Improved model of lanosterol 14α-demethylase by ligand-supported homology modeling: Validation by virtual screening and azole optimization

Article abstract of DOI:10.1002/cmdc.200900468

Lanosterol 14α-demethylase (CYP51) is an important target for antifungal drugs. An improved three-dimensional model of CYP51 from Candida albicans (CACYP51) was constructed by ligand-supported homology modeling and molecular dynamics simulations. The accuracy of the constructed model was evaluated by its performance in a small-scale virtual screen. The results show that known CYP51 inhibitors were efficiently discriminated by the model, and it performed better than our previous CACYP51 model. The active site of CACYP51 was characterized by multiple copy simultaneous search (MCSS) calculations. On the basis of the MCSS results, a series of novel azoles were designed and synthesized, and they showed good in vitro antifungal activity with a broad spectrum. The MIC₈₀ value of four of these compounds against C. albicans is 0.001 μgmL^-1, indicating that they are promising leads for the discovery of novel antifungal agents.

Full text of DOI:10.1002/cmdc.200900468

MED
DOI: 10.1002/cmdc.200900468
Improved Model of Lanosterol 14a-Demethylase by
Ligand-Supported Homology Modeling: Validation by
Virtual Screening and Azole Optimization
Chunquan Sheng,[a] Wenya Wang,[a] Xiaoying Che,[a] Guoqiang Dong,[a] Shengzheng Wang,[a]
Haitao Ji,[b] Zhenyuan Miao,[a] Jianzhong Yao,[a] and Wannian Zhang*[a]
Lanosterol 14a-demethylase (CYP51) is an important target for
antifungal drugs. An improved three-dimensional model of
CYP51 from Candida albicans (CACYP51) was constructed by
ligand-supported homology modeling and molecular dynamics
simulations. The accuracy of the constructed model was evalu-
ated by its performance in a small-scale virtual screen. The re-
sults show that known CYP51 inhibitors were efficiently discri-
minated by the model, and it performed better than our previ-
ous CACYP51 model. The active site of CACYP51 was charac-
terized by multiple copy simultaneous search (MCSS) calcula-
tions. On the basis of the MCSS results, a series of novel azoles
were designed and synthesized, and they showed good in vi-
tro antifungal activity with a broad spectrum. The MIC80 value
of four of these compounds against C. albicans is
0.001 mgmLÀ1, indicating that they are promising leads for the
discovery of novel antifungal agents.
Introduction
Lanosterol 14a-demethylase (P45014DM, CYP51) is a member of
the cytochrome P450 superfamily, which catalyzes the oxida-
tive removal of the 14a-methyl group (C32) of lanosterol to
give D14,15-desaturated intermediates in ergosterol biosynthe-
sis.[1] As an essential enzyme in the fungal life cycle, CYP51 has
been a primary target for the azole class of antifungal agents.
Eukaryotic CYP51 enzymes are membrane-associated proteins,
and thus collecting crystal structure data remains a challenge.
In the absence of structural information, homology modeling
has proven to be an effective alternative for constructing rea-
sonable three-dimensional (3D) protein models.
Evers, Gohlke, and Klebe developed a new approach toward
homology modeling termed MOBILE,[6] which incorporates ex-
perimental information available from site-directed mutagene-
sis or structure–activity relationship (SAR) studies in the pro-
cess of homology modeling and model refinement. Traditional
homology methods only consider structural information avail-
able from template proteins, while the MOBILE approach uses
the bound ligand molecules as spatial restraints and optimizes
the initial homology models iteratively by incorporating the
binding information of bioactive ligands. MOBILE has been
successfully used to construct reasonable 3D models of several
GPCR proteins,[7–9] and this method has been validated by
structure-based ligand design and virtual screening. Moreover,
several new methods for ligand-supported homology model-
ing have extended the methodology of the MOBILE ap-
proach.[10–11]
The incidence of systemic fungal infections has been in-
creasing dramatically in recent years, and CYP51 represents an
attractive target for antifungal therapy. In an attempt to design
highly potent CYP51 inhibitors, we constructed a 3D model of
CYP51 from Candida albicans (CACYP51) through homology
modeling on the basis of the crystallographic coordinates of
four prokaryotic P450 proteins (P450BM3, P450cam, P450terp,
and P450eryF) and CYP51 from Mycobacterium tuberculosis
(MTCYP51).[2,3] Detailed docking analysis led to a better under-
standing of the important interactions between CACYP51 and
azole antifungal agents.[3,4] From the results of molecular mod-
eling, successful structure-based optimization of azole antifun-
gal agents has been reported by our research group.[4] More-
over, novel non-azole lead compounds were designed and syn-
thesized by a structure-based de novo design approach.[5]
However, the sequence identity between CACYP51 and the
template protein is relatively low (29.5% for the whole protein
and 30.5% for binding site residues). The efficiency of struc-
ture-based design of antifungal inhibitors is limited by the use
of molecular models from traditional homology modeling
methods. Therefore, building a more accurate model of
CACYP51 is of great importance.
In the present study, we built an improved 3D model of
CACYP51 by ligand-supported homology modeling and molec-
ular dynamics (MD) simulation. The active site was character-
ized by multiple copy simultaneous search (MCSS)[12] calcula-
[a] Dr. C. Sheng, Dr. W. Wang, Dr. X. Che, G. Dong, S. Wang, Dr. Z. Miao,
Dr. J. Yao, Prof. W. Zhang
School of Pharmacy, Military Key Laboratory of Medicinal Chemistry
Second Military Medical University
325 Guohe Road, Shanghai 200433 (People’s Republic of China)
Fax: (+86)21-81871243
E-mail: zhangwnk@hotmail.com
[b] Dr. H. Ji
Department of Chemistry
Department of Biochemistry, Molecular Biology, and Cell Biology
and Center for Drug Discovery and Chemical Biology
Northwestern University, Evanston, IL 60208-3113 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.200900468.
390
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 390 – 397

Improved Model of CYP51
tions. The reliability of the model was validated by a virtual
screening test and structure-based optimization of azole anti-
fungal agents. To the best of our knowledge, this is the first
application of ligand-supported homology modeling to cyto-
chrome P450 proteins.
Generation of a ligand-supported homology model of
CACYP51
For generation and validation of a ligand-supported homology
model, details about the interaction between protein and
ligand should be available.[6] In our previous studies, we gener-
ated a topographical interaction model[4] between azole inhibi-
tors and CACYP51 (Figure 1) to guide the subsequent homolo-
gy modeling process. The robustness of the interaction model
has been validated by SAR studies[4,13] and site-directed muta-
genesis.[14] In comparison with our previous homology model-
Results and Discussion
Chemistry
The syntheses of compounds 8a–k and 9a–c are outlined in
Schemes 1 and 2, respectively. The oxirane intermediate 4 was
synthesized according to our reported procedure.[4] The N-
ing studies on CACYP51,
a more accurate structure of
MTCYP51 (PDB ID: 2VKU)[15] was used as the template. A set of
100 initial crude models of CACYP51 were generated
by using the MODELLER module within the Insight II
2000 software package.[16] An azole inhibitor was
then docked into each individual CACYP51 model by
using the flexible docking method, Affinity.[17] The ap-
propriate docking poses were determined by the in-
teraction energies of Affinity and those calculated by
the Cscore module of Sybyl 6.9.[18] To decrease search
space, the active site was kept rigid during the dock-
ing study. On the basis of the ensemble CACYP51–
azole complex, the best model was identified by the
following selection criteria: 1) the model with the
lowest interaction energy, 2) avoidance of any intra-
molecular clashes,[6] and 3) the generated binding
mode of the azole compound with CACYP51 should
agree with the derived ligand binding mode shown
in Figure 1. After energy minimization, the resulting
Scheme 1. Reagents and conditions: a) ClCH2COCl, AlCl3, CH2Cl2, 408C, 3 h, 50%; b) tria-
zole, K2CO3, CH2Cl2, RT, 24 h, 70.0%; c) (CH3)3SOI, NaOH, toluene, 608C, 3 h, 62.3%;
d) phenol, K2CO3, DMF, 708C, 2 h, 91.8–92.3%; e) methylamine alcohol solution, RT, 12 h,
97–98%; f) 4, Et3N, EtOH, reflux, 9 h, 44.4–58.5%.
protein–ligand complex was further refined by MD
simulation using the Charmm module[19] within the
Insight II 2000 software package. The objective of the
MD simulations was to relieve the structural strain
stemming from the replacement of non-conserved
methyl-substituted phenoxypropan-1-amine side chains 7a–k
were synthesized via two steps. Various substituted phenols
were treated with excess 1,3-dibromopropane to give 1-(3-bro-
mopropoxy)-substituted benzenes 6a–k. These were subse-
quently reacted with methylamine in ethanol at room temper-
ature to afford side chains 7a–k. The target compounds 8a–k
were obtained in moderate to high yields by treating epoxide
4 with side chains 7a–k in the presence of triethylamine and
ethanol at 808C. Compounds 9a–c were obtained by the re-
duction of the nitro groups of compounds 8j–k in the pres-
ence of Raney nickel and hydrazine hydrate (Scheme 2). All the
target compounds were obtained as racemates.
residues in the homology modeling process and to obtain a
reasonable conformation. From the deviation of the total
energy and heavy atom RMSD during 1 ns MD simulations (see
the Supporting Information), the total energy of the model de-
creased substantially in the first 100 ps of MD simulation and
was stabilized after 300 ps equilibration.
Model validation through small-scale virtual screening
Structure-based virtual screening was performed with the re-
fined CACYP51 model (Figure 2) to determine whether it could
differentiate the known CACYP51 ligands from the “drug-like”
decoy ligands. Following a well-accepted approach,[20] we con-
structed a test set containing 950 drug-like molecules
extracted from the MDDR (MDL Drug Data Report)
dataset and 32 known CACYP51 inhibitors with differ-
ent scaffolds (the dataset is given in the Supporting
Information). All the compounds in the test set were
docked into the CACYP51 model with FlexX[21] and
scored with four different scoring functions (D_
Score,[22] ChemScore,[23] PMF,[24] and G_Score[25]) as im-
plemented in the Cscore module of Sybyl 6.9. The en-
Scheme 2. Reagents and conditions: a) Raney Ni, NH2NH2, EtOH, RT, 5 h, 95.5–97.0%.
ChemMedChem 2010, 5, 390 – 397
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
391

W. Zhang et al.
MED
also plotted with PMF as the scoring function (Fig-
ure 3B). The structure constructed in this study per-
formed better than the previous model, which sug-
gests that ligand-supported homology modeling is
more efficient when the sequence identity is relative-
ly low between the template protein and the target
protein. The above results also indicate that the pres-
ent structure is suitable for virtual screening.
MCSS functional maps for the active site of
CACYP51
To explore the key regions in the active site that are
important for ligand binding, the MCSS program[12]
was employed to calculate energetically favorable
positions and orientations of given functional groups
in the active site. The total numbers and ranges of
energies of the minima found for the various molecu-
lar probes are summarized in Table 1. All of the
Figure 1. Schematic representation of the postulated interactions between CACYP51 and
the azole compound (derived from ref. [4]); the arrows indicate proposed key interac-
tions between receptor and ligand.
energy minima of the ring functional groups (such as hydro-
phobic benzene and cyclohexane) were distributed over the
S2 (above the heme ring), S3, and S4 (near the FG loop) sub-
sites. Moreover, alkyl probes (such as propyl and n-butyl) re-
vealed the same distribution profile. Polar groups (such as
methanol, ethanol, and water) were mainly scattered over the
S1 and S4 subsites (surrounding His377 and Ser378). The
energy minima of the polar group can form hydrogen bonds
His310 (S1), Thr311 (S1), and Ser378 (S4), indicating that they
are important hydrogen bonding (HB) sites. In combination
with our previous modeling studies,[2–5] the active site of
CACYP51 can be divided into four functional regions (Figure 4):
1) The S1 subsite represents the hydrophilic hydrogen bond
binding site. His310 and Thr311 are two important HB sites in
this area. 2) The S2 subsite is above the heme ring represent-
ing the core hydrophobic area. Hydrophobic scaffolds of the
inhibitors (such as the sterol group of the substrate) or groups
that coordinate with the heme (such as the triazole or imida-
zole group of the azole inhibitors) are favorable in this pocket.
3) The S3 subsite represents the narrow and hydrophobic cleft
(facing the BC loop). The alkyl side chain (such as that of the
substrate and chromene inhibitors) or substituted phenyl
Figure 2. Ribbon representation of the refined homology model of
CACYP51.
richment factor (EF) is plotted
representing the fraction of the
database that is screened versus
the percentage of the active
compounds found (Figure 3A).
The best result is obtained with
PMF as the scoring function;
analysis of the top 30 and 45%
of the scored dataset results in
81 and 100% of the CYP51 in-
hibitors being identified. To com-
pare the effectiveness of virtual
screening between the present
model and our previous 3D
Figure 3. Enrichment of CYCYP51 obtained from: A) four different scoring functions and B) the new and previous
structure of CACYP51,[3] EF was homology models of CACYP51.
392
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 390 – 397

Improved Model of CYP51
Structure-based azole
optimization
Table 1. Summary of results for the functional groups used in the MCSS calculations.
Functional
Groups[a]
DH/2
Initial Number
of Copies
Minima with
E Range
Minima with
Interaction <DH/2
[kcalmolÀ1
]
E
interaction <0
[kcalmolÀ1
]
To verify the reliability of the
binding site characterization of
BENZ
CHEX
PRPN
ILER
PHEN
MEOH
THRR
WATR
À3.80
À3.97
À2.69
À3.10
À6.80
À5.32
À6.26
À4.99
1000
1000
1000
1000
1000
1000
1000
1000
48
À2.72~À22.51
À0.01~À9.09
À0.28~À8.00
À5.33~À8.81
À8.91~À30.19
À0.41~À37.18
À5.37~À26.13
À0.10~À42.16
43
73
190
176
343
235
162
416
195
CACYP51,
a
series of novel
171
343
235
142
89
azoles were designed and syn-
thesized on the basis of the four
functional pockets in the active
site. The triazole ring (located in
the S2 subsite), difluorophenyl
group (located in the S3 subsite),
and C2 hydroxy group (located
in the S1 subsite) are common
scaffold components of triazole
antifungal agents. N-Methyl-N-
153
[a] MCSS functional groups: BENZ=benzene, CHEX=cyclohexane, PRPN=propane, ILER=butane, PHEN=
phenol, MEOH=methanol, THRR=ethanol, WATR=water.
phenoxyalkylamines were designed as the side chains attached
to C3. The steric and hydrophobic nature of this new type of
side chain fits well with the S4 subsite. Moreover, the oxygen
atom of the phenoxy group can function as a HB acceptor and
form a HB interaction with Ser378.
The in vitro antifungal activities of the synthesized com-
pounds are listed in Table 2. The antifungal activity of each
compound is expressed as the minimal inhibitory concentra-
tion (MIC) that effects 80% inhibition of the tested fungi, with
fluconazole, itraconazole, and voriconazole used as reference
drugs. In general, all the target compounds show excellent ac-
tivity against most of the tested fungal pathogens except A. fu-
migatus. They show the greatest activity against C. albicans
and C. tropicalis. Most of the compounds are more potent than
fluconazole, itraconazole, and voriconazole, with MIC80 values
in the range of 0.016–0.001 mgmLÀ1. On the C. neoformans
strain, the inhibitory activities of the target compounds are
similar or superior to that of fluconazole. In particular, the in-
hibitory activity of compound 8c (MIC80 =0.004 mgmLÀ1) is
fourfold higher than that of voriconazole. Fluconazole is not ef-
fective against A. fumigatus, whereas some of our compounds
show moderate activity. For example, the MIC80 value of com-
pound 8i against A. fumigatus is 16 mgmLÀ1; however, it is less
active than itraconazole and voriconazole. For the dermato-
phytes (i.e., T. rubrum and M. gypseum), most of the com-
pounds show moderate to good inhibitory activity. They reveal
higher inhibitory activity against M. gypseum than against
T. rubrum. Several compounds (such as compounds 8a and
8b) are more active than the reference antifungal drugs.
Among the synthesized triazoles, compounds 8a, 8c, 8 f, and
8i exhibit strong in vitro antifungal activity with a broad anti-
fungal spectrum, and are worthy of further evaluation.
Figure 4. Surface rendering of the active site of CACYP51: the S1 site is at
lower left; the S2 site is the region in the lower middle; the region at lower
right is the S3 site; and at top is the S4 site.
groups (such as the difluorophenyl phenyl group of flucona-
zole) can form hydrophobic and van der Waals interactions
with this region. 4) The S4 subsite represents a hydrophobic
hydrogen bond binding site (facing the FG loop), which is im-
portant for the optimization of the C3 side chain of azole anti-
fungal agents. A hydrophobic side chain is favorable in this
region, and the addition of a group forming a HB interaction
with Ser378 can improve the affinity for CACYP51. In compari-
son with the MCSS calculation results of our previous model,[5]
the four functional subsites are identical. The difference be-
tween them is the orientation of several residues in the active
site (such as Tyr118 and Ser378), which resulted in the im-
proved performance of the present model in virtual screening.
The results from MCSS calculation can provide important infor-
mation for the design of novel antifungal agents.
To clarify the binding mode and SAR of the synthesized
azoles, compound 8i was docked into the active site of
CACYP51. The result revealed that compound 8i binds
CACYP51 with an extended conformation (Figure 5). The C2
hydroxy group, triazole ring, and difluorophenyl group are lo-
cated in the S1, S2, and S3 subsites, respectively. They form
similar interactions with CACYP51 as observed in the docking
model of fluconazole.[4] The substituted alkoxybenzene side
chain is oriented into the S4 pocket. The N-methyl group of
ChemMedChem 2010, 5, 390 – 397
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
393

W. Zhang et al.
MED
8c, and 8 f showed excellent an-
tifungal activity. The nitro group
of compound 8i forms hydrogen
bonds with the hydroxy group
of Tyr64. The hydrogen bonds
would be broken if the 4-nitro
group of compound 8i were
moved to positions 3 or 2 (com-
pounds 8j and 8k, respectively),
or reduced to an amino group
(compound 9a), which resulted
in a 4–16-fold decrease in anti-
fungal activity.
Table 2. In vitro antifungal activities of the triazole compounds.[a]
MIC80 [mgmLÀ1
]
Compd
C. alb.
C. tro.
C. neo.
T. rub.
M. gyp.
A. fum.
8a
8b
8c
8d
8e
8 f
8g
8h
8i
0.001
0.004
0.001
0.004
0.016
0.001
0.004
0.016
0.001
0.004
0.004
0.016
0.0625
0.0625
0.5
0.004
0.16
1
0.0625
0.0625
0.004
0.0625
0.0625
0.25
0.25
0.0625
0.25
0.25
0.25
1
0.0625
0.25
0.25
0.25
0.25
64
>64
64
>64
64
>64
>64
64
0.625
0.0625
0.0625
0.25
0.0625
0.25
0.0625
0.0625
0.25
0.0625
1
0.25
0.25
0.0625
0.0625
0.0625
0.016
0.0625
0.004
0.016
0.016
1
0.0625
0.016
0.0625
16
0.0625
0.25
0.016
0.016
1
0.25
1
1
0.25
16
8j
1
4
4
4
1
4
>64
>64
>64
>64
>64
>64
2
8k
9a
9b
9c
FLZ
ITZ
VOR
1
1
1
Conclusions
0.0625
0.016
0.125
0.016
0.125
0.0625
In summary, ligand-supported
homology modeling is a useful
method to build accurate 3D
structures of proteins when the
sequence identity between the
template structure and the
target protein is low. An im-
0.25
[a] Abbreviations: C. alb.=Candida albicans, C. tro.=Candida tropicalis, C. neo.=Cryptococcus neoformans,
T. rub.=Trichophyton rubrum, M. gyp.=Microsporum gypseum, A. fum.=Aspergillus fumigatus; FLZ=fluconazole,
ITZ=itraconazole, VOR=voriconazole.
proved model of CACYP51 was built by ligand-supported ho-
mology modeling. The current model has several advantages
over the previous homology model of CACYP51.[2,3] First, the
ligand information was taken into account in the construction
of the molecular model, and MD simulations were used in
model refinement; this ensures a reasonable 3D structure with
low energy. Second, the CACYP51 model performs well in the
virtual screening test, which is beneficial for finding lead struc-
tures. Lastly, the active site of CACYP51 model was character-
ized by MCSS calculations. The results reveal its efficiency in
structure-based optimization of novel azole antifungal agents.
The new azoles designed show excellent activity with a broad
antifungal spectrum. In particular, the MIC80 values of the most
active compounds 8a, 8c, 8d, and 8i against C. albicans reach
0.001 mgmLÀ1. These compounds are more potent than fluco-
nazole, itraconazole, and voriconazole, and are promising leads
for the development of novel antifungal agents. Based on the
above results, it is anticipated that the improved homology
model of CACYP51 could be used as a suitable template for
the structure-based design of novel inhibitors.
Figure 5. The docking conformation of compound 8i in the active site of
CACYP51: Important residues that interact with 8i are indicated, and hydro-
gen bonds are represented by dashed lines. The image was generated with
the Insight II 2000 software package.
Experimental Section
Chemistry
General methods: NMR spectra were recorded on a Bruker 500
spectrometer with (CH3)4Si as internal standard and CDCl3 as sol-
vent. Chemical shift (d) values and coupling constants (J) are given
in ppm and Hz, respectively. ESI mass spectra were performed on
an API-3000 LC–MS instrument. High-resolution mass spectrometry
data were collected on a Kratos Concept mass spectrometer. Ele-
mental analyses were performed with a MOD-1106 instrument and
are consistent with theoretical values within Æ0.4%. TLC analysis
was carried out on silica gel plates GF254 (Qindao Haiyang Chemi-
cal, China). Column chromatography was performed with silica gel
the side chain forms hydrophobic and van der Waals interac-
tions with Phe228. The propyl group forms hydrophobic and
van der Waals interactions with Leu376, Ile379, Leu461, and
Val519. The oxygen atom is hydrogen bonded with Ser378.
The terminal phenyl group is located in a hydrophobic pocket
lined by Tyr64, Phe380, Leu403, and Met508. Small hydropho-
bic substituents on the phenyl group can improve the affinity
of the compounds for CACYP51. For example, compounds 8a,
394
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 390 – 397

Improved Model of CYP51
60 G (Qindao Haiyang Chemical, China). Commercial solvents were
used without pretreatment.
16.72. The syntheses of the subsequent compounds 9b and 9c
were carried out similarly to that of compound 9a.
2-(2,4-Difluorophenyl)-3-{methyl-[3-(o-tolyloxy)propyl]amino}-1-
(1H-1,2,4-triazol-1-yl)propan-2-ol (8b): Pale-yellow oil. Rf =0.30
(CH2Cl2/MeOH 50:1); 1H NMR (500 MHz, CDCl3): d=8.08 (s, 1H),
7.77 (s, 1H), 6.75–7.51 (m, 7H), 5.30 (br, 1H), 4.49 (d, J=14.2 Hz,
2H), 3.87 (t, J=6.0 Hz, 2H), 3.05 (d, J=13.6 Hz, 1H), 2.76 (d, J=
13.6 Hz, 1H), 2.55 (t, J=6.8 Hz, 2H), 2.16 (s, 3H), 2.09 (s, 3H), 1.77–
1.80 ppm (m, 2H); HRMS-FAB m/z [M+H]+: calcd for C22H26F2N4O2:
417.2024, found: 417.12018; Anal. calcd for C22H26F2N4O2: C 63.45,
H 6.29, N 13.45, found: C, 63.70; H, 6.31; N, 13.40.
1-(3-Propanoxy)-4-methylbenzene (6a): A solution of p-cresol
(10.81 g, 0.10 mol) in DMF (50 mL) was added dropwise to a stirred
suspension of 1,3-dibromopropane (40.38 g, 0.20 mol) and K2CO3
(20.73 g, 0.15 mol) in DMF (100 mL) at room temperature. The reac-
tion mixture was then stirred at room temperature for 2 h, and
heated at 708C for 2 h. The mixture was filtered, and the resulting
solution was diluted with EtOAc (200 mL) and washed with H2O
(3ꢁ100 mL). The organic layer was separated, dried with anhy-
drous Na2SO4, and concentrated under reduced pressure. The resi-
due was purified by column chromatography (hexane) to give 6a
as a colorless oil (21.04 g, 92.3%). 1H NMR (500 MHz, CDCl3): d=
6.80–7.09 (m, 4H), 4.07 (t, J=5.8 Hz, 2H), 3.60 (t, J=6.5 Hz, 2H),
2.30–2.34 (m, 2H), 2.28 ppm (s, 3H); MS (ESI) m/z: 229 [M+1]. The
syntheses of the subsequent compounds 6b–k were carried out
similarly to that of compound 6a.
1-{[3-(4-Chlorophenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8c): Pale-yellow oil.
1
Rf =0.11 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3): d=8.13 (s,
1H), 7.79 (s, 1H), 6.76–7.52 (m, 7H), 5.31 (br, 1H), 4.50 (d, J=
14.2 Hz, 2H), 3.83 (t, J=5.8 Hz, 2H), 3.05 (d, J=13.4 Hz, 1H), 2.74
(d, J=13.6 Hz, 1H), 2.49 (br, 2H), 2.06 (s, 3H), 1.75 ppm (br, 2H);
HRMS-FAB m/z [M+H]+: calcd for C21H23ClF2N4O2: 437.1478, found:
437.1470; Anal. calcd for C21H23ClF2N4O2: C 57.73, H 5.31, N 12.82,
found: C 57.84, H 5.29, N 12.84.
N-Methyl-3-(p-tolyloxy)propan-1-amine (7a): A solution of com-
pound 6a (3.42 g, 0.015 mol) in EtOH (20 mL) was added dropwise
to a solution of methylamine alcohol (40 mL). The mixture was
stirred at room temperature for 12 h, at which time the reaction
was nearly complete. The solvent was evaporated under reduced
pressure to give 7a as a white solid (2.60 g, 97%). The product
was used in the next step without further purification. 1H NMR
(500 MHz, CDCl3): d=6.77–7.09 (m, 4H), 3.97 (t, J=6.2 Hz, 2H),
2.94 (t, J=7.3 Hz, 2H), 2.53 (s, 3H), 2.27 (s, 3H), 1.93–1.95 ppm (m,
2H); MS (ESI) m/z: 180 [M+1]. The syntheses of the subsequent
compounds 7b–k were carried out similarly to that of compound
7a.
1-{[3-(2-Chlorophenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8d): Pale-yellow oil.
1
Rf =0.16 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3): d=8.08 (s,
1H), 7.76 (s, 1H), 6.75–7.36 (m, 7H), 4.50 (d, J=14.2 Hz, 2H), 3.93
(t, J=5.8 Hz, 2H), 3.05 (d, J=13.6 Hz, 1H), 2.78 (d, J=13.6 Hz, 1H),
2.57–2.60 (m, 2H), 2.10 (s, 3H), 1.80–1.83 ppm (m, 2H); HRMS-FAB
m/z [M+H]+: calcd for C21H23ClF2N4O2: 437.1478, found: 437.1470;
Anal. calcd for C21H23ClF2N4O2: C 57.73, H 5.31, N 12.82, found: C
57.55, H 5.30, N 12.86.
2-(2,4-Difluorophenyl)-3-{methyl-[3-(p-tolyloxy)propyl]amino}-1-
(1H-1,2,4-triazol-1-yl)propan-2-ol (8a): A solution of epoxide 4
(1.67 g, 0.005 mol), 7a (1.08 g, 0.006 mol), Et3N (3 mL), and EtOH
(30 mL) was heated at reflux for 9 h. The solvent was evaporated
under reduced pressure. The residue was purified by silica gel
column chromatography (CH2Cl2/MeOH 100:2, v/v) to give 8a as a
pale-yellow oil (1.20 g, 57.7%). Rf =0.27 (CH2Cl2/MeOH 50:1);
1H NMR (500 MHz, CDCl3): d=8.10 (s, 1H), 7.76 (s, 1H), 6.74–7.52
(m, 7H), 5.30 (br, 1H), 4.52 (d, J=14.3 Hz, 1H), 4.47 (d, J=14.2 Hz,
1H), 3.85 (t, J=6.0 Hz, 2H), 3.05 (d, J=13.4 Hz, 1H), 2.76 (d, J=
13.3 Hz, 1H), 2.51 (br, 2H), 2.29 (s, 3H), 2.08 (s, 3H), 1.74–1.76 ppm
(m, 2H); 13C NMR (500 MHz, CDCl3): d=162.56, 158.84, 156.47,
150.81, 144.57, 129.87, 129.48, 126.17, 114.20, 111.31, 104.02, 72.14,
65.29, 62.33, 56.27, 55.73, 43.55, 26.96, 20.32 ppm; HRMS-FAB m/z
[M+H]+: calcd for C22H26F2N4O2: 417.2024, found: 417.12020; Anal.
calcd for C22H26F2N4O2: C 63.45, H 6.29, N 13.45, found: C 63.26, H
6.31, N 13.49. The syntheses of the subsequent compounds 8b–k
were carried out similarly to that of compound 8a.
1-{[3-(4-tert-Butylphenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8e): Pale-yellow oil.
Rf =0.16 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3): d=8.11 (s,
1H), 7.76 (s, 1H), 6.76–7.53 (m, 7H), 4.50 (d, J=14.2 Hz, 2H), 3.95
(t, J=5.8 Hz, 2H), 3.06 (d, J=13.4 Hz, 1H), 2.80 (d, J=13.6 Hz, 1H),
2.60 (br, 2H), 2.09 (s, 3H), 1.58 (br, 2H), 1.30 ppm (s, 9H); HRMS-
FAB m/z [M+H]+: calcd for C25H32F2N4O2: 459.2493, found:
459.2489; Anal. calcd for C25H32F2N4O2: C 65.48, H 7.03, N, 12.22,
found: C 65.68, H 7.02, N 12.20.
1
2-(2,4-Difluorophenyl)-3-{methyl-[3-(m-tolyloxy)propyl]amino}-1-
(1H-1,2,4-triazol-1-yl)propan-2-ol (8 f): Pale-yellow oil. Rf =0.17
(CH2Cl2/MeOH 50:1); 1H NMR (500 MHz, CDCl3): d=8.16 (s, 1H),
7.77 (s, 1H), 6.66–7.57 (m, 7H), 5.30 (br, 1H), 4.53 (d, J=14.2 Hz,
2H), 3.85 (br, 2H), 3.07 (d, J=13.6 Hz, 1H), 2.76 (d, J=13.6 Hz, 1H),
2.51 (br, 2H), 2.08 (s, 3H), 2.05 (s, 3H), 1.75 ppm (br, 2H); HRMS-
FAB m/z [M+H]+: calcd for C22H26F2N4O2: 417.2024, found:
417.12028; Anal. calcd for C22H26F2N4O2: C 63.45, H 6.29, N 13.45,
found: C 63.25, H 6.28, N 13.50.
3-{[3-(4-Aminophenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (9a): Raney Ni (cat.)
was added to a solution of 8j (0.89 g, 2 mmol) in EtOH (25 mL) in
the presence of N2H4 (5 mL). The mixture was stirred at room tem-
perature for 5 h. The solid was separated, and the filtrate was
evaporated under reduced pressure. The residue was purified by
column chromatography (CH2Cl2/MeOH 100:2, v/v) to give 9a as a
yellow oil (0.79 g, 95.5%). Rf =0.08 (CH2Cl2/MeOH 50:1); 1H NMR
(500 MHz, CDCl3): d=8.10 (s, 1H), 7.76 (s, 1H), 6.63–7.52 (m, 7H),
5.30 (br, 1H), 4.50 (d, J=14.2 Hz, 2H), 3.81 (t, J=6.0 Hz, 2H), 3.06
(d, J=13.3 Hz, 1H), 2.79 (d, J=13.6 Hz, 1H), 2.54 (br, 2H), 2.10 (s,
3H), 1.73–1.76 ppm (m, 2H); HRMS-FAB m/z [M+H]+: calcd for
C21H25F2N5O2: 418.1976, found: 418.1980; Anal. calcd for
C21H25F2N5O2: C 60.42, H 6.04, N 16.78, found: C 60.54, H 6.05, N
2-(2,4-Difluorophenyl)-3-{[3-(4-fluorophenoxy)propyl]-
(methyl)amino}-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8g): Pale-
1
yellow oil. Rf =0.14 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3):
d=8.13 (s, 1H), 7.78 (s, 1H), 6.76–7.54 (m, 7H), 4.49 (d, J=14.2 Hz,
2H), 3.85 (t, J=5.9 Hz, 2H), 3.07 (d, J=13.6 Hz, 1H), 2.76 (d, J=
13.6 Hz, 1H), 2.52 (br, 2H), 2.08 (s, 3H), 1.76 ppm (br, 2H); HRMS-
FAB m/z [M+H]+: calcd for C21H23F3N4O2: 421.1773, found:
421.1762; Anal. calcd for C21H23F3N4O2: C 59.99, H 5.51, N 13.33,
found: C 59.87, H 5.53, N 13.36.
2-(2,4-Difluorophenyl)-3-{[3-(4-methoxyphenoxy)propyl]-
(methyl)amino}-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8h): Pale-
1
yellow oil. Rf =0.22 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3):
d=8.12 (s, 1H), 7.77 (s, 1H), 6.77–7.52 (m, 7H), 5.30 (br, 1H, OH),
ChemMedChem 2010, 5, 390 – 397
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
395

W. Zhang et al.
MED
4.50 (d, J=14.2 Hz, 2H), 3.85 (t, J=6.0 Hz, 2H), 3.77 (s, 3H), 3.05 (d,
J=13.2 Hz, 1H), 2.75 (d, J=13.3 Hz, 1H), 2.51 (br, 2H), 2.08 (s, 3H),
1.75 ppm (br, 2H); 13C NMR (500 MHz, CDCl3): d=162.58, 158.88,
157.27, 150.96, 144.65, 129.48, 129.31, 126.30, 125.65, 115.64,
111.41, 104.14, 72.09, 65.68, 62.19, 56.25, 55.70, 43.62, 26.91 ppm;
HRMS-FAB m/z [M+H]+: calcd for C22H26F2N4O3: 433.1973, found:
433.1979; Anal. calcd for C22H26F2N4O3: C 61.10, H 6.06, N 12.96,
found: C 60.92, H 6.08, N 12.93.
In vitro antifungal activity assays
In vitro antifungal activities are expressed as MIC80 values, and
were measured by the serial dilution method in 96-well microtiter
plates. Test fungal strains were either clinical isolates, or were ob-
tained from the American Type Culture Collection (ATCC). MIC80 de-
terminations were performed according to the National Committee
for Clinical Laboratory Standards (NCCLS) recommendations, with
RPMI 1640 (Sigma) buffered with 0.165m MOPS (Sigma) as the test
medium. The MIC80 value is defined as the lowest concentration of
test compound that results in a culture with turbidity ꢀ80% inhib-
ition relative to the growth of the control. Test compounds were
dissolved in DMSO serially diluted in growth medium. The yeast
strains were incubated at 358C, and the dermatophytes at 288C.
Growth MIC80 was determined at 24 h for Candida spp., at 72 h for
Cryptococcus neoformans, and at 7 days for filamentous fungi.
2-(2,4-Difluorophenyl)-3-{methyl-[3-(4-nitrophenoxy)propyl]ami-
no}-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8i): Pale-yellow oil. Rf =
0.11 (CH2Cl2/MeOH 50:1); 1H NMR (500 MHz, CDCl3): d=8.13 (s,
1H), 7.80 (s, 1H), 6.77–7.56 (m, 7H), 5.25 (br, 1H), 4.50 (d, J=
14.2 Hz, 2H), 3.94 (t, J=5.9 Hz, 2H), 3.07 (d, J=13.5 Hz, 1H), 2.76
(d, J=13.7 Hz, 1H), 2.50 (t, J=6.6 Hz, 2H), 2.09 (s, 3H), 1.78–
1.81 ppm (m, 2H); 13C NMR (500 MHz, CDCl3): 163.63, 162.68,
158.84, 150.97, 144.67, 141.49, 129.41, 126.23, 125.85, 114.28,
111.44, 104.20, 72.06, 66.28, 62.00, 56.13, 55.50, 43.66, 26.72 ppm;
HRMS-FAB m/z [M+H]+: calcd for C21H23F2N5O4: 448.1718, found:
448.1709; Anal. calcd for C21H23F2N5O4: C 56.37, H 5.18, N 15.65,
found: C 56.48, H 5.19, N 15.60.
Acknowledgements
This work was supported by the National Natural Science Foun-
dation of China (Grant No. 30930107), the Shanghai Rising-Star
Program (Grant No. 09A1407000), the National Major Special
Project for the Creation of New Drugs (Grant No. 2009ZX09301-
011), and the Shanghai Leading Academic Discipline Project
(Project No. B906).
2-(2,4-Difluorophenyl)-3-{methyl-[3-(3-nitrophenoxy)propyl]ami-
no}-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8j): Pale-yellow oil. Rf =
0.19 (CH2Cl2/MeOH 50:1); 1H NMR (500 MHz, CDCl3): d=8.12 (s,
1H), 7.79 (s, 1H), 6.81–7.82 (m, 7H), 5.29 (br, 1H), 4.51 (d, J=
14.2 Hz, 2H), 3.93 (t, J=6.0 Hz, 2H), 3.09 (d, J=13.5 Hz, 1H), 2.77
(d, J=13.6 Hz, 1H), 2.52 (t, J=6.6 Hz, 2H), 2.11 (s, 3H), 1.79–
1.82 ppm (m, 2H); 13C NMR (500 MHz, CDCl3): 162.45, 159.09,
158.90, 150.80, 148.99, 144.54, 129.82, 129.35, 126.12, 121.33,
115.61, 111.28, 108.50, 104.02, 72.07, 66.05, 62.06, 56.08, 55.44,
43.55, 26.63 ppm; HRMS-FAB m/z [M+H]+: calcd for C21H23F2N5O4:
448.1718, found: 448.1706; Anal. calcd for C21H23F2N5O4: C 56.37, H
5.18, N 15.65, found: C 56.26, H 5.19, N 15.68.
Keywords: active sites
·
antifungal agents
·
azoles
·
cytochromes · homology modeling
[1] Y. Aoyama, Y. Yoshida, R. Sato, J. Biol. Chem. 1984, 259, 1661–1666.
[2] H. Ji, W. Zhang, Y. Zhou, M. Zhang, J. Zhu, Y. Song, J. Lu, J. Med. Chem.
2000, 43, 2493–2505.
2-(2,4-Difluorophenyl)-3-{methyl-[3-(2-nitrophenoxy)propyl]ami-
no}-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (8k): Pale-yellow oil. Rf =
0.28 (CH2Cl2/MeOH 50:1); 1H NMR (500 MHz, CDCl3): d=8.09 (s,
1H), 7.76 (s, 1H), 6.76–7.85 (m, 7H), 5.30 (br, 1H), 4.50 (d, J=
14.2 Hz, 2H), 4.01 (br, 2H), 3.04 (d, J=13.2 Hz, 1H), 2.73 (d, J=
13.2 Hz, 1H), 2.52 (br, 2H), 2.10 (s, 3H), 1.78 ppm (br, 2H); HRMS-
FAB m/z [M+H]+: calcd for C21H23F2N5O4: 448.1718, found:
448.1713; Anal. calcd for C21H23F2N5O4: C 56.37, H 5.18; N 15.65,
found: C 56.20, H 5.17, N 15.70.
[3] C. Sheng, W. Zhang, M. Zhang, Y. Song, H. Ji, J. Zhu, J. Yao, J. Yu, S.
Yang, Y. Zhou, J. Lu, J. Biomol. Struct. Dyn. 2004, 22, 91–99.
[4] C. Sheng, W. Zhang, H. Ji, M. Zhang, Y. Song, H. Xu, J. Zhu, Z. Miao, Q.
Jiang, J. Yao, Y. Zhou, J. Lu, J. Med. Chem. 2006, 49, 2512–2525.
[5] H. Ji, W. Zhang, M. Zhang, M. Kudo, Y. Aoyama, Y. Yoshida, C. Sheng, Y.
Song, S. Yang, Y. Zhou, J. Lu, J. Zhu, J. Med. Chem. 2003, 46, 474–485.
[6] A. Evers, H. Gohlke, G. Klebe, J. Mol. Biol. 2003, 334, 327–345.
[7] A. Evers, T. Klabunde, J. Med. Chem. 2005, 48, 1088–1097.
[8] A. Evers, G. Klebe, Angew. Chem. 2003, 115, 250–253; Angew. Chem. Int.
Ed. 2004, 43, 248–251.
1-{[3-(3-Aminophenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (9b): Pale-yellow oil.
Rf =0.11 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3): d=8.10 (s,
[9] A. Evers, G. Klebe, J. Med. Chem. 2004, 47, 5381–5392.
[10] C. N. Cavasotto, A. J. Orry, N. J. Murgolo, M. F. Czarniecki, S. A. Kocsi,
B. E. Hawes, K. A. O’Neill, H. Hine, M. S. Burton, J. H. Voigt, R. A. Abagy-
an, M. L. Bayne, F. J. Monsma, Jr., J. Med. Chem. 2008, 51, 581–588.
[11] A. Patny, P. V. Desai, M. A. Avery, Proteins 2006, 65, 824–842.
[12] A. Miranker, M. Karplus, Proteins 1991, 11, 29–34.
[13] F. Giraud, C. Loge, F. Pagniez, D. Crepin, P. Le Pape, M. Le Borgne,
Bioorg. Med. Chem. Lett. 2008, 18, 1820–1824.
[14] S. H. Chen, C. Q. Sheng, X. H. Xu, Y. Y. Jiang, W. N. Zhang, C. He, Biol.
Pharm. Bull. 2007, 30, 1246–1253.
[15] A. N. Eddine, J. P. von Kries, M. V. Podust, T. Warrier, S. H. Kaufmann,
L. M. Podust, J. Biol. Chem. 2008, 283, 15152–15159.
[16] Insight II: Accelrys Inc., 2000: 10188 Telesis Court, Suite 100, San Diego,
CA 92121 (USA); http://www.accelrys.com/ (accessed December 22,
2009).
[17] B. A. Luty, Z. R. Wasserman, P. F. W. Stouten, C. N. Hodge, M. Zacharias,
J. A. McCammon, J. Comput. Chem. 1995, 16, 454–464.
[18] Tripos Associates Inc., 2001: St.Louis, MO (USA); http://www.tripos.com
(accessed December 22, 2009).
1
1H), 7.77 (s, 1H), 6.17–7.26 (m, 7H), 5.30 (br, 1H), 4.50 (d, J=
14.2 Hz, 2H), 3.84 (t, J=5.9 Hz, 2H), 3.05 (d, J=13.2 Hz, 1H), 2.78
(d, J=13.2 Hz, 1H), 2.52 (br, 2H), 2.101 (s, 3H), 1.75 ppm (br, 2H);
HRMS-FAB m/z [M+H]+: calcd for C21H25F2N5O2: 418.1976, found:
418.1967; Anal. calcd for C21H25F2N5O2: C 60.42, H 6.04, N 16.78,
found: C 60.24, H 6.06, N 16.75.
1-{[3-(2-Aminophenoxy)propyl](methyl)amino}-2-(2,4-difluoro-
phenyl)-1-(1H-1,2,4-triazol-1-yl)propan-2-ol (9c): Pale-yellow oil.
1
Rf =0.08 (CH2Cl2/MeOH 50:1); H NMR (500 MHz, CDCl3): d=8.10 (s,
1H), 7.77 (s, 1H), 6.70–7.54 (m, 7H), 5.30 (br, 1H), 4.50 (d, J=
14.2 Hz, 2H), 3.91 (t, J=6.0 Hz, 2H), 3.07 (d, J=13.0 Hz, 1H), 2.80
(d, J=13.0 Hz, 1H), 2.55 (br, 2H), 2.11 (s, 3H), 1.81 ppm (br, 2H);
HRMS-FAB m/z [M+H]+: calcd for C21H25F2N5O2: 418.1976, found:
418.1972; Anal. calcd for C21H25F2N5O2: C 60.42, H 6.04, N 16.78,
found: C 60.60, H 6.05, N 16.71.
[19] B. R. Brooks, R. E. Bruccoleri, B. D. Olafson, D. J. States, S. Swaminathan,
M. Karplus, J. Comput. Chem. 1983, 4, 187–217.
[20] C. Bissantz, P. Bernard, M. Hibert, D. Rognan, Proteins 2003, 50, 5–25.
396
www.chemmedchem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2010, 5, 390 – 397

Improved Model of CYP51
[21] M. Rarey, B. Kramer, T. Lengauer, G. Klebe, J. Mol. Biol. 1996, 261, 470–
489.
[25] G. Jones, P. Willett, R. C. Glen, A. R. Leach, R. Taylor, J. Mol. Biol. 1997,
267, 727–748.
[22] H. Gohlke, M. Hendlich, G. Klebe, J. Mol. Biol. 2000, 295, 337–356.
[23] M. D. Eldridge, C. W. Murray, T. R. Auton, G. V. Paolini, R. P. Mee, J.
Comput. Aided Mol. Des. 1997, 11, 425–445.
Received: November 14, 2009
Revised: December 15, 2009
[24] I. Muegge, Y. C. Martin, J. Med. Chem. 1999, 42, 791–804.
Published online on February 15, 2010
ChemMedChem 2010, 5, 390 – 397
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemmedchem.org
397