Fragment-based design of symmetrical bis-benzimidazoles as selective inhibitors of the trimethoprim-resistant, type II R67 dihydrofolate reductase

Article abstract of DOI:10.1021/jm201645r

The continuously increasing use of trimethoprim as a common antibiotic for medical use and for prophylactic application in terrestrial and aquatic animal farming has increased its prevalence in the environment. This has been accompanied by increased drug

Full text of DOI:10.1021/jm201645r

Article
pubs.acs.org/jmc
Fragment-Based Design of Symmetrical Bis-benzimidazoles as
Selective Inhibitors of the Trimethoprim-Resistant, Type II R67
Dihydrofolate Reductase
Dominic Bastien,† Maximilian C. C. J. C. Ebert,† Delphine Forge,‡ Jacynthe Toulouse,†
Natalia Kadnikova,§ Florent Perron,‡ Annie Mayence,∥ Tien L. Huang,∥ Jean Jacques Vanden Eynde,‡
and Joelle N. Pelletier*,†,§
†Dep
‡Laboratoire de Chimie Organique, Universite
§Dep
artement de Chimie, Universite de Montrea
́
artement de Biochimie, Universite
́
de Montrea
de Mons-UMONS, 20 Place du Parc, B-7000 Mons, Belgium
l, C.P. 6128, Succursale Centre-ville Montreal, Quebec H3C 3J7, Canada
́ ́ ́
l, C.P. 6128, Succ. Centre-ville Montreal, Quebec H3C 3J7, Canada
́
́
́
́
́
́
∥Division of Basic Pharmaceutical Sciences, Xavier University of Louisiana, 1 Drexel Drive, New Orleans, Louisiana 70125, United
States
S
* Supporting Information
ABSTRACT: The continuously increasing use of trimetho-
prim as a common antibiotic for medical use and for
prophylactic application in terrestrial and aquatic animal
farming has increased its prevalence in the environment.
This has been accompanied by increased drug resistance,
generally in the form of alterations in the drug target,
dihydrofolate reductase (DHFR). The most highly resistant
variants of DHFR are known as type II DHFR, among which
R67 DHFR is the most broadly studied variant. We report the
first attempt at designing specific inhibitors to this emerging
drug target by fragment-based design. The detection of
inhibition in R67 DHFR was accompanied by parallel monitoring of the human DHFR, as an assessment of compound
selectivity. By those means, small aromatic molecules of 150−250 g/mol (fragments) inhibiting R67 DHFR selectively in the low
millimolar range were identified. More complex, symmetrical bis-benzimidazoles and a bis-carboxyphenyl were then assayed as
fragment-based leads, which procured selective inhibition of the target in the low micromolar range (Ki = 2−4 μM). The putative
mode of inhibition is discussed according to molecular modeling supported by in vitro tests.
resistance and its impact on the long-term effectiveness of this
low-cost antibiotic.
INTRODUCTION
■
Trimethoprim (TMP) has been used as an antibiotic worldwide
for the past 50 years, frequently in combination with
sulfamethoxazole.1 Today, TMP remains a primary treatment
of urinary Escherichia coli infections and is used to treat
respiratory infections such as pneumonia. It has been used to
treat sexually transmitted infections such as gonorrhea, but this
has resulted in high TMP resistance.2 As well, TMP with
sulfamethoxazole shows potential for the treatment of tuber-
culosis,2b which causes an estimated 1.7 million deaths each
year, and was recommended by the WHO in 2011 to reduce
mortality due to opportunistic infections in adults living with
HIV/AIDS.3 In addition to medical use, TMP is also used for
veterinary applications, including prophylactic use in porcine,
bovine,4 salmon,5 and shrimp farming.5,6 These result in large-
scale environmental dissemination of the antibiotic,7 which is
inevitably accompanied by horizontal transfer of plasmids
carrying multiple drug resistance genes to other microbes,
including human pathogens.5,8 As a result, there is a growing
concern worldwide over the increased incidence in TMP
TMP is a specific inhibitor of bacterial dihydrofolate
reductases (DHFRs). DHFRs are oxidoreductases that catalyze
the reduction of dihydrofolate (DHF) to tetrahydrofolate, using
nicotinamide dinucleotide phosphate (NADPH) as a hydride-
donating cofactor. DHFR is an essential enzyme for the
synthesis of DNA precursors (purines, thymidylate) in all living
cells, where it is chromosomally encoded and has been highly
conserved throughout evolution.9 Because of its essential role in
cellular proliferation, DHFR is a central target in the control of
proliferative diseases, including microbial infections. There are
sufficient structural differences between human DHFR
(hDHFR) and its microbial counterparts to allow for selective
inhibition by compounds such as pyrimethamine and cyclo-
guanyl, with high clinical efficacy.10 TMP is such a selective
inhibitor of microbial DHFRs; it binds to E. coli DHFR
Received: December 5, 2011
Published: March 16, 2012
© 2012 American Chemical Society
3182
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
approximately 4 orders of magnitude more tightly than it binds
to hDHFR.11 Among the mechanisms of TMP resistance that
have been reported,12 R-plasmid-encoded type II DHFR13
constitutes an important health concern in both humans and
livestock by sustaining bacterial proliferation.14 Type II R67
DHFR catalyzes the same reaction as chromosomally encoded
DHFR15 albeit with a lower efficiency; yet, it evades the action
of TMP, providing bacterial drug resistance.13,16 Indeed, R67
DHFR is highly resistant to TMP, with IC50 and Ki > 106 fold
higher than its chromosomal counterpart.16,17 In contrast to the
DHF substrate, TMP has no amide with which to H-bond with
the Ile68 backbone. In addition, R67 DHFR has no carboxylate
in its hydrophobic active site18 and thus cannot form
electrostatic interactions with the protonated TMP as seen in
the chromosomal DHFR.19 Indeed, R67 DHFR is genetically
and structurally unrelated to chromosomal DHFRs, resulting in
important structural differences that form the basis of its
intrinsic TMP resistance. This enzyme is a homotetramer,
where four identical protomers form a symmetrical active-site
pore that binds a single DHF/NADPH pair for turnover.19 The
active-site cavity is a “binding hot-spot” of relatively low
specificity that can bind two DHF or two NADPH molecules at
once17,19 and reduce some folate analogues.15 This binding
promiscuity runs counter the excruciatingly high binding
stringency of hDHFR, which should allow for discovery of
selective inhibitors of R67 DHFR that will not significantly
inhibit hDHFR. Taken along with TMP, inhibitors of R67
DHFR would break TMP resistance granted by R67 DHFR.
Novobiocin, a DNA gyrase inhibitor, inhibits the activity of
R67 DHFR as does congo red, a diazo dye.17 Neither is specific
to R67 DHFR; yet, they provide structural information relative
to functional groups of interest in inhibitor design. In this work,
we applied fragment-based inhibitor screening20 to identify
simple molecules that inhibit R67 selectively. While such small
molecules may exhibit only a weak inhibitory effect due to their
small size, they may form high quality interactions as expressed
by the ratio of affinity to number of heavy atoms, or ligand
efficiency (LE).21 Inhibitory “fragments” allow the subsequent
identification or design of more complex molecules sharing
those substructures, which may display improved inhibition or
other properties relevant to drug design. Molecules resulting
from fragment-based design may require less optimization than
their counterparts identified by more conventional high-
throughput screening (HTS) campaigns, which may contain
nondesirable functional groups.21
serving features of the fragment molecules were subsequently
tested for increased inhibition. Kinetic analysis and molecular
modeling provided insight into the putative binding mode of
these selective, micromolar inhibitors. Preliminary in vivo tests
against adherent mouse fibroblast 3T6 cells revealed weak
cytotoxicity.
MATERIALS AND METHODS
■
Materials. Dimethylsulfoxide (DMSO) was purchased from Fisher
Scientific (Fair Lawn, NJ). The origin of the screened fragments is
indicated in Table S1 in the Supporting Information. Solvents and
reagents for synthesis are commercially available (Aldrich, Acros
Organics, Fisher Scientific, Sigma Chemical Co.) and were used
without further purification. DHF was synthesized from folic acid as
described.22 β-NADPH was purchased from Alexis biochemicals (San
Diego, CA). Unless otherwise indicated, other reagents for enzyme
expression and assays were from BioShop Canada Inc. (Burlington,
ON).
Synthesis of Symmetrical Bis-benzamidine, Bis-benzimida-
zoles, and Bis-p-hydroxybenzoate. Compounds 8 and 9 were
prepared and characterized following a known procedure.23 Briefly, a
mixture of 4-hydroxybenzaldehyde (5.12 g, 42 mmol), a dibromoal-
kane (C4 or C5) or dibromopropan-2-ol (20 mmol), and potassium
carbonate (2.76 g, 20 mmol) in ethanol (10 mL) was heated under
reflux for 8 h. After it was cooled, the precipitate was filtered and
successively washed with water, ethanol, and ether. The bisbenzalde-
hyde was sufficiently pure to proceed to the next step. A mixture of the
bisbenzaldehyde (3 mmol), sodium pyrosulfite (0.57 g, 3 mmol), 3,4-
diaminobenzoic acid (0.91 g, 6 mmol), and water (3 mL) in ethanol (9
mL) was microwave irradiated (Biotage) for 15 min at 140 °C. After it
was cooled, the precipitate was filtered and thoroughly washed with
water, ethanol, and ether. Yields: 8, 97%; 9, 94%. The synthesis of
compound 8b was the same except for substitution of the
diaminobenzoic acid with ortho-phenylenediamine (6 mmol).23 The
synthesis of compound 8a has been reported;24 however, a slightly
modified version was performed. In short, a mixture of 4-
hydroxybenzoic acid (0.69 g, 5 mmol), dibromopentane (7.5 mmol),
potassium hydroxyde (0.84 g, 15 mmol), and water (1.5 mL) in
ethanol (13.5 mL) was microwave irradiated for 20 min at 120 °C.
After it was cooled, the precipitate was filtered and thoroughly washed
with water, ethanol, and ether. Yield: 72%. Compounds were
1
characterized by IR and H NMR (see the Supporting Information).
Enzyme Purification. Recombinant type II R67 DHFR18,25 with
an N-terminal 6-histidine tag was overexpressed in E. coli BL21
containing plasmid pRep4 (Qiagen). Fresh overnight culture (3 mL)
was added to Terrific Broth (1 L; 100 μg/mL ampicillin and 50 μg/
mL kanamycin), and the flask was incubated at 37 °C and 250 rpm
until OD600 reached approximately 0.7. Protein expression was
induced by the addition of isopropyl β-D-1-thiogalactopyranoside
(IPTG) to a concentration of 1 mM, with further incubation for 3 h.
The cells were harvested by centrifugation (30 min, 2700g, 4 °C). The
cell pellet was resuspended in 30 mL of lysis buffer (0.1 M potassium
phosphate, 5 mM imidazole, pH 8.0), and the cells were disrupted by
one passage through a cell disrupter (Constant Systems) adjusted to
27 kpsi. An additional 10 mL of buffer washed residual lysate through.
Following centrifugation (30 min, 47500g, 4 °C) and filtration on a
0.22 μm filter, the supernatant was injected onto a 5 mL His-Trap HP
We seek molecules that bind within the active site in an effort
to increase the likelihood of selectivity. Thus, screening for
inhibition was carried out against a panel of approximately 100
simple compounds (or fragments) with a molecular weight of
150−250 g/mol. The fragments screened generally comprise
cyclesoften aromaticand were generally of an overall
elongated structure. Nitrogen and oxygen were preferred
heteroatoms. While broader chemical diversity was also
included, the above features were prioritized as they imitate
features of the nicotinamide ring of NADPH and of the pteroyl
ring of DHF, which are involved in binding to the center of the
active-site cavity.19 To restrict downstream efforts to molecules
showing the best potential for selective inhibitor development,
we screened the target bacterial R67 DHFR in parallel with
human DHFR, which should be spared. Here, we report seven
fragments that inhibit the R67 DHFR with an IC50 in the high
micromolar to low millimolar range. Symmetrical bis-
benzimidazole and bis-carboxyphenyl type compounds con-
̈
cartridge at a flow rate of ∼1 mL/min using an Akta FPLC (GE
Healthcare). The column was washed with ∼12 column volumes (CV)
of lysis buffer. A linear gradient (6 CV) of the same buffer + imidazole
from 0 to 30 mM with a plateau (6 CV) at 30 mM was followed by a
step to 300 mM imidazole for elution. Fractions containing R67
DHFR were identified according to activity assay and analysis on
tricine-SDS-PAGE26 and pooled for dialysis at 4 °C into to 0.1 M
phosphate buffer, pH 8.0, using 3500 Da molecular weight cutoff
dialysis tubing (Spectrum Laboratories). The protein concentration
was determined using the Bradford protein assay (Bio-Rad) using
bovine serum albumin (Bio-Rad) as a protein standard. Human
3183
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
chromosomal DHFR was overexpressed in E. coli BL21 (DE3) and
purified as described previously.27
DDR3 running either Molegro Virtual Docker 4.0.031 or Autodock
Vina 1.1.1.32 For Molegro, the Moldock Score [Grid] scoring function
was used. Grid resolution was 0.30 Å with a radius of 30. The search
algorithm used was Moldock Optimizer, with default settings. The
number of runs varied from 10 to 25, and the maximum iterations
varied from 2000 to 10000. For Autodock Vina, the default settings
were kept except the exhaustiveness parameter, which was increased
from 8 to 25. Structural visualization was performed with Moldock.
Eukaryotic Cell Proliferation Assay. The cytotoxicity of
compounds 8, 8a, 8b, and 9 was evaluated using 3T6 fibroblast cells
(kindly provided by the Laboratory of Biology and Embryology of the
University of Mons-UMONS) in a eukaryotic cell viability test
performed with the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-
tetrazolium bromide) reduction assay.33 MTT is a water-soluble
tetrazolium salt that is cleaved into an insoluble purple formazan by
the succinate dehydrogenase system of the mitochondrial chain, which
is active only in live cells. 3T6 Fibroblasts are a permanent mouse
embryonic fibroblast cell line. Cells were cultured in Dulbecco's
modified Eagle's medium (DMEM) supplemented with 10% heat-
inactivated fetal bovine serum (FBS) (Invitrogen), in a CO2 incubator
(37 °C, 5% CO2). Trypsinized 3T6 cells (200 μL, 5 × 103 cells) were
seeded in each well of a 96-well plate (except for the wells on the
edge). Eight successive dilutions of each compound (in DMSO, Sigma
Aldrich) were realized, from 200 to 0.05 μM. After 24 h of incubation,
the culture medium was replaced with 200 μL of fresh medium
containing the same compound dilutions. Culture medium (200 μL)
was also added to the cell growth control rows while the solvent
control wells received 200 μL of the relevant mixture of DMSO and
culture medium. Blanks (no cells) received the same mixture. After a
further 48 h of incubation, the cells were carefully washed with PBS,
and 100 μL of fresh culture medium were added into each well. MTT
(100 μL) (In vitro toxicology assay kit, MTT based, Sigma Aldrich)
was then added to all wells and incubated for 3 h. The reaction was
stopped by adding MTT solubilization reagent. After overnight
incubation at 37 °C, the formazan dye was quantified with a
microplate reader (Thermo Labsystems Multiskan Ascent 354) using a
test wavelength of 540 nm and a reference wavelength of 690 nm.
Eight concentrations of each test compound (0.05−200 μM) were
verified in triplicate. The % inhibition of cell proliferation was
calculated as follows: % IC = 100 − [corrected mean OD sample
×100/corrected mean OD solvent controls], where % IC = %
inhibition of cell proliferation, and corrected mean OD sample/solvent
= mean OD540−690 of samples/controls − mean OD540−690 of blanks.
For each compound, the % inhibition of activity was plotted against
the compound concentration scale. The concentration inhibiting 50%
of cell proliferation (IC50) was the x-axis value corresponding to one-
half of the maximal absorbance value.
Fragment-Based Inhibitor Screening against Purified En-
zymes. Substrates were quantified by spectrophotometry in 50 mM
potassium phosphate, pH 8.0 (ε340nm = 6200 M−1 cm−1 for NADPH
and ε282nm = 28400 M−1 cm−1 for DHF). For inhibitor screening, the
enzyme activity was determined by monitoring the depletion of
NADPH and DHF at 340 nm (Δε340nm = 12800 M−1 cm−1) in 400
mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)
buffer, pH 7.0, for R67 DHFR and 50 mM potassium phosphate,
pH 8.0, for hDHFR. The high concentration of HEPES was required
to offer adequate buffering in the presence of high concentrations of
acidic or basic compounds, because the activity of R67 DHFR is
exquisitely sensitive to pH changes.28
To provide adequate compound dissolution in a solvent system
where both DHFRs retained appreciable activity, the maximal
solubility of each test compound was determined in HEPES-buffered
10% DMSO. In that system, R67 DHFR retained 31% activity and
hDHFR retained 17% activity, allowing reliable assessment of
inhibition. For inhibition assays, each test compound was dissolved
in neat DMSO and taken up in the reaction buffer containing the
substrates, to a final DMSO concentration of 10%. For assay
automation, testing was performed in a 96-well plate format with a
reaction volume of 200 μL. When test compounds contributed to an
excessive background absorbance at 340 nm, the activity was
monitored in a volume of 100 μL to reduce the path length. Liquid
handling was carried out using a BioMek NX automated workstation
(Beckman Coulter), and data were collected with a Beckman DTX
880 plate-reader. Unless otherwise noted, final concentrations of
NADPH and DHF were 50 μM. As a result, NADPH was at 30 ×
18
NADPH
KM
.
The inclusion of 10% DMSO weakens DHF binding,
29
DHF
increasing the apparent KM
approximately 3-fold, which is taken
into account in all calculations. Here, DHF was at ∼2 × KMDHF. These
conditions provide a clear spectrophotometric signal in the case of
uninhibited activity, without masking potential competitive inhibition.
Reactions were initiated with the addition of 250 nM R67 DHFR.
Initial rates (generally less than 10% substrate conversion) were
determined. The determination of selectivity using hDHFR was
performed under the same conditions except the buffer (see above).
The concentration of hDHFR was 40 nM to allow measurement of
initial rates while compensating for the higher turnover constant of
hDHFR relative to R67 DHFR.
Determination of Binding and Kinetic Parameters. The IC50
values were determined with GraphPad Prism 5 using the
log[inhibitor] vs response equation; Ki values were determined using
the Cheng−Prusoff equation: Ki = IC50/([S]/KMNADPH + 1), assuming
a competitive mode of inhibition.30 KM
used for the calculation
NADPH
was 1.6 0.02 μM.18 For determination of the mode of inhibition, the
method of Dixon was used, in 50 mM potassium phosphate buffer, pH
7.0, with 10% DMSO. For inhibition relative to NADPH, DHF was
held at 50 μM, and NADPH was at 5 or 80 μM (3- and 50-fold
KMNADPH, respectively) and inhibitor over a range of concentrations
spanning Ki. For inhibition relative to DHF, NADPH was held at 50
RESULTS AND DISCUSSION
■
Fragment-Based Inhibitor Screening of R67 DHFR. To
date, two nonspecific micromolar inhibitors of R67 DHFR have
been reported: novobiocin (Ki = 60 μM), a DNA gyrase
inhibitor, and congo red, a diazo dye (Ki = 2 μM).17 Their
structural differences relative to the other known micromolar
ligands of R67 DHFR, namely, the DHF substrate and NADPH
cofactor, as well as the loose substrate specificity of the
enzyme,15 immediately suggest that the active site is amenable
to fragment-based inhibitor design. Accordingly, we selected a
panel of small organic compounds (fragments) for inhibitor
screening, based on structural similarity to the moieties of the
substrate and cofactor, which bind deepest in the active-site
pore.19 Approximately 100 commercially available compounds
that share some chemical features of the DHF substrate pteroyl
group or NADPH cofactor nicotinamide ring were screened for
selective inhibition of R67 DHFR relative to hDHFR (Table S1
in the Supporting Information). Because the compounds are
simple and possess a low molecular weight (150−250 g/mol),
DHF
DHF
μM, and DHF was at 25 or 164 μM (∼KM
and ∼7-fold KM
)
with inhibitor again spanning Ki. Initial rates were determined in 1 cm
path-length cuvettes using a Cary 100 Bio UV−visible spectropho-
tometer (Agilent). For determination of binding stoichiometry,
NADPH
NADPH was held at 16 μM and DHF at 80 μM (10-fold KM
and ∼3-fold KMDHF, respectively) and titrated with increasing
concentrations of 9. Binding data were analyzed according to the
Hill equation for inhibition: log[vi/(v0 − vi)] = −n log[Ii] + log Kd,
where vi is the velocity at inhibitor concentration [Ii], v0 is the
noninhibited velocity, n is the Hill coefficient giving an indication of
the number of binding sites, and Kd is the dissociation constant.
Molecular Docking Simulations. The crystal structure of a
ternary complex of R67 DHFR (PDB file 2RK119) was used for all
docking simulations. The ligand 3D structures were drawn using
ChemDraw 3D Pro 8.0. The geometry of the ligands was optimized
using HyperChem 8.0.3 with the default settings. Docking simulations
were run on a Dell XPS 1645 with an i7-720QM processor and 6 GB
3184
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
Figure 1. Structures of the weak inhibitor fragments and their inactive analogues. Structures of the DHF substrate and NADPH cofactor are
provided as a reference. Compounds: 1, 6-aminoisonaphthoic acid; 1a, 3-aminoisonaphthoic acid; 2, 3,5-dimethoxyisonaphthoic acid; 3, thieno[2,3-
b]pyrazine-6-carboxylic acid; 4, 1H-benzimidazole-5-carboxylic acid; 4a, benzimidazole; 4b, piperonylic acid; 4c, 2-methyl-1H-benzimidazole-5-
carboxylic acid; 5, 6-hydroxy-2-naphthoic acid; 5a, 6-methoxy-2-naphthoic acid; 6, 1H-benzotriazole-5-carboxylic; 6a, 1-methylethyl-1H-
benzotriazole-5-carboxylic acid; and 7, 1(2H)-phthalazinone. No inhibition was observed up to 5 (1a and 5a), 10 (4c and 6a), or 30 mM (4a
and 4b), which are the highest concentrations that could be tested due to constraints of poor solubility and high absorbance.
no strong inhibition was expected. Hence, they were tested for
inhibition at mM concentrations. Figure 1 illustrates com-
pounds 1−7, which provided inhibition with an IC50 in the low-
to midmillimolar range (Table 1), as well as some structural
analogues, which provided no inhibition.
amino groups of 1a have the potential for intramolecular H-
bonding, which would modulate the electronic effect of these
functional groups relative to 1. Compound 5 is hydroxylated at
position 6 and shows inhibition similar to 1 and 2 (IC50 = 6.1
mM), while a 6-methoxy substituent abrogated inhibition (5a).
These results suggest the positive contribution of a hydrophilic
substituent at C5 or C6 of the naphthoic acid, opposite the
carboxylic acid, although different substituents modulate the
inhibitory activity of the naphthoic acid framework.
Compound 3, a thienopyrazine framework substituted with a
carboxylic acid, inhibited R67 DHFR with an IC50 of 2.0 mM,
in the same range as the similarly substituted benzimidazole 4
(IC50 = 2.6 mM) and the carboxylate-substituted benzotriazole
6 (IC50 = 8.4 mM). Benzimidazole (4a) gave no inhibition,
confirming the contribution of the carboxylic acid substitution
to inhibition. Changing the aromatic diimino ring to a dioxole
ring (compound 4b) prevented inhibition, suggesting the
importance of aromaticity and/or polarity of heteroatoms. It is
not clear why an additional methyl substituent at C6 of the
acid-substituted benzimidazole (4c) abrogated inhibition, since
it will be shown below that C6 was used to build fragment 4
into a more complex inhibitor (vide infra). An additional
isopropyl at N5 of the acid-substituted benzotriazole (6a)
prevented inhibition, indicating that additional bulk or
hydrophobicity at positions opposite the carboxylic acid
substituent are to be avoided. The only inhibitor identified
that carried no carboxylic acid substituent was 7, the weakest
inhibitor with IC50 = 17 mM.
Table 1. Inhibition of R67 DHFR and hDHFR by Fragments
a
1−7
relative activity hDHFR
LE
b
compd IC50 (mM)
Ki (μM)
51
(%)
(kcal)
1
2
3
4
5
6
7
1.7 0.3
1.8 0.4
2.0 1.3
2.6 1.4
6.1 2.3
8.4 2.6
17 1.7
9
14 13
−0.32
−0.32
−0.48
−0.47
−0.36
−0.41
−0.32
55 14
61 41
80 42
190 71
260 80
530 54
14
37
75
13
4
9
5
2
69 13
97
3
a
Values are given as the average standard deviation from the mean.
Compounds were tested at the highest possible concentration (1 =
b
2.5 mM, 2 and 3 = 5 mM, 4 and 5 = 10 mM, and 6 and 7 = 30 mM),
taking into account constraints due to poor solubility and high
absorbance.
Compounds 1, 2, and 5 possess a naphthalene core
substituted with a carboxylic acid but are differently
functionalized. Despite the differences between the 6-amino
group of 1 and the 2,5-dimethoxy substitution of 2, their IC50
values are unchanged (1.7 and 1.8 mM for compounds 1 and 2,
respectively). However, with the amino group at position 2
(1a), no inhibition was observed. The vicinal carboxylate and
By way of fragment screening, we thus identified several weak
inhibitors of R67 DHFR. All are hydrophobic and aromatic and
possess a carboxylic acid (except the weaker inhibitor 7). The
3185
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
positions opposite the carboxylic acid modulate the activity of
the compound and could be targeted to improve these
inhibitors.
mostly hydrophobic active site,25 the center of which holds the
reactive groups of the DHF substrate and NADPH cofactor.
Upon binding of DHF and NADPH, Lys32 from the four
DHFR protomerswhich lie at the mouth of the active-site
channelbind to the negatively charged DHF carboxylates and
NADPH phosphates.34 We hypothesized that the sulfites of
congo red and carboxylic acid substituents of fragments 1−6
interact with these lysines or with the neighboring Lys33 (from
each of the four protomers), which lie further outside of the
active-site channel. Thus, we tested inhibition with symmetrical
compounds constituted of two fragments 4 symmetrically
connected with a semirigid hydrophobic linker (yielding 8) or a
shorter, slightly more hydrophobic linker (yielding 9) (Figure
2).
To direct inhibitor development toward selectivity for R67
DHFR while sparing human DHFR (hDHFR), we tested these
seven fragments against hDHFR. Compounds 4 and 6
displayed good selectivity: at 3 × IC50, approximately 80% of
hDHFR activity remained. Compound 7 did not affect the
activity of the hDHFR up to 30 mM (2 × IC50). However,
compounds 1, 2, 3, and 5 inhibited nonselectively: at <2 × IC50,
these compounds strongly inhibited hDHFR. Although 4 and 6
were not the most potent, they showed a good selectivity for
R67 DHFR, and they have a good LE, where LE is ΔG of
binding (calculated from Ki) divided by the number of heavy
atoms in the molecule (Table 1). LE indicates how efficiently a
compound binds to its target, relative to its molecular weight.
In a recent example, careful optimization of millimolar
fragments with a LE of −0.30 kcal and less provided nanomolar
inhibitors.21 Fragment 4 possesses a good potency, selectivity,
and a LE of −0.47 kcal; it was thus selected as a basis for
development of a more potent inhibitor.
Inhibition of R67 DHFR by Symmetrical Polyaromatic
Molecules. As mentioned above, novobiocin (Ki = 70 μM)
and congo red (Ki = 2 μM) (Figure 2) are the only two known
R67 DHFR inhibitors, and neither is specific to R67 DHFR.
Both structures are elongated and possess substituted aromatic
rings. Notably, the more potent congo red is symmetrical, with
a negatively charged substituent on each extremity and a central
hydrophobic linker. R67 DHFR possesses a symmetrical and
We observed an improvement in potency, with IC50 of 64
and 130 μM for compounds 8 and 9, respectively (as compared
to 2.6 mM for 4) (Table 2 and Figure S1 in the Supporting
Table 2. Inhibition of R67 DHFR and hDHFR by
a
Symmetrical Compounds 8 and 9
relative activity hDHFR
LE
(kcal)
b
c
compd IC50 (μM) Ki (μM)
(%)
8
64 11
≥1000
≥300
2.0 0.3
ND
95 11
−0.18
ND
8a
8b
9
100
1
ND
110 14
95 10
ND
130 11
4.0 0.3
−0.18
a
Values are given as the average standard deviation from the mean.
b
ND, not determined. Ki was calculated from IC50 values.
c
Compounds were tested at the highest possible concentration (8 =
400 μM, 8a = 1 mM, 8b = 300 μM, and 9 = 1 mM), taking into
account constraints due to poor solubility and high absorbance.
Information). To gain insight into the structure−activity
relationship (SAR), we tested two analogues of 8 (8a and
8b). Removing the benzimidazole moiety from 8 (yielding 8a)
provided no inhibition up to the highest concentration tested
(1 mM). Removing the carboxylic acid from the benzimidazole
moiety (yielding 8b) reduced the potency: at 300 μM, more
than 60% of R67 activity remained, demonstrating that the
carboxylic acid functions are important contributors to binding.
Nonetheless, lack of inhibition with 8a demonstrated that
carboxylic acids alone are not sufficient for binding to R67
DHFR. To verify if the selectivity of the inhibitors was
conserved, we tested inhibitors 8 and 9 against hDHFR (Table
2). No inhibition of hDHFR was observed up to the highest
concentrations tested (400 μM 8, 1 mM 8a, 300 μM 8b, and 1
mM 9).
Overall, starting from fragment 4, which gave a millimolar
range IC50 and a LE of −0.47 kcal (Table 1), we identified two
selective inhibitors of R67 DHFR (8 and 9), which showed
IC50 300−600-fold lower, in the micromolar range. Com-
pounds 8 and 9 possess a LE of −0.18 kcal (Table 2) and thus
bind less efficiently relative to their molecular weight than the
precursor fragment 4. This illustrates that they are poorly
optimized compounds. Thus, inhibitors 8 and 9 provide a
starting point to develop potent and selective inhibitors against
R67 DHFR.
Inhibition Is Competitive. To understand the binding
mode of the symmetrical inhibitors to their target, the mode of
inhibition of 8 toward R67 DHFR was determined according to
the method of Dixon, by performing inhibition at varying
NADPH and inhibitor concentrations.35 Figure 3 illustrates that
Figure 2. Structures of the R67 DHFR inhibitors and their inactive
analogues. Congo red and novobiocin are nonspecific inhibitors of
R67 DHFR. Compounds: 8, 2,2′-[1,5-pentanediylbis(4-oxypheny-
lene)]-bis-1H-benzimidazole-5-carboxylic acid; 8a, 4,4′-[1,5-pentane-
diyl-(oxy)]-bisbenzoic acid; 8b, 2,2′-[1,5-pentanediylbis(4-oxypheny-
lene)]-bis-1H-benzimidazole; and 9, 2′-(4,4′-(2-hydroxypropane-1,3-
diyl)-bis-(4-oxyphenylene)-bis-1H-benzimidazole-5-carboxylic acid.
3186
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
apoenzyme using the ligand NADPH. In the five poses
corresponding to the best docking scores, the nicotinamide
moiety and the three phosphates were roughly well localized,
although the adenine ring made contacts with different faces of
the channel. Thus, the docking of NADPH to the apoenzyme
could not be directly compared to the bound conformer in the
crystallized ternary complex but served as a guide for relative
docking scores of the inhibitors.
Compounds 8, 8a, 8b, and 9 were docked onto R67 DHFR
using Molegro Virtual Docker. The docking scores are reported
NADPH
in Table 3, with NADPH as the reference compound. KM
is 1.6 μM18 (for the ternary complex). Its docking score (for
the binary complex) was −190, where the score is a
dimensionless unit estimating the binding energy. The scores
for inhibitors 8 and 9 were similar, at −190 and −185,
respectively, consistent with their Ki (for 8: calculated from
Figure 3. Dixon plot for the determination of type of inhibition of R67
DHFR by 8, relative to the NADPH cofactor. The reciprocal rates of
DHFR activity were plotted as a function of inhibitor concentration.
DHF substrate was held at 50 μM. NADPH cofactor was held at 5 (3
NADPH
NADPH
■
▲
× KM
;
) or 80 μM (50 × KM
;
). Values are given as the
IC50 = 2.0 μM, experimentally determined = 4.2 μM; for 9:
mean standard deviation for triplicate results. The intercept of the
NADPH
calculated from IC50 = 4.0 μM), which are similar to KM
.
two slopes gave Ki (8) = 4.2 μM.
Compounds 8a and 8b provided weaker scores of −109 and
−156, respectively, reflecting the poor inhibition properties that
were apparent in their respective IC50 values.
8 is a competitive inhibitor, with Ki = 4.2 μM against the
cofactor NADPH. This value is consistent with that calculated
from IC50 (Table 2). Compound 8 showed similar results
toward the substrate DHF, with a Ki of 9.7 μM (Figure S2 in
the Supporting Information). The competitive nature of the
inhibition with respect to both the substrate and the cofactor is
consistent with the inhibitor binding within the active-site
cavity.
Typical poses obtained for these compounds are shown in
Figure 4D−K. Interestingly, each of the four R67-bound
inhibitors was predicted to adopt a U shape, with stacking of
the symmetry-related aromatic moieties. The aromatic stacking
would fill the volume of the cavity in a manner similar to the
juxtaposition of the NADPH and DHF during the native
reaction. This proposed binding mode is consistent with the
symmetry of the tetrameric active site, where each arm of the
inhibitor forms similar contacts with symmetry-related residues
of the enzyme (i.e., protomera and protomerd). Furthermore, it
is consistent with the hydrophobic nature of the center of the
active site, which would disfavor “threading” of the charged
extremity of any of these compounds through the active-site
channel. Overall, it is consistent with a competitive mode of
inhibition. Specific features of the predicted mode of binding
are discussed below.
Compound 9 has the most hydrophilic central spacer;
interestingly, the docked model predicts that 9 is deeply bound
within the active site, where the glycerol spacer appears to
strongly interact with the backbone of Ile68c (Figure 4J,K).
Ile68 is proposed to provide recognition of the carboxamide on
the nicotinamide ring and the amide on the pteridine ring.19 As
a result, the benzene and benzimidazole would form van der
Waals interactions with the central hydrophobic pocket. The
central pocket would be fully occupied, as shown in Figure 4K,
where the docked molecule is predicted to traverse the center
of symmetry of the enzyme. Lys32a has the potential to form an
electrostatic bond with a benzimidazole carboxylate (Figure 4J),
consistent with its role in binding the NADPH phosphate or
the DHF glutamate tail to orient these molecules in the active
site. Moreover, in the model, Gln67a H-bonds with the
benzimidazole moiety, and Ser65b H-bonds with the terminal
carboxylate.
Docked Model of Inhibitor Binding to R67 DHFR. To
help direct future efforts in improving the inhibitors,
compounds 8, 8a, 8b, and 9 were docked into the crystal
structure of R67 DHFR (2RK119). Each of the four protomers
of the R67 DHFR homotetramer plays a different role when a
molecule binds in the active site, thus breaking the tetramer's
symmetry.36 For clarity, we will refer to residues according to
their chain identity (protomera to protomerd), as required.
Figure 4A−C shows that R67 DHFR forms a symmetrical
homotetramer where the active site lies within the central
channel. The entrances to the pore are lined with the positively
charged Lys32 and Lys33, each contributed by symmetry-
related protomers (i.e., protomera and protomerd or protomerb
and protomerc) to provide four lysines at each entrance to the
pore (Figure 4C). They serve to bind and orient the
phosphates of the cofactor NADPH and the carboxylates of
the substrate DHF.34 The NADPH nicotinamide ring and the
DHF pteroyl group, which contain the hydride donor and
acceptor, respectively, meet at the mostly hydrophobic center
of the pore for reactivity, where Gln67, Ile68, and Tyr69 are key
binding residues. In particular, Gln67 forms a H-bond network
with Tyr69, providing a molecular “clamp” that holds the
substrates in place.19 The unusual symmetry of the active-site
pore and its binding promiscuity,15 in addition to the
cooperativity of NADPH and DHF binding,37 increase the
difficulty of obtaining meaningful docking results. Docking to
the apo-form of the crystallized ternary complex was shown to
give results that were inconsistent with that complex,38
although docking of one ligand in the presence of the second
has proven more reliable.38,39 Nonetheless, we assume that
inhibitor binding follows the simplest model of binding to the
apoenzymethis is consistent with inhibition being compet-
itive against both DHF and NADPH, although it does not
necessarily demonstrate it. DHF binding was not deemed a
meaningful control because only its pterin ring has been
resolved. We thus conducted a docking control on the
The docked model of compound 8 (Figure 4D,E) was bound
less deeply within the hydrophobic pocket, although it was also
predicted to traverse the lengthwise center of symmetry of the
channel. Interactions with the backbone of Ile68 were lost due
to the more hydrophobic nature of the spacer, but interactions
with Gln67b and Gln67c were gained. This position allows the
terminal carboxylates to H-bond with residues Thr51a, Tyr46a,
and Ala36a, in addition to allowing electrostatic interaction with
Lys32d. Compound 8b (Figure 4H,I) differs from 8 by its lack
3187
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
Figure 4. Docking of 8, 8a, 8b, and 9 R67 DHFR using Molegro Virtual Docker. (A) R67 DHFR is shown as an electrostatics surface with the basic,
acidic, and hydrophobic regions shown in blue, red, and white, respectively, and protomers labeled a−d. (B) Two protomers are shown, where the
tetramer was split as indicated in panel A. (C) A 90° counter-clockwise rotation of panel B, revealing the largely hydrophobic active-site cavity
(circled). (D−K) Docking of 8, 8a, 8b, and 9. (D and E) Docking of 8, (F and G) 8a, (H and I) 8b, and (J and K) 9. The left-hand panels show
putative contacts formed, with the compound in ball-and-stick representation and the residues of R67 DHFR in sticks representation. Hydrogen,
carbon, nitrogen, and oxygen are in white, gray, blue, and red balls, respectively. Red and green bonds represent the nonrotatable and rotatable
bonds, respectively, as defined for docking experiments. Yellow and green dashed lines represent electrostatic bonds and H-bonds, respectively. The
right-hand panels show the same result, with the enzyme active site oriented as in panel C. The vertical blue line indicates the center of symmetry for
the axis that runs lengthwise through the binding channel.
of terminal carboxylates on the benzimidazole moieties.
Compound 8b was also deeply docked in the active site. The
interactions with Thr51, Tyr46, Ala36, and Lys32 predicted
with 9 were lost with 8b due to the lack of the carboxylates.
This result is consistent with the IC50 for 8b, which is at least 5-
fold higher than 8. Compound 8a (Figure 4F,G) differs from 8
and 9 by its lack of benzimidazole moieties. In the docked
model, the terminal carboxylates formed the same H-bonding
and electrostatic interactions observed with 8, which suggests
that the benzimidazole moiety is not essential to the formation
of these interactions. Putative interactions with Gln67 and Ile68
were lost as a consequence of the loss of the benzimidazole
heteroatoms and the replacement of the glycerol moiety by
pentane, which could explain the lower docking score and the
lack of in vitro inhibition. Compound 8a was also the least
deeply docked into the active site.
Docking with Autodock Vina afforded essentially super-
imposable results (Figure S3 in the Supporting Information):
3188
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
Binding Stoichiometry. As mentioned above, if the
inhibitor molecules 8 or 9 were to bind to the active site in a
linear conformation, it is conceptually possible for two
molecules to bind simultaneously. In contrast, the model of
the U-shaped conformation of bound 8 or 9 would not allow
this (Figure 4). We examined binding stoichiometry by
performing competitive inhibition under conditions appropriate
for determining a Hill coefficient. As shown in Figure 5 (inset),
Table 3. Docking of NADPH and 8, 8a, 8b, and 9 to R67
DHFR
compd
Moldock score
NADPH
−190
−190
−109
−156
−185
8
8a
8b
9
the inhibitors folded into a U shape, slipping the center of the
molecule deep into the active-site channel with the carboxylates
generally held by Lys32; Tyr46, Thr51, and Ser65 also
participated in carboxylate binding (not shown). Similar
interactions were predicted with Gln67, Ile68, and Tyr69 as
when docking with Molegro.
While these models suggest that intramolecular stacking
occurs upon binding, this may result from limitations of the
docking algorithms used. Specifically, docking algorithms dock
a single molecule of test compound at a time. Thus, the U
shape that we observed may reflect a virtual solution to
maximize molecular packing within the active-site tunnel. An
alternative docking method of compounds 8, 8a, and 9 has
shown a linear mode of binding (Hogue, H., Purisima, E., and
Sulea, T. Personal communication). In that case, the molecule
threads through the length of the active-site pore, forming
essentially the same contacts as in the U shape, except that the
two R67 DHFR protomers that it interacts with are on opposite
ends of the channel (thus interacting with protomera and
protomerc rather than protomera and protomerd). Indeed, it is
possible that two molecules of a given compound bind
simultaneously, laying flat on top of each other, thus
maximizing the contacts with the large active-site tunnel.
That mode of binding remains consistent with the hydro-
phobicity of the tunnel, with the putative role of the 4-fold
Lys32 residues in binding the inhibitors' carboxylates, with the
volume occupied by the stacking of the natural DHF substrate
and NADPH cofactor and provides a more entropically favored
solution to binding; threading of the terminal carboxylate
through the hydrophobic active site may be promoted by the
water molecules known to populate the active-site pore.19
Attempts were made to soak inhibitor 9 into R67 DHFR
crystals that were obtained using a recently described
crystallization protocol.25 A change in the electron density
was observed in the active-site pore in the presence of 9 relative
to the apo structure of R67 DHFR (Yachnin, B., Colin, D.,
Berghuis, A. Personal communication). Unfortunately, because
the pore lies on the crystallographic symmetry axes, combined
with the fact that the ligand did not appear to be present at full
occupancy, it proved impossible to interpret the electron
density. From what could be seen, the density was not
consistent with the U-shaped conformation of compound 9 as
suggested by the dockings results, but the poor quality of the
electron density precludes drawing conclusions on the actual
bound conformation of the inhibitor.
Figure 5. Hill plot for the determination of the stoichiometry of
binding of compound 9 to R67 DHFR. Semilogarithmic representa-
tion of the relative activity vs the inhibitor concentration. DHF
substrate was held at 80 μM (∼3 × KMDHF) and NADPH cofactor at
16 μM (10 × KMNADPH). The calculated Hill coefficient was 2.1 0.3.
The inset represents the corresponding double logarithmic Hill plot.
The best fit to the data does not follow a single straight line but is a
composite of two straight lines with different slopes. The apparent Hill
coefficients were 0.39 0.06 for low inhibitor concentrations and 2.2
0.4 for high inhibitor concentrations. Values are given as the mean
standard deviation for triplicate results.
the Hill plot with 9 shows a clear inflection, consistent with
more than one inhibitor molecule binding to the target.40 The
apparent Hill coefficients obtained are 0.39
inhibitor concentrations and 2.2 0.4 at high inhibitor
concentrations, while the Hill coefficient obtained upon
plotting in semilogarithmic format is 2.1 0.3 (Figure 5).
0.06 at low
These values are consistent with R67 having the potential to
bind more than one inhibitor molecule simultaneously.40
However, there was no evidence of binding cooperativity
according to a replot of reaction rate relative to inhibitor
concentration (not shown).40
These results support binding of the inhibitors in a linear
mode, where it is sterically possible to stack two molecules into
the active site, rather than the U-shaped docked models (Figure
4), where inhibitors 8 and 9 both cross the center of symmetry
of the active site such that it would not be possible to fit in two
molecules simultaneously. It should be noted that a further
sterically plausible binding mode would be for one molecule of
inhibitor to bind in a linear mode concurrently with one
molecule of NADPH or DHF. We attempted to verify this
according to fluorescence binding measurements, but com-
pounds 8 and 9 both gave rise to strong fluorescence plagued
by intense quenching at the relevant micromolar concen-
trations.
In Vivo Assays. Determination of the minimum inhibitory
concentration (MIC) for compounds 8 and 9 was attempted
using E. coli XL1-Blue expressing DHFR R67 in minimum
media (M9) containing 50 μg/mL of TMP (to inhibit the
native, chromosomal E. coli DHFR). Bacterial growth was not
inhibited up to 1 mM 8 or 9 (maximal concentration tested;
data not shown), suggesting that these compounds do not
penetrate into E. coli. We considered testing esterified versions
Notwithstanding the precise mode of binding, symmetry
appears to play a key role in binding and selectivity for R67
DHFR. The proposed interaction of terminal carboxylates with
Lys32, Thr51, Tyr46, or Ala36 may be modulated by the
distance between the carboxylate and the hydrophobic pocket
as observed by comparing docked models of 8 and 9.
Interaction with the backbone of Ile68 and Val66 may be
promoted by a heteroatom near the hydrophobic pocket.
3189
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
of the inhibitors to increase their permeability in microbial cells,
but their solubility became too low to allow testing at
meaningful concentrations. Nonetheless, 8b has an antiprolifer-
ative effect on Leishmania donovani, Pneumocystis carinii,
Plasmodium falciparum, and Trypanosoma brucei rhodesiense,
although their mode of action has not yet been determined.23
Compounds 8, 8a, 8b, and 9 exhibited only weak cytotoxicity
against adherent mouse fibroblast 3T6 cells with the exception
of compound 8b, which exhibited an IC50 of approximately 1
μM (Table 4).
ASSOCIATED CONTENT
■
S
* Supporting Information
Full list of fragments screened against R67 DHFR, character-
ization of compounds 8, 8a, and 9, and additional kinetic and
molecular docking results. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*Tel: 514-343-2124. Fax: 514-343-7586. E-mail: joelle.
pelletier@umontreal.ca.
Table 4. Cytotoxicity of 8, 8a, 8b, and 9 toward 3T6
Fibroblast Cells
Notes
The authors declare no competing financial interest.
compd
IC50 (μM)
ACKNOWLEDGMENTS
■
8
40 6% mortality at 200 μM
34.0 0.4
We thank the Combinatorial Chemistry Laboratory at
8a
8b
9
l'Universite
́
de Montrea
́
l and the Medicinal Chemistry Platform
0.98 0.07
at the Institut de Recherche en Immunologie et en Cancer
́
-
40 5% mortality at 200 μM
ologie (IRIC) de l'Universite de Montreal for access to
compounds. This work was supported by the Natural Sciences
and Engineering Research Council of Canada (NSERC), as
́
́
CONCLUSIONS
■
well as PROTEO, the Queb
Protein Structure, Function and Engineering, and CGCC, the
Quebec Centre for Green Chemistry and Catalysis, which are
funded by les Fonds Quebecois pour la Recherche sur la Nature
́
ec Network for Research on
We have developed a functional screening platform for the
identification of selective inhibitors of the bacterial antibiotic-
resistance enzyme, R67 DHFR. The platform is based on
semiautomated determination of enzymatic activity in the
presence of a variety of fragments, or simple compounds similar
to the native ligands of the enzyme, to assess inhibition. We
observed that a variety of small aromatic compounds offer
millimolar range inhibition of R67 DHFR, which is consistent
with the proposed “primitive” nature of its relatively
promiscuous binding site.15,39 Weakly inhibiting molecules
provided the basis for testing compounds of greater complexity,
which provided an increase in affinity from millimolar to
micromolar. Importantly, we screened the human DHFR in
parallel with R67 DHFR to identify structures with the best
prospects for selective inhibition. Thus, we identified a new
class of selective, symmetrical, and competitive inhibitors of
R67 DHFR (8 and 9), which are as potent as congo red but are
selective toward R67 DHFR. Compounds 8 and 9 were both
based on 1H-benzimidazole-5-carboxylic acid (4), which had an
IC50 of 3 mM (Table 1). Joining two benzimidazolecarboxylic
acid units into symmetrical structures with a central hydro-
phobic and flexible linker increased affinity to the low
micromolar range. Testing of the analogues 8a and 8b
demonstrated that carboxylic acid groups and benzimidazoles
contribute to affinity. Preliminary cytotoxicity and selectivity
experiments demonstrated that 8 and 9 are not highly toxic and
thus can serve as lead compounds to develop more potent
selective inhibitors of R67 DHFR. We are attempting to gain
structural information to determine the mode of binding of the
bis-benzimidazole compounds to R67 DHFR, to direct further
efforts in discovery and development of selective inhibitors for
this emerging drug-resistance target. Future inhibitors should
take advantage of the important structural differences between
R67 DHFR and human DHFR. In particular, the large size and
symmetry of the R67 DHFR active site appears to allow
simultaneous binding of two inhibitor molecules in a manner
that is not coherent with the narrower, asymmetric active-site
cleft of human DHFR.41 This work provides inspiration for the
design of the next generation of inhibitors.
́
́
́
et les Technologies (FQRNT).
ABBREVIATIONS USED
■
DHFR, dihydrofolate reductase; DHF, dihydrofolate; NADPH,
nicotinamide adenine dinucleotide phosphate, reduced form;
IC50, concentration giving 50% inhibition; Ki, inhibition
constant; CV, column volume; HEPES, 4-(2-hydroxyethyl)-1-
piperazineethanesulfonic acid; TMP, trimethoprim; MTT, 3-
[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide;
DMEM, Dulbecco's modified Eagle's medium; FBS, fetal
bovine serum; LE, ligand efficiency
REFERENCES
■
(1) Huovinen, P.; Sundstrom, L.; Swedberg, G.; Skold, O.
Trimethoprim and sulfonamide resistance. Antimicrob. Agents Chemo-
ther. 1995, 39 (2), 279−289.
̈
(2) (a) Huovinen, P. Resistance to trimethoprim-sulfamethoxazole.
Clin. Infect. Dis. 2001, 32 (11), 1608−1614. (b) Forgacs, P.;
Wengenack, N. L.; Hall, L.; Zimmerman, S. K.; Silverman, M. L.;
Roberts, G. D. Tuberculosis and trimethoprim-sulfamethoxazole.
Antimicrob. Agents Chemother. 2009, 53 (11), 4789−4793.
(3) http://www.who.int/selection_medicines/committees/expert/
18/applications/Sulfamethoxazole_trimethoprim.pdf (accessed No-
vember 29, 2011).
(4) Kehrenberg, C.; Schwarz, S. Trimethoprim resistance in a porcine
Pasteurella aerogenes isolate is based on a dfrA1 gene cassette located
in a partially truncated class 2 integron. J. Antimicrob. Chemother. 2011,
66 (2), 450−452.
(5) Cabello, F. C. Heavy use of prophylactic antibiotics in
aquaculture: A growing problem for human and animal health and
for the environment. Environ. Microbiol. 2006, 8 (7), 1137−1144.
(6) Le, T. X.; Munekage, Y.; Kato, S. Antibiotic resistance in bacteria
from shrimp farming in mangrove areas. Sci. Total Environ. 2005, 349
(1−3), 95−105.
(7) Segura, P. A.; Garcia-Ac, A.; Lajeunesse, A.; Ghosh, D.; Gagnon,
C.; Sauve, S. Determination of six anti-infectives in wastewater using
́
tandem solid-phase extraction and liquid chromatography-tandem
mass spectrometry. J. Environ. Monit. 2007, 9 (4), 307−313.
3190
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
(8) (a) Burridge, L.; Weis, J. S.; Cabello, F.; Pizarro, J.; Bostick, K.
Chemical use in salmon aquaculture: A review of current practices and
possible environmental effects. Aquaculture 2010, 306 (1−4), 7−23.
(b) Medina, A.; Horcajo, P.; Jurado, S.; De La Fuente, R.; Ruiz-Santa-
Quiteria, J. A.; Domínguez-Bernal, G.; Orden, J. A. Phenotypic and
Genotypic Characterization of Antimicrobial Resistance in Enter-
ohemorrhagic Escherichia coli and Atypical Enteropathogenic E. coli
Strains from Ruminants. J. Vet. Diagn. Invest. 2011, 23 (1), 91−95.
(c) Barlow, R. S.; Pemberton, J. M.; Desmarchelier, P. M.; Gobius, K.
S. Isolation and characterization of integron-containing bacteria
without antibiotic selection. Antimicrob. Agents Chemother. 2004, 48
(3), 838−842. (d) Zhao, H. X.; Shen, J. Z.; An, X. P.; Fan, H. L.; Cao,
J. S.; Li, P. F. Characterization of integrons in multiple antimicrobial
resistant Escherichia coli isolates from bovine endometritis. Res. Vet. Sci.
2011, 91 (3), 412−414.
(9) Volpato, J. P.; Pelletier, J. N. Mutational 'hot-spots' in
mammalian, bacterial and protozoal dihydrofolate reductases asso-
ciated with antifolate resistance: Sequence and structural comparison.
Drug Resist. Updates 2009, 12 (1−2), 28−41.
(10) (a) Plowe, C. V.; Kublin, J. G.; Dzinjalamala, F. K.; Kamwendo,
D. S.; Mukadam, R. A.; Chimpeni, P.; Molyneux, M. E.; Taylor, T. E.
Sustained clinical efficacy of sulfadoxine-pyrimethamine for uncom-
plicated falciparum malaria in Malawi after 10 years as first line
treatment: five year prospective study. BMJ 2004, 328 (7439), 545.
(b) Musset, L.; Pradines, B.; Parzy, D.; Durand, R.; Bigot, P.; Le Bras,
J. Apparent absence of atovaquone/proguanil resistance in 477
Plasmodium falciparum isolates from untreated French travellers. J.
Antimicrob. Chemother. 2006, 57 (1), 110−115.
(22) Blakley, R. L. Crystalline Dihydropteroylglutamic Acid. Nature
1960, 188, 231−232.
(23) Mayence, A.; Pietka, A.; Collins, M. S.; Cushion, M. T.;
Tekwani, B. L.; Huang, T. L.; Eynde, J. J. V. Novel bisbenzimidazoles
with antileishmanial effectiveness. Bioorg. Med. Chem. Lett. 2008, 18
(8), 2658−2661.
(24) Donahoe, H. B.; Benjamin, L. E.; Fennoy, L. V.; Greiff, D.
Synthesis of Potential Rickettsiostatic Agents.1a I. 4,4′-Dicarboxy-α,ι-
diphenoxyalkanes1b. J. Org. Chem. 1961, 26 (2), 474−476.
(25) Yachnin, B. J.; Colin, D. Y.; Volpato, J. P.; Ebert, M.; Pelletier, J.
N.; Berghuis, A. M. Novel crystallization conditions for tandem variant
R67 DHFR yield a wild-type crystal structure. Acta Crystallogr., Sect. F:
Struct. Biol. Cryst. Commun. 2011, 67, 1316−1322.
(26) Schagger, H.; von Jagow, G. Tricine-sodium dodecyl sulfate-
̈
polyacrylamide gel electrophoresis for the separation of proteins in the
range from 1 to 100 kDa. Anal. Biochem. 1987, 166 (2), 368−379.
(27) Volpato, J. P.; Yachnin, B. J.; Blanchet, J.; Guerrero, V.; Poulin,
L.; Fossati, E.; Berghuis, A. M.; Pelletier, J. N. Multiple conformers in
active site of human dihydrofolate reductase F31R/Q35E double
mutant suggest structural basis for methotrexate resistance. J. Biol.
Chem. 2009, 284 (30), 20079−20089.
(28) Park, H. Y.; Zhuang, P.; Nichols, R.; Howell, E. E. Mechanistic
studies of R67 dihydrofolate reductase - Effects of pH and an H62C
mutation. J. Biol. Chem. 1997, 272 (4), 2252−2258.
(29) Chopra, S.; Dooling, R. M.; Horner, C. G.; Howell, E. E. A
Balancing Act between Net Uptake of Water during Dihydrofolate
Binding and Net Release of Water upon NADPH Binding in R67
Dihydrofolate Reductase. J. Biol. Chem. 2008, 283 (8), 4690−4698.
(30) Cheng, Y.; Prusoff, W. H. Relationship between the inhibition
constant (K1) and the concentration of inhibitor which causes 50%
inhibition (I50) of an enzymatic reaction. Biochem. Pharmacol. 1973,
22 (23), 3099−3108.
(31) Thomsen, R.; Christensen, M. H. MolDock: a new technique
for high-accuracy molecular docking. J. Med. Chem. 2006, 49 (11),
3315−3321.
(32) Trott, O.; Olson, A. J. AutoDock Vina: Improving the speed and
accuracy of docking with a new scoring function, efficient
optimization, and multithreading. J. Comput. Chem. 2010, 31 (2),
455−461.
(33) Mosmann, T. Rapid colorimetric assay for cellular growth and
survival: Application to proliferation and cytotoxicity assays. J.
Immunol. Methods 1983, 65 (1−2), 55−63.
(34) Hicks, S. N.; Smiley, R. D.; Hamilton, J. B.; Howell, E. E. Role of
ionic interactions in ligand binding and catalysis of R67 dihydrofolate
reductase. Biochemistry 2003, 42 (36), 10569−10578.
(35) Kakkar, T.; Boxenbaum, H.; Mayersohn, M. Estimation of Ki in
a competitive enzyme-inhibition model: Comparisons among three
methods of data analysis. Drug Metab. Dispos. 1999, 27 (6), 756−762.
(36) Smiley, R. D.; Stinnett, L. G.; Saxton, A. M.; Howell, E. E.
Breaking symmetry: mutations engineered into R67 dihydrofolate
reductase, a D2 symmetric homotetramer possessing a single active
site pore. Biochemistry 2002, 41 (52), 15664−15675.
(37) Bradrick, T. D.; Beechem, J. M.; Howell, E. E. Unusual binding
stoichiometries and cooperativity are observed during binary and
ternary complex formation in the single active pore of R67
dihydrofolate reductase, a D2 symmetric protein. Biochemistry 1996,
35 (35), 11414−11424.
(11) Appleman, J. R.; Prendergast, N.; Delcamp, T. J.; Freisheim, J.
H.; Blakley, R. L. Kinetics of the formation and isomerization of
methotrexate complexes of recombinant human dihydrofolate
reductase. J. Biol. Chem. 1988, 263 (21), 10304−10313.
(12) Skold, O. Resistance to trimethoprim and sulfonamides. Vet. Res.
̈
2001, 32 (3−4), 261−273.
(13) (a) Fling, M. E.; Walton, L.; Elwell, L. P. Monitoring of plasmid-
encoded, trimethoprim-resistant dihydrofolate reductase genes:
Detection of a new resistant enzyme. Antimicrob. Agents Chemother.
1982, 22 (5), 882−888. (b) Stone, D.; Smith, S. L. The amino acid
sequence of the trimethoprim-resistant dihydrofolate reductase
specified in Escherichia coli by R-plasmid R67. J. Biol. Chem. 1979,
254 (21), 10857−10861.
(14) Alonso, H.; Gready, J. E. Integron-sequestered dihydrofolate
reductase: A recently redeployed enzyme. Trends Microbiol. 2006, 14
(5), 236−242.
(15) Holland, J. C.; Linn, C. E.; DiGiammarino, E.; Nichols, R.;
Howell, E. E. Does R67 dihydrofolate reductase possess a proton
donor? Adv. Exp. Med. Biol. 1993, 338, 493−498.
(16) Huovinen, P. Trimethoprim resistance. Antimicrob. Agents
Chemother. 1987, 31 (10), 1451−1456.
(17) Howell, E. E. Searching sequence space: two different
approaches to dihydrofolate reductase catalysis. ChemBioChem 2005,
6 (4), 590−600.
(18) Schmitzer, A. R.; Lep
́
ine, F.; Pelletier, J. N. Combinatorial
exploration of the catalytic site of a drug-resistant dihydrofolate
reductase: Creating alternative functional configurations. Protein Eng.
Des. Sel. 2004, 17 (11), 809−819.
(19) Krahn, J. M.; Jackson, M. R.; DeRose, E. F.; Howell, E. E.;
London, R. E. Crystal structure of a type II dihydrofolate reductase
catalytic ternary complex. Biochemistry 2007, 46 (51), 14878−14888.
(20) (a) Zartler, E. R.; Shapiro, M. J. Fragonomics: Fragment-based
drug discovery [Review]. Curr. Opin. Chem. Biol. 2005, 9 (4), 366−
370. (b) de Kloe, G. E.; Bailey, D.; Leurs, R.; de Esch, I. J.
Transforming fragments into candidates: Small becomes big in
medicinal chemistry. Drug Discovery Today 2009, 14 (13−14), 630−
646. (c) Gozalbes, R.; Carbajo, R. J.; Pineda-Lucena, A. Contributions
of computational chemistry and biophysical techniques to fragment-
based drug discovery. Curr. Med. Chem. 2010, 17 (17), 1769−1794.
(21) Murray, C. W.; Rees, D. C. The rise of fragment-based drug
discovery. Nat. Chem. 2009, 1 (3), 187−192.
(38) Alonso, H.; Gillies, M. B.; Cummins, P. L.; Bliznyuk, A. A.;
Gready, J. E. Multiple ligand-binding modes in bacterial R67
dihydrofolate reductase. J. Comput.-Aided Mol. Des. 2005, 19 (3),
165−187.
(39) Howell, E. E.; Shukla, U.; Hicks, S. N.; Smiley, R. D.; Kuhn, L.
A.; Zavodszky, M. I. One site fits both: A model for the ternary
complex of folate + NADPH in R67 dihydrofolate reductase, a D2
symmetric enzyme. J. Comput.-Aided Mol. Des. 2001, 15 (11), 1035−
1052.
(40) Segel, I. H. In Enzyme Kinetics: Behavior and Analysis of Rapid
Equilibrium and Steady-State Enzyme Systems, Wiley Classics Library
ed.; John Wiley and Sons: New York, 1993; p 957.
3191
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192

Journal of Medicinal Chemistry
Article
(41) Oefner, C.; D'Arcy, A.; Winkler, F. K. Crystal structure of
human dihydrofolate reductase complexed with folate. Eur. J. Biochem.
1988, 174 (2), 377−385.
3192
dx.doi.org/10.1021/jm201645r | J. Med. Chem. 2012, 55, 3182−3192