Selective activation of PPARα maintains thermogenic capacity of beige adipocytes

Summary Beige adipocytes are inducible thermogenic adipocytes used for anti-obesity treatment. Beige adipocytes rapidly lose their thermogenic capacity once external cues are removed. However, long-term administration of stimulants, such as PPARγ and β-adrenergic receptor agonists, is unsuitable due to various side effects. Here, we reported that PPARα pharmacological activation was the preferred target for maintaining induced beige adipocytes. Pemafibrate used in clinical practice for dyslipidemia was developed as a selective PPARα modulator (SPPARMα). Pemafibrate administration regulated the thermogenic capacity of induced beige adipocytes, repressed body weight gain, and ameliorated impaired glucose tolerance in diet-induced obese mouse models. The transcriptome analysis revealed that the E-twenty-six transcription factor ELK1 acted as a cofactor of PPARα. ELK1 was mobilized to the Ucp1 transcription regulatory region with PPARα and modulated its expression by pemafibrate. These results suggest that selective activation of PPARα by pemafibrate is advantageous to maintain the function of beige adipocytes.


INTRODUCTION
Thermogenic adipocytes, including brown and beige, have emerged as targets for treating metabolic disorders. These thermogenic adipocytes have abundant mitochondria and a unique protein, uncoupling protein-1 (UCP1), which produces heat by uncoupling the electron transport chain from ATP synthesis. 1,2 In particular, beige (or brite) adipocytes can be induced in white adipose tissue (WAT) by various external cues such as chronic cold exposure or long-term administration of peroxisome proliferator-activated receptor g (PPARg) agonists in rodent models. In humans, 2-deoxy-2-( 18 F)fluoro-D-glucose ([ 18 F]FDG) PET scanning has revealed that metabolically active brown adipose tissue (BAT) exists even in adults [3][4][5][6][7] and possesses murine beige-like molecular characteristics in the supraclavicular and neck regions. [8][9][10] Therefore, understanding the mechanism of beige adipocyte development might be linked to its practical application in human BAT activation.
The development of brown and beige adipocytes is regulated by various transcriptional regulatory factors, such as PR (PRD1-BF1-RIZ1 homologous) domain-containing 16 (PRDM16), 11,12 early B cell factor 2 (EBF2), 13 or nuclear factor I-A (NF1A). 14 In addition, euchromatic histone lysine methyltransferase 1 (EHMT1) regulates the development of brown and beige adipocytes by interacting with the PRDM16 transcriptional complex and epigenetically modulating the reduced expression of myogenic and white adipogenic genes. [15][16][17] Although the thermogenic capacity of induced beige adipocytes is almost the same as that of brown adipocytes, beige adipocytes lose their thermogenic function and de-differentiate into white adipocytes rapidly once developmental cues have been withdrawn. 18 b3-Adrenocepttor agonists or certain PPARg agonists, such as rosiglitazone, can activate thermogenic programs. While these agonists strongly induce beige adipogenesis in rodent models, their long-term use in humans leads to adverse side effects, such as elevation of blood pressure or tachycardia by b3-adrenoceptor agonists and edema or heart failure by rosiglitazone. 19,20 Therefore, it is necessary to identify valuable targets for beige biogenesis and develop drugs that can be used for a long time without unfavorable side effects. iScience Article the expression of thermogenic genes, such as Ucp1, Cidea, and Elovl3, and UCP1 protein decreased significantly (Figures 2A and 2B). Next, we investigated whether the pharmacological activation of PPARa enables the maintenance of the thermogenic function of beige adipocytes using PPARa agonists. First, beige adipocytes were treated with rosiglitazone for 5 days. Accordingly, rosiglitazone was withdrawn (Rosi-off) or switched to a PPARa agonist, such as bezafibrate (Rosi-Beza), GW9578 (Rosi-GW9578), or pemafibrate (Rosi-Pema) ( Figure S2B). All PPARa agonists maintained Ucp1 gene expression in beige adipocytes. However, pemafibrate treatment exhibited a considerably higher increase in Ucp1 ratio compared to bezafibrate and GW9578 treatments ( Figure S2C). Based on these results, we focused on the ability of pemafibrate to maintain thermogenic genes.
The expression of thermogenic genes significantly increased along with UCP1 protein expression in the Rosi-Pema group ( Figures 2C and S2D). The oxygen consumption rate was also significantly increased in the Rosi-Pema group ( Figure 2D). From these results, pemafibrate is suggested to enable the maintenance of the thermogenic capacity of beige adipocytes.
To examine the effect of pemafibrate on beige adipocyte induction, preadipocytes were treated with a vehicle (control), pemafibrate (Pema), and rosiglitazone (Rosi) in addition to a regular adipogenic cocktail ( Figure S2E, left panel). We also overexpressed PPARa exogenously (PPARa) in preadipocytes at a significantly higher level than in the control, Pema, and Rosi groups ( Figure S2E, right panel). Pemafibrate treatment increased the expression of thermogenic genes, including Ucp1, Cidea, Cox8b, and Elovl3, as well as the UCP1 protein, in differentiated adipocytes. Notably, the expression levels of these genes and protein were marked higher in the Rosi group than in Pema group. In the PPARa group, thermogenic genes and UCP1 protein were slightly increased compared to the control group, although they were not as high as those in the Pema and Rosi groups ( Figures 2E and 2F).
These results suggest that the activation of thermogenic genes by pemafibrate treatment is not considerably greater than that by rosiglitazone treatment and that the induction of thermogenic genes is dependent on the activation of PPARa by its ligand, rather than the mere presence of the PPARa protein.
Pemafibrate treatment suppressed glucose intolerance and weight gain in DIO mice while maintaining beige adipocytes Next, we tried to elucidate whether pemafibrate could maintain the induced beige adipocytes in a mouse model. First, we investigated the thermogenic induction ability of pemafibrate (Pema) in inguinal adipose tissue compared to a vehicle (Vehicle), rosiglitazone (Rosi), or CL316,243 (CL) ( Figure S3A). We found that the administration of rosiglitazone or CL316,243 for 10 days significantly induced thermogenic genes, including Ucp1, Cidea, or Pgc1a, but that pemafibrate treatment did not ( Figure S3B). Hence, we aimed to examine the effect of pemafibrate after the induction of beige adipocytes by using CL316,243. After breeding mice at 30 C for a week, CL316,243 (1 mg kg À1 ) was administered for 1 week to induce beige adipocytes in the inguinal WAT. For the following 4 weeks, a vehicle (CL-Off) or pemafibrate (1 mg kg À1 ) (CL-Pema) was injected intraperitoneally. For control, a vehicle was administered intraperitoneally during all study periods (Vehicle). All mice were fed a high-fat diet (60% HFD) during the 4 weeks ( Figure 3A). There was no significant difference in dietary intake between the groups ( Figure S3C). HFD feeding induced body weight and glucose intolerance in the control and off groups; however, pemafibrate administration significantly repressed body weight gain and showed lower glucose levels, as estimated by the intraperitoneal glucose tolerance test ( Figures 3B and 3C). To confirm whether beige adipocytes were maintained in subcutaneous WAT, inguinal WAT was collected. Histologically, adipocytes in the CL-Pema group had smaller lipid droplets and were UCP1-positive, as assessed by hematoxylin and eosin and immunohistochemical staining ( Figure 3D). Thermogenic gene expression and protein levels of UCP1 and respiratory chain components were higher in the CL-Pema group than in the vehicle and off groups ( Figures 3E and 3F). We also iScience Article confirmed that oxygen consumption rates were increased in ex vivo inguinal adipose tissue in the CL-Pema group ( Figure 3G). In the CL-Pema group, the expression of Ucp1 in BAT increased slightly, and the expression of thermogenic genes in epididymal WAT increased significantly; however, UCP1 protein levels did not differ between BAT and epididymal WAT ( Figures S3D and S3E). A previous study reported that pemafibrate administration in mice increased FGF21 production, leading to improvements in terms of body weight gain and glucose intolerance. 35 Both liver Fgf21 expression levels and plasma FGF21 concentrations demonstrated a significant increase following pemafibrate treatment ( Figures 3H and 3I). These results suggest that induced beige adipocytes are sufficiently maintained with pemafibrate administration, which may contribute to the repression of glucose intolerance and inhibition of body weight gain in the DIO mouse model.

ELK1 is an important cofactor of PPARa for pemafibrate-regulated thermogenic gene regulation
PPARa agonists exert their transcription regulatory action via transcription cofactors, and each PPARa agonist has a unique agonist-specific type of cofactor mobilization. 36 To elucidate the mechanism by which SPPARMa induces thermogenic programs, we investigated the binding state of transcription regulators on thermogenic genes induced by various PPAR agonists ( Figure S4A). Rosiglitazone, pemafibrate, and GW9578, another selective PPARa agonist, significantly induced beige-enriched thermogenic and mitochondrial genes and repressed white adipocyte-specific markers, such as Retn, Agt, or Lep ( Figures S4B  and S4C). Motif analysis revealed the presence of ETS-related genes, such as ELK1, in the promoter regions of the Pema group ( Figure 4A). Interestingly, the transcriptional factors present in the promoter region of the GW9578 group differed remarkably from those in the Pema group, and a multidimensional scaling (MDS) analysis revealed that the position of pemafibrate-regulated genes was distant from that of GW9578-regulated genes ( Figures S4D and S4E). Motif analysis of rosiglitazone-regulated genes also revealed the presence of ETS-related genes in the promoter regions ( Figure S4F). Because the effects of ELK1 on thermogenic adipocytes have not been reported, we examined the contribution of ELK1 to the pemafibrate-induced thermogenic gene programs. Gene expression of Elk1 was also increased in drug-induced beige adipocytes in vitro and in vivo ( Figures S3G and S3H). The gene expression of ELK1 was also increased in the perirenal tissue of human pheochromocytoma and correlated with PPARA expression (Figures 4B and S4I). ELK1 reportedly forms a transcriptional complex together with MED23 and controls white fat differentiation; MED23 has been identified as a member of the PPARa transcription complex. 37,38 In this study, we confirmed that PPARa and ELK1 form transcriptional complexes together with MED23 in the nucleus of beige adipocytes ( Figure S4J). To determine whether ELK1 is required for the expression of thermogenic genes, we knocked down ELK1 using lentiviral shRNA in preadipocytes and differentiated them with pemafibrate treatment ( Figure 4C). Oil Red O staining showed that adipogenesis was not altered with ELK1 depletion ( Figure 4D); however, the expression of thermogenic genes and UCP1 protein were significantly reduced in ELK1-deficient cells ( Figures 4E-4F), indicating that ELK1 is required for thermogenic gene regulation by pemafibrate treatment. In contrast, ELK1 knockdown did not affect either the induction of thermogenic genes by PPARg agonists or the activation of PKA by forskolin, indicating that ELK1 is specifically required for pemafibrate-caused effects on thermogenic gene induction ( Figure S4K). When examining the effects of pemafibrate on ELK1 expression in differentiated adipocytes, pemafibrate slightly affected the ELK1 gene expression or protein levels ( Figures S4L and S4M). Finally, to clarify the transcriptional regulation of Ucp1, a representative thermogenic gene, by PPARa and ELK1 with pemafibrate, we performed ChIP-qPCR analyses with antibodies against ELK1, PPARa, and H3K27ac, an active histone marker. Interestingly, ELK1 was enriched in the Ucp1 transcription regulatory region with PPARa and H3K27Ac, and this enrichment was robustly increased by pemafibrate treatment, thereby suggesting that ELK1 and PPARa are co-localized in the regulatory region of Ucp1. PPARa agonism by pemafibrate treatment increases this co-localization and regulates Ucp1 expression ( Figure 4G). These results suggest that ELK1 is an important factor in the regulation of thermogenic genes transcriptionally controlled by pemafibrate treatment.  iScience Article PPARa is not essential for thermogenesis in BAT, 23,40 and the induction of beige adipocytes under cold stimuli. 41 The action of PPARa differs in the developmental process of thermogenic adipocytes, depending on various stimuli (environmental factors, agonists, etc.).
The activation of PPARa by synthetic PPARa agonists results in a conformational change in the PPARa complex. This complex binds to the target gene locus with cofactors to regulate transcriptional activity. 29 In the regulation of thermogenic genes, administration of WY-14643, a PPARa agonist, activates the transcriptional function of PPARa in the presence of PRDM16, 11 and GW7647, another PPARa agonist, recruits the PPARa-PRDM16 complex to the PGC1a modulating locus. 42 This suggests that the activation of PPARa by ligands is deeply involved in the regulation of thermogenic gene expression. In humans, pemafibrate has been suggested to be effective in the treatment of dyslipidemia and in reducing adverse events in liver and kidney function compared to fenofibrate. 43,44 By using an FGF21 KO mouse model to examine the effects of pemafibrate on adipocytes, it was evident that administering pemafibrate to a DIO mouse model increases thermogenic gene expression in inguinal WAT via increased production of Fgf21 in the liver. 35 FGF21 acts on paracrine/autocrine signaling and increases the expression of thermogenic genes in inguinal WAT. 45 The improvement in weight gain and glucose intolerance may be due to the synergistic effect on metabolic organs, such as adipose tissue and the liver.
However, there are currently no reports examining the direct effects of pemafibrate on adipocytes in detail. This study revealed that pemafibrate treatment induced beige adipogenesis in a ligand-dependent manner. In addition, pemafibrate and GW9578 differed in the recruitment of transcription factors to the transcriptional regulatory regions based on motif analysis of RNA-seq. Furthermore, as one of the transcriptional cofactors with PPARa recruited by pemafibrate, ELK1 was identified. ELK1 is a ternary complex factor belonging to the ETS family of transcription factors. The ETS gene is conserved in various species and regulates cell differentiation, development, and proliferation 46-48 ELK1 reportedly promotes the production of adipocytes via insulin-MAPK signaling in combination with MED23 (Sur2, DRIP130). 37 In the nucleus of hepatocytes, MED23 binds to PPARa via the LXXLL motif. 38 However, ELK1 has not been previously reported as a member of the PPARa transcription complex. The present study revealed that ELK1 forms a transcriptional complex with PPARa via MED23 in the nucleus of beige adipocytes and that pemafibrate treatment increases ELK1 enrichment in the transcriptional regulatory region of Ucp1. ELK1 may be a specific cofactor mobilized by pemafibrate administration, and the co-localization of PPARa and ELK1 may regulate the expression of Ucp1.
Clinically, pemafibrate can be safely used without rhabdomyolysis at various chronic kidney disease (CKD) stages. Several clinical studies have revealed that pemafibrate administration is not related to increased frequencies of side effects, such as tachycardia, edema, or body weight gain, which are commonly observed with b3-adrenergic agonists and PPARg agonists. 49,50 Furthermore, this study revealed that the pemafibrate treatment enhances thermogenic gene expression in human brown adipocytes. PPARa selective activation by pemafibrate might hold therapeutic potential for the treatment of obesity-related disorders.
In conclusion, we identified PPARa as a key factor for maintaining the thermogenic capacity of beige adipocytes. Further, we revealed that the transcription factor ELK1 co-localizes with PPARa and is involved in the transcriptional regulation of thermogenic genes. Pemafibrate is widely used in clinical practice, and our results shed light on its long-term usefulness in treating obesity-related disorders while avoiding side effects.

Limitations of the study
The findings of this study suggest that pemafibrate is useful for maintaining mature beige adipocytes.
However, the precise mechanism through which PPARa and ELK1 regulate the expression of thermogenic genes in response to pemafibrate remains unclear. Investigating the roles of PPARa and ELK1 in the mechanism of maintaining thermogenic capacity by pemafibrate was challenging in this study, particularly because of the technical difficulties in creating a loss-of-function model in mature beige adipocytes. In the future, studies using inducible ELK1 adipose tissue-specific knockout mice will demonstrate how ELK1 contributes to the mechanism by which pemafibrate induces the expression of thermogenic genes.

OPEN ACCESS
iScience 26, 107143, July 21, 2023 9 iScience Article STAR+METHODS Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
We declare that none of the authors have any financial interests related to this work.  iScience Article iScience Article were filtered through a 70 mm cell strainer and centrifuged at 700 3 g to collect SVF. SV cell pellets were rinsed and plated on collagen-coated plates. Preadipocytes were immortalized by infection with the pMSCV-puro retroviral vector encoding the SV40T antigen and selection with puromycin (Invivogen# ant-pr-1, 2 mg/mL).
Immortalized mouse adipocyte cultures were prepared according to previously reported methods. 16 Briefly, cells were cultured to confluence, and adipocyte differentiation was induced using a Dulbecco's Modified Eagle Medium (DMEM)/F-12 medium (with phenol red, Wako Pure Chemicals, Osaka, Japan) containing 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, 5 mg/mL insulin, 1 nM T3, 0.25 mM IBMX, 0.125 mM indomethacin, and 2 mg/mL dexamethasone. Two days after induction, the cells were switched to a maintenance medium containing 10% FBS, 5 mg/mL insulin, and 1 nM T3 for an additional 5-6 days. To induce beige adipocyte differentiation, cells were differentiated in the presence of 1 mM rosiglitazone, as outlined in a previous study. 30 The effects of PPARa transcriptional activity inhibition were determined using GW6471. The differentiated beige adipocytes were treated with 5 mM GW6471 to test its effect on thermogenic gene levels. The in vitro dosage of pemafibrate used in this study was based on a previous study. 55 Pemafibrate (10 mM) was used to differentiate adipocytes, which was switched to rosiglitazone on days 5-6. GW9578 (1 mM) and bezafibrate (500 mM) were used to test its effect on thermogenic gene levels, which was switched to rosiglitazone on day 5. Mycoplasma infections were not tested in any of these experiments.

Forward siRNA transfections
Using Dharmacon SmartPool On-Target PLUS RNA pools, immortalized inguinal preadipocytes were forward-transfected with siRNA as per the manufacturer's protocol when the cells reached 80% confluence on collagen I-coated 12-well cell culture plates (4815-010; IWAKI) ( Table S1). The siRNA and Lipofectamine RNAiMAX (13378-150; Invitrogen, Carlsbad, CA, USA) were diluted separately in Opti-MEMâ I Reduced Serum Medium (31985-062; Life Technologies, Carlsbad, CA, USA) and mixed by pipetting. The siRNA-RNAiMAX mixture was incubated for 20 min at room temperature and was then added onto the adherent cells, and the cells were gently shaken. The final concentrations of Lipofectamine RNAiMAX and siRNA were 3 mL/mL and 20 nM, respectively. The preadipocytes were then harvested to reach confluence 1 day after transfection. The cells were then subjected to adipogenic conditions to promote mature adipocyte differentiation, with the addition of pemafibrate (10 mM) for 4 days. The cells were forward-transfected 3 days after the first transfection.

DNA constructs and virus production
The constructs were subcloned into a pMEI5-neo retroviral vector (Genscript, Piscataway, NJ, USA). For retrovirus production, Phoenix packaging cells were transfected at 70% confluence using the ll OPEN ACCESS