Polyacetylenes from Notopterygium incisum–New Selective Partial Agonists of Peroxisome Proliferator-Activated Receptor-Gamma

Peroxisome proliferator-activated receptor gamma (PPARγ) is a key regulator of glucose and lipid metabolism and therefore an important pharmacological target to combat metabolic diseases. Since the currently used full PPARγ agonists display serious side effects, identification of novel ligands, particularly partial agonists, is highly relevant. Searching for new active compounds, we investigated extracts of the underground parts of Notopterygium incisum, a medicinal plant used in traditional Chinese medicine, and observed significant PPARγ activation using a PPARγ-driven luciferase reporter model. Activity-guided fractionation of the dichloromethane extract led to the isolation of six polyacetylenes, which displayed properties of selective partial PPARγ agonists in the luciferase reporter model. Since PPARγ activation by this class of compounds has so far not been reported, we have chosen the prototypical polyacetylene falcarindiol for further investigation. The effect of falcarindiol (10 µM) in the luciferase reporter model was blocked upon co-treatment with the PPARγ antagonist T0070907 (1 µM). Falcarindiol bound to the purified human PPARγ receptor with a Ki of 3.07 µM. In silico docking studies suggested a binding mode within the ligand binding site, where hydrogen bonds to Cys285 and Glu295 are predicted to be formed in addition to extensive hydrophobic interactions. Furthermore, falcarindiol further induced 3T3-L1 preadipocyte differentiation and enhanced the insulin-induced glucose uptake in differentiated 3T3-L1 adipocytes confirming effectiveness in cell models with endogenous PPARγ expression. In conclusion, we identified falcarindiol-type polyacetylenes as a novel class of natural partial PPARγ agonists, having potential to be further explored as pharmaceutical leads or dietary supplements.


Introduction
Peroxisome proliferator activated receptors (PPARs) are liganddependent nuclear receptors which play a major role in lipid and glucose metabolism [1][2][3][4][5]. Three subtypes of PPAR, namely PPARa, PPARb/d, and PPARc, have been identified in human and other species [6,7]. Upon ligand binding, PPARs form heterodimers with another nuclear receptor, the retinoid X receptor (RXR) and subsequently bind to response elements located in the promoter region of their target genes [8]. The binding of a PPAR/RXR heterodimer to its PPAR response element (PPRE) triggers the recruitment of nuclear receptor coactivators [9,10], and subsequent chromatin rearrangement allowing initiation of the transcription of target genes [11].
The three PPAR subtypes are expressed in different tissues and regulate distinct physiological functions. PPARa is mainly expressed in muscle, liver, heart, and kidney, and is involved in the regulation of genes that play a role in lipid and lipoprotein metabolism [12,13]. The PPARb/d subtype is ubiquitously expressed in various tissues and predominantly associated with lipid metabolism and energy expenditure [14]. Among the three PPAR subtypes, PPARc is the best studied one. PPARc is expressed in adipose tissue, lung, large intestine, kidney, liver, heart, and macrophages [15], and is engaged in the regulation of adipogenesis, insulin sensitivity, and inflammation [16]. The important functions of PPARc in the regulation of glucose and lipid metabolism make it an attractive pharmacological target for combating metabolic diseases [17][18][19]. It is established that the activation of this receptor mediates the action of the glitazone-type drugs clinically used for the treatment of type 2 diabetes [20]. Furthermore, recent findings supporting a role of PPARc in inflammation and cell growth have promoted the investigation of PPARc agonists as experimental drugs for some chronic diseases such as atherosclerosis and cancer, among others [21][22][23]. The glitazones currently used in clinics (e.g., pioglitazone) potently activate PPARc as full agonists. In spite of being effective drugs, their use is limited due to undesirable side effects [24], urging the identification and characterization of new PPARc agonists. A number of recent studies revealed that partial PPARc agonists inducing submaximal receptor activation are demonstrating great promise by exerting good hypoglycaemic activity with reduced side effects [25][26][27].
Natural products encompass a broad structural diversity of secondary metabolites, which often represent privileged structures serving a variety of biological functions [28]. Therefore, we focused on the identification of novel PPARc activators derived from natural sources. As part of an ethnopharmacology-based screening approach, extracts from medicinal plants used in traditional Chinese medicine to treat inflammation-related conditions were tested for PPARc activation. Extracts of the rhizomes and roots of Notopterygium incisum displayed significant PPARc activation. The plant has been traditionally used in China for the treatment of rheumatism, cold, and headache [29], and its lipophilic extracts have shown inhibitory effects on 5-lipoxygenase and cyclooxygenase [30]. Pure compounds isolated from the dichloromethane extracts were now examined for PPARc activation, leading to the identification of falcarindiol-type polyacetylenes as a novel group of natural selective partial PPARc agonists. transiently co-transfected with a plasmid encoding full-length human PPARc, a reporter plasmid containing PPRE coupled to a luciferase reporter and an EGFP plasmid as internal control. After re-seeding, cells were treated as indicated for 18 h. Since the positive control pioglitazone (5 mM), as well as the dried DCM-and MeOH-extract were reconstituted in DMSO, cells were treated with equal amount of the solvent vehicle DMSO (0.1%) as negative control. The luciferase activity was normalized to the EGFP-derived fluorescence, and the result is expressed as fold induction compared to the solvent vehicle control. The data shown are means 6 SD of three independent experiments each performed in quadruplet. ***p,0.001, *p,0.05 (compared to the solvent vehicle group; ANOVA/Bonferroni). (B) Chemical structures of the PPARc-activating polyacetylenes isolated from N. incisum. doi:10.1371/journal.pone.0061755.g001

Chemicals, Cell Culture Reagents, and Plasmids
Calf serum, L-glutamine, and Dulbecco's modified Eagle's medium (DMEM) containing 4.5 g/l glucose was purchased from Lonza (Basel, Switzerland). Fetal bovine serum (FBS) was supplied from Gibco (Lofer, Austria). PPARa and PPARb/d agonists GW7647 and GW0742, respectively, as well as the PPARc antagonist T0070907, were purchased from Cayman (Missouri, USA); pioglitazone was purchased from Molekula Ltd (Shaftesbury, UK). All other chemicals were obtained from Sigma-Aldrich (Vienna, Austria). The tested compounds, or dried plant extracts, were dissolved in dimethyl sulfoxide (DMSO), aliquoted, and stored at 220uC until use. The final concentration of DMSO in all experiments was 0.1% or lower. An equal amount of DMSO was always tested in each experiment to assure that the solvent vehicle does not influence the results. The three human PPAR subtype expression plasmids (pSG5-PL-hPPAR-alpha, pSG5-hPPAR-beta, pSG5-PL-hPPAR-gamma1) were a kind gift from Prof. Beatrice Desvergne and Prof. Walter Wahli (Center for Integrative Genomics, University of Lausanne, Switzerland), and the luciferase reporter plasmid (tk-PPREx3-luc) was kindly provided by Prof. Ronald M. Evans (Salk Institute for Biological Studies, San Diego, California).

PPAR Luciferase Reporter Gene Transactivation
HEK-293 cells (ATCC, USA) were cultured in DMEM with phenol red, supplemented with 100 U/ml benzylpenicillin, 100 mg/ml streptomycin, 2 mM L-glutamine, and 10% FBS. Cells were maintained in 75 cm 2 flasks containing 10 ml medium at 37uC and 5% CO 2 . The cells were seeded in 10 cm dishes at a density of 6 6 10 6 cells/dish, incubated for 18 h, and transfected by the calcium phosphate precipitation method [31] with 4 mg of the reporter plasmid (tk-PPREx3-luc), 4 mg from the respective PPAR subtype expression plasmid, and 2 mg green fluorescent protein plasmid (pEGFP-N1, Clontech, Mountain View, CA) as internal control. After 6 h, the transfected cells were harvested and re-seeded (5 6 10 4 cells/well) in 96-well plates containing DMEM without phenol red, supplemented with 100 U/ml benzylpenicillin, 100 mg/ml streptomycin, 2 mM L-glutamine, and 5% charcoal-stripped FBS. The cells were further treated with the indicated extracts or compounds or the solvent vehicle and incubated for 18 h. The medium was then discarded and the cells were lysed with a reporter lysis buffer (E3971, Promega, Madison, USA). Luciferase activity of the cell lysates was evaluated using a GeniosPro plate reader (Tecan, Grödig, Austria). The luminescence signals obtained from the luciferase activity measurements were normalized to the EGFP-derived fluorescence, to account for differences in the transfection efficiency or cell number. Luciferase reporter gene assays were performed with polyacetylene concentrations up to 30 mM as higher dosages led to a downregulation of the internal normalization control (EGFP, not shown).  (5) and 2Z,9Z-heptadecadiene-4,6-diyne-1-ol (6). The purity of the isolated compounds was over 95% for 1-4 and above 90% for 5 and 6. All analytical HPLC experiments were performed using an Agilent 1100 series system (Agilent 1100, Series Degaser G1311 A, Quat Pump G1311, Autosampler G1313 A, Colcom G1316 A, DAD 1315 B) equipped with a diode-array detector. Analyses were performed using a SB-C18 Zorbax column (3.5 mm; 15062.1 mm; Agilent Technologies) at a flow rate of 300 ml/min and a gradient elution program. The extract Table 1. Activity of the isolated polyacetylenes towards the three subtypes of human PPAR (a, b/d, c) in a luciferase reporter transactivation assay. was phytochemically investigated using a combination of different chromatographic techniques such as LC, SPE (RP-C18), and preparative HPLC (RP-C18). Isolated constituents were characterized and identified by multidimensional NMR spectroscopy ( 1 H, 13 C, COSY, HMBC, HSQC) and mass spectrometry.

Receptor Binding Study
The LanthaScreenH time-resolved fluorescence resonance energy transfer (TR-FRET) PPARc competitive binding assay (Invitrogen, Lofer, Austria) was performed according to the manufacturer's protocol. The test compounds dissolved in DMSO or the blank solvent vehicle (DMSO) were incubated together with the human PPARc ligand-binding domain (LBD) tagged with GST, a terbium-labeled anti-GST antibody, and a fluorescently labeled PPAR ligand (Fluormone Pan-PPAR Green, Invitrogen). In this assay, the fluorescently labeled ligand is binding to the human PPARc LBD, which brings it in close spatial proximity to the terbium-labeled anti-GST antibody. Excitation of the terbium at 340 nm results in energy transfer (FRET) and partial excitation of the fluorescently labeled PPAR ligand, followed by emission at 520 nm. Test-compounds binding to the human PPARc LBD are competing with the fluorescently labeled ligand and displacing it, resulting in a decrease of the FRET signal. The signal obtained at 520 nm is normalized to the signal obtained from the terbium emission at 495 nm; therefore, the decrease in the 520 nm/ 495 nm ratio is used as a measure for the ability of the tested compounds to bind to the human PPARc LBD. All measurements were performed with a GeniosPro plate reader (Tecan, Grödig, Austria).

Molecular Docking
To predict binding modes of the falcarindiol-type polyacetylene PPARc partial agonists, a docking study was performed, using the quantum mechanics-polarized ligand docking (QPLD) workflow [32] in Maestro version 9.2.112 (Schrödinger, LLC, New York, NY, 2011; http://www.schrodinger.com). In this docking study, the X-ray crystal structure of PPARc with two co-crystallized molecules of the natural product magnolol was used (Protein Data Bank entry: 3r5n; http://www.rcsb.org/pdb/) [33]. The ligand and protein preparation were conducted using the Ligprep module with default settings for the ligands and the Protein Preparation Wizard for the protein. The protein was prepared by adding hydrogens, removing water molecules, assigning atom and bond types, and a refinement by completing missing side chains. Then, an exhaustive exploration of the hydrogen bond assignment was conducted. The protein was minimized applying the OPLS-AA 2005 force field with a threshold of 0.3 Å RMSD. Afterward, atomic partial charges of the ligands were calculated within the QPLD workflow, applying the semi-empirical method AM1. Furthermore, the initial docking was conducted using Glide in standard precision (SP) mode, basically default settings, and returning up to five diverse poses for each ligand. These intermediate poses were submitted to the second docking procedure, after calculating the atomic partial charges of the ligands based upon the QM/MM approach performed by the module QSite. Mulliken population analysis was performed for the calculation of ligand atomic partial charges, which is based on the proposed orientation of the ligands relative to the receptor, and, therefore, influenced by the electrostatic field of the receptor. Afterwards, the final docking poses were retrieved using Glide in SP mode. For the final ranking of the docking poses, several scoring functions were evaluated, from which the scoring from Emodel calculations achieved the best results in several runs. To theoretically validate the docking workflow, the docking poses of magnolol were compared to its native pose in the X-ray crystal structure by calculating the geometric RMSD. The highest-ranked pose had an RMSD of 0.555 Å and the mean RMSD of all 10 calculated poses was 0.554 Å , respectively. Finally, the docking poses of falcarindiol were visually inspected and analyzed within the program LigandScout 3.03 [34] (Inte:Ligand GmbH, Maria Enzersdorf, Austria, 2012; http://www.inteligand.com).

Adipogenicity of Falcarindiol
3T3-L1 preadipocytes (ATCC, USA) were propagated in DMEM supplemented with 10% calf serum. For differentiation, preadipocytes were grown to confluence (day 22) and kept for two additional days before the medium was changed to DMEM supplemented with 10% FBS, 1 mg/ml insulin, and falcarindiol (10 mM) or the full PPARc agonist pioglitazone (10 mM) (day 0). The medium was renewed every two days until day 8. For an estimate of lipid accumulation, Oil Red O staining was performed. For this, cells were fixed in 10% formaldehyde for 1 h and stained with Oil Red O for 10 min. After washing off excessive dye, bound dye was solubilized with 100% isopropanol and photometrically quantified at 550 nm. For better appreciation of the adipogenic potential of the investigated PPARc agonists, a subset of preadipocytes was treated according to a common standard differentiation protocol (2 days in DMEM/10% FBS/500 mM IBMX/500 nM dexamethasone and 1 mg/mL insulin (MDI), followed by 2 days in DMEM/10% FBS/1 mg/ml insulin and 6 days in DMEM/10% FBS).  CA-630), and normalized to protein content assessed by the Bradford protein assay [35].

Novelty Evaluation
The novelty evaluation of the findings was performed with PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) and SciFinder (https://scifinder.cas.org/) using ''falcarindiol'' as keyword, resulting in 99 and 364 literature references, respectively (as of 13 th of December, 2012). None of the retrieved references was related to PPAR modulation. To further evaluate the novelty of the chemical scaffold of falcarindiol as PPAR ligand, chemical similarity search was performed with SciFinder, using the query structure of falcarindiol with a Tanimoto similarity score .80%. As outcome 714 similar chemical structures, associated with 1367 literature references were retrieved. None of the obtained literature references was related to PPAR.

Statistical Methods and Data Analysis
All statistical analyses were done with the GraphPad Prism software version 4.03. One-way analysis of variance (ANOVA) with Bonferroni post hoc test, or two-tailed paired t-test was used to determine statistical significance. Nonlinear regression (sigmoidal dose response) was used to calculate the EC 50 values and maximal fold activation. K i values of the investigated compounds in the receptor binding assay were calculated with the Cheng-Prusoff equation: (K i ) = IC 50 /(1+ L/K D )), where IC 50 is the concentration of competitor that displaces 50% of the ligand, L is the concentration of Fluormone TM Pan-PPAR Green used in the assay (5 nM), and K D is the binding constant (2.8 nM) of Fluormone TM Pan-PPAR Green to PPARc LBD [36].

Identification of Polyacetylenes from Notopterygium Incisum as Selective Partial PPARc Agonists
Aiming to identify novel PPARc ligands from natural sources with an ethnopharmacological background, we used a PPARcdriven luciferase reporter gene assay (Fig. 1A) and observed a significant PPARc activation exerted by the DCM (2.1660.28fold activation, p,0.001) and MeOH (1.6860.49-fold activation, p,0.05) extracts from N. incisum roots and rhizomes applied at 10 mg/ml (Fig. 1A). A bioactivity-guided approach using the DCM extract led to the isolation of six active polyacetylenes: 8acetoxyfalcarinol (1), falcarindiol (2), 9-epoxy-falcarindiol (3), crithmumdiol (4), 9Z-heptadecene-4,6-diyne-1-ol (5), and 2Z,9Zheptadecadiene-4,6-diyne-1-ol (6) (Fig. 1B). All polyacetylenes activated PPARc as partial agonists with maximal activation ranging from 1.73-to 3.26-fold compared to the blank treatment control, while the full agonist pioglitazone activating PPARc up to 9.28-fold. The observed activating effect was specific for PPARc with no activation of PPARa or PPARb/d (Table 1). Falcarindiol (2) is one of the most abundant polyacetylenes that we identified in N. incisum. Furthermore, it is one of the best investigated polyacetylenes in general, and is broadly found in many other plant species. Since all active compounds that we identified from N. incisum were polyacetylenes, and since PPARc activating properties of this class of compounds have not been reported so far (see also ''Novelty evaluation'' in the ''Materials and Methods'' section), we have chosen the prototypical polyacetylene falcarindiol for further pharmacological characterization.

Transactivation Activity of Falcarindiol in the Luciferase Reporter Model
As a partial PPARc agonist, falcarindiol transactivated the luciferase reporter with an EC 50 of 3.29 mM (95% CI: 2.33-4.64 mM) inducing a maximum of 3.26 (95% CI: 2.90-3.61)-fold activation, whereas the full PPARc agonist pioglitazone displayed an EC 50 of 0.41 mM (95% CI: 0.25-0.67 mM) and a maximum of 9.28 (95% CI: 8.08-10.48)-fold activation (Fig. 2A). The stimulating effect of 10 mM falcarindiol was blocked upon cotreatment with the PPARc antagonist T0070907 confirming the PPARc dependence of the observed activation (Fig. 2B). Noteworthy, the antagonistic effect of T007090907 was more pronounced when the receptor was activated with pioglitazone than with falcarindiol, an observation which is in agreement with previous studies demonstrating that the antagonistic potency of T007090907 is dependent on the type of ligand used to activate PPARc [37]. Furthermore, co-treatment with falcarindiol blocked up to 70-80% of the stimulating effect of pioglitazone (1 mM) with an IC 50 of 18.14 mM (95% CI: 14.32 to 22.97 mM), suggesting competition of the two compounds for the same binding site of the receptor (Fig. 2C).

Receptor Binding and Molecular Modeling
To further demonstrate the direct receptor binding of falcarindiol we performed a TR-FRET competitive binding assay with purified human PPARc-LBD (Fig. 2D). In this model falcarindiol and pioglitazone displayed K i values of 3.07 mM and 0.27 mM, respectively. Noteworthy, there was a very good correlation between the EC 50 values obtained with falcarindiol and pioglitazone in the cell based luciferase reporter model (3.29 mM and 0.41 mM, respectively) and the K i values obtained in the receptor binding assay with purified PPARc-LBD (3.07 mM and 0.27 mM, respectively).
In order to gain insight into the molecular interaction of the tested polyacetylene-type PPARc agonists with the receptor binding pocket of the protein, docking studies were performed (Fig. 3). The composition and overall structure of the PPARc LBD have been analyzed previously, describing the PPARc LBD with a Y-shaped topology [38]. The entrance includes several polar residues (e.g., Arg288, Glu291, Glu295, and Glu343), while the remaining two branches of the binding pocket, arm I and arm II, are composed mostly of hydrophobic residues, except a few moderately polar residues in arm I (e.g., Cys285, Ser289, His323, His449, and Tyr473) [38]. Consistent with the relatively similar structure (Fig. 1B) and bioactivity (Table 1) of the investigated polyacetylenes, all compounds (1-6) docked to the PPARc LBD in a quite similar manner as depicted for falcarindiol ( Fig. 3A and  3B). The molecular binding mode of the falcarindiol-type polyacetylene PPARc partial agonists is proposed to consist of hydrophobic contacts of the ligand alkyl chains with residues of arm II (e.g., Ile281) and arm I (e.g., Ile326, Tyr327, Phe363) as well as residues of the entrance region (e.g., Ala292, Met329, Leu330, and Leu333). The hydroxyl groups form hydrogen bonds with the backbone amide of Cys285 in arm I, and the carboxylate group of Glu295 in the entrance (Fig. 3A and 3B).

Effectiveness of Falcarindiol in Cell Models with Endogenous Expression of PPARc
A main target site involved in the insulin-sensitizing action of the clinically used PPARc agonists is the fat tissue [39,40]. To further confirm the effectiveness of falcarindiol in a functionally relevant cell model with endogenous expression of PPARc, we examined the effect of this compound on 3T3-L1 adipocyte differentiation and glucose uptake. As evident by the accumulation of lipid droplets and subsequent Oil Red O staining, the treatment with 10 mM falcarindiol induced the differentiation to adipocytes as expected, although to a smaller extent than pioglitazone at the same concentration (Fig. 4A). Furthermore, falcarindiol significantly enhanced the insulin-mediated glucose uptake in differentiated 3T3-L1 adipocytes (Fig. 4B).

Discussion
In this study we identify polyacetylenes from N. incisum as representatives of a novel group of natural products activating PPARc as partial agonists, and characterize in more detail the mode of action of the prototypical polyacetylene falcarindiol.
C(17)-polyacetylenes of the falcarindiol-type are found in many plants of the Apiaceae family, including commonly used vegetables and seasonings such as carrots, celery, fennel, dill, parsley, and parsnip [41,42]. A variety of bioactivities have been reported so far for this compound class which may contribute to the health promoting effects of certain edible plants as well as medicinal plants used in traditional medicine [43]. The possibility that this compound class may modulate the activity of PPARc, however, was so far not investigated. Since PPARc is a key regulator of a number of pathways related to lipid and carbohydrate metabolism, this nuclear receptor represents an important pharmacological target for the treatment of diabetes type 2 and the metabolic syndrome. Furthermore, PPARc appears to be a suitable target for modulation by diet or food supplements, since PPARc activity was reported to be enhanced by a number of common food constituents or natural products present in certain seasonings or medicinal plants [44][45][46]. All polyacetylenes from N. incisum that were investigated in this study displayed a partial PPARc agonism with a maximal activation several folds lower than the activation induced by the full PPARc agonist pioglitazone (Table 1, Fig. 1B, and Fig. 2A). Docking studies ( Fig. 3A and 3B) suggest receptor binding in a manner similar to other partial PPARc agonists, which have been co-crystallized with the receptor, indicating that a hydrogen-bonding network formed by full agonists to residues of arm I is not formed in a comparable manner or even lacking in case of partial agonists [47,48]. The partial agonism makes the polyacetylenes an interesting class of novel PPARc ligands, since it is currently considered that the development of new partial agonists that do not activate PPARc as strongly as the glitazonetype of full agonists might be a way to retain the needed pharmaceutical effectiveness with reduced side effects [49,50]. Furthermore, the investigated polyacetylenes selectively activated PPARc without activating PPARa or PPARb/d (Table 1). This is another rather favorable profile of action since the PPARc agonists currently approved on the market are isoform-specific PPARc activators, although there is some research evidence suggesting possible advantages for PPAR dual agonists or panagonists [50]. Very interestingly, previous studies have reported an antidiabetic activity of polyacetylenes isolated from Bidens pilosa upon oral administration in mice [51], possibly indicating effective modulation of PPARc activity in vivo.
In summary, we report the identification of polyacetylenes as novel class of natural products activating PPARc. The observed partial activation mode makes these compounds highly interesting PPARc ligands, having the potential to be further explored as pharmaceutical leads or dietary supplements.