Received: 18 February 2019 

|
  Revised: 9 April 2019 
|  Accepted: 18 April 2019

DOI: 10.1111/cbdd.13536

RESEARCH ARTICLE

Structure‐based drug design and in vitro testing reveal new

inhibitors of enoyl-acyl carrier protein reductases

Mohammad A. Ghattas1   | Nermin A. Eissa1  | Francesca Tessaro2,3  | Remo Perozzo2,3  |

Leonardo Scapozza   2,3
| Dana Obaid   1
| Noor Atatreh 1

1
College of Pharmacy, Al Ain University
of Science and Technology, Abu Dhabi,
Abstract
United Arab Emirates The need for new antibacterial agents is increasingly becoming of great importance
2
Pharmaceutical Biochemistry as bacterial resistance to current drugs is quickly spreading. Enoyl‐acyl carrier pro-
Group, University of Geneva, Geneva,
tein reductases (FabI) are important enzymes for fatty acid biosynthesis in bacte-
Switzerland
3 ria and other micro‐organisms. In this project, we conducted structure‐based virtual
University of Lausanne, Lausanne,
Switzerland screening against the FabI enzyme, and accordingly, 37 compounds were selected for
experimental testing. Interestingly, five compounds were able to demonstrate antimi-
Correspondence
Mohammad A. Ghattas and Noor Atatreh,
crobial effect with variable inhibition activity against various strains of bacteria and
College of Pharmacy, Al Ain University of fungi. Minimum inhibitory concentrations of the active compounds were determined
Science and Technology, P.O. Box 112612 and showed to be in low to medium micromolar range. Subsequently, enzyme inhibi-
Abu Dhabi, United Arab Emirates.
Emails: mohammad.ghattas@aau.ac.ae and tion assay was carried out for our five antimicrobial hits to confirm their biological
noor.atatreh@aau.ac.ae target and determine their IC50 values. Three of these tested compounds exhibited
inhibition activity for the FabI enzyme where our best hit MN02 had an IC50 value
Present address
Nermin A. Eissa, Department of of 7.8 μM. Furthermore, MN02 is a small bisphenolic compound that is predicted
Pharmacology & Therapeutics, College of to have all required features to firmly bind with the target enzyme. To sum up, hits
Medicine & Health Sciences, United Arab
discovered in this work can act as a good starting point for the future development of
Emirates University, Al Ain, United Arab
Emirates new and potent antimicrobial agents.

Funding information KEYWORDS

Al Ain University of Science and antimicrobial agents, bacterial resistance, bisphenol, dichlorophen, diphenylmethane, docking, drug
Technology discovery, enoyl-acyl carrier protein reductase, FabI

1  |   IN T RO D U C T ION wall production. This step is crucial for micro‐organism sur-

The search for new antimicrobial agents is still of great impor-
tance as bacterial resistance to current drugs is increasingly
spreading (Chitsaz & Brown, 2017; Furusawa, Horinouchi,
& Maeda, 2018). Enoyl‐acyl carrier protein reductases (FabI)
are associated in non‐covalent complexes with their cofac-
tor NAD(P)H, and they play an important role in bacterial
survival (Yao & Rock, 2017). FabI reduces fatty acids using
NAD(P)H in order to complete their biosynthesis during cell
vival and cell homeostasis (Heath & Rock, 1995; Marrakchi,
Zhang, & Rock, 2002; McInnes, 2007).
The fatty acid biosynthesis pathway has been considered
as increasingly attractive for antimicrobial drug discovery
(Eissa, Farrag, Shawer, & Ammar, 2017; Kronenberger
et al., 2017; Mehboob et al., 2015; Mistry, Truong, Ghosh,
Johnson, & Mehboob, 2017; Wadhwa, Saha, & Sharma,
2015). This is mainly because of the major structural differ-
ences between mammalian enzymes and their corresponding

Ghattas and Eissa authors contributed equally to this work.

Chem Biol Drug Des. 2019;00:1–11. wileyonlinelibrary.com/journal/cbdd © 2019 John Wiley & Sons A/S     1 |
 |
2       GHATTAS et al.
microbial isoforms (Massengo‐Tiasse & Cronan, 2009).
FabI inhibitors can be classified into three groups accord-
ing to their mechanism of action (Yao & Rock, 2016): The
first group is the bi‐substrate FabI inhibitors that bind co-
valently to the cofactor NAD(P) to form an adduct complex
(e.g., isoniazid). The second group consists of inhibitors
that bind with FabI‐NAD(P) complex in a reversible manner
such as triclosan (Figure  1) and its closely related deriva-
tives (Jiangwei Yao & Rock, 2016). The third group consists
of inhibitors that reversibly interact with the FabI‐NAD(P)
complex in the presence of the substrate molecule such
as CG400549 (Yum et  al., 2007) and AFN‐1252 (Zitko &
Doležal, 2016). The latter compound is one of the very few
FabI inhibitors that has reached phase II clinical trial (Zitko
& Doležal, 2016).
FabI are reductase enzymes that act on short‐chain fatty
acids (Massengo‐Tiasse & Cronan, 2009). They have con-
served triad in their catalytic pocket (Figure 1) that is com-
posed of lysine, tyrosine, and phenylalanine (or tyrosine).
The lysine residue has a role in stabilizing the binding of the
cofactor, while Tyr is actively involved in the redox catalytic
mechanism by donating a proton to the substrate (Parikh,
Moynihan, Xiao, & Tonge, 1999). The fatty acyl U‐shaped
conformation is directed by the third conserved residue,
that is Phe or Tyr, which is positioned nearby the NAD(P)
nicotinamide (Rozwarski, Vilchèze, Sugantino, Bittman, &
Sacchettini, 1999). Out of these conserved residues, only the
catalytic Tyr is involved in binding the FabI ligands (Ghattas,
Mansour, Atatreh, & Bryce, 2015). According to previ-
ous research, this conserved Tyr in addition to the cofactor
F I G U R E 1   The FabI active site (yellow cartoon) along with the
catalytic triad (yellow sticks), the NAD cofactor (orange sticks), and
the standard FabI inhibitor triclosan (blue sticks)
molecule was observed to be essential for binding of nearly
all studied FabI inhibitors (Ghattas et al., 2015). The bound
NAD(P) nicotinamide ring was found to be involved in π–π
stacking interactions with the aromatic ring of various FabI
inhibitors. On the other hand, NAD(P) ribose along with the
Tyr hydroxyl group was observed to be constantly involved in
hydrogen bonding interactions.
In this study, we aim to discover new antimicrobial agents
via conducting structure‐based virtual screening against the
FabI active site based on a previously validated and tested
docking approach (Ghattas, Eissa, Bardaweel, Abu Mellal,
& Atatreh, 2017; Ghattas et  al., 2015). Subsequently, top‐
ranked compounds identified by virtual screening and vi-
sual inspection were tested for their antimicrobial activity
against a panel of microbial strains. The minimum inhibi-
tory concentrations were then identified for the most active
hits. Subsequently, these hits were biologically tested via
FabI enzyme assay to confirm their binding with the target
protein.
2  |  EXPERIM ENTAL
2.1  |  Protein and ligand library preparation
Escherichia coli FabI (ecFabI) crystal structure was ob-
tained from the protein data bank (http://www.rcsb.org/),
PDB: 1QSG (Stewart, Parikh, Xiao, Tonge, & Kisker, 1999).
Solvent atoms were removed; then, MOE protein prepara-
tion module (MOE, 2016)b was used to check out the pro-
tein‐cofactor complex for any missing atoms or residues and
correct them accordingly (Halgren, 1996). Then, a previously
validated hydrogen bonding constraint (Ghattas et al., 2015)
was created on the NAD ribose 2‐hydroxyl group using the
protein preparation module in Maestro (Schrödinger Suite,
2015)d. Triclosan bound conformation was used to identify
the active site on the FabI‐NAD complex. Another FabI crys-
tal structure brought from staphylococcus aureus (saFabI),
PDB: 4ALI (Schiebel et  al., 2012), was used for analysis
purposes and has been prepared using the above‐mentioned
procedure.
The ligand library of National Cancer Institute/USA
(www.cancer.gov) was used for this screening and was fil-
tered based on the Veber's rules (Veber et  al., 2002) and
the Lipinski's rule of five (Lipinski, Lombardo, Dominy,
& Feeney, 1997). Hence, filtered ligand library had ligands
with the following criteria: rotatable bonds ≤10, logP ≤5,
hydrogen bond donor (HBD) ≤5, hydrogen bond acceptor
(HBA) ≤10, molecular weight ≤500 and polar surface area
(PSA) ≤140. After filtration, ligand library size was re-
duced from 273,885 to 191,929 ligands. Ligands were then
assigned the appropriate protonation states and tautomer
forms via LigPrep module in Maestro (Schrödinger Release,
2015)c
GHATTAS et al.   
   3
|

Star ng Database of 273,885 Ligands

Filtraon

Filtered Database of 191,929 Ligands

Const. Docking via

Drug-like characteristics:
Rotatable bond ≤ 10
Top 20,000 ranked List GLIDE-HTVS
Molecular weight ≤ 500
LogP ≤ 5.0 Const Docking via
PSA ≤ 140 Top 2000 ranked list GLIDE-SP
HB donor ≤ 5 acceptor ≤ 10

Select Const. Docking via

37 GLIDE-XP

5 hits
In vitro Evaluaon

F I G U R E 2   Virtual screening protocol used for the discovery of new FabI inhibitors. PSA: polar surface area; HB: hydrogen bond

(Small‐Molecule Drug Discovery Suite, 2015)e. A three‐step

2.2  |  Virtual screening docking protocol
docking protocol was employed to allow gradual increase
Constrained docking into ecFabI enzyme active site for the in docking precision: High‐throughput virtual screening
prepared drug‐like ligand library was employed using GLIDE (HTVS), standard precision (SP), and then extra‐precision

T A B L E 1   The antimicrobial activity of the 37 compounds tested against various strains of bacterial and fungal organisms, using the disk
diffusion test

Zone of Inhibition (mm)

S. aureus B. subtilis E. coli C. albicans

Compounds (ATCC259230) MRSA (ATCC3359) (ATCC6633) (ATCC25922) (ATCC10231)
MN01 − NT NT − NT
MN02 +++ +++ +++ + +
MN03 – MN04 − NT NT − NT
MN05 +++ +++ +++ − +
MN06 – MN13 − NT NT − NT
MN14 +++ +++ ++ − −
MN15 – MN31 − NT NT − NT
MN32 + − + − −
MN33 – MN35 − NT NT − NT
MN36 + +++ + − +
MN37 − NT NT − NT
chloramphenicol +++ ++ +++ +++ NT
Ketoconazole NT NT NT NT +++
Note: Key to symbols: − inactive (inhibition zone < 6 mm); slightly active = + (inhibition zone 7–9 mm); moderately active = + + (inhibition zone 10–13 mm);
highly active = + + + (inhibition zone > 14 mm) (NT= not tested).
 |
4       GHATTAS et al.

T A B L E 2   The MIC values (μg/ml) of

MIC (μg/ml)
our five hits along with the positive standard
S. aureus MRSA B. subtilis E. coli tested against different bacterial strains
Compounds (ATCC259230) (ATCC3359) (ATCC6633) (ATCC25922)
MN02 4.0 8.0 4.0 32.0
MN05 13.9 13.9 13.9 NT
MN14 15.9 15.9 15.9 NT
MN32 66.7 NT 66.7 NT
MN36 NT 41.2 NT NT
Chloramphenicol 4.0 74.0 4.0 4.0
a a a
Triclosan 0.125 64.0 0.4 0.5a
Abbreviation: NT, not tested.
a
MIC values obtained from literature.

(XP) docking algorithms. In GLIDE_XP, flexible ligand
sampling was enabled, taking into account nitrogen inver-
sions and ring conformations. Post‐docking minimization
was performed for top‐ten poses with a threshold value of
0.5  kcal/mol for rejecting minimized poses. Afterward, all
produced poses were scored and ranked via the Glide_XP
scoring function which takes accounts of several terms (i.e.,
van der Waals, hydrogen bond, electrostatic interactions, de-
solvation penalty, and penalty for intra‐ligand contact). The
2000 ligands docked and ranked by GLIDE‐XP were then
clustered based on MACCS algorithm (MDL, 2015)a and
were then visually inspected. Finally, 37 ligands belonging to
various structural scaffolds and showing convenient binding
modes in the ecFabI active site were selected for antibacterial
testing.
2.3  |  Antimicrobial testing
Antimicrobial susceptibility testing of 37 compounds se-
lected from virtual screening was performed. Screening
was initially conducted by a disk diffusion test against the
Gram‐positive micro‐organism Staphylococcus aureus and
to dry at room temperature in order to remove any residual
solvent. A sterile cotton swab was dipped into the inoculum
suspension that was adjusted to 0.5 McFarland standard tur-
bidity, and lawns were made on nutrient agar (S.  aureus,
MRSA, B. subtilis, and E. coli) and sabouraud dextrose agar
(C. albicans). Disks were then placed carefully on the agar
plates that had been already inoculated with micro‐organ-
isms. The plates were incubated at 37°C for 24 hr (bacteria)
and 30°C for 48 hr (fungi). After incubation, the diameter of
inhibition zones was measured to the nearest whole millim-
eter. Chloramphenicol (Sigma) and ketoconazole (Bio‐rad,
Marnes‐la‐Coquette, France) were used as reference antibac-
terial and antifungal agents, respectively. All experiments
were made in duplicate.
Minimum inhibitory concentrations (MICs) for the most
active compounds MN02, MN05, MN14, and MN36 against
E.  coli, S.  aureus, MRSA, and B.  subtilis (where applica-
ble) were determined by using the broth dilution method
150
Remaining activity %

the Gram‐negative bacteria Escherichia coli. Compounds 97.4% 100.0%

showed growth inhibition were further evaluated for antibac- 100
terial activity against a selected range of bacterial strains and
fungi, including methicillin‐resistant Staphylococcus aureus
(MRSA, ATCC33591), Bacillus subtilis (ATCC6633), and 45.7%
50
Candida albicans (ATCC10231). All strains were purchased
from Remel Europe (Datford, UK). Preliminary screening
was carried out by using disk diffusion method as per Clinical 0
Laboratory Standards Institute (CLSI, 2006, 2012) and litera-
µM
µM

µM

µM

µM
0µ

ture (Reller, Weinstein, Jorgensen, & Ferraro, 2009) where

0
0

10
10

10

10

10

10

nutrient agar medium (LABM, Lancashire, UK) was used

L
02

05

14

32

36

TC
N

for bacteria and sabouraud dextrose agar medium (LABM,

M

Lancashire, UK) was used for fungi. Sterile filter paper disks F I G U R E 3   Inhibition activity of the tested compounds shown as
(6 mm, sigma, st. Louis, MO or Whatman Number 1) were remaining activity expressed in percentage of MN02, MN05, MN14,
soaked in the test compounds dissolved in dimethyl sulfoxide MN36, and triclosan. The standard deviation bars for MN0, MN32,
(20 mM, Sisco Research Lab., Mumbai, India) and were left and triclosan were lower than 0.1% thus not visible in the graph
GHATTAS et al.   
|
   5

150
IC50 = 27.4 ± 1.3 µM 150
IC50 = 7.8 ± 1.2 µM 150 IC50 = 2.7 ± 0.8 µM

100 100 100

Activity
Activity
Activity

50 50
50

0 0
0 –2 –1 0 1 2 3
–4 –2 0 2 4
–4 –2 0 2 4
Conc [Log]
Conc [Log] –50
Conc [Log] –50
–50

MN14 MN02 TCL

F I G U R E 4   Dose–response curve of molecules MN02 and MN14 as well as the positive control triclosan. The values reported in the graph
represent the mean values of a triplicate

(Wiegand, Hilpert, & Hancock, 2008). Tested compounds
were dissolved in dimethyl sulfoxide where twofold dilutions
of solution were prepared and then were diluted in culture
medium (Nutrient broth; LABM, Lancashire, UK) to obtain
a final concentration range of 0.56–256 μg/ml. The dimethyl
sulfoxide final concentration never exceeded 1% v/v. Each
test‐tube was inoculated with micro‐organism suspensions
at 1 × 106 CFU/ml (colony forming unit/ml) to obtain final
desired inoculum of 5 × 105 CFU/ml. MICs were read after
incubation at 37°C for 24  hr (bacteria). Growth controls
consisting of media with 1% v/v dimethyl sulfoxide were
employed. Chloramphenicol was used as a reference for an-
tibacterial activity. All experiments were repeated two times.
MIC values were determined as the lowest concentrations of
compound that completely inhibited the visible growth of
each strain. That was detected by unaided eye after overnight
incubation.
2.4  |  Enzymatic assay against saFabI
Enzyme activity at a fixed concentration of compound was
assessed using spectrophotometric method at 25°C. That was
done by monitoring the decrease in NADPH absorbance at
340 nm in the presence of trans‐2‐decenoyl‐n‐acetylcysteam-
ine substrate during 2 min (Slepikas et al., 2016). The standard
experimental conditions in a total volume of 1 ml consisted of
assay buffer (20 mM Hepes, 150 mM NaCl, pH 7.4), 9.6 μg
of purified recombinant SaFabI, 200 μM of NADPH, 100 μM
of substrate, and 100 μM of compound (20 mM stock solu-
tion in DMSO). IC50 values were determined using the same

T A B L E 3   The GLIDE‐XP docking scores of our hits obtained from docking into two ENR enzymes, ecFabI and saFabI, along with their
ligand efficiency

Docking into ecFabI (kcal/mol) Docking into saFabI (kcal/mol)

Compound code Structure GLIDE‐XP score Ligand efficiencya GLIDE‐XP score Ligand efficiencya
MN02 −9.24 −0.54 −9.14 −0.52

MN05 −8.07 −0.47 −6.87 −0.40

MN14 −7.88 −0.39 −8.56 −0.43

MN32 −7.82 −0.37 −6.96 −0.32

MN36 −7.97 −0.35 −8.46 −0.37

triclosan −8.15 −0.47 −9.20 −0.51

a
Ligand efficiency is calculated via dividing docking score over molecular weight.
 |
6       GHATTAS et al.
method but by varying the compound concentration. DMSO
was kept constant at 0.5%. Data were analyzed by non‐lin-
ear fit to the sigmoidal dose–response curve using GraphPad
Prism software. All measurements were carried out in tripli-
cate from the same batch of protein.
3  |  R ES U LTS A N D D IS C U S S ION
Constrained docking has reportedly been showing en-
hanced success rates of virtual screening in many protein
families, such as protein tyrosine phosphatases and kinases
(Ghattas, Atatreh, Bichenkova, & Bryce, 2014; Perola,
2006). Similarly, a previous study showed that constrain-
ing interaction, namely to the NAD(P) ribose 2‐hydroxyl
group, can significantly increase hit rates and improve
docking accuracy for FabI virtual screens (Ghattas et  al.,
2015). Accordingly, hydrogen bonding constraint was also
applied in this work to raise our chance for discovering
new FabI inhibitors.
The virtual screening protocol employed here is schemat-
ically shown in Figure  2. The NCI ligand library was first
filtered for druglikeness before docking into the ecFabI ac-
tive site, reducing the library size from 273,885 ligands to
191,929 ligands. The drug‐like library was initially screened
using the GLIDE‐HTVS docking algorithm. The top 20,000
ligands were re‐docked in a slower but more accurate dock-
ing algorithm using GLIDE‐SP. Finally, docking results were
enhanced using the extra‐precision docking and scoring
F I G U R E 5   The best predicted pose of
MN02 (pink sticks) docked into the active
site of saFabI (a) and ecFabI (b) where
the surrounding residues (including the
cofactor) are shown as green or blue sticks,
respectively. Interactions are shown as blue
dotted lines
algorithm (GLIDE‐XP) in the Schrodinger software package.
The resultant docked poses of these 2000 ligand were visu-
ally inspected inside the ecFabI binding pocket. Further, 37
compounds from various chemical clusters were selected for
antibacterial testing based on their ligand–protein interaction
profiles and according to their fitting into the catalytic pocket
of the target enzyme.
The best hits selected from virtual screening were screened
initially against the E. coli and S. aureus strains using disk
diffusion test. Amongst the 37 tested compounds, five hits
(i.e., MN02, MN05, MN14, MN32, and MN36) showed clear
inhibition zones in at least one of the cultured bacterial plates,
whereas the rest of compounds revealed no significant inhibi-
tion. Those five hits were further screened against two more
bacterial strains (MRSA and B. subtilis) and one fungal strain
(C. albicans) to check their spectrum of activity. As shown in
Table 1, MN02 and MN05 showed high inhibition activity in
particular against S. aureus, MRSA, and B. subtilis along with
weak inhibition activity against C.  albicans. However, the
growth of the Gram‐negative bacteria (i.e., E. coli) was only
inhibited by the former compound (MN02) which showed the
broadest spectrum of activity amongst all tested compounds.
MN14 showed similar bacterial inhibition as MN02
and MN05 against S.  aureus and MRSA with lesser activ-
ity against B. subtilis and none against C. albicans. On the
other hand, MN36 exhibited similar inhibition activity as
MN02 against MRSA and C. albicans but was less effective
against S. aureus and B. subtilis. Finally, MN32 exhibited the
weakest antimicrobial activity amongst the five compounds
F I G U R E 6   The best predicted pose of
MN05 (pink sticks) docked into the active
site of saFabI (a) and ecFabI (b) where
the surrounding residues (including the
cofactor) are shown as green or blue sticks,
respectively. Interactions are shown as blue
dotted lines
GHATTAS et al.
F I G U R E 7   The best predicted pose of
MN14 (pink sticks) docked into the active
site of saFabI (a) and ecFabI (b) where
the surrounding residues (including the
cofactor) are shown as green or blue sticks,
respectively. Interactions are shown as blue
dotted lines
demonstrating weak inhibition activity against S. aureus and
B. subtilis with no observed activity against E. coli or C. al-
bicans. In conclusion, all discovered hits exhibited more pro-
nounced activity against Gram‐positive bacteria compared
with Gram‐negative and fungal strains.
Active hits obtained from disk diffusion testing were fur-
ther screened to determine their minimum inhibitory concen-
tration (MIC) using the broth dilution method. As shown in
Table 2, amongst the five tested compounds, MN02 exhibited
the best antibacterial potency with MIC values ranging from
low to intermediate micromolar range, which is compara-
ble to antibacterial activity of chloramphenicol. In addition
to showing the best inhibitory activity, MN02 demonstrated
the best spectrum of activity against the tested panel of
bacterial strains, being more active though against Gram‐
positive (MIC  =  4–8  μg/ml) than Gram‐negative bacteria
(MIC = 32 μg/ml).
The other two hits, MN05 and MN14, showed anti-
bacterial activity in the middle micromolar range against
all Gram‐positive bacterial strains, obtaining MIC values
of 14 and 16  μg/ml, respectively. The final two inhibi-
tors, MN32 and MN36, had weaker inhibition activities
than previous compounds, exhibiting MIC values in the
range of 40–67  μg/ml. Most interestingly, four out of
the five discovered hits (i.e., MN02, MN05, MN14, and
MN36) demonstrated better inhibition activities against
the penicillin‐resistant strain MRSA than chloramphenicol
(MIC = 74 μg/ml) and triclosan, MIC reported as 64 μg/ml
(Regös, Zak, Solf, Vischer, & Weirich, 1979). To sum up,
all discovered hits exhibited an interesting anti‐Gram‐posi-
tive activity (particularly with regard to their MRSA activ-
ity), highlighting MN02 as the most potent antimicrobial
agent with the broadest spectrum of activity.
The five active compounds were further screened against
enzymatic inhibition assay (Figure  3) in order to confirm
whether or not they exert their antibacterial activity through
inhibiting FabI. Amongst the five tested compounds, MN02
and MN14 showed complete inhibition for saFabI enzyme at
100 μM, while MN05 exhibited a lesser inhibition percent-
age of 45% at the aforementioned concentration (Figure 3).
The other two compounds, MN32 and MN36, exhibited no
  
   7
|
significant inhibition for the target enzyme. Triclosan was
used as a positive control, and it showed full inhibition at
100 μM.
The three hits with most significant saFabI inhibition
were undergone further screening to determine their IC50.
As shown in Figure 4, while MN14 exhibited an IC50 value
of 27.4  μM, MN02 showed enzyme inhibition activity in
the low micromolar range with an IC50 value of 7.8  μM,
which is comparable to the standard inhibitor triclosan
(IC50 = 2.7 μM). The IC50 value of MN05 could not be mea-
sured because of solubility issues related to the compound;
however, since MN05 was able to inhibit saFabI activity by
45.7% at the 100 μM concentration, its IC50 value is estimated
to be around 100 μM. In summary, MN02, MN05, and MN14
have been all confirmed to exert their antimicrobial activity
via inhibiting the FabI enzyme, while MN32 and MN36 seem
to act through other routes.
To explain and correlate in vitro results with the in silico
data, docking results of the discovered hits were studied and
analyzed. Looking at Table 3, MN02 was predicted to have
the best GLIDE_XP docking scores for both FabI enzymes,
E.  coli (−9.24  kcal/mol) and S.  aureus (−9.14  kcal/mol),
scoring at least as good as reference ligand triclosan (−8.15
to 9.20 kcal/mol). Interestingly, MN02's docking data seem
to be consistent with the in vitro findings that recognize it
as the most active hit. The other four hits exhibited conve-
nient docking scores with comparable or higher binding en-
ergies than the reference ligand, in term of how these ligands
are scoring as per their size, as shown in Table 3. Our best
hit MN02 was also able to obtain the best ligand efficiency
score (−0.5  kcal/mol), and relatively similar scoring to tri-
closan (−0.47 to −0.51 kcal/mol). This suggests that MN02
possesses lead‐like characteristics and offers possibility for
structural modifications and improvement future work.
Considering the docking pose, MN02 forms all key inter-
actions with the cofactor and the catalytic Tyr residue. It had
its hydroxyl group simultaneously interacting via hydrogen
bonding with the NAD ribose and the Tyr156 side chain, in
addition to having its phenyl group forming π–π interaction
with the NAD nicotinamide ring (Figure  5). Interestingly,
MN02 bears structural similarity with triclosan which could
 |
8       GHATTAS et al.

T A B L E 4   The GLIDE_XP docking

GLIDE‐XP Score (kcal/
scores of dichlorophen and other derivatives
mol)
into ecFabI and saFabI
Compound code Structure ecFabI saFabI
Resorcinol sulfide −7.52 −7.62

Dichlorophen −5.81 −6.66

Bromochlorophen −5.41 −5.66

Bithionol −5.00 −5.26

Hexachlorophene −4.43 −5.07

MN02 −9.24 −9.14

explain why both ligands possess similar binding modes and
antimicrobial activity profiles (Stewart et al., 1999).
MN05 belongs to the monohydroxydiphenylmethane class
of compounds that are known for their bactericidal activity
(Florestano & Bahler, 1953; Klarmann, Gates, & Shternov,
1932). However, bacterial targets for such compounds have
never been identified and it is the first time to report FabI as
a drug target for them. In terms of binding, MN05 possesses
a very similar binding mode to MN02 and triclosan, form-
ing the two necessary hydrogen bonding with the NAD ri-
bose and Tyr156 (Figure 6). However, unlike the latter two
ligands, MN05 does not make the commonly seen π‐π inter-
action with the NAD nicotinamide ring. That is probably due
to steric hindrance caused by the meta substitution of MN05,
which is slightly pushing away the aromatic moiety from the
deepest point in the FabI binding cavity.

(a) (b)

F I G U R E 8   (a) The best predicted pose of dichlorophen (red) along with its para derivative MN02 (pink). (b) The best predicted pose of
bithionol (brown) along with its para derivative resorcinol sulfide (green). The reference ligand triclosan is shown as blue sticks in both pictures,
and the active site of the ecFabI‐NAD complex is shown as blue ribbon
GHATTAS et al.
MN14 is another hit confirmed to exert its antibacterial
action via inhibiting the FabI enzyme. As shown in Figure 7,
MN14 forms two hydrogen bonds with the NAD ribose via
its phenol and ketone functionalities. Additionally, the phenol
group was shown to make an extra hydrogen bonding with
the Tyr156 side chain. MN14 was also predicted to form the
usual interaction with NAD nicotinamide ring in addition to
several hydrogen–π interactions and vdW contacts with the
surrounding residues. It is worth mentioning that interaction
between the ketone group and the NAD ribose was only ob-
served in MN14 binding with ecFabI. However, this interac-
tion was compensated in saFabI by an alternative hydrogen
bonding observed between Ser197 and the distant phenol
group of the compound.
In fact, MN02 is structurally close to the known dichloro-
phen with the only difference in the position of chloro substi-
tutions. Dichlorophen is a compound used in many veterinary
preparations (Gilvydis, 1972) as anthelmintic agent and an-
tiprotozoal agent (Kawasaki & Takeuchi, 1984; Takeuchi,
Kobayashi, Tanabe, & Fujiwara, 1985). As the protein tar-
get of dichlorophen and its derivatives (i.e., bithionol, bro-
mochlorophen, resorcinol sulfide, and hexachlorophene) has
never been reported before, it would be interesting to check
the binding potential of these compounds via docking them
into the FabI active site. As shown in Table  4, all of these
compounds have similar structures to MN02; however, none
of them (except resorcinol) were able to score as low docking
score as MN02.
Figure 8 shows the binding mode of bithionol and dichlo-
rophen along with their para‐substituted derivatives (i.e., re-
sorcinol sulfide and MN02, respectively). It is evident that
the usual interaction with the cofactor and the catalytic Tyr
residue is hindered by the meta substitution on bithionol and
dichlorophen. This explains why such structurally similar
compounds could not fit as good as MN02 in the FabI binding
site. Hence, it can be suggested that bisphenolic compounds
with para substitutions have more potential to inhibit the FabI
enzyme, while those with meta substitutions are less likely
to do so and they could rather exert their antimicrobial ef-
fect via binding to other targets in the fatty acid biosynthesis
pathway, such as 3‐oxoacyl‐ACP reductase (Wickramasinghe
et al., 2006).
4  |   CO NC LU SION S
A validated structure‐based virtual screening has been car-
ried out against the FabI catalytic pocket. Consequently,
a total number of 37 hits were selected for experimental
testing, five of which showed antibacterial activity particu-
larly against Gram‐positive strains. The obtained hits ex-
hibited MIC values in the micromolar range, showing very
interesting activity against MRSA. Our best hit, MN02,
  
   9
|
exhibited the broadest spectrum of activity, eradicating
both Gram‐positive and Gram‐negative bacteria. Three
out of the discovered hits were confirmed to bind and in-
hibit the FabI enzyme, showing IC50 values in the low to
medium micromolar range. These three active compounds
showed convenient docking binding modes inside the FabI
catalytic pocket, satisfying the key interactions with the
bound cofactor as well as the catalytic Tyr residue. The
obtained hits, particularly MN02, look promising leads for
future development of antibacterial agents.
ACKNOWLEDGMENTS
This work was funded by Deanship of Scientific Research
and Graduate Studies at Al Ain University of Science and
Technology, Al Ain, United Arab Emirates. The authors ac-
knowledge National Cancer Institute, USA, for providing the
compounds used in this work. We would like to thank Dr
Nezar Al Bataineh for proofreading the manuscript.
CONFLICT OF INTEREST
The authors declare that no conflict of interest is associated
with this work.
ORCID
Mohammad A. Ghattas  https://orcid.
org/0000-0002-2240-8037
ENDNOTES
1
MDL. (2015). MACCS Keys, MDL Information Systems Inc., 14600
Catalina Street, San Leandro, CA 94577.
2
MOE. (2016). manual version 2016.09, Molecular Operating
Environment (MOE), Chemical Computing Group, http://www.chemc​
omp.com, Montreal, Canada.
3
Schrödinger Release 2015‐2: LigPrep, version 3.4, Schrödinger, LLC,
New York, NY, 2015.
4
Schrödinger Suite 2015‐2 Protein Preparation Wizard; Epik version 3.2,
Schrödinger, LLC, New York, NY, 2015; Impact version 6.7, Schrödinger,
LLC, New York, NY, 2015; Prime version 4.0, Schrödinger, LLC, New
York, NY, 2015.
5
Small‐Molecule Drug Discovery Suite 2015‐2: Glide, version 6.7,
Schrödinger, LLC, New York, NY, 2015.
R E F E R E NC E S
Chitsaz, M., & Brown, M. H. (2017). The role played by drug efflux
pumps in bacterial multidrug resistance. Essays In Biochemistry,
61(1), 127–139. https​://doi.org/10.1042/ebc20​160064
Clinical Laboratory Standards Institute (CLSI) (2006). Performance
standards for antimicrobial disk susceptibility tests; Approved
 |
10       GHATTAS et al.
standard‐9th ed. CLSI document M2‐A9. 26:1. Wayne, PA: Clinical
Laboratory Standards Institute.
Clinical Laboratory Standards Institute (CLSI) (2012). Methods for di-
lution antimicrobial susceptibility tests for bacteria that grow aer-
obically; approved standard—9th edition. CLSI document M07–A9
32: 2. Wayne, PA: Clinical Laboratory Standards Institute.
Eissa, S. I., Farrag, A. M., Shawer, T. Z., & Ammar, Y. A. (2017).
Design, synthesis, 3D pharmacophore, QSAR, and docking stud-
ies of some new (6‐methoxy‐2‐naphthyl) propanamide derivatives
with expected anti‐bacterial activity as FABI inhibitor. Medicinal
Chemistry Research, 26(10), 2375–2398. https​://doi.org/10.1007/
s00044-017-1939-1
Florestano, H. J., & Bahler, M. E. (1953). Antibacterial activity of a se-
ries of diphenylmethanes. Journal of the American Pharmaceutical
Association, 42(9), 576–578. https​://doi.org/doi:10.1002/jps.30304​
20916​
Furusawa, C., Horinouchi, T., & Maeda, T. (2018). Toward pre-
diction and control of antibiotic‐resistance evolution. Current
Opinion in Biotechnology, 54, 45–49. https​ ://doi.org/10.1016/j.
copbio.2018.01.026
Ghattas, M. A., Atatreh, N., Bichenkova, E. V., & Bryce, R. A. (2014).
Protein tyrosine phosphatases: Ligand interaction analysis and op-
timisation of virtual screening. Journal of Molecular Graphics and
Modelling, 52, 114–123.
Ghattas, M. A., Eissa, N. A., Bardaweel, S. K., Abu Mellal, A., &
Atatreh, N. (2017). Computer‐aided discovery of antimicrobial
agents as potential enoyl-acyl carrier protein reductase inhibitors.
Tropical Journal of Pharmaceutical Research, 16(2), 397–405.
Ghattas, M. A., Mansour, R. A., Atatreh, N., & Bryce, R. A. (2015).
Analysis of Enoyl‐Acyl carrier protein reductase structure and inter-
actions yields an efficient virtual screening approach and suggests
a potential allosteric site. Chemical Biology & Drug Design, 87(1),
131–142. https​://doi.org/10.1111/cbdd.12635​
Gilvydis, D. M. (1972). Collaborative study of the determination of
dichlorophene in veterinary preparations by ultraviolet spectropho-
tometry. Journal of the Association of Official Analytical Chemists,
55(1), 163–165.
Halgren, T. A. (1996). Merck molecular force field. V. Extension of
MMFF94 using experimental data, additional computational data,
and empirical rules. Journal of Computational Chemistry, 17(5–6),
616–641. https​://doi.org/10.1002/(SICI)1096-987X(19960​4)17:5/6
<616: AID-JCC5>3.0.CO;2-X
Heath, R. J., & Rock, C. O. (1995). Enoyl‐Acyl carrier protein reductase
(fabI) plays a determinant role in completing cycles of fatty acid
elongation in Escherichia coli. Journal of Biological Chemistry,
270(44), 26538–26542. https​://doi.org/10.1074/jbc.270.44.26538​
Kawasaki, H., & Takeuchi, T. (1984). Entamoeba histolytica: Effects of
dichlorophene and hexachlorophene on respiratory activities, growth
in vitro and ultrastructure. Japanese Journal of Parasitology, 33(6),
545–554.
Klarmann, E., Gates, L. W., & Shternov, V. A. (1932). Halogen deriv-
atives of monohydroxydiphenylmethane and their antibacterial ac-
tion. Journal of the American Chemical Society, 54(8), 3315–3328.
https​://doi.org/10.1021/ja013​47a045
Kronenberger, T., Asse, L. R., Wrenger, C., Trossini, G. H. G., Honorio,
K. M., & Maltarollo, V. G. A. (2017). Studies of Staphylococcus
aureus FabI inhibitors: Fragment‐based approach based on holo-
graphic structure‐activity relationship analyses. Future Medicinal
Chemistry, 9(2), 135–151. https​://doi.org/10.4155/fmc-2016-0179
Lipinski, C. A., Lombardo, F., Dominy, B. W., & Feeney, P. J. (1997).
Experimental and computational approaches to estimate solubil-
ity and permeability in drug discovery and development settings.
Advanced Drug Delivery Reviews, 23(1–3), 3–25.
Marrakchi, H., Zhang, Y.‐M., & Rock, C. O. (2002). Mechanistic di-
versity and regulation of Type II fatty acid synthesis. Biochemical
Society Transactions, 30, 1050–1055.
Massengo‐Tiasse, R. P., & Cronan, J. E. (2009). Diversity in enoyl‐acyl
carrier protein reductases. Cellular and Molecular Life Sciences,
66(9), 1507–1517. https​://doi.org/10.1007/s00018-009-8704-7
McInnes, C. (2007). Virtual screening strategies in drug discovery.
Current Opinion in Chemical Biology, 11(5), 494–502.
Mehboob, S., Song, J., Hevener, K. E., Su, P.‐C., Boci, T., Brubaker, L.,
… Johnson, M. E. (2015). Structural and biological evaluation of
a novel series of benzimidazole inhibitors of Francisella tularensis
enoyl‐ACP reductase (FabI). Bioorganic and Medicinal Chemistry
Letters, 25(6), 1292–1296.
Mistry, T. L., Truong, L., Ghosh, A. K., Johnson, M. E., & Mehboob,
S. (2017). Benzimidazole‐based FabI inhibitors: A promising
novel scaffold for anti‐staphylococcal drug development. ACS
Infectious Diseases, 3(1), 54–61. https​ ://doi.org/10.1021/acsin​
fecdis.6b00123
Parikh, S., Moynihan, D. P., Xiao, G., & Tonge, P. J. (1999). Roles of
tyrosine 158 and lysine 165 in the catalytic mechanism of InhA,
the Enoyl‐ACP reductase from Mycobacterium tuberculosis.
Biochemistry, 38(41), 13623–13634. https​://doi.org/10.1021/bi990​
529c
Perola, E. (2006). Minimizing false positives in kinase virtual screens.
Proteins: Structure, Function, and Bioinformatics, 64(2), 422–435.
https​://doi.org/10.1002/prot.21002​
Regös, J., Zak, O., Solf, R., Vischer, W. A., & Weirich, E. G. (1979).
Antimicrobial spectrum of triclosan, a broad‐spectrum antimicro-
bial agent for topical application. Dermatologica, 158, 72–79.
Reller, L. B., Weinstein, M., Jorgensen, J. H., & Ferraro, M. J. (2009).
Antimicrobial susceptibility testing: A review of general principles
and contemporary practices. Clinical Infectious Diseases, 49(11),
1749–1755.
Rozwarski, D. A., Vilchèze, C., Sugantino, M., Bittman, R., &
Sacchettini, J. C. (1999). Crystal structure of the Mycobacterium
tuberculosis Enoyl‐ACP reductase, InhA, in complex with NAD+
and a C16 Fatty Acyl substrate. Journal of Biological Chemistry,
274(22), 15582–15589. https​://doi.org/10.1074/jbc.274.22.15582​
Schiebel, J., Chang, A., Lu, H., Baxter Michael, V., Tonge Peter, J., &
Kisker, C. (2012). Staphylococcus aureus Fab I: Inhibition, sub-
strate recognition, and potential implications for in vivo essentiality.
Structure, 20(5), 802–813. https​://doi.org/10.1016/j.str.2012.03.013
Slepikas, L., Chiriano, G., Perozzo, R., Tardy, S., Kranjc, A.,
Patthey‐Vuadens, O., … Scapozza, L. (2016). In silico driven
design and synthesis of rhodanine derivatives as novel antibac-
terials targeting the enoyl reductase InhA. Journal of Medicinal
Chemistry, 59(24), 10917–10928. https​ ://doi.org/10.1021/acs.
jmedc​hem.5b01620
Stewart, M. J., Parikh, S., Xiao, G., Tonge, P. J., & Kisker, C. (1999).
Structural basis and mechanism of enoyl reductase inhibition by tri-
closan. Journal of Molecular Biology, 290(4), 859–865. https​://doi.
org/10.1006/jmbi.1999.2907
Takeuchi, T., Kobayashi, S., Tanabe, M., & Fujiwara, T. (1985). In vitro
inhibition of Giardia lamblia and Trichomonas vaginalis growth
by bithionol, dichlorophene, and hexachlorophene. Antimicrobial
GHATTAS et al.
Agents and Chemotherapy, 27(1), 65–70. https​://doi.org/10.1128/
aac.27.1.65
Veber, D. F., Johnson, S. R., Cheng, H.‐Y., Smith, B. R., Ward, K. W., &
Kopple, K. D. (2002). Molecular properties that influence the oral
bioavailability of drug candidates. Journal of Medicinal Chemistry,
45(12), 2615–2623. https​://doi.org/doi:10.1021/jm020​017n
Wadhwa, P., Saha, D., & Sharma, A. (2015). Combined 3D‐QSAR and
molecular docking study for identification of diverse natural prod-
ucts as potent Pf ENR inhibitors. Current Computer‐Aided Drug
Design, 11(3), 245–257.
Wickramasinghe, S. R., Inglis, K. A., Urch, J. E., Müller, S., Aalten, D.
M. F. V., & Fairlamb, A. H. (2006). Kinetic, inhibition and structural
studies on 3‐oxoacyl‐ACP reductase from Plasmodium falciparum,
a key enzyme in fatty acid biosynthesis. Biochemical Journal,
393(2), 447–457. https​://doi.org/10.1042/bj200​50832​
Wiegand, I., Hilpert, K., & Hancock, R. E. (2008). Agar and broth di-
lution methods to determine the minimal inhibitory concentration
(MIC) of antimicrobial substances. Nature Protocols, 3(2), 163–
175. https​://doi.org/10.1038/nprot.2007.521
Yao, J., & Rock, C. O. (2016). Resistance mechanisms and the future
of bacterial Enoyl‐Acyl carrier protein reductase (FabI) antibiot-
ics. Cold Spring Harbor Perspectives in Medicine, 6(3), https​://doi.
org/10.1101/cshpe​rspect.a027045
  
   11
|
Yao, J., & Rock, C. O. (2017). Bacterial fatty acid metabolism in mod-
ern antibiotic discovery. Biochimica et Biophysica Acta ‐ Molecular
and Cell Biology of Lipids, 1862(11), 1300–1309.
Yum, J. H., Kim, C. K., Yong, D., Lee, K., Chong, Y., Kim, C. M., …
Cho, J. M. (2007). In vitro activities of CG400549, a novel FabI
inhibitor, against recently isolated clinical staphylococcal strains in
Korea. Antimicrobial Agents and Chemotherapy, 51(7), 2591–2593.
https​://doi.org/10.1128/aac.01562-06
Zitko, J., & Doležal, M. (2016). Enoyl-acyl carrier protein reduc-
tase inhibitors: An updated patent review (2011–2015). Expert
Opinion on Therapeutic Patents, 26(9), 1079–1094. https​ ://doi.
org/10.1080/13543​776.2016.1211112
How to cite this article: Ghattas MA, Eissa NA,
Tessaro F, et al. Structure‐based drug design and in
vitro testing reveal new inhibitors of enoyl-acyl carrier
protein reductases. Chem Biol Drug Des. 2019;00:
1–11. https​://doi.org/10.1111/cbdd.13536​