The thermogenic characteristics of adipocytes are dependent on the regulation of iron homeostasis

IRP/IRE signaling and iron metabolism in adipose tissue modulate the thermogenic potential

Cellular iron homeostasis is tightly regulated by two cytoplasmic RNA-binding proteins, iron regulatory protein (IRP) 1 and 2 (Fig. 1A). IRPs/IRE regulatory system is likely operated in a cell-type-specific manner to satisfy metabolic needs. Based on the high metabolic rate of brown fat, we first hypothesized that iron demands and IRP/IRE interaction would be higher in brown adipose tissue than white adipose tissue. To investigate IRP/IRE binding, we performed an electrophoretic mobility shift assay (EMSA) using an IRDye700-labeled IRE probe with the cytosolic fractions prepared from epididymal (eWAT), inguinal white adipose tissue (iWAT), and interscapular brown fat (iBAT) that represent visceral, subcutaneous fat and classical brown fat, respectively. The IRP/IRE binding was barely noticeable, and only free IRE was observed in eWAT, while IRP/IRE binding was obvious in iWAT. In contrast to WAT, the IRP/IRE band was intense in the cytosolic fraction of iBAT, indicating the higher demands for iron in BAT than WAT (Fig. 1B). We next examined the expression levels of iron-handling proteins and mitochondrial respiratory components in each depot. In the basal levels, iron-deficiency responses (i.e., IRP2 accumulation, increased TfR1, and reduced FT light chain), were apparent in iBAT, but not in eWAT and iWAT (Fig. 2A, Left), which is consistent with the intense IRP/IRE-binding pattern (Fig. 1A). Notably, the endogenous levels of Fe-S cluster scaffold protein ISCU were distinctively lower in iBAT compared with iWAT or eWAT (Fig. 2A). The expression levels of the respiratory chain components were significantly higher in the iBAT than eWAT or iWAT, which is consistent with more mitochondrial OxPhos proteins in iBAT than WAT (7). Reflecting the increased TfR1 protein expression in iBAT, the total iron content in iBAT was approximately threefold higher than in eWAT or iWAT (Fig. 2B).

Next, we questioned whether the environmental stresses that alter the thermogenic potential could also modulate the intracellular iron content. To test this, we administrated the C57BL/6 mice with β3-adrenergic agonist CL 316,243 (CL) for 5 days to induce adipose tissue browning, and depot-specific adipose tissues were collected. In agreement with the fact that eWAT is less susceptible to adipose browning than iWAT, CL treatment did not affect the expression levels of UCP1 or IRP2 in eWAT. Importantly, enhanced browning in iWAT was accompanied by a significant increase of iron-starvation responses, including stabilized IRP2, increased TfR1, and decreased FTL and ISCU expression along with increased UCP1 expression (Fig. 2A, Right). Consistently, the iron content in iWAT was increased in response to CL treatment (Fig. 2C), with the rank order of iron content as eWAT < iWAT < iWAT_CL < iBAT. The increase of iron content in response to CL treatment was consistent with increased mitochondrial respiratory chain proteins (Fig. 2A, Right) and mitochondrial DNA content (Fig. 2D). Conversely, when the mice were kept in thermoneutral conditions (30 °C) for 2 weeks to dampen the thermogenesis, the iron content of adipose tissue was significantly reduced compared with iron levels at ambient temperature (22 °C) in all three adipose depots (p < 0.001) (Fig. 2B). Within the same temperature setting, iBAT maintained the highest iron content when compared with eWAT and iWAT (p < 0.001) (Fig. 2B). The decrease of UCP1 levels of iBAT at 30 °C induced an increase of FTL, suggesting an iron deposition into the cytoplasm (Fig. 2E). Intriguingly, the reduction of IRP2 and TfR1 was not readily detectable in thermoneutral iBAT (Fig. 2E). Instead, we identified 1) upregulation of ER stress proteins, including CHOP, p-eIF2a, p-JNK, and p-38 (Fig. 2F); and 2) augmented autophagy, including the upregulation of autophagosome machinery (Beclin 1 and ATG5) and LC3II formation (Fig. 2G). These data imply that thermoneutral conditions induce ER stress in the iBAT and autophagic clearance of unnecessary mitochondria. Further, the iron released from the mitochondria seems to be sequestered into ferritin to avoid iron toxicity and exported out of iBAT.

Taken together, the changes in the thermogenic capacity of adipose tissues are correlated with the control of the IRP/IRE signaling and iron homeostasis.

Differential regulation of IRPs during white and brown adipocyte differentiation

Next, we questioned whether the regulation of IRP/IRE signaling and iron levels are the cell-autonomous mechanism that dictates white versus brown adipocyte differentiation. To address this question, 3T3-L1 and H1B1B cells, murine white and brown preadipocytes, respectively, were induced to differentiate (Fig. 3A), and we analyzed the kinetic changes of the proteins (Fig. 3) and mRNA (Fig. 4) related to iron regulation and mitochondrial biogenesis. Both 3T3-L1 and HIB1B cells were differentiated into lipid-laden mature adipocytes after 10 days of differentiation, showing the differentiation efficiency of ∼90% in both adipocytes (Fig. S1A). The expression of peroxisome proliferator-activated receptor-gamma (PPARγ), a key regulator in the complex process of adipocyte differentiation, was significantly increased over the differentiation process in both types of adipocytes. The protein levels of UCP1 and PRDM16, brown-specific markers, were markedly increased upon differentiation of brown adipocytes, which were negligible in white adipocytes. (Fig. 3B). In response to adipocyte differentiation, TfR1 expression continuously rises in brown adipocytes, while it showed only a transient increase of TfR1 in white adipocytes. In contrast to TfR1 expression, FTL was maintained remarkably low throughout the brown adipocyte differentiation, while FTL levels were kept high in white adipocytes with a transient decrease at day 4 (Fig. 3B). Consistently, the total iron content of 3T3L1 preadipocytes was higher than HIB1B preadipocytes, but it remained unchanged after 10 days of differentiation. However, the iron content of HIB1B cells was significantly increased after differentiation (62.5 versus 82.3 ng Fe/mg protein) (Fig. 3C). These results suggest that iron demands consistently rise for brown adipocyte differentiation, while iron demands are transiently increased only at the early differentiation process of white adipocytes, and the excess iron is stored as ferritin.

We also examined the differential regulation of IRP1 and IRP2 during brown versus white adipocyte differentiation. The abundance of IRP1 was gradually increased in both white and brown adipocytes, with a higher expression in brown than white adipocytes. Interestingly, the ISCU protein, a scaffold for synthesis of Fe-S cluster that masks IRP1-binding affinity to IRE, was higher in white adipocytes than in brown, suggesting that a bigger portion of IRP1 would be free to serve as RNA-binding protein in the brown adipocytes (Fig. 3D). In turn, IRP2 protein levels were barely detectable in white adipocytes upon full differentiation at day 10, while IRP2 accumulation showed an overt increase as early as 4 days postdifferentiation and continuously rose during brown adipocyte differentiation (Fig. 3D). IRP2, which does not bind a Fe-S cluster, is known to undergo targeted degradation by the iron-stabilized E3 ligase component FBXL5 (8, 9). This inverse relationship between FBXL5 and IRP2 was evident during adipocyte differentiation; the abundance of FBXL5 in 3T3-L1 white adipocytes correlated with a near absence of IRP2. Conversely, reduced FBXL5 protein in HIB1B adipocytes was enabled to escape FBXL5-mediated degradation and accumulate IRP2 within the cytoplasm (Fig. 3D). Consequently, the augmented iron influx and low iron storage during brown adipocytes (Fig. 3, B and C) are under the coordinated regulation of IRP1 and 2 (Fig. 3D). In parallel with an increase of iron import during brown adipocyte differentiation, there was a significant increase of mitochondrial respiratory chain proteins, suggesting that mitochondrial biogenesis is a key driving force for iron influx during brown adipocyte formation (Fig. 3E).

Next, we compared the kinetic changes of mRNA levels related to white- and brown adipocyte formation and mitochondrial biogenesis. The analysis was conducted with two-way ANOVA to determine the individual effect of cell type and time. All genes were significantly increased during the differentiation process in both white and brown adipocytes (time < 0.0001) (Fig. 4). The increase of Pparγ and adiponectin (adipoQ) shows no statistical significance between two cell types, although adipoQ levels reached a higher plateau in white adipocytes (Fig. 4A). On the contrary, leptin gene expression was significantly higher in white adipocyte differentiation than brown adipocytes, confirming that leptin is a white adipocyte marker (Fig. 4A). The increase of brown-specific markers, such as Prdm16 and Ucp1, was significantly higher in brown adipocytes than white adipocytes (Fig. 4B), consistent with their protein expression (Fig. 3B). The mRNA expression levels of Tfr1 were distinctively higher in brown adipocytes than in white adipocytes (Fig. 4B), which conforms with the posttranscriptional control by IRPs to stabilize their target mRNAs that possess IRE at the 3’UTR (6). The changes of mRNAs that are crucial for transcriptional regulation of mitochondrial biogenesis (e.g., Pgc1α and Tfam) were dramatically higher in brown adipocytes than white adipocytes (Fig. 4C). Moreover, the increase of mRNA of cardiolipin synthase 1 (Crls1) that is a critical phospholipid for mitochondrial biogenesis was dramatically higher in concomitant with brown differentiation compared with white differentiation (Fig. 4C).

Collectively, these results indicate that brown adipocyte, but not white, differentiation is concurrent with the expansion of LIP (increased uptake but decreased storage). This brown-specific iron metabolism is under the control of IRPs to promote iron influx via TfR1 and to meet the demands for mitochondrial biogenesis.

Iron deficiency interfered with cell-autonomous brown/beige adipocyte differentiation

The cellular LIP reflects transitory and chelatable iron. We hypothesized that limiting the LIP compromises the cell-autonomous brown adipocyte differentiation. To test this hypothesis, we depleted the cellular iron by adding deferoxamine (DFO, 100 nM), a cell-permeable iron chelator, during HIB1B adipocyte differentiation (Fig. 5A). Chelation of intracellular iron right after adipocyte commitment step in HIB1B cells dampened brown adipocyte differentiation, including dramatic reductions in mRNA levels of Prdm16, Pparγ, Pgc1α, and Ucp1 (Fig. 5B) and their protein levels (Fig. 5C). The decrease of LIP was confirmed by decreased iron storage protein ferritin (Fig. 5C) and augmented Tfr1 mRNA (Fig. 5B) in the DFO-received cells compared with control. Accordingly, protein expression of mitochondrial oxidative phosphorylation (OxPhos) components I to V, as well as cytochrome C, was significantly reduced in response to DFO treatment (Fig. 5, C and D). Notably, the abundance of respiratory components I and II, which contain eight and three moieties of the Fe-S clusters, is more strongly affected by DFO treatment than components III and IV, which contain one moiety of Fe-S cluster or only heme irons (10). In contrast, component V, which has no iron, was minimally affected by DFO. These results are reminiscent of the iron limitation in myotubes by DFO treatment (11). These observations indicate that the chelation of labile iron significantly interferes with the entire differentiation process of brown adipocytes.

To determine the iron requirements during the beige differentiation, a similar experimental scheme was employed to C3H/10T1/2 cells, a pluripotent stem cell that is commonly used to obtain beige adipocytes (Fig. 5E). The DFO treatment in C3H/10T1/2 cells did not cause a significant difference in Prdm16 mRNA levels (Fig. 5F), but PRDM16 protein levels were significantly lower than the control (Fig. 5G). DFO treatment significantly depleted the LIP, evidenced by the iron starvation responses displaying FTL suppression and TfR1 and IRP2 induction (Fig. 5G). Compared with the complete inhibition of mitochondrial proteins in HIB1B cells (Fig. 5D), DFO treatment only affects complex I and II and cytochrome C levels without changes in complexes III, IV, and V (Fig. 5H). This result reflects the higher iron requirement of CI and CII. More importantly, UCP1 expression was almost completely blunted upon DFO treatment. Collectively, limited iron availability during adipocyte differentiation significantly disrupts the brown and beige adipocyte differentiation program per se, a lesser degree to beige, and posed a detrimental impact on thermogenic mitochondrial biogenesis.

To reduce LIP without causing serious effects on the adipocyte transcriptional program, we selectively depleted the transferrin receptor 1 by adding the siRNA targeting TfR1 (siTfR1) in comparison with the nontargeting siRNA (siCont) during HIB1B differentiation (Fig. 6A). The addition of siTfR1 resulted in a decrease of Tfr1 mRNA levels by 40% (Fig. 6B) and its protein levels by ∼20% (Fig. 6, C and D). Reduction in ∼20% of TfR1 showed no significant challenge on adipocyte differentiation program, exhibiting similar levels of Prdm16, Pparγ, and Pgc1α compared with control (Fig. 6B). However, depletion of ∼20% of TfR1 expression resulted in ∼40% decrease of ferritin, indicating the reduced LIP size. Despite no noticeable differences in PPARγ (Fig. 6C) and mitochondrial respiratory chain proteins (Fig. S2), there was a significant reduction in UCP1 expression by 60% and a slight but significant reduction in cytochrome C by ∼15% (Fig. 6, C and D). Collectively, iron availability is a crucial determinant not only for brown adipocyte differentiation per se but also for competent thermogenic function.

Iron uptake and compartmentalization into mitochondria occur autonomously during beige adipocyte differentiation

To confirm the role of IRP-dependent iron distribution in thermogenic adipocytes, we sought to compare the physical movement of iron and intracellular translocation during the white or beige adipocyte differentiation. To address this question, uncommitted human stromal vascular (hSV) cells were differentiated into either white or beige adipocytes (adding T3 and indomethacin) for 3 days, and then we measured iron uptake for 12 h by using ⁵⁹Fe-Tf (transferrin bound ⁵⁹Fe) as an iron tracer with or without cAMP analog. (Fig. 7A, upper). There was a substantial increase of ⁵⁹Fe-Tf uptake after 3 days of incubation with beige differentiation medium compared with that of white differentiation (Fig. 7A, lower). Upon addition of dibutyryl cAMP (Bt₂-cAMP), a cell-permeable nondegradable cAMP analog as key downstream signaling of β3-adrenergic receptor stimulation, the ⁵⁹Fe-Tf uptake was further augmented by 1.8-fold in beige differentiation compared with white adipocytes (Fig. 7A). The intracellular distribution of ⁵⁹Fe-iron between cytosol and mitochondria was tracked for 12 h. There was a stepwise decrease of cytosolic ⁵⁹Fe, but a corresponding increase of mitochondrial ⁵⁹Fe in a time-dependent manner (Fig. 7B), indicating the redistribution of cytosolic iron into mitochondria and the maintenance of low-cytosolic iron status as a consequence.

To obtain a better understanding of the relationship among iron demand, IRP activation, and adapted metabolic responses in adipocytes, we compared the iron-starvation responses induced by DFO (iron immobilization) versus Bt₂-cAMP (adipocyte browning) stimulation. Similar to DFO responses in HIB1B and C3H/10T1/2 cells (Fig. 5), DFO treatment resulted in depletion of iron-storage protein FTL, stabilization of IRP2, and a robust increase of TfR1 expression in adipocytes as a collective sign of iron deficiency (Fig. 7C). However, DFO treatment decreased oxygen consumption rate (OCR) (Fig. 7D) as well as OxPhos protein expression (Fig. 7E), confirming the notion that adipocyte iron is required for mitochondrial function and respiratory protein transcription (7). The stimulation of Bt₂-cAMP with human SV cells during beige differentiation triggered a substantial accumulation of IRP2, upregulation of TfR1, and slight reduction of FTL, which was comparable with DFO treatment (Fig. 7C). In contrast to DFO, Bt₂-cAMP treatment resulted in increased OCR (Fig. 7D) and OxPhos protein expression compared with the control (Fig. 7E). These results suggest that IRP/IRE signaling should be coupled with external iron import and subsequent relocation into mitochondria for thermogenic mitochondrial development.

Based on these results, we propose a novel iron-regulatory network that β3-adrenergic receptor stimulation instigates cell-autonomous iron movement into mitochondria, which creates an iron-gradient between cytosol and mitochondria, thereby leading to subsequent IRP binding to IRE and iron uptake for thermogenic activation.

Animals

All protocols and procedures were approved by the Institutional Animal Care and Use Committee of the University of Nebraska-Lincoln and the University of Massachusetts-Amherst. To induce adipose tissue browning, C57BL/6 male mice were intraperitoneally injected daily with saline or β-3 adrenergic receptor agonist CL 316,243 (Santa Cruz Biotechnology, 1 mg/kg BW) for 5 days (n = 8 per group). To investigate the iron regulation at thermoneutral conditions, the mice were housed at either 22 °C (ambient) or 30 °C (thermoneutral) for 2 weeks (n = 8 per group). At necropsy, each mouse was fully perfused with 20 ml of ice-cold saline to remove the residual blood to avoid the potential contamination of iron from the blood. Three different depots of adipose tissues were collected from C57BL/6 mice: epididymal white adipose tissue (eWAT) as visceral fat, inguinal white adipose tissue (iWAT) as subcutaneous fat, and interscapular brown adipose tissue (iBAT) as constitutively active brown fat.

Iron determination

For determination of the iron content, ∼50 mg of the adipose depot is digested in 70% nitric acid for 2 h at 75 °C. Total iron content in adipose tissues was analyzed by inductively coupled plasma mass spectrometry (ICP-MASS). The collected data are normalized by tissue weight. To avoid exogenous iron contamination, all glassware and plasticware were soaked in 10% nitric acid and washed with distilled water.

Electrophoresis mobility shift assay (EMSA)

EMSA for IRP/IRE binding was performed based on the published protocol (48) with the following modification. The cytosolic fractions were prepared from eWAT, iWAT, and iBAT. Two micrograms of cytosolic protein was incubated with 16 μl of reaction mixture containing 5% glycerol, 0.2 units Super RNasin (Promega, N2511), 2 μg of yeast tRNA (Thermo Fisher, AM7119), 50 uM DTT, and 50 nM of IRDye700-IRE consensus RNA oligonucleotides in 10 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, and 100 mM KCl for 30 min at room temperature. The resulting reaction mixture was mixed with 2 μl of Orange loading dye (Li-Cor, P/N 927-10100), and load on a 4 to 12% acrylamide/TBE gel (Invitrogen) and run at 120 V for 3 h in the dark. The gel was scanned using Odyssey CLX System (Li-Cor). The RNA oligonucleotide consensus to IRE labeled with IRDye700 probe was synthesized from Integrated DNA Technology based on the published sequence; 5′-UCCUGCUUCAACAGUGCUUGGACGGAAC-3′ (49).

Cell culture

3T3-L1 preadipocytes (ATCC, CRL173) were maintained with 10% bovine calf serum (BCS, GIBCO), and 2 days after reaching confluency, differentiation was induced with DMEM/H containing 10% FBS, 1 μg/ml of insulin, 390 ng/ml of dexamethasone, and 115 μg/ml of isobutyl-methylxanthine. After 2 days, media were changed to DMEM/H containing 10% FBS and 1 μg/ml of insulin and harvested on given days. The murine brown H1B1B preadipocytes were from Dr Johannes Klein (University of Lubeck, Lubeck, Germany) (50). HIB1B cells were differentiated as we described previously (51) and harvested on given days during differentiation. C3H/10T1/2 (ATCC CCL-226) cells, a pluripotent cell line derived from mouse embryos, display fibroblastic morphology and are functionally similar to mesenchymal stem cells (52). C3H/10T1/2 cells were included in this study as commonly used inducible beige adipocytes (53, 54). To induce beige differentiation, C3H/10T1/2 cells were maintained for 3 days in a medium supplemented with 5 μg/ml insulin, 0.5 mM 3-isobutyl-1-methylxanthine (IBMX, Sigma Aldrich), 2 μg/ml dexamethasone (Sigma Aldrich), 2 nM triiodothyronine (T3), 30 μM indomethacin, 5 μM rosiglitazone (Cayman Chemicals), 5 μM rosiglitazone (Cayman Chemicals), and then switched to medium supplemented with 5 μg/ml insulin for 6 days as described previously (53, 54).

Primary hSV cells were isolated from abdominal adipose tissue (subcutaneous fat) from adult females who had liposuction or abdominal plastic surgeries. To reduce the individual variation, pooled samples of 4 to 5 different human subjects were used for the experiments (Fig. 7). All protocols and procedures for human SV cell isolation were approved by the Institutional Review Board (IRB) at the University of Nebraska. For beige differentiation, confluent human SV cells were added to a differentiation mixture containing 500 nM insulin, 2 nM triiodothyronine (T3), 0.25 mM IBMX, 1 μM dexamethasone, 30 μM indomethacin (Indo), 1 μM rosiglitazone BRL49653, 17 μM pantothenate, and 33 μM biotin in 2% FBS for 3 days. For white differentiation, the same medium except for the absence of T3 and Indo was used for 3 days. After the initial commitment step, cells were changed to adipocyte sustaining medium that contains reduced insulin (100 nM) and the absence of IBMX before using for Tf-Fe uptake experiments (Fig. 7A). The experimental details for ⁵⁹Fe-Tf uptake were found in the next session.

To chelate intracellular iron, either HIB1B or C3H/10T1/2 was incubated with 100 nM of DFO (Calbiochem) after 3 days of adipocyte commitment step and kept for 6 days throughout the terminal differentiation (Fig. 5, A and E)

Knockdown of transferrin receptor 1 (TfR1)

siRNAs (On-Target plus, Dharmacon) targeting the mouse transferrin receptor 1 (TfR1) or nontargeting siRNA controls were transfected into HIB1B brown preadipocytes at a final concentration of 100 nM using DharmaFECT4 transfection reagent (GE Dharmacon, CO). Twenty-four hours posttransfection, HIB1B cells were washed and induced differentiation as described above.

⁵⁹Fe-Tf uptake

To prepare ⁵⁹Fe-Tf, a bioavailable form of iron bound with transferrin, 10 mg of apo-transferrin was incubated with 100 μCi of ⁵⁹FeCl₃ (PerkinElmer; Cat. #NEZ037500UC) in 100 mM of disodium nitrilotriacetate at room temperature for 1 h and then passed through a PD-10 column (GE Healthcare) that was equilibrated with 10 μM NaHCO₃/0.25 M Tris-HCl buffer, pH 8.0 to remove unbounded ⁵⁹Fe. ⁵⁹Fe-Tf was eluted with 0.2 M sodium acetate buffer of pH 5.1 (flow rate of about 4 ml/min) and 5 ml volume fractions were collected and made a pool with fractions having absorbance at 280 nm. The ratio of A₄₆₅/A₂₈₀ was 0.046. The molar ratio of Fe/Tf was ∼1.6. For iron uptake experiments, hSV cells that are committed to either white or beige adipocytes were incubated with ⁵⁹Fe-Tf (1 μCi/0.5 × 10⁶ cells) for 12 h. To determine the effects of β3-adrenoceptor stimulation, adipocytes were costimulated with dibutyryl cAMP (Bt₂-cAMP, 0.5 mM) for 12 h (Fig. 7A). To determine the iron compartmentalization, hSV cells that are committed to beige adipocytes were incubated with ⁵⁹Fe-Tf (1 μCi/0.5 × 10⁶ cells) for the first 12 h. Cells were washed thoroughly and nonradioactive beige differentiating medium with Bt₂-cAMP (0.5 mM) for additional 12 h. Cells were harvested every 3 h and used for separation into a nuclear, mitochondrial, and cytosolic fraction (Fig. 7B). For the preparation of subfraction of cytosol and mitochondria, mitochondria/cytosol fractionation kit (Abcam 65320) was used according to the manufacturer’s instruction. ⁵⁹Fe radioactivity was quantified from cytosol and mitochondria by gamma countering (Wizard 2, PerkinElmer).

Western blot analysis

Total cell and tissue extract preparation and western blot procedures were conducted as we described previously (5). Each lane was loaded with 25 μg proteins of cell or adipose tissue lysates. The proteins were separated on 10 to 12% SDS-PAGE gels, transferred to polyvinyl difluoride membranes and polyvinylidene fluoride (PVDF) membrane, and immunoblotted with the relevant primary antibodies. Chemiluminescence from ECL solution (Western Lightning) was detected using an Odyssey FC Imaging System (Li-Cor) and analyzed with Image Studio Lite Software Ver 5.2.5 (Li-Cor). Details of antibody information are available in Table S1.

qPCR

Total RNA was extracted from ∼100 mg of tissue or 0.5 million cells using TRIzolTM Reagent and treated with DNA-free DNA removal kit (ThermoFisher Scientific). Then the RNA was reverse-transcribed for cDNA synthesis (iScript, BioRad). Real-time PCR was carried out on a QuantStudio 6 Flex (Applied Biosystems) using SYBR Green (Fisher scientific). The relative gene expression was calculated based on the 2⁻ΔΔCT method with normalization of the raw Ct values by either 36b4 or Gapdh. For quantification of mitochondrial DNA content (mtDNA) in iWAT, DNA was isolated using DNAzol (Life Technologies). Quantitative PCR was performed in duplicate using mtDNA specific primer (16s rRNA) and nuclear DNA-specific primer (intron 9). Primer sequences are available in Table S2.

Oxygen consumption rate (OCR)

To determine the mitochondrial respiration activities, the O₂ concentration in the human adipocytes (derived from hSV cells) was measured using the XF24 extracellular flux analyzer (Seahorse), as we described previously (51).

Statistics

All results are presented as means ± SD. The Student’s t-test was used to compare two independent groups, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001. Multigroup comparisons were performed by a one-way ANOVA followed by Tukey’s multiple comparisons. The depot-specific iron content at different temperature settings (22 °C and 30 °C) was analyzed by two-way ANOVA with Tukey’s multiple comparisons (Fig. 2B). The gene expression patterns of mitochondria development and iron transport during white and brown adipocyte differentiation (Fig. 4) were analyzed by two-way ANOVA with two main factors of cell type and differentiation time. All statistical analyses were performed using GraphPad Prism 7 (Version 7.03).