Kojyl cinnamate esters are peroxisome proliferator-activated receptor α/γ dual agonists

Article abstract of DOI:10.1016/j.bmc.2018.10.010

Adiponectin is an adipocytokine with insulin-sensitizing, anti-inflammatory, anti-atherosclerotic, and anti-aging properties. Compounds with the ability to promote adiponectin secretion are of interest for the development of anti-aging drugs to improve skin-aging phenotypes. In the phenotypic assay to measure adiponectin secretion during adipogenesis in human adipose tissue-derived mesenchymal stem cells (hAT-MSCs), kojyl cinnamate ester derivatives increased adiponectin secretion. A target identification study showed that the kojyl cinnamate ester derivatives competitively bound to peroxisome proliferator-activated receptor α/γ (PPARα/γ). The upregulation of adiponectin production induced by kojyl cinnamate ester derivatives was significantly correlated with PPARα and PPARγ binding activities. Kojyl cinnamate ester derivatives significantly increased the transcription of genes encoding cholesterol and fatty acid synthesizing enzymes in hAT-MSCs. Notably, the kojyl cinnamate esters upregulated the gene transcription of lipid metabolic enzymes in human epidermal keratinocytes, which are important in the integrity of skin permeability barrier. In addition, the kojyl cinnamate esters that function as PPARα/γ dual modulators inhibited ultraviolet B irradiation-induced inflammation in human epidermal keratinocytes. Therefore, kojyl cinnamate ester derivatives are a novel class of PPARα/γ dual agonists with the potential to improve skin-aging phenotypes.

Full text of DOI:10.1016/j.bmc.2018.10.010

Accepted Manuscript
Kojyl cinnamate esters are peroxisome proliferator-activated receptor α/γ dual
agonists
Sae On Kim, Yujia Han, Sungjin Ahn, Seungchan An, Jeayoung C. Shin,
Hyunjung Choi, Hyoung-June Kim, Nok Hyun Park, Yong-Jin Kim, Sun Hee
Jin, Ho Sik Rho, Minsoo Noh
PII:
S0968-0896(18)31446-9
https://doi.org/10.1016/j.bmc.2018.10.010
BMC 14569
DOI:
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
12 August 2018
13 October 2018
16 October 2018
Please cite this article as: On Kim, S., Han, Y., Ahn, S., An, S., Shin, J.C., Choi, H., Kim, H-J., Hyun Park, N., Kim,
Y-J., Hee Jin, S., Sik Rho, H., Noh, M., Kojyl cinnamate esters are peroxisome proliferator-activated receptor α/γ
dual agonists, Bioorganic & Medicinal Chemistry (2018), doi: https://doi.org/10.1016/j.bmc.2018.10.010
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.

Kojyl cinnamate esters are peroxisome proliferator-activated receptor α/γ dual agonists
Sae On Kima, #, Yujia Hana, #, Sungjin Ahna, Seungchan Ana, Jeayoung C. Shina, Hyunjung Choib,
Hyoung-June Kimb, Nok Hyun Parkb, Yong-Jin Kimb, Sun Hee Jina, Ho Sik Rhoc,*, Minsoo Noha,*
aCollege of Pharmacy and Natural Products Research Institute, Seoul National University, 1 Gwanak-ro,
Gwanak-gu, Seoul 08826, Republic of Korea; bAmorePacific Corporation R&D Center, Yongin,
Gyeounggi-do 17074, Republic of Korea; cDepartment of Chemical and Material Engineering, The
University of Suwon, Gyeounggi-do 18323, Republic of Korea
# These authors contributed equally to this work
*Corresponding Authors
Minsoo Noh, PhD
College of Pharmacy, Natural Products Research Institute, Seoul National University, 1 Gwanak-ro,
Gwanak-gu, Seoul 08826, Republic of Korea
Tel: 82-2-880-2481
E-mail: minsoonoh@snu.ac.kr
Ho Sik Rho, PhD
Department of Chemical and Material Engineering, The University of Suwon, Gyeounggi-do 18323,
Republic of Korea
Tel: 82-31-220-2360
E-mail: thiocarbon@naver.com
Abbreviations: ACACB, Acetyl-CoA carboxylase beta; ACAT, Acetyl-CoA acetyltransferase;
ADIPOQ, Adiponectin; CDK5, cyclin-dependent kinase 5; COX-2, cyclooxygenase-2; DMEM,
Dulbecco Modified Eagle’s Medium; EC50, effective concentration 50; ECM, extracellular
matrix; FABP, Fatty acid binding protein; FADS, Fatty acid desaturase; FASN, Fatty acid
synthase; FBS, fetal bovine serum; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GR,
glucocorticoid receptor; hAT-MSCs, mesenchymal stem cells; HMGCR, 3-hydroxy-3-
methylglutaryl-COA reductase; HMGCS, 3-hydroxy-3-methylglutaryl-CoA synthase; hMSCs,
human mesenchymal stem cells; IBMX, 3-isobutyl-1-methylxanthine; LBD, ligand-binding
domain; MMP, matrix metalloproteinase; NHEKs, normal human epidermal keratinocytes; PBS,
phosphate-buffered saline; PDB, protein data bank; PG, prostaglandin; PPAR, peroxisome
proliferator-activated receptor; PTGS, prostaglandin endoperoxide synthase; SCD, stearoyl-CoA
desaturase; Seletinoid G, 5-hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(benzo[d][1,3]dioxol-5-
yl)acrylate; TR-FRET, time-resolved fluorescence resonance energy transfer; UV, ultraviolet

Abstract
Adiponectin is an adipocytokine with insulin-sensitizing, anti-inflammatory, anti-atherosclerotic, and
anti-aging properties. Compounds with the ability to promote adiponectin secretion are of interest for the
development of anti-aging drugs to improve skin-aging phenotypes. In the phenotypic assay to measure
adiponectin secretion during adipogenesis in human adipose tissue-derived mesenchymal stem cells
(hAT-MSCs), kojyl cinnamate ester derivatives increased adiponectin secretion. A target identification
study showed that the kojyl cinnamate ester derivatives competitively bound to peroxisome proliferator-
activated receptor α/γ (PPARα/γ). The upregulation of adiponectin production induced by kojyl
cinnamate ester derivatives was significantly correlated with PPARα and PPARγ binding activities. Kojyl
cinnamate ester derivatives significantly increased the transcription of genes encoding cholesterol and
fatty acid synthesizing enzymes in hAT-MSCs. Notably, the kojyl cinnamate esters upregulated the gene
transcription of lipid metabolic enzymes in human epidermal keratinocytes, which are important in the
integrity of skin permeability barrier. In addition, the kojyl cinnamate esters that function as PPARα/γ
dual modulators inhibited ultraviolet B irradiation-induced inflammation in human epidermal
keratinocytes. Therefore, kojyl cinnamate ester derivatives are a novel class of PPARα/γ dual agonists
with the potential to improve skin-aging phenotypes.
Key words: Kojyl cinnamate ester derivatives; Human adipose tissue-derived mesenchymal stem cells;
Adiponectin; Peroxisome proliferator-activated receptor dual modulators
1. Introduction
Phenotypic drug screening is an alternative approach to molecular target-based methods for
identifying novel drug candidate molecules.1 Compared to the conventional target-based approach, cell-
or organism-based phenotypic assays have the advantage of being able to evaluate candidate molecules in
a more biologically relevant disease context. Despite this advantage, the phenotype-based approach
requires follow-up studies to elucidate the direct molecular targets responsible for the induction of the

observed responses.1 The differentiation of human mesenchymal stem cells (hMSCs) has been
investigated for the development of phenotype-based pharmacological assays for various
pathophysiological conditions associated with adipose tissue.2 For example, the adipogenesis of human
adipose tissue-derived mesenchymal stem cells (hAT-MSCs) can be used as a model system to screen for
active compounds that may prevent the gradual decrease of subcutaneous adipose tissue with age. The
age-dependent decline in subcutaneous fat volume and the subsequent regression of adipocyte lipid
metabolism contribute to the progressive thinning and wrinkling of the skin.3,4
Adiponectin, which is mainly expressed in adipose tissue, is an adipocytokine with insulin-
sensitizing, anti-inflammatory, anti-atherosclerotic, and anti-aging properties.5 In addition, adiponectin
can inhibit age-associated breakdown of extracellular matrix proteins, such as type 1 collagen and elastin,
in both chronologically aged and photo-aged skin.6,7 Thus, adiponectin production can be measured to
provide a quantitative indication for phenotypic assays during adipogenesis in hAT-MSCs. Adipogenesis
is regulated by an interacting network of transcription factors, including PPARγ and CCAAT/enhancer
binding proteins (C/EBPs), which collectively enhance the expression of adipocyte specific genes.3,4
Pharmacological agents that modulate this transcriptional network can affect adiponectin production.
PPARγ agonists, such as rosiglitazone and pioglitazone, stimulate adiponectin production in
adipogenesis-inducing hMSCs. Adiponectin expression can be also regulated by cyclin-dependent kinase
5 (CDK5), which phosphorylates the nuclear receptor PPARγ at serine 273.8 Therefore, adiponectin-
secretion promoting molecules identified in the adipogenesis phenotypic model of hMSCs should be
investigated to determine whether they interact with PPARs or CDK5.
Previously, we reported that a topical kojyl cinnamate ester derivative, 5-hydroxy-4-oxo-4H-
pyran-2-yl-methyl (E)-3-(benzo[d][1,3]dioxol-5-yl)acrylate (also called as seletinoid G, 1), improved the
mass balance of collagen in aged human skin.7 Compound 1 was designed for ameliorating skin aging
phenotypes by achieving the prevention of gradual subcutaneous fat loss and the inhibition of the reactive

oxygen species.9 The cinnamate moiety was identified to induce adipocyte differentiation in a natural
product library screnning.9 The γ-pyramone ring of kojic acid has a free radical scavenging activity and is
also associated with the attenuation of photodamage.7,9 The introduction of a cinnamate moiety to kojic
acid was expected to have an improved anti-skin aging activity. Compound 1 was the most potent kojyl
cinnamate ester derivative to increase collagen synthesis in human dermis.7 In addition, compound 1
significantly promoted adiponectin production during adipogenesis in hAT-MSCs .9 Because adiponectin
promotes ECM synthesis and prevents age-related decrease of subcutaneous fat volume, the adiponectin-
secretion promoting activity of compound 1 was suggested as an indirect molecular mechanism
associated with the compound 1-mediated clinical improvement of the dermal mass balance in aged
human skin.7 However, the direct molecular target responsible for the compound 1-induced upregulation
of adiponectin was not clearly determined because of the relatively weak potency of its effect on PPARs
or CDKs.9
In this study, we aimed to synthesize more potent adiponectin-secreting kojyl cinnamate ester
derivatives for use in a target identification studies investigating their molecular targets during
adipogenesis in hAT-MSCs. Our results demonstrated that the adiponectin secreting activity of the kojyl
cinnamate esters are associated with dual agonism of PPARγ and PPARα.
2. Materials and methods
2.1. Chemical synthesis.
All chemicals were purchased from Sigma-Aldrich (St Louis, MO, USA) and used without further
purification. Proton (1H) and carbon (13C) NMR spectra were obtained on a Verian FT-NMR (500/125
MHz) spectrometer. The chemical shifts are reported in  units (ppm) relative to tetramethylsilane (TMS)
and the coupling constants (J) quoted in Hz. HRMS spectrometry was performed on a negative method to
obtain high-resolution mass spectra.
2.1.1. General procedure

To a stirred solution of kojyl chloride 9 (4.80 g, 30.0 mmol) in 100 mL of dimethylformamide (DMF)
under N2 was added potassium salt of 3,4-methylenedioxycinnamic acid 10 (6.88 g, 30.0 mmol) with 3,4-
methylenedioxycinnamic acid 2 (4.32 g, 22.5 mmol). The reaction mixture was stirred for 1 h at 110°C -
120°C, after which DMF was evaporated in vacuo. The residue was extracted with ethyl acetate (500 mL)
and washed with water. The organic layer was dried with anhydrous MgSO4 and concentrated to give a
crude product. The resultant was purified by crystallization from ethyl acetate-hexane to give the desired
compound 1 (7.40 g, 78%).
2.1.2. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(benzo[d][1,3]dioxol-5-yl)acrylate (1)
1H NMR (500 MHz, DMSO-d6) : δ 9.29 (s, 1H), 8.12 (s, 1H), 7.68 (d, 1H, J = 16.2 Hz), 7.46 (s, 1H), 7.25
(d, 1H, J = 8.1 Hz), 6.98 (d, 1H, J = 8.1 Hz), 6.64 (d, 1H, J = 16.2 Hz), 6.52 (s, 1H), 6.08 (s, 2H), 5.07 (s,
13
2H). C NMR (125 MHz, DMSO-d6) : δ 173.5, 165.6, 161.6, 149.5, 148.0, 146.0, 145.7, 139.8, 128.2,
125.3, 114.6, 112.5, 108.4, 106.7, 101.6, 61.2.
2.1.3. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(3-methoxyphenyl)acrylate (2)
1H NMR (500 MHz, DMSO-d6) : δ 9.27 (s, 1H), 8.10 (s, 1H), 7.70 (d, 1H, J = 16.0 Hz), 7.22 – 7.33 (m,
3H), 6.98 (d, 1H, J = 7.0 Hz), 6.74 (d, 1H, J = 8.0 Hz), 6.52 (s, 1H), 5.07 (s, 2H), 3.77 (s, 3H). 13C NMR
(125 MHz, DMSO-d6) : δ 173.6, 165.5, 161.5, 159.6, 146.09, 145.82, 139.9, 135.2, 129.9, 121.2, 117.31,
116.96, 113.05, 112.73, 61.4, 55.2. HRMS (ESI) m/z (M-H)+ calcd for C16H14O6 = 301.0790, found
301.0718.
2.1.4. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(3,4-dimethoxyphenyl)acrylate (3)
1H NMR (500 MHz, DMSO-d6) : δ 9.24 (s, 1H), 8.09 (s, 1H), 7.64 (d, 1H, J = 16.0 Hz), 7.37 (s, 1H), 7.25
(d, 1H, J = 7.0 Hz), 6.96 (d, 1H, J = 7.0 Hz), 6.62 (d, 1H, J = 16.0 Hz), 6.51 (s, 1H), 5.05 (s, 2H), 3.77 (s,
3H). 13C NMR (125 MHz, DMSO-d6) : δ 173.6, 165.8, 161.7, 151.2, 148.9, 146.11, 146.08, 139.9, 126.6,

123.4, 114.3, 112.6, 111.4, 110.4, 61.2, 55.62, 55.57. HRMS (ESI) m/z (M-H)+ calcd for C17H16O7 =
331.0896, found 331.0824.
2.1.5. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(3,4,5-trimethoxyphenyl)acrylate (4)
1H NMR (500 MHz, DMSO-d6) : δ 9.24 (s, 1H), 8.10 (s, 1H), 7.64 (d, 1H, J = 16.0 Hz), 7.10 (s, 2H), 6.73
(d, 1H, J = 16.0 Hz), 6.52 (s, 1H), 5.04 (s, 2H), 3.79 (s, 6H), 3.67 (s, 3H). 13C NMR (125 MHz, DMSO-d6)
: δ 173.6, 165.6, 161.6, 153.0, 146.1, 139.96, 139.67, 129.4, 116.2, 112.7, 106.1, 61.4, 60.1, 56.0. HRMS
(ESI) m/z (M-H)+ calcd for C18H19O8 = 361.1002, found 361.0930.
2.1.6. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(3,5-dimethoxyphenyl)acrylate (5)
1H NMR (500 MHz, DMSO-d6) : δ 9.26 (s, 1H), 8.09 (s, 1H), 7.63 (d, 1H, J = 16.0 Hz), 6.92 (s, 2H), 6.76
(d, 1H, J = 16.0 Hz), 6.94 (s, 1H), 6.52 (s, 1H), 5.07 (s, 2H), 3.75 (s, 6H). 13C NMR (125 MHz, DMSO-d6)
: δ 173.6, 165.5, 161.5, 160.6, 146.10, 145.89, 139.9, 135.7, 117.6, 112.7, 106.3, 103.0, 61.4, 55.4.
HRMS (ESI) m/z (M-H)+ calcd for C17H16O7 = 331.0896, found 331.0824.
2.1.7. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(4-methoxyphenyl)acrylate (6)
1H NMR (500 MHz, DMSO-d6) : δ 9.25 (s, 1H), 8.09 (s, 1H), 7.66 – 7.71 (m, 3H), 6.95 (d, 2H, J = 8.0
Hz), 6.55 (d, 1H, J = 16.0 Hz), 6.50 (s, 1H), 5.05 (s, 2H), 3.76 (s, 3H). 13C NMR (125 MHz, DMSO-d6) :
δ 173.6, 165.7, 161.72, 161.39, 146.06, 145.69, 139.9, 130.4, 126.4, 114.40, 114.11, 112.6, 61.2, 55.3.
HRMS (ESI) m/z (M-H)+ calcd for C16H14O6 = 301.0790, found 301.0717.
2.1.8. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2-methoxyphenyl)acrylate (7)
1H NMR (500 MHz, DMSO-d6) : δ 9.26 (s, 1H), 8.09 (s, 1H), 7.91 (d, 1H, J = 16.0 Hz), 7.73 (d, 1H, J =
7.0 Hz), 7.42 (m, 1H), 7.08 (d, 1H, J = 8.0 Hz), 6.98 (m, 1H), 6.68 (d, 1H, J = 16.0 Hz), 6.49 (s, 1H), 5.06
(s, 2H), 3.85 (s, 3H). 13C NMR (125 MHz, DMSO-d6) : δ 173.7, 165.9, 161.7, 158.0, 146.1, 140.67,

140.02, 132.5, 129.0, 122.0, 120.8, 116.9, 112.7, 111.8, 61.4, 55.7. HRMS (ESI) m/z (M-H)+ calcd for
C16H14O6 = 301.0790, found 301.0717.
2.1.9. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2,3-dimethoxyphenyl)acrylate (8)
1H NMR (500 MHz, DMSO-d6) : δ 9.27 (s, 1H), 8.10 (s, 1H), 7.91 (d, 1H, J = 16.0 Hz), 7.36 (d, 1H, J =
7.0 Hz), 7.10 (m, 2H), 6.69 (d, 1H, J = 16.0 Hz), 6.50 (s, 1H), 5.07 (s, 2H), 3.81 (s, 3H), 3.76 (s, 3H). 13C
NMR (125 MHz, DMSO-d6) : δ 173.6, 165.6, 161.5, 152.7, 147.8, 146.0, 139.98, 139.95, 127.2, 124.4,
119.1, 117.9, 115.1, 112.7, 61.5, 60.8, 55.7. HRMS (ESI) m/z (M-H)+ calcd for C17H16O7 = 331.0896,
found 331.0823.
2.1.10. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2,3,4-trimethoxyphenyl)acrylate (9)
1H NMR (500 MHz, DMSO-d6) : δ 9.25 (s, 1H), 8.09 (s, 1H), 7.78 (d, 1H, J = 16.0 Hz), 7.54 (d, 1H, J =
7.0 Hz), 6.85 (d, 1H, J = 7.0 Hz), 6.59 (d, 1H, J = 16.0 Hz), 6.48 (s, 1H), 5.05 (s, 2H), 3.829 (s, 3H),
3.820 (s, 3H), 3.74 (s, 3H). 13C NMR (125 MHz, DMSO-d6) : δ 173.6, 165.9, 161.7, 155.8, 152.7, 146.0,
141.6, 140.34, 140.01, 123.5, 120.0, 115.2, 112.6, 108.5, 61.4, 61.3, 60.4, 56.1. HRMS (ESI) m/z (M-H)+
calcd for C18H19O8 = 361.1002, found 361.0929.
2.1.11. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2,5-dimethoxyphenyl)acrylate (10)
1H NMR (500 MHz, DMSO-d6) : δ 9.24 (s, 1H), 8.10 (s, 1H), 7.85 (d, 1H, J = 16.0 Hz), 7.68 (d, 1H, J =
7.0 Hz), 6.55 – 6.63 (m, 3H), 6.43 (s, 1H), 5.05 (s, 2H), 3.87 (s, 3H), 3.82 (s, 3H). 13C NMR (125 MHz,
DMSO-d6) : δ 173.5, 165.8, 161.6, 153.2, 152.3, 146.0, 140.2, 139.9, 122.4, 118.4, 117.2, 113.0, 112.73,
112.68, 61.3, 56.1, 55.5. HRMS (ESI) m/z (M-H)+ calcd for C17H16O7 = 331.0896, found 331.0822.
2.1.12. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2,4,5-trimethoxyphenyl)acrylate (11)
1H NMR (500 MHz, DMSO-d6) : δ 9.24 (s, 1H), 8.09 (s, 1H), 7.88 (d, 1H, J = 16.0 Hz), 7.29 (s, 1H), 6.70
13
(s, 1H), 6.59 (d, 1H, J = 7.0 Hz), 6.45 (s, 1H), 5.04 (s, 2H), 3.85 (s, 3H), 3.82 (s, 3H), 3.78 (s, 3H). C

NMR (125 MHz, DMSO-d6) : δ 173.6, 166.2, 161.8, 153.7, 152.7, 146.0, 143.0, 140.2, 139.9, 113.5,
113.2, 112.6, 111.06, 110.97, 61.2, 56.4, 56.0, 55.7. HRMS (ESI) m/z (M-H)+ calcd for C18H19O8 =
361.1002, found 361.0932.
2.1.13. 5-Hydroxy-4-oxo-4H-pyran-2-yl-methyl (E)-3-(2,4,6-trimethoxyphenyl)acrylate (12)
1H NMR (500 MHz, DMSO-d6) : δ 9.23 (s, 1H), 8.09 (s, 1H), 7.96 (d, 1H, J = 16.0 Hz), 6.63 (d, 1H, J =
7.0 Hz), 6.44 (s, 1H), 6.28 (s, 2H), 5.02 (s, 2H), 3.85 (s, 9H). 13C NMR (125 MHz, DMSO-d6) : δ 173.6,
167.1, 163.4, 161.9, 161.1, 146.0, 139.9, 136.5, 114.6, 112.4, 104.1, 90.9, 61.0, 56.0, 55.5. HRMS (ESI)
m/z (M-H)+ calcd for C18H19O8 = 361.1002, found 361.0930.
2.2. Cell culture and differentiation.
hAT-MSCs and their culture media were purchased (Lonza, Inc., Walkersville, MD). hAT-MSCs were
cultured in Dulbecco Modified Eagle’s Medium (DMEM) containing low glucose (1 g/L) and
supplemented with 10% fetal bovine serum (FBS), antibiotics, and GlutamaxTM (Invitrogen, Carlsbad,
CA). To induce adipocyte differentiation, the growth medium was replaced with DMEM containing high
glucose (4.5 g/L) and supplemented with 10% FBS, 10 g/mL insulin, 0.5 M dexamethasone, and 0.5 mM
3-isobutyl-1-methylxanthine (IBMX) (IDX condition). Dexamethasone, insulin, IBMX, glibenclamide,
troglitazone, Wy-14643, and pioglitazone were purchased (Sigma–Aldrich, St. Louis, MO, USA). hBM-
MSCs were grown in the IDX condition and treated twice with either glibenclamide, troglitazone, Wy-
14643, or compounds 1 through 12 for a total of 6 days. Primary human keratinocytes from neonatal
foreskin (Lonza) were cultured in KBM medium with KGM2 growth supplements consisting of insulin,
hydrocortisone, epinephrine, transferrin, human epidermal growth factor, bovine pituitary extract and
gentamicin/amphotericin B (Sigma-Aldrich).
2.3. Oil Red O and hematoxylin staining.
The level of adipocyte differentiation was determined using Oil Red O stain (Sigma Aldrich) as an
indicator of intracellular lipid accumulation. Differentiated adipocytes were rinsed twice with phosphate-

buffered saline (PBS), fixed with 10% formalin in PBS (pH 7.4) for 1 h, and then washed once with 60%
isopropanol. Cultures were stained with 0.2% Oil Red O reagent for 10 min at room temperature and
washed four times with H2O. To visualize the nucleus, the differentiated hBM-MSCs were counterstained
with hematoxylin reagent from Sigma–Aldrich (St. Louis, MO, USA) for 1 min and then washed four
times with H2O. The level of adipocyte differentiation was photographed using an inverted phase
microscope.
2.4. Enzyme-linked immunosorbent assay (ELISA).
The R&D Systems QuantikineTM immunoassay kit (Minneapolis, MN, USA) was used for quantitative
measurement of adiponectin in cell culture supernatants. Cytokine concentrations were determined
according to the manufacturer’s instructions.
2.5. Nuclear receptor binding (NR) assays.
LanthascreenTM’s TR-FRET-based competitive binding assay kits (Invitrogen) were used to evaluate the
PPARγ and PPARα binding activities of ligands as previously described.10 All assay measurements were
performed using CLARIOstar (BMG LABTECH, Ortenberg, Germany).
2.6. Computational docking analysis
Computational docking analysis was performed using both AutoDock Vina 1.1.2 software (The Scripps
Research Institute, La Jolla, CA, USA) and Accelrys Discovery Studio (Dassault Systems, BIODIVA
Corporation, San Diego, CA, USA) with crystal structures of PPARα (PDB code: 4BCR) and PPARγ
(PDB code: 2PRG) that were downloaded from the Protein Data Bank. The protein crystal structure was
prepared for docking by removing native ligand from the ligand-binding domain. Polar hydrogens were
then added using MGLTools 1.5.6. All ligands were prepared by ChemBioDraw Ultra 14.0 and
MGLTools 1.5.6. We focused on key amino acid residues in the hydrophilic region of the ligand-binding
domain for successful docking. Docking success was evaluated by the lowest affinity value.
2.7. RNA isolation and quantitative real-time PCR (Q-RT-PCR).

Total RNA samples were isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, United States),
followed by a purification step using the Qiagen RNeasy kit (Qiagen, Valencia, CA, United States). The
concentration of RNA in each sample was determined spectrophotometrically at 260 nm. The integrity of
each RNA sample was evaluated using Agilent 2100 BioAnalyzer (Agilent Technologies, Santa Clara,
CA, USA). Total RNA from each sample was reversed transcribed to cDNA using the Superscript
Reverse Transcriptase (RT) II Kit (Invitrogen). TaqMan Universal Master Mix II and Q-RT-PCR primer
sets (Applied Biosystems, Foster City, CA, USA) were used to analyze cDNA samples for 3-hydroxy-3-
methylglutaryl-CoA synthase 1 (HMGCS1, Hs00940429_m1), 3-hydroxy-3-methylglutaryl-CoA
synthase 2 (HMGCS2, Hs00985427_m1), 3-hydroxy-3-methylglutaryl-COA reductase (HMGCR,
Hs00168352_m1), acetyl-CoA acetyltransferase 2 (ACAT2, Hs00255067_m1), fatty acid desaturase 1
(FADS1, Hs01096546_m1), fatty acid desaturase 2 (FADS2, Hs00927433_m1), fatty acid synthase
(FASN, Hs01005622_m1), acetyl-CoA carboxylase beta (ACACB, Hs01565914_m1), fatty acid binding
protein 4 (FABP4, Hs01086177_m1), and adiponectin (ADIPOQ, Hs00605917_m1). Human
glyceraldehyde-3-phosphate dehydrogenase (GAPDH, 4333764F) was used to normalize sample
variations. Q-RT-PCR was performed with an Applied Biosystems 7500 Real-Time PCR System
(Applied Biosystems).
2.8. Statistical Analyses.
The experimental values were expressed as the mean ± standard deviation. Statistical analysis was
performed using one-way ANOVA followed by Bonferroni’s post-test. All statistical analyses were
conducted using RStudio® for Windows (RStudio Inc., Boston, MA, USA). Experiments were repeated
independently three times, unless otherwise stated.
3. Results and Discussion
3.1 Chemical synthesis
In order to find more potent kojyl cinnamate ester compounds, we synthesized kojyl cinnamate
ester derivatives and examined their effect on adiponectin production during adipogenesis in hAT-MSCs.

The synthetic pathways are shown in scheme 1. Cinnamate derivatives (1 - 12) were synthesized by the
condensation of kojyl chloride (14) with the potassium salt of cinnamic acids. The reaction of kojic acid
(13) with thionyl chloride produced kojyl chloride (14). Kojyl chloride (14) was reacted with potassium
salts of cinnamic acids and cinnamic acids in dimethylformamide at 110 - 120 °C to give the
corresponding cinnamate derivatives (1 - 12).
R1
O
O
O
R2
R3
O
O
a
b
HO
Cl
O
OH
OH
R5
OH
O
O
R4
13
14
1 : R2, R3 = -OCH2O-, R1, R4, R5 = H
2 : R2 = OMe, R1, R3, R4, R5 = H
3 : R2, R3 = OMe, R1, R4, R5 = H
4 : R2, R3, R4 = OMe, R1, R5 = H
5 : R2, R4 = OMe, R1, R3, R5 = H
6 : R3 = OMe, R1, R2, R4, R5 = H
7 : R1 = OMe, R2, R3, R4, R5 = H
8 : R1, R2 = OMe, R3, R4, R5 = H
9 : R1, R2, R3 = OMe, R4, R5 = H
10 : R1, R4 = OMe, R2, R3, R5 = H
11 : R1, R3, R4 = OMe, R2, R5 = H
12 : R1, R3, R5 = OMe, R2, R4 = H
Scheme 1. (a) SOCl2, DMF, rt ; (b) potassium salt of cinnamic acids, cinnamic acids, DMF, 110 - 120 °C.
3.2. Kojyl cinnamate esters promote adiponectin-secretion promoting activity in hAT-MSCs during
adipogenesis
Kojyl cinnamate ester derivatives were initially designed through the synthesis of kojic acid and
cinnamates in order to improve skin-aging phenotypes.9 In the previous study, the adiponectin-secretion
promoting activity of kojyl cinnamate ester compounds was reduced by the introduction of an altered
kojic acid moiety.9 In this study, we modified the phenyl acrylate moiety in kojyl cinnamate ester
compounds through the introduction of methoxy groups into a benzene ring to improve adiponectin-
secretion promoting activity. To evaluate the adiponectin secretion-promoting activity, we treated hAT-
MSCs with compounds 1-12 in the presence of the IDX adipogenesis-inducing medium (Table 1).
Compared with the IDX control condition, compounds 2, 7, 8, and 10 promoted adiponectin production
during adipogenesis in hAT-MSCs by 2.65, 2.21, 2.81, and 3.83-fold, respectively (Table 1). The

intracellular lipid accumulation and the number of lipid droplets were also upregulated in hAT-MSCs
treated with compounds 2, 7, 8, and 10 compared with the IDX control (Fig. 1A). When the adiponectin
secretion-promoting activities were compared, compounds 2, 7, 8, and 10 were more potent than
compound 1. The introduction of a 2-methoxy group into the benzene ring of the phenyl acrylate structure
improved an adiponectin promoting activity during adipogenesis in hAT-MSCs (Table 1).
Table 1. Adiponectin-secreting activity of kojyl cinnamate ester compounds
Adiponectin (pg/ml)
IDX + Cpd 30 µM
164 ± 12*
265 ± 10**
120 ± 41
Compound (Cpd)
R1
R2
R3
R4
R5
1
2
3
4
5
6
7
8
9
H
H
H
H
H
-OCH2O(1,3dioxolyl)-
H
H
H
H
H
H
H
H
H
H
H
H
OCH3
OCH3
OCH3
OCH3
H
H
OCH3
OCH3 OCH3
H
OCH3
H
H
OCH3
H
R1
137 ± 22
67 ± 34
138 ± 72
221 ± 30**
281 ± 14**
137 ± 22
383 ± 18**
103 ± 30
O
R2
O
O
OCH3
H
H
H
H
O
H
R3
R5
OH
OCH3
OCH3
OCH3
OCH3
OCH3
OCH3
H
R4
OCH3
OCH3
H
H
H
10
11
12
OCH3
H
H
OCH3
OCH3 OCH3
OCH3
H
87 ± 32
IDX control
100 ± 15
IDX + aspirin 300 µM
IDX + troglitazone 10 µM
197 ± 21*
722 ± 72**
Notably, compounds with 2,3- or 2,5-dimethoxy groups in the benzene ring of the phenyl acrylate
structure, such as 8 and 10, resulted in more potent in promoting adiponectin secretion than those with
trimethoxy-phenyl acrylate derivatives, such as 9, 11, and 12. (Fig. 1B). The adiponectin secretion-
promoting activity of these kojyl cinnamate ester compounds was stronger than that of aspirin, but weaker
than that of a PPARγ agonist troglitazone (Fig. 1B). In a concentration-response analysis, the adiponectin
secretion-promoting activity of troglitazone reached a maximum at concentrations above 3 µM. At this
maximal level, the effective concentration 50 (EC50) value of troglitazone in the adiponectin secretion
phenotype-based assay was 0.86 µM, and those of 2, 7, 8, and 10 were 36.6, 20.6, 14.8, and 7.6 µM,
respectively. The EC50 value of compound 1 was reported as 40.3 µM.9 Therefore, the efficacy of
compounds 8 and 10 in the upregulation of adiponectin production was significantly greater than that of

compound 1. We performed a target identification study with compounds 8 and 10 to elucidate the direct
molecular mechanism responsible for the adiponectin secretion-promoting activity of kojyl cinnamate
ester derivatives.
Figure 1. Effects of kojyl cinnamate ester compounds on adipogenesis in hAT-MSCs. (A) hAT-MSCs
were co-treated with kojyl cinnamate ester compounds when adipocyte differentiation was induced with
an IDX medium. On the 3rd day of the differentiation, adipogenic media containing the compounds were
exchanged. On the 5th day , Oil red O staining was performed to visualize lipid droplets formed during
adipogenesis in hAT-MSCs. (B) On the 5th day, cell culture supernatants were collected and adiponectin
levels were determined by ELISA. Values represent mean ± SD (n = 3). * p ≤ 0.05 and ** p ≤ 0.01.
3.3. Kojyl cinnamate ester compounds bind both PPARα and PPARγ
The most potent kojyl cinnamate ester, compound 10, was used to identify the molecular targets
responsible for the adiponectin secretion-promoting activity during adipogenesis in hAT-MSCs.
Adiponectin production and mammalian adipogenesis are primarily regulated by nuclear receptors, such
as PPARα, PPARβ, PPARγ, glucocorticoid receptor (GR), and LXRα.10 To elucidate the molecular
targets associated with the adiponectin secretion-promoting activity of compound 10 during adipogenesis
of hAT-MSCs, we investigated the effects on nuclear receptor binding or coactivation (Fig. 2).

Figure 2. Binding activity of kojyl cinnamate ester compounds on nuclear receptors. (A) TR-FRET based
nuclear receptor binding assays were performed with peroxisome proliferator-activated receptor (PPAR)
α, PPARγ, PPARδ, and glucocorticoid receptor (GR). The receptor co-activation analysis for liver X
receptor (LXR) α was determined. The TR-FRET-based competitive binding activities of kojyl cinnamate
ester compounds were determined for PPARα (B) and PPARγ (C). (D) A Pearson’s correlation
coefficient (r2) was calculated between adiponectin production inducing activities of kojyl cinnamate
ester compounds and PPARα. (E) The correlation coefficient was determined between adiponectin
secretion- promoting activity and PPARγ. Values represent mean ± SD (n = 3). * p ≤ 0.05 and ** p ≤ 0.01.
In a receptor binding assay based on time-resolved fluorescence resonance energy transfer (TR-FRET),
compound 10 (30 μM) competitively replaced the binding of the labeled PPARα and PPARγ ligands by
59% and 77%, respectively (Figure 2A). In contrast, compound 10 had no significant effects on PPARδ,

GR, or LXRα, compared with the appropriate positive control of GW501516, dexamethasone, and
T0901317, respectively (Figure 4A). To determine the pharmacological correlation between the
adiponectin secretion-promoting effects and the PPAR binding activities, we measured the nuclear
receptor binding of kojyl cinnamates to PPARα or PPARγ (Fig. 2B and 2C). The correlation coefficient
between the adiponectin secretion-promoting activity of kojyl cinnamates and PPARα was 0.39 (p = 0.03)
(Fig. 2D). More significant correlation was observed between the PPARγ binding at 30 μM and the
adiponectin secretion-promoting activities of kojyl cinnamates during adipogenesis in hAT-MSCs (r2 =
0.74, p = 0.004) (Fig. 2E). The significant correlation suggests that the activation of both PPARα and
PPARγ by kojyl cinnamate esters was associated with the adiponectin-promoting activity in hAT-MSCs.
Next, we determined the concentration-dependent receptor binding of the two most potent compounds, 8
and 10. Compounds 8 and 10 exhibited significant competitive binding activity to both PPARα and
PPARγ in a concentration-dependent manner (Fig. 3). The Ki values for PPARα binding by 8 and 10 were
8.0 and 9.5 μM, respectively, compared with that of the PPARα agonist Wy-14643, which was 4.5 μM
(Fig. 3A). The Ki values of PPARγ binding by 8 and 10 were 6.7 and 7.0 μM, respectively, which were
less potent as the binding of the PPARγ agonist troglitazone (Fig. 3B). Therefore, the adiponectin
secretion-promoting kojyl cinnamates 8 and 10 were dual modulators that could regulate the cellular
functions of PPARα and PPARγ.

Figure 3. Concentration-dependent binding of kojyl cinnamates 8 and 10 to PPARα and PPARγ. TR-
FRET competitive binding assays were performed with kojyl cinnamates, 8 and 10, with as a ligand of
PPARα (A) and PPARγ (B). Ki values were derived for compounds 8 and 10 based on the Cheng and
Prusoff equation. Values represent mean ± SD (n = 3). * p ≤ 0.05 and ** p ≤ 0.01.
3.4. Molecular docking analyses of compound 10 to PPARα and PPARγ
Computational docking analysis was performed with the PPARα ligand-binding model from the
protein data bank (PDB) entry 4BCR, a PPARα structure in complex with Wy-14643 (Fig. 4A and 4B).11
The optimal docking mode of compound 10 was determined from an interaction model with the lowest
free energy score, which was -8.5 kcal/mol in the AutoDock Vina software (Fig 4B). The CDOCKER
interaction energy levels for Wy-14643 and compound 10 to the PPARα ligand-binding domain (LBD)
were -38.0 kcal/mol and -41.9 kcal/mol, respectively. Structural analysis showed that the interaction
between compound 10 and the LBD of PPARα was similar to that of a PPARα agonist Wy14643. Similar
to Wy-14643, compound 10 was stabilized through a combination of hydrogen bonds and hydrophobic
interactions with amino acid residues in helices, H3, H5, H7, and H10/11 of the PPARα LBD. Hydrogen
bonds with Ser280 (H3), Tyr314 (H5), and His440 (H10/H11) existed in both Wy-14643 and compound
10 (Fig. 4A and 4B). The Cys276 residue on H3 formed a hydrogen bond with Wy-14643 in the PPARα

LBD interaction model, whereas compound 10 formed a hydrophobic interaction with Cys276. In contrast,
compound 10 formed a hydrogen bond with the Thr279 residue on H3, but Wy-14643 did not. Notably,
many amino acid residues in the hydrophobic pocket of the PPARα LBD including Phe318 (H5), Leu321
(H5), Val324 (H5/β-sheet S2), Val332 (S2), Met355 (H7), and Lys358 (H7) hydrophobically interacted
with Wy-14643. Among these amino acid residues that hydrophobically interacted with Wy-14643,
Leu321 formed a hydrogen bond with compound 10 in the PPARα LBD interaction model and the sulfur
atom in Met355 formed pi-sulfur bonding with compound 10. Therefore, the molecular docking
simulation of Wy-14643 and compound 10 to the PPARα LBD suggested that the relative stabilized free
energy score for Wy-14643 compared with that of compound 10 is mainly composed of the hydrophobic
interactions with residues Phe318, Val324, Val332, and Lys358. Despite minor differences, the
conformation of compound 10 in the PPARα LBD interaction model was similar to that of the PPARα
agonist Wy-14643. Notably, the 2-methoxy group in the phenyl acrylate moiety of compound 10
interacted with Cys276 and Met355 of the PPARα LBD, suggesting that this interaction may be
responsible for the improvement of its adiponectin-secreting activity.

Figure 4. The binding modes of compound 10 to PPARα and PPARγ. Molecular docking analysis of Wy-
14643 (A) and compound 10 (B) against the ligand binding domains (LBD) of PPARα. Interactions of
troglitazone (C) and compound 10 (D) with the LBD of PPARγ. Key amino acid residues involved in the
molecular interactions with PPARα or PPARγ have been labeled. Docking simulations of the selective
compounds were performed using BIOVIA Discovery Studio Visualizer. The CDOCKER energy was
calculated using the Discovery Studio. AutoDock Vina binding energy was derived using AutoDock Vina
version 1.1.2.

As the adiponectin secretion-promoting activity of compound 10 was not as potent as that of
troglitazone, the PPARγ LBD interaction models for both troglitazone and compound 10 were
constructed with the PDB entry 4PRG, the co-crystal structure between a partial agonist GW0072 and
PPARγ (Fig. 4C and 4D).12 The free energy scores of optimized docking models for troglitazone and
compound 10 against the PPARγ LDB interaction model in AutoDock Vina analysis were -9.8 kcal/mol
and -7.9 kcal/mol, respectively (Fig. 4C and 4D). In the CDOCKER docking analysis, free energy scores
of -59.6 kcal/mol and -43.5 kcal/mol were calculated for troglitazone and compound 10, respectively (Fig.
4). In the two different ligand docking modeling studies, the free energy levels were correlated to the
receptor binding affinity of troglitazone and compound 10. Troglitazone, a full PPARγ agonist, formed a
hydrogen bond network with Gln286 (H3), Ser289 (H3), His323 (H5), Tyr473 (H12), Ser342 (S3/S4),
and Glu343 (S3/S4) while also forming hydrophobic interactions with Arg288 (H3), Tyr327 (H5),
Leu330 (H5), Val339 (S3), and Ile341 (S3) in the PPARγ LBD interaction model. Hydrogen bonds with
Ser289, His323, and Tyr473 existed in the compound 10-optimized PPARγ LBD interaction model,
whereas compound 10 did not interact with Gln286, Ser342, and Gln343, which contributed to binding in
the troglitazone-optimized model. Compound 10 formed only a single hydrophobic interaction with
Leu330 in the hydrophobic pocket of the PPARγ LBD. Interestingly, the 2-methoxy group in the phenyl
acrylate moiety of compound 10 contributed to the interaction with Leu330, suggesting that the 2-
methoxy group may stabilize the interaction in the hydrophobic binding pocket of PPARγ. Compound 10
formed both hydrogen and hydrophobic bonds with Arg288 (H3), but troglitazone did not. In spite of
differences in numbers and interactions of associated amino acid residues, both troglitazone and
compound 10 formed a horseshoe-shaped conformation spanning both the hydrophilic and the
hydrophobic pockets around helix H3 of the PPARγ LBD, which is a common structural motif of a full

PPARγ agonist.13 The stabilization of helix H12 through the interaction between Tyr473 and the
thiazolidinedione group of PPARγ full agonists is well known as its importance in PPARγ coactivator
binding and transactivation.13 Like troglitazone, compound 10 also formed a hydrogen bond with Tyr473
in the PPARγ LBD interaction model, which suggested that compound 10 acts as a PPARγ agonist.
3.5. The characterization of kojyl cinnamate ester compounds as PPARα and PPARγ dual
modulators in hAT-MSCs
PPARγ is one of the major transcription factors that regulates the adipogenic transcriptional
cascade and is essential for the promotion of adipocyte differentiation.14,15 PPARα is an important
regulator of the lipogenic processes.16 To confirm the dual activation of PPARα and PPARγ by kojyl
cinnamate derivatives during adipogenesis in hAT-MSCs, we measured the mRNA expression of major
PPARα- and/or PPARγ-regulated genes at 24 h after treatments with Wy-14643, troglitazone, compounds
8 and/or 10 (Fig. 5). In the adipogenesis-induced hAT-MCSs, the transcription of adiponectin, a
quantitative biomarker of adipogenesis, was significantly increased by Wy-14643, troglitazone,
compound 8, compound 10, or by co-treatment with Wy-14643 and troglitazone (Fig. 5A). A similar
change was also observed in the expression of fatty acid binding protein 4 (FABP4), a well-established
PPAR target protein associated with the transport of fatty acids in differentiated adipocytes. These results
confirm that the adiponectin secretion-promoting activity of kojyl cinnamate esters 8 and 10 at the level
of gene transcription (Fig. 5B).17

Figure 5. Effects of compounds 8 and 10 on mRNA levels of PPARα- and/or PPARγ-regulated genes
during adipogenesis in hAT-MSCs. hAT-MSCs were differentiated in IDX media and co-treated with
kojyl cinnamate ester compounds 8 and 10. On the third day of the induction, total RNA was extracted
and Q-RT-PCR was performed for adiponectin (A), FABP4 (B), ACAT1 (C), HMGCR (D), HMGCS2
(E), ACACB (F), FASN (G), SCD1 (H) and FADS2 (I). GAPDH was used as an internal control. Values
represent mean ± SD (n = 3). * p ≤ 0.05 and ** p ≤ 0.01.
Next, we analyzed the effects of PPAR agonists and kojyl cinnamate ester compounds on the
mRNA levels of key enzymes regulating lipid metabolism in adipogenesis-induced hAT-MSCs. First, we
evaluated the transcriptional change associated with cholesterol biosynthesis, cytosolic acetyl-coenzyme
A cholesterol acyltransferase (ACAT), 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase (HMGCR),
and 3-hydroxy-3-methylglutaryl-CoA synthase (HMGCS). Co-treatment with Wy-14643 and troglitazone
significantly increased the transcription of ACAT1 (2.07 fold), a mitochondrial ACAT subtype
ubiquitously expressed in tissues (Fig. 5C). Compounds 8 and 10, PPARα/γ dual modulators, also
significantly upregulated the expression of ACAT1 mRNA (1.91 and 2.50 fold, respectively).
Troglitazone increased the transcription of ACAT1, but Wy-14643 did not. The transcription of HMGCR,
a rate-determining enzyme of the cholesterol biosynthetic pathway, was upregulated in adipogenesis-
induced hAT-MSCs in response to Wy-14643, troglitazone, compound 8, compound 10, or by co-

treatment with Wy-14643 and troglitazone (Fig. 5D). Wy-14643, a PPARα agonist, induced a significant
upregulation of HMGCS2 transcription (4.53 fold) compared to the other conditions (Fig. 5E).
Troglitazone did not affect the transcription of HMGCS2. It is well known that HMGCS2, a
mitochondrial enzyme that regulates the first step of cellular ketogenesis, is under the control of
PPARα.18 PPARα/γ dual modulators 8 and 10 increased HMGCS2 transcription, suggesting that they
regulate ketogenesis. We also evaluated whether compounds 8 and 10 affected the transcription of
enzymes regulating cellular fatty acid biosynthesis, such as acetyl-CoA carboxylase (ACACB) and fatty
acid synthase (FASN) (Fig. 5F and 5G). PPARα/γ dual modulators 8 and 10 upregulated mRNA levels of
ACACB and FASN in adipogenesis-induced hAT-MSCs. Co-stimulation of adipogenesis-induced hAT-
MSCs with Wy-14643 and troglitazone also significantly increased the transcription of ACACB and
FASN. In contrast to the effect on HMGCS2, Wy-14643 did not significantly affect the transcription of
ACACB and FASN. Notably, Wy-14643 significantly upregulated the transcription of stearoyl-CoA
desaturase 1 (SCD1) and fatty acid desaturase 2 (FADS2), key genes in the regulation of lipogenesis,
whereas troglitazone did not (Fig. 5H and 5I). PPARα/γ dual modulators 8 and 10 also significantly
increased mRNA levels of both SCD1 and FADS2 in adipogenesis-induced hAT-MSCs. In adipogenesis-
induced hAT-MSCs, HMGCS2, SCD1, and FADS2 showed a similar expression pattern in response to
PPARα and PPARγ modulators and their expression pattern was dependent on PPARα activation. The
expression profile of ACAT1 and ACACB showed similar characteristics, but was different from that of
HMGCS2, SCD1, and FADS2. Interestingly, dual activation of PPARα/γ by either PPARα/γ dual
modulators or co-treatment with mono-agonists significantly upregulated the transcription of FASN.
These gene expression patterns for lipid metabolic enzymes in response to PPARα and/or PPARγ

modulators in adipogenesis-induced hAT-MSCs suggest that PPARα/γ dual modulators 8 and 10 have
differentiated pharmacological features to be used for therapeutic purposes.
3.6. The pharmacological characterization of kojyl cinnamate ester compounds as PPARα and
PPARγ dual modulators in human keratinocytes
JAa2u0cgn0lngi8nee;ri5rcsT8at:el.2dLp5eiJ5prM–sidp2,se6Hc2ate.nilvdlegs.rkeCinnoLnbtIaa,rcJrtieeDmrefrucmnGcBtioEn, –
In skin, the integrity of the skin permeability barrier is associated with various inflammatory
responses.19 With aging, the skin permeability barrier function gradually degenerates and the epidermis of
aged skin exhibits an abnormal lipid composition.20,21 In human epidermis, PPARα and PPARγ are
present in the suprabasal layer of epidermis where keratinocytes differentiate and mature to form the skin
permeability barrier.22 It has also been reported that PPARα and PPARγ activators play a role in repairing
impaired epidermal barrier function by enhancing the biosynthesis of cholesterols, fatty acids and
ceramides.23 Here, we investigated whether PPARα/γ dual modulators 8 and 10 affected the transcription
of these lipid metabolic enzymes in normal human epidermal keratinocytes (NHEKs) (Fig. 6).
Figure 6. Effects of compounds 8 and 10 on mRNA levels of PPARα- and/or PPARγ-regulated genes in
normal human epidermal keratinocytes (NHEKs). Compounds 8 and 10 were treated when the culture
confluency of NHEKs was 100%. Twenty-four hours after treatment with the compounds, total RNA was
extracted and Q-RT-PCR was performed for ACAT1 (A), HMGCS2 (B), FADS1 (C), SCD1 (D), PTGS2

(COX-2) (E), and MMP1 (F). GAPDH was used as an internal control in Q-RT-PCR analysis. Twenty-
four hours after treating NHEKs with compounds 8 and 10, culture supernatants were harvested. The
levels of PGE2 (G) and MMP-1 (H) in the culture supernatants were measured by ELISA. Values
represent mean ± SD (n = 3). * p ≤ 0.05 and ** p ≤ 0.01.
In NHEKs, mRNA expression levels of ACAT1, HMGCS2, FADS1, and SCD1 were
significantly upregulated as a result of PPARα and PPARγ dual activation, either through co-treatment
with 10 μM Wy-14643 and 1 μM troglitazone or with PPARα/γ dual modulators 8 and 10 (Fig. 6A-6D).
In contrast to adipogenesis-induced hAT-MSCs, a significant change was not observed for the
transcription of HMGCR, ACACB, and FASN in NHKEs treated with Wy-14643, troglitazone, or
PPARα/γ dual modulators 8 and 10 (data not shown). It has been reported that topical applications of
either PPARα or PPARγ agonists increase epidermal lipogenesis in hairless mice.23 However, except the
effect of Wy-14643 on the HMGCS2 gene transcription, either a single treatment with 10 μM Wy-14643
or 1 μM troglitazone did not significantly change mRNA levels of other lipid metabolic enzymes in
NHEKs, although significant changes were observed with these concentrations in adipogenesis-induced
hAT-MSCs. The difference in sensitivity to PPARα and PPARγ agonists may be due to the differential
sensitivity of PPAR signaling pathways between NHEKs and adipogenesis-induced hAT-MSCs. The
mRNA levels of ACAT1, HMGCS2, FADS1, and SCD1 were upregulated in NHEKs when co-treated
with 10 μM Wy-14643 and 1 μM troglitazone (Fig. 6A-6D). These results suggest that the dual activation
of PPARα and PPARγ is more effective at increasing the cellular level of lipid metabolic enzymes in
NHEKs compared to a single treatment with a PPARα or a PPARγ agonist. Similar to the co-treatment,
PPARα/γ dual modulators 8 and 10 upregulated the expression of lipogenic enzymes such as ACAT,
HMGCS, FADS, and SCD in NHEKs (Fig. 6A-6D). ACAT mediates the esterification of cholesterol.
HMGCS catalyzes the reaction to generate HMG-CoA, an essential metabolite in both cholesterol

synthesis and ketogenesis. Cholesterol and cholesterol esters contribute to the lamellar structures during
skin barrier formation.19 FADS, a well-known PPAR target gene product, modulates cutaneous
lipogenesis and is crucial for formation of essential long-chain fatty acids, which is important in the
formation of skin barrier lipids.24 Reduced levels of FADS are associated with impaired skin barrier
function.25 The phenotypes of skin specific SCD1-deficient mice are characterized by downregulated
sebaceous lipids, increased cutaneous inflammation, and dysregulated energy metabolism.26 In this regard,
kojyl cinnamate ester PPARα/γ dual modulators 8 and 10 have a therapeutic potential for improving the
weakened skin permeability barrier with skin-aging or during various dermatological conditions.
PPARα or PPARγ agonists have been reported to inhibit cutaneous inflammation in various animal
models.27,28 We also investigated whether PPARα/γ dual modulators 8 and 10 affected ultraviolet (UV) B-
induced proinflammatory responses in NHEKs. UVB-irradiation increased the transcription of
prostaglandin endoperoxide synthase 2 (PTGS2, also known as cyclooxygenase-2, COX-2) and matrix
metalloproteinase 1 (MMP1) in NHEKs (Fig. 6E-6F). Wy-14643, co-stimulation with mono-agonists, and
PPARα/γ dual modulators 8 and 10 significantly inhibited the UVB-induced upregulation of PTGS2
transcription, whereas the single troglitazone treatment condition did not. The inhibition profile of
prostaglandin (PG) E2 in the culture supernatants in NHEKs treated with Wy-14643, troglitazone,
compound 8 and/or compound 10 was correlated to the PTGS2 transcription pattern (Fig. 6G). The UVB
irradiation-induced upregulation of MMP1 transcription in NHEKs was antagonized by all treatment
conditions and a correlated pattern was observed in MMP1 protein levels in the culture supernatants of
NHEKs treated with Wy-14643, troglitazone, compound 8 and/or compound 10 (Fig. 6H). Therefore,
PPARα/γ dual modulators 8 and 10 have pharmacological activity to inhibit UVB-induced upregulation
of proinflammatory PGE2 and MMP1 in NHEKs.

4. Conclusion
Compound 1, a kojyl cinnamate ester derivative, was reported to attenuate ultraviolet (UV) B-induced
inflammation and increase dermal collagen synthesis in a human clinical study, leading to the positive
mass balance of collagen biosynthesis.7 The molecular mechanism for the effect of compound 1 on
dermal collagen synthesis was not identified.7 We recently found that compound 1 increases adiponectin
production during adipogenesis in hAT-MSCs, which explains one of phenotypic pharmacological
mechanisms to increase collagen synthesis.9 However, the adiponectin mechanism of action by compound
1 was not enough to explain the phenotypic inhibition of cutaneous inflammation in human skin. In this
study, we synthesized more potent adiponectin secretion-promoting kojyl cinnamate ester compounds 8
and 10 in order to identify the responsible molecular target of compound 1 to explain its adiponectin-
promoting activity. In receptor binding assays and molecular docking simulations, we elucidated
compounds 8 and 10 as PPARα/γ dual modulators. Compounds 8 and 10 upregulated the transcription of
ACAT1, ACACB, HMGCR, HMGCS2, FASN, FADS2, and SCD1, lipid metabolic enzymes in
adipogenesis-induced hAT-MSCs. Compared to the effects of Wy-14643, a PPARα agonist, or
troglitazone, a PPARγ agonist, on the transcription of lipid metabolic enzymes in hAT-MSCs, PPARα/γ
dual modulators 8 and 10 have differentiated characteristics. In addition, PPARα/γ dual modulators 8 and
10 led to greater increases in the mRNA levels of ACAT1, HMGCS2, FADS1, and SCD1 in NHEKs
compared to Wy-14643 and/or troglitazone. The upregulation of ACAT1, HMGCS2, FADS1, and SCD1
in human keratinocytes induced by the PPARα/γ dual modulators 8 and 10 may improve the weakened
integrity of the skin permeability barrier with skin aging. Importantly, PPARα/γ dual modulators 8 and 10
inhibited the UVB-irradiation-induced increase in PGE2 and MMP-1 in NHEKs. In conclusion, the
adiponectin secretion-promoting activity of kojyl cinnamate ester compounds was significantly correlated

with their PPARα/γ dual modulator activity. The clinically proven anti-skin aging activity of kojyl
cinnamate ester compounds may be explained by their PPARα/γ dual modulation in both subcutaneous
adipocytes and epidermal keratinocytes.
Acknowledgement
This study was supported by a grant from the Korea Healthcare Technology R&D Project, Ministry of
Health & Welfare, Republic of Korea (Grant No. HN14C0088), a National Research Foundation of Korea
(NRF) grant (2015R1A2A2A01008408), the MRC grant through NRF Korea (NRF-
2018R1A5A2024425), and a grant by Promising-Pioneering Researcher Program through the Seoul
National University (SNU) in 2015.
References and notes
1. Schenone M, Dancik V, Wagner BK, et al. Target identification and mechanism of action in chemical
biology and drug discovery. Nat Chem Biol. 2013;9:232-240.
2. Noh M. Interleukin-17A increases leptin production in human bone marrow mesenchymal stem cells
Biochem Pharmacol. 2012;83:661-670.
3. Cartwright MJ, Tchkonia T, Kirkland JL. Aging in adipocytes: Potential impact of inherent, depot-
specific mechanisms. Exp Gerontol. 2007;42:463-471.
4. Cristancho AG, Lazar MA. Forming functional fat: a growing understanding of adipocyte
differentiation. Nat Rev Mo Cell Biol. 2011;12:722-734.
5. Rosen ED, MacDougald OA. Adipocyte differentiation from the inside out. Nat Rev Mol Cell Biol.
2006;7:885-896.
6. Dadson K, Chasiotis H, Wannaiampikul S, et al. Adiponectin Mediated APPL1‐AMPK Signaling
Induces Cell Migration, MMP Activation, and Collagen Remodeling in Cardiac Fibroblasts. J Cell
Biochem. 2014;115:785-793.

7. Kim MS, Lee S, Rho HS, et al. The effects of a novel synthetic retinoid, seletinoid G, on the expression
of extracellular matrix proteins in aged human skin in vivo. Clin Chim Acta. 2005;362:161-169.
8. Choi JH, Banks AS, Estall JL, et al. Anti-diabetic drugs inhibit obesity-linked phosphorylation of
PPARγ by Cdk5. Nature. 2010;466:451-456.
9. Rho HS, Hong SH, Park J, et al. Kojyl cinnamate ester derivatives promote adiponectin production
during adipogenesis in human adipose tissue-derived mesenchymal stem cells. Bioorg Med Chem Lett.
2014;24:2141-2145.
10. Yu J, Ahn S, Kim HJ, et al. Polypharmacology of N6-(3-Iodobenzyl)adenosine-5'-N-
methyluronamide (IB-MECA) and Related A3 Adenosine Receptor Ligands: Peroxisome Proliferator
Activated Receptor (PPAR) γ Partial Agonist and PPARδ Antagonist Activity Suggests Their
Antidiabetic Potential. J Med Chem. 2017;60:7459-7475.
11. Bernardes A, Souza PC, Muniz JR, et al. Molecular mechanism of peroxisome proliferator-activated
receptor α activation by WY14643: a new mode of ligand recognition and receptor stabilization. J Mol
Biol. 2013;425:2878–2893.
12. Oberfield JL, Collins JL, Holmes CP, et al. A peroxisome proliferator-activated receptor gamma
ligand inhibits adipocyte differentiation. Proc Natl Acad Sci USA.1999;96:6102–6106.
13. Ahn S, Lee M, An S, et al. 2-Formyl-komarovicine promotes adiponectin production in human
mesenchymal stem cells through PPARγ partial agonism. Bioorg Med Chem. 2018;26:1069-1075.
14. Farmer SR. Regulation of PPARgamma activity during adipogenesis. Int J Obes.(Lond). 2005;29
Suppl 1:S13-6.
15. Ahmadian M, Suh JM, Hah N, et al. PPARγ signaling and metabolism: the good, the bad and the
Future. Nat Med. 2013;19:557-566.
16. Rivier M, Castiel I, Safonova I, et al. Peroxisome Proliferator-Activated Receptor-α Enhances Lipid
Metabolism in a Skin Equivalent Model. J Invest Dermatol. 2000;114:681-687.

17. Furuhashi M, Saitoh S, Shimamoto K, et al. Fatty Acid-Binding Protein 4 (FABP4):
Pathophysiological Insights and Potent Clinical Biomarker of Metabolic and Cardiovascular Diseases.
Clin Med Insights Cardiol. 2015;8:23–33.
18. Rodríguez JC, Gil-Gómez G, Hegardt FG, et al. Peroxisome proliferator-activated receptor mediates
induction of the mitochondrial 3-hydroxy-3-methylglutaryl-CoA synthase gene by fatty acids. J Biol
Chem.1994;269:18767-18772.
19. Jungersted JM, Hellgren LI, Jemec GB, et al. Lipids and skin barrier function--a clinical perspective.
Contact Dermatitis. 2008;58:255-262.
20. Ghadially R, Brown BE, Sequeira-Martin SM, et al. The aged epidermal permeability barrier.
Structural, functional, and lipid biochemical abnormalities in humans and a senescent murine model. J
Clin Invest. 1995;95:2281-2290.
21. Ghadially R, Brown BE, Hanley K, et al. J Invest Dermatol. 1996;106:1064-1069.
22. Schmuth M, Moosbrugger-Martinz V, Blunder S, et al. Role of PPAR, LXR, and PXR in epidermal
homeostasis and inflammation. Biochim Biophys Acta. 2014;1841:463-473.
23. Man MQ, Choi EH, Schmuth M, et al. Basis for improved permeability barrier homeostasis induced
by PPAR and LXR activators: liposensors stimulate lipid synthesis, lamellar body secretion, and post-
secretory lipid processing. J Invest Dermatol. 2006;126:386-392.
24. Echeverría F, Ortiz M, Valenzuela R, et al. Long-chain polyunsaturated fatty acids regulation of
PPARs, signaling: Relationship to tissue development and aging. Prostaglandins Leukot Essent Fatty
Acids. 2016;114:28-34.
25. Sampath H, Ntambi JM. The role of fatty acid desaturases in epidermal metabolism.
Dermatoendocrinol. 2011;3:62-64.
26. Sampath H, Ntambi JM. Role of stearoyl-CoA desaturase-1 in skin integrity and whole body energy
balance. J Biol Chem. 2014;289:2482-2488.