All of the animal experiments in this study were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at UCSF and Beth Israel Deaconess Medical Center. Unless otherwise specified, all of the mice had free access to food and water, and were housed under 12 h–12 h light–dark cycle, at 22 °C, and 45% humidity on average. Cul2flox/− mice in the C57BL/6J background were generated by Applied StemCell using CRISPR–Cas9 technology. Appbp2flox/− mice in the C57BL/6J background were made by Cyagen with CRISPR–Cas9. Adipocyte-specific Cul2-KO (Adipo-Cul2-KO mice) and Appbp2-KO mice (Adipo-Appbp2-KO mice) were developed by crossing Cul2 or Appbp2 floxed mice with Adiponectin-Cre mice (B6; FVB-Tg (Adipoq-Cre)1Evdr/J, 028020). Appbp2 knock-in mice that carry the S561N mutation (equivalent to the human SNP in the APPBP2 gene) were generated by Cyagen. To recapitulate the human S561N variant of APPBP2 in mice, we mutated Ser561 to Asn and Ser562 to Thr by co-injecting the gRNA and the donor oligo containing p.Y551 (TAT to TAC) and p.A556 (GCC to GCG) into fertilized mouse oocytes. A list of the primer sequences used for genotyping and gRNA is provided in Supplementary Table 1.
The stromal vascular fraction (SVF) from the inguinal WAT of C57BL/6J mice, Prdm16flox/flox mice and Appbp2flox/flox mice were immortalized by expressing the SV40 large T antigen as described previously29. For the generation of Prdm16-KO and Appbp2-KO cells, immortalized Prdm16flox/flox or Appbp2flox/flox preadipocytes were infected with retrovirus expressing cre (34565, Addgene), followed by hygromycin selection at a dose of 200 μg ml−1. For the generation of inguinal preadipocytes stably expressing Cul2, immortalized preadipocytes were infected with retrovirus expressing codon-optimized mouse Cul2, followed by blasticidin selection at a dose of 10 µg ml−1. For AAV infection of primary SVF cells, cells isolated from Cul2flox/flox or Appbp2flox/flox mice were infected with AAV-cre (105537-AAV8, Addgene) or AAV-GFP (105530-AAV8, Addgene) in growth medium overnight. After changing to a fresh growth medium, the cells were cultured for another 60 h. Preadipocytes were seeded into coated plates, and differentiation was induced by culturing cells with DMEM medium containing 10% FBS (Atlanta Biologicals), 5 µg ml−1 insulin, 1 nM T3, 0.5 µM rosiglitazone, 0.5 mM isobutylmethylxanthine, 125 nM indomethacin and 2 µg ml−1 dexamethasone. After 48 h, cells were cultured in medium containing 10% FBS, 5 µg ml−1 insulin, 1 nM T3 and 0.5 μM rosiglitazone for another 5–7 days. HEK293T cells, HEK293 virus packaging cells and C2C12 cells were maintained in DMEM high-glucose medium containing 10% FBS and 1% penicillin–streptomycin.
Cul1 (19896), Cul2 (19892), Cul3 (19893), Cul4a (19951), Cul4b (19922), Cul5 (19895) and Cul7 (20695) were obtained from Addgene. The EHMT1, HA-Ub, HA-PRDM16, Flag-PRDM16, GST-PRDM16 constructs were developed in our laboratory. Human APPBP2 cDNA and mouse Appbp2 cDNA constructs were amplified using standard PCR techniques from plasmids (MHS6278-202757182 for human APPBP2, MMM1013-202761267 for mouse Appbp2, Horizon) and subsequently inserted into mammalian expression vectors. Codon-optimized mouse Cul2 cDNA and codon-optimized human Appbp2 cDNA were synthesized by GenScript and cloned into a PMSCV-blasticidin vector (75085, Addgene). Point mutations used in the study were introduced using site-directed mutagenesis with the In-Fusion HD Cloning Plus (638909, Takara Bio) kit. All of the constructs were confirmed by sequencing.
Antibodies and reagents
The following reagents were used in this study: MG132 (474790, Sigma-Aldrich), MLN4924 (501146629, Thermo Fisher Scientific), anti-Flag(R) M2 affinity gel (A2220, Sigma-Aldrich), Pierce anti-HA magnetic beads (88836, Thermo Fisher Scientific), 3×Flag peptide (F4799, Sigma-Aldrich), 3×HA peptide (AS-63764, Anaspec). The following antibodies were used in this study: anti-UCP1 (ab-10983, Abcam), anti-UCP1 (U6382, Sigma-Aldrich), anti-CUL2 (sc-166506, Santa Cruz), anti-recombinant CUL2 (EPR3104(2)) (ab166917, Abcam), anti-APPBP2 (NBP2-81781, Novus), anti-Flag-HRP (A8592, Sigma-Aldrich), anti-HA (sc-7392, Santa Cruz), anti-MYC (sc-40, Santa Cruz), anti-PPARγ (E-8) (sc-7273, Santa Cruz), OXPHOS cocktail (ab110413, Abcam), anti-ubiquitin (sc-8017, Santa Cruz), anti-GST (sc-138, Santa Cruz), anti-GAPDH (sc-32233, Santa Cruz), anti-β-actin (A3854, Sigma-Aldrich), anti-MUM1 (12682-1-AP, Proteintech), anti-SMAD4 (10231-1-AP, Proteintech), anti-EHD2 (11440-1-AP, Proteintech), anti-GTF2I (10499-1-AP, Proteintech), anti-HDAC1 (10197-1-AP, Proteintech), anti-CNOT1 (14276-1-AP, Proteintech), anti-CNOT9 (22503-1-AP, Proteintech), anti-CHD4 (14173-1-AP, Proteintech), anti-PRDM3 (C50E12) (2593S, Cell Signaling Technology), anti-EHMT1(E6Q8B) (35005S, Cell Signaling Technology), PhosphoPlus Akt (Ser473) Antibody Duet (8200S, Cell Signaling Technology), anti-HSP90 (4874S, Cell Signaling Technology), normal mouse IgG (sc-2025, Santa Cruz), rabbit IgG, polyclonal-isotype control (ab37415, Abcam), goat anti-rabbit light chain secondary antibody (NBP2-75935, Novus), goat anti-mouse IgG, light-chain specific antibody (91196S, Cell Signaling Technology), anti-PRDM16 (AF6295, R&D systems), polyclonal antibody against PRDM16 generated by immunizing rabbit with recombinant human PRDM16 (GenScript).
shRNA constructs and virus production
The following lentiviral shRNA clones were purchased from GeneCopoeia: scrambled control (CSHCTR001-LVRH1H); Cul2 (MSH036079-LVRH1H); Klhdc2 (MSH034863- LVRH1H); Lrrc14 (MSH038419-LVRH1H). Lentiviral shRNA constructs were generated by cloning into pLKO.1-hygromycin (24150, Addgene) or pLKO.1-blasticidin (26655, Addgene). A list of the sequences used in the study is provided in Supplementary Table 1.
For lentivirus production, HEK293 packaging cells were transfected with 10 µg lentiviral plasmids and 10 µg packaging plasmids (psPAX2 and pMD2.G) using the calcium phosphate method. After 48 h, the culture supernatant was collected and filtered using a 0.45 µm filter. Inguinal WAT-derived SVF cells or HEK293T cells were incubated with the viral supernatant supplemented with 10 µg ml−1 polybrene for 24 h. Subsequently, stable cell lines were obtained by selection with the indicated antibiotics: hygromycin B (10687010, Thermo Fisher Scientific) at a dose of 200 µg ml−1, puromycin (A1113803, Gibco) at 1 µg ml−1 or blasticidin (A1113903, Thermo Fisher Scientific) at 10 µg ml−1.
Overexpression of Cul2 and Appbp2 in adipocytes
Flag-tagged codon-optimized mouse Cul2 cDNA and HA-tagged codon-optimized human Appbp2 cDNA were synthesized by GenScript and cloned into a PMSCV-blasticidin vector (75085, Addgene). For retrovirus production, HEK293 packaging cells were transfected with 10 µg above plasmid and 10 µg packaging plasmids (VSV and gag-pol) using the calcium phosphate method. After 48 h, the culture supernatant was collected and filtered by a 0.45 µm filter. Inguinal WAT-derived SVFs or Appbp2-KO preadipocytes were infected with retroviruses with 10 µg ml−1 polybrene for 24 h. Subsequently, cells were selected by blasticidin at a dose of 10 µg ml−1.
Affinity purification of protein complex and proteomics
Protein complex purification was performed as previously described4,6. For CUL2 complex purification, immortalized preadipocytes derived from mouse inguinal WAT were infected with retrovirus expressing Flag-tagged Cul2 or an empty vector. Adipocytes were grown to post-confluence and differentiated for 4 days. Cell extracts were prepared using lysis buffer (50 mM Tris-Cl pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA) supplemented with a protease inhibitor cocktail (Roche). The supernatant from lysates was incubated with anti-Flag M2 affinity gel for 2 h. The immunoprecipitants were washed three times and subsequently eluted by 3×Flag peptide (Sigma-Aldrich). The eluants were then TCA-precipitated, separated in a 4–15% acrylamide gradient gel, and visualized by silver staining or Coomassie blue staining.
For APPBP2 complex characterization, immortalized Appbp2-KO preadipocytes derived from mouse inguinal WAT were infected with retrovirus expressing HA-tagged Appbp2 or an empty vector. The adipocytes were grown to post-confluence and differentiated for 4 days, followed by MG132 treatment. The cells were homogenized to prepare nuclear extracts. The nuclear extracts were incubated with Pierce anti-HA magnetic beads (Thermo Fisher Scientific) for 2 h and then washed in a binding buffer (50 mM Tris-Cl pH 7.4, 150 mM NaCl, 0.5% Triton X-100). The complexes were eluted by 3×HA peptide (Anaspec), underwent TCA-precipitation, separated in a 4–15% acrylamide gradient gel, and subsequently visualized by silver staining or Coomassie blue staining.
Gel-resolved proteins were excised, digested with trypsin and individually analysed using reverse-phase liquid chromatography with tandem mass spectrometry (LC–MS/MS) using a high-resolution hybrid mass spectrometer (LTQ-Orbitrap, Thermo Fisher Scientific) with the TOP10 method at the Taplin Biological Mass Spectrometry Facility in Harvard Medical School. The LC–MS/MS data were searched against the IPI mouse database37. Proteins identified with at least two unique valid peptides were considered to be significant, and the FDR was estimated to be 0% using the target–decoy approach38.
Protein expression and purification
SF9 cells were obtained from the UC Berkeley Cell Culture Facility. SF9 cells were cultured in Sf-900 II SFM (10902088, Thermo Fisher Scientific) supplemented with 1% penicillin–streptomycin (Gibco) and were maintained at 27 °C without CO2. Baculovirus packaging and amplification were performed according to the established commercial protocol of the Bac-to-Bac C-His TOPO Expression System (A11100, Thermo Fisher Scientific). In brief, C-terminally His-tagged Appbp2 and Flag–Prdm16 cDNAs were cloned into the pFastBac TOPO vector and transformed into DH10Bac Escherichia coli competent cells to form a recombinant expression bacmid. The bacmid was then transfected into SF9 cells using ExpiFectamine Sf Transfection Reagent for the production of recombinant baculovirus particles (P0 virus). The high titre baculovirus was amplified by infection of more insect cells with P0/P1 virus. For the purification, SF9 insect cells in 1 l medium were infected and collected by centrifugation and frozen at −80 °C. Cells were incubated and stirred with the lysis buffer (50 mM Tris-Cl pH 7.4, 150 mM KCl, 1 mM PMSF, 10 mM imidazole and 0.1% Triton X-100) supplemented with a protease inhibitor cocktail (Roche) using a stir bar at 4 °C for 1 h. Subsequently, cells were Dounce-homogenized 50 times, sonicated and then centrifuged at 26,000 rpm for 1 h. The supernatant was mixed with Ni-NTA slurry and bound for 1 h at 4 °C. Beads were washed in wash buffer (50 mM Tris-Cl pH 7.4, 150 mM KCl, 20 mM imidazole) three times for 15 min and eluted in a buffer containing 50 mM Tris-Cl pH 7.4, 150 mM KCl and 250 mM imidazole. Eluted proteins were exchanged into storage buffer (50 mM Tris-Cl pH 7.4, 100 mM KCl, 1 mM DTT, PMSF, 10% glycerol), aliquoted and flash-frozen.
In vitro ubiquitination assays
Purified His–Flag–PRDM16 (400 nM), purified His–APPBP2 (600 nM), CUL2 (neddylated)/RBX1 (600 nM), or CUL1 (neddylated)/RBX1 (600 nM), or CUL5 (neddylated)/RBX2 (600 nM), UBE1 (120 nM), ubiquitin (20 µM), UBE2D1, or UBE2D3 or UBE2R1 (400 nM) were mixed in a reaction buffer containing 50 mM Tris–HCl, pH 7.4, 5 mM MgCl2, 2 mM ATP and 1 mM DTT. The reaction was carried out in a 30 μl volume at 37 °C for 60 min, and then resolved by SDS–PAGE. Ubiquitinated products were detected by immunoblotting. All of the proteins were obtained from Boston Biochem unless specified.
Ubiquitination assays in cells
HEK293T cells were transfected with the indicated plasmids using the calcium phosphate method. After 42 h, 20 µM MG132 was directly added to the medium. After incubation for 6 h, the cells were collected and lysed with RIPA buffer (9806S, Cell Signaling Technology) containing 1% SDS supplemented with EDTA-free protease inhibitors. Cell lysates were briefly sonicated, boiled at 95 °C for 10 min, and then centrifuged at 12,000 rpm for 15 min. The supernatants were then diluted 1:9 with RIPA lysis buffer to reduce the SDS concentration to 0.1%. Pierce anti-HA magnetic beads (88837, Thermo Fisher Scientific) were incubated with diluted lysates for 2 h at 4 °C. Immunoprecipitates were washed four times with RIPA lysis buffer and analysed by immunoblotting.
Identification of ubiquitination sites
Purified recombinant Flag–PRDM16 protein was processed for the in vitro ubiquitination reaction using methyl-Ub. The ubiquitinated Flag–PRDM16 was separated by SDS–PAGE, and the gel was stained with Coomassie blue. Protein bands were excised, destained and reduced in 1 mM DTT at 60 °C for 30 min, followed by alkylation in 5 mM iodoacetamide. After in-gel digestion, peptides were extracted from the gel with 5% formic acid/50% acetonitrile and dried completely in a speed-vac. At the time of analyses, samples were resuspended in a 2%/0.1% acetonitrile/formic acid solution. The extracted peptides were separated by an analytical capillary column (100 μm × 25 cm) packed with 2.6 μm spherical C18-reversed-phase silica beads (Accucore, Thermo Fisher Scientific). The Accela 600 HPLC pump was used to generate the following HPLC gradient: 5–35% in 60 min (A, 0.1% formic acid in water; B, 0.1% formic acid in acetonitrile). The eluted peptides were sprayed into a LTQ Orbitrap Velos Pro ion-trap mass spectrometer (Thermo Fisher Scientific) equipped with a nano-electrospray ionization source. The mass spectrometer was operated in data-dependent mode with one MS scan followed by 20 collision-induced dissociation for each cycle. Database searches were performed using Sequest (Thermo Fisher Scientific) against the PRDM16 protein sequence using the following search parameters: 10 ppm mass tolerance for precursor ions; 1 Da mass tolerance for product ions. The modification of 114.0429 mass units to lysine was included in the database searches to determine ubiquitin-modified peptides. All databases include a reversed version of all of the sequences, and the data were filtered to 1% or lower peptide FDR. The tandem MS data of matched ubiquitinated peptides were checked manually for their validity.
Immunoprecipitation and immunoblot analyses
Two-step immunoprecipitation and ubiquitination assays were performed as described previously39. In the first-round immunoprecipitation assays, cell lysates were prepared using lysis buffer (50 mM Tris-Cl pH 7.4, 150 mM NaCl, 1% Triton X-100 and 1 mM EDTA) supplemented with a protease inhibitor cocktail (Roche), and mixed with the anti-Flag M2 affinity gel for 2 h. The agarose gels were washed extensively using the same buffer. In the second-round immunoprecipitation assays, the bound proteins on the agarose were denatured by boiling for 5 min in the lysis buffer containing 1% SDS. The solution was diluted 1:10 using lysis buffer. The diluted elutes were re-immunoprecipitated with the anti-Flag M2 affinity gel. After four washes, the bound proteins were separated by SDS–PAGE and analysed using immunoblotting.
A published protocol was used in our study40. In brief, differentiated adipocytes or HEK293T cells transfected with the indicated plasmids were treated with cycloheximide at a final concentration of 20 µg ml−1 and 10 µg ml−1, respectively. The cells were collected at the indicated time points. Cell lysates were analysed by immunoblotting for PRDM16. The intensity of PRDM16-specific protein expression was quantified by Image J software and normalized to that of β-actin signals. For the statistical analyses, two-tailed unpaired Student’s t-tests were performed on the basis of n = 3 biologically independent samples.
Various forms of GST-tagged PRDM16 protein were expressed in E. Coli and purified as previously described4. The GST–PRDM16 protein fragments were incubated with pre-equilibrated Glutathione-Sepharose beads (GE Healthcare) for 2 h, followed by extensive washing. The preloaded GST resins were incubated with Myc-tagged APPBP2 protein for 2 h at 4 °C. Precipitates were washed four times and separated by SDS–PAGE and then analysed using immunoblotting.
OCR was measured using the Seahorse XFe Extracellular Flux Analyzer (Agilent) in a 24-well plate or 96-well plate. Differentiated inguinal-WAT-derived adipocytes were seeded and differentiated for 4 or 5 days. For the measurement of noradrenaline-induced respiration, differentiated adipocytes were stimulated with 1 µM noradrenaline. For the measurement of uncoupled respiration in a 24-well plate, cells were treated with 5 µM oligomycin, followed by phenylhydrazone (5 µM) and antimycin (5 µM). For the measurement of uncoupled respiration in a 96-well plate, cells were treated with 1 µM oligomycin, followed by phenylhydrazone (3 µM) and antimycin (0.5 µM). For the measurement of noradrenaline-induced tissue OCR, adipose tissues (0.5 mg for BAT, 1.5 mg for inguinal and 2.5 mg for epididymal WAT) were placed into XF24 Islet Capture Microplates and stimulated with 10 µM noradrenaline.
RNA-seq and analyses
For RNA-seq analysis of inguinal cells with stable Cul2 knockdown or scramble control, total RNA was isolated using the RNeasy Micro Kit (Qiagen). High-throughput sequencing was performed using the HiSeq 3000 instrument (Illumina) at the Technology Center for Genomics & Bioinformatics at UCLA. The reads were mapped to the latest UCSC transcript set using Bowtie2 v.2.1.0 and the gene expression level was estimated using RSEM (v.1.2.15)41. Trimmed mean of M-values (from edgeR) were used to normalize the gene expression. Gene Ontology analysis was performed using Enrichr42. RNA-seq and library construction were conducted by technical staff at the UCLA genome core who were blinded to the experimental groups. For RNA-seq analysis of Prdm16-KO mouse-derived inguinal adipocytes, total RNA was isolated using the Zymo Direct-zol RNA preparation kit (R2052, Zymo). Extracted RNA (400 ng) was treated with the NEBNext rRNA Depletion Kit v2 (E7400X) to deplete ribosomal RNA and then converted into double-stranded cDNA using the NEBNext mRNA Second Strand Synthesis Module (E6111L). cDNA was analysed using Qubit and BioAnalyzer and subsequently amplified for 12 cycles using the Nextera XT DNA Library Preparation Kit (Illumina FC-131). Generated libraries were analysed by Qubit and Agilent Bioanalyzer, pooled at a final concentration of 1.35 pM, and sequenced on the NextSeq 500 system. Sequencing reads were demultiplexed and trimmed for adapters using bcl2fastq (v.2.20.0). Secondary adapter trimming, NextSeq/Poly(G) tail trimming and read filtering were performed using fastp (v.0.20.1); low-quality reads and reads shorter than 24 nucleotides after trimming were removed from the read pool. Salmon (v.1.4.0)43 was used to simultaneously map and quantify reads to transcripts in the GENCODE M24 genome annotation of GRCm38/mm10 mouse assembly. Salmon was run using full selective alignment, with sequence-specific and fragment GC-bias correction turned on (the –seqBias and –gcBias options, respectively). Transcript abundances were collated and summarized to gene abundances using the tximport package for R44. Normalization and differential expression analysis were performed using edgeR. For differential gene expression analysis, genes were considered to be significant if they passed an FDR cut-off of FDR ≤ 0.05. The heat map of the RNA-seq transcriptome was generated using MetaboAnalyst (v.5.0)45.
ChIP assays were performed according to the established commercial protocol using the Thermo Fisher Scientific Pierce Magnetic ChIP Kit (26157, Thermo Fisher Scientific). In brief, differentiated Appbp2-KO adipocytes and control adipocytes were fixed in 1% formaldehyde for 10 min by gently swirling the dish and quenched with 1× glycine for 5 min at room temperature. The samples were washed twice with ice-cold PBS supplement with Halt Cocktail, collected and then placed into membrane extraction buffer containing protease/phosphatase inhibitors. After centrifugation at 9,000g for 3 min, the supernatant was removed. The nuclei were digested with MNase (ChIP grade) in MNase Digestion Buffer Working Solution at 37 °C for 15 min, followed by sonication to break the nuclear membrane on ice. Digested chromatin was centrifuged at 9,000g for 5 min, and antibodies were added for overnight incubation at 4 °C on a rotating platform. ChIP grade protein A/G magnetic beads were added for 2 h at 4 °C. The samples were washed with the IP wash buffer provided with the kit. The samples were eluted with 1× IP elution buffer at 65 °C for 30 min by vigorous shaking. The samples were subsequently treated with proteinase K followed by column-purification to recover the DNA. Target enrichment was calculated as the percentage of input. The target loci of PRDM16 were chosen on the basis of the previous study that performed ChIP–seq of PRDM16 in brown adipocytes46. A list of the primer sequences is provided in Supplementary Table 1.
Human SNP analyses
The metabolic traits of the APPBP2 genetic variant (rs34146848) were obtained from the FinnMetSeq exome sequence data33. The analysis also can be found in the type 2 diabetes knowledge portal (https://t2d.hugeamp.org/) and Pheweb (http://pheweb.sph.umich.edu/FinMetSeq/variant/17:58525018-C-T). Population frequencies of the APPBP2 genetic variant (rs34146848) can be found in the gnomAD browser (https://gnomad.broadinstitute.org/variant/17-58525018-C-T?dataset=gnomad_r2_1)34 in which the SNP is more frequently found in African and African American individuals (17,099 out of 41,342 alleles, for a frequency of 41.4%) compared with in other ethnic groups. Accordingly, we independently tested for the association between genetic variants at the APPBP2 gene locus and measures of obesity (body mass index, waist–ratio and waist–hip ratio adjusted for BMI) in individuals of African ancestry (n = 7,447) in the UK Biobank35 using the BOLT linear mixed model GWAS software. Phenotypes were adjusted for age, age squared, sex, genotyping array and genetic principal components, followed by inverse normal transformation.
Glucose homeostasis in mice
Male Adipo-Cul2-KO, Adipo-Appbp2-KO and the respective littermate control mice in the C57BL/6J background at 6 weeks old were fed on an HFD (60% fat, D12492, Research Diets) at 22 °C. Body weight was measured every week. The fat mass and lean mass of mice were measured on an HFD for 8 weeks using the Body Composition Analyzer EchoMRI (Echo Medical Systems) system. For glucose-tolerance tests, mice on an HFD for 3 weeks or 9 weeks and fasted for 6 h from 09:00 to 15:00 were administered glucose intraperitoneally (1.5 g kg−1 body weight). For the insulin-tolerance tests, mice on an HFD for 10 weeks and fasted for 3 h from 09:00 to 12:00 were injected intraperitoneally with insulin (1 U kg−1 body weight). For the pyruvate-tolerance tests, mice on an HFD for 11 weeks and fasted for 16 h were injected intraperitoneally with pyruvate (1 g kg−1 body-weight). Blood samples were collected at the indicated time points before and after injection, and glucose levels were measured using blood glucose test strips (Freestyle Lite).
Energy expenditure in mice
Whole-body energy expenditure (VO2, VCO2), food intake and locomotor activity (beam break counts) of Adipo-Cul2-KO mice and littermate control mice were monitored using the Comprehensive Laboratory Animal Monitoring System (CLAMS, Columbus Instruments) after 3 weeks of HFD. For the analyses of Adipo-Appbp2-KO mice, the whole-body metabolic rate was measured using the Promethion Metabolic Cage System (Sable Systems) at 30 °C. Adipo-Appbp2-KO and their littermate control mice were acclimatized to 30 °C for 3 days before transferring to metabolic cages. During the measurement of energy expenditure, mice received a single intraperitoneal injection of CL-316,243 (Sigma-Aldrich; 0.1 mg per kg body weight). Obtained indirect calorimetry data were analysed by CaIR-ANCOVA (https://calrapp.org/), a regression-based analysis of energy expenditure in mice47.
Fatty acid oxidation assay
Fatty acid oxidation assays were performed according to the protocol described by our previous work48. In brief, BAT, inguinal WAT and gastrocnemius muscle tissues were isolated from Adipo-Cul2-KO and control mice after exposure to 8 °C, or Adipo-Appbp2-KO on an HFD. The tissues were minced to small pieces, placed into a polypropylene round-bottom tube and then incubated in the 1 ml KRB-HEPES buffer containing 0.5 μCi ml−1 [1-14C]oleic acid at 37 °C at 60 rpm for 1 h. After adding 350 μl 30% hydrogen peroxide into the reaction mixture, [14C]CO2 was trapped in the centre well supplemented with 300 μl of 1 M benzethonium hydroxide solution for 20 min at room temperature. 14C radioactivity was measured using a liquid scintillation counter and normalized to tissue mass.
For the measurement of liver triglyceride contents, liver tissues from Adipo-Cul2-KO or Adipo-Appbp2-KO mice were collected and homogenized in 350 ml ethanolic KOH (100% ethanol and 30% KOH at a ratio of 2:1) and incubated overnight at 55 °C. Subsequently, tissue lysates were supplemented with 50% ethanol to 1 ml final volume. After centrifugation, the supernatant was mixed with 1 M MgCl2 and incubated on ice for 10 min. The amounts of triglycerides were measured using the Infinity Triglycerides kit (Thermo Fisher Scientific). Serum cholesterol and serum triglyceride measurement were performed by the Longwood Small Animal Imaging Facility at BIDMC.
Peripheral insulin signalling in vivo
Adipo-Appbp2-KO mice and their littermate controls at 4 weeks of HFD were fasted for 4 h followed by intraperitoneal injection of insulin at 1.3 U kg−1 body weight. Liver, EpiWAT and inguinal WAT were removed 10 min after the injection and lysed in RIPA lysis buffer, supplemented with protease and phosphatase inhibitor cocktails. The lysates were separated by SDS–PAGE and analysed using immunoblotting. PhosphoPlus Akt (Ser473) Antibody Duet was used for western blot analysis.
Adipo-Cul2-KO mice, Adipo-Appbp2-KO mice and their respective littermate control mice were kept on a regular chow diet. Mice were acclimatized to 30 °C for 11 days and subsequently exposed to 8 °C for 6 h. The rectal temperatures of mice were monitored every 1 h using the TH-5 thermometer (Physitemp).
Adipose tissues and liver were fixed in 4% paraformaldehyde overnight at 4 °C, followed by dehydration in 70% ethanol. After the dehydration procedure, tissues were embedded in paraffin and cut into sections at a thickness of 5 μm. The sections were processed for haematoxylin and eosin staining according to the standard protocol at the BIDMC pathology core. Images were acquired using the Revolve microscope (ECHO Laboratories).
Total RNA was prepared from cells using TRIzol reagents (Invitrogen) according to the manufacturer’s instructions. Total RNA extracted from tissues was obtained using TRIzol reagents plus RNeasy Mini Kit (Qiagen). RNA samples were reverse-transcribed using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories) according to the provided protocol. The quantifications of gene transcripts were performed by qPCR using the ABI ViiA 7 PCR or QS6 cycler (Applied Biosystems). 36B4 or TBP served as an internal control. A list of the PCR primers used to amplify the target genes is provided in Supplementary Table 1.
Quantification of the mtDNA copy number
Total DNA was isolated from mature adipocytes with the SpeeDNA Isolation Kit (MB6918, ScienCell) according to the manufacturer’s instructions. DNA concentrations were measured using the Nanodrop 2000 (Thermo Fisher Scientific) and diluted to final concentrations of 20 ng ml−1 with double-distilled H2O. The mtDNA copy number was amplified using primers specific for the mitochondrial Cox1, Cox2, Cox3, Atp6 and Atp8 genes and normalized to genomic DNA by amplification of the β-globin gene. A list of the primer sequences is provided in Supplementary Table 1.
Lipid staining by Oil Red O
Cells were washed once with PBS, fixed in 4% paraformaldehyde for 15 min and then stained with Oil-Red-O solution for 10–20 min at ambient temperature. Subsequently, cells were washed three times with PBS followed by imaging with a Revolve microscope (ECHO Laboratories).
Immortalized inguinal preadipocytes were differentiated for 5 days cultured on 12-well plates. Cells were fixed for 2 h at room temperature with fixative solution (2.5% glutaraldehyde, 1.25% paraformaldehyde, 0.03% picric acid in 0.1 M sodium cacodylate buffer, pH 7.4), washed in 0.1 M cacodylate buffer and post-fixed with 1% osmiumtetroxide (OsO4)/1.5% potassium ferrocyanide (KFeCN6) for 1 h. Samples were washed in water twice, 1× maleate buffer (MB) one time, and incubated in 1% uranyl acetate in MB for 1 h followed by two washes in water and subsequent dehydration in grades of alcohol (10 min each at 50%, 70% and 90%, and twice for 10 min at 100%). After dehydration, propyleneoxide was added to the dish and the cells were lifted off using a transfer pipet, pelleted and infiltrated overnight in a 1:1 mixture of propyleneoxide and TAAB Epon (TAAB Laboratories Equipment; https://taab.co.uk). The samples were then embedded in TAAB Epon and polymerized at 60 °C for 48 h. Ultrathin sections (about 60 nm) were cut on the Reichert Ultracut-S microtome, picked up onto copper grids stained with lead citrate and examined on the JEOL 1200EX transmission electron microscope or the TecnaiG2 Spirit BioTWIN system and images were recorded with the AMT 2k CCD camera.
Statistics and reproducibility
All the biological experiments were repeated at least twice and reproduced. RNA-seq was performed once but three independent samples were analysed and further validated using alternative approaches, such as RT–qPCR. Western blotting data were confirmed by two or three independent samples. The presented data were collected from biologically independent samples. Statistical analyses were performed using GraphPad Prism v.7.0 (GraphPad). All data are represented as mean ± s.e.m. unless otherwise specified. Unpaired Student’s t-tests were used for two-group comparisons. One-way ANOVA followed by the Dunnett’s test was used for multiple-group comparisons. Two-way ANOVA was used for Seahorse measurements from multiple groups. Two-way repeated-measures ANOVA followed by Fisher’s LSD test was applied to determine the statistical differences in body-weight gain, whole-body energy expenditure results, glucose-tolerance tests, insulin-tolerance tests and pyruvate-tolerance tests between genotypes. The statistical parameters and mouse numbers used per experiment are specified in the figure legends. No statistical methods were used to predetermine sample size. P < 0.05 was considered to be significant throughout the study.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.