Synthesis, biological evaluation, and molecular docking investigation of benzhydrol- and indole-based dual PPAR-γ/FFAR1 agonists

Article abstract of DOI:10.1016/j.bmcl.2018.03.051

Type-2 diabetes mellitus is a progressive cluster of metabolic disorders, representing a global public health burden affecting more than 366 million people worldwide. We recently reported the discovery of three series of novel agents showing balanced activity on two metabolic receptors, peroxisome proliferator activated receptor-γ (PPAR-γ) and free fatty acid receptor 1 (FFAR1), also known as GPCR40. Our designing strategy relied on linking the thiazolidinedione head with known GPCR privilege structures. To further investigate this concept, two new scaffolds, the benzhydrol- and indole-based chemotypes, were introduced here in. Our optimization campaign resulted in three compounds; 15a, 15c, and 15d, with affinities in the low micromolar range on both targets. In vivo study of selected test compounds, revealed that 15c possesses a significant anti-hyperglycemic and anti-hyperlipidemic activities superior to rosiglitazone in fat-fed animal models. Molecular docking analysis was conducted to explain the binding modes of both series. These compounds could lead to the development of the unique antidiabetic agent acting as insulin sensitizer as well as insulin secretagogue.

Full text of DOI:10.1016/j.bmcl.2018.03.051

Accepted Manuscript
Synthesis, biological evaluation, and molecular docking investigation of benz-
hydrol- and indole-based dual PPAR-γ/FFAR1 agonists
Khaled M. Darwish, Ismail Salama, Samia Mostafa, Mohamed S. Gomaa, El-
Sayed Khafagy, Mohamed A. Helal
PII:
S0960-894X(18)30244-0
https://doi.org/10.1016/j.bmcl.2018.03.051
BMCL 25705
DOI:
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
22 June 2017
18 March 2018
20 March 2018
Please cite this article as: Darwish, K.M., Salama, I., Mostafa, S., Gomaa, M.S., Khafagy, E-S., Helal, M.A.,
Synthesis, biological evaluation, and molecular docking investigation of benzhydrol- and indole-based dual PPAR-
γ/FFAR1 agonists, Bioorganic & Medicinal Chemistry Letters (2018), doi: https://doi.org/10.1016/j.bmcl.
2018.03.051
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Synthesis, biological evaluation, and molecular docking
investigation of benzhydrol- and indole-based
dual PPAR-γ/FFAR1 agonists
Khaled M. Darwisha,*, Ismail Salamaa, Samia Mostafaa, Mohamed S. Gomaaa, El-Sayed
Khafagyb, Mohamed A. Helala,c
a
Medicinal Chemistry Department, Faculty of Pharmacy, Suez Canal University, Ismailia
41522, Egypt
b
Department of Pharmaceutics and Industrial Pharmacy, Faculty of Pharmacy, Suez Canal
University, Ismailia 41522, Egypt
c
Biomedical Sciences Program, University of Science and Technology, Zewail City of
Science and Technology, Giza 12587, Egypt
*Corresponding author;
Business Tel: +20100-5330114
Fax number: +20-064-3230741
E-mail address: khaled_darwish@pharm.suez.edu.eg
1

Abstract
Type-2 diabetes mellitus is a progressive cluster of metabolic disorders, representing a global public health
burden affecting more than 366 million people worldwide. We recently reported the discovery of three series of
novel agents showing balanced activity on two metabolic receptors, peroxisome proliferator activated receptor-γ
(PPAR-γ) and free fatty acid receptor 1 (FFAR1), also known as GPCR40. Our designing strategy relied on
linking the thiazolidinedione head with known GPCR privilege structures. To further investigate this concept,
two new scaffolds, the benzhydrol- and indole-based chemotypes, were introduced here in. Our optimization
campaign resulted in three compounds; 15a, 15c, and 15d, with affinities in the low micromolar range on both
targets. In vivo study of selected test compounds, revealed that 15c possesses a significant anti-hyperglycemic
and anti-hyperlipidemic activities superior to rosiglitazone in fat-fed animal models. Molecular docking analysis
was conducted to explain the binding modes of both series. These compounds could lead to the development of
the unique antidiabetic agent acting as insulin sensitizer as well as insulin secretagogue.
Keywords: Type-2 diabetes mellitus, PPAR-γ, FFAR1, thiazolidindione, biological evaluation.
Although type-2 diabetes mellitus (T2DM) is not a recent disease, with a history dated
back to the middle of the sixth century, its prevalence is rising every day. The disease is a
progressive cluster of metabolic disturbances characterized by chronic hyperglycemia arising
from insufficient pancreatic insulin secretion and resistance to insulin action or a combination
of both. As for complications, T2DM is one of the primary causes of renal injury, eye
problems, amputations and hospitalizations and is associated with high rates of morbidity and
mortality [1]. Currently, T2DM is a global public health burden affecting more than 367
million people worldwide. According to the World Health Organization (WHO), the number
is expected to reach 557 million by the year 2030 [2].
An efficient management of T2DM comprises of several stages [3]. Typically, the first step
could be achieved through lifestyle modifications including healthy diet and exercise.
However, when changing lifestyle is not sufficient, pharmacotherapy is required. A variety of
oral hypoglycemic medications were developed either to increase insulin secretion
(secretagogues), enhance insulin action (insulin sensitizers), or slow glucose absorption [4].
Examples of insulin sensitizers include metformin and thiazolidinediones (TZDs), while oral
sulfonylureas and meglitinides are prescribed as insulin secretagogues. Since introduced in
the late 90’s, TZDs are considered as potent agonists of peroxisome proliferator-activated
receptors (PPARs). The latter represent a group of nuclear hormone receptors responsible for
the expression of a number of genes concerning with glucose and fat metabolism [5]. The
PPARs are abundant in the key target tissues for insulin action including skeletal muscles,
2

adipose tissues, and liver. Out of the three distinct PPAR isoforms, PPAR-α, -β and -γ, the
activation of PPAR-γ has the major role in regulating glucose homeostasis, inflammatory
responses, cellular differentiation, and apoptosis [6].
Till date, the first line treatment of T2DM is oral metformin followed by oral sulfonylureas,
dipeptidyl peptidase-IV-inhibitors, and TZDs. As the disease progresses, injectable glucagon-
like peptide-I analogues or ultimately insulin become necessary to maintain glycaemic
control [3]. Despite the established guidelines, there is still no cure for T2DM. Additionally,
many of the traditional current therapies are associated with gastro-intestinal discomforts
(metformin), hypoglycaemia, and weight gain (sulfonylurea and to lesser extent
meglitinides), or edema and increased risk of bone fractures (TZDs) [7].
In this regard, there has been an influx of novel antidiabetic agents influencing a variety of
cellular targets. Recently, free fatty acid receptor-I (FFAR1) has emerged as a promising
target, belonging to the class A family of G-protein-coupled receptors (GPCRs). FFAR1 is
mainly expressed in the pancreatic β-cells, and to a lesser extent in gastrointestinal
enteroendocrine cells, immune cells, and parts of the brain [8-10]. This lipid-binding receptor
enhances glucose-stimulated insulin secretion (GSIS) by direct action on β-cells and possibly
indirectly through the incretin system. Since FFAR1-induced GSIS is glucose-dependent,
FFAR1’s mechanism potentially mitigates the risk of hypoglycemia seen with standard
insulin secretagogues [11,12].
In our recent publication, we described the discovery of novel hybrid molecules possessing a
balanced dual agonism on PPAR-γ and FFAR1 receptors [13]. These hybrids would act as
insulin sensitizers and secretagogues through their action on PPAR-γ and FFAR1,
respectively. Unlike marketed secretagogues/sensitizers combination; Amaryl MTM
(metformin/glimepiride) and glucovanceTM (metformin/glibenclamide), these hybrids could
provide less complex pharmacokinetic/pharmacodynamic relationships. Such profiles would
lead to more predictable variability between patients and less extensive clinical studies [14].
Additionally, the afforded moderate activity of these hybrids, down to one digit µM, could be
beneficial to limit the potential adverse effects raised with potent PPAR-γ agonists.
Nevertheless, such hybrids could keep an acceptable overall anti-diabetic efficacy as a result
of their collective synergism on both receptors.
The design of our lead compounds depends on combining the 5-benzyl-thiazolidinedione
scaffold, as TZDs common pharmacophore, with diverse GPCRs’ privileged structures, as
3

hydrophobic tails, through a simple alkyl linker (Fig.1). In another previous study, we
illustrated through homology modelling and molecular dynamics simulations that the TZD
acidic head is capable of binding to the critical polar residues within both PPAR-γ and
FFAR1 binding domains [15]. The biphenyl, phenoxybenzyl, and benzimidazole-based
derivatives represent excellent choices as versatile hydrophobic tails (series MK-1, -2 and -5;
Fig.2). Such choice was based on preliminary molecular docking studies conducted in our
late publication [13]. These three scaffolds showed preferential binding, with both receptors,
out of a total 17 docked GPCRs’ privileged structures. Notably, our rationalized design came
in great consistency with a recently published mini-review article by Yasmin et al. [16],
summarizing the chemical modifications around TZDs in past two decades. Through
extensive mining, the authors depicted that the 5-[4-alkyloxybenzyl]-thiazolidinedione
scaffold, being adopted herein, is considered an optimal motif for furnishing favourable
PPAR-γ transactivation.
Simple Alkyl
Spacer
Linker
Hydrophobic
S
GPCR Class A
Glitazones
O
Privileged Structures
Pharmacophore
N
H
O
O
n
Fig. 1. General topology proposed for dual PPAR-γ/FFAR1 agonists.
Regarding our preliminary in-silico study, eight compounds showed balanced agonistic
activity on both PPAR-γ and FFAR1 receptors. Both the benzimidazole series MK-5 and
biphenyl series MK-1, ranked second in activity, were proved to be the most suitable
scaffolds for this purpose. Assessment of activity was done using luciferase bioluminescence
and calcium flux/fluorometric imaging plate reader (FLIPR) assays for PPAR-γ and FFAR1
transactivation, respectively. High intrinsic activities were noticed, with significantly larger
maximum fold-induction compared to that of the PPAR-γ full agonist reference,
rosiglitazone.
4

Fig. 2. Leads with the best docking score in both FFAR1 and PPAR-γ receptors, expressing the common
phrarmacophore in red.
Thrilled by these promising results, our quest for discovering new PPAR-γ/FFAR1 co-
agonists was directed towards exploring the other two new scaffolds. So, in this proposed
manuscript, we will describe our efforts regarding the synthesis, biological evaluation and
molecular docking studies of the benzhydrol- and indole-based hybrids (series MK-3 and -4,
respectively, Fig.2). Moreover, the effect on blood glucose and plasma triglyceride levels
were studied in vivo using the appropriate animal models.
Despite the novelty of the previously introduced series MK-2, a branched biaryl
chemotype with a stereogenic centre, it furnished mediocre potencies regarding the PPAR-γ
receptor. It was expected that the Y-shaped structures of this series could help the compounds
to fit into the two hydrophobic arms of PPAR-γ, while maintaining a suitable size that is not
too bulky [17-19]. Nevertheless, promising FFAR1 receptor activity at low micromolar range
persuaded us for further investigations. Thus, the benzhydrol-based series, MK-3, was
introduced herein as a forked biphenyl chemotype. The phenyl shift in series MK-3 was
designed to allow the terminal oxygen to be widely open for optimized H-bonding interaction
within the PPAR-γ binding pocket. This was based on predicted polar interaction between
linker oxygen of the most active members of series MK-2 and PPAR-γ central pocket residue
(R288).
As mentioned earlier, our explored scaffolds showed satisfactory results within previous
in silico investigations [13]. In a similar fashion to series MK-5, the indole based-chemotype
is another representative for the fused heterocycles having similarity with some reported
glitazones showing comparable euglycemic and hypolipidemic activities to rosiglitazone [20-
22]. However, this newly introduced series MK-4 was designed with a two-carbon spacer,
instead of the methoxy linker in series MK-5. Such linker extension was to bury the
5

hydrophobic tail deeper into the binding site for better lodging. This might help to achieve
significant anchoring of the TZD head and allow favoured conformations that could minimize
the steric clashes.
Initial retrosynthetic dissection of the leads revealed that any of them can be broken
down into three fragments; 5-benyl-thiazolidindione head, linker, and hydrophobic synthon.
The key thiazolidinedione head group was prepared byan economic two-stage cyclization
reaction [23] (Scheme 1).
Scheme 1. (a) (i) 0-5 °C, water, stirring for 15 min (ii) 2 N HCl, H2O, reflux, 12 h, 87 %; (b) 4-
hydroxybenzaldehyde, benzoic acid, piperidine, toluene, reflux, overnight, 75 %.
Later, a parallel synthesis of both the hydrophobic-linker fragment and the 5-(4-
hydroxybenylidene)-thiazolidindione head (4), prior to their coupling, was attempted. Such a
way would permit us to introduce diversity at the penultimate step (forward design).
Unfortunately, the ionizable TZD head was found to interfere with the nucleophilic
substitution or Mitsunobu coupling reactions needed for installing the hydrophobic-linker
fragment. Moreover, the easily ionizable functionalities at (4), the phenolic –OH (pKa ~ 9.5)
and amidic –NH (pKa ~ 6.5), represent another challenge that could complicate the
purification step of the products even if the problem of imide reactivity was tackled by
protection. Thus, all designed compounds were prepared via a stepwise addition of
fragments, starting from the hydrophobic trunk ‘left-hand-side’ and proceeding with building
up the molecule, while keeping the TZD head as the last step (backward design) [24,25]. It is
worth noting that the linker was introduced along with hydrophobic trunk or with central
aromatic moiety. The later tactic was more convenient regarding the yield and/or commercial
availability of starting materials.
Synthesis of series MK-3 started with the acid-mediated condensation of the corresponding
substituted benzhydrols (5a-e) with 2-bromoethanol (Scheme 2). Subsequent SN2 with 4-
hydroxybenzaldehyde yielded the ether-aldehyde analogues (7a-e). Finally, Knoevenagel
condensation with 3 followed by catalytic hydrogenation of the olefin moiety using
NaBH4/CoCl2-dimethylglyoxime (Co-DMG) complex provided the target compounds (9a-e)
6

in good yields. The reduction of the benzylidenes was confirmed by the appearance of three
double doublets in range of δ 3.00–5.50 ppm which appeared due to protons present on C-5
of TZD ring and methylene bridge.
CHO
R
R
R
OH
O
O
a)
b)
c)
Br
O
7a-e
6a-e
5a-e
S
S
O
O
R
R
O
O
NH
NH
O
O
O
O
d)
R
a
b
c
d
e
H
8a-e
9a-e
2-Me
4-MeO
4-Cl
4-CF3
Scheme 2. (a) 2-bromoethanol, Conc. H2SO4, toluene, rt-60 °C, 0.5-3 h, 89-95 %; (b) 4-
hydroxybenzaldehyde, K2CO3, HPLC grade ACN, reflux, 3-5 h, 52-73 %; (c) 3, benzoic acid,
piperidine, toluene, reflux, overnight; (d) Co-DMG/DMF, NaBH4, 1 M NaOH aq, THF, 35-40 °C, 24
h, 56-62 %.
Scheme 3 depicts the synthesis of series MK-4 with a four-step sequence. First, the
commercially available 4-hydroxybenzaldehyde (10) was condensed with 2-bromoethanol
through Mitsunobu reaction to give the key intermediate (11). Next, conventional
nucleophilic substitution reaction with the privileged structures (12a-f) using a variety of
deprotonating bases, yielded their congruent aldehydes (13). The R groups implicated within
the scaffolds (12a-f) were used as a handle to introduce structural diversity. Finally, the
obtained aldehydes were converted to their corresponding target molecules (15a-f) in an
analogous fashion to that previously described in Scheme 2. It is worth noting that the
hydrophobic scaffolds (12a-f) were either commercially available or synthesized in-house.
The benzimidazole ring (12d) or its 2-methanol derivative (12e) were constructed using
Phillips procedure through an acid-catalyzed condensation of phenylenediamine with formic
or glycolic acid, respectively [13,26].
7

X
Y
R1
R2
a
b
c
d
e
f
C
C
C
N
N
N
N
N
N
-C-O-
H
H
6-Cl
H
H
H
Ph
H
H
N
-CH20H
H
-CH=CH-
H
→ rt, overnight, 62 %; (b) KOH, dry
Scheme 3. (a) 2-bromoethanol, PPh3, DIAD, dry THF, 0 °C
DMSO, rt-40 °C, 24-72 h, 60-95 %; (c) NaH, dry DMF, 60-80 °C, 48 h, 65 %; (d) Cs2CO3, dry DMF,
rt, 24 h, 55 %; (e) 3, benzoic acid, piperidine, toluene, reflux, overnight. (f) Co-DMG/DMF, NaBH4, 1
M NaOH aq, THF, 35-40 °C, 24 h, 54 % -quantitative.
All synthesized target compounds were tested for their ability to activate PPAR-γ using
luciferase-based genetic reporter assay [27]. In this cell-based transcription assay, proprietary
bioreporter cells were transfected with two genetic elements; the luciferase reporter gene
plasmid and GAL4-PPAR-γ expression plasmid. These cells express hybrid receptors, GAL4-
PPAR-γ chimeric protein, which can bind to the GAL4 binding sites located upstream of the
firefly luciferase reporter gene (Fig. S1, Supplementary data). Activities are reported as fold-
induction as well as EC50. Regarding fold-induction, results obtained by test compounds were
compared to negative control sample of the vehicle DMSO. Additionally, the fold-induction
was also calculated as a percentage of that of rosiglitazone as a measure of compounds’
intrinsic activity (Table 1). Values ≥ five-fold-induction, relative to DMSO, were considered
statistically significant and may warrant further consideration.
8

In series MK-3, modest activities were afforded with an exception of 9c (5.2-fold activation).
The reason for such modest potency might be attributed to the bulky branched hydrophobic
tail. In both series, MK-2 and -3, four atoms separate the large biaryl tail from the central
core. This might push the terminal aromatic group too far downwards into the receptor
hydrophobic pocket. On the other hand, the presence of the methoxy group in 9c might
favour polar interactions with the receptor pocket. Interestingly, the same observation was
also noticed with the methoxy derivative of our reported series MK-2 (phenoxybenzyl-
chemotype). It is worth noting that, introducing a different electron withdrawing H-bonding
group (CF3 in 9e) completely abolished the activity.
To our delight, seven compounds showed future prospection with members of series MK-4
showed more significant PPAR-γ transactivation. The activities afforded by this series were
greater than or equal five-fold-induction with three compounds (15a,c,d) exceeding the 30-
fold-induction border. Interestingly, 15a and 15c showed the full agonist profile with
maximum fold-activation up to 130 %, relative to that of rosiglitazone. As reported, a
hydrophobic tail bearing H-bond acceptors may enhance the PPAR-γ transactivation [28].
This could explain the improved activity of this series, together with suitable size and shape
of the indole privileged scaffold.
Regarding our synthesized compound 15e, its design rational was set to further explore the
benzimidazole motif as being an advantageous indole-bioisostere capable of providing extra-
hydrogen bonding at the tail pocket of both receptors. Moreover, the possibility of different
linking position at the benzimidazole ring, in comparison to the MK-5, would introduce more
spacial-occupation for the 15e overall structure allowing more favoured pocket
accommodation. However, the low activities depicted by 15e suggested a non-tolerable
substitution at C-2, possibly due to steric clashes, the thing being assumed since 15b also
depicted low activity range.
9

Table 1
In vitro data of the human PPAR-γ transactivation assay.a
EC50 (µM)c
Compound
Fold-inductionb
Compared to vehicle (1 fold)
Percentage compared to Rosiglitazone
Vehicle
1
─
─
Rosiglitazone
43.1 1.1
4.0 0.1
1.9 0.3
5.2 0.6
4.6 0.2
IA
100
0.24
ND
ND
ND
ND
ND
3.94
ND
4.51
2.32
ND
ND
9.3 0.2
4.4 0.7
12.1 1.4
10.7 0.4
IA
9a
9b
9c
9d
9e
15a
15b
15c
15d
15e
15f
56.1 2.4
15.1 1.1
36.2 2.4
29.7 2.0
17.1 1.3
5.70 0.5
130.2 5.5
34.9 2.6
83.7 5.5
69.8 4.6
39.5 3.1
13.3 1.2
a Reported in vitro values are the average of two replicates SD, significant at p < 0.05when compared to control.
b Fold-induction values for the prepared compounds were obtained at 30 µM and calculated relative to that of the vehicle (DMSO 0.10%) and the
reference compound (Rosiglitazone; normalized to 100%).
EC50 value represents the effective concentration for 50% response of a given compound's intrinsic maximum response; ND denotes not
c
determined; IA denotes inactive (> 0.01 fold-induction).
10

A calcium flux FLIPRTM assay was implemented to assess the FFAR1 transactivation
capability of our synthesized compounds (Table 2). This functional cell-based assay utilized
CHO-K1/FFA1/Gα15 stable cell line (Fig. S2, Supplementary data) [29]. Comparing both
series, the fused ring scaffold (series MK-4) showed better activities in the low micromolar
range. This could be related to the resembleance between both the most potent FFAR1
ligands (biphenyl-based agonists) and fused ring chemotype regarding linearity.
On the contrary, the benzhydrol series showed modest activity with EC50 of two digit
micromole. The bulkiness of the hydrophobic tail may be the reason for the low activity.
Interestingly, introduction of H-bonding groups, such as OMe in 9c and CF3 in 9e, slightly
improved the activity compared to the unsubstituted benzhydrol (9a). This was in agreement
with what noticed in the PPAR-γ transactivation assay. The improvement of activity with
these polar substituents might be attributed to a favored H-bonding interaction at the FFAR1
binding site.
On general bases, four compounds (9c, 15a, 15c and 15d) showed moderate and balanced
dual activities with EC50 < 20 µM in both PPAR-γ and FFAR1 agonist assays. Remarkably,
compounds 15a and 15d, furnished affinities in the lowmicromolar range on both targets.
Table 2
In vitro data for the human FFAR1 calcium flux FLIPR assaya.
Compound
EC50 (µM)b
Vehicle
─
γ-linolenic acid
9a
0.24
21.34
IA
9b
9c
17.75
IA
9d
11.33
2.19
IA
9e
15a
15b
15c
15d
15e
15f
5.06
4.31
IA
IA
a Reported in vitro values are the average of two replicates SD, significant at p < 0.05.
bEC50 value means the effective concentration for 50 % response of a given compound’s
intrinsic maximum response; IA denotes inactive (> 9.00 M).
11

To obtain further insights into the differential activities of the members of both series, we
examined their virtual binding within both receptors using GOLD Suite V.4.1.2. The co-
crystal structures of hPPAR-γ (PDB code: 3DZY) and hFFAR1 (PDB code: 4PHU) were
employed for this docking workflow. The first crystal form is an intact PPAR-γ/RXR-α
nuclear receptor complex with rosiglitazone, 9-cis-retinoic acid, and NCoA-2, while hFFAR1
is bound to the partial agonist fasiglifam (TAK-875). The ligands polar heads, TZDs,
interacts significantly with the H-bonding triad (H323, H449 and Y473) within the hPPAR-γ
polar arm containing the important AF-2helix. On similar bases, extensive polar network was
predicted between the TZD war-head and both arginine residues (R1835.39 and R2587.35) in
the hFFAR1 hydrophilic pocket. Further stabilization through complex charge networking
with Y913.37, Y2406.51, E172ECL2, S1875.44, N2416.52 and N2446.55 was also noticed.
Beside these polar networks, hydrophobic interactions with the ligands tail parts also
contribute for effective occupancy of both receptors binding sites. However, many reports
have suggested that interactions with such conserved H-bonding rich areas are of much
greater importance for the activities of PPAR-γ and hFFAR1 agonists. These H-bonding
network could stabilize the AF-2 helix in a conformation favoring the binding of co-
activators to PPAR-γ and, consequently, enhancing their recruitment leading to gene
transcription [30-33]. Similarly, polar interactions with the critical arginines breaks the
arginine-glutamate ionic locks at the extracellular surface of FFAR1 allows the receptor to
adopt an active conformation through the outward movement of transmembrane (TM) helices
[34]. It is worth noting that the ligand hydrophobic tails could indirectly affect the activities
through influencing the distances and angles of H-bonding interactions between the TZD
heads and polar residues [32]. The best docking modes were selected by considering both the
GOLD docking scores as well as visual investigation.
Throughout our virtual studies, the predicted poses regarding the ligand’s hydrophobic tail
represents the significant mark for each receptor. In case of hPPAR-γ, the hydrophobic tails
are extended deep into either of the two hydrophobic arms of the Y-shaped orthosteric site
(Fig.3a). Typically, compounds of series MK-3 occupied the larger hydrophobic arm (II)
favoring the least sterically hindered binding mode. They, with their branched benzhydrol-
based tails, manage to make contacts with I326, L228, M329, F226 and P227. Surprisingly,
compound 9c, despite its large tail fragment, could manage to accommodate the tighter arm I
(Fig.3b). Through careful examination, the methoxy substituent was found to be directed into
the exit terminal of arm (I) (ligand entrance site). Such orientation could allow a good chance
12

of forming water-mediated H-bonding. Moreover, the involvement of methoxy-substituted
benzhydrol in π-cation interaction with R280 and π–π stacking with F287, could stabilize this
unexpected fitting. Despite this significant complementary of the tails with either
hydrophobic cavities, only 9c and 9d showed predicted H-bondings with crucial residues plus
an extra bonding with S289, only for 9c. This gives more emphasis on the importance of H-
bonding interactions with the AF-2 helix for PPAR-γ activation.
Moving forward to series MK-4, the energy minimized conformation of compound 15a
within the binding pocket is quite interesting. The indole-based tail guides the TZD head for a
highly-favored interaction with all residues of the H-bonding triad (Fig. S3, Supplementary
data). In addition, an extra bonding with H449 in the AF-2 helix was achieved. Such
preferential settling furnished the highest docking score (Table S1, Supplementary data)
among this series which was interpreted through the significantly improved in vitro result.
Compounds 15c and 15d, the second in rank, were accommodated in the hydrophobic arm
(II) showing π-stacking with F226, H-bonding between the linker oxygen and R288, and
hydrophobic interactions with L228, L333 and P227. Such accommodations drag 15c and
15d deeper into the hydrophobic pocket anchoring the TZD far from the triad with only a
single interaction with Y473 (Fig.3c).
13

Fig 3. Docked poses of synthesized compounds at hPPAR-γ active site (PDB ID: 3DZY); (a) Overlay of MK-4
series (magenta), MK-4 series (blue) and crystallized rosiglitazone (yellow); (b) Overlay of 9c (magenta) and
crystallized rosiglitazone (yellow); (c) Overlay 15d (blue) and crystallized rosiglitazone (yellow). Protein–
ligand interactions are only visualized for crystallized rosiglitazone in (a) while for 9c and 15d in (b) and (c),
respectively. Hydrogen bonds discussed in text are depicted as red dashed-lines. Each helix is labeled in red
roman numerals. Only residues located within 4 Å radius of bound ligand are displayed (green) and labeled with
sequence number.
We then turned our attention to the proposed docking poses in hFFAR1. All our ligands were
predicted to occupy the allosteric binding site rather than the canonical orthosteric site. In a
similar fashion to the crystallized fasiglifam, the proposed docking modes force the
hydrophobic tails to extend out through the space between TM3 and TM4 of the hFFAR1
receptor. Both the hydrophobic and steric interactions along the hydrophobic tails influence
14

the anchoring of the ligands’ polar heads with the arginine-tyrosine polar network. The edge-
on π-π interaction of W174ECL2 or π-stacking of F1424.61 and the respective ligand’s central
aromatic ring, orient the TZD head for a proper H bonding with the crucial arginines.
In the benzhydrol-based series (MK-3), one of the aromatic rings shows good overlapping
with fasiglifam’s terminal phenyl groups, while the other points deeper into the lipid
membrane layer. In addition to these hydrophobic interactions, the terminal methoxy in 9c
shows upward direction forming an H-bond with P803.26 (Fig. 4a). Actually, H-bonding
groups on the hydrophobic tails play a crucial role in ligand binding in this series. This is
obvious when comparing the activities of 9c and 9e to those of 9b and 9d. On the contrary,
steric clashes with L1584.77 and P803.26 with the o-methyl of the inactive 9b caused a marked
deterioration of the activity.
As expected, series MK-4 show extensive complex charge networks with the critical arginine
residues of FFAR1 (Table S2, Supplementary data). This was in advent for our later series
keeping their EC50 down to one digit micromolar potencies. Compounds 15a, 15c and 15d,
the front-runner representatives of series MK-4, showed great overlapping with the
crystallized fasiglifam. Their leading compound 15a displayed favorable π-π interaction with
F1424.61 and W174ECL2. Moreover, its hydrophobic tail expressed interaction with L1354.54
,
L1384.57, L1584.77, P803.26, F1424.61, V813.27 and V843.30, similar to fasiglifam’s biphenyl
scaffold. These favored tail housing, furnish minimal clashes with TM3 and TM4 residues. In
addition, it ensures the desired TZD anchoring guided by perfect π-π interaction with F873.33
(Fig.4b). On the other hand, the large and rigid phenyl substituent in the inactive compound
15b is particularly prone to steric clashes with V813.27, P803.26 and F1424.61. In case of 15f, the
longer linker together with structural rigidity of its naphthyl moiety shift the TZD little far
from R1835.39 while force the central aromaticity to be in a vertical plane disrupting its π-π
stacking with F1424.61 and lose the H-bonding with Y913.37 (Fig. S4a, Supplementary data).
Interestingly, compound 15e showed unfamiliar orientation in which the TZD head was
inverted bringing its 2ry nitrogen too close to R2587.35 side chain nitrogen prone to steric
clashes rather than H-bonding (Fig. S4b, Supplementary data). Moreover, such anchoring
prevents the desirable H-bonding interaction with Y913.37. This binding mode could be forced
by the free hydroxyl group on benzimidazole scaffold. It is believed that etherification of
such hydroxy group with alkyl sulphone analogues as in fasiglifam, would bring the TZD to a
more favored orientation with the sulphone directed upwards [35].
15

Fig. 4. Docked poses of at hFFAR1 active site (PDB code: 4PHU); (a) Overlay 9c (magenta) and crystallized
TAK-875 (yellow); (b) Overlay of 15a (blue) and crystallized TAK-875 (yellow). Protein–ligand interactions
are only visualized for 15a and 9c in (a) and (b), respectively. Hydrogen bonds discussed in text are depicted as
red dashed-lines. Each transmembrane helix is labeled in red roman numerals. Only residues located within 4 Å
radius of bound ligand are displayed (green) and labeled with sequence number and Ballesteros−Weinstein
notations [36] where the later are given as superscripts.
The encouraging in vitro agonistic activities were anticipated to translate well into an in
vivo efficacy. We decided to scrutinize compounds 9a, 9c, 15a, and 15c using in vivo blood
glucose and plasma triglycerides (TG) testing. Rosiglitazone, a well-recognized PPAR-γ full
agonist, was used as our positive control for its capability to lower both the circulating
glucose and TG levels [37]. On the other hand, selection of these four compounds was
rationalized by their high in vitro potencies and diverse chemical structures.
For our test subjects, we used established high fat diet–induced obese (eDIO) male mices
T2DM models.The establishment of eDIO mice was accomplished with a 60 % high fat diet
ad libitum over a 14 week period (initiated at 6 weeks of age) [38]. Compared with regularly-
dieted mice, the eDIO ones weigh significantly heavier (51 0.9 vs. 25 0.7 g; n = 5; p <
0.05) and show glucose intolerance manifestation (Random blood glucose levels: 131 0.9
vs. 87 0.8 mg/dL; n = 5; p < 0.05). Our test compounds, rosiglitazone, and the vehicle
16

(DMSO, as a negative control) were intravenousely dosed through tail at 10 mg/kg/day for 14
days run. As depicted in Fig. 5, both indole-based compounds (15a and 15c) presented much
greater euglycemic capabilities than series MK-3 compounds (9a and 9c). This could be
attributed to the much greater transactivation of the PPAR-γ receptor furnished by these
fused-ring derivatives. It is worth noting that compound 15c showed a 6-fold reduction in
blood glucose levels compared that of rosiglitazone.
(a)
(b)
Rosiglitazone
Rosiglitazone
9a
9a
9c
9c
15a
15c
15a
15c
1% 0% -1% -2% -3% -4% -5% -6% -7%
Hundreds
14% 11% 8% 5% 2% -1% -4% -7%
Hundreds
% Reduction of blood glucose
% Reduction of plasma TG
Fig. 5. Comparison of percent reduction in (a) random blood glucose and (b) plasma TG in
eDIO male mice, relative to rosiglitazone, using 10 mg/kg/day dosing for 14 days.
Interestingly, both 9c and 15c offered statistically significant reduction of 5.5- and 3.5-fold,
respectively, relative to rosiglitazone, in the TG plasma levels. These augmented potencies
might be ascribed to better pharmacokinetics profiles when compared to their respective
unsubstituted series members (9a and 15a). The above in vivo results also reflect the
excellent anti-hyperglycemic/hyperlipidemic profiles of 15c that came in accordance with the
in vitro activities.
Seeking for better credibility, we further investigate the dual-agonistic activity of our
promising candidate (15c) against both the native insulinagogue Glimepride, a sulphonylurea
derivative, and the insulin sensitizer Rosiglitazone, a PPAR-γ agonist. Throughout this
second in vivo assay, we adopted the same previous (eDIO) mice model where we have
evaluated the hypoglycemic/hypolipidemic activities via assessing the random blood glucose
and plasma TG levels, as well as, the mice body weights. As depicted in Figure 6, lower
blood glucose level is noticed with 15c compared to Rosiglitazone and Glimepride,
17

individually or in combination. Notably, the body weight reductions, in case of 15c, was
higher than that of Rosiglitazone but lower than the Glipepiride's. Concerning plasma TG,
satisfactory reduced levels were detected for our candidate (Fig. S5, Supplementary data).
The overall results, form the two in vivo investigations, suggest that 15c is a promising lead
compound that could pave the way for further development of the first clinically useful
PPAR-γ/FFAR1 dual agonists.
200
(a)
180
160
140
Rosiglitazone
120
100
80
60
40
20
0
Glimepiride
15c
Combination
Control
0
7
14
21
Time (Days)
40
35
30
25
20
15
10
(b)
Rosiglitazone
Glimepiride
15c
Combination
Control
0
7
14
21
Time (Days)
Fig. 6. Comparison of (a) random blood glucose levels and (b) mice body weight, for 15c,
rosiglitazone, glimepiride, combination of rosiglitazone\glimepiride, and DMSO (as negative
control), using a total daily dosing of 10 mg/kg/day dose for 14 days.
18

In our continued quest for dual PPAR-γ/FFAR1 agonists, we herein introduced the
benzhydrol- and indole-tailed thiazolidindiones chemotypes. Both series were rationally
designed as better successors of our recently published biphenyl-, phenyloxybenzyl-, and
benzimidazoles. Down to low micromolar potencies, compounds 9c, 15a, 15c and 15d
exhibited moderate and balanced dual activities on both targets. Among them, Compound 15c
showed better anti-hyperglycemic and anti-hyperlipidemic activities in fat-fed animal models
compared to rosiglitazone. This was rationally explained using our molecular docking
simulation studies. Those qualities made compound 15c a promising candidate for more
advanced preclinical studies.
Acknowledgments
This research was entirely supported by the Science and Technology Development Fund
(STDF), Egypt (project grant # 4244) awarded to the Principal Investigator Dr. Mohamed A.
Helal. The authors are grateful to Prof. Dr. Soad H. Abou El-Ela, Dean of the Faculty of
Pharmacy and Pharmaceutical Industries, Sinai University, El-Arish, Egypt; for providing the
access to the HPLC/diiodo array instrumentation used in the analysis of our compounds’
purity. The authors would also like to thank Dr. Mohamed S. Elgawish, Graduate School of
Biomedical Sciences, Course of Pharmaceutical Sciences, Nagasaki University, 1-14
Bunkyo-machi, Nagasaki 852-8521, Japan, for performing the mass measurements.
References
[1] Hyae GC. Arch. Pharm. Res. 2013;36:145.
[2] Diabetes factsheet from WHO providing key facts and information on types of diabetes,
symptoms, common consequences, economic impact, diagnosis and treatment, WHO
response. Updated March 2013. WHO | Diabetes. WHO.
[3] Home P, Mant J, Diaz J, Turner C. BMJ. 2008;336:1306.
[4] Mizuno CS, Chittiboyina AG, Kurtz TW, Pershadsingh HA, Avery MA. Curr. Med.
Chem. 2008;15:61.
[5] Lemay DG, Hwang DH. J. Lipid Res.2006;47:1583.
[6] Whalen K, Kelley SS. Insulinand Oral Hypoglycemic Drugs. In: Harvey RC, Champe PC
eds. Lippincott's Illustrated Reviews: Pharmacology, 4th ed. Philadelphia, Baltimore:
Lippincott Williams & Wilkins; 2009:292.
19

[7] Ahrén B. Nat. Rev. Drug Discovery. 2009;8:369.
[8] Briscoe CP, Tadayyon M, Andrews JL, Benson WG, Chambers JK, Eilert MM, Ellis
C, Elshourbagy NA, Goetz AS, Minnick DT, Murdock PR, Sauls HR Jr, Shabon
U, Spinage LD, Strum JC, Szekeres PG, Tan KB, Way JM, Ignar DM, Wilson S, Muir AI.
J. Biol. Chem.2003;278:11303.
[9] Edfalk S, Steneberg P, Edlund H. Diabetes. 2008;57:2280.
[10] Nakamoto K. Brain Res. 2012;1432:74.
[11] Wellendorph P, Johansen LD, Brauner-Osborne H. Mol. Pharmacol. 2009;76:453.
[12] A Salehi A, Flodgren E, Nilsson NE, Jimenez-Feltstrom J, Miyazaki J, Owman C, Olde
B. Cell Tissue Res. 2005;322:207.
[13] Darwish KM, Salama I, Mostafa S, Gomaa MS, Helal MA. Eur. J. Med. Chem.
2016;109:157.
[14] Morphy R, Kay C, Rankovic Z. Drug Discov. Today. 2004;9:641.
[15] Helal MA, Darwish KM, Hammad MA. Bioorg. Med. Chem. Lett. 2014;24:5330.
[16] Yasmin S, Jayaprakash V. Eur. J. Med. Chem. 2017;126:879.
[17] Hara T, Hirasawa A, Ichimura A, Kimura I, Tsujimoto G. J. Pharm. Sci. 2011;100:3594.
[18] Tikhonova IG, Sum CS, Neumann S, Engel S, Raaka BM, Costanzi S, Gershengorn MC.
J. Med. Chem. 2008;51:625.
[19] Walker CG, Goff L, Bluck LJ, Griffin BA, Jebb SA, Lovegrove JA, Sanders TA, Frost
GS. PLoS One. 2011;6:19146.
[20] Chakrabarti R, Vikramadithyan RK, Dileepkumar T, Kumar KB, Kumar MP, Misra
P, Rao PB, Lohray VB, Lohray BB, Rajagopalan R. Arzneimittelforschung. 1999;
49:905.
[21] Izumi T, Tsuruta F, Ishizuka T, Nakamura K, Kothuma M, Takahashi M. J. Pharm. Sci.
2013;102:3174.
[22] Lohray BB, Bhushan V, Rao BP, Madhavan GR, Murali N, Rao KN, Reddy AK, Rajesh
BM, Reddy PG, Chakrabarti R, Vikramadithyan RK, Rajagopalan R, Mamidi RNVS,
Jajoo HK, Subramaniamr S. J. Med. Chem. 1998;41:1619.
[23] Prashantha Kumar BR, Karvekar MD, Adhikary L, Nanjan MJ, Suresh B. Heterocycl.
Chem. 2006;43:897.
[24] Sugimoto O, Arakaki T, Kamio H, Tanji K. Chem. Commun. (Camb). 2014;50:7314.
20

[25] Ramachary DB, Anebouselvy K, Chowdari NS, Barbas CF. J. Org. Chem.
2004;69:5838.
[26] Elshihawy H, Helal MA, Said M, Hammad MA. Bioorg. Med. Chem. 2014;22:550.
[27] Kojo H, Fukagawa M, Tajima K, Suzuki A, Fujimura T, Aramori I, Hayashi K,
Nishimura S. J. Pharmacol. Sci. 2003;93:347.
[28] Kumar BR, Nanjan MJ. Bioorg. Med. Chem. Lett. 2010;20:1953.
[29] Zhang Y, Kowal D, Kramer A, Dunlop J. J. Biomol. Screen. 2003;8:571.
[30] Nolte RT, Wisely GB, Westin S, Cobb JE, Lambert MH, Kurokawa R, Rosenfeld
MG, Willson TM, Glass CK, Milburn MV. Nature. 1998;395:137.
[31] Pissios P, Tzameli I, Kushner P, Moore DD. Mol. Cell. 2000;6:245.
[32] Sheu SH, Kaya T, Waxman DJ, Vajda S. Biochemistry. 2005;44:1193.
[33] Kallenberger BC, Love JD, Chatterjee VK, Schwabe JW. Nat. Struct. Biol. 2003;10:136.
[34] Costanzi S. Curr. Opin. Struct. Biol. 2013;23:185.
[35] Krasavin M, Lukin A, Bagnyukova D, Zhurilo N, Zahanich I, Zozulya S, Ihalainen J,
Forsberg MM, Lehtonen M, Rautio J, Moore D, Tikhonova IG. Bioorg. Med. Chem.
2016;24:5481.
[36] Ballesteros J, Weinstein H. Biophys. J. 1995;68:A446.
[37] Tamori Y, Masugi J, Nishino N, Kasuga M. Diabetes. 2002;51:2045.
[38] Tan CP, Feng Y, Zhou Y-P, Eiermann GJ, Petrov A, Zhou C, Lin S, Salituro G, Meinke
P, Mosley R, Akiyama TE, Einstein M, Kumar S, Berger JP, Mills SG, Thornberry NA,
Yang L, Howard AD. Diabetes. 2008;57:2211.
21

Highlights
•
•
T2DM is a progressive cluster of metabolic disorders, with a global health burden.
Continuing our previous work, we report benzhydrols and indoles on FFAR1 and
PPARγ
•
•
•
15a, 15c, and 15d, depicted affinities in low micromolar range on both targets.
In vivo study, 15c of superior hypoglycemic/lipidemic activities to rosiglitazone.
These compounds could develop unique insulin sensitizers as well as secretagogues.
22