Design, synthesis, and biological evaluation of novel thiazolidinediones as PPAR3/FFAR1 dual agonists

Article abstract of DOI:10.1016/j.ejmech.2015.12.049

Diabetes mellitus is a chronic metabolic disorder that affects more than 180 million people worldwide. Peroxisome proliferator activated receptors (PPARs) are a group of nuclear receptors that have been targeted by the thiazolidinedione (TZD) class of compounds for the management of type II diabetes. PPAR3 is known to regulate adipogenesis and glucose metabolism. Another emerging target for the design of antidiabetic agents is the free fatty acid receptor 1 (FFAR1), previously known as GPR40. Agonists of this receptor were found to enhance insulin secretion in diabetic patients. It has been reported that some thiazolidinediones (TZDs) activate FFAR1 with micromolar potency. In this study, based on docking studies into the crystal structure of PPAR3 and a homology model of FFAR1, nineteen compounds were designed, synthesized, and biologically tested for agonistic activity on both receptors. Nine compounds showed promising dual activity, with two compounds, 11a and 5b, having affinities in the low micromolar range on both targets. These molecules represent the first antidiabetic agents that could act as insulin sensitizers as well as insulin secretagogues.

Full text of DOI:10.1016/j.ejmech.2015.12.049

Accepted Manuscript
Design, Synthesis, and Biological Evaluation of Novel Thiazolidinediones as PPARγ/
FFAR1 Dual Agonists
Khalid M. Darwish, Ismail Salama, Samia M. Mostafa, Mohamed S. Gomaa,
Mohamed A. Helal
PII:
S0223-5234(15)30435-9
10.1016/j.ejmech.2015.12.049
EJMECH 8289
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 19 August 2015
Revised Date: 27 December 2015
Accepted Date: 28 December 2015
Please cite this article as: K.M. Darwish, I. Salama, S.M. Mostafa, M.S. Gomaa, M.A. Helal, Design,
Synthesis, and Biological Evaluation of Novel Thiazolidinediones as PPARγ/FFAR1 Dual Agonists,
European Journal of Medicinal Chemistry (2016), doi: 10.1016/j.ejmech.2015.12.049.
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.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Design, Synthesis, and Biological Evaluation of Novel Thiazolidinediones as
PPARγ/FFAR1 Dual Agonists
a
a
a
a
Khalid M. Darwish , Ismail Salama , Samia M. Mostafa , Mohamed S. Gomaa , and Mohamed
a,*
A. Helal
a
Medicinal Chemistry Department, Faculty of Pharmacy, Suez Canal University, Ismailia
41522, Egypt.
Corresponding Author: Mohamed A. Helal
Tel: +201201122213
E-mail address: mohamed_hilal@pharm.suez.edu.eg
Abstract
Diabetes mellitus is a chronic metabolic disorder that affects more than 180 million people
worldwide. Peroxisome proliferator activated receptors (PPARs) are a group of nuclear
receptors that have been targeted by the thiazolidinedione (TZD) class of compounds for the
management of type II diabetes. PPARγ is known to regulate adipogenesis and glucose
metabolism. Another emerging target for the design of antidiabetic agents is the free fatty acid
receptor 1 (FFAR1), previously known as GPR40. Agonists of this receptor were found to
enhance insulin secretion in diabetic patients. It has been reported that some thiazolidinediones
(TZDs) activate FFAR1 with micromolar potency. In this study, based on docking studies into
the crystal structure of PPARγ and a homology model of FFAR1, nineteen compounds were
designed, synthesized, and biologically tested for agonistic activity on both receptors. Nine
compounds showed promising dual activity, with two compounds, 11a and 5b, having affinities
in the low micromolar range on both targets. These molecules represent the first antidiabetic
agents that could act as insulin sensitizers as well as insulin secretagogues.
Key words: Diabetes, PPARγ, FFAR1, Thiazolidinedione, Docking.
1

ACCEPTED MANUSCRIPT
1. Introduction
Diabetes represents a major health concern, especially in developing countries. According
to the World Health Organization (WHO), more than 180 million people worldwide have
diabetes and this number is expected to reach 366 million in 2030. Half of diabetes-related
death cases occur in people under the age of 70 and this number is expected to increase by
more than 50% in the next 10 years.[1] Moreover, the number of diabetic patients is
continuously increasing due to several factors including population growth, increased life
expectancy, increased rates of obesity, and lack of physical activity.[2]
The most widely used antidiabetic medications are insulin secretagogues and insulin
sensitizers. Examples of insulin secretagogues include sulfonylureas and meglitinides, while
metformin and thiazolidinediones (TZDs) are insulin sensitizers. TZDs were introduced in
the late 1990s as the first agents that control blood glucose level by acting on Peroxisome
proliferator-activated receptors (PPARs).[3-5] These represent a group of nuclear receptors
that control cellular metabolism through the modulation of gene expression.[6] There are
three distinct subtypes of PPARs: PPARα, PPARδ, and PPARγ. Activation of PPARγ has
been shown to regulate glucose homeostasis, cellular differentiation, apoptosis, and
inflammatory responses.[7] Over the past few years, there has been an influx of new
antidiabetic agents acting on a variety of cellular targets. Free fatty acid receptors (FFARs)
are a family of G-protein coupled receptors (GPCRs) that act as fatty acid sensors and play a
crucial role in glucose homeostasis. Recent studies have demonstrated that both dietary fatty
acids and synthetic agonists can stimulate glucose-dependent insulin secretion by acting on
FFAR1 highly expressed in pancreatic β-cells.[8, 9]
Several studies reported that modulation of PPARγ using TZDs was not efficient at
controlling diabetes in some patients. This led to the use of combination therapies containing
®
both insulin secreting and insulin sensitizing agents, such as Amaryl M (glimepiride and
2

ACCEPTED MANUSCRIPT
®
metformin) and glucovance (glibenclamide and metformin).[10] These findings suggest the
need for co-administration of insulin sensitizers and insulin secretagogues for the
management of diabetes in some situations.
Interestingly, it has been reported that some TZDs activate FFAR1 expressed in human
HeLa cells with micromolar potency.[11, 12] In 2007, Owman and co-workers exploited the
TZD scaffold for the design of FFAR1 ligands by combining fatty acids substructures with
TZD heads. Two of the synthesized compounds in this study showed activity in the
micromolar range (compounds A and B, Figure 1).[13] In a recent study, 2,000 TZDs from
the Merck compound collection were screened using the FLIPR assay in human GPR40-CHO
cells. This led to the discovery of a partial agonist for GPR40 (compound C, Figure 1)
possessing an EC of 0.5 ꢀM. The initial GPR40 activity of this compound was improved
5
0
after subsequent optimization. However, it was inactive in binding assays against all human
PPARα, -δ, and -γ isoforms.[14]
Figure 1: TZD-fatty acid hybrids
The objective of this study is to design and synthesize drug-like molecules with agonistic
activity on both receptors; PPARγ and FFAR1. These agents would act as insulin sensitizers
and insulin secretagogues through their action on PPARγ and FFAR1, respectively. The design
of drugs with dual mode of action is a valid approach. Aleglitazar, a PPAR modulator with
affinity for both PPARα and PPARγ, is currently in phase III clinical trials.[15] Similarly,
Asenapine, a dual antagonist of dopamine D and serotonin 5-HT receptors, was launched in
2
2
2009 by Schering-Plough for the acute treatment of schizophrenia.[16] It is worth noting that
3

ACCEPTED MANUSCRIPT
the endogenous ligands of both PPARγ and FFAR1 are fatty acids. Therefore, simultaneous
activation of both PPARγ and FFAR1 should be well tolerated by the body, simulating the
metabolic response after a fatty meal. Moreover, the concept of increasing insulin sensitivity
and insulin secretion simultaneously has been successfully implemented using various
marketed drug combinations as mentioned above.
2
2
. Results and discussion
.1. Compounds design
The basis for compound design was the similarity in structural requirements of the
agonists of these two receptors. A careful literature survey revealed an evident similarity
between the ligands reported for FFAR1 and those active on PPARγ.[17-19] A typical
FFAR1 agonist consists of an acidic head attached to an aromatic scaffold, a heteroalkyl
linker, and a hydrophobic tail. TZDs have very similar pharmacophoric features and are valid
candidates for drug design because of their well-studied pharmacokinetics.[20-22]
Representative examples of both FFAR1 agonists and PPARγ agonists are depicted in Figure
2
. The design strategy was to combine the common 5-benzyl-thiazolidinedione head from
TZDs with diverse hydrophobic fragments (tails). In a previous study, we illustrated through
homology modeling and molecular dynamics simulations that the thiazolidinedione ring is
capable of binding to the critical polar residues within both PPAR-γ and FFAR1 binding
sites.[23]
Figure 2: Representative examples of FFAR1 agonists and PPARγ agonists
4

ACCEPTED MANUSCRIPT
Selection of the appropriate hydrophobic substructures was based on the “privileged
structures” approach, which was introduced in the late 1980s by Evans and co-workers.[24]
These structures can be defined as molecular scaffolds that bind with high affinity to multiple
receptor families. To prove this concept, Bemis and Murcko performed an exhaustive
analysis of all known drugs in the CMC database. This study revealed that the 5120
compounds in the CMC database contain 1179 different frameworks; however, only 32 (3%)
of these frameworks account for 50% of all drugs.[25] An interesting example of the useful
application of the privileged structures concept was reported in 1998. The 2-arylindole
privileged molecular fragments were utilized as a starting point for the design of ligands for
five different receptors; serotonin, melanocortin, chemokine, NPY, and neurokinin.[26] This
phenomenon could be attributed to the presence of conserved regions deep in the binding site
of most Class A GPCRs that are predominantly hydrophobic and recognized by the
privileged structures. The previous findings suggest that empirically observed privileged
structures are viable starting points for drug design, especially when targeting multiple
receptors. Recently, there has been a large number of publications describing molecular
frameworks considered as privileged structures for GPCRs.[27-32] In this study, the criteria
for the selection of suitable privileged structures were; 1) similarity to previously reported
FFAR1 and/or PPARγ agonists, 2) similarity to reported agonists for the purine receptor
(which is the closest neighbor to FFAR1), 3) high frequency of occurrence among GPCR
ligands.[33] Depending on these criteria, 17 fragments were selected (Figure S1,
Supplementary Content) and utilized for the design of 17 initial scaffolds by combining these
privileged structures with the 5-benzyl-thiazolidinedione head of TZDs through an ethoxy or
methoxy linkers (Figure 3).
Figure 3: Design strategy of the dual PPARγ/FFAR1 agonists.
The 17 initial compounds were sketched, minimized, and docked into the ligand binding
sites of FFAR1 and PPARγ. Docking studies were carried out using the co-crystal structure
of PPARγ with rosiglitazone (PDB code: 3DZY) and the homology model of FFAR1
developed in our laboratory.[23] Careful examination of the binding sites of both receptors
revealed an obvious similarity. Interestingly, both receptors had a hydrogen bonding triad at
one end of the binding site and a Y-shaped hydrophobic pocket at the other end. This could
explain, at least in part, the similarity between the agonists of FFAR1 and PPARγ. The
5

ACCEPTED MANUSCRIPT
hydrogen bonding triad is formed by Arg183, Arg258, and Asn244 in FFAR1; and by
His323, His449, and Tyr473 in PPARγ. Of the 17 docked compounds, 5 compounds showed
satisfactory binding with both receptors (Figure 4). As expected, the thiazolidinedione ring
formed hydrogen bonds with at least 2 of the 3 amino acid residues at the hydrogen bonding
triad. In addition, the hydrophobic tail (privileged structure) showed favorable hydrophobic
and π-π stacking interactions with the Y-shaped hydrophobic pocket of both FFAR1 and
PPARγ (Figure S2 and S3, Supplementary Content). It is worth noting that the biphenyl
group in a meta arrangement (series 1) made better hydrophobic contact with FFAR
compared to its para counterpart. In addition, in terms of stereochemistry, our docking studies
showed that the (S)-enantiomer of scaffold 2 ranked higher than its (R)-isomer. In this study,
we decided to start a lead optimization campaign for three of these scaffolds; 1, because of
the reported importance of the biphenyl moiety for the FFAR binding; 2, as a representative
of the “branched biphenyl” chemotype; and 5, as a representative for fused heterocycles.[34,
3
5] We selected 5 due to its similarity with rivoglitazone, a benzimidazole-based glitazone,
which was reported in a recent study to be 3-times more active than rosiglitazone in a PPAR-
γ transfection assay.[36]
S
O
S
O
S
O
N
H
O
O
O
O
N
H
O
O
N
H
(
S)
O
O
Series 1
Series 2
Series 3
O
NH
O
S
S
O
N
O
N
N
H
O
O
N
H
Series 4
Series 5
Figure 4: Scaffolds with the highest docking score in both FFAR1 and PPARγ.
2.2. Chemistry
The key thiazolidinedione head group was prepared by refluxing chloroacetic acid and
thiourea in water for 12 h to yield pure white crystals of thiazolidine-2,4-dione after cooling
Scheme 1).[37] Subsequently, we attempted to synthesize the 5-(4-hydroxybenzylidene)-
(
thiazolidinedione moiety first and attach the linker and the hydrophobic tail later on. This
would allow us to introduce diversity at the last step of the synthesis. Unfortunately, the imide
6

ACCEPTED MANUSCRIPT
group of the TZD head was found to interfere with the Mitsunobu coupling or nucleophilic
substitution reactions required for installing the linker. In addition, the polarity of the 5-(4-
hydroxybenzylidene)-thiazolidinedione moiety could complicate the purification step even if
the problem of imide reactivity was tackled by protection/deprotection. Using this “forward”
approach, only benzyl derivative 5 could be prepared in satisfactory yield (Scheme 1).
Therefore, we decided to prepare all the designed compounds starting from the hydrophobic
privileged structure, followed by the linker and 4-hydroxy benzaldehyde. Finally, Knoevenagel
condensation with the TZD ring and reduction would furnish the target compounds (backward
design).[38, 39]
Synthesis of the biphenyl-based series 1 was achieved using Suzuki-Miyaura cross-coupling
(Scheme 2).[40] First, commercially available 3-bromobenzyl alcohol (6a) and 4-hydroxy
benzaldehyde were coupled under Mitsunobu conditions to give the key
benzyloxybenzaldehyde intermediate (7a). Second, Suzuki-coupling of this intermediate with
the appropriate arylboronic acids provided the biaryl analogues (8a-d). Finally, Knoevenagel
condensation with 2,4-thiazolidinedione followed by catalytic reduction using sodium
borohydride/cobalt chloride-dimethylglyoxime (Co-DMG) complex provided the desired
compounds in good yields.[41] In addition, we decided to exploit the size and lipophilicity of
the bromine atom by synthesizing 3- and 4-bromophenyl derivatives instead of biphenyls.
Compounds 11a and 11b were prepared using the appropriate bromobenzyl alcohols and the
same procedure mentioned above but without the Suzuki coupling step (Scheme 2).
For the preparation of 2, an asymmetric synthetic route was adopted by reducing the
commercially available 3-chloropropiophenone 12 under Corey-Bakshi-Shibata (CBS)
conditions using (S)-oxazaborolidine to furnish the enantiomerically pure alcohol (R)-(+)-3-
chloro-1-phenyl-1-propanol.[42, 43] Then 4-hydroxybenzaldehyde was installed using
Mitsunobu coupling to give the key intermediate alcohol 14. This alcohol underwent
7

ACCEPTED MANUSCRIPT
Mitsunobu coupling with the appropriate substituted phenols leading to compounds 15a-g with
inversion to the desired (S)-configuration.[44-46] These intermediates were condensed under
Knoevenagel conditions with 2,4-thiazolidinedione to give benzylidenes 16a-g. Reduction to
the saturated 2,4-thiazolidinediones was accomplished using sodium borohydride and catalytic
Co/DMG (Scheme 3). It is worth noting that the Y-shaped structures of these series could help
the compounds to fit into the two hydrophobic arms of PPARγ and FFAR1, while maintaining
a suitable size that is not too bulky.[9, 47, 48] Biological evaluation of series 1 and 2 enabled
us to compare between straight and branched hydrophobic moieties and also gave insights into
the binding preferences of FFAR1.
5
derivatives were prepared according to scheme 4. The benzimidazole ring was
constructed using Phillips procedure by the condensation of phenylenediamine with glycolic
acid in acidic medium to give compound 18.[49] Alkylation of the NH of this intermediate with
diverse alkyl halides was accomplished using sodium hydride in DMF. Then the resulting
alcohol derivatives 19a-e were coupled with 4-hydroxybenzaldehyde under Mitsunobu
conditions to give aldehydes 20a-e. Finally, condensation under Knoevenagel conditions with
2
,4-thiazolidinedione followed by reduction using NaHB and catalytic Co/DMG furnished the
4
target compounds 5a-e.
2
2
.3. Biological evaluation
.3.1. Human PPAR-γ Transactivation Assay
All the synthesized target compounds were tested for their ability to activate PPAR-γ using
luciferase-based genetic reporter assay.[50] In this assay, mammalian cells were transfected
with the ligand binding domain of PPAR-γ fused to a heterogonous DNA binding domain
linked to the luciferase gene (Figure S4, Supplementary Content). The fold activation caused
by test compounds was compared to a negative control sample of the vehicle DMSO. In
addition, as a measure of intrinsic activity, the test compounds fold activation was also
8

ACCEPTED MANUSCRIPT
calculated as a percentage of that of rosiglitazone (Table 1). Values ≥ 5-fold activation were
considered statistically significant and may warrant further consideration.
As discussed above, this study focuses only on the modification of the lipophilic tail part
because it is the moiety responsible for the unique pharmacodynamics of each glitazone. The
bromophenyl and biphenyl derivatives caused moderate PPAR-γ transactivation activity with
five compounds showing more than 5-fold activation. It was noticed that the biphenyl
derivatives, 1a and 1b, were less active than the bromophenyl derivatives, 11a and 11b. This
can be attributed to the rigidity of the biphenyl nucleus. Adding more bulk and rigidity using
CF group completely abolished the activity (1d). Unfortunately, series MK-2 had modest
3
activity with only compound MK-2d showing a 5-fold activation, perhaps due to the presence
of the methoxy hydrogen bonding group which is able to make H bonding as well as
hydrophobic contact with the binding pocket. In this series, 4 atoms separate the bulky biaryl
tail from the core. This may push the terminal phenoxy group too far downwards into the
receptor hydrophobic pocket. To our delight, four compounds of series 5 showed high PPAR-γ
transactivation. The allyl derivative, 5c, had higher intrinsic activity compared to rosiglitazone
with 55-fold activation and an EC of 4.95 µM. Contrary to the 2 series, bulky substituents on
5
0
the 5 scaffold, like the benzyl group in 5a, were well tolerated by the receptor. It was reported
that the presence of an H-bond acceptor in the hydrophobic tail may enhance the activity
towards PPAR-γ.[51] This could be the reason for the improved activity of this series, together
with the suitable size and shape of the benzimidazole privileged structure.
2
.3.2. FFAR1 Calcium flux FLIPR Assay:
The target compounds were also tested for FFAR1 activation using calcium flux FLIPR
assay (Table 2).[52] This study utilized a CHO-K1/FFA1/Gα15 stable cell line and a
fluorometric imaging plate reader system (Figure S5, Supplementary Content). As expected,
series 1 showed the highest activity towards FFAR1 in accordance with the reported biphenyl-
9

ACCEPTED MANUSCRIPT
based FFAR1 agonists.[34, 35] Also, comparing m-Br and p-Br derivatives 11a and 11b, the
meta arrangement was proved superior. 1c is the most active compound in this series perhaps
due to restricted rotation imparted by the two methyl groups in the ortho positions. Introduction
of an H-bonding group, CF in 1d, did not significantly improve the activity. Series MK-2, the
3
least active on PPAR-γ, showed moderate activity against FFAR1. The bulky α-phenoxy
benzyl tail of this series seems to be more complementary to the FFAR1 hydrophobic pocket.
Furthermore, increasing the steric bulk on ortho and/or para positions of the phenoxy group
enhanced the activity as in compound 2g. Again, the H-bonding group had little impact on the
activity (compound 2e). Fortunately, series 5, the most active on PPAR-γ, showed acceptable
activity on FFAR1 with the isobutyl derivative, 5b, being the most active. Increasing the bulk
of the substituent on the benzimidazole ring was well tolerated as the benzyl derivative, 5a,
showed good activity (4.11 µM). However, the cyclopentyl derivative was inactive perhaps due
to its direct attachment to the benzimidazole scaffold that might hinder it from adapting the
proper conformation.
2.4. Computational study
To examine the mode of binding and understand the structure-activity relationship of the
synthesized ligands, all the compounds were docked into the crystal structures of PPARγ and
FFAR1. To date, 138 crystal structures of human PPARγ have been reported and deposited into
the Brookhaven protein data bank.[53] For the docking study, we used the crystal structure
3DZY because it is bound with rosiglitazone which possesses a similar pharmacophore to our
compounds.[54] The Y-shaped cavity of PPARγ has two hydrophobic arms. The tighter and
more hydrophobic arm I, which binds rosiglitazone, is formed by Met348, Ile281, Ile341,
Leu228, Leu330, Leu353, and Val339. Arm II, lined with Ile326, Leu228, Leu330, Leu333
Phe226 and Glu295, is larger and relatively more hydrophilic. The TZD head of glitazones
binds to the H bonding triad in the third arm which contains the important AF-2 helix. These
1
0

ACCEPTED MANUSCRIPT
polar contacts stabilize the conformation of AF-2 that allows the binding of co-activator protein
leading eventually to gene transcription.[55]
Our docking simulation study showed that series 1, similar to rosiglitazone, binds to the
more hydrophobic arm I. This might be due to the high lipophilicity of the biphenyl privileged
structure which makes contacts with Ile281, Leu330, Val339, Ile341, Met348 and Leu353.
Figure 5a shows the proposed binding mode of 1a into the crystal structure of PPARγ. In
addition to the lipophilic biphenyl moiety, central phenoxy group makes favourable contacts
with Leu330 and Met364. However, the rigid nature of this series does not allow the TZD head,
on the other end, to form the ideal H bonding network. Concerning series 2, the docking poses
of compound 2d within PPARγ completely fills the lower arm II. Its terminal phenyl groups
could efficiently make hydrophobic contacts with Phe226, Leu228, Ala292, and Met329. Its
methoxy group is located close to both Glu295 and Glu343 with a good chance of forming
water-mediated H bonds (Figure 5b). The overall orientation of this compound is similar to
that of rosiglitazone but a bit lower to fill arm II. Unfortunately, the polar head of 2d made H
bond with His323 only and failed to interact with the other H bonding residues. In addition, the
greatly reduced potency of 5e and 5g could be attributed to the bulky nature of the hydrophobic
substituents that may cause steric clashes with the residues of the hydrophobic arm II. Finally,
compound 5b, which has balanced activity on both receptors, forms a near-perfect extensive
network of H bonds with the key polar residues of the AF-2 helix of PPARγ, namely, His323,
His449, and Tyr473. On the other end, its lipophilic part could make contacts with both
hydrophobic arms. The benzimidazole moiety fits nicely in a tight pocket formed by Leu228,
Leu330, and Leu333, while the isobutyl group extends downwards between Pro227 and
Met329 without any steric clashes (Figure 5c). The previous observations emphasize the
importance of H bonding interactions with the AF-2 helix for PPARγ activation. As a result,
some of the prepared compounds showed obvious complementarity with the hydrophobic
1
1

ACCEPTED MANUSCRIPT
cavity but failed to cause significant activation of the receptor due to the lack of H bonding
with the critical polar residues.
As discussed before, a homology model of FFAR1, developed in our laboratory, was used
for the design of our compounds library. However, during the course of our study, the crystal
structure of FFAR1 in complex with the allosteric agonist TAK-875 was released.[56] Hence,
we decided to use this crystal structure in the docking studies of the final compounds to
question the possibility of binding of the prepared ligands to the observed allosteric site and
compare the results to our previous docking and MD simulation study.[23] It is worth noting
that the studied compounds were designed by docking them into the canonical orthosteric
pocket among transmembrane helices (TM) 3, 5, and 7 as in other Class A GPCRs. In addition,
prior to docking, this model had been subjected to MD simulation for 20 ns in an NPγT
ensemble using Langevin dynamics with the agonist GW9508 in the orthosteric site.
Compound GW9508 was selected because it is the most studied FFAR1 agonist and it lacks
unnecessary functional groups. This previous simulation study helped to open up the putative
binding site, break the ionic lock between Glu172 and Arg258, and transform the receptor into
the active conformation.[23] However, the recently published crystal structure is bound to an
allosteric ligand, TAK-875, binding in a pocket between TM3 and TM4 which is distinctive
from the orthosteric site. In addition, contrary to our model, the whole receptor is in an
“inactive-like” state with the ionic lock intact and TM7 close to the helical bundle core
narrowing the putative binding site. Therefore, docking of our compounds into the FFAR1
crystal structure, with only one constrain to Arg258, led to the orientation of all ligands into the
non-canonical site between transmembrane helices 3, 4, and 5 (Figure 6). Moreover, bulky
compounds were found to protrude from the receptor between TM3 and TM4 in a similar
fashion to the crystallized ligand TAK-875. On the other end, the TZD head of most of our
1
2

ACCEPTED MANUSCRIPT
compounds were able to make H bonds with the critical polar residues of FFAR1; Arg183 and
Arg258.
Series 1, the most active against FFAR1, shows complete overlap with TAK-875 due to the
similarity to the hydrophobic biphenyl tail of the crystallized ligand. We did not notice a
significant effect of substitution on the terminal phenyl ring on activity. Similarly, it has been
reported that the terminal phenyl ring of TAK-875 could accommodate variable substituents
with negligible effect on the potency. These observations suggest an allosteric binding mode of
series 1, similar to the crystal structure, where the terminal substituents do not make contacts
with the protein (Figure 6a). Likewise, series 2 adapts a binding mode in which the two phenyl
rings protruding between TM3 and TM4. Within this series, we noticed that para-substitution is
preferred to meta. Also, the ortho substitution in case of compound 2g could rationalize its
highest activity due to its ability to make hydrophobic interaction with Val84 (Figure 6b).
However, the relatively weak activity of this series could be attributed to the sub-optimal H
bonding network imparted by its bulky nature (weak interactions with Tyr91 and Arg258).
Contrary to the other series, 5 shows a notable differential effect of substituents on the FFAR1
agonistic activity. This could be explained by the proximity of these substituents to the protein
(Figure 6c). It is expected that the butyl and benzyl groups could make stronger hydrophobic
interactions with Pro80, Val84, and Phe142 than the allyl and isopropyl groups. However, in
case of 5e, the combination of size, rigidity, and direct attachment of the cyclopentyl ring to the
benzimidazole scaffold leads to loss of activity.
1
3

ACCEPTED MANUSCRIPT
Figure 5: Proposed binding mode of 1a (a), 2d (b), and 5b in the crystal structure of PPARγ. Protein is
represented as grey cartoon with amino acid residue colored green. Compounds are displayed as yellow sticks.
Only residues having significant interaction with the ligand are shown.
1
4

ACCEPTED MANUSCRIPT
Figure 6: Proposed binding mode of 1a (a), 2g (b), and 5b in the crystal structure of FFAR1. Protein is
represented as grey cartoon with amino acid residue colored green. The crystallized ligand TAK-875 (cyan) and
the prepared compound (yellow) are displayed as sticks. A part of TM4 was removed for clarity.
1
5

ACCEPTED MANUSCRIPT
3. Conclusions
TZDs represent a well-studied group of antidiabetic agents. Based on previous observations
of the ability of some TZDs to activate FFAR1, we designed five scaffolds consisting of the
TZD head, a linker, and a carefully selected privileged structure. For the present study, we
decided to explore three of these scaffolds and nineteen compounds were synthesized. Nine of
the prepared compounds showed acceptable activity on both receptors with series 5, the
benzimidazole-based series, being the most promising. Our docking study indicated that polar
interactions with the H bonding triad of PPARγ or FFAR1 is more important for receptor
activation than hydrophobic interactions. It should be emphasized that the main objective of
this work was not to obtain a highly potent agonist on PPARγ or FFAR1, but to design a lead
compound with a balanced activity on both targets. The benzimidazole series 5, followed by
the biphenyl series 1, were proved to be the most suitable scaffolds for this purpose. They
provide a privileged structure with suitable shape and size for further fine-tuning of the activity
on both receptors. Future studies in our laboratory will aim to optimize these lead compounds
as well as to explore the remaining scaffolds, 3 and 4. We hope that these efforts could lead to
the discovery of a dual acting antidiabetic agent to replace some of the currently used drug
combinations.
4
. Experimental
.1. Chemistry
Reaction progress was monitored by thin layer chromatography (TLC) analysis on Merck
4
silica gel 60 F254 plates. Visualization was performed with UV light (254 nm) or iodine.
Yields are of purified compounds and were not optimized. Reagents and solvents were
obtained from commercial sources and used without any further purification. Flash column
chromatography was performed on Loba Chemie silica gel (200–400 mesh) as a stationary
phase. Melting points (mp) were determined in open capillaries on an Electrothermal IA9100
1
6

ACCEPTED MANUSCRIPT
melting point apparatus and were uncorrected. Optical rotations were measured with a
JASCO, DIP-370 Digital polarimeter at 29 °C. Concentrations are expressed as g/100 ml.
Low-resolution mass spectra (MS) were recorded with a Shimadzu QP2010-Plus gas
+
chromatograph/mass spectrometer (GC/MS), electron impact (EI ) 70 eV maintained at 250
°
C. Infrared (IR) spectra were recorded on a Bruker Tensor 27 FT-IR spectrophotometer,
-1
driven by OPUSTM version 1.1 managing software (Vmax in cm , using KBr pellets).
1
13
The proton and carbon nuclear magnetic resonance ( H- and C-NMR) spectra were
recorded in CDCl or DMSO-d as a solvent on a Varian-MERCURY 300 (300 MHz)
3
6
spectrometer or a Bruker Ascend Aeon 400 (400 MHz) spectrometer at room temperature
within the range (20-25 °C). Chemical shifts (δ) are expressed as parts per million (ppm),
using tetramethylsilane (TMS) as the internal standard. For Signal multiplicities, the
following abbreviations are used as follows; s (singlet), d (doublet), dd (doublets of doublet),
t (triplet), q (quartet), br s (broad singlet) and m (multiplet). Coupling constants (J values) are
given in hertz (Hz). The acidic protons of the amidic –NH in the thiazolidindione ring,
1
alcohols or aromatic amines were not frequently observed in H-NMR spectra.
4.1.1. General Procedure A (Mitsunobu reaction).
To a stirred and ice-bath cooled mixture of the phenol (1 equiv) in anhydrous THF (9 ml),
was added the alcohol (1 equiv). Triphenylphosphine (PPh3) (1.1 equiv) was added
portionwise to this mixture and the system was kept under nitrogen by evacuation and filling
with nitrogen three times. Subsequently, diisopropyl azodicarboxylate (DIAD) (94 wt %
solution in toluene, 1.1 equiv) was added in a strictly dropwise fashion. The reaction mixture
was first stirred for 1 h at 0 °C and then overnight at room temperature. After completion of
the reaction as indicated by TLC, the reaction mixture was concentrated under reduced
pressure. The residue was extracted with EtOAc (2x) and the combined extract was washed
successively with saturated aqueous NaHCO and sodium chloride (NaCl) solutions, dried
3
1
7

ACCEPTED MANUSCRIPT
over anhydrous MgSO and concentrated in-vacuo affording an oily residue. The crude oil
4
was chromatographed on flash column to yield the desired ether derivative.
4
.1.2. General Procedure B (Suzuki-Miyaura Coupling).
A solvent mixture of MeOH (5 ml), toluene (20 ml) and potassium carbonate
aqueous solution (2 equiv in 1 ml) were sonicated at room temperature for 30 min and then
flushed with nitrogen for 30 min. Bromobenzene derivative (1 equiv), substituted
phenylboronic acid (1.25 equiv), and tetrakis(triphenylphosphine) palladium(0) [Pd(PPh ) ]
3
4
(0.024 equiv) were added sequentially at room temperature. The reaction mixture was
refluxed with stirring under nitrogen atmosphere overnight. The reaction was ceased through
stirring in open air for 30 min at room temperature forming a black emulsion. The insoluble
material was filtered off through celite pad and the filtrate was extracted with EtOAc (2x).
The combined organic phases were washed successively with water, saturated aqueous
sodium bicarbonate (NaHCO ), and brine. After drying over anhydrous magnesium sulphate
3
(
MgSO ), the solvent was evaporated to dryness in-vacuo affording a brownish yellow sticky
4
mass. The residue was purified by chromatography on silica gel with stepwise gradient
elution (n-hexane/DCM; 8:2 to 3:7) to give the desired biaryl derivatives.
4.1.3. General procedure C (Knoevenagel Condensation).
To a stirred solution of the appropriate benzaldehyde derivative (1 equiv) in toluene (50
ml) were sequentially added thiazolidine-2,4-dione (3; 1 equiv), benzoic acid (0.5 equiv) and
piperidine (0.5 equiv). The flask was connected to a Dean-Stark apparatus and a reflux
condenser. After refluxing overnight, the reaction mixture is reduced then cooled and placed
in the fridge overnight. In case of formed precipitant, the reaction suspension was filtered and
the residue was washed with cold water and hexane. The washed residue was dried in oven at
4
5 °C overnight affording the desired derivatives of the 5-benzylidenethiazolidine-2,4-dione
as canary yellow residues. However, in case of absence of precipitant, the reaction mixture
1
8

ACCEPTED MANUSCRIPT
was extracted with EtOAc (2x), washed successively with saturated aqueous NaHCO and
3
brine, dried over anhydrous MgSO and concentrated in vacuo affording the canary yellow
4
,
crude residues. In both cases, the residue was used for the next step without further
purification or analysis.
4.1.4. General Procedure D (olefin reduction)
To a rapidly stirred mixture of 5-benzylidenethiazolidine-2,4-dione derivatives
Knoevenagel condensation products; 1 equiv) in THF (2ml) and 1 M sodium hydroxide
(
solution (0.06 equiv), was added CoCl -DMG complex solution (0.6 equiv), made from
2
cobaltus chloride hexahydrate (0.07 g) and dimethylglyoxime (0.75 g) in DMF (15 ml) at
room temperature. After stirring for 30 min, sodium borohydride aqueous solution (5.8 equiv)
was added and the mixture was stirred at 35-40 °C overnight. The reaction mixture was
quenched with water and aqueous HCl at 0 °C and extracted with EtOAc (2x). The combined
organic fractions were washed with brine, dried over anhydrous MgSO and concentrated
4
under reduced pressure. Purification of the afforded yellowish crude residue by flash column
chromatography furnished the desired end products of the three KM Series.
4.1.5. Thiazolidine-2,4-dione (2)
To a solution of chloroacetic acid (5 g, 52.91 mmol) in 6 ml of water was added thiourea
(1) (4 g, 52.91 mmol) in 6 ml of water. After stirring for 30 min accompanied by considerable
cooling, a white precipitant was separated out. The product was filtered, washed with water
to remove traces of HCl and finally dried. It was purified by recrystallization in hot water
affording the above titled compound as white crystals (5.39 g). Yield: 87%; mp: 125-127 °C;
1
+
+
H-NMR (300 MHz, DMSO-d ): δ 4.14 (s, 2H), 12.02 (br s, 1H); MS (EI ) m/z: 117 [M ].
6
4.1.6. 5-(4-hydroxybenzylidene)thiazolidine-2,4-dione (3)
1
9

ACCEPTED MANUSCRIPT
Prepared as reported above, by the general procedure C for Knoevenagel condensation,
using the commercially available 4-hydroxybenzaldehyde (2 g, 16.38 mmol), the previously
prepared 2 (1.9 g, 16.38 mmol), benzoic acid (0.98 g, 8.19 mmol), and piperidine (0.79 ml,
8
.19 mmol) in toluene. The precipitant was filtered, washed with cold water, EtOH, and
hexane. The above addressed compound was obtained as canary yellow powder (2.72 g).
Yield: 75%; mp: > 300 °C; IR (KBr): 3402, 3120, 2998, 2864, 1721, 1677, 1570, 1504, 1332,
1
1
213, 900, 692; H-NMR (300 MHz, DMSO-d ): δ 6.89-6.92 (d, J = 8.67 Hz, 2H), 7.43-7.46
6
13
(
d, J = 8.67 Hz, 2H), 7.70 (s, 1H), 10.31 (br s, 1H); C-NMR (300 MHz, DMSO-d ): δ
6
+
+
1
16.31, 118.97, 123.93, 132.29, 132.35, 159.86, 167.48, 168.04; MS (EI ) m/z: 211 [M ].
.1.7. 5-(4-(benzyloxy)benzylidene)thiazolidine-2,4-dione (4)
To a stirred suspension of 3 (0.30 g, 1.36 mmol) and Cs CO (0.88 g, 2.71 mmol) in
4
2
3
anhydrous DMF (10 ml) was added benzyl chloride (0.19 ml, 1.63 mmol) dropwise. The
reaction mixture was stirred at room temperature under nitrogen atmosphere for 4 h. After
completion, the reaction mixture was extracted with EtOAc (2x) and the combined organic
phases were washed successively with water and brine, and dried over anhydrous MgSO4.
The solvent was evaporated to dryness under reduced pressure affording the desired title
compound in quantitative yield as yellowish green solid (0.42 g). Yield: 67%; mp: 75-77 °C;
1
IR (Nujol): 1684, 1597, 1506, 1445, 1309, 1152, 1035, 673; H-NMR (300 MHz, CDCl ): δ
3
5
7
1
.13 (s, 2H), 7.03-7.13 (m, 2H), 7.23-7.31 (m, 2H), 7.34-7.39 (m, 1H), 7.43-7.53 (m, 2H),
1
3
.58-7.66 (m, 1H), 7.80-7.98 (m, 2H), 9.91 (s, 1H); C-NMR (300 MHz, CDCl ): δ 69.28,
3
+
15.09, 122.79, 125.82, 130.26, 130.33, 131.38, 131.99, 138.26, 163.32, 190.65; MS (EI )
+
m/z: 290 [M ], 292 [M+2] showing relative intensity ratio ~ 1:1.
4.1.8. 5-(4-(benzyloxy)benzyl)thiazolidine-2,4-dione (5)
2
0

ACCEPTED MANUSCRIPT
This compound was accessed in a straightforward fashion from 4 (0.20 g, 0.64 mmol) by
means of the general procedure D for reduction except that the reaction time was set for 8 h
instead of overnight. Column purification (100% DCM to DCM/MeOH; 97:3, stepwise
gradient) furnished the desired reduced end product as a pale yellow solid (0.19 g). Yield:
1
9
3
6
5%; mp: > 300 °C; H-NMR (300 MHz, DMSO-d ): δ 3.06 (dd, J = 9.03, 14.20 Hz, 1H),
6
.33 (dd, J = 4.20, 14.20 Hz, 1H), 4.87 (dd, J = 4.20, 9.03 Hz, 1H), 5.07 (s, 1H), 5.19 (s, 1H),
1
3
.91-6.99 (m, 2H), 7.13-7.20 (m, 2H), 7.26-7.51 (m, 4H), 7.74 (s, 1H), 12.25 (br s, 1H); C-
NMR (400 MHz, DMSO-d ): δ 36.76, 53.48, 69.64, 115.12, 116.17, 126.21, 128.20, 128.88,
6
+
+
1
30.84, 132.52, 136.97, 137.54, 157.90, 168.55, 176.18; MS (ESI ) m/z: 336 [M+Na] .
.1.9. 4-((3-bromobenzyl)oxy)benzaldehyde (7a)
Mitsunobu coupling of 3-bromobenzyl alcohol (6a) (2.30 g, 12.28 mmol) with 4-
4
hydroxybenzaldehyde, (1.5 g, 12.28 mmol) was performed as described in the general
procedure B affording yellow oily crude residue. Purification via silica gel column
chromatography (n-hexane/DCM; 8:2 to 1:9 stepwise gradient) gave the product as a fluffy
icy-white solid (2.40 g). Yield: 67%; mp: 75-77 °C; IR (Nujol): 1684, 1597, 1506, 1445,
1
1
7
2
1
309, 1152, 1035, 673; H-NMR (300 MHz, CDCl ): δ 5.13 (s, 2H), 7.03-7.13 (m, 2H),
3
.23-7.31 (m, 1H), 7.34-7.39 (m, 1H), 7.43-7.53 (m, 1H), 7.58-7.66 (m, 1H), 7.80-7.98 (m,
1
3
H), 9.91 (s, 1H); C-NMR (300 MHz, CDCl ): δ 69.28, 115.09, 122.79, 125.82, 130.26,
3
+
+
30.33, 131.38, 131.99, 138.26, 163.32, 190.65; MS (EI ) m/z: 290 [M ], 292 [M+2] showing
relative intensity ratio ~ 1:1.
.1.10. 4-((4-bromobenzyl)oxy)benzaldehyde (7b)
4
Mitsunobu coupling of 4-bromobenzyl alcohol (6b) (0.77 g, 4.09 mmol)with 4-
hydroxybenzaldehyde (0.5 g, 4.09 mmol), as described in the general procedure B, afforded
an orange oily crude residue. Purification via silica gel column chromatography with
2
1

ACCEPTED MANUSCRIPT
stepwise gradient elution (n-hexane/DCM; 8:2 to 2:8) gave icy-white crystals (0.83 g). Yield:
1
7
2
1
0%; mp: 95-97 °C; H-NMR (400 MHz, DMSO-d ): δ 5.12 (s, 2H), 7.08 (d, J = 8.80 Hz,
6
H), 7.33 (d, J = 8.31 Hz, 2H), 7.55 (d, J = 8.31 Hz, 2H), 7.86 (d, J = 8.80 Hz, 2H), 9.91 (s,
+
+
H); MS (EI ) m/z: 290 [M ], 292 [M+2] showing relative intensity ratio ~ 1:1.
.1.11. 4-([1,1'-biphenyl]-3-ylmethoxy)benzaldehyde (8a)
Aryl bromide key intermediate 7a (0.30 g, 1.03 mmol), obtained from the reaction
4
described above, was coupled with phenyl boronic acid (0.16 g, 1.29 mmol) as described in
the general procedure B for Suzuki-Miyaura coupling, to give a yellowish white solid (0.19
1
g). Yield: 63%; mp: 73-75 °C; H-NMR (300 MHz, CDCl ): δ 5.23 (s, 2H), 7.12 (d, J = 8.79
3
13
Hz, 2H), 7.31-7.55 (m, 5H), 7.55-7.76 (m, 4H), 7.81 (d, J = 8.79 Hz, 2H), 9.91 (s, 1H); C-
NMR (300 MHz, CDCl ): δ 70.29, 115.14, 126.30, 126.35, 127.17, 127.53, 128.81, 129.18,
3
+
+
1
30.20, 132.02,136.45, 140.66, 141.83, 163.68, 190.78; MS (EI ) m/z: 288 [M ].
.1.12. 4-((2'-methyl-[1,1'-biphenyl]-3-yl)methoxy)benzaldehyde (8b)
This compound was prepared from 7a (0.30 g, 1.03 mmol) and o-tolyl boronic acid (0.18
4
g, 1.29 mmol) using a procedure similar to that used for 8a, and obtained as an off-white
1
solid (0.23 g). Yield: 75%; mp: 83-85 °C; H-NMR (300 MHz, CDCl ): δ 2.26 (s, 3H), 5.22
3
(s, 2H), 7.04-7.16 (m, 2H), 7.19-7.26 (m, 1H), 7.28-7.37 (m, 2H), 7.39-7.58 (m, 4H), 7.60-
1
3
7
1
1
.74 (m, 1H), 7.81-7.92 (m, 2H), 9.90 (s, 1H); C-NMR (300 MHz, CDCl ): δ 20.40, 70.22,
3
15.17, 125.82, 127.47, 128.25, 128.49, 129.14, 129.70, 130.15, 130.36, 131.97, 132.14,
+
+
35.26, 135.79, 141.35, 142.44, 163.70, 190.76; MS (EI ) m/z: 302 [M ].
.1.13. 4-((2',6'-dimethyl-[1,1'-biphenyl]-3-yl)methoxy)benzaldehyde (8c)
This compound was prepared from 7a (0.30 g, 1.03 mmol) and (2,6-dimethyl
4
phenyl)boronic acid (0.19 g, 1.29 mmol) in a manner similar to that used for 8a, and
1
furnished the end product as a yellowish white oil (0.23 g). Yield: 71%; H-NMR (300 MHz,
2
2

ACCEPTED MANUSCRIPT
CDCl ): δ 2.02 (s, 6H), 5.22 (s, 2H), 7.11 (d, J = 9.03 Hz, 2H), 7.14-7.19 (m, 4H), 7.23 (s,
3
1
3
1
H), 7.37-7.53 (m, 2H), 7.84 (d, J = 9.03 Hz, 2H), 9.90 (s, 1H); C-NMR (300 MHz,
CDCl ): δ 20.83, 70.17, 115.22, 125.56, 127.19, 127.32, 127.96, 128.85, 128.95, 130.06,
3
+
+
1
31.94, 135.94, 136.17, 141.25, 141.52, 163.65, 190.80; MS (EI ) m/z: 316 [M ].
.1.14. 4-((4'-(trifluoromethyl)-[1,1'-biphenyl]-3-yl)methoxy)benzaldehyde (8d)
This compound was prepared from 7a (0.30 g, 1.03 mmol) and (4-
4
(trifluoromethyl)phenyl)boronic acid (0.25 g, 1.29 mmol) using the procedure described for
1
8
a, and obtained as a thick orange oil (0.29 g). Yield: 79%; mp: 95-96 °C; H-NMR (300
MHz, CDCl ): δ 5.24 (s, 2H), 7.12 (d, J = 8.79 Hz, 2H), 7.44-7.64 (m, 3H), 7.64-7.79 (m,
3
1
3
5
1
1
H), 7.88 (d, J = 8.79 Hz, 2H), 9.92 (s, 1H); C-NMR (300 MHz, CDCl ): δ 70.07, 115.11,
3
25.80, 126.36, 127.22, 127.28, 127.48, 129.42, 130.23, 132.06, 136.78, 140.35, 144.28,
+
+
63.57, 190.80; MS (EI ) m/z: 356 [M ].
.1.15. 5-(4-([1,1'-biphenyl]-3-ylmethoxy)benzyl)thiazolidine-2,4-dione (1a)
The corresponding Knoevenagel product 9a (0.15 g, 0.39 mmol), obtained using the
4
general procedure C illustrated above and without further purification, was selectively
reduced with NaBH (0.09 g, 2.26 mmol) and CoCl -DMG complex (0.09 ml, 0.24 mmol) as
4
2
described in the general procedure D. The purification using silica gel column
chromatography (100% DCM to DCM/MeOH; 98:2 stepwise gradient) afforded the desired
reduced end product in sub-quantitative yield as a pale yellowish white solid (0.12 g). Yield:
1
8
3
6
1%; mp: >300 °C; H-NMR (300 MHz, DMSO-d ): δ 3.06 (dd, J = 9.03, 13.92 Hz, 1H),
6
.29 (dd, J = 4.39, 13.92 Hz, 1H), 4.87 (dd, J = 4.39, 9.03 Hz, 1H), 5.14 (s, 1H), 5.26 (s, 1H),
.98 (d, J = 8.30 Hz, 2H), 7.14-7.23 (m, 2H), 7.34-7.68 (m, 8H), 7.73-7.77 (m, 1H), 12.22 (br
1
3
s, 1H); C-NMR (300 MHz, CDCl ): δ 36.28, 52.95, 69.12, 114.67, 125.72, 126.00, 126.67,
3
2
3

ACCEPTED MANUSCRIPT
+
1
28.91, 130.33, 137.20, 137.77, 139.82, 140.29, 140.37, 157.40, 167.92, 175.61; MS (ESI )
+
m/z: 412 [M+Na] .
4.1.16. 5-(4-((2'-methyl-[1,1'-biphenyl]-3-yl)methoxy)benzyl)thiazolidine-2,4-dione
(1b)
Reduction of the corresponding Knoevenagel product 9b (0.15 g, 0.37 mmol), obtained
from the general procedure C, was achieved in a manner similar to that described for 10a.
The title end product was obtained as yellow thick solid (0.09 g). Yield: 61%; mp: 148-150
1
°
C; H-NMR (300 MHz, DMSO-d ): δ 2.02 (s, 3H), 3.07 (dd, J = 9.05, 13.92 Hz, 1H), 3.32
6
(dd, J = 4.15, 13.92 Hz, 1H), 4.89 (dd, J = 4.15, 9.05 Hz, 1H), 5.14 (s, 1H), 5.26 (s, 1H),
1
3
6
.92-7.00 (m, 2H), 7.15-7.32 (m, 2H), 7.38-7.66 (m, 7H), 7.72 (s, 1H), 12.19 (br s, 1H); C-
NMR (300 MHz, CDCl ): δ 20.04, 36.27, 52.96, 69.35, 115.96, 124.84, 125.85, 125.90,
3
1
1
26.16, 128.60, 129.41, 130.30, 134.62, 136.49, 137.04, 140.91, 141.39, 157.35, 171.58,
+
+
75.64; MS (EI ) m/z: 403 [M ].
4.1.17. 5-(4-((2',6'-dimethyl-[1,1'-biphenyl]-3-yl)methoxy)benzyl)thiazolidine-2,4-
dione (1c)
Reduction of 9c (0.15 g, 0.36 mmol), obtained from the general procedure C, was achieved
in a similar manner to that described for 10a. The title product was afforded as yellow thick
1
solid (0.10 g). Yield: 68%; mp: 149-151 °C; H-NMR (300 MHz, DMSO-d ): δ 1.94 (s, 6H),
6
3
.05 (dd, J = 9.03, 13.92 Hz, 1H), 3.33 (dd, J = 4.15, 13.92 Hz, 1H), 4.85 (dd, J = 4.15, 9.03
Hz, 1H), 5.14 (s, 1H), 5.26 (s, 1H), 6.95 (d, J = 8.79 Hz, 2H), 7.08-7.21 (m, 7H), 7.41-7.57
1
3
(
m, 1H), 7.70 (s, 1H), 12.01 (br s, 1H); C-NMR (300 MHz, CDCl ): δ 20.43, 36.79, 52.97,
3
6
1
6.96, 114.78, 124.84, 127.22, 127.92, 130.24, 135.08, 135.10, 136.51, 139.10, 140.39,
+
+
41.06, 157.26, 171.59, 175.67; MS (EI ) m/z: 417 [M ].
2
4

ACCEPTED MANUSCRIPT
4.1.18. 5-(4-((4'-(trifluoromethyl)-[1,1'-biphenyl]-3-yl)methoxy)benzyl)
thiazolidine-2,4-dione (1d)
Reduction of the corresponding Knoevenagel product 9d (0.15 g, 0.33 mmol) was
achieved in a procedure similar to that used for 10a. The title product was afforded as
1
yellowish brown solid (0.10 g). Yield: 69%; mp: >300 °C; H-NMR (400 MHz, DMSO-d ): δ
6
3
.07 (dd, J = 9.05, 14.06 Hz, 1H), 3.30 (dd, J = 4.16, 14.06 Hz, 1H), 4.88 (dd, J = 4.16, 9.05
Hz, 1H), 5.17 (s, 2H), 7.00 (d, J = 8.56 Hz, 2H), 7.15-7.22 (m, 2H), 7.50-7.60 (m, 2H), 7.67-
1
3
7
.76 (m, 1H), 7.83 (d, J = 8.56 Hz, 3H), 7.87-7.96 (m, 2H), 12.09 (br s, 1H); C-NMR (400
MHz, CDCl ): δ 36.74, 53.48, 69.48, 115.17, 126.26, 126.30, 126.89, 127.02, 127.22, 128.00,
3
1
28.27, 129.39, 129.77, 130.88, 132.55, 138.56, 139.21, 144.40, 157.86, 172.15, 176.18; MS
+
+
(
ESI ) m/z: 457 [M ].
4.1.19. 5-(4-((3-bromobenzyl)oxy)benzyl)thiazolidine-2,4-dione (11a)
Reduction of the olefin in the corresponding Knoevenagel product 10a (0.15 g, 0.38
mmol) was achieved according to the general procedure D. The title end product was
1
obtained as yellow solid (0.11 g). Yield: 76%; mp: >300 °C; H-NMR (300 MHz, DMSO-
d6): δ 3.06 (dd, J = 9.03, 14.06 Hz, 1H), 3.33 (dd, J = 4.39, 14.06 Hz, 1H), 4.88 (dd, J = 4.39,
9
.03 Hz, 1H), 5.09 (s, 2H), 6.92-6.99 (m, 2H), 7.14-7.20 (m, 2H), 7.35 (dd, J = 7.69, 8.79 Hz,
13
1
H), 7.45 (d, J = 7.69 Hz, 1H), 7.53 (d, J = 8.79 Hz, 1H), 7.65 (s, 1H), 11.98 (br s, 1H); C-
NMR (400 MHz, CDCl ): δ 36.74, 53.45, 69.14, 115.15, 121.39, 129.45, 130.87, 131.82,
3
+
1
31.90, 132.55, 136.47, 137.04, 157.70, 160.31, 172.13, 175.67; MS (ESI ) m/z: 414
+
[
M+Na] .
4.1.20. 5-(4-((4-bromobenzyl)oxy)benzyl)thiazolidine-2,4-dione (11b)
Reduction of the olefin in the corresponding Knoevenagel product 10b (0.15 g, 0.38
mmol) was achieved in a manner similar to according to the general procedure D. The title
2
5

ACCEPTED MANUSCRIPT
1
product was afforded as yellow solid 10.12 g). Yield: 80%; mp: >300 °C; H-NMR (400
MHz, DMSO-d ): δ 3.07 (dd, J = 8.93, 14.06 Hz, 1H), 3.29 (dd, J = 4.28, 14.06 Hz, 1H), 4.88
6
(dd, J = 4.28, 8.93 Hz, 1H), 5.06 (s, 2H), 6.95 (d, J = 8.56 Hz, 2H), 7.17 (d, J = 8.56 Hz, 2H),
1
3
7
3
1
.39-7.44 (m, 2H), 7.57-7.62 (m, 2H), 12.02 (br s, 1H); C-NMR (400 MHz, CDCl ): δ
3
6.74, 53.45, 68.82, 115.15, 121.39, 129.45, 130.29, 131.82, 131.90, 132.55, 136.47, 137.04,
+
+
57.70, 172.13, 175.16; MS (ESI ) m/z: 414 [M+Na] .
.1.21. (R)-(+)-3-chloro-1-phenyl-propan-1-ol (13)
Step 1: a flask was charged with (S)-(–)-α,α-diphenylprolinol (0.30 g, 1.19 mmol) and
4
anhydrous toluene (7 ml). The mixture was placed under nitrogen by evacuation and filling
three times. The borane tetrahydrofuran complex solution [BH •THF; 1 M solution in THF,
3
stabilized with 5 mM NaBH ] (0.35 ml, 3.56 mmol) was added in a dropwise fashion at 30
4
°
C and the clear solution was stirred for 30 min. Step 2: to this stirred solution, 3-chloro-1-
phenylpropan-1-one 12 (1 g, 5.93 mmol) in anhydrous toluene (1 ml) was added. The
reaction mixture was then stirred at room temperature for 15 min followed by the addition of
BH •THF (6 ml, 61.44 mmol) over a period of 3 min. After stirring for additional 90 min, the
3
reaction flask was cooled in ice bath and quenched successively with MeOH (10 ml),
isopropanole (10 ml), and HCl (3 ml), and was passed through a pad of Celite. The filtrate
was concentrated to dryness and crystallized from hexane. The title compound was obtained
2
5
1
as white fluffy solid (4.91 g). Yield: 97%; m.p. 57-59 ℃; [α] = + 25.7 ° (c 1, CHCl ); H-
D
3
NMR (400 MHz, CDCl ): δ 2.07-2.16 (m, 2H), 2.20-2.32 (m, 1H), 3.51-3.62 (m, 1H), 3.71-
3
1
3
3
4
.80 (m, 1H), 4.91-4.98 (m, 1H), 7.37-7.43 (m, 5H), C-NMR (400 MHz, CDCl ): δ 41.45,
3
+
+
1.73, 71.34, 125.80, 127.94, 128.69, 143.71; MS (EI ) m/z: 170 [M ], 172 [M+2] showing
relative intensity ratio ~ 3:1.
.1.22. (R)-4-(3-hydroxy-3-phenylpropoxy)benzaldehyde (14)
4
2
6

ACCEPTED MANUSCRIPT
After nitrogen substitution, 4-hydroxybenzaldehyde (2.15 g, 17.58 mmol) was dissolved in
a mixture of anhydrous THF/DMF (2:1v/v, 15 ml), and Cs CO (5.73 g, 17.58 mmol) was
2
3
added at room temperature. The mixture was stirred at 50 °C in a water bath for 1 h.
Compound 13 (1 g, 5.86 mmol) was added and the system was stirred at 50 °C overnight.
The reaction mixture was cooled, diluted with EtOAc (2x) and washed successively with
saturated aqueous NaHCO and brine. The combined organic layer was washed then with 1
3
M aqueous NaOH (3 x 25 ml) in order to remove the excess 4-hydroxybenzaldehyde. Finally,
the mixture was dried over MgSO4 and evaporated to dryness affording the title key
2
5
1
intermediate as a thick yellow oil (2.26 g). Yield: 95 %; [α] = – 3.7 ° (c 2.3, CHCl ); H-
D
3
NMR (300 MHz, CDCl ): δ 2.19-2.38 (m, 2H), 2.99 (br s, 1H), 4.13 (dt, J = 5.37, 9.52 Hz,
3
1
2
3
H), 4.28 (ddd, J = 5.37, 7.08, 9.52 Hz, 1H), 5.02 (dd, J = 5.37, 7.08 Hz, 1H), 6.94-7.08 (m,
13
H), 7.29-7.47 (m, 5H), 7.79-7.90 (m, 2H), 9.89 (s, 1H); C-NMR (400 MHz, CDCl ): δ
3
7.14, 64.38, 70.34, 113.78, 124.73, 126.78, 127.59, 128.91, 131.01, 142.99, 162.88, 189.92 ;
+
+
MS (EI ) m/z: 256 [M ].
4.1.23. (S)-4-(3-phenoxy-3-phenylpropoxy)benzaldehyde (15a)
Starting from 14 (0.50 g, 1.95 mmol) and phenol (0.18 g, 1.95 mmol), the steps of the
general procedure A described above for Mitsunobu coupling provided the title compound.
After flash column purification (n-hexane/EtOAc; 95:5 to 7:3 stepwise gradient), the target
1
compound was obtained as a colorless oil solidified in the fridge (0.38 g). Yield: 59%; H-
NMR (300 MHz, CDCl ): δ 2.24-2.42 (m, 1H), 2.42-2.60 (m, 1H), 4.14 (dt, J = 4.43, 9.42
3
Hz, 1H), 4.33 (ddd, J = 4.43, 8.06, 9.42 Hz, 1H), 5.42 (dd, J = 4.43, 8.06 Hz, 1H), 6.81-6.92
(m, 3H), 7.00 (d, J = 8.06 Hz, 2H), 7.26-7.47 (m, 7H), 7.83 (d, J = 8.06 Hz, 2H), 9.89 (s, 1H);
+
+
MS (EI ) m/z: 332 [M ].
4.1.24. (S)-4-(3-phenyl-3-(m-tolyloxy)propoxy)benzaldehyde (15b)
2
7

ACCEPTED MANUSCRIPT
The title compound was prepared from 14 (0.50 g, 1.95 mmol) and m-cresol (0.21 g, 1.95
mmol) using the same method as 15a giving a thick yellow oil which solidifies in the fridge
1
(
0.43 g). Yield: 63%; H-NMR (300 MHz, CDCl ): δ 2.25 (s, 3H), 2.28-2.38 (m, 1H), 2.41-
3
2
8
2
.54 (m, 1H), 4.06-4.19 (dt, J = 4.83, 9.39 Hz, 1H), 4.25-4.37 (m, 1H), 5.41 (dd, J = 4.83,
.46 Hz, 1H), 6.60-6.72 (m, 4H), 6.95-7.07 (m, 3H), 7.33-7.40 (m, 4H), 7.83 (d, J = 8.86 Hz,
+
+
H), 9.89 (s, 1H); MS (EI ) m/z: 346 [M ].
.1.25. (S)-4-(3-phenyl-3-(p-tolyloxy)propoxy)benzaldehyde (15c)
The title compound was prepared from 14 (0.50 g, 1.95 mmol) and p-cresol (0.21 g, 1.95
4
mmol), in a manner similar to that described for 15a, as yellow thick oil which solidifies in
1
the fridge (0.38 g). Yield: 56%; H-NMR (300 MHz, CDCl ): δ 2.22 (s, 3H), 2.31-2.56 (m,
3
2
6
H), 4.13 (dt, J = 4.64, 7.55 Hz, 1H), 4.25-4.40 (m, 1H), 5.37 (dd, J = 4.64, 8.75 Hz, 1H),
.68-6.79 (m, 2H), 6.93-7.07 (m, 4H), 7.28-7.47 (m, 5H), 7.75-7.93 (m, 2H), 9.89 (s, 1H);
1
3
C-NMR (400 MHz, CDCl ): δ 19.38, 37.12, 63.66, 70.24, 113.79, 114.79, 124.88, 126.71,
3
+
+
1
27.70, 128.78, 130.96, 140.32, 154.78, 162.88, 189.79; MS (EI ) m/z: 346 [M ].
.1.26. (S)-4-(3-(3-methoxyphenoxy)-3-phenylpropoxy)benzaldehyde (15d)
The title compound was prepared from 14 (0.50 g, 1.95 mmol) and 3-methoxyphenol (0.24
4
g, 1.95 mmol), in a manner similar to that described for 15a, as a thick colorless oil which
1
solidifies in the fridge (0.29 g). Yield: 41%; H-NMR (300 MHz, CDCl ): δ 2.22-2.41 (m,
3
1
1
4
H), 2.41-2.57 (m, 1H), 3.67-3.75 (s, 3H), 4.04-4.17 (m, 1H), 4.30 (dd, J = 5.54, 9.75 Hz,
H), 5.40 (dd, J = 5.54, 7.55 Hz, 1H), 6.39-6.50 (m, 4H), 6.95-7.08 (m, 3H), 7.33-7.42 (m,
+
+
H), 7.77-7.87 (m, 2H), 9.88 (s, 1H); MS (EI ) m/z: [M ].
.1.27. (S)-4-(3-phenyl-3-(4-(trifluoromethyl)phenoxy)propoxy)benzaldehyde (15e)
The title compound was prepared from 14 (0.50 g, 1.95 mmol) and 4-(trifluoromethyl)
phenol (0.32 g, 1.95 mmol), in a manner similar to that described for 15a, as a thick yellow
4
2
8