
e0.125
2
e0.125
0.25
e0.125
e0.125
0.25
e0.125
e0.125
e0.125
e0.125
e0.125
1
0.25
0.5
0.5
e0.125
32
0.25
0.25
e0.125
e0.125
0.5
e0.125
e0.125
e0.125
e0.125
e0.125
0.25
0.5
8
16
e0.125
4
0.5
4
4
4
8
0.5
8
8
e0.125
4
8
64
64
32
1
0.25
1
0.25
2
e0.125
0.25
1
0.5
32
0.25
e0.125
e0.125
e0.125
e0.125
0.5
8
16
8
e0.125
e0.125
32
64
64
64
64
64
32
64
16
32
64
64
64
>64
0.5
2
0.25
e0.125
e0.125
e0.125
0.25
1
0.25
1
2
1
e0.125
0.25
e0.125
e0.125
e0.125
0.125
1
4
64
e0.125
e0.125
4
2
0.5
e0.125
16
0.0625
16
0.25
a Abbreviations: C. alb., Candida albicans; C. par., Candida paropsilosis; C. tro., Candida tropicalis; C. neo., Cryptococcus neoformans; A. fum.,
Aspergillus fumigatus; F. ped., Fonsecaea pedrosoi; T. rub., Trichophyton rubrum; S. sch., Sporothrix schenckii; Flu, fluconazole; Ter, terbinafine; Ket,
ketoconazole.
a
Table 5. Antifungal in Vitro Activities of Benzylamine Compounds against Dermatophytes (MIC80, µg·mL-1
)
compd
M. gyp.
T. Vio.
M. lan.
T. men.
E. flo.
45
47
48
49
50
51
52
53
54
Flu
Ket
Ter
0.5
32
32
8
32
4
0.125
<0.0625
16
0.25
2
4
2
2
0.5
<0.0625
<0.0625
4
32
0.25
4
<0.0625
<0.0625
0.125
0.5
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
4
8
0.5
1
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
<0.0625
>32
<0.0625
>32
0.25
<0.0625
<0.0625
8
8
>32
>32
<0.0625
>32
<0.0625
<0.0625
<0.0625
32
a Abbreviations: M. gyp., Microsporum gypseum; T. Vio., Trichophyton Violaceum; M. lan, Microsporum lanosum; T. men., Trichophyton mentagrophytes;
E. flo., Epidermophyton floccosum.
a
Table 6. Antifungal in Vitro Activities of the Amide-Benzylamine Compounds (MIC80, µg·mL-1
)
compd
C. alb.
C. par.
C. tro.
C. neo.
A. fum.
F. ped.
T. rub.
M. lan.
58
59
60
61
62
63
64
65
66
67
Flu
Ket
Itz
Ter
0.031
0.016
0.0625
0.016
0.031
0.016
0.0625
0.031
0.125
0.031
0.0625
0.125
16
0.5
0.25
0.016
0.5
0.0625
0.25
0.125
0.016
0.5
0.016
1
0.5
4
0.25
0.5
0.016
1
1
0.25
0.5
0.5
1
0.0625
0.5
e0.031
2
0.125
4
0.5
2
32
32
>64
32
16
16
64
>64
32
16
1
0.5
e0.125
0.5
e0.125
e0.125
0.5
0.25
e0.125
4
4
8
4
2
0.25
0.5
8
e0.125
0.25
0.25
0.5
e0.125
0.5
e0.125
2
e0.125
32
16
0.0625
0.0625
0.008
0.25
0.0625
0.125
2
e0.125
2
0.25
64
4
1
e0.125
64
4
>64
0.0625
0.125
2
1
0.25
0.5
e0.125
1
>16
>16
1
e0.125
e0.125
e0.125
a Abbreviations: C. alb., Candida albicans; C. par., Candida paropsilosis; C. tro., Candida tropicalis; C. neo., Cryptococcus neoformans; A. fum.,
Aspergillus fumigatus; F. ped., Fonsecaea pedrosoi; T. rub., Trichophyton rubrum; M. lan, Microsporum lanosum; Flu, fluconazole; Ket, ketoconazole; Itz,
itraconazole; Ter, terbinafine.
conazole, itraconazole, and ketoconazole, with their MIC80
values in the range of 0.125 µg/mL to 0.008 µg/mL. Compounds
59, 61, 63, and 67 exhibited strong in vitro antifungal activity
with broad antifungal spectrum, which was worthy of further
evaluation.
Third, we chose the triazolone group to replace the amide
group as a hydrogen bond acceptor because triazolone group
can not only form a hydrogen bond with S378 but also can
adjust the phsicochemical properties of the molecules and
improve the water solubility. Various substituted benzyl groups
and substituted 2-oxophenethyl groups were attached to the
triazolone group to form hydrophobic interaction with the
residues in the ligand access channel 2 (FG loop). These
compounds showed better 3D-QSAR predictive activities and
interaction energies with CACYP51 than the corresponding
benzylamine or amide-benzylamine compounds. Therefore, we
synthesized 34 new azoles with triazolone-benzylamine side
chains (compounds 76-109). In vitro antifungal assay revealed
that their antifungal activities were improved again (Table 7),
and most of them showed the best antifungal activities among
the synthesized azoles with their MIC80 values against C.
albicans in the range of 0.004 µg/mL to 0.001 µg/mL.
Compound 94 was the most active among the designed
compounds with MIC80 value of 0.001 µg/mL against C.
albicans, and further pharmacological and toxicological evalu-
ation is in progress.

2522 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
S. sch.
Table 7. Antifungal in Vitro Activities of the Triazolone-Benzylamine Compounds (MIC80, µg·mL-1
)
a
compd
C. alb.
C. par.
C. tro.
C. neo.
A. fum.
F. ped.
T. rub.
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
Flu
Ket
Ter
0.002
0.0625
0.004
0.031
0.031
0.004
0.016
0.008
0.002
0.002
0.031
0.031
0.002
0.002
0.004
0.002
0.004
0.002
0.001
0.002
0.002
0.031
0.0625
0.002
0.002
0.25
0.125
0.25
4
4
2
0.5
0.0625
1
1
16
4
2
1
1
32
64
64
8
32
32
8
8
2
4
64
2
0.25
4
1
1
16
8
0.5
1
4
0.5
0.5
0.25
0.5
16
0.25
0.25
0.25
0.5
0.5
1
64
64
32
32
64
32
32
64
4
1
0.25
0.0625
0.5
0.125
0.25
0.125
0.125
0.5
0.0625
0.0625
0.0625
1.25
0.125
0.25
0.25
0.0625
0.125
0.125
4
0.0625
0.125
0.5
0.06
4
0.5
0.5
0.008
0.5
64
0.25
0.25
0.008
1
0.125
0.125
0.5
0.25
0.125
0.016
16
0.008
0.008
0.008
0.016
0.031
0.008
0.008
0.008
0.008
0.008
0.008
0.008
0.008
0.008
1
0.25
0.0625
0.0625
1
64
64
64
64
64
64
64
64
16
16
64
32
64
64
64
64
64
64
64
64
64
64
64
64
64
>64
0.5
2
1
0.0625
0.0625
0.125
0.0625
0.125
0.25
0.0625
0.0625
0.03125
0.125
0.5
2
8
16
4
16
32
8
16
8
64
64
16
8
64
64
64
16
32
32
64
32
64
32
1
2
0.5
0.5
0.25
32
64
1
0.5
32
4
2
0.25
4
2
32
2
0.008
0.125
2
0.03
0.06
1
0.03
0.06
0.008
0.5
0.016
0.5
0.03125
0.0625
0.5
0.125
0.5
16
2
1
0.25
8
1
8
64
0.125
0.0625
8
0.25
0.0625
0.0625
1
0.125
1
1
0.004
0.004
0.002
0.016
0.004
0.016
0.125
0.008
0.25
0.0625
0.0625
0.0625
0.125
0.125
0.0625
0.0625
0.0625
0.5
1
8
64
4
8
16
0.031
4
2
64
0.0625
e0.125
8
0.0625
16
0.031
16
0.0625
32
0.0625
e0.125
0.25
a Abbreviations: C. alb., Candida albicans; C. par., Candida paropsilosis; C. tro., Candida tropicalis; C. neo., Cryptococcus neoformans; A. fum.,
Aspergillus fumigatus; F. ped., Fonsecaea pedrosoi; T. rub., Trichophyton rubrum; S. sch., Sporothrix schenckii; Flu, fluconazole; Ter, terbinafine; Ket,
ketoconazole.
Conclusion
the powerfulness of the combination of 3D-QSAR and molecular
docking in the optimization of azole antifungal agents. Several
compounds (such as 53, 67, and 94) showed strong in vitro
antifungal activity against most of the tested pathogenic fungi.
However, the in vitro antifungal activity of the azoles does not
always correlate well with their in vivo efficiency because many
azoles are very hydrophobic and their bioavailability is relatively
low. Thus, only when our synthesized azoles have good
pharmacokinetic and toxicological profiles can they be devel-
oped into novel triazole antifungal agents. Further pharmaco-
logical and toxicological evaluation of these promising com-
pounds is in progress. Our study demonstrates the efficiency
of 3D-QSAR and molecular docking not only for understanding
the molecular basis of drug-enzyme interaction but also for
the discovery and optimization of the lead compounds.
In response to the emergent need for novel azole antifungal
agents with improved activity and broader antifungal spectrum,
we sought to understand the necessary structural features for
antifungal activities and then to design new potent inhibitors
by use of a structure-based strategy. In the present study, we
have established predictive CoMFA and CoMSIA 3D-QSAR
models for a set of newly synthesized triazole antifungal
compounds from our laboratory. The best derived CoMFA and
CoMSIA models showed a predictive q2 value of 0.718 and
0.655, and the activities of compounds in the data set were
predicted with high accuracy. Moreover, molecular docking was
used to clarify the binding mode of the azoles in the data set.
The side chain attached to C3 had a major impact on antifungal
activity, which was found to be located in the ligand access
channel 2 in the active site of CACYP51. Y118 and S378 could
form specific π-π interactions and hydrogen bonding interac-
tions with the inhibitors, which were very important for drug
design. The results obtained from 3D-QSAR and molecular
docking yielded reliable clues on how to further optimize the
azoles in the data set. Therefore, a receptor-based pharmacoph-
ore model for the rational optimization of azole antifungal agents
was established. A three-step optimization process was per-
formed based on this model with 3D-QSAR predictive values
and interaction energies as criteria to select compounds for
synthesis. A total of 57 novel azoles with three different types
of side chains were designed and synthesized, and their
antifungal activities were greatly improved, which confirmed
Experimental Section
Chemistry. General Methods. Melting points were measured
on an electrically heated XT4A instrument and are uncorrected.
Infrared spectra (IR) were recorded on a Brucker Vector II
instrument. Mass spectra (MS) were measured on a Micromass
Qtof-Micro LC-MS instrument. Nuclear magnetic resonance (NMR)
spectra were recorded on a Bruker 500 spectrometer with TMS as
an internal standard and CDCl3 as solvent. Elemental analyses were
performed with a MOD-1106 instrument and were consistent with
theoretical values within (0.4%. Silica gel thin-layer chromatog-
raphy was performed on precoated plates GF254 (Qindao Haiyang
Chemical, China). Silica gel column chromatography was performed

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2523
with Silica gel 60 G (Qindao Haiyang Chemical, China). Com-
mercial solvents were used without any pretreatment.
Methyl-{4-[5-oxo-1-(4-trifluoromethyl-benzyl)-1,5-dihydro-
[1,2,4]triazol-4-yl]-benzyl}-carbamic Acid tert-Butyl Ester (74).
A mixture of compound 73 (1.5 g, 0.005 mol), 4-trifluoromethyl-
benzyl chloride (2.9 g, 0.015 mol), and potassium carbonate (1.4
g, 0.01 mol) in DMF (15 mL) was heated at 80 °C for 18 h. The
solvent was removed under reduced pressure. The residue was
diluted with H2O (50 mL) and extracted with ethyl acetate (50 mL
× 3). The combined organic layers were washed with H2O (100
mL × 3) and brine (100 mL × 3), dried over anhydrous MgSO4,
and filtrated, and the solvent was evaporated under reduced pressure.
The resulting product was purified by silica gel column chroma-
tography gradient elution, EtOAc/hexane 1/4 toward 1/1 to give
1-(1H-1,2,4-triazole-1-yl)-2-(2,4-difluorophenyl)-3-[N-methyl-
N-(3-chlorobenzyl)-amino]-2-propanol (45). The mixture of ep-
oxide 44 (3.3 g, 0.01 mol), MeOH (20 mL), NaOH (0.40 g, 0.01
mol), and 3-chloro-N-methyl-benzylamine (1.90 g, 0.012 mol) was
stirred under the refluxing condition for 8 h. The solvent was
evaporated under reduced pressure. The residue was triturated with
H2O (50 mL) and extracted with ethyl acetate (100 mL × 3). The
combined organic layers were washed with H2O (100 mL × 3)
and brine (100 mL × 3), dried over anhydrous MgSO4, and filtrated,
and the solvent was evaporated under reduced pressure. The residue
was purified by silica gel column chromatography (EtOAc:hexane
1:1, v/ v) to give 45 (2.67 g, 67.9%) as a white solid: mp 82-83
°C. IR (film): ν (cm-1) ) 3106, 1615, 1518, 1503. 1H NMR
(CDCl3): δ 2.04 (s, 3H), 2.81 (d, J ) 13.2 Hz, 1H), 3.07 (d, J )
13.2 Hz, 1H), 3.40 (d-d, J1 ) 13.2 Hz, J2 ) 38.8 Hz, 2H), 4.47
(d-d, J1 ) 14.4 Hz, J2 ) 28.4 Hz, 2H), 6.74∼7.60 (m, 7H), 7.74
(s, 1H), 8.06 (s, 1H). MS (ESI) m/z: 393 (M + 1). Anal. (C19H19-
ClF2N4O) C, H, N.
1
74 (1.80 g, 78.3%) as a white solid. H NMR (CDCl3): 1.42 (s,
9H), 2.78 (s, 3H), 4.41 (s, 2H), 5.07 (s, 2H), 7.35∼7.74 (m, 8H),
8.48 (s, 1H). MS (ESI) m/z: 463 (M + 1). Anal. (C23H25F3N4O3)
C, H, N.
4-(4-Methylaminomethyl-phenyl)-2-(4-trifluoromethyl-benzyl)-
2,4-dihydro-[1,2,4]triazol-3-one (75). To a solution of compound
74 (1.8 g, 0.004mol) in EtOAc (25 mL) was added 20 mL of 10%
HCl. The resulting mixture was stirred at room temperature for 24
h. The solvent was evaporated under reduced pressure. The residue
was diluted with water (50 mL), basified with K2CO3, and extracted
with CHCl3 (50 mL × 3). The combined extract was washed with
water (50 mL × 3) and brine (50 mL×3), dried over anhydrous
MgSO4, and filtrated, and the solvent was evaporated under reduced
pressure to give compound 74 as a white solid. 1H NMR (CDCl3):
2.50 (s, 3H), 4.10 (s, 2H), 5.08 (s, 2H), 7.53∼7.80 (m, 8H), 8.55
(s, 1H). MS (ESI) m/z: 363 (M + 1). Anal. (C18H17F3N4O) C, H,
N.
4-[4-({[2-(2,4-Difluoro-phenyl)-2-hydroxy-3-[1,2,4]triazol-1-
yl-propyl]-methyl-amino}-methyl)-phenyl]-2-(4-methyl-benzyl)-
2,4-dihydro-[1,2,4]triazol-3-one (76). The mixture of epoxide 44
(1.1 g, 0.0033 mol), EtOH (20 mL), triethylamine (2 mL), and 4-(4-
methylaminomethyl-phenyl)-2-(4-methyl-benzyl)-2,4-dihydro-[1,2,4]-
triazol-3 -one (1.0 g, 0.0033 mol) was stirred under refluxed for 8
h. The solvent was evaporated under reduced pressure. The residue
was diluted with H2O (50 mL) and extracted with ethyl acetate
(50 mL × 3). The combined organic layers were washed with H2O
(100 mL × 3) and brine (100 mL × 3), dried over anhydrous
MgSO4, and filtrated, and the solvent was evaporated under reduced
pressure. The residue was purified by silica gel column chroma-
tography (CH2Cl2:MeOH 95:5, v/v) to give 76 (1.3 g, 74.1%) as a
white solid: mp 80-82 °C. IR (film): ν (cm-1) ) 3413, 3126,
3054, 1710, 1615, 1554, 1500. 1H NMR (CDCl3): δ 1.24 (s, 3H),
2.03 (s, 3H), 2.40 (d, 2H), 3.72 (d-d, J1 ) 7.0 Hz, J2 ) 14.0 Hz,
2H), 4.47 (d-d, J1 ) 14.2 Hz, J2 ) 33.7 Hz, 2H), 5.00 (d, 2H),
6.80∼7.76 (m, 13H), 8.06 (s, 1H). MS (ESI) m/z: 547 (M + 1).
Anal. (C29H29F2N7O2) C, H, N.
The synthetic methods for the following compounds 46-57 were
similar to the synthesis of compound 45.
N-[4-({[2-(2,4-Difluoro-phenyl)-2-hydroxy-3-[1,2,4]triazol-1-
yl-propyl]-methyl-amino}-methyl)-phenyl]-benzamide (58). A
solution of benzoyl chloride (0.14 g, 0.001 mol) in 1,4-dioxane
(10 mL) was added dropwise to a stirred mixture of compound 57
(0.37 g, 0.001 mol) and triethylamine (1.0 mL) in dioxane (15 mL)
at 0 °C under nitrogen flow. The mixture was stirred at room
temperature for 5 h, and the solvent was evaporated under reduced
pressure. The residue was purified by silica gel column chroma-
tography (CH2Cl2:MeOH 95:5, v/v) to give 58 (0.42 g, 67.9%) as
a white solid: mp 88-90 °C. IR (film): ν (cm-1) ) 3226, 3109,
1659, 1599, 1520, 1500, 1469. 1H NMR (CDCl3): δ 2.50 (s, 3H),
2.85 (d, J ) 4.2 Hz, 1H), 3.16 (d, J ) 4.7 Hz, 1H), 3.59 (d, J )
4.7 Hz, 1H), 3.85 (d, J ) 4.7 Hz, 1H), 4.55 (d-d, J1 ) 14.4 Hz,
J2 ) 20.0 Hz, 2H), 7.08∼8.90 (m, 14H), 10.30 (s, 1H). MS (ESI)
m/z: 478 (M + 1). Anal. (C26H25F2N5O2) C, H, N.
The synthetic methods for the following compounds 59-67 were
similar to the synthesis of compound 58.
{4-[(tert-Butoxycarbonyl-methyl-amino)-methyl]-phenyl}-car-
bamic Acid Phenyl Ester (71). Phenyl chloroformate (17.2 g, 0.11
mol) was added dropwise to a stirred mixture of compound 70 (23.6
g, 0.1 mol), pyridine (8.5 g, 0.11 mol), and EtOAC (200 mL) at 0
°C. The mixture was stirred for 3 h at room temperature, then
washed with water (200 mL × 3), dried over anhydrous MgSO4,
and filtrated. The solvent was evaporated under reduced pressure,
and the deposited crystals were collected and washed with hexane
1
to give 70 (34.8 g, 97.7%) as a white solid: mp 109-110 °C. H
The synthetic methods for the following compounds 77-109
were similar to the synthesis of compound 76.
NMR (CDCl3): 1.48 (s, 9H), 2.82 (s, 3H), 4.39 (s, 2H), 7.00 (br,
1H), 7.18∼7.42 (m, 9H). MS (ESI) m/z: 357 (M + 1). Anal.
(C20H24N2O4) C, H, N.
Computational Details. General Method. The crystallographic
coordinates of MTCYP51 in complex with fluconazole (0.22 Å
resolution, Rcryst ) 0.204) were obtained from the Brookheaven
Protein Databank as entries 1EA1. The 3D model of CYP51 from
C. albicans was constructed in our previous studies.25 All calcula-
tions were performed with the commercially available SYBYL6.951
software package and InsightII 200052 software package. All
calculations were performed on an Origin 300 server.
Data Sets. A total of 40 azole antifungal compounds from our
laboratory were used as a data set in the 3D-QSAR and docking
analysis. In the CoMFA and CoMSIA study, five compounds
(labeled with asterisks in Figure 1) were randomly selected as a
test set, which represented a range of antifungal activity similar to
that of a training set and was used to evaluate the predictive power
of the resulting models. The biological activity of each compound
was expressed as minimal inhibitory concentrations (MIC80) against
C. albicans, and -log(MIC80) was used for the 3D-QSAR analysis.
Molecular Modeling and Alignment. The 3D structures of all
compounds in the training set and test set were built from the
SYBYL fragment library. Energy minimization was performed
using the Tripos force field, Powell optimization method, and
4-(N-methyl-N-Boc-aminomethyl)phenylsemicarbazide (72).
A mixture of compound 71 (35.6 g, 0.1 mol), hydrazine hydrate
(10 mL), and dimethoxyethane (150 mL) was stirred for 24 h. The
solvent was evaporated under reduced pressure, and the residue
1
was washed with EtOAc to give 72 as a white solid. H NMR
(CDCl3): 1.48 (s, 9H), 2.81 (s, 3H), 3.84 (br, 2H), 4.36 (s, 2H),
6.76 (br, 1H), 7.16∼7.42 (m, 4H), 8.21 (s, 1H). MS (ESI) m/z:
295 (M + 1). Anal. (C14H22N4O3) C, H, N.
Methyl-[4-(5-oxo-1,5-dihydro-[1,2,4]triazol-4-yl)-benzyl]-car-
bamic Acid tert-Butyl Ester (73). A mixture of compound 72 (15.0
g, 0.05 mol), formamidine acetate (22.0 g, 0.2 mol), and DMF (250
mL) was stirred at room temperature for 30 min. AcOH (14.3 mL)
was added, and the resulting mixture was heated for 6 h at 80 °C.
The solvent was evaporated under reduced pressure, and the residue
was poured onto the ice water. After filtration, the residue was
recrystallized from EtOAC-hexane to give 73 (12.5 g, 82.2%) as
a pale yellow solid. 1H NMR (CDCl3): 1.42 (s, 9H), 2.74 (s, 3H),
4.40 (s, 2H), 7.33∼7.67 (m, 4H), 8.31 (s, 1H), 11.80 (br, 1H). MS
(ESI) m/z: 305 (M + 1). Anal. (C15H20N4O3) C, H, N.

2524 Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8
Sheng et al.
MAXIMIN2 minimizer with a convergence criterion 0.001 kcal/
mol·Å. Charges were calculated by the Gasteiger-Hu¨ckel method.
The docked conformation of each compound in the active site of
CACYP51 was used as pharmacophoric conformation for the 3D-
QSAR studies. Compound 3, one of the most active compounds,
was used as the alignment template. The reference atoms used for
alignment were three identified pharmacophoric points: N4 of the
azole ring, the centroid of the phenyl ring, and the oxygen attached
to C2.
CoMFA and CoMSIA Setup. The CoMFA descriptors, steric
(Lennard-Jones 6-12 potential) and electrostatic (Coulombic po-
tential) field energies, were calculated by the following param-
eters: an sp3 carbon probe atom with +1 charge and a van der
Waals radius of 1.52 Å, and energy cutoff of 30 kcal/mol. CoMSIA
similarity indices descriptors (steric, electrostatic, hydrophobic,
H-bond donor, and H-bond acceptor fields) were calculated using
a C1+ probe atom with a radius of 1.0 Å placed at grid spacing of
2 Å. CoMSIA similarity indices (AF) for a molecule j with atoms
i at a grid point q are calculated by eq 1:
accepted on the basis of energy check, which used the Metropolis
criterion, and also a check of rms distance of the new structure
versus the structure found so far. The final conformation was
obtained through a simulation annealing procedure from 500 K to
300 K, and then 5000 rounds of energy minimization were
performed to reach a convergence, where the resulting interaction
energy values were used to define a rank order. Each energy-
minimized final docking position of the compounds was evaluated
using the interaction score function in the LUDI54 module.
In Vitro Antifungal Activity Assay. In vitro antifungal activity
was measured by means of the minimal inhibitory concentrations
(MIC) using the serial dilution method in 96-well microtest plates.
Test fungal strains were obtained from the ATCC or were clinical
isolates. The MIC determination was performed according to the
National Committee for Clinical Laboratory Standards (NCCLS)
recommendations with RPMI 1640 (Sigma) buffered with 0.165
M MOPS (Sigma) as the test medium. The MIC value was defined
as the lowest concentration of test compounds that resulted in a
culture with turbidity less than or equal to 80% inhibition when
compared with the growth of the control. Test compounds were
dissolved in DMSO serially diluted in growth medium. The yeasts
were incubated at 35 °C and the dermatophytes at 28 °C. Growth
MIC was determined at 24 h for Candida species, at 72 h for Crypt.
neoformans, and at 7 days for filamentous fungi.
2
AF,Kq(j) ) - ωprobe′kωik e-Rr
(1)
iq
∑
where k represents the following physicochemical properties: steric,
electrostatic, hydrophobic, H-bond donor, and H-bond acceptor. A
Gaussian type distance dependence was used between the grid point
q and each atom i of the molecule. Here, steric indices are related
to the third power of the atomic radii, electrostatic descriptors are
derived from atomic partial charges, hydrophobic fields are derived
from atom-based parameters, and H-bond donor and acceptor
indices are obtained by a rule-based method based on experimental
results. The attenuation factor (R) was set to the default value of
0.3.
The CoMFA and CoMSIA descriptors were used as independent
variables, and pMIC80 values were used as dependent variables in
partial least squares (PLS)53 regression analyses to derive 3D-QSAR
models using the standard implementation in the SYBYL package.
To obtain the optimum number of principle components, the leave-
one-out (LOO) cross-validation53 was utilized, which is used to be
a measure of how good the model represents the data in the training
set. The cross-validated coefficient, q2, was calculated using eq 2:
Acknowledgment. We thank Professor Yuanyin Jiang of
The Second Military Medical University for determining in vitro
antifungal activities. This work was supported by the National
Natural Science Foundation of China (Grant No. 30430750).
Supporting Information Available: IR, NMR, MS, and
elemental analysis data for the compounds in this study. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Fridkin, S. K.; Jarvis, W. R. Epidemiology of nosocomial fungal
infections. Clin. Microbiol. ReV. 1996, 9, 499-511.
(2) Latge, J. P. Aspergillus fumigatus and aspergillosis. Clin. Microbiol.
ReV. 1999, 12, 310-350.
(3) Steenbergen, J. N.; Casadevall, A. Prevalence of Cryptococcus
neoformans var. neoformans (Serotype D) and Cryptococcus neo-
formans var. grubii (Serotype A) isolates in New York City. J. Clin.
Microbiol. 2000, 38, 1974-1976.
2
(Ypredicted - Yobserved
)
∑
∑
q2 ) 1 -
(2)
2
(4) Sheehan, D. J.; Hitchcock, C. A.; Sibley, C. M. Current and emerging
azole antifungal agents. Clin. Microbiol. ReV. 1999, 12, 40-79.
(5) Gallis, H. A.; Drew, R. H.; Pickard, W. W. Amphotericin B: 30
years of clinical experience. ReV. Infect. Dis. 1990, 12, 308-329.
(6) Arikan, S.; Rex, J. H. Nystatin LF (Aronex/Abbott). Curr. Opin.
InVest. Drugs 2001, 2, 488-495.
(Yobserved - Ymean
)
where Ypredicted, Yobserved, and Ymean are predicted, actual (observed),
and mean values of the target property (pMIC80), respectively.
Conventional correlation coefficient r2 and its standard error (s)
were also computed for the final PLS models. To graphically
interpret the 3D-QSAR results in terms of field contributions,
isocontour maps were generated using the field type “stdev*coeff”
and the contour levels were set to default values.
Docking Analysis. Fluconazole and the antifunal azoles in the
data set were manually docked into the active site of CACYP51
on the basis of the binding mode of fluconazole with MTCYP51
and our previous docking studies.17 First, the N4 atom of the azole
ring was initially placed at 2.34 Å from the heme iron, and the
distance was the same as that in the crystal structure of MTCYP51
in complex with fluconazole. Second, the difluorophenyl group was
placed at the hydrophobic pocket lined with F126, L139, P145,
I304, and M306. Third, the relatively reasonable positions of the
azoles were obtained by monitoring the interaction energy between
the ligands and CACYP51.
Then, the flexible ligand docking procedure in the Affinity
module within InsightII was used to define the lowest energy
position for the substrate using a Monte Carlo docking protocol.
All the atoms within a defined radius (8 Å) of the substrate were
allowed to move. The solvation grid supplied with the affinity
Program was used. If the resulting substrate/enzyme system was
within a predefined energy tolerance of the previous structure, the
system was subjected to minimization. The resulting structure was
(7) Denning, D. W. Echinocandins: a new class of antifungal. J.
Antimicrob. Chemother. 2002, 49, 889-891.
(8) Birnbaum, J. E. Pharmacology of the allylamines. J. Am. Acad.
Dermatol. 1990, 23, 782-785.
(9) Cha, R.; Sobel, J. D. Fluconazole for the treatment of candidiasis:
15 years experience. Expert. ReV. Anti-Infect. Ther. 2004, 2, 357-
366.
(10) Hoffman, H. L.; Ernst, E. J.; Klepser, M. E. Novel triazole antifungal
agents. Expert Opin. InVest. Drugs 2000, 9, 593-605.
(11) Casalinuovo, I. A.; Di Francesco, P.; Garaci, E. Fluconazole resistance
in Candida albicans: a review of mechanisms. Eur. ReV. Med.
Pharmacol. Sci. 2004, 8, 69-77.
(12) Chandrasekar, P. H.; Manavathu, E. Voriconazole: A second-
generation triazole. Drugs Today (Barc.) 2001, 37, 135-148.
(13) Herbrecht, R. Posaconazole: a potent, extended-spectrum triazole
anti-fungal for the treatment of serious fungal infections. Int. J. Clin.
Pract. 2004, 58, 612-624.
(14) Arikan, S.; Rex, J. H. Ravuconazole Eisai/Bristol-Myers Squibb. Curr.
Opin. InVest. Drugs 2002, 3, 555-561.
(15) Aoyama, Y.; Yoshida, Y.; Sato, R. Yeast cytochrome P-450
catalyzing lanosterol 14R-demethylation. II. Lanosterol metabolism
by purified P-450(14)DM and by intact microsomes. J. Biol. Chem.
1984, 259, 1661-1666.
(16) Lamb, D. C.; Kelly, D. E.; Venkateswarlu, K.; Manning, N. J.; Bligh,
H. F.; Schunck, W. H.; Kelly, S. L. Generation of a complete, soluble,
and catalytically active sterol 14R-demethylase-reductase complex.
Biochemistry 1999, 38, 8733-8738.

Optimization of Azole Antifungal Agents
Journal of Medicinal Chemistry, 2006, Vol. 49, No. 8 2525
(17) Ji, H.; Zhang, W.; Zhou, Y.; Zhang, M.; Zhu, J.; Song, Y.; Lu, J. A
three-dimensional model of lanosterol 14R-demethylase of Candida
albicans and its interaction with azole antifungals. J. Med. Chem.
2000, 43, 2493-2505.
(18) Ravichandran, K. G.; Boddupalli, S. S.; Hasermann, C. A.; Peterson,
J. A.; Deisenhofer, J. Crystal structure of hemoprotein domain of
P450BM-3, a prototype for microsomal P450’s. Science 1993, 261,
731-736.
(19) Poulos, T. L.; Finzel, B. C.; Howard, A. J. High-resolution crystal
structure of cytochrome P450cam. J. Mol. Biol. 1987, 195, 687-
700.
(20) Hasemann, C. A.; Ravichandran, K. G.; Peterson, J. A.; Deisenhofer,
J. Crystal structure and refinement of cytochrome P450terp at 2.3 A
resolution. J. Mol. Biol. 1994, 236, 1169-1185.
(21) Cupp-Vickery, J. R.; Poulos, T. L. Structure of cytochrome P450eryF
involved in erythromycin biosynthesis. Nat. Struct. Biol. 1995, 2,
144-153.
(22) Ji, H.; Zhang, W.; Zhang, M.; Kudo, M.; Aoyama, Y.; Yoshida, Y.;
Sheng, C.; Song, Y.; Yang, S.; Zhou, Y.; Lu, J.; Zhu, J. Structure-
based de novo design, synthesis, and biological evaluation of non-
azole inhibitors specific for lanosterol 14R-demethylase of fungi. J.
Med. Chem. 2003, 46, 474-485.
(23) Podust, L. M.; Poulos, T. L.; Waterman, M. R. Crystal structure of
cytochrome P450 14R-sterol demethylase (CYP51) from Mycobac-
terium tuberculosis in complex with azole inhibitors. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 3068-3073.
(24) Podust, L. M.; Yermalitskaya, L. V.; Lepesheva, G. I.; Podust, V.
N.; Dalmasso, E. A.; Waterman, M. R. Estriol bound and ligand-
free structures of sterol 14R-demethylase. Structure. (Camb.) 2004,
12, 1937-1945.
(25) Sheng, C.; Zhang, W.; Zhang, M.; Song, Y.; Ji, H.; Zhu, J.; Yao, J.;
Yu, J.; Yang, S.; Zhou, Y.; Zhu, J.; Lu, J. Homology modeling of
lanosterol 14R-demethylase of Candida albicans and Aspergillus
fumigatus and insights into the enzyme-substrate Interactions. J.
Biomol. Struct. Dyn. 2004, 22, 91-99.
(26) Sheng, C. Q.; Zhu, J.; Zhang, W. N.; Song, Y. L.; Zhang, M.; Ji, H.
T.; Yu, J. X.; Yao, J. Z.; Yang, S.; Miao, Z. Y. Synthesis and
antifungal activity of novel triazole antifungal agents. Yaoxue Xuebao
2004, 39, 984-989.
(27) Sheng, C. Q.; Zhang, W. N.; Ji, H. T.; Song, Y. L.; Yang, S.; Zhou,
Y. J.; Zhu, J.; Lu, J. G. Synthesis and antifungal activity of 1-(1,2,4-
triazolyl-1H-1-yl)-2-(2,4-diflurophenyl)-3-(4-substituted benzyl-1-
piperazinyl)-2-propanols. Yaoxue Xuebao 2003, 38, 665-670.
(28) Fratev, F.; Benfenati, E. 3D-QSAR and molecular mechanics study
for the differences in the azole activity against yeastlike and
filamentous fungi and their relation to P450DM inhibition. 1.
3-substituted-4(3H)-quinazolinones. J. Chem. Inf. Model. 2005, 45,
634-644.
(29) Kidwai, M.; Venkatarmanan, R.; Rastogi, S.; Sapra, P. Discovery
and development of antifungal compounds. Curr. Med. Chem. - Anti-
Infect. Agents 2003, 2, 27-71.
(30) Liu, J.; Pan, D.; Tseng, Y.; Hopfinger, A. J. 4D-QSAR analysis of
a series of antifungal p450 inhibitors and 3D-pharmacophore
comparisons as a function of alignment. J. Chem. Inf. Comput. Sci.
2003, 43, 2170-2179.
(36) Lewis, D. F.; Wiseman, A.; Tarbit, M. H. Molecular modelling of
lanosterol 14R-demethylase (CYP51) from Saccharomyces 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.
(37) Xiao, L.; Madison, V.; Chau, A. S.; Loebenberg, D.; Palermo, R.
E.; McNicholas, P. M. Three-dimensional models of wild-type and
mutated forms of cytochrome P450 14R-sterol demethylases from
Aspergillus fumigatus and Candida albicans provide insights into
posaconazole binding. Antimicrob. Agents Chemother. 2004, 48,
568-574.
(38) Gollapudy, R.; Ajmani, S.; Kulkarni, S. A. Modeling and interactions
of Aspergillus fumigatus lanosterol 14-Rdemethylase ‘A’ with azole
antifungals. Bioorg. Med. Chem. 2004, 12, 2937-2950.
(39) Cramer, R. D.; Patterson, D. E.; Bunce, J. D. Comparative molecular
field analysis (CoMFA). 1. Effect of shape on binding of steroids to
carrier proteins. J. Am. Chem. Soc. 1988, 110, 5959-5967.
(40) Klebe, G.; Abraham, U.; Mietzner, T. Molecular similarity indices
in a comparative analysis (CoMSIA) of drug molecules to correlate
and predict their biological activity. J. Med. Chem. 1994, 37, 4130-
4146.
(41) Sheng, C.; Chen, Y.; Zhang, W.; Song, Y.; Ji, H.; Zhou, Y.; Zhu, J.;
Lu, J. Optimization of the synthesis of the key intermediateof
fluconazole by the orthogonal experimental design. Zhongguo Yaowu
Huaxue Zazhi 2002, 12, 344-346.
(42) Singer, G. M.; Andrews, A. W.; Guo, S. M. Quantitative structure-
activity relationship of the mutagenicity of substituted N-nitroso-N-
benzylmethylamines: possible implications for carcinogenicity. J.
Med. Chem. 1986, 29, 40-44.
(43) Bringmann, G.; Rummey, C. 3D QSAR investigations on antimalarial
naphthylisoquinoline alkaloids by comparative molecular similarity
indices analysis (CoMSIA), based on different alignment approaches.
J. Chem. Inf. Comput. Sci. 2003, 43, 304-316.
(44) Bohm, M.; St rzebecher, J.; Klebe, G. Three-dimensional quantitative
structure-activity relationship analyses using comparative molecular
field analysis and comparative molecular similarity indices analysis
to elucidate selectivity differences of inhibitors binding to trypsin,
thrombin, and factor Xa. J. Med. Chem. 1999, 42, 458-477.
(45) Bush, B. L.; Nachbar, R. B., Jr. Sample-distance partial least
squares: PLS optimized for many variables, with application to
CoMFA. J. Comput.-Aided Mol. Des. 1993, 7, 587-619.
(46) Luty, B. A.; Wasserman, Z. R.; Stouten, P. F. W.; Hodge, C. N.;
Zacharias, M.; McCammon, J. A. A molecular mechanics/grid method
for evaluation of ligand-receptor interactions. J. Comput. Chem.
1995, 16, 454-464.
(47) Bohm, H. J. The development of a simple empirical scoring function
to estimate the binding constant for a protein-ligand complex of
known three-dimensional structure. J. Comput.-Aided Mol. Des. 1994,
8, 243-256.
(48) Bohm, H. J. Prediction of binding constants of protein ligands: a
fast method for the prioritization of hits obtained from de novo design
or 3D database search programs. J. Comput.-Aided Mol. Des. 1998,
12, 309-323.
(49) Macchiarulo, A.; Costantino, G.; Fringuelli, D.; Vecchiarelli, A.;
Schiaffella, F.; Fringuelli, R. 1,4-Benzothiazine and 1,4-benzoxazine
imidazole derivatives with antifungal activity: a docking study.
Bioorg. Med. Chem. 2002, 10, 3415-3423.
(50) Lepesheva, G. I.; Virus, C.; Waterman, M. R. Conservation in the
CYP51 family. Role of the B′ helix/BC loop and helices F and G in
enzymatic function. Biochemistry 2003, 42, 9091-9101.
(51) SYBYL 6.9: Tripos Associates, Inc. 1699S. Hanley Road, Suite 303,
St. Louis, MO 63144. 2003.
(31) 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.
(32) Talele, T. T.; Kulkarni, V. M. Three-dimensional quantitative
structure-activity relationship (QSAR) and receptor mapping of
cytochrome P-450(14 R DM) inhibiting azole antifungal agents. J.
Chem. Inf. Comput. Sci. 1999, 39, 204-210.
(33) Talele, T. T.; Kulkarni, S. S.; Kulkarni, V. M. Development of
pharmacophore alignment models as input for comparative molecular
field analysis of a diverse set of azole antifungal agents. J. Chem.
Inf. Comput. Sci. 1999, 139, 958-966.
(34) Fukuoka, T.; Johnston, D. A.; Winslow, C. A.; de Groot, M. J.; Burt,
C.; Hitchcock, C. A.; Filler, S. G. Genetic basis for differential
activities of fluconazole and voriconazole against Candida krusei.
Antimicrob. Agents Chemother. 2003, 47, 1213-1219.
(35) Holtje, H. D.; Fattorusso, C. Construction of a model of the Candida
albicans lanosterol 14-R-demethylase active site using the homology
modelling technique. Pharm. Acta HelV. 1998, 72, 271-277.
(52) InsightII 2000: Molecular Simulation Inc. 9685 Scranton Road, San
Diego, CA 92121-3752. 1999.
(53) Cramer, R. D.; Bunce, J. D.; Patterson, D. E. Crossvalidation,
bootstrapping and partial least squares compared with multiple
regression in conventional QSAR studies. QSAR 1988, 7, 18-25.
(54) Bohm, H. J. The computer program LUDI: a new method for the
de novo design of enzyme inhibitors. J. Comput.-Aided Mol. Des.
1992, 6, 61-78.
JM051211N