Structure-activity relationships of rosiglitazone for peroxisome proliferator-activated receptor gamma transrepression

Article abstract of DOI:10.1016/j.bmcl.2017.04.061

Anti-inflammatory effects of peroxisome proliferator-activated receptor gamma (PPRAγ) ligands are thought to be largely due to PPARγ-mediated transrepression. Thus, transrepression-selective PPARγ ligands without agonistic activity or with only partial agonistic activity should exhibit anti-inflammatory properties with reduced side effects. Here, we investigated the structure-activity relationships (SARs) of PPARγ agonist rosiglitazone, focusing on transrepression activity. Alkenic analogs showed slightly more potent transrepression with reduced efficacy of transactivating agonistic activity. Removal of the alkyl group on the nitrogen atom improved selectivity for transrepression over transactivation. Among the synthesized compounds, 3l exhibited stronger transrepressional activity (IC₅₀: 14?μM) and weaker agonistic efficacy (11%) than rosiglitazone or pioglitazone.

Full text of DOI:10.1016/j.bmcl.2017.04.061

Accepted Manuscript
Structure-activity relationships of rosiglitazone for peroxisome proliferator-ac-
tivated receptor gamma transrepression
Yosuke Toyota, Sayaka Nomura, Makoto Makishima, Yuichi Hashimoto,
Minoru Ishikawa
PII:
S0960-894X(17)30438-9
DOI:
http://dx.doi.org/10.1016/j.bmcl.2017.04.061
BMCL 24914
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
29 March 2017
18 April 2017
19 April 2017
Please cite this article as: Toyota, Y., Nomura, S., Makishima, M., Hashimoto, Y., Ishikawa, M., Structure-activity
relationships of rosiglitazone for peroxisome proliferator-activated receptor gamma transrepression, Bioorganic &
Medicinal Chemistry Letters (2017), doi: http://dx.doi.org/10.1016/j.bmcl.2017.04.061
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.

Structure-activity
relationships
of
rosiglitazone
for
peroxisome
proliferator-activated receptor gamma transrepression
Yosuke Toyota,a Sayaka Nomura,a Makoto Makishima,b Yuichi Hashimoto,a Minoru
Ishikawaa*
a Institute of Molecular and Cellular Biosciences, The University of Tokyo, 1-1-1 Yayoi,
Bunkyo-ku, Tokyo 113-0032, Japan
b Division of Biochemistry, Department of Biomedical Sciences, Nihon University School
of Medicine, Itabashi-ku, Tokyo 173-8610, Japan
Corresponding author: Minoru ISHIKAWA
Institute of Molecular and Cellular Biosciences, The University of Tokyo
1-1-1, Yayoi, Bunkyo-ku, Tokyo, 113-0032, Japan
Tel.: +81 3 5841 7853; fax: +81 3 5841 8495
E-mail address: m-ishikawa@iam.u-tokyo.ac.jp

Abstract
Anti-inflammatory effects of peroxisome proliferator-activated receptor gamma
(PPRAγ) ligands are thought to be largely due to PPARγ-mediated transrepression. Thus,
transrepression-selective PPARγ ligands without agonistic activity or with only partial
agonistic activity should exhibit anti-inflammatory properties with reduced side effects.
Here, we investigated the structure-activity relationships (SARs) of PPARγ agonist
rosiglitazone, focusing on transrepression activity. Alkenic analogs showed slightly more
potent transrepression with reduced efficacy of transactivating agonistic activity.
Removal of the alkyl group on the nitrogen atom improved selectivity for transrepression
over transactivation. Among the synthesized compounds, 3l exhibited stronger
transrepressional activity (IC50: 14 µM) and weaker agonistic efficacy (11%) than
rosiglitazone or pioglitazone.

Nuclear receptors (NRs) are ligand-dependent transcription factors that are involved in
control of diverse biological functions, including reproduction, differentiation,
homeostasis, and immunity. Upon binding of an agonist to the ligand-binding domain
(LBD), the NR heterodimer binds to NR response elements (NRE) in promoter regions of
specific genes, and helix 12 in the LBD adopts a conformation that closes the ligand
binding pocket and forms a groove to which coactivators can bind, allowing gene
transcription to occur. This ligand-mediated activation of NRE-dependent gene
transcription is called transactivation (Fig. 1a). On the other hand, binding of antagonists
results in a conformation that favors binding of corepressors.
Among the NRs, peroxisome proliferator-activated receptors (PPARs) include three
subtypes, PPARα, PPARγ and PPARδ. PPARγ plays a critical role in the differentiation
of preadipocytes to adipocytes, promotes lipid storage, and enhances glucose disposal to
peripheral tissues through binding to PPAR response elements (PPRE) in the promoter
region of target genes. Thiazolidinediones (TZDs), such as pioglitazone (1) and
rosiglitazone (2), are PPARγ agonists used clinically to treat type 2 diabetes by enhancing
insulin sensitivity in target tissues and lowering glucose and fatty acid levels (Fig. 2).
However, despite their proven benefits, clinical application of these drugs has been
plagued by adverse effects, such as weight gain, increased rate of bone fractures, fluid
accumulation, and pulmonary edema, leading to increased frequency of congestive heart
failure.1,2 These adverse effects are at least partially associated with transactivation of
PPARγ.1,3 Therefore, partial agonists have been developed with the aim of retaining the

beneficial effects while diminishing the adverse effects of full agonists.4,5 Indeed, phase
II clinical trials showed that metaglidasen, a PPARγ partial agonist, significantly
improved metabolic parameters without the side effects of fluid retention/edema or
weight gain.2
Fig. 1. Proposed mechanisms of a) transactivation and b) transrepression of PPARγ.
Fig. 2. Chemical structures of PPARγ agonists.

Recent studies have revealed that PPARγ also functions in the regulation of inflammation,
by transrepression of proinflammatory genes. For example, clinical efficacy of TZDs on
asthma6 and rheumatoid arthritis7 has been reported. Several PPARγ agonists inhibit the
expression of proinflamatory mediators such as interleukin-6 (IL-6), inducible nitric
oxide synthase (iNOS) and tumor necrosis factor α (TNFα) in macrophages stimulated
by bacterial infection, TNFα, or lipopolysaccaride (LPS) in vitro.8,9,10 In the absence of
ligand-bound PPARγ, other transcription factors such as NF-κB or AP-1 are able to
transactivate target genes through binding to NF-κB response element (NFκBRE) or
AP-1 response element in the promoter regions of the genes. In contrast, SUMOylation of
PPARγ induced by binding of a PPARγ ligand stabilizes the transrepressional complex,
including PPARγ and co-repressor, and prevents NF-κB and AP-1 from binding the target
genes, thereby inhibiting expression of these genes11 (Fig. 1b). This indirect mechanism
of action is referred to as transrepression. Because the promoter regions of TNF α, IL-6,
iNOS, and other proinflammatory genes do not contain PPRE, the mechanisms of the
transactivating and transrepressional actions are clearly different.
Transrepression-selective PPARγ ligands (without agonistic activity or with only partial
agonistic activity) should exhibit anti-inflammatory properties with reduced side effects.
However, the SARs of rosiglitazone (2) focusing on PPARγ transrepression and those of
transrepression-selective PPARγ ligands have not yet been fully investigated.12
Benzothiazolylamino analogs 4f and 4g were reported as PPARγ full agonists with

transrepressional activity.13 On the other hand, transrepressional activities of other NRs,
glucocorticoid receptor (GR) and liver X receptor (LXR),14 have been reported.15
Separation of transactivating and transrepressional activities was first investigated in the
case of GR (Fig. 3).16,17,18 We have also reported two distinct chemical classes of
transrepression-selective LXR ligands.19,20 We hypothesized that the transactivating and
transrepressional activities of PPARγ are also separable. Here, we report SARs of PPARγ
agonist rosiglitazone (2), focusing on transrepression.
Fig. 3. Transrepression-selective GR and LXR ligands
We selected rosiglitazone (2) as a lead compound, because its transrepressional activity
has been extensively investigated, including clinical studies. Surprisingly, however, there
is relatively little information on SARs for the transactivating activity (PPARγ-mediated

transcription) of rosiglitazone,21,22,13,23 probably because the molecular target of TZDs
was unknown when these compounds were discovered. Acidic hydrogen on the
thiazolidinedione ring of 2 forms a hydrogen bond with the phenol group of Tyr473 on
helix 12 (Fig. 4a), and this interaction is important for PPARγ agonistic activity and
receptor binding.23 Thus, this acidic hydrogen on the thiazolidinedione ring was fixed
with the aim of maintaining PPARγ binding affinity. On the other hand, the pocket to
which the pyridyl group binds is relatively wide and includes unoccupied space (Fig. 4b).
Therefore, we focused on the methylpyridylamino moiety, and designed several
heterocyclic analogs including a benzothiazole analog.13 In addition, introduction of a
double bond at the 5-position of the thiazolidinedione was planned to remove the
asymmetric carbon, which is prone to isomerization.24
Figure 4. X-Ray crystal structure of PPARγ (green) and rosiglitazone (white stick)
complex (PDB ID: 1FM6). Helix 12 is shown in cartoon form. a) Only amino acids
within 8 angstroms of the compound are shown. The important hydrogen bond is shown
as a yellow line. Tyr473 is shown in stick form. b) The binding pocket of the pyridyl
group of rosiglitazone.

Rosiglitazone analogs 3e-l and 4e-h were synthesized as outlined in Scheme 1.
Nucleophilic substitution of aryl chlorides with aminoethanols 5a-d afforded 6e-l.
Subsequent nucleophilic substitution of 4-fluorobenzaldehyde with alcohols 6e-l yielded
7e-l. Aldol reaction of aldehydes 7e-l and 2,4-thiazolidinedione gave 3e-l. Reduction of
3e-l with CoCl2·DMG (dimethylglyoxime) and NaBH424 gave 4e-h.
Scheme 1. Reagents and conditions: a) ArCl, 120 ºC, 6-99%, b) 4-fluorobenzaldehyde,
KOt-Bu, DMF, r.t., 9-78%, c) 2,4-thiazolidinedione, pyrrolidine, MeOH, 45 ºC, 38-74%,
d) CoCl2·DMG, NaBH4, MeOH, 35 ºC, 20-33%.
Cell-based transrepressional activity was evaluated with an NF-κB reporter system in
HEK293 (human embryonic kidney) cells transiently expressing human PPARγ under
stimulation with TNFα.25 The positive controls pioglitazone (1) and rosiglitazone (2)
showed transrepressional activity with IC50 values of 22 µM (Table 1).
PPARγ-transactivating agonistic activity was measured with the Gal4N-human PPARγ

LBD reporter system in HEK293 cells, as previously reported.26 Percent efficacy is
estimated as the maximal stimulatory response in relation to the maximal activity of 1.
The positive controls 1 and 2 showed transactivating agonistic activity with EC50 values
of 1.1 µM (Table 1). In both these assays, cells were assayed for luciferase activities and
β-galactosidase activities after 7 h (for transrepression) and 24 h (for transactivation)
treatment with test compounds. None of our compounds reduced β-galactosidase activity
under these conditions, suggesting that none of them is toxic at the concentrations tested.
Table 1. SARs of rosiglitazone
Compound
R
Ar
Bonda
Transactivating agonistic activity
Transrepression
% efficacyb
EC50 (µM)
IC50 (µM)
-
-
-
1.1
1.1
100
94
22
22
1
2
Me
s
Et
Et
s
d
s
0.54
0.36
110
38
21
15
20
4e
3e
4f
Me
0.057
120

Me
Et
d
s
0.31
0.052
0.82
40
130
44
13
25
17
20
3f
4g
3g
4h
Et
d
s
i-Pr
0.53
92
a s, single bond; d, double bond. b Emax values are given relative to positive control 1.
As regards transactivating agonistic activity, we found that benzothiazolyl analogs 4f and
4g showed 19- and 10-fold increased EC50 values compared with pyridyl analogs 2 and 4e,
respectively, whereas maximum efficacy was almost the same, in accordance with the
previous report.13 Next, we examined the effect of the alkyl group on transactivation.
Ethyl analogs 4e and 4g showed more potent transactivating agonistic activity (EC50)
than the methyl analogs 2 and 4f, and i-Pr analog 4h. On the other hand, transrepressional
activity was roughly equal among these analogs 2 and 4e-g (IC50 in the range of 20-25
µM). These results indicate that substituents on the amino group influence transactivating
activity more sensitively than transrepressional activity.
Alkenic analogs 3e-g showed slightly increased transrepressional activity compared with
the saturated analogs 4e-g, and 3f showed an IC50 value of 13 µM. On the other hand,
EC50 values of 3e-g were decreased compared with the saturated analogs 4e-g. In addition,

3e-g exhibited weak efficacy of around 40% compared to the full agonist pioglitazone.
These results indicate that introduction of the double bond into the 5-position of
thiazolidindione increases the selectivity for transrepression over transactivation.
These results prompted us to synthesize alkenic analogs bearing other heterocycles.
Thiazol-2-yl 3i, benzoimidazol-2-yl 3j, and 2-naphthyl 3k analogs showed decreased
EC50 values and efficacy compared with 3f (Table 2). But these analogs 3i-k also showed
diminished transrepressional activity (IC50 in the range of 38-42 µM). On the other hand,
a smaller alkyl substituent (3f) tends to show slightly increased transrepressional activity
with lower efficacy compared to 3g (Table 1). Therefore, we synthesized non-substituted
analog 3l in an attempt to improve the selectivity. In the range of 0.03-30 µM, compound
3l exhibited a more potent EC50 value (0.053 µM) but lower efficacy of 11% than methyl
analog 3f. On the other hand, the transrepressional activity of 3l (14 µM) was comparable
to that of 3f. It has been proposed that the molecular mechanisms of PPARγ partial
agonism are stabilization of helix 3 and β sheet in the PPARγ LBD, based on X-ray
crystallographic studies, docking studies, and amide H/D exchange kinetics (Fig. 5).27,28
The molecular mechanism of PPARγ partial agonism of 3l is unclear, but the SAR
described here indicates that the substituent on the amino group and the location of the
aromatic heterocycle near helix 3 and the β sheet might be important for partial agonism.

Table 2. SARs of benzylidene-thiazolidine-2 4-diones
Compound
R
Ar
Transactivating agonistic activity
Transrepression
IC50 (µM)
13
% efficacya
EC50 (µM)
Me
Me
Me
Me
H
0.31
0.70
3.4
40
31
12
8.7
11
3f
3i
42
38
40
14
3j
3k
3l
1.6
0.053
a Emax values are given relative to the positive control 1.

Fig. 5. Location of helix 3 (magenta) and the β sheet (magenta) proposed to be important
for partial agonism. PPARγ: green, rosiglitazone: white stick (PDB ID: 1FM6).
TNFα transmits inflammatory signaling intracellularly through the TNFα receptor in the
plasma membrane, and this induces degradation of the transrepressional complex on the
promoter region of proinflammatory genes (e.g., IL-6 and iNOS), thereby activating gene
transcription. So, inhibitory activity toward TNFα-induced NF-κB inhibition is also
exhibited by TNF receptor antagonists and TNFα-signaling inhibitors. Therefore, we
examined whether the NF-κB luciferase inhibitory activity of our compounds was
PPARγ-dependent. As shown in Figure 6, inhibition of TNFα-induced NF-κB luciferase
by 3l (IC50: 29 µM) was partially abolished in the absence of transfection of PPARγ. This
partial abolition of NF-κB inhibitory activity would be due to endogenous expression of
PPARγ in HEK293 cell lines. This result indicates that PPARγ contributes to the
inhibition of NF-κB luciferase by 3l at least in part, although we cannot exclude the

possibility that our compounds also inhibit TNFα-induced NF-κB transcription through
other mechanisms.
Fig. 6. PPARγ dependency of NF-kB-inhibitory activity of compound 3l. Red: PPARγ
transfection (+); blue: PPARγ transfection (-).
To examine the separability of transactivation and transrepression, we plotted –Log (IC50)
against –Log (EC50) (Fig. 7). Neither –Log (EC50) nor Emax was correlated to –Log
(IC50), (R2 = 0.2 and 0.07), despite the similarity of chemical structures. This result
indicates that strong transactivation is not necessary for strong transrepression. Thus, it
can be anticipated that transrepression and transactivation should be separable by
chemical modification of PPARγ ligands.

Fig. 7. Correlation between transactivation and transrepression among risoglitazone
analogs. (a) –Log (IC50) and –Log (EC50). (b) –Log (IC50) and Emax.
In summary, we investigated SARs of the PPARγ agonist rosiglitazone, focusing on
transrepressional activity. Among the synthesized compounds, 3l showed
PPARγ transrepressional activity with low transactivating efficacy (11%). Therefore, this
compound might be a promising candidate for exhibiting anti-inflammatory properties
with reduced side effects. Biological evaluations of 3l and further chemical modifications
to improve selectivity are in progress.
ACKNOWLEDGMENT
The work described in this paper was partially supported by Grants-in Aid for Scientific
Research (B) No. 25293027 and 26293025 from The Ministry of Education, Culture,
Sports, Science and Technology, Japan.
References and notes

1.
2.
3.
4.
5.
Grey, A. Skeletal consequences of thiazolidinedione therapy. Osteoporos. Int.
2008, 19, 129–137.
Rubenstrunk, A.; Hanf, R.; Hum, D. W.; Fruchart, J.-C.; Staels, B. Biochim.
Biophys. Acta 2007, 1771, 1065–81.
Guan, Y.; Hao, C.; Cha, D. R.; Rao, R.; Lu, W.; Kohan, D. E.; Magnuson, M. A.;
Redha, R.; Zhang, Y.; Breyer, M. D. Nat. Med. 2005, 11, 861–866.
Garcia-vallve, S.; Guasch, L.; Tomas-herna, S.; Maria, J.; Ollendor, V.; Pujadas,
G.; Mulero, M. J. Med. Chem. 2015, 58, 5381–5394.
Ohashi, M.; Oyama, T.; Putranto, E. W.; Waku, T.; Nobusada, H.; Kataoka, K.;
Matsuno, K.; Yashiro, M.; Morikawa, K.; Huh, N. H.; Miyachi, H. Bioorganic
Med. Chem. 2013, 21, 2319–2332.
6.
7.
Rinne, S. T.; Feemster, L. C.; Collins, B. F.; Au, D. H.; Perkins, M.; Bryson, C. L.;
O’Riordan, T. G.; Liu, C.-F. Allergy, Asthma Clin. Immunol. 2014, 10, 34.
Ormseth, M. J.; Oeser, A. M.; Cunningham, A.; Bian, A.; Shintani, A.; Solus, J.;
Tanner, S.; Stein, C. M. Arthritis Res. Ther. 2013, 15, R110.
Ricote, M.; Glass, C. K. Biochim. Biophys. Acta 2007, 1771, 926–935.
Jiang, C.; Ting, A. T.; Seed, B. Nature 1998, 391, 82–86.
8.
9.
10. Ricote, M.; Li, A. C.; Willson, T. M.; Kelly, C. J.; Glass, C. K. Nature 1998, 391,
79–82.
11. Pascual, G.; Fong, A. L.; Ogawa, S.; Gamliel, A.; Li, A. C.; Perissi, V.; Rose, D.
W.; Willson, T. M.; Rosenfeld, M. G.; Glass, C. K. Nature 2005, 437, 759–63.
12. Barros, C. D.; Amato, A. A.; Oliveira, T. B. de; Iannini, K. B. R.; Silva, A. L. da;

Silva, T. G. da; Leite, E. S.; Hernandes, M. Z.; Lima, M. do C. A. de; Galdino, S.
L.; Neves, F. de A. R.; Pitta, I. da R. Bioorganic Med. Chem. 2010, 18, 3805–3811.
13. Jeon, R.; Kim, Y.; Cheon, Y.; Ryu, J. Arch. Pharm. Res. 2006, 29, 394–399.
14. Ghisletti, S.; Huang, W.; Ogawa, S.; Pascual, G.; Lin, M.-E.; Willson, T. M.;
Rosenfeld, M. G.; Glass, C. K. Mol. Cell 2007, 25, 57–70.
15. Glass, C. K.; Saijo, K. Nat. Rev. Immunol. 2010, 10, 365–376.
16. Sundahl, N.; Bridelance, J.; Libert, C.; De Bosscher, K.; Beck, I. M. Pharmacol.
Ther. 2015, 152, 28–41.
17. Carson, M. W.; Luz, J. G.; Suen, C.; Montrose, C.; Zink, R.; Ruan, X.; Cheng, C.;
Cole, H.; Adrian, M. D.; Kohlman, D. T.; Mabry, T.; Snyder, N.; Condon, B.;
Maletic, M.; Clawson, D.; Pustilnik, A.; Coghlan, M. J. J. Med. Chem. 2014, 57,
849–860.
18. Xiao, H. Y.; Wu, D. R.; Malley, M. F.; Gougoutas, J. Z.; Habte, S. F.; Cunningham,
M. D.; Somerville, J. E.; Dodd, J. H.; Barrish, J. C.; Nadler, S. G.; Dhar, T. G. M. J.
Med. Chem. 2010, 53, 1270–1280.
19. Aoyama, A.; Endo-Umeda, K.; Kishida, K.; Ohgane, K.; Noguchi-Yachide, T.;
Aoyama, H.; Ishikawa, M.; Miyachi, H.; Makishima, M.; Hashimoto, Y. J. Med.
Chem. 2012, 55, 7360–7377.
20. Nomura, S.; Endo-Umeda, K.; Aoyama, A.; Makishima, M.; Hashimoto, Y.;
Ishikawa, M. ACS Med. Chem. Lett. 2015, 6, 902–907.
21. Cantello, B. C. C.; Cawthorne, M. A.; Haigh, D.; Hindley, R. M.; Smith, S. A.;
Thurlby, P. L. Bioorganic Med. Chem. Lett. 1994, 4, 1181–1184.

22. Cantello, B. C. C.; Cawthorne, M. A.; Cottam, G. P.; Duff, P. T.; Haigh, D.;
Hindley, R. M.; Lister, C. A.; Smith, S. A.; Thurlby, P. L. J. Med. Chem. 1994, 37,
3977–3985.
23. Liu, K.; Black, R. M.; Acton, J. J.; Mosley, R.; Debenham, S.; Abola, R.; Yang,
M.; Tschirret-Guth, R.; Colwell, L.; Liu, C.; Wu, M.; Wang, C. F.; MacNaul, K.
L.; McCann, M. E.; Moller, D. E.; Berger, J. P.; Meinke, P. T.; Jones, A. B.; Wood,
H. B. Bioorganic Med. Chem. Lett. 2005, 15, 2437–2440.
24. Jamali, B.; Bjørnsdottir, I.; Nordfang, O.; Hansen, S. H. J. Pharm. Biomed. Anal.
2008, 46, 82–87.
25. HEK293 cells were plated at a density of 10% confluent/96 well plate 24 h prior to
transfection. Human PPARγ-full-length, pGL4.32[luc2P/NF-κB-RE/Hygro]
(Promega) and β-galactosidase expression plasmids were co-transfected by the
calcium phosphate coprecipitation method. After 48 h, cells were treated with test
compounds or DMSO. After 4 h, cells were stimulated with TNF-α
(PEPROTECH). After 3 h, cells were assayed for luciferase activity in a
luminometer. The luciferase activity of each sample was normalized by the level of
β-galactosidase activity. Each transfection was carried out in triplicate.
26. Kasuga, J.; Ishikawa, M.; Yonehara, M.; Makishima, M.; Hashimoto, Y.; Miyachi,
H. Bioorg. Med. Chem. 2010, 18, 7164–7173.
27. Montanari, R.; Saccoccia, F.; Scotti, E.; Crestani, M.; Godio, C.; Gilardi, F.;
Loiodice, F.; Fracchiolla, G.; Laghezza, A.; Tortorella, P.; Lavecchia, A.;
Novellino, E.; Mazza, F.; Aschi, M.; Pochetti, G. J. Med. Chem. 2008, 51,

7768–7776.
28. Bruning, J. B.; Chalmers, M. J.; Prasad, S.; Busby, S. A.; Kamenecka, T. M.; He,
Y.; Nettles, K. W.; Griffin, P. R. Structure 2007, 15, 1258–1271.

GA