Benzoic acid derivatives with improved antifungal activity: Design, synthesis, structure-activity relationship (SAR) and CYP53 docking studies

Article abstract of DOI:10.1016/j.bmc.2015.06.042

Previously, we identified CYP53 as a fungal-specific target of natural phenolic antifungal compounds and discovered several inhibitors with antifungal properties. In this study, we performed similarity-based virtual screening and synthesis to obtain benzo

Full text of DOI:10.1016/j.bmc.2015.06.042

Accepted Manuscript
Benzoic acid derivatives with improved antifungal activity: design, synthesis,
structure–activity relationship (SAR) and CYP53 docking studies
Sabina Berne, Lidija Kovačič, Matej Sova, Nada Kraševec, Stanislav Gobec,
Igor Križaj, Radovan Komel
PII:
S0968-0896(15)00536-2
DOI:
http://dx.doi.org/10.1016/j.bmc.2015.06.042
BMC 12398
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
20 February 2015
15 June 2015
16 June 2015
Please cite this article as: Berne, S., Kovačič, L., Sova, M., Kraševec, N., Gobec, S., Križaj, I., Komel, R., Benzoic
acid derivatives with improved antifungal activity: design, synthesis, structure–activity relationship (SAR) and
CYP53 docking studies, Bioorganic & Medicinal Chemistry (2015), doi: http://dx.doi.org/10.1016/j.bmc.
2015.06.042
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
Benzoic acid derivatives with improved
antifungal activity: design, synthesis,
structure–activity relationship (SAR) and
CYP53 docking studies
Leave this area blank for abstract info.
Sabina Bernea, ∗, Lidija Kovačičb, c, Matej Sovad, Nada Kraševece, Stanislav Gobecd, Igor Križajb, f and Radovan
Komela, e,
∗
aInstitute of Biochemistry, Faculty of Medicine, University of Ljubljana, Vrazov trg 2, SI-1000 Ljubljana,
Slovenia
bDepartment of Molecular and Biomedical Sciences, Jožef Stefan Institute, Jamova 39, SI-1000 Ljubljana,
Slovenia
cUCD Conway Institute, School of Medicine and Medical Science, University College Dublin, Belfield, Dublin
4, Ireland
dChair of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, SI-1000
Ljubljana, Slovenia
eNational Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia
fDepartment of Chemistry and Biochemistry, Faculty of Chemistry and Chemical Technology, University of
Ljubljana, Večna pot 113, SI-1000 Ljubljana, Slovenia

Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com
Benzoic acid derivatives with improved antifungal activity: design, synthesis,
structure–activity relationship (SAR) and CYP53 docking studies
Sabina Bernea, ∗, Lidija Kovačičb, c, Matej Sovad, Nada Kraševece, Stanislav Gobecd, Igor Križajb, f and
Radovan Komela, e,
∗
aInstitute of Biochemistry, Faculty of Medicine, University of Ljubljana, Vrazov trg 2, SI-1000 Ljubljana, Slovenia
bDepartment of Molecular and Biomedical Sciences, Jožef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia
cUCD Conway Institute, School of Medicine and Medical Science, University College Dublin, Belfield, Dublin 4, Ireland
dChair of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Ljubljana, Aškerčeva cesta 7, SI-1000 Ljubljana, Slovenia
eNational Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia
fDepartment of Chemistry and Biochemistry, Faculty of Chemistry and Chemical Technology, University of Ljubljana, Večna pot 113, SI-1000 Ljubljana,
Slovenia
ARTICLE INFO
ABSTRACT
Article history:
Received
Received in revised form
Accepted
Previously, we identified CYP53 as a fungal-specific target of natural phenolic antifungal
compounds and discovered several inhibitors with antifungal properties. In this study, we
performed similarity-based virtual screening and synthesis to obtain benzoic acid-derived
compounds and assessed their antifungal activity against Cochliobolus lunatus, Aspergillus
niger and Pleurotus ostreatus. In addition, we generated structural models of CYP53 enzyme
and used them in docking trials with 40 selected compounds. Finally, we explored CYP53-
ligand interactions and identified structural elements conferring increased antifungal activity to
facilitate the development of potential new antifungal agents that specifically target CYP53
enzymes of animal and plant pathogenic fungi.
Available online
Keywords:
Benzoic acid;
CYP53;
Cytochrome P450;
SAR;
2009 Elsevier Ltd. All rights reserved.
Molecular modeling;
Molecular docking;
Antifungal activity.
1. Introduction
natural products and chemical libraries provide rapid selection
and evaluation of potential antifungal drug candidates.
Fungal diseases are not only becoming an increasing
economic burden in healthcare and jeopardizing food security
1
2
Numerous and highly diverse cytochromes P450 (CYPs) have
been found in fungal genomes.11 Apart from their roles in the
primary and secondary metabolism, certain fungal CYPs are
targets of antifungal drugs and agrochemicals (e.g., CYP51).12
We previously established that benzoate 4-monooxygenase
(CYP53A15) from the plant pathogenic Dothidiomycete
Cochliobolus lunatus (Curvularia lunata) is a target of natural
phenolic antifungal compounds and proposed CYP53 family
members as potential novel antifungal targets.13 The CYP53
family is fairly conserved and widely distributed among
ascomycete and basidiomycete fungi but so far unidentified in
ascomycete yeasts.14 Initially, fungal CYP53 family members
were characterized as detoxifying enzymes that convert benzoate
and certain mono-substituted benzoate derivatives to 4-
3
but also pose a threat to plant and animal biodiversity . Despite
the high demand for antifungal compounds with novel chemical
scaffolds and modes of action, the development of antifungal
agents is relatively slow.4 Due to the eukaryotic nature of fungal
cells, it is difficult to find appropriate fungal-specific targets.5
Furthermore, structure-based drug design is limited by the scarce
availability of three-dimensional (3D) structures of known fungal
drug targets.4 Nevertheless, significant technological advances
have been made to accelerate antifungal drug discovery. Genome
information from over 350 fungal species
6
enables the
identification of new molecular targets by comparative genomics
tools 7. Modern computer-aided drug design approaches
8
9,10
combined with new high-throughput screening assays
against
———
∗ Corresponding author. Tel.: +386-1-320-3296; fax: +386-1-423-1088; e-mail: sabina.berne@mf.uni-lj.si or sabina.berne@bf.uni-lj.si
Present address: Department of Agronomy, Biotechnical Faculty, Jamnikarjeva 101, SI-1000 Ljubljana, Slovenia.
∗ Corresponding author. Tel.: +386-1-543-7662; fax: +386-1-543-7641; e-mail: radovan.komel@mf.uni-lj.si

hydroxylated products.15–18 These are further metabolized via the
β-ketoadipate pathway for aromatic compound degradation to
produce intermediates that finally enter the tricarboxylic acid
cycle and are utilized as a source of carbon.19 However, some
CYP53 members also exhibit O-demethylation activity of 3-
methoxybenzoate derivatives.13,16 Recently,
a peculiar O-
demethylation of stilbene derivatives has been demonstrated for
CYP53D2 from Postia placenta.20 In addition to a proposed
biological role of CYP53 in the degradation of plant-based
aromatic compounds (e.g., released in soil or dead plant material)
21, it is possible that benzoate detoxification in plant pathogenic
fungi contributes to their pathogenicity. Indeed, during infection
of maize, up-regulated expression of benzoate 4-monooxygenase
gene (BPH) from the pathogenic Dothideomycete fungus
Cochliobolus heterostrophus (Bipolaris maydis) has been
reported.22
Figure 1. The shape and atom types of query compound 1a used for
similarity-based virtual screening. The compound shape is presented in grey,
atom types are colored: ring systems (green), anions (red), H-bond acceptors
(red net).
2.2. Compound characterization
Compound
1a
was
obtained
from
ChemBridge
Building on these findings, we have already identified new
antifungal compounds that specifically inhibit the catalytic
activity of CYP53A15 and restrain mycelial growth of
Cochliobolus lunatus, Aspergillus niger and Pleurotus
ostreatus.23,24 Since benzoic and cinnamic acid derivatives tested
against these fungi displayed only moderate antifungal activity
the present study focuses primarily on the structure-antifungal
activity analyses to determine key structural features of the
inhibitors that bestow increased antifungal effects. Exploration of
the nature of CYP53-ligand interactions was performed by
HADDOCK25 that requires prior knowledge of biochemical and
biophysical interaction data and to these end the results obtained
from our previous studies23,24 served as a good basis. By
homology modeling we first generated structural models of three
fungal CYP53 enzymes: i) CYP53A1 from Aspergillus niger, the
first fungal benzoate para-hydroxylase to be purified and
biochemically characterized, ii) CYP53A15 from Cochliobolus
lunatus, also well characterized protein and an object of our
research for the past decade and iii) CYP53C11 from Pleurotus
ostreatus selected due to higher identity to CYP53A15 (54%)
and CYP53A1 (50%) in comparison to the other two predicted
CYP53 sequences in its genome. We then carried out ligand-
based virtual screening with best hit compound23 1a as a template
and obtained 40 compounds that were screened for antifungal
activity against these fungal species. We finally correlated
antifungal activity with compounds’ structural features and
investigated CYP53 enzyme-ligand interactions by docking
experiments using HADDOCK. The structural insights into
individual/common characteristics of the CYP53 active sites and
enzyme-inhibitor interactions will contribute to the design and
development of potential new antifungal compounds.
(http://www.chembridge.com), compounds 14 and 17 from
Maybridge (http://www.maybridge.com), compounds 11, 12, 28
and 33 from Vitas-M Laboratories (http://www.vitasmlab.com),
compounds 13, 19, 21, 22, 23 and 29 from Key Organics
(http://www.keyorganics.net), compounds 3, 8, 16, 18 and 32
from Specs (http://www.specs.net), compounds 4, 7 and 10 from
Mir Biotech (http://www.mirbiotech.com), compounds 5, 6 and
20
from
Princeton
BioMolecular
Research
(http://www.princetonbio.com), compounds 15, 24, 25, 26 and 30
from Combi-Block (http://www.combi-blocks.com), compound
31 from UkrOrgSynthesis (http://www.ukrorgsynth.com),
compounds
34
and
36
from
Fluka
(http://www.sigmaaldrich.com) and compounds 36 and 35 from
Acros Organics (http://www.acros.com). The reagents and
solvents for chemical synthesis were purchased from Acros
Organics, Sigma-Aldrich and Maybridge and used without
further purification. Analytical TLC was performed on Merck
silica gel (60F254
) pre-coated plates (0.25 mm) and the
compounds were visualized under UV light. Column
chromatography was carried out on the Merck silica gel 60 (mesh
70-230). The yields of purified products were not optimized. The
compounds’ melting points were determined on a Reichert hot-
stage apparatus and are presented as uncorrected values. The
molecular formulas of the compounds were identified and
confirmed by mass spectral data and high-resolution mass
measurements on
spectrometer. IR spectra were recorded on a Perkin-Elmer FTIR
1600 spectrometer.
a
VG-Analytical Autospec
Q
mass
2.2.1. NMR spectra.
1H and 13C NMR spectra were recorded on a Bruker Avance
400 DPX spectrometer at 298 K. The values are reported in ppm
using tetramethylsilane or solvent as internal standard (DMSO-d6
at 2.50 ppm, CDCl3 at 7.26 ppm). The coupling constants (J) are
in Hz and the splitting patterns are designated: s, singlet; br s,
broad singlet; d, doublet; dd, double doublet; ddd, doublet of
doublet of doublet; t, triplet; dt, double triplet; and m, multiplet.
2. Materials and methods
2.1. Compound selection using ligand-based virtual screening
For the ligand-based virtual screening, we selected the “drug-
like” subset of the ZINC database 26, with 13 million compounds,
and downloaded it in SD format. We first prepared the database
with Omega 2.4 software (OpenEye Scientific Software, Santa
Fe, NM. http://www.eyesopen.com) to yield an average of 152
conformations per compound. We then screened this database
using ROCS 3 software (OpenEye Scientific Software, Santa Fe,
NM. http://www.eyesopen.com) to reduce it to compounds that
fit well to the query compound in terms of size and shape (Figure
1), 3-methyl-4-(1H-pyrrol-1-yl)benzoic acid (1a; the best
compound from the first round of virtual screening 23).
2.2.2. HPLC analysis.
The purities of all active compounds were determined by
reversed-phase high-performance liquid chromatography (HPLC)
analysis on an Agilent 1100 system (Agilent Technologies, Santa
Clara, CA, USA) equipped with a quaternary pump and a
multiple-wavelength detector, using an Agilent Eclipse Plus C18,
5 µm column (4.6 × 50 mm, 5 µm), with a flow rate of 1.0
mL/min, detection at 254 nm and an eluent system of: A = 0.1%
TFA in H2O; B = MeOH. The following gradient was applied:

i) method 1: 0–16 min, 10% B → 90% B in A; 16–19 min,
90% B in A; 19–20 min, 90% B → 10% B in A; post time 6 min.
The run time was 26 min, at a temperature of 25 °C;
separation, the organic phase was dried with anhydrous Na2SO4,
filtered and the solvent was removed by evaporation. The residue
was
purified
by
column
chromatography
using
ethylacetate/hexane (1/6 to 1/4) as an eluent to obtain compound
42 as white needle crystals. Yield 44 %; mp 27-28°C; H-NMR
ii) method 2: 0–12 min, 50% B → 90% B in A; 12–16 min,
90% B in A; 16-20 min, 90% B → 50% B in A; post time 6 min.
The run time was 26 min, at a temperature of 25 °C.
1
(400 MHz, CDCl3): δ(ppm) 3.93 (s, 3H, CH3), 5.16 (s, 2H, CH2),
6.24 (t, J= 2.0 Hz, 2H, Ar-H), 6.72 (t, J= 2.0 Hz, 2H, Ar-H), 7.17
(d, J= 8.0 Hz, 2H, Ar-H); 8.01 (d, J= 8.0 Hz, 2H, Ar-H); HR-MS
(ESI): m/z calcd. for C13H14NO2 [M+H]+ 216.1025, found:
216.1029; HPLC tR = 9.832 min (Method 2). The data are
consistent with those reported previously.28
The relative purity of all active compounds was above 95.0%
as determined by HPLC.
2.3. Synthesis of compounds 1b, 2, 9 and 27.
4-(pyrrol-1-ylmethyl)benzoic acid (2). Methyl 4-(pyrrol-1-
ylmethyl)benzoate (42) (196 mg, 0.91 mmol) was suspended in 1
M NaOH (10.0 mL, 10.0 mmol) and stirred at room temperature
for 24 hours; 2.0 ml of dioxane were then added and stirred at
room temperature for an additional 72 hours. The pH was
adjusted to 1 by the addition of 1 M HCl and the white solid was
filtered off to obtain compound 2 as white needles. Yield 94%;
mp 152-155°C; 1H-NMR (400 MHz, DMSO-d6): δ(ppm) 5.19 (s,
CH2), 6.04 (s, 2H, Ar-H), 6.82 (s, 2H, Ar-H), 7.23 (d, J= 7.6 Hz,
2H, Ar-H), 7.89 (d, J= 7.6Hz, 2H, Ar-H), 12.94 (br s, 1H,
COOH); HR-MS (ESI): m/z calcd. for C12H10NO2 [M-H]-
200.0712, found: 200.0710; HPLC tR = 7.616 min (Method 2).
Scheme 1. Synthesis of compound 1b.
Ethyl 4-(1H-pyrrol-1-yl)benzoate (40): Compound 40 was
synthesized according to the published procedure 27. Cesium
carbonate (978 mg, 12.0 mmol), copper (75 mg, 1.2 mmol) and
anhydrous acetonitrile (24 ml) were added to a dry flask
(equipped with a magnetic stirrer and a rubber septum) filled
with argon. Ethyl 4-iodobenzoate 38 (1.0 ml, 6.0 mmol) and
pyrrole 39 (0.7 mM, 9.0 mmol) were added to the obtained
suspension and stirred under reflux for 72 hours. Since TLC
analysis of the reaction mixture showed an incomplete reaction,
cesium carbonate (978 mg, 12.0 mmol) and copper (75 mg, 1.2
mmol) were added again and the reaction mixture was stirred
under reflux for an additional 48 hours and then cooled to room
temperature. The solid was filtered off and washed with
ethylacetate. The combined organic phases were removed by
evaporation and the obtained residue was purified by column
chromatography using diethylether/hexane (1/10) as an eluent to
obtain 0.581 g (45%) of compound 40 as white crystals. Mp 78-
Scheme 3. Synthesis of compound 9.
Methyl 4-(pyrrolidin-1-ylmethyl)benzoate (44). Pyrrolidine 43
(0.246 mL, 3.0 mmol) and NaH (108 mg, 4.5 mmol) were
dissolved in anhydrous DMF (10 mL) and cooled to 0ºC under an
argon
atmosphere.
After
30
minutes,
methyl
4-
1
79°C; H-NMR (400 MHz, CDCl3) δ(ppm): 1.32 (t, J = 7.2 Hz,
(bromomethyl)benzoate 41 (0.756 g; 3.3 mmol) was added, the
ice bath was removed and the mixture was stirred at room
temperature for 24 hours; 30 ml of saturated NaHCO3 were then
slowly added and the obtained solution was washed with
ethylacetate (50 mL). After layer separation, the organic phase
was dried with anhydrous Na2SO4, filtered and the solvent was
removed by evaporation. The oil residue was purified by column
chromatography using methanol/ethylacetate (7/1) as an eluent to
obtain compound 44 as a slightly yellow viscous liquid. Yield
52%; 1H-NMR (400 MHz, CDCl3): δ(ppm) 1.78-1.81 (m, 4H, 2x
CH2), 2.49-2.52 (s, 4H, 2xCH2), 3.66 (s, 2H, CH2-Ph), 3.91 (s,
3H, CH3), 7.41 (d, J= 7.6 Hz, 2H, Ar-H), 7.99 (d, J= 7.6 Hz, 2H,
Ar-H); HR-MS (ESI): m/z calcd. for C13H18NO2 [M+H]+
220.1338, found: 220.1333. The data are consistent with those
reported previously.29
3H, CH3), 4.30 (q, J = 7.2 Hz, 2H, CH2), 6.29 (t, J = 2.2 Hz, 2H,
pyrrole-H), 7.07 (t, J = 2.2 Hz, 2H, pyrrole-H), 7.35 (d, J = 8.8
Hz, 2H, Ar-H), 8.01 (d, J = 8.8 Hz, 2H, Ar-H); HR-MS (ESI):
m/z calcd. for C13H14NO2 [M+H]+ 216.1025, found: 216.1025.
4-(1H-pyrrol-1-yl)benzoic acid (1b): Compound 40 (490 mg,
1.75 mmol) was suspended in 10 ml of 1 M NaOH and stirred at
room temperature for 48 hours; 1 M HCl was then added
dropwise to adjust the pH to 1 and the white solid was filtered off
and dried to obtain 0.083 g (19%) of compound 1b as a light
brown solid. Mp 198-202°C; 1H-NMR (400 MHz, DMSO-d6)
δ(ppm): 6.32 (t, J = 2.4 Hz, 2H, pyrrole-H), 7.50 (t, J = 2.2 Hz,
2H, pyrrole-H), 7.70-7.74 (m, 2H, Ar-H), 7.98-8.01 (m, J = 8.8
Hz, 2H, Ar-H); 1H from COOH is exchanged; 13C-NMR (100
MHz, DMSO-d6: δ(ppm) 111.4, 118.6, 119.1, 127.1, 131.0,
143.0, 166.0; HR-MS (ESI): m/z calcd. for C11H10NO2 [M+H]+
188.0712, found: 188.0716; HPLC tR = 16.054 min (95.4%,
Method 1).
4-(pyrrolidin-1-ylmethyl)benzoic acid (9): Methyl 4-
(pyrrolidin-1-ylmethyl)benzoate (44) (193 mg, 0.88 mmol) was
suspended in 1 M NaOH (5.0 mL, 5.0 mmol) and stirred at room
temperature for 72 hours. The alkaline solution was washed with
ethylacetate (10 mL) and 1 M HCl was then added to adjust the
pH to 6. The solvent was removed by evaporation and the residue
was
purified
by
column
chromatography
using
Scheme 2. Synthesis of compound 2.
methanol/ethylacetate (2/1) as an eluent to obtain compound 9 as
a white powder. Yield 95%; mp 229-231°C; 1H-NMR (400 MHz,
D2O): δ(ppm) 1.72-1.76 (m, 4H, 2xCH2), 2.53-2.57 (s, 4H,
2xCH2), 3.68 (s, 2H, CH2), 7.39 (d, J= 7.6 Hz, 2H, Ar-H), 7.81
(d, J=7.6 Hz, 2H, Ar-H), H from COOH is exchanged; HR-MS
(ESI): m/z calcd. for C12H14NO2 [M-H]- 204.1025, found:
204.1019.
Methyl 4-(pyrrol-1-ylmethyl)benzoate (42). Pyrrole 39 (0.201
g, 3.0 mmol) and NaH (108 mg, 4.5 mmol) were dissolved in
anhydrous DMF (10 mL) and the mixture was cooled to 0ºC
under an argon atmosphere. After 30 minutes, methyl 4-
(bromomethyl)benzoate 41 (0.756 g; 3.3 mmol) was added, the
ice bath was removed and the mixture was stirred for 72 hours at
55°C. The reaction mixture was then cooled to room temperature
and 30 ml of saturated NaHCO3 were slowly added. The obtained
solution was washed with ethylacetate (50 mL). After layer

Scheme 4. Synthesis of compound 27.
parameters: radial growth rate (RGR) and initial growth
inhibition (IGI).30 To compare growth inhibition between the
different fungi, fungal growth was expressed as percentage of the
matching solvent control. All compounds were tested in
triplicate. The most active compounds (15, 20 and 27) were
assessed in 2-4 independent experiments.
We synthesized compound 47 from Grignard reagent,
prepared from 1-bromo-2,4,5-trifluorobenzene and ethyl 4-
iodobenzoate (38).
Ethyl 2',4',5'-trifluoro-[1,1'-biphenyl]-4-carboxylate (47): To
a dry flask (equipped with a magnetic stirrer and a rubber
septum) filled with argon, 1-bromo-2,4,5-trifluorobenzene 45
(4.39 ml, 37.5 mol) and anhydrous THF (10.6 ml) were added
and cooled to -20ºC; 2 M solution of iPrMgCl in THF (22.5 ml)
was slowly added to the reaction mixture, the reaction
temperature was adjusted to -10ºC and the mixture was stirred for
one hour until the Grignard exchange reaction was completed.
The resulting solution of 46 was added dropwise to the solution
of ethyl 4-iodobenzoate 38 (0.50 ml, 3.0 mmol),
tetrakis(triphenylphosphine)palladium (299 mg, 0.15 mmol) and
HMTA (21 mg, 0.15 mmol) in anhydrous THF (20 ml) at 0ºC.
The reaction mixture was maintained at 0ºC for one hour and
then stirred overnight at room temperature. The solvent was
removed by evaporation and the residue was purified by column
chromatography using diethylether/hexane (1/10) as an eluent to
obtain 0.70 g (83%) of compound 47 as white crystals. Mp 73-
The antifungal activity of compounds was assessed also
against a deletion mutant (∆bph) of CYP53A15 from C. lunatus,
prepared as described in 13, to determine if observed antifungal
effects can be ascribed solely to the CYP53A15 or compounds
interact also with other molecular targets contributing to the
antifungal activity.
2.5. Sequential and structural conservation of CYP53 linked to
functional relevance
FASTA formatted protein sequences of Cochliobolus lunatus
CYP53A15 (protein ID: 52559), Aspergillus niger CYP53A1
(protein ID: 162813) and Pleurotus ostreatus CYP53C11
(protein ID: 175456) were acquired from the Joint Genome
Institute at the U.S. Department of Energy (DOE-JGI;
http://genome.jgi.doe.gov/programs/fungi/index.jsf) and aligned
with the T-Coffee program at the European Bioinformatics
1
Institute (EMBL-EBI). The CYP modules tool in the Cytochrome
75°C; H-NMR (400 MHz, CDCl3): δ(ppm) 1.42 (t, J = 7.2 Hz,
31
P450 Engineering Database
was used to predict structurally
3H, CH3), 4.41 (q, J = 7.2 Hz, 2H, CH2), 7.02-7.08 (m, 1H, Ar-
H), 7.25-7.32 (m, 1H, Ar-H), 7.56 (d, J = 8.8 Hz, 2H, Ar-H), 8.12
(d, J = 8.8 Hz, 2H, Ar-H); 13C-NMR (100 MHz, DMSO-d6):
conserved regions of CYP53. Functionally relevant amino acid
residues and regions of CYP53 (Figure 3) were marked, based on
information from 32,33
.
2
2
δ(ppm) 14.08, 60.88, 106.80 (dd, 1H, JC,F = 21.2 Hz, JC,F’ =
29.7 Hz), 118.51 (dd, 1H, 3JC,F = 4.2 Hz, 2JC,F = 19.9 Hz), 124.00
(ddd, JC,F = 4.5 Hz, JC,F = 5.7 Hz, JC,F = 15.7 Hz), 129.14 (d,
2.6. Modeling of CYP53 and selected compounds
3
3
2
3
4
1H, JC,F = 3.2 Hz), 129.35, 129.54, 137.71, 146.53 (ddd, JC,F
=
By comparative modeling of fungal benzoate 4-
monooxygenases, based on the available template three-
dimensional (3D) model of Cochliobolus lunatus (PMDB ID:
PM0075149; http://www.biocomputing.it/PMDB/) 13, 3D models
of the enzyme soluble parts of CYP53A15 (Cochliobolus
lunatus), CYP53A1 (Aspergillus niger) and CYP53C11
(Pleurotus ostreatus) were generated. Homology modeling was
performed with the default parameters in Modeller 9v4
(http://www.salilab.org/modeller/), using the “allHmodel”
protocol to include hydrogen atoms and the “HETATM” protocol
to include HEM.
3.3 Hz, 2JC,F = 13.6 Hz, 1JC,F = 241,4 Hz), 149.08 (dt, 3JC,F = 13.2
Hz, 2JC,F = 13.6 Hz, 1JC,F = 248.5 Hz), 154.39 (ddd, 4JC,F = 1.2 Hz,
3JC,F = 9.8 Hz, JC,F = 245,1 Hz), 165.28; HR-MS (ESI): m/z
1
calcd. for C15H12O2F3 [M+H]+ 281.0789, found: 281.0787; HPLC
tR = 13.669 min (97.2%, Method 2).
2',4',5'-trifluoro-[1,1'-biphenyl]-4-carboxylic
acid
(27):
Compound 47 (490 mg, 1.75 mmol) was suspended in 12 ml of 1
M NaOH and 5 ml of dioxane and stirred for 24 hours; 1 M HCl
was added dropwise to adjust the pH to 1 and the white solid was
then filtered off and dried to obtain 0.43 g (97%) of compound
1
27 as white needles. Mp >300°C; H-NMR (400 MHz, DMSO-
After initial docking trials with some of the previously
published compounds13, the models have been optimized to better
fit the experimental data. For that purpose, several structures of
Bacillus megaterium BM3 cytochrome P450 (PDB IDs: 1YQP,
4HGF, 4HGG, 4HGH and 4KEW), sharing 34% identity with our
published template CYP53A15, were aligned and used for
optimization of models in Modeller. All three models were then
compared to the only available crystal structure of fungal
cytochromes P450, CYP51 from Saccharomyces cerevisiae
(PDB ID: 4LXJ; 20% sequence identity to CYP53A15).
d6): δ(ppm) 7.68-7.83 (m, 4H, Ar-H), 8.04 (dt, J = 8.4, 2.0 Hz,
2H, Ar-H), 13.13 (s, 1H, COOH); 13C-NMR (100 MHz, DMSO-
2
2
d6): δ(ppm) 106.84 (dd, 1H, JC,F = 21.3 Hz, JC,F’ = 29.5 Hz),
118.59 (dd, 1H, 3JC,F = 4.4 Hz, 2JC,F = 19.8 Hz), 124.22 (ddd, 3JC,F
3
2
3
= 4.4 Hz, JC,F = 6.2 Hz, JC,F= 15.4 Hz), 129.10 (d, 1H, JC,F
3.4 Hz), 129.62, 130.55, 137.45, 146.42 (dd, 4JC,F= 3.2 Hz, 2JC,F
=
=
12.4 Hz, 1JC,F = 241,0 Hz), 149.07 (dt, 3JC,F= 12.9 Hz, 2JC,F = 14.3
Hz, JC,F = 249.0 Hz), 154.42 (ddd, 4JC,F = 2.2 Hz, 3JC,F = 9.8 Hz,
1
1JC,F = 245,9 Hz), 166.96; HR-MS (ESI): m/z calcd. for
C13H8O2F3 [M+H]+ 253.0476, found: 253.0477; HPLC tR
18.250 min (100%, Method 1).
=
Additionally, we checked models’ quality by online version of
SWISS-MODEL Workspace34. All models were subjected to
Local Model Quality Estimation (Anolea - atomic mean force
potential, QMEAN6 - Composite scoring function for model
quality estimation), Global Model Quality Estimation (DFire -
All-atom distance-dependent statistical potential) and to
Stereochemistry Check (Procheck). Results (in Supporting
Information) indicated that on average 99.9 % of all atoms in
models were within the limits.
2.4. Antifungal screening
The antifungal activity of compounds against Cochliobolus
lunatus (Curvularia lunata) strain MUCL 38696 (m118),
Aspergillus niger N402 cspA1 and Pleurotus ostreatus Plo5
(ZIM collection of the Biotechnical Faculty, University of
Ljubljana, Slovenia) was examined with a fungal growth based
assay 24. MEA (Blakeslee’s formula) agar plates supplemented
with inhibitor (0.1 mM final concentration) or control solvent
were centrally inoculated with mycelial discs taken from 5-day-
old fungal culture and fungal growth at 28°C was monitored
daily. The fungal growth kinetics was determined using two
The atomic coordinates of most of the compounds were
26
downloaded from ZINC database
. When the atomic
coordinates of the compounds were not available in the database,

they were constructed based on the homology of the compound
with a known structure.
2.7. Docking compounds in the active site of CYP53 models
25
The HADDOCK modeling program was used for docking
selected compounds based on previously known biochemical
data and predicted biophysical interaction 13,23. Docking was
performed using the web server version of HADDOCK with a
Guru interface and with most of the parameters set to default. For
the enzyme-compound docking trials, compound active residues
were set to all and passive residues to determine automatically.
The active residues of the enzyme were predicted based on the
23
CYP53A15-compound 1a model
. For the clCYP53A15
enzyme model, the active residues were 41, 68, 69, 80, 83, 199,
364, 451 and 452; for the anCYP53A1 enzyme model, the active
residues were 112, 136, 137, 138, 150, 153, 262 and 452; and for
the poCYP53C11 enzyme model, the active residues were 79, 84,
108, 109, 123, 240, 243, 492 and 493, while the passive residues
were set to determine automatically. In order to gain the van der
Waals, electrostatic and desolvation energies for each enzyme-
compound model, HADDOCK automatically performed
molecular dynamics before and after each docking trial by
including water in the calculation.
3. Results and discussion
3.1. Compound selection using Rapid Overlay of Chemical
Structures (ROCS)
From the “drug-like” subset of the ZINC database 26, with
over 13 million compounds, we retained 70 compounds with the
highest shape and chemical similarity to 3-methyl-4-(1H-pyrrol-
1-yl)benzoic acid (1a; Figure 1). We experimentally evaluated
the 40 compounds listed in Supporting Information (SI) Table
S1. To guide the structure-antifungal activity relationship studies,
this list contains 30 compounds from ROCS screening selected
according to market availability and price, 7 additionally
synthesized or purchased compounds, the query compound (1a),
as well as CYP53 substrate (benzoic acid; BA) and product (4-
hydroxybenzoic acid; 4-HBA).
3.2. Structural features of benzoic acid derivatives conferring
antifungal activity
3.2.1. Inhibition of Cochliobolus lunatus growth.
Among the 40 compounds (SI Table S1) assayed in vitro as
inhibitors of fungal growth, we identified 8 (1b, 14, 15, 20, 21,
25, 27 and 28) with improved antifungal activity against
Cochliobolus lunatus compared to the parent compound 1a
(Figure 2). We have tested these compounds also against the
deletion mutant (∆bph) of CYP53A15. Since the observed
antifungal activity of compounds against wild type and mutant
fungal strain (not shown) is not statistically significantly different
(Two way ANOVA with Bonferroni post test; p<0.05), we
presume these compounds bind to and/ or inhibit also other, yet
unidentified, molecular targets in C. lunatus.
Figure 2. Antifungal activity of benzoic acid derivatives against
Cochliobolus lunatus, Aspergillus niger and Pleurotus ostreatus. Fungal
growth is expressed as a percentage of the matching control (C). Error bars
indicate standard error from the mean (n = 3). Initial growth inhibition (IGI)
is defined as an intercept on the time axis, whereby the linear growth phase is
extrapolated to a zero increase in diameter.
We observed improved inhibition of C. lunatus fungal growth
from the synthesized close analogue of 1a (compound 1b)
compared to the parent compound, indicating steric hindrance
when additional substituents, such as a methyl group (in 1a), are
present at the meta position of the BA phenyl ring. Substitution
with two or more methyl groups (compounds 3-6) significantly
decreased the antifungal activity against all tested fungi.
35
Similarly, in a study with Aspergillus fumigatus and A. flavus
,
moving the methyl, methoxy or chloro group on the BA phenyl
ring in the direction orto > meta > para increased the antifungal
activity of the tested compounds.
We also noticed that antifungal activity against C. lunatus is
greatly influenced by para substitution on the phenyl ring. For
example, the substitution of a pyrrole ring with less reactive
thiophene increased the antifungal activity and produced the most

3.2.3. Pleurotus ostreatus growth inhibition.
active compound (15) among all 37 derivatives of carboxylic
acid. However, the thiophene-3-il derivative (14) exhibited
almost 50% lower antifungal activity in comparison to the
thiophene-2-il derivative (15). Similarly, the thiophene
compounds extracted from the plant Echinops ritro (Astraceae)
were reported previously as highly effective against fungal plant
pathogens Fusarium oxysporum, Phomopsis viticola, P.
obscurans and three different Colletotrichum species.36
We observed a different pattern of antifungal properties of
benzoic acid derivatives against P. ostreatus (Figure 2). This was
expected, since P. ostreatus belongs to the basidiomycetes,
which have multiple CYP53 sequences in their genomes
20,21
and
are speculated to have acquired other functions in addition to
benzoate detoxification, such as involvement in the degradation
of wood components.14 It is possible that compounds interact
with other two CYP53 predicted in the P. ostreatus genome and
contribute in part to the observed antifungal effects.
Nevertheless, compound 15 remained the most active benzoic
acid derivative, almost completely inhibiting fungal growth at a
100 µM concentration. We also obtained promising results for
compounds 14, 20 and 21, which exhibited around 75%
inhibition of fungal growth. The presence of a thiophene ring
(compound 14) and fluoro substituted phenyl ring in the para
position of BA (compound 20) was again confirmed as the best
selection for promising antifungal activity. Interestingly,
compound 1b showed lower antifungal activity than the parent
compound 1a, indicating that steric effects are less important in
P. ostreatus.
Furthermore, we detected increased antifungal activity by the
addition of fluoro or chloro substituents on the second phenyl
ring (compounds 20, 21, and 25) in comparison to the
unsubstituted biphenyl derivative 24. Fluoro derivatives were
more active than chloro, and the para position was superior to the
meta position on the second phenyl ring. Based on these findings,
we synthesized a 2,4,5-trifluoro derivative (compound 27) that
confirmed our initial structure-antifungal activity information and
showed the highest antifungal activity among biphenyl-4-
carboxylic acid derivatives. The presence of three fluorines on
the second phenyl ring significantly increased antifungal activity
against C. lunatus, confirming that fluorine, in addition to
increasing ligand metabolic stability and membrane permeation,
can also enhance binding efficacy 37. In contrast, 4-fluorobenzoic
acid (compound 34) had no antifungal properties, presumably
due to the lack of a second ring system.
3.3. CYP53 sequence and structure conservation
Cytochromes P450 are highly diverse proteins and sequence
identities between distant CYP families may be less than 15%.38
Using the T-Coffee alignment tool, we determined the amino
acid sequence conservation of selected CYP53 members (Figure
3). Ascomycete CYP53A15 from Cochliobolus lunatus and
CYP53A1 from Aspergillus niger belong to CYP53 subfamily A
and share 66% sequence identity. Basidiomycete Pleurotus
ostreatus CYP53C11 is a member of CYP53 subfamily C and
has 54% and 50% amino acid residues identical to CYP53A15
and CYP53A1, respectively. Among 94 protein sequences in a
A non-aromatic 4-methylpiperidine derivative (compound 28)
also showed some antifungal properties; however, other
derivatives with an aliphatic ring (pyrrolidine, piperidine,
cyclohexyl) in the para position of BA (compounds 8-13, 29, 31)
lacked antifungal activity against C. lunatus. Most of the
pyrrazole derivatives, purchased as close analogues of 1a, were
also inactive and we ascribed this to methylation of the pyrrazole
ring (compounds 16-19). When the carboxylic acid was replaced
by sulphonic acid (compound 33), the antifungal activity was
lost. Interestingly, the presence of a polar acidic group on the
second phenyl ring is not preferred and results in the loss of
antifungal properties, as indicated by 4,4'-biphenyl-dicarboxylic
acid (compound 35). As expected, we detected no antifungal
activity against C. lunatus for 1,4-benzenedicarboxylic acid
(compound 36) and naphthalene-2-carboxylic acid (compound
37).
protein
alignment
of
the
CYP53A
subfamily
(http://www.cyped.uni-stuttgart.de/CONTENT/aln/aln971.html)
in the CYPED database 31, only 18 amino acid residues are
identical, while 78 amino acid positions were conserved.
The available crystal structures of cytochromes P450 show a
similar protein tertiary structure, which is predominantly
composed of α-helices labeled A to L (starting from the N-
terminus) and β-strands numbered 1-5; with the enzyme active
site buried deep in the core of the protein next to the heme
prosthetic group.39–41 The spatial conservation of CYP structures
is the highest in the regions associated with binding of heme and
protein redox partners.38,41 The characteristic P450 consensus
sequence (CxG) with an absolutely conserved Cys residue (as
fifth ligand to the heme iron) is located just before the L helix in
the β-bulge region (called the Cys pocket).33,38 The second
conserved motif (ExxR in the K helix) is important for the
stabilization of the meander loop and probably for the
maintenance of the core structure.33 The central part of the I helix
(A/G-G-x-D/E-T-T/S) corresponds to the putative oxygen
binding motif.42 The reductase interaction sites are proposed in
the helices J/J’ and in the insertion between the meander loop and
the Cys pocket.39 The highest variability and flexibility are in the
3.2.2. The antifungal activity of benzoic acid
derivatives against Aspergillus niger.
The inhibition of Aspergillus niger growth (Figure 2) by
selected benzoic acid derivatives produced a similar pattern of
antifungal properties as observed for C. lunatus but resulted in
six moderately active compounds. Compound 15 inhibited only
30% of fungal growth, compared to almost 80% inhibition of C.
lunatus growth. Compounds 1b, 20 and 25 showed similar
antifungal properties in both fungi and in the same range as
compound 15. Compound 13 was equally effective, but selective
towards A. niger. We noticed decreased antifungal activity of
2',4',5'-trifluoro-[1,1'-biphenyl]-4-carboxylic acid (compound 27)
against A. niger. Compounds with fluoro substituents on the
second phenyl ring (20, 25 and 27) again showed promising
antifungal activity, while chloro substituents were less effective
(compounds 21-23). We detected no antifungal activity against A.
niger for other benzoic acid derivatives (pyrrolidine, pyrrazole,
piperidine, etc).
43
substrate recognition sites (SRS 1-6; Figure 3) that are located
in the termini of helices F and G, forming the “roof” of the active
site cavity, and helix I on one side of the heme and helix K, β-
strands 1-4 and 4-1 and the BC loop on the opposite side.39,41

Figure 3. Amino acid sequence alignment of selected CYP53 enzymes indicating conserved secondary structure elements and variable substrate recognition
sites (SRS; boxed). Red letters mark residues that are conserved and strictly conserved (bold) among 94 proteins of the CYP53A subfamily in the CYPED
database.31 Predicted secondary structure elements 39,32 are highlighted green (α-helices) and yellow (β-sheets). Underlined residues indicate cytochrome P450
signature motifs ExxR and CxG.33 The Cys pocket region is highlighted blue and the meander loop is colored dark green.
3.4. CYP53 modeling
enter the buried active site from the membrane through the pw2a
channel and the hydroxylated water-soluble product exits through
pw2c. The pw2a channel opening is in the lipid membrane and
lies between the FG loop and the B’ helix/BC loop, while pw2c
is located between the B’ helix/BC loop and helices G and I.40
We noticed that in the clCYP53A15 model, the channel with
inhibitor 1a traveling to the active site cavity above the heme
plane is opened and easily accessible (Figure 4a). However, in
the anCYP53A1 model, two regions confine access to the
catalytically active heme iron (Figure 4b). The first region is part
of SRS5 and consists of residues P-R-E-I-P-A (positions 389-
394) in the β1-4 strand. Due to the flanking proline residues, the
flexibility of this region is limited and restricts the passage of the
inhibitor to the active site cavity. The second region, comprised
of residues S-D-F-Y-D (positions 100-104) in the B-1 helix, is
part of SRS1. It protrudes towards the heme, preventing the
inhibitor from making a turn and approaching the heme at an
appropriate angle. Access to the active site of the anCYP53A1
model is additionally sterically restricted by residue F-43 and the
charge restricted by residue R-77, both residues in this way
acting as a gating mechanism for substrate/inhibitor specificity.
In the poCYP53C11 model (Figure 4c), passage of the inhibitor
is restricted to a certain extent by S-E-F-Y-D (positions 106-110)
in the SRS1 region. Unlike in anCYP53A1, this part is not rigid
and allows a slight movement of the loop to accommodate the
inhibitor. The second region, narrowing the inhibitor pathway
towards the heme iron, resides in SRS6 and consists of residues
R-E-G (positions 489-491) in the β4-1 strand. Although, the M-
The enzymatic activity of CYP53 family members is mainly
restricted to 4-hydroxylation of benzoate and some mono-
substituted benzoate derivatives, indicating precision in substrate
recognition and catalysis. Using the Cochliobolus lunatus
13
CYP53A15 model
with inhibitor 1a as a template 23, we
constructed three CYP53 homology models labeled
clCYP53A15, anCYP53A1 and poCYP53C11 (Figure 4) to
provide an explanation of such precision, based on protein
structural information. All models generated in this work were
aligned to the crystal structure of Saccharomyces cerevisiae
44
CYP51
(PDB 4LXJ, data not shown), currently the only
available fungal cytochrome P450 crystal structure. The overall
RMSD parameter for backbone residues, which reflects the
similarity between the compared structure and Saccharomyces
cerevisiae CYP51, were calculated to be 3.0, 5.1, 7.7 Å for
anCYP53A1, poCYP53C11 and clCYP53A15, respectively. We
noticed that the amino acid residues involved in the binding of
heme and stabilization of the CYP core structure are always
placed in similar positions and this structural folding is
energetically favorable. This trend has also been observed in the
folding of over 6,000 artificial chimeric P450 heme proteins
created by SCHEMA-guided recombination of parent
cytochromes P450.45
Most eukaryotic cytochromes P450 are membrane bound
proteins and presumably utilize various substrate access and
product egress routes.40,46 A hydrophobic substrate is likely to

233 residue in poCYP53C11 is incapable of sterically blocking
inhibitor entrance, as F-43 and R-77 residues in anCYP53A1 do,
this residue adds a charge restriction for inhibitor specificity.
CYP53 enzymes are membrane-bound proteins and solving
their crystal structure remains a challenging task. However, with
homology modeling based on a previously published template 13
,
we generated energy minimized 3D models of selected CYP53
47
enzymes and performed HADDOCK solvated docking
of
CYP53 substrate (BA), product (4-HBA) and 38 ligands with a
potential antifungal activity. A total of 200 models was generated
for each compound in the rigid body docking step. The top 100
structures were then subjected to refinement steps with a flexible
interface and a water layer around the complex to obtain high
quality predictions of complex structures in clusters sorted
according to HADDOCK score (SI Table S1). HADDOCK
docking scores for the CYP53A15 model ranged from −22.9
6.7 for substrate (BA) to −63.2
8.9 for compound 3. We
obtained similar results for the CYP53A1 model with substrate
and compound 34, which had the lowest scores of −16.9 1.7
and −14.4 2.6, respectively; while the highest score (−41.0
2.7) was attributed to compound 33. Docking the compounds into
the CYP53C11 model resulted in HADDOCK scores ranging
from −45.0 0.7 for compound 34 to −82.6 3.9 for compound
9. The size of the interface in a protein-ligand complex is
indicated by the buried surface area (BSA).25 In the CYP53A15
model, the BSA ranges between 308.2 16.3 Å2 for BA and
545.8
21.7 Å2 for compound 27. A similar range of BSA
values, from 330.5 8.0 Å2 (BA) to 585.8 25.7 Å2 (27), was
also calculated for the CYP53A1 model, while the ligand-
CYP53C11 complex interface varied from 302.9 10.0 Å2 (BA)
to 575.8 24.7 Å2 (33).
We grouped the compounds into 7 categories, based on the
nature of the ring system at the para position of the BA phenyl
ring, and compared the energy contributions (electrostatic, van
der Waals and desolvation) for each ligand-CYP53 combination
(Figure 5). The electrostatic interactions are the most significant
contributor in the binding of compounds to all three CYP53
models indicated by large negative energy values. However,
there are considerable differences among CYP53 enzymes. While
CYP53C11 favors displacement of water molecules from the
active site cavity on ligand binding, whereby water molecules
possibly mediate H-bonds interactions between the enzyme and
ligands, the desolvation of CYP53A1 and CYP53A15 is
energetically unfavorable and water molecules possibly mediate
charge-charge interactions between the enzyme and ligands.
Furthermore, we noticed favorable van der Waals contributions
(E value less than −10 kJ/mol), possibly stabilizing CYP53A1
and CYP53C11 interactions with the compounds, while these
effects are less pronounced in CYP53A15-ligand interactions,
with a few exceptions.
Figure 4. Comparison of CYP53 homology models from (a)
Cochliobolus lunatus clCYP53A15 (brown) and the previously published
model 23 (gray), (b) Aspergillus niger anCYP53A1 (violet) and (c) Pleurotus
ostreatus poCYP53C11 (cyan). Inhibitor 1a (green) and heme (yellow) are
indicated in stick mode, while CYP53 enzymes are shown in a cartoon
presentation. The residues narrowing the inhibitor pathway in anCYP53A1
and poCYP53C11 enzyme models are shown in violet and cyan sticks,
respectively.
3.5. Docking reveals differences in the CYP53-ligand
interactions

Figure 5. Van der Waals, electrostatic and desolvation energy as calculated by HADDOCK 47 for each CYP53 enzyme model and compound combination.
As with other cytochromes P450 40,46, with the molecular
docking of the most active compounds in the clCYP53A15
model and molecular dynamics simulation (Figure 6), we
predicted various pathways leading to the CYP53A15 active site
cavity above the heme plane. The substrate (BA) probably enters
the pw2a channel lined by αF- and αI-helices and β1-4, β2-1, β4-
1 and β4-2-sheets. The BA in the active site cavity is oriented
perpendicular to the heme plane, sloped by 80°, with COOH
group oriented away from the heme plane (Figure 6a). The query
compound (1a) and compounds 15, 27, and 20 all access the
CYP53A15 active site cavity via the pw2 channel formed by
similar helices and strands as described above (Figure 6b-e).
Their orientation towards the heme plane in the active site cavity
is horizontal, sloped by 20°, with the COOH group pointing away
from the heme. We observed that the most active compound (15)
travels through the pw2a channel and then turns inside the active
site cavity, finally to position itself with the carboxyl group
towards the substrate access channel (Figure 6c; circled) thereby
preventing the passage of the BA to the catalytic core. Docking
of inactive compound 18 (Figure 6f) indicates that this compound
only interacts with the residues at the CYP53A15 protein surface.
Figure 6. The most likely passage of substrate and putative inhibitors
from the protein surface to the active site of clCYP53A15. The 3D
HADDOCK models of inhibitor-clCYP53A15 enzyme complexes were
superimposed on the clCYP53A15 enzyme alone using the Pymol 1.3
program (DeLano WL, http://www.pymol.org). The clCYP53A15 enzyme is
shown in a gray cartoon presentation; inhibitors (cyan), substrate (brown),
heme (yellow) are in stick presentation. a, benzoic acid; b, compound 1a; c,
compound 15; d, compound 27; e, compound 20; f, compound 18.

Using the CYP53A1 model and antifungal active compounds
in the molecular docking and molecular dynamics simulation
(Figure 7) provided an explanation for the lower activity. The
substrate (BA) can enter the CYP53A1 active site cavity via 2
different channels. One is similar to the CYP53A15 pw2a
channel, but at a 15° slope, lined by αF- and αI-helices and the
β1-4-sheet, and the second is channel linked by αI-helix and β4-1
and β4-2-sheets. BA binding to both sites seems equally possible,
since the energy of the models in both states is similar. The final
position of BA above the heme plane is perpendicular, sloped by
60°, with the COOH group oriented away from the heme. The
majority of compounds that moderately inhibit A. niger growth
are bound at the rim of channel pw2a. The only compound that
inserts into the active site cavity is compound 27, which positions
with the COOH group away from the heme and with rings
oriented horizontally to the heme at a slope of 15°, by also
blocking the second BA pathway. Although other compounds
enter the pw2a channel, the reason for their inactivity or partial
inactivity may be an inability to block the BA secondary channel
path.
Molecular docking of active benzoic acid derivatives in the
CYP53C11 model revealed that all compounds travel via the
same access channel as BA but to different depths (Figure 8).
The BA channel in this enzyme is similar to the pw2a
anCYP53A1 channel and BA is oriented into the active cavity,
similarly as in clCYP53A15. All compounds position with the
COOH group away from the heme and with rings oriented
horizontally to the heme at a slope of 15°.
Figure 8. The most likely passage of substrate and putative inhibitors
from the protein surface to the active site of poCYP53C11. The 3D
HADDOCK models of inhibitor- poCYP53C11 enzyme complexes were
superimposed on the poCYP53C11 enzyme alone using the Pymol program
(DeLano WL, http://www.pymol.org). The poCYP53C11 enzyme is shown in
a light purple cartoon presentation; inhibitors (cyan), substrate (brown), heme
(yellow) are in stick presentation. a, benzoic acid; b, compound 1a; c,
compound 15; d, compound 27; e, compound 20: f, compound 35.
4. Conclusions
The antifungal activity of BA and its derivatives is well
documented in yeasts and is based on the undissociated lipophilic
acid form, which can permeate the plasma membrane and cause
intracellular acidification, which leads to oxidative stress, protein
aggregation, lipid peroxidation and inhibition of membrane
trafficking.48,49 In an adaptive response, yeasts counteract these
effects by activating H+-ATPases, changing the lipid composition
of the plasma membrane, remodeling the cell wall structure and
detoxification via multidrug resistance transporters (in particular
Pdr12p).49 However, in the filamentous ascomycetes and
basidiomycetes, the BA is primarily detoxified by CYP53
enzymes.13,16,18 The critical role of CYP53 in the primary fungal
metabolism, together with the absence of homologues in higher
eukaryotes and relatively conserved active site cavity, make
CYP53 an attractive target in antifungal research.14,23
In this study, we generated three structural models of CYP53
enzymes and evaluated the antifungal activity of benzoic acid
derivatives against Cochliobolus lunatus, Aspergillus niger and
Pleurotus ostreatus. We found 8 compounds with improved
antifungal activity compared to the parent compound 1a.
Moreover, we determined the key structural features giving the
compounds’ antifungal properties: i) the carboxylic group is
important for biological activity, ii) the small group substitutions
on the BA phenyl ring at positions orto and meta decrease
activity due to steric hindrance, iii) substitution of the pyrrole
ring with the less reactive thiophene or fluorinated phenyl ring is
beneficial, while other substitutions abolish the antifungal
activity.
Figure 7. The most likely passage of substrate and putative inhibitors
from the protein surface to the active site of anCYP53A1. The 3D
HADDOCK models of inhibitor- anCYP53A1 enzyme complexes were
superimposed on the anCYP53A1 enzyme alone using the Pymol program
(DeLano WL, http://www.pymol.org). The anCYP53A1 enzyme is shown in
a golden cartoon presentation; inhibitors (cyan), substrate (brown), heme
(yellow) are in stick presentation. a, benzoic acid; b, compound 1a; c,
compound 15; d, compound 27; e, compound 20; f, compound 33.
In addition to these findings, in our molecular docking and
molecular dynamics experiments we predicted
a gating
mechanism involving π-stacked interactions (present also in other
CYPs; e.g., human CYP3A4 50), the amino acid residues

contributing to H-bonding and charge-charge interactions with
CYP53. The ligand COOH group, important for biological
activity, is always positioned away from the catalytically active
heme iron. In C. lunatus CYP53A15 (Figure 9; a-c), we observed
that residues R-110 and T-212 direct the most active antifungal
compound (15) deep towards the heme plane, R-109 and E-224
then mediate the turning and positioning of the compound at such
an angle that it blocks the pathway for BA. As in our previous
study 23, we found a possible second access channel for BA. The
occurrence of two access channels for BA is even more probable
in CYP53A1 from A. niger, but to be a successful inhibitor of
CYP53A1 the compound has to close both channels. As seen in
Figure 9 (d-f), compound 15 cannot penetrate deep into the
enzyme active site due to the specific gating mechanism
comprised of a π-stacking interaction with F-210 and H-bonding
with R-77. In Pleurotus ostreatus CYP53C11 (Figure 9; g-i),
amino acid residues W-200, R-39 and R-83 mediate H-bonding
interactions with compound 15, while T-69 contributes to the π-
stacking interaction.
Supplementary Material
Supplementary data (Table S1, Chemical characterization,
Model quality evaluation and 1H and 13C NMR spectra)
associated with this article can be found, in the online version, at
.
References and notes
1.
In Fungal Research Trust 20 th Anniversary meeting.; London,
2011.
2.
3.
Pennisi, E. Science 2010, 327, 804–5.
Fisher, M. C.; Henk, D. A.; Briggs, C. J.; Brownstein, J. S.;
Madoff, L. C.; McCraw, S. L.; Gurr, S. J. Nature 2012, 484, 186–
94.
4.
5.
Sheng, C.; Zhang, W. Curr. Med. Chem. 2011, 18, 733–66.
We believe that these data contribute to a better understanding
of CYP53-ligand interactions by identifying possible substrate/
inhibitor channels and assist in the design of more potent and
selective inhibitors targeting CYP53, thereby facilitating
antifungal drug research.
Ostrosky-Zeichner, L.; Casadevall, A.; Galgiani, J. N.; Odds, F. C.;
Rex, J. H. Nat. Rev. Drug Discov. 2010, 9, 719–27.
6.
Grigoriev, I. V; Nikitin, R.; Haridas, S.; Kuo, A.; Ohm, R.; Otillar,
R.; Riley, R.; Salamov, A.; Zhao, X.; Korzeniewski, F.; Smirnova,
T.; Nordberg, H.; Dubchak, I.; Shabalov, I. Nucleic Acids Res.
2014, 42, D699–704.
7.
8.
9.
Cools, H. J.; Hammond-Kosack, K. E. Mol. Plant Pathol. 2013, 14,
197–210.
Liao, C.; Sitzmann, M.; Pugliese, A.; Nicklaus, M. C. Future Med.
Chem. 2011, 3, 1057–85.
Gassner, N. C.; Tamble, C. M.; Bock, J. E.; Cotton, N.; White, K.
N.; Tenney, K.; St Onge, R. P.; Proctor, M. J.; Giaever, G.; Nislow,
C.; Davis, R. W.; Crews, P.; Holman, T. R.; Lokey, R. S. J. Nat.
Prod. 2007, 70, 383–90.
10.
Srinivasan, A.; Leung, K. P.; Lopez-Ribot, J. L.;
Ramasubramanian, A. K. MBio 2013, 4.
11.
12.
Črešnar, B.; Petrič, S. Biochim. Biophys. Acta 2011, 1814, 29–35.
Kelly, S. L.; Kelly, D. E. Philos. Trans. R. Soc. Lond. B. Biol. Sci.
2013, 368, 20120476.
Figure 9. Compound 15 interactions with CYP53 enzyme models of
Cochliobolus lunatus (a-c), Aspergillus niger (d-f) and Pleurotus ostreatus
(g-i). The atomic coordinates of the HADDOCK models were superimposed
on the model of clCYP53A15 using the Pymol program (DeLano WL,
http://www.pymol.org). Inhibitor 15 (cyan) and heme (yellow) are in stick
representation. Enzyme models are presented in cartoon representation.
13.
14.
Podobnik, B.; Stojan, J.; Lah, L.; Kraševec, N.; Seliškar, M.;
Rižner, T. L.; Rozman, D.; Komel, R. J. Med. Chem. 2008, 51,
3480–6.
Jawallapersand, P.; Mashele, S. S.; Kovačič, L.; Stojan, J.; Komel,
R.; Pakala, S. B.; Kraševec, N.; Syed, K. PLoS One 2014, 9,
e107209.
Acknowledgments
15.
16.
17.
18.
Fukuda, H.; Nakamura, K.; Sukita, E.; Ogawa, T.; Fujii, T. J.
Biochem. 1996, 119, 314–8.
We sincerely thank Dr. Samo Turk for the similarity-based
virtual screening. We wish to acknowledge the help provided by
Marko Bunc and Žan Koren in the synthesis and analysis of
compounds as well as with antifungal screening. Thanks to David
R. Nelson from Cytochrome P450 Nomenclature Committee for
the naming of CYP53C11 from Pleurotus ostreatus. We thank
Martin Cregeen for language editing. We thank OpenEye
Scientific Software, Inc. for free academic licenses for their
software. We are grateful for the financial support of the
Slovenian Research Agency by Grants J4-2212, P1-0104 and P1-
0207.
Faber, B. W.; van Gorcom, R. F.; Duine, J. A. Arch. Biochem.
Biophys. 2001, 394, 245–54.
Fraser, J. A.; Davis, M. A.; Hynes, M. J. Fungal Genet. Biol. 2002,
35, 135–46.
Matsuzaki, F.; Wariishi, H. Biochem. Biophys. Res. Commun. 2005,
334, 1184–90.

19.
20.
21.
Harwood, C. S.; Parales, R. E. Annu. Rev. Microbiol. 1996, 50,
553–90.
35.
36.
Kim, J. H.; Campbell, B. C.; Mahoney, N.; Chan, K. L.; Molyneux,
R. J.; Balajee, A. Fungal Biol. 2010, 114, 817–24.
Ide, M.; Ichinose, H.; Wariishi, H. Arch. Microbiol. 2012, 194,
243–53.
Fokialakis, N.; Cantrell, C. L.; Duke, S. O.; Skaltsounis, A. L.;
Wedge, D. E. J. Agric. Food Chem. 2006, 54, 1651–5.
Moktali, V.; Park, J.; Fedorova-Abrams, N. D.; Park, B.; Choi, J.;
Lee, Y.-H.; Kang, S. BMC Genomics 2012, 13, 525.
37.
38.
Müller, K.; Faeh, C.; Diederich, F. Science 2007, 317, 1881–6.
Werck-Reichhart, D.; Feyereisen, R. Genome Biol. 2000, 1,
22.
23.
Degani, O. Adv. Biosci. Biotechnol. 2014, 05, 340–352.
REVIEWS3003.
Berne, S.; Podobnik, B.; Zupanec, N.; Novak, M.; Kraševec, N.;
Turk, S.; Korošec, B.; Lah, L.; Šuligoj, E.; Stojan, J.; Gobec, S.;
Komel, R. J. Chem. Inf. Model. 2012, 52, 3053–63.
39.
40.
41.
42.
43.
44.
Sirim, D.; Widmann, M.; Wagner, F.; Pleiss, J. BMC Struct. Biol.
2010, 10, 34.
Schleinkofer, K.; Sudarko; Winn, P. J.; Lüdemann, S. K.; Wade, R.
C. EMBO Rep. 2005, 6, 584–9.
24.
Korošec, B.; Sova, M.; Turk, S.; Kraševec, N.; Novak, M.; Lah, L.;
Stojan, J.; Podobnik, B.; Berne, S.; Zupanec, N.; Bunc, M.; Gobec,
S.; Komel, R. J. Appl. Microbiol. 2013.
Kirchmair, J.; Williamson, M. J.; Tyzack, J. D.; Tan, L.; Bond, P.
J.; Bender, A.; Glen, R. C. J. Chem. Inf. Model. 2012, 52, 617–48.
25.
26.
27.
28.
29.
30.
31.
De Vries, S. J.; van Dijk, M.; Bonvin, A. M. J. J. Nat. Protoc. 2010,
5, 883–97.
Poulos, T. L.; Madrona, Y. Biotechnol. Appl. Biochem. 2013, 60,
128–33.
Irwin, J. J.; Sterling, T.; Mysinger, M. M.; Bolstad, E. S.; Coleman,
R. G. J. Chem. Inf. Model. 2012, 52, 1757–68.
Lepesheva, G. I.; Waterman, M. R. Mol. Cell. Endocrinol. 2004,
215, 165–70.
Zhu, R.; Xing, L.; Wang, X.; Cheng, C.; Su, D.; Hu, Y. Adv. Synth.
Catal. 2008, 350, 1253–1257.
Monk, B. C.; Tomasiak, T. M.; Keniya, M. V; Huschmann, F. U.;
Tyndall, J. D. A.; O’Connell, J. D.; Cannon, R. D.; McDonald, J.
G.; Rodriguez, A.; Finer-Moore, J. S.; Stroud, R. M. Proc. Natl.
Acad. Sci. U. S. A. 2014, 111, 3865–70.
Molander, G. A.; Ryu, D.; Hosseini-Sarvari, M.; Devulapally, R.;
Seapy, D. G. J. Org. Chem. 2013, 78, 6648–56.
Sheng, R.; Xu, Y.; Hu, C.; Zhang, J.; Lin, X.; Li, J.; Yang, B.; He,
Q.; Hu, Y. Eur. J. Med. Chem. 2009, 44, 7–17.
45.
46.
47.
48.
49.
50.
Otey, C. R.; Landwehr, M.; Endelman, J. B.; Hiraga, K.; Bloom, J.
D.; Arnold, F. H. PLoS Biol. 2006, 4, e112.
López-Malo, A.; Alzamora, S. M.; Argaiz, A. Food Microbiol.
1995, 12, 213–219.
Wade, R. C.; Winn, P. J.; Schlichting, I.; Sudarko J. Inorg.
Biochem. 2004, 98, 1175–82.
Fischer, M.; Knoll, M.; Sirim, D.; Wagner, F.; Funke, S.; Pleiss, J.
Bioinformatics 2007, 23, 2015–7.
Van Dijk, A. D. J.; Bonvin, A. M. J. J. Bioinformatics 2006, 22,
2340–7.
32.
33.
34.
Gricman, Ł.; Vogel, C.; Pleiss, J. Proteins 2014, 82, 491–504.
Syed, K.; Mashele, S. S. PLoS One 2014, 9, e95616.
Hazan, R.; Levine, A.; Abeliovich, H. Appl. Environ. Microbiol.
2004, 70, 4449–57.
Mira, N. P.; Teixeira, M. C.; Sá-Correia, I. OMICS 2010, 14, 525–
40.
Arnold, K.; Bordoli, L.; Kopp, J.; Schwede, T. Bioinformatics
2006, 22, 195–201.
Fishelovitch, D.; Shaik, S.; Wolfson, H. J.; Nussinov, R. J. Phys.
Chem. B 2009, 113, 13018–25.