Abstract Brown adipose tissue (BAT) is a major site of non-shivering thermogenesis in mammals and plays an important role in energy homeostasis. Nuclear factor-erythroid 2-related factor 1 (NFE2L1, also known as Nrf1), a master regulator of cellular metabolic homeostasis and numerous stress responses, has been found to function as a critical driver in BAT thermogenic adaption to cold or obesity by providing proteometabolic quality control. Our recent studies using adipocyte-specific Nfe2l1 knockout [Nfe2l1(f)-KO] mice demonstrated that NFE2L1-dependent transcription of lipolytic genes is crucial for white adipose tissue (WAT) homeostasis and plasticity. In the present study, we found that Nfe2l1(f)-KO mice develop an age-dependent whitening and shrinking of BAT, with signatures of down-regulation of proteasome, impaired mitochondrial function, reduced thermogenesis, pro-inflammation, and elevated regulatory cell death (RCD). Mechanistic studies revealed that deficiency of Nfe2l1 in brown adipocytes (BAC) primarily results in down-regulation of lipolytic genes, which decelerates lipolysis, making BAC unable to fuel thermogenesis. These changes lead to BAC hypertrophy, inflammation-associated RCD, and consequently cold intolerance. Single-nucleus RNA-sequencing of BAT reveals that deficiency of Nfe2l1 induces significant transcriptomic changes leading to aberrant expression of a variety of genes involved in lipid metabolism, proteasome, mitochondrial stress, inflammatory responses, and inflammation-related RCD in distinct subpopulations of BAC. Taken together, our study demonstrated that NFE2L1 serves as a vital transcriptional regulator that controls the lipid metabolic homeostasis in BAC, which in turn determines the metabolic dynamics, cellular heterogeneity and subsequently cell fates in BAT. Keywords: NFE2L1, brown adipocyte, BAT whitening, Lipolysis, snRNA-seq Highlights * • Adipocyte-specific knockout of Nfe2l1 results in cold intolerance in mice. * • Nfe2l1(f)-KO mice display an age-dependent BAT whitening and shrinking. * • Nfe2l1(f)-KO mice show severe inflammation and elevated RCD in BAT. * • Aberrant lipid metabolism is involved in BAT inflammation in Nfe2l1(f)-KO mice. * • NFE2L1 determines the metabolic dynamics and cellular heterogeneity of BAC. Abbreviations: Adgre1 adhesion G protein-coupled receptor E1 Adipoq adiponectin ANOVA analysis of variance APP antigen processing and presentation AT adipose tissues Atgl/Pnpla2 adipocyte triglyceride lipase/patatin-like phospholipase domain containing 2 BAC brown adipocyte BAT brown adipose tissue Casp1 caspase 1 CNC-bZIP cap ‘n’ collar-basic leucine zipper protein CO[2] carbon dioxide COX cytochrome c oxidase Cpt1b carnitine palmitoyltransferase 1 B DEGs differentially expressed genes Elovl ELOVL fatty acid elongase ER endoplasmic reticulum FFAs free fatty acids Fis1 mitochondrial fission protein 1 GPR3 G protein-coupled receptor 3 GSEA gene set enrichment analysis gWAT gonadal WAT H&E hematoxylin and eosin Hsl hormone-sensitive lipase IHC immunohistochemistry Ifng interferon γ Il1b interleukin 1β iWAT inguinal WAT KEGG Kyoto encyclopedia of genes and genomes LDs lipid droplets MAC macrophages Mgl monoacylglycerol lipase NC nerve cells NK/T natural killer/T cells Ndufs4 NADH dehydrogenase (ubiquinone) Fe–S protein 4 NFE2L1/NRF1 nuclear factor erythroid 2-related factor 1 Nfe2l1(f)-KO adipocyte-specific Nfe2l1 knockout Nlrp3 NLR family pyrin domain containing 3 NST non-shivering thermogenesis OCR oxygen consumption rate Opa1 optic atrophy 1 mitochondrial dynamin like GTPase ORO oil red-O OXPHOS oxidative phosphorylation Pgc-1α peroxisome proliferator-activated receptor gamma coactivator-1 α PKA protein kinase A Plin1 perilipin 1 PPAR peroxisome proliferator-activated receptor Prdm16 PR domain containing 16 Pycard apoptosis-associated speck-like protein containing a CARD RCD regulatory cell death RT-qPCR reverse transcription-quantitative polymerase chain reaction snRNA-seq single-nucleus RNA sequencing SNP single nucleotide polymorphism TAG triacylglycerol/triglyceride TEM transmission electron microscopy Ucp1 uncoupling protein 1 VEC vascular endothelial cells WAT white adipose tissue WAC white adipocyte 1. Introduction The epidemic outburst of obesity and related metabolic disorders has directed research attention to the physiology of distinct adipocytes and fat depots. The adipose tissue (AT) is not only a depot for storing excessive energy but also a crucial metabolic sensor and endocrine organ coordinating a wide range of physiological processes [[51][1], [52][2], [53][3]]. In contrast to white adipocytes (WAC), which store energy as triglycerides (TAG) in unilocular lipid droplets, brown adipocytes (BAC) have highly enriched mitochondria and multilocular lipid droplets. BAC utilizes free fatty acids (FFAs) and glucose at a high rate mainly mediated by uncoupling protein 1 (UCP1) which uncouples the energy of mitochondrial proton gradient from ATP synthesis, dissipating the energy instead in the form of heat. BAC thermogenesis, also referred to as non-shivering thermogenesis (NST), is a dynamic and precisely controlled process, the failure of which may bring devastating consequences to the body [[54]4]. Thus, the function and plasticity of BAT are crucial for the human body to adapt to environmental cues. Dysfunctional BAT is implicated in the pathogenesis of various metabolic disorders, including obesity and diabetes [[55]5]. While BAT has long been recognized as a highly homogeneous population of BAC, cellular heterogeneity of BAT was recently revealed by single-cell/nucleus RNA-sequencing (sc/nRNA-seq) technique showing that distinct subpopulations of BAC exist in the tissue [[56]6,[57]7]. The cap ‘n’ collar-basic leucine zipper protein (CNC-bZIP) nuclear factor erythroid 2-related factor 1 (NFE2L1, also known as NRF1) is a master regulator controlling the transcription of a suite of genes involved in proteasomal homeostasis, cell metabolism and numerous stress responses, antioxidant response in particular [[58][8], [59][9], [60][10], [61][11]]. A single nucleotide polymorphism (SNP), rs3764400, in the 5′-flanking region of the human NFE2L1 gene, appears to be associated with obesity in humans [[62][12], [63][13], [64][14], [65][15]]. In addition, an epigenome-wide association study identified human NFE2L1 as one of the genes nearest to the sentinel methylation markers of a locus linked to BMI and adverse outcomes of obesity [[66]15,[67]16]. In line with the findings that NFE2L1 is potentially involved in the pathogenesis of obesity, BAC-specific deletion of Nfe2l1 in mice resulted in endoplasmic reticulum (ER) stress, tissue inflammation, markedly diminished mitochondrial function and whitening of the BAT, highlighting NFE2L1 as a critical driver in proteasomal adaption under thermogenic conditions [[68]9,[69]17]. Our recent studies demonstrated that adipocyte-specific Nfe2l1 knockout [Nfe2l1(f)-KO] mice, in which the Nfe2l1 gene was disrupted specifically in the adiponectin-expressing adipocytes, exhibit a dramatically reduced subcutaneous adipose tissue mass, adipocyte hypertrophy, and severe adipose inflammation mediated via disturbed expression of lipolytic genes in adipocytes [[70][18], [71][19], [72][20]]. In addition, NFE2L1 regulates adipogenesis in an isoform-specific manner. Specifically, the long isoforms of NFE2L1 negatively regulate the transcription of peroxisome proliferator-activated receptor γ (Pparγ), Pparγ2 in particular, and thereby suppress adipogenesis [[73]21]. Given that NFE2L1 controls WAC plasticity by regulating the transcription of multiple lipolytic genes and Pparγ, which are also crucial to the function and homeostasis of BAC, we speculate that Nfe2l1 deficiency may cause downregulation of lipolytic genes and induction of Pparγ in BAC, and thus results in lipid overload leading to inflammatory response and other consequences in BAT. Although the crucial roles of NFE2L1 in adipocyte biology have been investigated by using two distinct lines of mouse models with BAC or mature adipocyte-specific knockout of Nfe2l1, the effect of Nfe2l1 loss on the cellular heterogeneity and metabolic dynamics of BAT, BAC subpopulations in particular, has not been investigated at the single-cell level. Such studies would be helpful to understand the molecular mechanisms of adaptive NST response and cellular dynamics of BAT. In the present study, we found that deletion of Nfe2l1, mediated by adiponectin promoter-driven Cre recombinase in mice, induces an age-dependent whitening and shrinking of BAT and cold intolerance. The mRNA profiles of BAT showed signatures of down-regulated proteasome subunits, mitochondrial dysfunction, pro-inflammatory response, and augmented inflammation-related regulatory cell death (RCD). Mechanistic studies revealed that deficiency of Nfe2l1 in BAC primarily decreases the expression of multiple lipolytic genes and thus disturbs lipid metabolism leading to BAC hypertrophy and inflammatory responses in BAT. snRNA-seq analysis of BAT reveals that NFE2L1 is a key factor determining the cellular heterogeneity and dynamics and thus homeostasis of BAC. 2. Materials and methods 2.1. Animals and treatments Nfe2l1(f)-KO mice were produced by crossing the mice carrying Nfe2l1^flox allele [[74]22] and Cre recombinase gene driven by the promoter of adiponectin gene (Adipoq-Cre, 010,803, Jackson Labs Technologies, Inc., Sacramento, CA) as described previously [[75]18]. Nfe2l1(f)-KO mice carrying the genotype of Nfe2l1^flox/flox;Adipoq-Cre^+ and their littermates with the genotype of Nfe2l1^flox/flox;Adipoq-Cre^- (Flox) were used in the current study. Genotyping was performed using PCR on genomic DNA isolated from tail snips. The sequence of genotyping primers is given in [76]Supplementary Table S1. Mice were housed on a 12 h/12 h light/dark cycle, 24 °C and 50% humidity. Mice were given distilled water and a standard laboratory chow diet (Cat# SWS9102, Jiangsu Xietong Organism Co, Ltd, Nanjing, China) ad libtum unless otherwise specified. For acute cold exposure, mice fasted for 12 h were placed in a cage maintained at 4 ± 1 °C for 4 h. For chronic cold adaptation (8 °C for 14 days), mice were housed in cages with ad libitum access to a chow diet and water with a 12 h light/dark cycle. In the intervention studies, mice were administered with [77]CL316243 (1 mg/kg/d, C5976, Sigma) by intraperitoneal injection for 7 days, followed by metabolic measurements and tissue collection. Mice were euthanized and various relevant tissues were collected and weighed after relevant treatments. Tissue samples were stored at –80 °C until subsequent determination. Group size was 7–8 mice per group. All animal procedures followed the ARRIVE guidelines and were approved by the Animal Ethics Committee of China Medical University (#1408 M). 2.2. Measurements of core temperature and energy metabolism in mice Rectal body core temperature was measured using a probe attached to a digital thermometer (BAT-10 thermometer, Physitemp, Clifton, NJ, USA). An 8-channel PromethION etabolic system (Sable Systems International, Las Vegas, Nevada, USA) was used to monitor energy expenditure and respiration in mice. Mice were housed at 24 °C or 4 °C in the PromethION metabolic cages (1 mouse per cage) placed in a temperature-controlled cabinet. The basic metabolic indicators, including diet consumption, O[2] consumption and CO[2] production, were monitored in real-time according to the manufacturer's instructions. 2.3. Histological examination The interscapular BAT sections were stained with hematoxylin and eosin (H&E) and immunohistochemistry (IHC) as described previously [[78]18,[79]19]. In brief, BAT was dissected and subjected to fixation in 4% paraformaldehyde for at least 24 h. Then, tissues were dehydrated step by step with ethanol and embedded in paraffin. As the liquid paraffin solidified, serial sections (4 μm) were obtained via the HistoCore AUTOCUT R-Automated Rotary Microtome (Leica, Nussloch, Germany) followed by H&E staining. For IHC staining, the primary antibody against F4/80 (sc-25830, Santa Cruz Biotechnology Inc., CA, USA) was used and followed by DAB staining (ZLI-9019, Zhongshan Golden Bridge BioTech, Co., Ltd., Beijing, China). Regarding oil red-O (ORO) staining, the BAT flash-frozen with liquid N[2] immediately after collection were embedded in O·C.T compound and sectioned into 10-μm-thick slices. ORO stain (0.3% w/v ORO in 60% v/v isopropanol) was added and incubated for 40 min at room temperature [[80]23]. Imaging of pathological changes was performed using an optical microscope (200x, DMi8, Leica). 2.4. Transmission electron microscopy (TEM) Tissue pieces (2 mm × 2 mm × 2 mm) isolated from the centric part of interscapular BAT were fixed in 2.5% glutaraldehyde-0.1 M cacodylate buffer at 4 °C for 4 h at least, followed by washing with 0.1 M cacodylate buffer for 15 min twice. Then the samples were further fixed with 1% (w/v) osmium tetroxide (OsO[4]) at 4 °C for 2 h, cleared in 0.1 M cacodylate buffer twice, and serially dehydrated with 50, 70, 80, 90, and 95% acetone for 15 min each followed by 3 times of 100% acetone dehydration. The resulting specimens were preserved in a mixture of epoxy resins and acrylic overnight, then embedded with beam capsules filled with 100% resin and aggregated at 60 °C for 48 h. Sections (60 nm) prepared using an ultramicrotome (Leica EM UC6, Germany) were examined by TEM (JEM-1400 Flash, Japan) and photographed at different magnifications ranging from 2000 to 20,000 × . 2.5. Bulk RNA-sequencing and data analysis RNA isolation, purification, quality control, and sequencing library preparation and data pre-analysis were conducted as detailed in Supplementary Materials and Methods, Section [81]2.5. Genes differentially expressed between groups were identified using the DESeq2 package (version 1.36.0). A p-value cutoff of 0.05 and fold-change cutoff of 2 were used for statistical significance of gene expression changes. Gene set enrichment analysis (GSEA) was completed using clusterProfiler package (version 4.4.1), using pathways from Kyoto Encyclopedia of Genes and Genomes (KEGG) database. 2.6. RNA isolation and RT-qPCR RNA isolation and RT-qPCR were performed as described previously [[82]24]. Briefly, according to manufacturer's instructions, total RNA of BAT was extracted using TissueLyser II with TRIzol (#15596, Invitrogen Corporation, Carlsbad, CA, USA), subsequently precipitated with isopropanol and subjected to clean-up using 75% ethanol. RNA concentrations were measured using the NanoDrop 2000c Spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). Reverse transcription was conducted using a commercial kit (#RR037A, Takara Biomedical Technology, Co., Ltd., Dalian, China). SYBR Premix Ex Taq (#RR420A, Takara Biomedical Technology, Co., Ltd., Dalian, China) was used for qPCR. Amplification of cDNA and fluorescence measurement was performed using an ABI Q6 Flex Real-time PCR System (Applied Biosystems Inc., Foster City, CA). Each tissue sample was measured in triplicate and resulting average was used for further statistical analysis. Primers were designed using Primer Express 4 and Primer Blast ([83]http://www.ncbi.nlm.nih.gov/tools/primer-blast/) software and synthesized by Life Technologies (Shanghai, China). The sequences are shown in [84]Table S2. The expression level of 36B4 was used as the loading control and the relative expression of genes is shown as fold-change over control (2-ΔΔCt). 2.7. Protein extraction and western blotting Water soluble proteins were isolated by pulverizing frozen adipose tissues in a lysis buffer purchased from Cell Signaling Technology Inc. (#9803, Danvers, MA, USA) containing both proteinase and phosphatase inhibitors (P8340, P5726 and P0044, Sigma). Protein concentrations were determined using a BCA assay kit (#P0010, Beyotime Biotechnology, Suzhou, China). Immunoblot analysis was performed as previously described [[85]20]. Equal amounts of protein were resolved using SDS-PAGE gels. The following antibodies were used for detecting specific proteins: UCP1 (#ab23841, 1:1000), PGC-1α (#ab54481, 1:1000), OXPHOS (#ab110413, 1:1000), FIS1 (#ab71498, 1:1000) and OPA1 (#ab42364, 1:1000) were purchased from Abcam Inc. (Cambridge, MA, USA); COX IV (#4844, 1:1000), PLIN1 (#9349, 1:1000) and ATGL (2138, 1:1000) were from Cell Signaling Technology; NDUFS4 ([86]O43181, 1:1000) was from Thermo Fisher Scientific; HSL (sc-25843, 1:1000), MGL (sc-398,942, 1:1000) and β-ACTIN (sc-47778, 1:3000) were obtained from Santa Cruz Biotechnology. Quantification of immunoreactive bands was carried out with the Image J software (National Institutes of Health, USA). 2.8. Assay of lipolytic activity BAT minced at around 2 mm in diameter were incubated with a glycerol-free basic medium containing 10 μM [87]CL316243 (C5976, Sigma). Following a 2-h incubation at 37 °C and 5% CO[2], the resulting media were collected and applied to determine the content of glycerol with a commercial kit (K622, BioVision Inc., Milpitas, CA). Glycerol release was calculated according to the instructions and normalized by tissue weight or protein content. 2.9. snRNA-seq The mixed BAT samples from 3 mice of each genotype (4 weeks old, female) were rinsed with pre-cooled RNase-free saline, cut into small pieces on ice and stored at −80 °C. Nuclear suspensions were loaded on a 10x Genomics GemCode Single-cell instrument that generates single-cell Gel Bead-In-EMlusion (GEMs). Libraries were generated and sequenced from the cDNAs with Chromium Next GEM Single Cell 3′ Reagent Kits v3.1. Silane magnetic beads were used to remove leftover biochemical reagents and primers from the post-GEM reaction mixture. Full-length, barcoded cDNAs were then amplified by PCR to generate sufficient mass for library construction. R1 (read 1 primer sequence) was added to the molecules during GEM incubation. P5 and P7, the sample indexes, and R2 (read 2 primer sequence) were added during library construction via End Repair, Atailing, Adaptor Ligation, and PCR. The final libraries contained the P5 and P7 primers used in Illumina bridge amplification. The Single Cell 3′ Protocol produced Illumina-ready sequencing libraries. A Single Cell 3′ Library comprised standard Illumina paired-end constructs which begin and end with P5 and P7. The Single Cell 3’ 16 bp 10x Barcode and 10 bp UMI were encoded in Read 1, while Read 2 was used to sequence the cDNA fragment. Sample index sequences were incorporated as the i7 index read. Read 1 and Read 2 were standard Illumina® sequencing primer sites used in paired-end sequencing. 2.10. snRNA-seq data analysis The count matrix generation, filtering, normalization, clustering and t-SNE generation were conducted as detailed in Supplementary Materials and Methods, Section [88]2.10. The enrichment analysis, gene set score, and trajectory analysis were performed as follows. 2.10.1. Enrichment analysis The differentially expressed genes (DEGs) were further divided into up- and down-regulated groups for KEGG pathway enrichment analysis by using clusterProfiler package and pathways with FDR <0.05 were shown. 2.10.2. Gene set score Gene set scores were calculated using the Seurat function AddModuleScore which calculates the average expression of a given signature per cell and subtracts it from the average expression of randomly selected control features. AddModuleScore was run using 500 control features. The proteasome gene set was formed by Psmd1, Psmd11, Psmd14, Psmc6, Psmd12, Psma1, Psmd3, and Psmd4. This list contains genes from the mmu 03050 of KEGG ‘‘Proteasome’’ that were downregulated in the combination of A0, A1, A2 subclusters in Nfe2l1(f)-KO (KO) compared to Nfe2l1^flox/flox (Flox). The AMPK gene set was formed by Pparg, Igf1r, Ppp2r3a, Eef2k, Camkk2, and Prkag2, containing genes from the mmu 04152 of KEGG ‘‘AMPK signaling pathway’’ that were downregulated in the combination of A0, A1, A2 subclusters in KO. The APP gene set was formed by Tap1, Tap2, H2-T22, Tapbp, H2-T23, Psme1, Ctsb, B2m, H2–K1 and necroptosis gene set was formed by Stat1, Stat2, Mlkl, Stat3, Glul, Eif2ak2, Bid, Irf9, Ifngr1, Camk2d, and Fas. These two gene sets are from the mmu 04612 and mmu 04217 of KEGG and they are upregulated in A3 of Flox compared to the combination of A0, A1, and A2 subclusters in Flox. Statistical analysis was performed by a two-sided Wilcoxon rank sum test followed by BH correction. 2.10.3. Trajectory analysis RNA velocity analysis was conducted by using scVelo v.0.2.3. In particular, to count spliced and unspliced reads for each sample, the 10 × velocyto pipeline was run in the filtered cell ranger-generated BAM files, while for single-nucleus RNA velocity inference, the dynamical model of scVelo was applied [[89]25]. 2.11. Statistics All statistical analyses of non-omics data were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA), with p < 0.05 considered as significant. Data were expressed as mean ± standard deviation (SD). For comparisons between the two groups, a two-tailed unpaired Student's t-test was performed. When the difference among three or more groups was evaluated, one-way analysis of variance (ANOVA) or, when appropriate, two-way ANOVA followed by Bonferroni's test was used. Statistical analyses of omics data were all performed in R, and details of the packages used and statistical test methods can be viewed in the corresponding sections. 2.12. Data and code availability Our snRNA-seq and bulk RNA-seq data have been deposited in NCBI Gene Expression Omnibus (GEO) database (accession number: [90]GSE225281). The dataset of low- and high-thermogenic adipocytes can be downloaded from GEO databases with accession number of [91]GSM3567479. The dataset of adipocytes that regulates thermogenesis can be accessed from public database ArrayExpress with accession number E-MTAB-8562. All the code for analysis and plotting in R is available on GitHub ([92]https://github.com/SEVEN1003/BAT). 3. Results 3.1. Thermogenesis is impaired in adult Nfe2l1(f)-KO mice To verify the efficiency and specificity of Adipoq-Cre mediated Nfe2l1 knockout, we first analyzed the mRNA levels of Nfe2l1 in different tissues of Nfe2l1(f)-KO mice. As shown in [93]Fig. 1A, the mRNA levels of Nfe2l1 were significantly decreased in interscapular BAT, gonadal WAT (gWAT) and inguinal WAT (iWAT) of Nfe2l1(f)-KO mice, but not in the tissues with low Adipoq expression, including liver, lung and skeletal muscle. In addition, the protein levels of NFE2L1 in BAT were analyzed by western blotting. As illustrated in [94]Supplementary Fig. S1, multiple protein bands around 95–110, 85, 65, and 42–45 kDa decreased substantially in the KO mice compared to their littermate controls. To investigate whether the loss of Nfe2l1 in adipocytes affects the thermogenic function of adipose tissues, we analyzed the core temperature of adult Nfe2l1(f)-KO mice under different conditions. While the rectal temperature of mice at room temperature showed no significant differences between genotypes with or without fasting, 4 h cold exposure induced a significant reduction in the core body temperature in fasting Nfe2l1(f)-KO mice, but not in Flox controls with the same temperature treatments ([95]Fig. 1B). In agreement with findings with acute cold exposure, the survival rates of male ([96]Fig. 1C) and female ([97]Fig. 1D) Nfe2l1(f)-KO mice following sustained cold exposure for up to 10 days were much lower than those in Flox mice. To ascertain the impaired adaptive thermogenesis in Nfe2l1(f)-KO mice, we monitored O[2] consumption and CO[2] production using metabolic cages. At conventional rearing temperatures (24 °C), the O[2] consumption and CO[2] production of Nfe2l1(f)-KO mice were slightly lower than those of Floxed mice, especially at night ([98]Supplementary Fig. S2). After the mice were challenged with cold (4 °C), Nfe2l1(f)-KO mice displayed a substantially subdued increase of O[2] consumption and CO[2] production compared to Flox mice ([99]Fig. 1E–H). Fig. 1. [100]Fig. 1 [101]Open in a new tab The core temperature, survival rate and energy expenditure of Nfe2l1(f)-KO mice under cold challenge. (A) mRNA levels of Nfe2l1 in the tissues of Nfe2l1(f)-KO (KO) mice relative to their Nfe2l1^flox/flox (Flox) littermates. sMuscle, skeletal muscle; BAT, interscapular brown adipose tissue; gWAT, gonadal white adipose tissue; iWAT, inguinal white adipose tissue. N = 6–8. *p < 0.05 vs the same tissue of Flox mice. (B) The rectal temperature of mice under room tempreture (RT, 24 °C) or following acclimation to cold for 4 h (Cold, 4 °C). The mice were either fasted for 16 h overnight or given ad libtum access to chow diet before the cold exposure. N = 6–8. *p < 0.05 vs Flox. (C and D) The survival rates of male and female mice under sustained cold exposure (8 °C). N = 6–8. *p < 0.05 vs Flox. (E and G) The oxygen consumption rate (E) and CO[2] production rate (G) of male mice in response to acute cold exposure. VO[2], volume of O[2] consumed; VCO[2], volume of CO[2] produced. N = 8. (F and H) The quantitative areas under curves (AUC) of (E) and (G), respectively. *p < 0.05 vs Flox. (For interpretation of the references to color in