Protein arginine methyltransferase (PRMT) 1 is involved in the regulation of various metabolic pathways such as glucose metabolism in liver and atrophy in the skeletal muscle. However, the role of PRMT1 in the fat tissues under the disease state has not been elucidated to date. In this study, we delineate the function of this protein in adipocytes in vivo. PRMT1 expression was abundant in the white adipose tissues (WAT), which was induced upon a high-fat diet in mice and by obesity in humans. We found that adipocyte-specific depletion of Prmt1 resulted in decreased fat mass without overall changes in body weight in mice. Mechanistically, the depletion of Prmt1 in WAT led to the activation of the AMPK pathway, which was causal to the increased lipophagy, mitochondrial lipid catabolism, and the resultant reduction in lipid droplet size in WAT in vivo. Interestingly, despite the increased energy expenditure, we observed a promotion of adipose tissue inflammation and an ectopic accumulation of triglycerides in the peripheral tissues in Prmt1 adipocyte-specific knockout mice, which promoted the impaired insulin tolerance that is reminiscent of mouse models of lipodystrophy. These data collectively suggest that PRMT1 prevents WAT from excessive degradation of triglycerides by limiting AMPK-mediated lipid catabolism to control whole-body metabolic homeostasis in diet-induced obesity conditions.
Introduction
White adipose tissue (WAT) functions as a reservoir for excess nutrients, such as triglycerides (TG) (1). Upon nutrient deprivation, TG is mobilized via lipolysis, resulting in the production of free fatty acids that can be used for peripheral tissues. Under pathological conditions of diet-induced obesity (DIO), regulation of TG homeostasis in WAT is altered, resulting in the increased adipocyte size in WAT and the subsequent infiltration of M1 macrophages (2–4). This remodeling process is not only detrimental to whole body lipid homeostasis but is also critical in instigating insulin resistance via the increased secretion of proinflammatory cytokines. On the other hand, defects in TG storage in WAT, either due to the dysregulation of TG synthesis or accelerated lipid breakdown, could result in lipodystrophy. This latter process could lead to the ectopic lipid accumulation in the peripheral tissues, also stimulating systemic insulin resistance.
Protein arginine methyltransferases (PRMTs) comprise a family of proteins that promote arginine methylation of various cellular proteins (5,6). To date, 11 family members of PRMTs have been identified, which can be divided into three subgroups. Type 1 PRMTs include PRMT1, 3, 4, 6, and 8, which generate asymmetric dimethyl arginine, while type 2 PRMTs, containing PRMT5, 7, and 9, produce symmetric dimethyl arginine. The function of the remaining PRMTs has yet to be determined. PRMT1 is a predominant PRMT in mammals, representing >85% of all PRMT activity. Substrates of PRMT1 include histones H3 and H4, transcription factors, and other cellular signaling proteins (7–10). Recently, PRMT1 was shown to be associated with metabolic processes such as hepatic gluconeogenesis, lipid homeostasis, and skeletal muscle atrophy (11–14). However, the direct role of PRMT1 in adipose tissues under the disease state has not been fully elucidated to date.
In this study, we attempted to delineate the role of PRMT1 in WAT functions under the DIO state by using adipocyte-specific Prmt1 functional knockout (FKO) mice. We found that PRMT1 is essential in maintaining the normal function of WAT in DIO conditions by limiting excessive lipid catabolism, preventing the resultant lipid accumulation in the peripheral tissues, and the dysregulation of metabolic homeostasis.
Research Design and Methods
Animal Studies
To generate Prmt1 FKO mice, Prmt1 f/f mice (European Conditional Mouse Mutagenesis Program [EUCOMM]) were bred with adiponectin-Cre mice and were backcrossed at least 10 times to maintain the C57BL/6N background. Male mice were fed a normal chow diet (NCD) or high-fat diet (HFD) (60% kcal from fat) and used for the study in a specific pathogen-free facility at the Central Laboratory Animal Research Center, Korea University.
For insulin tolerance test, 6-h fasted mice were intraperitoneally injected with insulin (0.5 units/kg; Gibco). Blood glucose was measured using a OneTouch blood glucose monitor (LifeScan) from the tail blood. Lipid extraction was performed using the Folch method with a slight modification (15,16). TG levels were determined using L-Type Triglyceride M kit (Wako). Plasma nonesterified fatty acid (NEFA) levels were assessed using the NEFA-HR(2) kit (Wako). Plasma tumor necrosis factor-α (TNF-α) and interleukin 6 (IL-6) levels were measured using Mouse TNFα ELISA and Mouse IL-6 ELISA kits, respectively (Invitrogen). Plasma insulin levels were determined using the Mouse Insulin ELISA Kit (ALPCO). For measuring insulin signaling, 6-h fasted mice were intraperitoneally injected with PBS or insulin (0.2 units/mouse) 10 min before being sacrificed. For in vivo autophagic flux assay, 20-h fasted mice were injected intraperitoneally with leupeptin (40 mg/kg; Sigma-Aldrich) or PBS 4 h before being sacrificed (17,18).
Quantitative PCR
Mouse tissues and adipocytes were lysed in QIAzol Lysis Reagent (Qiagen), and total RNA was extracted using RNeasy Plus Mini kit (Qiagen). For human adipose tissues, total RNA was extracted from frozen samples using TRIzol (Ambion).
For RT-quantitative (q)PCR, cDNA was generated using GoScript Reverse Transcription System (Promega), followed by qPCR using the SensiFAST SYBR No-ROX Kit (Bioline). mRNA levels were normalized to L32 (mouse) or CYPA (human).
Western Blot Analysis
Tissues and adipocytes were lysed in modified radioimmunoprecipitation assay buffer with proteinase inhibitor cocktail (GenDEPOT). Proteins were separated by SDS-PAGE and immunoblotted against the specific antibodies shown in Supplementary Table 1.
Isolation of Mature Adipocytes, Stromal Vascular Fraction, and Adipogenic Progenitor Cells
Adipose tissues were digested in Ham’s F10 media (Gibco) containing 1.5% BSA (GenDEPOT) and 0.1% collagenase type I (Worthington) at 37°C for 30 min and were filtered through a cell strainer. After centrifugation, floating adipocytes were harvested. For stromal vascular fraction (SVF) isolation, pellets were filtered through a cell strainer, centrifuged at 4°C, and then were harvested as the SVF. Adipogenic progenitor cells were isolated using anti–platelet-derived growth factor receptor-α–phycoerythrin (PE)/anti-PE microbeads with the magnetic column according to the manufacturer's instructions (Miltenyi Biotec). Cells were cultured in DMEM with 20% FBS (HyClone) and 10 ng/mL human fibroblast growth factor 2 (FGF2) (Miltenyi Biotec), and were differentiated in DMEM with 20% FBS, 0.5 mmol/L 3-isobutyl-1-methylxanthine (Sigma-Aldrich), 125 μmol/L indomethacin (Sigma-Aldrich), 2 μmol/L dexamethasone (Sigma-Aldrich), 5 μg/mL insulin, and 0.5 μmol/L rosiglitazone (Sigma-Aldrich) for 2 days.
siRNA Transfection
At 100% confluency, cells were transfected with control, mouse Prmt1, or rodent Prkaa1/Prkaa2 siRNA pools (Dharmacon) using Lipofectamine RNAiMAX Transfection Reagent (Invitrogen) according to the manufacturer's instructions. Cells were then induced with the differentiation media shown above.
Adenoviruses
3T3-L1 cells (male origin) were infected with adenovirus (Ad) green fluorescent protein, Ad-mFOXO3, or Ad-hPRMT6 for 48 h. For knockdown of PRMT1, 3T3-L1 cells were infected with Ad-US (nonspecific RNAi) or Ad-Prmt1i for 72 h.
RNA-Sequencing Analysis
Total RNAs from epididymal WAT (eWAT) adipocytes of Prmt1 f/f and FKO mice under 10-week HFD were prepared with the RNeasy Plus Mini Kit. To assess the integrity of the total RNA, samples were run on the TapeStation RNA ScreenTape (Agilent Technologies).
A library was prepared by the TruSeq mRNA Sample Prep kit (Illumina) according to the manufacture’s protocol. Indexed libraries were sequenced using the HiSeq 4000 platform (Illumina) by Macrogen Incorporated, Korea. The raw reads were processed and were aligned to the Mus musculus (mm10) using Hierarchical Indexing for Spliced Alignment of Transcripts (HISAT v2.1.0) (19). Transcript assembly of known transcripts was processed by StringTie v1.3.4d (20). Abundance of transcripts and genes was calculated as read count or fragments per kilobase per million value per sample.
Volcano plot was analyzed by calculating fold change (log2[Prmt1 FKO/Prmt1 f/f]) and significance (−log10[P value]), wherein differentially expressed genes (DEG) were selected by cutoffs (P < 0.01 or P < 0.05; >1.3-fold). Gene ontology (GO) analysis was conducted by GOrilla (https://cbl-gorilla.cs.technion.ac.il/) with supply of upregulated DEG (P < 0.01, log2[fold change] >0.4) under the background of expressed transcripts in a matched control set (selected as read count ≥100). GO analysis results (P < 10−6) were visualized by Reduce & Visualize Gene Ontology (REVIGO; https://revigo.irb.hr/). Functional enrichment analysis was performed by gene set enrichment analysis (https://software.broadinstitute.org/gsea), correlated with two sets of FOXO3 targets, downregulated transcripts by expression of a constitutively active FOXO3 (n = 68) and FOXO3-induced transcriptional changes identified by RNA polymerase II (RNAPII) occupancy (n = 447) to calculate the normalized enrichment score and adjusted P value. The raw data for the RNA-seq analysis was deposited in the Gene Expression Omnibus repository (GSE167446).
mtDNA Quantification
Genomic DNA was prepared using the Exgene Tissue mini kit (GeneAll). Relative mtDNA copy number was determined as a ratio of mtDNA marker (Nd1, Cytb, and Co1) to nuclear DNA marker (H19) by RT-qPCR.
Histology and Tissue Staining
For the staining of mitochondria, deparaffinized adipose tissue sections were dehydrated and stained with 200 nmol/L MitoTracker Green FM (Invitrogen) at room temperature for 1 h. For identifying the intact mitochondria, tissue sections were blocked with 1% BSA, incubated with anti-pyruvate dehydrogenase (PDH) at 4°C for 20 h, and then incubated with Alexa Fluor 488 goat anti-mouse (IgG) secondary antibody at room temperature for 1 h. For detecting the evidence for lipophagy, tissue sections were incubated with anti-light chain 3B (LC3B) and antiperilipin-1 at 4°C for 20 h and then incubated with Alexa Fluor 488 goat anti-mouse (IgG) and Alexa Fluor 568 goat anti-rabbit (IgG) secondary antibody at room temperature for 1 h. For F4/80 staining, tissue sections were incubated with anti-F4/80 at 4°C for 20 h and then incubated with anti-rat IgG-horseradish peroxidase at room temperature for 1 h.
Ex Vivo Lipolysis
Fat pads were isolated from 20-week HFD-fed mice and incubated in DMEM containing 2% BSA with or without 10 μmol/L forskolin (Sigma-Aldrich) at 37°C for 2 h. Secreted glycerol levels were measured using the Free Glycerol Assay kit (Abcam).
Oxygen Consumption Rate
Fat pads were isolated from 14-week HFD-fed mice, minced into 10-mg pieces, loaded on to Seahorse XF24 Islet Capture Microplates, and cultured in DMEM containing 25 mmol/L glucose and 1 mmol/L sodium pyruvate. Basal oxygen consumption rate and maximal respiration values were measured using Seahorse XF Analyzers (Agilent Technologies).
Indirect Calorimetry Analysis
Energy expenditure of HFD-fed Prmt1 f/f mice and Prmt1 FKO mice were analyzed by using a metabolic cage (OxyletPro System–Physiocage [Panlab]; Harvard Apparatus) as described (21). The energy expenditure ANCOVA analysis done for this work was provided by the National Institute of Diabetes and Digestive and Kidney Diseases Mouse Metabolic Phenotyping Centers (www.mmpc.org) using their Energy Expenditure Analysis page (https://www.mmpc.org/shared/regression.aspx; supported by grants DK076169 and DK115255O).
Luciferase Assay
Putative FOXO3 binding motifs in the Prkaa2 promoter were identified using the Eukaryotic Promoter Database. Three promoter regions were inserted into the pGL4.12 vector (Promega). Luciferase activities were measured using the Luciferase Assay System (Promega).
Chromatin Immunoprecipitation
Statistical Analysis
All data are expressed as means ± SEM. Statistical analysis was performed using the Student t test or one-way ANOVA with the Tukey post hoc test. P < 0.05 was considered statistically significant.
Study Approval
The Korea University Institutional Animal Care and Use Committee (KUIACUC-2018-0031) approved all animal experimental procedures.
All work on human subjects was performed as approved by the Seoul National University Bundang Hospital Institutional Review Boards (IRBs). Male nonobese subjects were enrolled from normal adipose tissue banking who underwent surgery for kidney donation (IRB #B-1801445301). Male obese subjects were enrolled from bariatric bypass surgery for rescuing morbid obesity (IRB #B-1812513302). The cutoff for human obesity (in BMI) is based on the definition from the World Health Organization (www.who.int/news-room/fact-sheets/detail/obesity-and-overweight) and the previous report (23). All informed consents were obtained from enrolled subjects.
Data and Resource Availability
The data in the current work are available upon request to the author.
Results
PRMT1 Expression Is Predominant in WAT
To investigate the role of PRMT1 in adipose tissues, we first attempted to determine PRMT1 expression in various adipose tissues. We found that PRMT1 expression is most abundant in eWAT among adipose tissues tested. The level of PRMT1 expression was similar in other visceral WAT compared with inguinal WAT (iWAT), whereas only a small fraction of PRMT1 was detected in brown adipose tissue (BAT) (Fig. 1A). In the eWAT, we found that Prmt1 expression was 1.8-fold more enriched in the mature adipocytes compared with SVF (Supplementary Fig. 1A). Thus, we decided to generate adipocyte-specific Prmt1 KO mice (Prmt1 FKO mice) to explore the role of PRMT1 in adipose tissues, and wanted to mainly focus on the WAT.
PRMT1 expression was specifically reduced in adipose tissues but not in other tissues in Prmt1 FKO mice compared with the control (Supplementary Fig. 1B). Moreover, we detected the nearly complete elimination in PRMT1 expression in mature adipocytes from Prmt1 FKO mice, showing that the depletion of Prmt1 was indeed specific to the mature adipocytes (Supplementary Fig. 1C). Compared with the control, Prmt1 FKO mice showed a slight increase in body weight with a small reduction in eWAT weight (Supplementary Fig. 2A and B), displaying reduced fat mass and enhanced lean mass compared with Prmt1 f/f mice (Supplementary Fig. 2C). We did not observe any changes in blood glucose levels or plasma NEFA and TG levels upon adipocyte-specific depletion of Prmt1 (Supplementary Fig. 2D–H). In addition, insulin tolerance testing revealed no differences between the two genotypes (Supplementary Fig. 2I), suggesting that the depletion of Prmt1 in adipocytes did not cause changes in glucose metabolism under NCD-fed conditions.
WAT plays a critical role in regulating whole-body metabolism under DIO conditions. Interestingly, HFD feeding dramatically induced expression of PRMT1 in eWAT and to a lesser extent in iWAT (Fig. 1B), prompting us to explore the potential significance of our finding in humans. We therefore measured PRMT1 expression in human visceral WAT of obese individuals as well as lean counterparts. Interestingly, we observed a slight induction of PRMT1 mRNA levels in the visceral WAT of obese humans compared with the control subjects (Fig. 1C). Furthermore, protein levels of PRMT1 were significantly higher in visceral WAT of obese humans compared with the control subjects (Fig. 1D), and more importantly, we found a significant correlation between PRMT1 protein levels and BMI (Fig. 1E). These data suggested that a positive correlation between the obesity and PRMT1 expression in the visceral WAT is conserved between species.
Depletion of Prmt1 in Adipocytes Reduced Lipid Droplet Size in WAT
To interrogate the role of PRMT1 in WAT under DIO conditions, we fed Prmt1 f/f and Prmt1 FKO mice the HFD for up to 16 weeks. Body weight of Prmt1 FKO mice was heavier than that of Prmt1 f/f mice under NCD, which disappeared within 2 weeks of HFD feeding (Fig. 2A and B). On the other hand, after 16 weeks of HFD feeding, we found that eWAT weight was significantly reduced upon adipocyte-specific depletion of Prmt1 without changes in skeletal muscles. In general, WAT from Prmt1 FKO mice looked smaller and darker compared with the WAT from Prmt1 f/f mice (Fig. 2C and Supplementary Fig. 3A–G). In line with this result, we observed a slight reduction in the fat mass of Prmt1 FKO mice compared with the control, although the difference did not reach statistical significance (Supplementary Fig. 3H). Interestingly, we observed a potential compensatory increase in liver weight of Prmt1 FKO mice compared with Prmt1 f/f mice.
To delineate a potential mechanism by which adipocyte-specific depletion of Prmt1 affects WAT mass, we performed histological analysis. Hematoxylin-eosin staining revealed that eWAT of Prmt1 FKO mice showed reduced adipocyte size compared with the control, without drastic changes in adipocyte size in iWAT (Fig. 2D and E, and Supplementary Fig. 4). In contrast, adipocytes with increased lipid droplets were observed in BAT of Prmt1 FKO mice (Fig. 2D), suggesting that adipocyte-specific depletion of Prmt1 might promote a milder form of lipodystrophy and could result in the appearance of TG-rich BAT upon HFD feeding in this setting.
Depletion of Prmt1 Induced AMPK Expression via PRMT6-FOXO Pathway
Depletion of Prmt1 in adipocytes led to the reduction of adipocyte size in eWAT, prompting us to explore a potential mechanism by which Prmt1 deficiency promotes the reduced lipid contents in adipocytes. We found that expression of Atgl, but not Hsl, was elevated in eWAT and iWAT of Prmt1 FKO mice compared with the control (Supplementary Fig. 5A). We did not observe significant changes in phosphorylated hormone-sensitive lipase levels or ex vivo lipolysis in Prmt1 FKO eWAT compared with the control (Supplementary Fig. 5B). Interestingly, phosphorylated hormone-sensitive lipase levels and lipolysis were significantly higher in iWAT from Prmt1 FKO mice compared with Prmt1 f/f mice (Supplementary Fig. 5C). Since we observed more pronounced changes in tissue mass or adipocyte size of eWAT compared with iWAT upon depletion of Prmt1, we suspected that a slight enhancement of lipolysis might not be causal to the current phenotype shown in Prmt1 FKO mice.
To gain further mechanistic insight, we performed RNA-seq analysis by using mature adipocytes from eWAT of Prmt1 f/f or Prmt1 FKO mice. Biological processes of GO that were significantly enriched (P < 10−6) in upregulated transcripts by Prmt1 KO (P < 0.05; fold change >1.3) include autophagy (e.g., Atg7, Sqstm1, Lamp1, Lamp2, and Atg16l1) as well as metabolic pathways in the mitochondria (e.g., Tfam, Ndufs8, Cpt1b, and Cox5b) (Fig. 3A and B). Intriguingly, these genes were known targets of AMPK, prompting us to look for the potential involvement of this kinase in the regulation of the signaling pathway in Prmt1-depleted adipocytes (24,25). We found that expression of Prkaa2, which encodes AMPKα2, was profoundly induced in adipocytes of eWAT from the Prmt1 FKO mouse compared with the control, suggesting that the AMPK pathway could be enhanced via a transcriptional mechanism (Fig. 3C). Increased expression of Prkaa2, but not Prkaa1, upon depletion of Prmt1 was recapitulated in 3T3-L1 adipocytes, confirming that the enhanced expression of Prkaa2 by Prmt1 knockdown can be attributed to the cell-autonomous mechanism (Supplementary Fig. 6A). As a result of the increased Prkaa2 expression, total and phosphorylated AMPK levels were both enhanced upon depletion of Prmt1 in eWAT and 3T3-L1 adipocytes (Fig. 3D and Supplementary Fig. 6B).
We previously showed that the depletion of Prmt1 in the skeletal muscle promotes catabolic signaling pathways by the PRMT6-FOXO3 transcriptional axis (11). Thus, we wanted to look for the expression of PRMT families as well as FOXO proteins in eWAT. We observed an increase in mRNA levels of type I PRMTs, with the highest increase in Prmt6, upon depletion of Prmt1 in adipocytes (Supplementary Fig. 6C). Indeed, we observed a dramatic increase in PRMT6 protein levels in Prmt1-depleted eWAT (Supplementary Fig. 6D). Furthermore, we also observed an enhanced expression of FOXO3 as well as FOXO1, showing that the PRMT6-FOXO axis could be activated upon depletion of Prmt1 in adipocytes (Supplementary Fig. 6B and D). To determine whether the PRMT6-FOXO3 axis regulates AMPK expression, we ectopically expressed PRMT6 and FOXO3 in 3T3-L1 adipocytes. Indeed, together with the increased AMPK protein levels, we were able to observe increased mRNA levels of Prkaa2 but not that of Prkaa1 upon coexpression of PRMT6 and FOXO3 in 3T3-L1 adipocytes (Fig. 3E and Supplementary Fig. 6F). Luciferase assay revealed that coexpression of PRMT6 and FOXO3 increased Prkaa2 promoter activity (Fig. 3F). By deletion analysis, we found that the promoter sequences containing −300 to +40 possessed an essential element for a transcriptional response to PRMT6/FOXO3 (Fig. 3F), which was confirmed by chromatin immunoprecipitation assay (Fig. 3G). These data suggest that increased expression of PRMT6 upon depletion of Prmt1 could specifically activate the transcriptional activity of FOXO3 to enhance expression of Prkaa2, resulting in the activation of the AMPK pathway.
Prmt1 KO Promotes Lipid Mobilization by Autophagy
Autophagy has emerged as an important regulator of lipid homeostasis (26). We found that genes involved in the process of autophagy as well as Prkaa2 were enhanced upon depletion of Prmt1 in adipocytes (Figs. 3B and C and 4A and Supplementary Fig. 7A). In addition, we also observed a consistent activation of key proteins involved in the process of autophagy, such as ATG5, ATG7, ATG16L1, and LC3-I/LC3-II, as well as pT172-AMPK in Prmt1-depleted eWAT (Figs. 3D and 4B and Supplementary Fig. 7B).
To investigate whether depletion of Prmt1 promotes enhanced autophagy, we performed autophagic flux assay in vivo. We noticed slightly higher LC3-II levels in Prmt1-depleted WAT compared with the control, perhaps indicating an increased basal autophagy (Fig. 4C and D and Supplementary Fig. 8A). Furthermore, leupeptin-induced inhibition of lysosomal activity further showed higher LC3-II protein levels in eWAT (and a less extent in iWAT) of Prmt1 FKO mice compared with the control, showing that autophagic flux was enhanced in Prmt1-depleted WAT. On the other hand, we did not observe differences in autophagic flux in the liver or BAT between the two genotypes (Supplementary Fig. 8B), suggesting that such changes occurred in a cell-autonomous manner. Increased autophagic flux in Prmt1 FKO mice might promote lipophagy in WAT. In line with this speculation, we were able to observe an increased localization of LC3-II in the lipid droplet of Prmt1-depleted WAT compared with the control, showing a sign of an increased lipophagy (Fig. 4E and Supplementary Fig. 8C and D). These data collectively suggest that increased autophagic flux might promote reduced lipid droplet size in eWAT (and a lesser extent in iWAT) via lipophagy upon adipocyte-specific depletion of Prmt1.
Depletion of Prmt1 Enhances Oxidative Metabolism in Adipocytes
Increased lipophagy could result in enhanced lipid catabolism in the mitochondria. GO analysis showed that Prmt1 depletion significantly induced genes that are predominantly localized and associated with mitochondria, including mitochondrion, mitochondrial part, and mitochondrial protein complex (>1.2 fold enrichment, P < 10−11) (Fig. 5A). The color of eWAT of Prmt1 FKO mice looked darker than that of control mice (Supplementary Fig. 3A), hinting that the depletion of Prmt1 in adipocytes could result in the increase in mitochondria. Indeed, we were able to observe an increase in mitochondria in eWAT (and a lesser extent in iWAT) of Prmt1 FKO mice compared with the control, as evidenced by MitoTracker staining and the measurement of mtDNA content (Fig. 5B and Supplementary Fig. 9A and B). Furthermore, we showed that mitochondria in eWAT of Prmt1 FKO mice were intact as displayed by PDH staining (Fig. 5B). These data suggest that a population of healthy mitochondria might be maintained in Prmt1-depleted adipocytes, perhaps via the increased expression of genes involved in mitochondrial biogenesis (Figs. 3B and 5C and Supplementary Fig. 9C).
Among the target genes that were upregulated upon Prmt1 depletion include nuclear genes encoding mitochondrial oxidative phosphorylation (OxPhos) machinery, suggesting that mitochondrial catabolism might be enhanced in eWAT (Fig. 5C). Indeed, we were able to confirm that mRNA levels for mitochondrial energy metabolism as well as all the components of mitochondrial OxPhos proteins were upregulated in Prmt1-depleted eWAT (Fig. 5D and Supplementary Fig. 9C). As a result, Prmt1-deficient eWAT showed an increased oxygen consumption rate, indicating the enhanced mitochondrial oxidation compared with the control (Fig. 5E). Furthermore, we also observed that mitochondrial OxPhos protein expression was higher in the differentiated primary adipocytes from Prmt1 FKO eWAT compared with the control, confirming that the effect of Prmt1 depletion on OxPhos proteins in eWAT was mainly due to the changes in the mature adipocytes (Fig. 5F). Interestingly, we observed that OxPhos protein expression in other visceral WAT as well as iWAT, but not in BAT, was similarly altered upon depletion of Prmt1 (Supplementary Fig. 10A–D). These data collectively suggest that depletion of Prmt1 would enhance mitochondrial activity, leading to the increased lipid catabolism specifically in adipocytes of WAT.
Since we found that the depletion of Prmt1 enhanced expression of AMPK in adipocytes, we wanted to confirm whether the increased AMPK expression was casual to the current phenomenon. Indeed, we observed that increased LC3-II levels upon depletion of Prmt1 were restored to the basal level with the concomitant depletion of Prkaa in adipocytes. Furthermore, increased OxPhos proteins with Prmt1 knockdown in adipocytes were also reduced to the control level in the Prmt1/Prkaa double-knockdown conditions, supporting our hypothesis that the increased lipid catabolism in Prmt1-depleted adipocytes might be due to the enhanced expression of AMPK (Supplementary Fig. 11).
Adipocyte-Specific Depletion of Prmt1 Enhances Energy Expenditure
Increased oxidative metabolism could promote enhanced energy expenditure, due in part to the activation of adaptive thermogenesis via an uncoupling protein 1 (UCP1)mediated mechanism. While we did not observe changes in UCP1 proteins levels from BAT of Prmt1 FKO mice and the control, we found that UCP1 protein levels were significantly elevated in iWAT upon adipocyte-specific depletion of Prmt1 in DIO conditions (Supplementary Fig. 12A). Our finding might be in a stark contrast to the recent study by Wu and colleagues (27), who suggested that adipocyte-specific depletion of Prmt1 led to the defects in the adaptive thermogenesis both in BAT and iWAT. While we also observed the increased lipid accumulation in BAT in Prmt1 FKO mice compared with the control, we suspected that the phenotype shown in BAT might be secondary to the changes in WAT since Prmt1 expression is not prominent in BAT (Fig. 1A). Indeed, cold exposure effectively reduced the lipid accumulation in BAT in both genotypes, showing that Prmt1 might not play a significant role in adaptive thermogenesis in BAT at least in DIO conditions (Supplementary Fig. 12B).
To address whether the increased mitochondria in WAT led to the enhanced energy expenditure, we performed indirect calorimetric analysis. As expected, Prmt1 FKO mice showed an enhanced energy expenditure in comparison with Prmt1 f/f mice, without thereby changes in food intake (Supplementary Fig. 13). These data suggest that the increased mitochondrial activity in WAT led to the enhanced energy expenditure in Prmt1 FKO mice in DIO setting.
Depletion of Prmt1 in Adipocytes Perturbs Glucose and Lipid
Enhanced energy expenditure together with decreased adipocyte size in eWAT is often associated with the decreased infiltration of proinflammatory M1 macrophages and the increased appearance of anti-inflammatory M2 macrophages (4). Surprisingly, we observed an increased macrophage infiltration of both eWAT and iWAT of Prmt1 FKO mice compared with the control, as indicated by more evident showing of crown-like structures with F4/80 immunostaining (Supplementary Fig. 14A). Interestingly, markers of M1 and M2 macrophages were generally elevated in eWAT of Prmt1 FKO mice compared with the control, although most of changes did not reach statistical significance due in part to the variability of each sample (Supplementary Fig. 14B). In contrast, we observed increased plasma levels of proinflammatory cytokines IL-6 and TNF-α in Prmt1 FKO mice compared with the control (Supplementary Fig. 14C), suggesting that adipocyte-specific depletion of Prmt1 could promote adipose tissue inflammation.
This phenotype is reminiscent of mouse models of lipodystrophy, which displayed impaired metabolic homeostasis despite the reduced adipocyte size due to the adipose tissue inflammation. Indeed, we observed increased fasting blood glucose levels in Prmt1 FKO mice compared with the control, without significant changes in plasma insulin levels (Fig. 6A and Supplementary Fig. 15A and B). Furthermore, we also observed an impaired insulin tolerance in Prmt1 FKO mice compared with the control (Fig. 6B). In addition, plasma TG was significantly elevated by adipocyte-specific depletion of Prmt1 in DIO mice, without discernible changes in plasma NEFA levels (Fig. 6C and Supplementary Fig. 15C). These data collectively suggest that adipocyte-specific depletion of Prmt1 perturbs glucose and lipid metabolism in DIO mice.
In addition to the increased plasma TG levels, we observed an increased TG accumulation in liver upon adipocyte-specific Prmt1 depletion (Fig. 6C and D). Interestingly, while lipogenic gene expression (Srebf1c, Fasn, and Scd1) was generally reduced by adipocyte-specific Prmt1 depletion, genes involved in the fatty acid uptake and lipid droplet fusion (Pparg2, Cd36, and Cidea) were upregulated in liver (Supplementary Fig. 15D), suggesting that the accumulation of hepatic lipid might be mainly derived from the dietary or endogenous lipid instead of the increased de novo lipogenesis. These data corroborate our hypothesis that the increased mitochondrial catabolism of eWAT resulted in the milder form of lipodystrophy, leading to the ectopic lipid accumulation in the peripheral tissues including liver and BAT.
Accumulation of TG in the peripheral tissues is associated with the perturbation of insulin signaling. Thus, we performed Western blot analysis for phosphorylated (p)-Akt in metabolic tissues. Indeed, we observed a significant reduction in p-Akt in Prmt1-depleted eWAT and liver (Fig. 6E and F). We observed a similar trend in p-Akt levels from BAT or skeletal muscles but not from iWAT upon adipocyte-specific depletion of Prmt1, although it did not reach the statistical significance. (Supplementary Fig. 16E and F). These data collectively suggest that the perturbation of lipid metabolism in WAT of Prmt1-depleted mice impaired systemic insulin signaling and promoted the resultant dysregulation of glucose homeostasis in DIO mice.
Discussion
Recent studies revealed the role of PRMT1 in metabolism by using KO mouse models. While knockdown of Prmt1 led to the reduction of FOXO1-dependent hepatic gluconeogenesis (12), liver-specific depletion of Prmt1 promotes the incidence of alcohol-induced fatty liver diseases by reducing HNF4 expression (13). Skeletal muscle-specific depletion of Prmt1 induces muscle atrophy phenotypes due to the dysregulation of catabolic pathways via the PRMT6-FOXO3 pathway (11), and cardiomyocytes-specific deletion of Prmt1 promotes CAMKII-mediated hypertrophy and fibrosis, resulting in cardiomyopathy in mice (28). Despite the importance of adipocytes in the energy metabolism, the role of Prmt1 in the adipocyte-derived metabolic pathway in DIO conditions has not been clearly demonstrated.
Macroautophagy in adipocytes has been shown to be critical in the process of adipogenesis (29,30). Proper degradation of cellular organelles, including lipid droplets, leads to the formation of unilocular lipid droplets, which is characteristic of adipocytes of WAT. Interestingly, recent studies suggested that the defect in autophagy in mature adipocytes might be linked to the metabolic disorder. Adipocyte-specific Atg7 KO mice showed reduced adiposity and weight gain in response to the HFD (29,30). Furthermore, these mice exhibited improved insulin sensitivity that is associated with reduced plasma lipid levels, suggesting that adipocyte-specific reduction of autophagy could be beneficial in reducing obesity-related metabolic disease. However, these studies used the aP2-Cre system for the generation of the adipocyte-specific KO model, which has been proven to be problematic due to the ubiquitous expression of aP2 (Fabp4) in various tissues. Thus, the fortuitous deletion of Atg7 in nonadipose tissues could affect the metabolic phenotype in this mouse model. Indeed, a recent study used an adiponectin-Cre driver to achieve adipocyte-specific Atg3 KO to specifically inhibit the autophagy in mature adipocytes. The authors showed that adipocyte-specific Atg3 KO mice displayed peripheral insulin resistance as opposed to the previous study (31), suggesting that the inhibition of autophagy in mature adipocytes rather impaired the insulin sensitivity. Interestingly, enhanced autophagy in adipocytes could also instigate insulin resistance as in our model, suggesting that the tight regulation of autophagy in adipocytes is necessary to maintain metabolic homeostasis.
In the current study, we explored the role of PRMT1 in adipocytes under metabolic stress. Surprisingly, we found that Prmt1 depletion in adipocytes promoted reduced adipocyte size with increased functional mitochondria in eWAT via the induction of the AMPK pathway. Despite the reduced adipocyte size and the enhanced mitochondrial function, we rather detected increased adipose tissue inflammation, ectopic TG accumulation in the peripheral tissues, and dysregulation of glucose homeostasis in Prmt1 FKO mice, which is reminiscent of a case of lipodystrophy. In support of our hypothesis, we observed an impaired insulin signaling in the peripheral tissues in Prmt1 FKO mice in the DIO setting.
One of the limitations of our current study is the lack of the rigorous analysis of overall energetics of Prmt1 f/f mice and Prmt1 FKO mice. Although we performed indirect calorimetry assay, key experiments that are necessary to fully understand the complex energetic phenotypes of Prmt1 FKO mice, such as measuring dietary efficiency between different diets, were not conducted. In addition, further study is necessary to test multiple conditions, including the thermoneutral temperature, cold exposure, as well as NCD feeding to interrogate the functional cross talk among different adipose tissues in the control of whole-body energy metabolism in our mouse model.
In summary, we propose that HFD-induced activation of PRMT1 in WAT could prevent this tissue from excessive lipid mobilization, which thwarts the accumulation of lipids in the peripheral tissues and the resultant insulin resistance in the DIO setting (Supplementary Fig. 16). It is tempting to speculate whether activation of PRMT1 in WAT could alleviate the symptoms of moderate lipodystrophy that is associated with the fatty liver and insulin resistance in animal models as well as humans.
This article contains supplementary material online at https://doi.org/10.2337/figshare.14665905.
Article Information
Acknowledgments. The authors thank Dr. Chul-Ho Lee (Korea Research Institute of Bioscience & Biotechnology) for providing adiponectin-Cre transgenic mice.
Funding. This research was supported by the National Research Foundation of Korea grants funded by the Korean Government (MSIT) (NRF-2018R1A2B3001540, NRF-2015R1A5A1009024, NRF-2019M3A9D5A01102794, and NRF-2021R1A2C3003435). S.C. was supported by NRF-2019R1A6A3A01096171, and S.W.C. was supported by NRF-2016R1C1B2014390. S.-H.K. was also supported by a grant from Korea University.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. S.C., D.C., and Y.-K.L. performed experiments. S.C., J.K.S., S.W.C., T.J.O., S.H.C., and S.-H.K. interpreted data. S.C. and S.-H.K. conceived the idea and developed the study design. S.C. and S.-H.K. wrote the manuscript. S.H.A. and S.W.C. analyzed the RNA sequencing data. S.-H.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.