Co-crystal structure guided array synthesis of PPARγ inverse agonists

Article abstract of DOI:10.1016/j.bmcl.2007.04.111

PPARγ-activating thiazolidinediones and carboxylic acids such as farglitazar exert their anti-diabetic effects in part in PPARγ rich adipose. Both pro- and anti-adipogenic PPARγ ligands promote glucose and lipid lowering in animal models of diabetes. Herein, we disclose representatives of an array of 160 farglitazar analogues with atypical inverse agonism of PPARγ in mature adipocytes.

Full text of DOI:10.1016/j.bmcl.2007.04.111

Bioorganic & Medicinal Chemistry Letters 17 (2007) 3916–3920
Co-crystal structure guided array synthesis of PPARc
inverse agonists
Ryan P. Trump, Jeffrey E. Cobb, Barry G. Shearer, Millard H. Lambert, Robert T. Nolte,
Timothy M. Willson, Richard G. Buckholz, Sumin M. Zhao, Lisa M. Leesnitzer,
Marie A. Iannone, Kenneth H. Pearce, Andrew N. Billin and William J. Hoekstra*
Drug Discovery, GlaxoSmithKline, Research Triangle Park, NC 27709, USA
Received 8 March 2007; revised 27 April 2007; accepted 30 April 2007
Available online 18 May 2007
Abstract—PPARc-activating thiazolidinediones and carboxylic acids such as farglitazar exert their anti-diabetic effects in part in
PPARc rich adipose. Both pro- and anti-adipogenic PPARc ligands promote glucose and lipid lowering in animal models of dia-
betes. Herein, we disclose representatives of an array of 160 farglitazar analogues with atypical inverse agonism of PPARc in mature
adipocytes.
Ó 2007 Elsevier Ltd. All rights reserved.
Peroxisome proliferator-activated receptor-gamma
(PPARc) agonists such as the thiazolidinedione (TZD)
drugs and related carboxylic acids such as farglitazar
(GI262570) are effective insulin sensitizers in patients with
type 2 diabetes mellitus.1,2 Insulin resistance is an early
marker of this type of adult onset diabetes. Oral adminis-
tration of PPARc agonists to these patients reverses
hyperglycemia and hyperinsulinemia by increasing glu-
cose metabolism in muscle and reducing glucose biosyn-
thesis in the liver. PPARc also plays a critical role in
adipocyte function. PPARc agonists promote differentia-
tion of pre-adipocytes to mature adipocytes as measured
by lipid accumulation, changes in cell morphology, and
the expression of adipocyte-specific genes.3
Adipogenesis involves a complex interplay between
PPARc and a variety of genes, notably fatty acid bind-
ing protein-4 (FABP4).4 Gene knockout studies have
demonstrated that PPARc is essential for adipocyte dif-
ferentiation in vivo.5,6 Paradoxically, the PPARc anta-
gonist GW9662 exerts a modest insulin sensitizing
effect but, unlike agonists, decreases weight gain in obese
mice.7 These results are consistent with knockout of
adipocytic PPARc in a mouse study.8 Despite these
reports, little is known about the interruption of
PPARc-induced target gene expression.
O
O
R1
N
H
H
HN
N
O
CO2H
O
N
O
H
HN
O
Ph
O
2
R
R = Ph
R = OMe
Farglitazar (GI262570)
GW7845
To date, ligands have been identified primarily for their
propensity to stimulate or inhibit adipogenesis, or re-
cruit co-activators to PPARc.2,9 TZDs and carboxylic
acids (e.g., farglitazar) form a network of hydrogen
bonds with Tyr-473, His-323, His-449, and Ser-289 adja-
cent to the AF2 helix.1,10 Herein, we have devised and
synthesized a mini-library of 160 diverse analogues 2
Keywords: PPARc; Inverse agonist.
Corresponding author. Fax: +1 919 315 0430; e-mail: william.j.
hoekstra@gsk.com
*
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.04.111

R. P. Trump et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3916–3920
3917
Table 2. PPARc CV-1 cell GAL4 reporter activity of representative
amides
of the PPARc agonist farglitazar to attempt to exert dif-
ferential effects at the AF-2 helix domain, thus leading
to some atypical cofactor peptide interactions.11,12
These compounds differentially recruited a set of 49
coactivator or corepressor-derived peptides to PPARc
ligand binding domain (LBD) in a multiplexed fluores-
cent microsphere based binding assay.11,13 We deter-
mined the functional profiles of these PPARc ligands
in both a GAL4-based reporter assay and mature adipo-
cytes. The synthetic approaches to these ligands and bio-
logical results are revealed below.
GAL4 PPARc
a
b
Compound
pEC50
%max
pIC50
%max
2g
<5.0
<5.0
<5.0
<5.0
6.2
—
—
—
—
29
27
37
92
5.2
5.2
5.7
6.0
5.6
5.6
5.8
<5.0
55
48
80
92
59
57
54
—
2i
2n
2o
2q
2s
6.2
6.0
2t
GW7845
9.2
Amide 2t (R1 = Me) was prepared as a singleton to aid
in design of the array. Following X-ray co-crystal anal-
ysis of 2t, amides 2 were prepared as described below.16
Optically pure farglitazar was converted to its corre-
sponding activated ester with isobutyl chloroformate
and reduced to the alcohol with sodium borohydride.
This alcohol was treated with diphenylphosphoryl azide
and subsequently reduced to its corresponding amine by
hydrogenation over 10% palladium-on-carbon; the
amine was protected in situ using di-tert-butyl-dicarbon-
ate to afford shelf-stable 1. N-Boc-protected amine 1 was
treated with hydrochloric acid and then quickly acylated
with a variety of polystyrene-bound activated hydrox-
ybenzotriazole esters of carboxylic acids to furnish an
array of 160 farglitazar inverse amide products 2. For
products containing terminal carboxylic acids or
amines, t-butyl ester or N-tert-butoxycarbonyl protec-
tion, respectively, was employed and cleaved as a final
step in a hydrochloric acid/dioxane medium.
a Assay measures transactivation of human PPARc in CV-1 cells
(n P 2, SD = 0.10).15
b Assay measures the inhibition of BRL-49653 stimulated (100 nM)
transactivation of human PPARc in CV-1 cells (n P 2, SD = 0.15).9
occupy minimal AF2 domain space. Thus, partial intrin-
sic efficacy was observed for this class of amides in the
GAL4 reporter assay. Conversely, larger hydrogen-
bonding amines and carboxylic acids were designed to
displace the AF2 helix and occupy the coactivator cleft
to interrupt coactivator binding (2g, 2i, 2n, and 2o).
Herein, these amides exhibited antagonist GAL4 activ-
ity (Fig. 1).
We also examined the co-regulator affinity signatures of
the 102 array members with pKi > 6.0 using a multi-
plexed interaction assay to determine PPARc LBD
interactions with 49 peptides derived from the binding
motifs of known co-regulator proteins.11,13 In the assay,
mean fluorescence intensity units (MFI) correlate with
affinity of peptide binding. TRAP-220 is a co-activator
that plays a central role in PPARc-mediated adipocyte
gene expression pathways. While the farglitazar/PPARc
complex bound tightly to the TRAP-220 derived peptide
(MFI = 499),13 amide analogues 2i (MFI = À99) and 2n
(MFI = À296) exhibited markedly reduced, atypical co-
activator peptide binding relative to apo-PPARc
(MFI = 0) and a neutral effect on corepressor binding.11
Based on these unusual PPARc/co-regulator interaction
Farglitazar analogues 2 were tested in PPAR scintilla-
tion proximity assay (SPA) binding and a transiently
transfected PPARc CV-1 cell GAL4 reporter assay (Ta-
bles 1 and 2).15 Approximately two-thirds of the array
members exhibited PPARc pKi > 6.0 (102/160). All
compounds exhibited >5-fold binding selectivity for
PPARc (PPARa, PPARd data not shown). Small
hydrophobic amides furnished high PPARc affinity
(2p–t). Carboxylic acids exhibited somewhat lower affin-
ity (2k–o). Small hydrophobic amides were devised to
Table 1. Monomer examples (R1CO2H) selected from co-crystal structure analysis for acylationa of a primary amine analogue of farglitazar that
furnished amides 2a–t with PPARc pKi > 6b
R1 Groups
Alcohols
pKic
Amines
pKic
a CH2NHCO-3-C6H4CH2OH
b CH2OH
c 4-C6H4CH2OH
7.00
6.95
6.80
6.76
6.51
f 3-piperidinyl
g 3-C6H4-O(CH2)2NMe2
h 2-pyridyl
7.29
7.00
6.78
6.72
6.71
d CH2NHCO-4-C6H4OH
e 4-c-C6H10OH
i CH2O-3-C6H4NH2
j 4-c-C6H10CH2NH2
Carboxylic acids
pKic
Hydrophobes
pKic
k 2-C6H4CO2H
l CH(Me)CH2CH(Me)CO2H
m (CH2)2CO2H
6.94
6.84
6.57
6.60
6.04
p 4-C6H4OMe
q c-C3H5
7.32
7.26
7.21
7.06
6.86
r (CH2)5Me
s (CH2)2Me
t Me
n (CH2)2-4-C6H4SC(Me)2CO2H
o CH2CH(Me)CO2H
a See conversion of 1–2 in Scheme 1.
b Listed in descending order of PPARc affinity for each of the four functional classifications, n = 2.
c The scintillation proximity assay measures compound interaction with the ligand binding domain of PPARc by displacement of tritiated BRL-
49653 (n P 2, SD = 0.15).15

3918
R. P. Trump et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3916–3920
Figure 1. X-ray co-crystal structure of farglitazar bound to the PPARc
LBD overlaid on the co-crystal structure of 2t (R1 = Me) bound to the
PPARc LBD.14 Whereas the farglitazar carboxylate forms a network
of hydrogen bonds with His-323, His-449 (side chain above/behind
label), and the AF2 helix’s Tyr-473, the amide of 2t merely approaches
His-323 and Tyr-473, without actually making any strong hydrogen
bonds. The AF2 helix is modestly unwound (not shown) consistent
with some destabilization of the activated conformation of the AF2
helix.
Figure 2. Gene expression by PPARc agonist GW7845, partial agonist
GW0072,9 and inverse agonists GSK5737 (2n) and GSK5775 (2i) in
mature 3T3L1 adipocytes.18
profiles,13 the phenotypes of these ligands and other
array members in Table 2 were characterized in a ma-
ture 3T3L1 adipocyte gene expression assay to compare
them to ligands with known activities. The highly effica-
cious agonist GW7845 positively regulated six of seven
structures. These amides were evaluated in PPARc bind-
ing and GAL4 reporter assays, and a mature 3T3L1 adi-
pocyte gene expression assay. Strikingly, sterically bulky
members of this array (2i, 2n), designed to repress coac-
tivator binding in the AF2 helix domain, suppressed
induction of adipogenic genes relative to basal
expression levels.11 Low efficacy partial agonists
(e.g., GW0072)9 and antagonists (e.g., GW9662,7
T0070907,19 PD-068235)20 of PPARc transcription that
inhibit PPARc agonist-induced adipogenesis yet do not
invert the PPARc phenotype in adipocytes have been
disclosed in the literature. In our adipocyte assay, the
competitive antagonist GW9662 exhibited a neutral,
marker genes of fatty acid synthesis, transport, storage,
17
and oxidation (Fig. 2).
Partial activators regulated
these same genes less efficaciously.11 By contrast, inverse
agonists GSK5737 (2n) and GSK5775 (2i) exhibited
atypical activity in that genes were driven to expression
levels below basal PPARc.11 For example, the PEPCK
and FABP4 genes were down-regulated significantly
(Scheme 1).
In summary, a synthetic array of farglitazar analogues
was synthesized with guidance from X-ray co-crystal
H
H
1) iBuOCOCl, NMM; NaBH4
CO2H
O
N
NHBoc
O
N
Ph
HN
Ph
Ph
HN
Ph
2) (PhO)2PON3, DBU
O
O
O
O
3) H2, 10% Pd/C; Boc2O
1
Farglitazar
O
H
R1
O
1) HCl, dioxane
N
H
Ph
HN
2) R1CO2H, DIC, PS-HOBT
O
N
O
Ph
2
Scheme 1. Synthesis of inverse amide analogues of farglitazar.

R. P. Trump et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3916–3920
3919
pound (at 50 mM in DMSO) using the vapor phase
diffusion method. The two co-crystals’ PDB codes are
2pob (2t) and 1fm9 (farglitazar).
basal-like effect on PPARc target gene expression. Thus,
inverse agonists of PPARc-mediated activity (e.g., 2n)
herein appear to be uncommon. These amides are repre-
sentatives of a new class of ligands that are available to
evaluate co-activator destabilization and inverse agon-
ism at PPARc in biological models of diabetes, obesity,
and other diseases.21
15. Henke, B. R.; Blanchard, S. G.; Brackeen, M. F.; Brown,
K. K. J. Med. Chem. 1998, 41, 5020.
16. Array members were prepared as described for compound
2t: To a stirred, cooled (À15 °C) solution of farglitazar
(compound 20 described in reference 15; 5.00 g,
9.15 mmol) in DME (50 mL) were added N-methylmorpho-
line (0.92 g, 9.15 mmol) and isobutyl chloroformate
(1.26 g, 9.25 mmol). This suspension was stirred for
5 min, filtered, and cooled (À15 °C). To this solution
was added a solution of NaBH4 (0.35 g, 9.30 mmol) in
water (5 mL). After 5 min the reaction mixture was diluted
with water (250 mL) and warmed to RT. The mixture was
poured into HCl (1 N; 250 mL) and extracted with DCM.
The organic layer was washed with brine and dried over
MgSO4. Removal of solvent in vacuo and recrystallization
of the resultant solid from EtOAc/hexane (1:1, 250 mL)
provided the desired primary alcohol as a yellow solid
(3.88 g, 80% yield). To a stirred solution of the yellow solid
(0.50 g, 0.94 mmol) and diphenylphosphoryl azide (1.28 g,
4.64 mmol) in DMF (30 mL) was added DBU (0.76 g,
5.02 mmol). The mixture was heated at 90 °C for 18 h,
cooled to rt, and the DMF removed in vacuo. The residue
was partitioned between EtOAc and HCl (1 N), and the
organic layer washed with water, saturated NaHCO3, and
brine, dried over MgSO4, and concentrated under reduced
pressure to give a solid. The crude material was purified by
silica gel chromatography eluting with 1:4 ethyl acetate/
hexanes to give the azide as a yellow solid (0.48 g, 91%
yield). The azide (0.39 g, 0.70 mol) was dissolved in THF
(10 mL) under nitrogen, and di-tert-butyl-dicarbonate
(0.46 g, 2.12 mmol) and 10% Pd/C (0.070 g) were added.
The solution was stirred under one atmosphere of hydro-
gen for 18 h, filtered (Celite), and concentrated in vacuo to
furnish crude, oily 1. This oil was purified by silica gel
chromatography (2:1 EtOAc/hexanes) to give intermediate
Acknowledgments
We thank Jeffrey D. Harless, Julie B. Stimmel, and
James M. Way for helpful discussions.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2007.04.111.
References and notes
1. Willson, T. M.; Brown, P. J.; Sternbach, D. D.; Henke, B.
R. J. Med. Chem. 2000, 43, 527.
2. Shearer, B. G.; Hoekstra, W. J. Curr. Med. Chem. 2003,
10, 267.
3. Tontonoz, P.; Hu, E.; Spiegelman, B. M. Cell 1994, 79,
1147.
4. Lowell, B. B. Cell 1999, 99, 239.
5. Barak, Y.; Nelson, M. C.; Ong, E. S.; Jones, Y. Z.; Ruiz-
Lozano, P.; Chien, C. R.; Koder, A.; Evans, R. M. Mol.
Cell 1999, 4, 585.
6. Rosen, E. D.; Sarraf, P.; Troy, A. E.; Bradwin, G.; Moore,
K.; Milstone, D. S.; Spiegelman, B. M.; Mortensen, R. M.
Mol. Cell 1999, 4, 611.
1
1 as a yellow solid (0.22 g, 50% yield): H NMR (CDCl3,
300 MHz) d 8.6 (d, J = 8.5 Hz, 1H), 8.0 (m, 2H), 7.6 (m,
2H), 7.5–7.2 (m, 8H), 7.1 (m, 2H), 6.8 (m, 1H), 6.8 (m,
2H), 6.5 (t, J = 7.4, 1H), 4.8 (br s, 1H), 4.2 (t, J = 6.6, 2 H),
4.0 (m, 1H), 3.5 (m, 1H), 3.3 (m, 1H), 3.0 (t, J = 6.6, 2H),
2.8 (m, 2H), 2.4 (s, 3 H), 1.4 (s, 9H). MS (APCI) m/z 632
(M+1). Intermediate 1 (0.22 g, 0.35 mmol) was dissolved
in HCl (4 M dioxane, 4.0 mL). The solution was stirred for
1 h and concentrated in vacuo to a yellow solid (0.20 g,
99% yield). Polystyrene-bound hydroxybenzotriazole
(0.029 g, 0.030 mmol) was suspended in DMF
(1 mL). Reagent monomer example acetic acid (0.0027 g,
0.045 mmol), dimethylaminopyridine (0.0022 g, 0.018 mmol),
and diisopropylcarbodiimide (0.017 g, 0.135 mmol) were
added, and the resulting suspension was shaken for 2 h.
The resin was filtered, washed with DCM, and suspended
in DCM (1 mL). The yellow, de-protected HCl salt of
intermediate 1 (0.012 g, 0.021 mmol) and polystyrene-
bound N-methylmorpholine (0.022 g, 0.09 mmol) were
added to the suspension. After 2 h of rotation, the organic
solution was collected by filtration and concentrated in
vacuo to give a crude oil that was purified by silica gel
chromatography to furnish 2t as a yellow solid (0.009 g,
75% yield): 1H NMR (CDCl3, 300 MHz) d 8.1 (d, J = 6.1,
2H), 7.5 (m, 2H), 7.4 (m, 6H), 7.3 (m, 1H), 7.2 (s, 3H), 7.0
(m, 2H), 6.8 (d, J = 8.6, 1H), 6.7 (d, J = 8.6, 2H), 4.2 (t,
J = 6.1, 2H), 3.0 (d, J = 5.8, 2H), 2.8 (m, 1H), 2.4 (s, 3H),
2.0 (s, 1H), 1.9 (s, 3H), 1.8 (m, 2H), 1.2 (m, 1H). MS
(APCI) m/z 574 (M+1).
7. Nakano, R.; Kurosaki, E.; Yoshida, S.; Yokono, M.;
Shimaya, A.; Maruyama, T.; Shibasaki, M. Biochem.
Pharm. 2006, 72, 42.
8. Jones, J. R.; Barrick, C.; Kim, K.-A.; Lindner, J.;
Blondeau, B.; Fujimoto, Y.; Shiota, M.; Kesterson, R.
A.; Kahn, B. B.; Magnuson, M. A. Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 6207.
9. Oberfield, J.; Collins, J. L.; Holmes, C. P.; Goreham, D.
M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6102.
10. Xu, H. E.; Lambert, M. H.; Montana, V. G.; Plunket, K.
D.; Moore, L. B.; Collins, J. L.; Oplinger, J.; Kliewer, S.
A.; Gampe, R. T.; McKee, D. D.; Willson, T. M. Proc.
Natl. Acad. Sci. U.S.A. 2001, 98, 13919.
11. Buckholz, R. G.; Trump, R. P.; Zhao, S.; Iannone, M. A.;
Leesnitzer, L. M.; Merrihew, R. V.; Lobe, D. C.; Winegar,
D. A.; Lazar, M. A.; Cobb, J. E.; Willson, T. M.;
Hoekstra, W. J.; Billin, A. B. Biochem. Biophys. Res.
Commun., submitted for publication; also see supplemen-
tal material.
12. Xu, H. E.; Stanley, T. B.; Montana, V. G.; Lambert, M.
H. Nature 2002, 415, 813.
13. Iannone, M. A.; Simmons, C. A.; Kadwell, S. H.;
Svoboda, D. L.; Vanderwall, D. E.; Deng, S. J.; Consler,
T. G.; Shearin, J.; Gray, J. G.; Pearce, K. H. Mol.
Endocrinol. 2004, 18, 1064.
14. Co-crystals of compound 2t with the PPARc LBD were
grown from a complex containing 15 mg/ml protein
(20 mM Tris, pH 8.0, 5 mM DTT, 100 mM NaCl, 2 mM
EDTA) complexed at a 5:1 molar ratio with the com-
17. For more information on these genes, see Way, J. M.;
Harrington, W. W.; Brown, K. K.; Gottschalk, W. K.;

3920
R. P. Trump et al. / Bioorg. Med. Chem. Lett. 17 (2007) 3916–3920
Sundseth, S. S.; Mansfield, T. A.; Ramachandran, R. K.;
Willson, T. M.; Kliewer, S. A. Endocrinology 2001, 142,
1269.
shown in Table 2 were run in this assay, results for other
high MW array members with an arene/two-carbon tether
motif are not available.
18. Adipocytes were treated with differentiation cocktail to
generate mature adipocytes, and then dosed with test
compounds at the indicated concentrations. GW7845
(1 lM) displays agonism on the tested genes, whereas
GW0072 (1 lM) acts as a partial agonist. Contrast this
with the inverse agonism exhibited for most of the genes
by GSK5737 (10 lM) and GSK5775 (10 lM): (*) p < 0.01,
(#) p > 0.01 versus vehicle.11 Because only array members
19. Lee, G.; Elwood, F.; McNally, J.; Weiszmann, J.; Lind-
strom, M.; Amaral, K.; Nakamura, M.; Miao, S.; Cao, P.;
Learned, R. M.; Chen, J.-L.; Li, Y. J. Biol. Chem. 2002,
277, 19649.
20. Camp, H. S.; Chaudhry, A.; Leff, T. Endocrinology 2001,
142, 3207.
21. The study of a PPARc inverse agonist in a diabetic animal
model will be disclosed in a future communication.