Structural characterization of amorfrutins bound to the peroxisome proliferator-activated receptor γ

Article abstract of DOI:10.1021/jm3013272

Amorfrutins are a family of natural products with high affinity to the peroxisome proliferator-activated receptor γ (PPARγ), a nuclear receptor regulating lipid and glucose metabolism. The PPARγ agonist rosiglitazone increases insulin sensitivity and is effective against type II diabetes but has severe adverse effects including weight gain. Amorfrutins improve insulin sensitivity and dyslipidemia but do not enhance undesired fat storage. They bear potential as therapeutics or prophylactic dietary supplements. We identified amorfrutin B as a novel partial agonist of PPARγ with a considerably higher affinity than that of previously reported amorfrutins, similar to that of rosiglitazone. Crystal structures reveal the geranyl side chain of amorfrutin B as the cause of its particularly high affinity. Typical for partial agonists, amorfrutins 1, 2, and B bind helix H3 and the β-sheet of PPARγ but not helix H12.

Full text of DOI:10.1021/jm3013272

Article
pubs.acs.org/jmc
Structural Characterization of Amorfrutins Bound to the Peroxisome
Proliferator-Activated Receptor γ
Jens C. de Groot,† Christopher Weidner,‡ Joern Krausze,† Ken Kawamoto,§ Frank C. Schroeder,§
Sascha Sauer,*,‡ and Konrad Bussow*,†
̈
†Department of Molecular Structural Biology, Helmholtz Centre for Infection Research, 38214 Braunschweig, Germany
‡Otto Warburg Laboratory, Max Planck Institute for Molecular Genetics, 14195 Berlin, Germany
§Boyce Thompson Institute and Department of Chemistry and Chemical Biology, Cornell University, Ithaca, New York,
United States
S
* Supporting Information
ABSTRACT: Amorfrutins are a family of natural products with high
affinity to the peroxisome proliferator-activated receptor γ (PPARγ), a
nuclear receptor regulating lipid and glucose metabolism. The PPARγ
agonist rosiglitazone increases insulin sensitivity and is effective against
type II diabetes but has severe adverse effects including weight gain.
Amorfrutins improve insulin sensitivity and dyslipidemia but do not
enhance undesired fat storage. They bear potential as therapeutics or
prophylactic dietary supplements. We identified amorfrutin B as a novel
partial agonist of PPARγ with a considerably higher affinity than that of
previously reported amorfrutins, similar to that of rosiglitazone. Crystal
structures reveal the geranyl side chain of amorfrutin B as the cause of
its particularly high affinity. Typical for partial agonists, amorfrutins 1, 2,
and B bind helix H3 and the β-sheet of PPARγ but not helix H12.
heterodimerization with RXR, and activates or represses
transcription of target genes in a ligand-dependent manner.
The LBD consists of 13 α-helices and a small, four-stranded
β-sheet. The lower half of the unbound LBD (Figure 2A) is
structurally less rigid than the upper portion.10,11 Helix H12 and
the loop between helices H2′ and helix H3 are the most mobile
segments. Ligand binding stabilizes the LBD and leads to a more
compact and rigid conformation, which in turn causes recruitment
of coactivators like SRC1 to the LBD’s AF2 effector binding site.1,12
PPARγ bound to the promoter of a target gene activates
transcription of that target gene upon coactivator recruitment.
The synthetic agonist rosiglitazone stabilizes helices H3 and H12,
which are part of the AF2 site, and thereby induces coactivator
recruitment.10 Binding of rosiglitazone to the LBD also inhibits
CDK5-mediated phosphorylation of Ser273, which is located in
the LBD (Figure 2A).1 Reduced Ser273 phosphorylation alters
the expression of a subset of genes with regulatory functions in
metabolism; for example, it increases expression of the insulin-
sensitizing adipokine adiponectin.1 The antidiabetic effects of
rosiglitazone are connected with the inhibition of Ser273
phosphorylation.
INTRODUCTION
■
The peroxisome proliferator-activated receptor γ (PPARγ) is a
nuclear receptor that regulates transcription with two effector
binding sites called activation function 1 (AF1) and activation
function 2 (AF2). AF1 is localized within the N-terminal
regulatory domain. The receptor’s central DNA binding domain
is followed by the C-terminal ligand binding domain (LBD),
which comprises AF2. PPARγ is regulated by a phosphorylation
site in the LBD at Ser273 (Ser245 in the shorter isoform 1).1
The ligand-activated transcription factor PPARγ acts in the
nucleus as a heterodimer with the retinoid X receptor RXR.2
PPARγ interacts with prostaglandins and fatty acids and their
metabolites.3−5 It acts as a sensor and regulator with a dominant
role in glucose and lipid metabolism and adipose cell differ-
entiation. Activation of the receptor improves insulin sensitivity
through different metabolic actions, including regulation of
adipokines.3 PPARγ is a well-established drug target for type II
diabetes. It plays also a key role in inflammation, atherosclerosis,
and cancer.3,6
The two other PPAR family members, PPARα and PPARδ (or β),
also bind fatty acids and are involved in fatty acid metabolism. In
general, PPARα and PPARδ promote fatty acid catabolism in several
tissues, whereas PPARγregulates fatty acid storage in adipose tissues.7
Dual PPARα and PPARγ agonists have been reported that correct
glucose and lipid abnormalities in patients with type 2 diabetes.8,9
The LBD of PPARγ has several regulatory functions. It
determines the receptor’s subcellular localization, initiates
Rosiglitazone and other glitazones (thiazolidinediones, TZDs)
strongly activate transcription of a large number of genes in various
tissues. This unspecific action of rosiglitazone is associated with its
Received: September 15, 2012
Published: January 3, 2013
© 2013 American Chemical Society
1535
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
a
severe side effects including weight gain, osteoporosis, cardiovas-
cular complications, and edema. Partial PPARγ agonists, which
activate PPARγ only weakly, are more selective PPARγ modulators
(SPPARγMs) and avoid side effects.13,14 The partial agonists
BVT.13, MRL-24, nTZDpa, and amorfrutin 1 block Ser273
phosphorylation as effectively as rosiglitazone but activate tran-
scription of target genes only to a moderate level.1,15 In contrast to
full agonists, these partial agonists do not contact helix H12 but
rather stabilize helix H3 and the β-sheet region of the binding pocket
of the PPARγ LBD.13
Amorfrutins are a group of natural products that have recently
been identified as PPARγ agonists with the characteristics of
SPPARγMs.15,16 Amorfrutins are nontoxic ingredients of edible
roots of licorice, Glycyrrhiza foetida, and of the fruits of Amorpha
fruticosa, an ingredient of condiments. They are high-affinity
PPARγ agonists with dissociation constants of around 300 nM.
The name “amorfrutin” is derived from the plant Amorpha
fruticosa, in which the molecules were originally identified. The
small, lipophilic amorfrutin class consists of a 2-hydroxybenzoic
acid core structure decorated with phenyl and isoprenyl moieties.
Amorfrutin 1 was shown to increase insulin sensitivity with
concomitant weight loss in mice.15 In contrast to rosiglitazone,
amorfrutin 1 did not cause weight gain or fluid retention.
Amorfrutins have potential for treatment of type II diabetes,
obesity, and the metabolic syndrome in general. As components of
edible biomaterials, they might also serve as prophylactic dietary
supplements.
Table 1. Dissociation Constants (Kd), Effective
b
c
Concentrations (EC50), and Transcriptional Activation
(TA) of Investigated Compounds Binding to PPAR
d
Subtypes
PPARγ
PPARα, Kd PPARδ, Kd
Kd
[μM]
EC50
[μM]
[μM]
[μM]
TA [%]
e
e
amorfrutin 1
amorfrutin 2
27
27
0.236
0.287
0.019
0.007
0.029
nd
0.458
1.200
0.050
0.004
nd
39
25
17
30
amorfrutin B
2.6
nd
1.8
nd
20
e
rosiglitazone
100
>50
>50
50−80
nd
ef
,
nTZDpa
nd
nd
f
MRL-24
nd
nd
nd
f
BVT.13
nd
nd
nd
nd
e
GW7647
GW0742
0.001
nd
nd
0.180
nd
nd
e
0.0004
nd
nd
a
Kd values were obtained by using a competitive TR-FRET assay.
EC50 and transcriptional activation values were determined from a
b
c
reporter gene assay. TA is the maximum activation of PPARγ relative
d
e
f
to rosiglitazone. nd, not determined. Value reported in ref 15. Value
reported in ref 13.
Structural Comparison. The crystal structures of the LBD
in complex with amorfrutins B and 2 were solved by molecular
replacement and were both refined to a resolution of 2.0 Å
(Table 2). The electron density maps clearly reveal the domain’s
canonical three-layer α-helical sandwich composed of 13 α-
helices and a small, four-stranded β-sheet (Figure 2A). The
structures are in accord with previously published PPARγ LBD
structures8,10,13 with a calculated rms deviation of 0.87 Å to
Protein Data Bank (PDB) entry 1PRG (258 common Cα-
positions of chain A aligned).
Here we present the novel PPARγ agonist amorfrutin B, which
has a considerably higher affinity to PPARγ than previously
described amorfrutins.15 For understanding the cause of this high
affinity, crystal structures of PPARγ in complex with amorfrutins
2 and B were solved and compared to the amorfrutin 1 complex
structure reported previously.15 The structures led to a detailed
description of amorfrutin recognition by PPARγ and revealed a
helix H12-independent interaction that is typical for the group of
SPPARγMs.
The asymmetric units of crystals of PPARγ in complex with
amorfrutins 1, 2, and B all contained two amorfrutin molecules
bound to two LBD molecules forming a homodimer. The
structures of the two monomers, denoted chain A and chain B,
differ because of crystal contacts of chain B to a neighboring
molecule in the crystal lattice. This conformational difference is
commonly observed in PPARγ LBD structures.8,10 In the
“inactive” conformation of chain B, helix H12 is dislocated and
forms crystal contacts with an adjacent molecule.8 The “active”
conformation of chain A corresponds to the structure of the
receptor in complex with the full agonist rosiglitazone and the
coactivator SRC1.10 This conformation is commonly regarded as
a suitable model for PPARγ LBD activation. The next sections
refer to chain A and the active conformation. The inactive
conformation is described in the Supporting Information file.
The electron density maps of the complex structures are well-
defined for each amorfrutin molecule (Figure 2C) and clearly
reveal the ligand positions in the cavity of the LBD.
Corresponding to their similar structures, the three amorfrutins
are bound with almost identical localization and orientation
(Figure 2A,B). Whereas the full agonists rosiglitazone and MRL-
20 stabilize helices H3 and H12 of the LBD,10,13 the amorfrutins
bind the receptor between helix H3 and the β-sheet, close to the
ligand entry site. The recognition of the amorfrutins by PPARγ is
strikingly similar to that of the partial agonists BVT.13, MRL-24,
and nTZDpa.13 The amorfrutins and the other three partial
agonists all bind helix H3 and the β-sheet by a combination of
hydrogen bonds to Ser342 and Arg288 and by extensive van der
Waals interactions with Ile341 of the β-sheet and Cys285 of helix
H3 (Figures 3, 4, and S1). The mechanism of stabilization and
RESULTS
■
Amorfrutin B Is a Novel High-Affinity Selective PPARγ
Agonist. Amorfrutin B17 was identified by a previously reported
screen for potential antidiabetic substances in edible biomate-
rials.15 Amorfrutin B and the previously reported amorfrutin 2
were synthesized. The dissociation constant of amorfrutin B to
human PPARγ was determined as 19 nM by a time-resolved
FRET assay (Table 1). In comparison to amorfrutins 1 and 2, the
affinity of amorfrutin B for PPARγ is an order of magnitude
stronger, comparable to that of rosiglitazone (Figure 1A, Table 1).
Amorfrutins 1, 2, and B are selective for PPARγ. Two other human
PPARs, PPARα and PPARδ, were bound with 60- to 140-fold
lower affinity in the FRET assay (Table 1).
A fusion construct of the PPARγ LBD and the DNA binding
domain of yeast GAL4 was used to measure the effect of
amorfrutin B on transcription of a reporter gene. Amorfrutin B
clearly activated transcription of the reporter gene (Figure 1B).
Despite its high affinity for the LBD, amorfrutin B binding
resulted only in a moderate level of transcriptional activation.
Reporter protein production induced by amorfrutin B was 20%
in comparison to rosiglitazone (Figure 1B, Table 1). Amorfrutin
B therefore belongs to the group of partial agonists of PPARγ.
Transcriptional activation by amorfrutin B (20%) was lower than
by the partial agonists amorfrutins 1 and 2 (39%, 30%), BVT.13,
MRL24, and nTZDpa (>50%) (Table 1).
1536
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
Figure 1. PPARγ binding and transcriptional activation by amorfrutins and rosiglitazone: (A) binding of compounds to the LBDs of PPARα, -γ, and -δ in
a competitive, time-resolved FRET assay; (B) cellular activation of PPARγ determined in a reporter gene assay in HEK 293H cells. Error bars represent
the standard deviation around the mean value (n = 3).
the low transcriptional activation (<50%) compared to the full
agonist rosiglitazone15 (Table 1) implicate amorfrutins as partial
PPARγ agonists.
(Figure 3). Overall, Arg288 and Ser342 form a network of
hydrogen bonds upon binding of amorfrutin 1 that is supported
by extensive van der Waals contacts of the ligand’s phenyl and
isoprenyl moieties (Figure 3).
Amorfrutins 2 and B also form direct hydrogen bonds to Ser342
of the β-sheet. However, their carboxyl group does not form a salt
bridge to Arg288 and they do not trigger the alternative
conformation of this residue. This observation is reflected by
increased crystallographic B-factors of the 2-hydroxybenzoic acid
cores of amorfrutins 2 and B, which were probably caused by lower
stabilization of the cores and stronger atom vibration in comparison
to amorfrutin 1 (Figure 2C). Amorfrutins 2 and B use the same
Despite their similar structures, the amorfrutins interact
differently with the LBD. Arg288 of the amorfrutin 1-bound
PPARγ adopts two alternative conformations (Figure 3), which
are also observed in nTZDpa-bound PPARγ but not in unbound
PPARγ (Figure 4). In the amorfrutin 1 complex, the positively
charged guanidinium group of Arg288 contacts different sets of
negatively charged groups depending on its conformation. The
residue forms electrostatic bonds either to Glu295 of helix H3 or
to Glu343 of the β-sheet and to the ligand’s carboxyl group
1537
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
a
and they do not affect the ligands’ position in the binding pocket.
The effect of the phenethyl/pentyl groups on Arg288 might be
mediated by the flexible loop between helices H2′ and H3 or by
solvent molecules.
The three amorfrutins were selectively recognized by PPARγ.
Their dissociation constants were lower by a factor of 60−140 in
comparison to the other PPAR subtypes in the FRET assay
(Table 1).15 The conservation of the residues of the lower
portion of helix H3 and the β-sheet is quite low (Figure S1).
Arg288 of PPARγ plays a key role in amorfrutin binding. The
replacement of Arg288 by a threonine in PPARα and PPARδ
(Figure S1) is most likely the reason why the three amorfrutins
are selective for the PPARγ subtype.
Table 2. Summary of Crystallographic Analysis
parameter
space group
amorfrutin B
amorfrutin 2
C2
C2
cell dimensions a, b, c (Å)
monoclinic angle β (deg)
X-ray source
92.14, 60.86, 117.6
102.58
92.29, 60.97, 117.9
102.65
Rigaku MicroMax-007 HF
wavelength (Å)
1.5419
1.5419
resolution range (Å)
29−2.0
27−2.0
last shell (Å)
2.10−2.00
8.7 (45.4)
148018 (18473)
42829 (5697)
17.68 (3.41)
99.2 (97.7)
3.5(3.2)
2.10−2.00
10.9 (56.3)
149072 (18878)
43005 (5857)
13.71 (2.82)
99.0 (99.5)
3.4(3.2)
b
R
merge (%)
observations
unique reflections
mean (I)/σ(I)
completeness
DISCUSSION
■
multiplicity
The present study identified amorfrutin B as a novel high-affinity
ligand of PPARγ with a dissociation constant of 19 nM, which is
similar to that of rosiglitazone but significantly lower than the
dissociation constants of amorfrutins 1 and 2. Therefore,
amorfrutin B currently represents the natural product with the
highest affinity for PPARγ.
structure refinement
resolution range (Å)
29−2.0
20.3
27−2.0
20.89
25.9
c
R
R
work (%)
d
free (%)
23.3
total number of
non-hydrogen atoms
protein atoms
4670
4339
60
4671
4380
44
Amorfrutins 1 and 2 and the novel amorfrutin B are recognized
by PPARγ in a similar way. They bind close to the ligand entry
site of the PPARγ LBD with almost identical localization and
orientation. All three amorfrutins form similar, extensive van der
Waals contacts with the LBD’s β-sheet and helix H3. The high
affinity of amorfrutins to PPARγ is caused by their carboxyl
group, which interacts with the main chain nitrogen of Ser342
and, as shown for amorfrutin 1, the guanidinium group of
Arg288. This is supported by the finding that esterification of the
carboxyl group of the structurally related amorfrutin 5 resulted in
a dramatic increase of the dissociation constant from 0.59 to
23 μM.15 The three amorfrutins contact Ser342 by hydrogen
bonds of similar strength (N−O distances 2.7−3.0 Å). The
increased binding affinity of amorfrutin B in comparison to
amorfrutin 1 arises from additional hydrophobic contacts of the
long geranyl side chain to Arg288 of helix H3 and to helix H4/5
(Figures 3 and S1). Arg288 also seems to be crucial for the
general selectivity of the amorfrutins for PPARγ over PPARα and
PPARδ. Arg288 is not conserved among the PPARs. PPARα and
PPARδ have a threonine at this position, which cannot interact
with the amorfrutins in the same way as arginine.
ligand atoms
water molecules
271
247
rmsd
bond length (Å)
bond angle (deg)
0.010
1.186
0.010
1.219
B-factors (Å2)
main chain
39.0
42.3
40.7
43.6
44.3
37.9
41.3
39.7
47.6
44.9
side chain
average protein atoms
average ligand atoms
average solvent
Ramachandran statistics
most favored regions (%) 98.3
97.3
2.7
allowed regions (%)
1.7
0
disallowed regions (%)
0
a
b
Values in parentheses are for the highest resolution shell. Rmerge
∑hkl∑i|Ihkl,i − I /∑ ∑iIhkl,i. Rwork = ∑hkl||Fobs| − |Fcalc||/∑ |Fobs|.
Rfree is calculated as Rwork but using Fobs derived from 5% randomly
=
c
̅
hkl
hkl
hkl
d
selected reflections excluded from refinement.
Structural comparison of amorfrutins 1, 2, and B with the
previously published partial PPARγ agonists BVT.13, MRL-24,
and nTZDpa reveals closely related interactions with PPARγ.
All of these ligands occupy the same binding site and interact
similarly with helix H3 and the β-sheet. Their carboxyl groups
interact directly with Ser342 of the β-sheet. Helix H3 is always
stabilized by a ligand carboxyl or hydroxyl group forming a direct
or a water-mediated hydrogen bond. Amorfrutins 2 and B most
likely mediate PPARγ stabilization and inhibition of Ser273
phosphorylation in the same way as amorfrutin 1,15 BVT.13,
MRL-24, and nTZDpa.1 It was shown by hydrogen/deuterium
exchange experiments that MRL-24 “freezes” the region of the
CDK5 phosphorylation site, apparently in a less favorable
conformation for the kinase.1
Despite their related structures, amorfrutins 1 and 2 reveal
distinct gene activation profiles.15 Therefore, it will be important
to study gene activation profiles and insulin sensitizing effects
also for amorfrutin B. Amorfrutin B binds to PPARγ at markedly
lower concentration than amorfrutin 1,15 and lower doses of
amorfrutin B might be sufficient for obtaining the same beneficial
pharmacological effect as amorfrutin 1.
residues of the LBD to form a water network that stabilizes Arg288
of helix H3 (Figure 3 and Figure S1). The partial agonist BVT.13 is
bound in a similar way (Figure 4).13 Amorfrutin 2 has a dissociation
constant of 287 nM, similar to amorfrutin 1 with 236 nM, whereas
binding of amorfrutin B is markedly stronger with a dissociation
constant of 19 nM (Table 1). Amorfrutin B is distinguished from
amorfrutin 1 by its long geranyl side chain, which forms additional
hydrophobic contacts especially to Arg288 of helix H3 and to helix
H4/5 (Figure 3). These additional contacts of amorfrutin B to
Arg288 prevent Arg288 from adopting the alternative conformation
that is required for direct contact to the compound’s carboxyl group
(Figure 3). The additional hydrophobic contacts are reflected by
lower crystallographic B-factors of the geranyl side chain compared
to the isoprenyl side chain of amorfrutin 1 (Figure 2C) and
rationalize the almost 12-fold higher binding affinity of amorfrutin B.
Amorfrutins 1 and 2 are distinguished only by the 6-phenethyl
and 6-pentyl substituents (Figure 2). The alternative con-
formation of Arg288 in the amorfrutin 1 complex obviously
depends on the presence of its phenethyl group. However, the
phenethyl/pentyl substituents are located distant from Arg288
1538
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
Figure 2. Structures of amorfrutins 1, 2, and B bound to the LBD of PPARγ. (A) The aligned LBD structures bound to the three amorfrutins are almost
indistinguishable. Only amorfrutin B (orange sticks) is shown. The amorfrutins bind between helix H3 and the β-sheet. Helices (blue) and β-strands
(green) are labeled H1−H12 and S1−S4, respectively. The phosphorylation site at Ser273 is represented by sticks (Ser273 of isoform 2 corresponds to
Ser245 of isoform 1). (B) Superposition of the three amorfrutins in the binding pocket, generated by aligning the three respective protein structures.
Amorfrutins 1 (PDB code 2YFE), 2, and B are shown as white, yellow, and orange sticks. Only the amorfrutin B-bound domain is displayed as ribbons.
(C) 2FO − FC electron density total omit maps calculated around the amorfrutins using SFCHECK,26 contoured at 1.0σ. The amorfrutin structures are
color-coded according to the crystallographic B-factors (the colors range from blue to red corresponding to increasing fluctuation of atom positions).
Note the low B-factors of the amorfrutin 1 2-hydroxybenzoic acid core in comparison to those of amorfrutins 2 and B.
geranyl side chain. Because of its high affinity, amorfrutin B has a
high potential for treatment or prevention of type II diabetes and
the metabolic syndrome.
CONCLUSIONS
■
Amorfrutins are powerful antidiabetic molecules from edible
biomaterials. Amorfrutin B was identified in this study as the
natural product with the highest reported affinity for PPARγ.
Crystal structures of PPARγ in complex with amorfrutins 1, 2,
and B reveal a highly similar binding mechanism that is related to
other partial PPARγ agonists. The high affinity of amorfrutin B is
caused by its carboxyl group and hydrophobic contacts of the
EXPERIMENTAL SECTION
■
Time-Resolved Fluorescence Resonance Energy Transfer
(TR-FRET) Assay. PPARγ ligands were characterized by a TR-FRET-
based competitive binding assays according to the manufacturer’s
1539
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
Figure 3. Structural details of amorfrutins bound to the LBD of PPARγ. Amorfrutin 1 (white) forms direct hydrogen bonds to Ser342 of the β-sheet and
Arg288 of helix H3 (PDB code 2YFE). Both alternative conformations of Arg288, marked by red arrows, are stabilized by hydrogen bonds. Amorfrutins
2 and B (yellow and orange) coordinate water molecules to form hydrogen-bonding networks between Arg288 of helix H3 and the β-sheet but also
directly interact with Ser342. Additional hydrophobic interactions of amorfrutin B’s longer geranyl side chain to Arg288 are the likely cause of this
ligand’s higher binding affinity. The geranyl moiety also prevents the alternative conformation of Arg288, which was observed in the complex with
amorfrutin 1. Only main chain atoms are shown for residues of the β-sheet region, excepting Glu343. In the schematic diagrams of atomic interactions on
the right, calculated using LIGPLOT,24 the ligands and PPARγ residues are drawn with black and orange bonds, respectively. Hydrogen bonds are
depicted by dashed lines, and van der Waals contacts are indicated by spoked arcs and atoms with spokes. Oxygen atoms are colored red, nitrogen atoms
blue, and carbon atoms black. S: β-sheet. H: helix. L: loop. Residue numbers correspond to PPARγ isoform 1. Hydrogen bonds are labeled with donor−
acceptor distances in angstrom.
protocol (LanthaScreen TR-FRET PPAR Alpha/Gamma/Delta com-
petitive binding assay kits, Invitrogen, Darmstadt, Germany), as
described.15 In the assay, increasing the concentration of potential
ligands resulted in a displacement of a fluorescent PPAR ligand from
fusion proteins of human PPAR LBDs with a DNA binding domain and
hence a decrease of the FRET signal.
Reporter Gene Assay. Cellular activation of PPARγ was assessed in
a reporter gene assay according to the manufacturer’s protocol
(GeneBLAzer PPARγ DA assay, Invitrogen), as described.15 In brief,
the assay uses HEK 293H cells stably expressing a GAL4-PPARγ-LBD
fusion protein and an β-lactamase reporter gene under the transcrip-
tional control of an upstream activator sequence. Cells were incubated
with indicated concentrations of compounds, and β-lactamase activity
was measured.
Protein Expression and Purification. The PPARγ-LBD was
expressed with a previously described expression plasmid kindly provided
by Krister Bamberg.8 Transformed BL21(DE3) cells were induced with
0.1 mM IPTGat18°C for 20 h. Harvestedcellswere disruptedwith a high
1540
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
at a final concentration of 1 mg/mL. DMSO and unbound ligand were
removed by gel filtration. The complex was concentrated to 11 mg/mL
and crystallized using hanging drop vapor diffusion by mixing it with an
equal volume of reservoir solution (0.8 M trisodium citrate and 0.1 M
imidazole, pH 8.0). Crystals appeared after 2−5 days at 19 °C and were
improved by microseeding. The crystals were transferred into a mildly
hypertonic cryoprotectant solution (0.84 M trisodium citrate, 25% v/v
glycerol, and 0.1 M imidazole, pH 8.0) and immediately thereafter flash
cooled in liquid nitrogen.
Structure Determination and Refinement. The data sets were
collected in-house on a Saturn 944+ detector (Rigaku) using X-rays
from a Rigaku MicroMax-007 HF rotating-anode X-ray generator with a
VariMax Optic (Rigaku). Images were indexed and processed with
XDS,18 and the structures were solved by molecular replacement using
CCP4 MOLREP19 with ligand-free PPARγ (PDB code 1PRG)10 as the
search model. REFMAC520 and Phenix.refine21 were used for
refinement and Coot22 for model building. Figures were prepared
with PyMOL.23 The schematic diagrams of atomic interactions were
calculated with LIGPLOT.24 Root mean square deviation (rmsd) values
between common Cα-positions were calculated with ProFit using the
McLachlan algorithm.25
Synthesis of Amorfrutins. Amorfrutin 1 and starting materials for
the synthesis of amorfrutin B and amorfrutin 2 were synthesized as
described15 or purchased from Sigma-Aldrich or Acros Organics and
used without further purification. Anhydrous solvents were prepared
using 4 Å molecular sieves. Silica gel flash column chromatography was
performed on Combiflash RF (Teledyne ISCO) chromatography
systems using normal-phase silica gel and ethyl acetate/hexanes
1
mixtures as solvents. H and 13C NMR spectra were obtained on 400,
500, and 600 MHz Varian NMR spectrometers using CDCl3 and
acetone-d6 (Cambridge Isotope Laboratories) as solvents. The purity of
the synthetic materials was assessed by proton NMR spectroscopy and
HPLC. All samples were of greater than 98% purity.
(E)-Methyl 3-(3,7-Dimethylocta-2,6-dien-1-yl)-2-hydroxy-4-
methoxy-6-phenethylbenzoate (Amorfrutin B Methyl Ester).
Figure 4. Comparison of related PPARγ complex structures. Shown are
the unbound LBD of PPARγ (PDB code 1PRG), with Arg288 hydrogen
bonded to Ser289,10 and the LBD in complex with the partial agonists
BVT.13 (2Q6S), nTZDpa (2Q5S), and MRL-24 (2Q5P).13 The full
agonist rosiglitazone (2PRG) is shown for comparison.10 The agonists
BVT.13, nTZDpa, and MRL-24 form a hydrogen bond network
comprising their carboxyl groups, LBD’s Arg288 and Ser342, and water
molecules. These interactions are similar to the interactions of the
amorfrutins. The alternative conformation of Arg288 in the nTZDpa
bound state is marked by red arrows. Only main chain atoms are shown
for residues of the β-sheet region, excepting Glu343. Oxygen atoms are
colored red, nitrogen atoms blue, carbon atoms black, and chloride ions
green. Residue numbers correspond to PPARγ isoform 1. Hydrogen
bonds are labeled with donor−acceptor distances in angstrom.
To a solution of methyl ester 1 (7.50 g, 26.2 mmol) in dry, distilled
toluene (200 mL) under argon in a 1 L Schlenk flask was added
potassium hydride (1.13 g, 28.2 mmol). The solution was stirred for
20 min at 20 °C and then heated to 70 °C. After 20 min at 70 °C the
yellow reaction mixture was cooled to 20 °C, and geranyl chloride dried
over molecular sieves (5.25 g, 5.64 mL, 30.4 mmol) was added. The
mixture was then heated to 80 °C. The reaction was monitored by thin-
layer chromatography using 3:1 hexane/ethyl acetate. After 2 h at 80 °C,
the mixture was cooled to 20 °C, diluted with diethyl ether (250 mL),
washed with three 100 mL portions of saturated aqueous NaHCO3, and
concentrated in vacuo. The crude residue was chromatographed over
silica gel using a hexane/ethyl acetate solvent gradient (0% → 15%),
yielding amorfrutin B methyl ester (5.00 g, 11.8 mmol, 45%) as a white
solid, in addition to 3.26 g (7.7 mmol, 29%) of the O-geranylated
derivative as a yellow oil. 1H NMR (400 MHz, acetone-d6): δ 11.80 (s,
1H, OH), 7.24−7.32 (m, 4H, phenyl), 7.16−7.22 (m, 1H, phenyl), 6.52
(s, 1H, 5-H), 5.20 (m, 1H, CHC(CH3CH2)−CH2−CH2−CH
C(CH3)2), 5.06(m, 1H, CHC(CH3)2), 4.01 (s, 3H, COOCH3), 3.85
(s, 3H, OCH3), 3.32 (d, J = 7.23 Hz, 2H, CH2−CHC(CH3)2), 3.17−
3.24 (m, 2H, CH2−CH2−phenyl), 2.85−2.91 (m, 2H, CH2−CH2−
phenyl), 2.06−2.07 (m, 2H, CH2−CH2−CHC(CH3)2), 1.91−1.98
(m, 2H, CH2−CH2−CHC(CH3)2), 1.77 (br s, 3H, CH3), 1.61 (br s,
3H, CH3), 1.55 (br s, 3H, CH3) ppm. 13C NMR (126 MHz, acetone-d6):
δ 173.04, 162.46, 162.26, 145.29, 143.01, 135.05, 131.56, 129.28, 129.15,
126.67, 125.12, 123.33, 115.54, 107.10, 105.91, 56.01, 52.61, 40.51,
39.96, 39.14, 27.39, 25.80, 22.48, 17.68, 16.17 ppm.
pressure cell disrupter in 20 mM Tris-HCl, 150 mM NaCl, 10% glycerol,
1 mM tris(2-carboxyethyl)phosphine HCl (TCEP), 10 mM imidazole,
pH 8.0, in the presence of protease inhibitors and then centrifuged. The
supernatant was loaded on a 5 mL HisTrap HP column and eluted with an
imidazole gradient. The eluate was diluted to 20 mM NaCl, immediately
loaded onto a MonoQ HR 10/10 column, and eluted with a NaCl
gradient, followed by incubation with thrombin protease (1 U/mg) at
4 °C for 20 h to cleave off the His-tag. His-tag peptides and uncleaved
material were removed by rechromatography with Ni-NTA agarose,
followed by Superdex 75 gel filtration in 20 mM Tris-HCl, 100 mM NaCl,
0.5 mM DTT, and 2 mM EDTA, pH 8.0.
Crystallization. The complex was prepared by adding a 10-fold
molar excess of amorfrutin 1, 2, or B to the purified human PPARγ LBD
1541
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
(E)-3-(3,7-Dimethylocta-2,6-dien-1-yl)-2-hydroxy-4-me-
thoxy-6-phenethylbenzoic Acid (Amorfrutin B).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Thorsten Luhrs for asymmetric flow field-flow
fractionation and Manfred Nimtz and Undine Felgentrager for
̈
̈
mass spectrometry experiments. This work is part of the Ph.D.
theses of J.C.d.G. and C.W. Our work is supported by the
German Ministry for Education and Research (BMBF, Grant
0315082), the National Genome Research Net (NGFN, Grant
01 GS 0828), the European Union ([Grant FP7/2007-2011],
under Grant Agreement No. 262055 (ESGI)), the Max Planck
Society and the Helmholtz Centre for Infection Research.
A solution of amorfrutin B methyl ester (8.57 g, 20.3 mmol) in methanol
(100 mL) was added to a solution of potassium hydroxide (44 g) in a
mixture of methanol (350 mL) and water (50 mL). The mixture was
heated under reflux (78 °C), and progress was monitored by thin-layer
chromatography. After 7 h, the mixture was cooled to 20 °C and
concentrated in vacuo to a volume of 150 mL. The concentrated mixture
was diluted with 200 mL of water, cooled 0 °C, and acidified (to pH 3)
with 2 N aqueous hydrochloric acid. The acidic suspension was
extracted with three 100 mL portions of diethyl ether. The combined
ether extracts were washed with brine, dried over MgSO4, and
evaporated to dryness. The residue was chromatographed over silica
gel using a hexane/ethyl acetate solvent gradient (0% → 20%) and then
recrystallized from hexane/ethyl acetate, yielding amorfrutin B (5.75 g,
14.1 mmol, 70%) as white, sticky crystals. 1H NMR (400 MHz, acetone-
d6): δ 7.22−7.31 (m, 4H, phenyl), 7.14−7.20 (m, 1H, phenyl), 6.49
(s, 1H, 5-H), 5.21 (m, 1H, CHC(CH3CH2)−CH2−CH2−CH
C(CH3)2), 5.06 (m, 1H, CHC(CH3)2), 3.85 (s, 3H, OCH3), 3.32
(d, J = 7.6 Hz, 2H, CH2−CHC(CH3)2), 3.26−3.30 (m, 2H, CH2−
CH2-phenyl), 2.89−2.95 (m, 2H, CH2−CH2−phenyl), 2.06−2.07
(m, 2H, CH2−CH2−CHC(CH3)2), 1.91−1.98 (m, 2H, CH2−
CH2−CHC(CH3)2), 1.77 (br.s, 3H, CH3), 1.69 (d, 3H, CH3), 1.55
(br s, 3H, CH3) ppm. 13C NMR (126 MHz, acetone-d6): δ 174.31,
163.48, 162.32, 145.94, 143.09, 134.95, 131.52, 129.24, 129.09, 126.62,
125.15, 123.32, 115.44, 106.91, 105.27, 56.00, 40.52, 39.93, 39.13, 27.38,
25.79, 22.47, 17.70, 16.21 ppm.
ABBREVIATIONS USED
■
AF, activation function; LBD, ligand binding domain; PPAR,
peroxisome proliferator-activated receptor; RXR, retinoid X
receptor; SPPARγM, selective peroxisome proliferator-activated
receptor γ modulator
REFERENCES
■
(1) Choi, J. H.; Banks, A. S.; Estall, J. L.; Kajimura, S.; Bostrom, P.;
Laznik, D.; Ruas, J. L.; Chalmers, M. J.; Kamenecka, T. M.; Bluher, M.;
Griffin, P. R.; Spiegelman, B. M. Anti-diabetic drugs inhibit obesity-
linked phosphorylation of PPARγ by Cdk5. Nature 2010, 466, 451−456.
(2) Berger, J.; Moller, D. E. The mechanisms of action of PPARs. Annu.
Rev. Med. 2002, 53, 409−435.
(3) Tontonoz, P.; Spiegelman, B. M. Fat and beyond: the diverse
biology of PPARγ. Annu. Rev. Biochem. 2008, 77, 289−312.
(4) Kliewer, S. A.; Sundseth, S. S.; Jones, S. A.; Brown, P. J.; Wisely, G.
B.; Koble, C. S.; Devchand, P.; Wahli, W.; Willson, T. M.; Lenhard, J. M.;
Lehmann, J. M. Fatty acids and eicosanoids regulate gene expression
through direct interactions with peroxisome proliferator-activated
receptors α and γ. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 4318−4323.
(5) Malapaka, R. R.; Khoo, S.; Zhang, J.; Choi, J. H.; Zhou, X. E.; Xu, Y.;
Gong, Y.; Li, J.; Yong, E. L.; Chalmers, M. J.; Chang, L.; Resau, J. H.;
Griffin, P. R.; Chen, Y. E.; Xu, H. E. Identification and mechanism of 10-
carbon fatty acid as modulating ligand of peroxisome proliferator-
activated receptors. J. Biol. Chem. 2012, 287, 183−195.
3-(3-Methyl-2-buten-1-yl)-2-hydroxy-4-methoxy-6-pentyl-
benzoate (Amorfrutin 2).
Amorfrutin 2 was prepared analogous to amorfrutin B, using 2-hydroxy-
4-methoxy-6-pentylbenzoate instead of 2-hydroxy-4-methoxy-6-phene-
thylbenzoate (compound 1). 1H NMR (600 MHz, acetone-d6): δ 6.49
(s, 1H, 5-H), 5.18 (m, 1H, CHC(CH3)2), 3.89 (s, 3H, OCH3), 3.29
(br. d, J = 7.6 Hz, 2H, CH2−CHC(CH3)2), 2.97−3.00 (m, 2H, CH2−
CH2−phenyl), 1.74 (br s, 3H, CH3), 1,58−1.63 (m, 5H, CH2−CH2-
phenyl and CH3), 1.32−1.39 (m, 4H, CH3−CH2−CH2), 0.87−0.91 (m,
3H, CH3−CH2) ppm. 13C NMR (126 MHz, acetone-d6): δ 174.4, 163.3,
162.2, 147.2, 131.2, 123.5, 115.0, 106.5, 105.3, 56.0, 37.6, 32.8, 32.7,
25.8, 23.2, 22.4, 17.8, 14.4 ppm.
(6) Murphy, G. J.; Holder, J. C. PPAR-γ agonists: therapeutic role in
diabetes, inflammation and cancer. Trends Pharmacol. Sci. 2000, 21,
469−474.
(7) Varga, T.; Czimmerer, Z.; Nagy, L. PPARs are a unique set of fatty
acid regulated transcription factors controlling both lipid metabolism
and inflammation. Biochim. Biophys. Acta 2011, 1812, 1007−1022.
(8) Cronet, P.; Petersen, J. F.; Folmer, R.; Blomberg, N.; Sjoblom, K.;
Karlsson, U.; Lindstedt, E. L.; Bamberg, K. Structure of the PPARα and
-γ ligand binding domain in complex with AZ 242; ligand selectivity and
agonist activation in the PPAR family. Structure 2001, 9, 699−706.
(9) Ebdrup, S.; Pettersson, I.; Rasmussen, H. B.; Deussen, H. J.; Frost
Jensen, A.; Mortensen, S. B.; Fleckner, J.; Pridal, L.; Nygaard, L.;
Sauerberg, P. Synthesis and biological and structural characterization of
the dual-acting peroxisome proliferator-activated receptor α/γ agonist
ragaglitazar. J. Med. Chem. 2003, 46, 1306−1317.
(10) Nolte, R. T.; Wisely, G. B.; Westin, S.; Cobb, J. E.; Lambert, M. H.;
Kurokawa, R.; Rosenfeld, M. G.; Willson, T. M.; Glass, C. K.; Milburn,
M. V. Ligand binding and co-activator assembly of the peroxisome
proliferator-activated receptor-γ. Nature 1998, 395, 137−143.
(11) Nagy, L.; Kao, H. Y.; Love, J. D.; Li, C.; Banayo, E.; Gooch, J. T.;
Krishna, V.; Chatterjee, K.; Evans, R. M.; Schwabe, J. W. Mechanism of
corepressor binding and release from nuclear hormone receptors. Genes
Dev. 1999, 13, 3209−3216.
ASSOCIATED CONTENT
* Supporting Information
■
S
Sequence alignment of the human PPAR LBDs; description of
the inactive conformation of the PPARγ LBD. This material is
available free of charge via the Internet at http://pubs.acs.org.
Accession Codes
PDB codes are the following: 4A4V for amorfrutin 2:PPARγ;
4A4W for amorfrutin B:PPARγ.
AUTHOR INFORMATION
Corresponding Author
■
(12) Nagy, L.; Schwabe, J. W. Mechanism of the nuclear receptor
molecular switch. Trends Biochem. Sci. 2004, 29, 317−324.
(13) Bruning, J. B.; Chalmers, M. J.; Prasad, S.; Busby, S. A.;
Kamenecka, T. M.; He, Y.; Nettles, K. W.; Griffin, P. R. Partial agonists
*For S.S.: phone, +49 30 8413-1661; e-mail, sauer@molgen.
mpg.de. For K.B.: phone, +49 531 6181-4550; e-mail, konrad.
buessow@helmholtz-hzi.de.
1542
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543

Journal of Medicinal Chemistry
Article
activate PPARγ using a helix 12 independent mechanism. Structure
2007, 15, 1258−1271.
(14) Rangwala, S. M.; Lazar, M. A. The dawn of the SPPARMs? Sci.
STKE 2002, 2002, pe9.
(15) Weidner, C.; de Groot, J. C.; Prasad, A.; Freiwald, A.; Quedenau,
C.; Kliem, M.; Witzke, A.; Kodelja, V.; Han, C. T.; Giegold, S.; Baumann,
M.; Klebl, B.; Siems, K.; Muller-Kuhrt, L.; Schurmann, A.; Schuler, R.;
̈
̈
̈
Pfeiffer, A. F.; Schroeder, F. C.; Bussow, K.; Sauer, S. Amorfrutins are
̈
potent antidiabetic dietary natural products. Proc. Natl. Acad. Sci. U.S.A.
2012, 109, 7257−7262.
(16) Lefebvre, P.; Staels, B. Naturally improving insulin resistance with
amorfrutins. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 7136−7137.
(17) Mitscher, L. A.; Park, Y. H.; Al-Shamma, A.; Hudson, P. B.; Haas,
T.; Amorfrutin, A. and B, bibenzyl antimicrobial agents from Amorpha
fruticosa. Phytochemistry 1981, 20, 781−785.
(18) Kabsch, W. Automatic processing of rotation diffraction data from
crystals of initially unknown symmetry and cell constants. J. Appl.
Crystallogr. 1993, 26, 795−800.
(19) Vagin, A.; Teplyakov, A. MOLREP: an automated programm for
molecular replacement. J. Appl. Crystallogr. 1997, 30, 1022−1025.
(20) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Refinement of
macromolecular structures by the maximum-likelihood method. Acta
Crystallogr. D 1997, 53, 240−255.
(21) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis, I.
W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; Grosse-
Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R. J.;
Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H.
PHENIX: a comprehensive Python-based system for macromolecular
structure solution. Acta Crystallogr. D 2010, 66, 213−221.
(22) Emsley, P.; Cowtan, K. Coot: model-building tools for molecular
graphics. Acta Crystallogr. D 2004, 60, 2126−2132.
(23) The PyMOL Molecular Graphics System, version 1.3r1;
Schrodinger, LLC: New York, 2010.
(24) Wallace, A. C.; Laskowski, R. A.; Thornton, J. M. LIGPLOT: a
program to generate schematic diagrams of protein-ligand interactions.
Protein Eng. 1995, 8, 127−134.
(25) McLachlan, A. D. Rapid comparison of protein structures. Acta
Crystallogr. A 1982, 38, 871−873.
(26) Vaguine, A. A.; Richelle, J.; Wodak, S. J. SFCHECK: a unified set
of procedures for evaluating the quality of macromolecular structure-
factor data and their agreement with the atomic model. Acta Crystallogr.
D 1999, 55, 191−205.
1543
dx.doi.org/10.1021/jm3013272 | J. Med. Chem. 2013, 56, 1535−1543