Functional and Structural Insights into Human PPARα/δ/γ Subtype Selectivity of Bezafibrate, Fenofibric Acid, and Pemafibrate

Among the agonists against three peroxisome proliferator-activated receptor (PPAR) subtypes, those against PPARα (fibrates) and PPARγ (glitazones) are currently used to treat dyslipidemia and type 2 diabetes, respectively, whereas PPARδ agonists are expected to be the next-generation metabolic disease drug. In addition, some dual/pan PPAR agonists are currently being investigated via clinical trials as one of the first curative drugs against nonalcoholic fatty liver disease (NAFLD). Because PPARα/δ/γ share considerable amino acid identity and three-dimensional structures, especially in ligand-binding domains (LBDs), clinically approved fibrates, such as bezafibrate, fenofibric acid, and pemafibrate, could also act on PPARδ/γ when used as anti-NAFLD drugs. Therefore, this study examined their PPARα/δ/γ selectivity using three independent assays—a dual luciferase-based GAL4 transactivation assay for COS-7 cells, time-resolved fluorescence resonance energy transfer-based coactivator recruitment assay, and circular dichroism spectroscopy-based thermostability assay. Although the efficacy and efficiency highly varied between agonists, assay types, and PPAR subtypes, the three fibrates, except fenofibric acid that did not affect PPARδ-mediated transactivation and coactivator recruitment, activated all PPAR subtypes in those assays. Furthermore, we aimed to obtain cocrystal structures of PPARδ/γ-LBD and the three fibrates via X-ray diffraction and versatile crystallization methods, which we recently used to obtain 34 structures of PPARα-LBD cocrystallized with 17 ligands, including the fibrates. We herein reveal five novel high-resolution structures of PPARδ/γ–bezafibrate, PPARγ–fenofibric acid, and PPARδ/γ–pemafibrate, thereby providing the molecular basis for their application beyond dyslipidemia treatment.


Introduction
Peroxisome proliferator-activated receptors (PPARs) belong to the nuclear receptor (NR) superfamily and ligand-activated transcription factors that sense intracellular free fatty acids [1]. Three subtypes (PPARα, PPARβ/δ, and PPARγ) with considerable amino acid identity (54-71% in humans) have been identified in mammals. PPARα regulates lipid metabolism mainly in the liver and skeletal muscle and glucose homeostasis via direct transcriptional control of genes involved in peroxisomal/mitochondrial β-oxidation, fatty acid uptake, and triglyceride (TG) catabolism [2]. PPARδ is ubiquitously expressed and controls energy metabolism and cell survival [3]. PPARγ is most highly expressed in white/brown adipose tissues, where it acts as a master regulator of adipogenesis and manner with maximal effect (×5.73 of the basal activity) at 0.1 µM and an EC 50 value of 8.18 nM ( Figure 1A); a potent PPARδ-selective agonist GW501516 induced PPARδ-LBDmediated transactivation with maximal effect (×18.3) at 0.02 µM and an EC 50 of 1.59 nM ( Figure 1B); and a potent PPARγ-selective agonist GW1929 induced PPARγ-LBD-mediated transactivation with maximal effect (×2.26) at 1 µM and an EC 50 of 18.3 nM ( Figure 1C). only by the GAL4-human PPARα/δ/γ-LBD chimera, and the potentially confounding effects of endogenous receptors were eliminated. Transfection efficiency was normalized by Renilla luciferase activity. First, we confirmed that a potent PPARα-selective agonist GW7647 induced PPARα-LBD-mediated transactivation in a concentration-dependent manner with maximal effect (×5.73 of the basal activity) at 0.1 µ M and an EC50 value of 8.18 nM ( Figure 1A); a potent PPARδ-selective agonist GW501516 induced PPARδ-LBDmediated transactivation with maximal effect (×18.3) at 0.02 µ M and an EC50 of 1.59 nM ( Figure 1B); and a potent PPARγ-selective agonist GW1929 induced PPARγ-LBD-mediated transactivation with maximal effect (×2.26) at 1 µ M and an EC50 of 18.3 nM ( Figure  1C). Fenofibrate is a prodrug that is metabolized by tissue and plasma carboxylesterases [22] to its active form, fenofibric acid. (E-G) PPARα/δ/γ-LBD-mediated transactivation by bezafibrate (E), fenofibric acid (F), and pemafibrate (G). Data are means ± standard error (SE) of three independent experiments with duplicate samples, and calculated EC50 values are shown. Fenofibrate is a prodrug that is metabolized by tissue and plasma carboxylesterases [22] to its active form, fenofibric acid. (E-G) PPARα/δ/γ-LBD-mediated transactivation by bezafibrate (E), fenofibric acid (F), and pemafibrate (G). Data are means ± standard error (SE) of three independent experiments with duplicate samples, and calculated EC 50 values are shown.

Fibrates Induce PPARγ Coactivator 1α (PGC1α) or Steroid Receptor Coactivator 1 (SRC1) Recruitment via PPARα/δ/γ-LBD
In the nucleus, PPARs remain largely in repressed states due to the presence of corepressors, such as nuclear receptor corepressor 1 (NCoR1) and NCoR2 (SMRT), bound to the cis-elements (PPREs: PPAR responsive elements) located in the promoter region of their multiple target genes, irrespective of their ligand binding status [8]. Ligand binding initiates a complicated transcription process, which includes the dissociation of the corepressor protein complexes and the association of coactivator protein complexes for linking to the basal transcription machinery [8]. The ligand-induced AF-2 helix 12 formation, which recruits coactivators such as PGC1α and SRC1, is a hallmark of PPAR activation. PGC1α has a preference for PPARα/γ and is highly expressed in brown adipose tissues and cardiac and skeletal muscles, whereas SRC1 has a preference for all PPAR subtypes and is highly expressed in brown and white adipose tissues and the brain [8]. Thus, timeresolved fluorescence resonance energy transfer (TR-FRET)-based detection of the physical association between PPARα/δ/γ-LBD and coactivators becomes a highly sensitive cell-free assay for evaluating PPAR ligand activities. First, we confirmed that GW7647 activates the recruitment of both coactivators in a concentration-dependent manner, with maximal effect (×8.87 of the basal activity) at 1 µM and 44.1 nM EC 50 for PGC1α and ×6.12 at 1 µM and 81.7 nM EC 50 for SRC1 ( Figure 2A). GW501516 induced a maximal effect of ×13.4 at 1 µM with 8.48 nM EC 50 for PGC1α and ×3.10 at 1 µM with 5.82 nM EC 50 for SRC1 ( Figure 2B), whereas GW1929 induced a maximal effect of ×5.80 at 1 µM with 32.2 nM EC 50 for PGC1α and ×9.26 at 1 µM with 75.6 nM EC 50 for SRC1 ( Figure 2C). Next, we again compared the PGC1α/SRC1 recruitment activity of the three fibrates considering the maximal effects by the GW compounds as 100%. Bezafibrate and pemafibrate recruited the PGC1α peptide to all PPARα/δ/γ-LBD; however, fenofibric acid did not recruit PGC1α to PPARδ-LBD ( Figure 2D-F). The situation was the same for SRC1 recruitment in that the efficacy and efficiency varied greatly between agonists, PPAR subtypes, and coactivator species ( Figure 2G-I).

Fibrates Induce the Thermostability of PPARα/δ/γ-LBD
PPARs show increased thermostability upon ligand binding, which can be detected using circular dichroism (CD) spectroscopy [6]. Ligand-induced alterations in Tm values at 222 nm were investigated as types of reflection of α-helical stable structures of PPARs because PPAR ligand binding induces stabilization of the LBP [23,24]. The basal (solvent

Fibrates Induce the Thermostability of PPARα/δ/γ-LBD
PPARs show increased thermostability upon ligand binding, which can be detected using circular dichroism (CD) spectroscopy [6]. Ligand-induced alterations in T m values at 222 nm were investigated as types of reflection of α-helical stable structures of PPARs because PPAR ligand binding induces stabilization of the LBP [23,24]. The basal (solvent

Structures of the PPARα/δ/γ-LBD-Bezafibrate Complexes
We have recently revealed the structures of 34 PPARα-LBD complexed with 17 ligands, including the three fibrates [6] and the structure of PPARγ-LBD-saroglitazar (a PPARα/γ dual agonist in clinical trials for NAFLD treatment) [20]. To gain structural insight into the PPARα/δ/γ selectivity of the fibrates, we aimed to obtain the structures of PPARδ/γ-fibrate complexes by X-ray crystallography and compare them with PPARα complex structures that we obtained (Figure 4A-D; reprinted from Kamata et al. [6]). We first screened various cocrystallization buffer conditions based on previous literature and our experience with PPARα/γ [19] (see Materials and Methods). A PPARδ-LBD-bezafibrate structure was obtained without using any coactivators ( Figure 4E-H), whereas PPARγ-LBD-bezafibrate cocrystals were obtained using the SRC1 peptide (Figure 4I-L; SRC1 is indicated by an arrow in Figure 4I). The complex structures of bezafibrate bound to PPARδ/γ-LBD were solved in a monoclinic space group P21 at 2.09 Å resolution ( Figure  4E; deposited in the Protein Data Bank (PDB) with ID: 7WGL) and an orthorhombic space group P212121 at 2.36 Å resolution (PDB ID: 7WGO), respectively (Supplementary Table  S1). The electron density map for bezafibrate in all PPARα/δ/γ-LBD indicated the presence of a single molecule in the protein monomer ( Figure 4B,F,J).

Structures of the PPARα/δ/γ-LBD-Bezafibrate Complexes
We have recently revealed the structures of 34 PPARα-LBD complexed with 17 ligands, including the three fibrates [6] and the structure of PPARγ-LBD-saroglitazar (a PPARα/γ dual agonist in clinical trials for NAFLD treatment) [20]. To gain structural insight into the PPARα/δ/γ selectivity of the fibrates, we aimed to obtain the structures of PPARδ/γ-fibrate complexes by X-ray crystallography and compare them with PPARα complex structures that we obtained (Figure 4A-D; reprinted from Kamata et al. [6]). We first screened various cocrystallization buffer conditions based on previous literature and our experience with PPARα/γ [19] (see Materials and Methods). A PPARδ-LBD-bezafibrate structure was obtained without using any coactivators ( Figure 4E-H), whereas PPARγ-LBD-bezafibrate cocrystals were obtained using the SRC1 peptide (Figure 4I-L; SRC1 is indicated by an arrow in Figure 4I). The complex structures of bezafibrate bound to PPARδ/γ-LBD were solved in a monoclinic space group P2 1 at 2.09 Å resolution ( Figure 4E; deposited in the Protein Data Bank (PDB) with ID: 7WGL) and an orthorhombic space group P2 1 2 1 2 1 at 2.36 Å resolution (PDB ID: 7WGO), respectively (Supplementary Table S1). The electron density map for bezafibrate in all PPARα/δ/γ-LBD indicated the presence of a single molecule in the protein monomer ( Figure 4B,F,J).

Structures of the PPARα/γ-LBD-Fenofibric Acid Complexes
Cocrystals of PPARδ-LBD-fenofibric acid were not obtained probably because of its low binding affinity ( Figures 1F and 2E,H), although increased thermostability was observed at high concentrations ( Figure 3B). However, cocrystals with PPARγ-LBD were obtained in the presence of the SRC1 peptide. The complex structure was resolved in the orthorhombic space group P2 1 2 1 2 1 at 2.53 Å resolution (PDB ID: 7WGP) (Supplementary Table S1). The electron density map for fenofibric acid bound to PPARα-LBD indicated the presence of two molecules in the protein monomer ( Figure 5A-D; reprinted from Kamata et al. [19]); however, that of fenofibric acid bound to PPARγ-LBD indicated the existence of three molecules in the protein monomer ( Figure 5E-H). Its overall structure was basically identical to the previously reported active conformations that form the AF-2 helix 12 (arrowheads in Figure 5A,E). The two structures were similar, with an RMS distance of 0.57 Å (215 common Cα positions).
In PPARα-LBD, two molecules were located at the Center/Arm I and Arm II/X ( Figure 5A,B). In PPARγ-LBD, the first molecule was located beside the Center region, the second molecule at Arm II/III, and the third molecule at Arm II/X ( Figure 5F). Only the first molecule was stabilized by the four consensus amino acids (Ser289/His323/His449/Tyr473) via hydrogen bonds (red dotted lines) and electrostatic interactions (blue dotted lines) in PPARγ-LBD ( Figure 5G). Hydrogen bonds were also observed between PPARα K257 and the carbonyl group of the first molecule (water-mediated) and between PPARα T279 and the carbonyl group of the second molecule ( Figure 5D). No hydrogen bond or electrostatic interaction was observed between PPARγ and the first molecule, but hydrogen bonds were observed between PPARγ R288 and the oxygen atom of the second molecule and between PPARγ S342 and the carboxylic acid of the third molecule. Furthermore, two halogen bonds were observed between PPARγ E259/R280 and the chlorine atom of the third molecule ( Figure 5H). Such interactions combined with hydrophobic interactions (Supplementary Figure S1D,E) may stabilize the location of the second and third fenofibric acid molecules.
Pemafibrate was located at the almost identical Y-shaped structures comprising the Center and Arm II/III regions in all PPARs ( Figure 6B,F,J) although its phenoxyalkyl group seemed to be pushed toward helix 5 in PPARδ-LBD ( Figure 6E) and its 2-aminobenzoxazole group seemed to be pushed toward helix 3 in PPARγ-LBD ( Figure 6I). Several hydrogen bonds and electrostatic interactions with the four consensus amino acids were observed ( Figure 6C,G,K) with a very close proximity (2.0 Å) to PPARγ Y473 ( Figure 6K). Hydrogen bonds between PPARα T279 and water molecules were also observed in PPARα ( Figure 6D), whereas a water-mediated hydrogen bond or electrostatic interaction with 2-aminobenzoxazole group of pemafibrate was observed in PPARδ-LBD ( Figure 6H). Furthermore, no such bonds or interactions were observed in PPARγ-LBD ( Figure 6L

Various Binding Modes to PPARα/δ/γ-LBD Pockets
We have previously defined five LBPs (Center and Arm I-III/X) in PPARα-LBD [19] ( Figure 7A) and four LBPs (Center and Arm I-III) in PPARδ/γ-LBD [20]. Bezafibrate was located at the Center/Arm II regions of PPARα-LBD ( Figure 7B), where endogenous fatty acids and many other ligands bind [19]. However, one fenofibric acid molecule was located at the Arm I and another was located at the Arm X of PPARα-LBD ( Figure 7B); only the Arm I pocket is known to be occupied by relatively low affinity fibrates, such as ciprofibrate, clofibric acid, and gemfibrozil [19]. Pemafibrate was located at the Y-shape structures comprising the Center and Arm II/III ( Figure 7B), similar to GW7647, the potent PPARα-selective agonist [19].
acid residues (that recognize the carboxyl moiety of pemafibrate) are indicated by red and blue dotted lines, respectively, with their distances (Å ). (D,H,L) Hydrogen bonds and electrostatic interactions between pemafibrate (in van der Waals spheres) and all surrounding amino acid residues located within a distance of 5 Å .

Various Binding Modes to PPARα/δ/γ-LBD Pockets
We have previously defined five LBPs (Center and Arm I-III/X) in PPARα-LBD [19] ( Figure 7A) and four LBPs (Center and Arm I-III) in PPARδ/γ-LBD [20]. Bezafibrate was located at the Center/Arm II regions of PPARα-LBD ( Figure 7B), where endogenous fatty acids and many other ligands bind [19]. However, one fenofibric acid molecule was located at the Arm I and another was located at the Arm X of PPARα-LBD ( Figure 7B); only the Arm I pocket is known to be occupied by relatively low affinity fibrates, such as ciprofibrate, clofibric acid, and gemfibrozil [19]. Pemafibrate was located at the Y-shape structures comprising the Center and Arm II/III ( Figure 7B), similar to GW7647, the potent PPARα-selective agonist [19].
In PPARδ-LBD, bezafibrate was located at the Center/Arm II regions ( Figure 7C), where the potent PPARδ-selective agonists such as GW501516 and GW0742 bind [26,27], and only pemafibrate was located at the Arm III region ( Figure 7C) as in PPARα-LBD ( Figure 7B). In PPARγ-LBD, a single bezafibrate was located at the Center/Arm III regions; three fenofibric acids were arranged in a parallel fashion within the Center and Arm II/III/X; and a single pemafibrate was in the Y-shape structures comprising the Center and Arm II/III ( Figure 7D). Notably, bezafibrate and fenofibric acid (the second molecule) were located in the Arm III regions only in PPARγ-LBD ( Figure 7D). In PPARδ-LBD, bezafibrate was located at the Center/Arm II regions ( Figure 7C), where the potent PPARδ-selective agonists such as GW501516 and GW0742 bind [26,27], and only pemafibrate was located at the Arm III region ( Figure 7C) as in PPARα-LBD ( Figure 7B). In PPARγ-LBD, a single bezafibrate was located at the Center/Arm III regions; three fenofibric acids were arranged in a parallel fashion within the Center and Arm II/III/X; and a single pemafibrate was in the Y-shape structures comprising the Center and Arm II/III ( Figure 7D). Notably, bezafibrate and fenofibric acid (the second molecule) were located in the Arm III regions only in PPARγ-LBD ( Figure 7D).

Discussion
Most fibrates were developed in the 1960s-1980s before their molecular target, PPARα, was identified. Therefore, neither the information regarding PPAR agonistic activity (including subtype selectivity) nor the results of molecular docking studies on the threedimensional structures of PPARs were employed for designing and developing the fibrates. Classical fibrates, such as bezafibrate and fenofibrate, are known to have relatively low PPAR activity and selectivity [28]. Bezafibrate was launched in 1978 by Boehringer Mannheim in Germany (current F. Hoffmann-La Roche AG [Roche], Switzerland) and is currently approved in many countries but not the United States. It is considered to be a "balanced" pan agonist that activates all PPARα/δ/γ at comparable doses and improves dyslipidemia via its actions against PPARα; insulin sensitivity via its actions against PPARγ, and overweightness by enhancing fatty acid oxidation, energy consumption, and adaptive thermogenesis via its actions against PPARδ [29]. Bezafibrate activated all PPARα/δ/γ in the cell-based transactivation assay ( Figure 1E), cell-free (both PGC1α and SRC1) coactivator recruitment assay ( Figure 2D,G), and cell-free thermostability assay ( Figure 3A) by binding to the similar binding pockets (Center/Arm II) of PPARα/δ ( Figure 4B,F) or the other binding pocket (Center/Arm III) of PPARγ ( Figure 4J). Interestingly, depending on the assays, bezafibrate induced responses with altered affinity against PPARα/δ/γ: the orders of EC 50 values were α (30.4 µM) < δ (86.7 µM) < γ (178 µM) in transactivation ( Figure 1E; equivalent to the values [α (25 µM); δ (95 µM); γ (>100 µM)] in previous transactivation experiments [30]), δ < γ < α in PGC1α recruitment ( Figure 2D), α < δ < γ in SRC1 recruitment ( Figure 2G), and α < δ = γ in thermostability assay ( Figure 3A). In addition, bezafibrate exhibited altered efficacy: the orders of the maximal responses were α > γ > δ in transactivation ( Figure 1E), δ > α > γ in PGC1α recruitment ( Figure 2D), α > δ > γ in SRC1 recruitment ( Figure 2G), and α > δ = γ in the thermostability assay ( Figure 3A). In the phase 3 study of "Bezafibrate in Combination with Ursodeoxycholic Acid in Primary Biliary Cirrhosis (BEZURSO; NCT01654731)" completed in December 2016, its dual actions on PPARα/δ were considered important for the improvement of biochemical measures, such as liver stiffness [31]. However, the action of bezafibrate on PPARγ could also be indispensable for the effect. Another phase 2 study investigating the effect of bezafibrate on bipolar depression (NCT02481245) depended on the concept that mitochondrial dysregulation is attributable to bipolar depression, which could be ameliorated by PPAR pan agonists, such as bezafibrate.
Fenofibrate was introduced into clinical practice in 1974 and launched in France in 1975 [32]. More than 200 clinical trials on fenofibrate have been completed and are ongoing, and it has been approved in the United States and many other countries. Unlike bezafibrate, fenofibric acid has been recognized as a PPARα-selective agonist in clinical settings, although it was shown to activate PPARγ in several in vitro experiments [30,33]. Fenofibric acid induced PPARα/γ-LBD-mediated transactivation ( Figure 1F), PGC1α/SRC1 recruitment ( Figure 2E,H), increases in thermostability ( Figure 3B), and bound to PPARα/γ-LBD ( Figure 5). Interestingly, fenofibric acid did not induce transactivation ( Figure 1F) and PGC1α/SRC1 recruitment ( Figure 2E,H) but increased thermostability ( Figure 3B) of PPARδ-LBD, and its cocrystals with PPARδ-LBD were not obtained. As observed in the thermostability assay ( Figure 3B), fenofibric acid might bind to the non-LBP regions of PPARδ-LBD and stabilize the PPARδ-LBD but fail to functionally activate it. Unlike bezafi-brate, fenofibric acid displayed a consistent order (α > γ >> δ = no response) of affinity and efficacy in transactivation ( Figure 1F) and PGC1α/SRC1 recruitment ( Figure 2E,H), except for thermostability ( Figure 3B). To our knowledge, no publication has reported EC 50 values of fenofibric acid or fenofibrate in PPARδ(-LBD)-mediated reactions; therefore, fenofibric acid is considered a genuine PPARα/γ dual agonist. Pemafibrate (K-877) was approved in 2018 in Japan as a highly selective PPARα agonist that is safe for simultaneous use with statins even in patients with mild adverse effects [34]. Pemafibrate activated PPARα/δ/γ-LBD-mediated transactivation ( Figure 1G), PGC1α/SRC1 recruitment ( Figure 2F,I), increases in thermostability ( Figure 3C), and bound to PPARα/δ/γ-LBD with the similar Y-shaped forms (Figure 6), although it acted at much lower concentrations on PPARα-LBD than on PPARδ/γ-LBD.
This study demonstrated that bezafibrate and pemafibrate could bind to and activate all PPAR subtypes, whereas fenofibric acid could bind to and activate only PPARα/γ. Whether this may happen in clinical settings is an issue of importance and needs to be investigated. Therapeutic doses in Japan are 800 mg, 106.6-160 mg, and 0.2 mg per day for bezafibrate, fenofibrate, and pemafibrate, respectively. The maximal plasma concentration (C max ) after the administration of a single dose (200 mg) of bezafibrate was 20.4 µM (7.39 µg/mL) [35], which is roughly equivalent to its EC 50 values in transactivation ( Figure 1E) and PGC1α/SRC1 recruitment ( Figure 2D,G); thus, bezafibrate can act as a PPAR pan activator in its clinical doses. According to the package insert of TRICOR (Takeda, Tokyo, Japan), the C max after the administration of a single one-day dose (160 mg) of fenofibrate was 37.0 µM (11.8 µg/mL). Fenofibric acid (30 µM) activated PPARα and slightly activated PPARγ but did not activate PPARδ ( Figures 1F and 2E,H). Therefore, fenofibric acid is considered a relatively weak PPARα agonist. According to the package insert of PARMODIA (Kowa, Nagoya, Japan), the C max after the administration of a single one-day maximal dose (400 µg in Japan) of pemafibrate was 7.28 nM (3.57 ng/mL) after seven-day repeats. Pemafibrate (7 nM) induced transactivation only in PPARα ( Figure 1G); and the coactivator recruitment assay hardly detected signals produced by <10 nM ligands owing to too many non-responding PPAR-LBDs within a total of 100 or 200 nM proteins. Therefore, pemafibrate is considered a selective agonist of PPARα at its clinical doses.
The global prevalence of NAFLD is estimated to be 25% of the global population, which is consistent with the substantial increases in the number of patients with diabetes and metabolic syndrome. PPAR dual/pan agonists are expected to be some of the most promising therapeutic drugs for NAFLD [36]. Although saroglitazar (α/γ dual agonist; Zydus Discovery, Dubai, UAE) [37] and lanifibranor (α/δ/γ pan agonist; Iventiva Pharma, Daix, France) [17] remain under clinical trials investigating their use against NAFLD/NASH, the development of most of the dual PPARα/γ agonists (mostly against type 2 diabetes)-muraglitazar (Bristol-Myers Squibb/Merck), tesaglitazar (AstraZeneca), aleglitazar (Roche), MK0767 (Kyorin/Banyu/Merck), naveglitazar (Ligand Pharmaceuticals, San Diego, CA, USA), ONO-5219 (Ono Pharma, Osaka, Japan), and DSP-8658 (Sumitomo Dainippon Pharma, Osaka, Japan)-has been discontinued due to PPARγ-related side effects (e.g., heart failure) or no effects [11,38]. Therefore, repositioning of bezafibrate [39] and fenofibrate, which have proven safety, as anti-NAFLD/NASH drugs might be a favorable option. Although three clinical trials investigating the use of fenofibrate against NAFLD/NASH (NCT02781584, NCT00262964, NCT02891408) and a single trial investigating pemafibrate (NCT03350165) have failed, there have been no clinical trials on the use of bezafibrate against NAFLD/NASH. However, bezafibrate has been shown to exert beneficial effects on NASH in tamoxifen-or its analog toremifene-treated breast cancer patients [40,41]. Pemafibrate improved the FibroScan-aspartate aminotransferase (FAST) scores (a novel index of NASH conditions) [42] and the similar scores [43] in NAFLD/NASH patients in some retrospective studies. The differential (in terms of efficiency and efficacy) recruitment of coactivators (PGC1α and SRC1) by bezafibrate ( Figure 2D,G) may induce the expression of the specific sets of genes different from those induced by fenofibric acid or pemafibrate [44] in organ-specific manners [45].
In conclusion, this study has highlighted the PPAR dual/pan agonistic aspects of the three approved fibrates-bezafibrate, fenofibric acid, and pemafibrate-using functional and structural analyses. PPAR dual/pan agonists could be a radical remedy for NAFLD/NASH, and our findings contribute toward the fine-tuning of PPAR subtype selectivity.

Recombinant PPARα/δ/γ-LBD Expression and Purification
Human PPARα-LBD (amino acids 200-468), PPARδ-LBD (amino acids 170-441), and PPARγ-LBD (amino acids 203-477 in isoform 1) were expressed as amino-terminal Histagged proteins by the pET28a vector (Merck KGaA (Novagen), Darmstadt, Germany) in Rosetta (DE3) pLysS competent cells (Novagen) and purified using three-step chromatography as described in our PPARα-LBD preparation [6,19]. Transformed cells were cultured at 30 • C in an LB medium (with 15 µg/mL kanamycin and 34 µg/mL chloramphenicol), and 50 mL of overnight culture was seeded in 1 L of a TB medium (with 15 µg/mL kanamycin), which was cultured at 30 • C for 1.5 h and then at 15 • C for 2 h. Protein overexpression was induced by adding 0.5 mM isopropyl β-D-thiogalactopyranoside, which were later cultured at 15 • C for 48 h. The cells were harvested and resuspended in 40 mL of buffer A (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM Tris 2-carboxyethylphosphine [TCEP]-HCl, and 10% glycerol) for PPARα/γ or buffer A' (20 mM Tris-HCl [pH 8.0], 500 mM ammonium acetate, 1 mM TCEP-HCl, and 10% glycerol) for PPARδ; both buffers contained a complete EDTA-free protease inhibitor (Sigma-Aldrich). The cells were then lysed by sonication five times, for 2 min each time, using a UD-201 sonicator (Tomy, Tokyo, Japan) at an output of eight; they were clarified by centrifugation at 12,000× g at 4 • C for 20 min (these conditions were used throughout the study unless otherwise noted); then, polyethyleneimine, at a final concentration of 0.15% (v/v), was added to the supernatant to remove nucleic acids. After centrifugation, 35 mL of the supernatant was mixed with 20 g of ammonium sulfate at 4 • C for 30 min using gentle rotation. After centrifugation, the pellet was resuspended in