TFAM expression in brown adipocytes confers obesity resistance by secreting extracellular vesicles that promote self-activation

Summary The occurrence of diet-induced obesity has been increasing worldwide and has become a major health concern. Mitochondria are densely distributed in brown adipose tissue and are involved in lipid consumption. Therefore, increasing energy expenditure through the activation of brown adipocytes may be a potential therapy for obesity. Our findings showed that mitochondrial transcription factor A (TFAM) homozygous transgenic (TgTg) mice had highly activated brown adipocytes and increased expression of oxidative phosphorylation, leading to resistance to obesity. Transplantation models of TFAM-expressing brown adipocytes could mimic the phenotype of TFAM TgTg mice, and proving their anti-obesity effect. We found that brown adipocytes secrete exosomes which enable self-activation in an autocrine and paracrine manner. The secretion was enhanced in TFAM TgTg brown adipocytes, resulting in a higher activation. These findings may lead to a promising treatment strategy for obesity through selective stimulation of exosome secretion.


INTRODUCTION
Obesity is expanding worldwide and is accompanied by serious medical and economic issues (Blü her, 2019). Most approved anti-obesity medications have side effects. Consequently, there is an urgent requirement for alternative therapeutic approaches (Daneschvar et al., 2016).
The brown adipose tissue (BAT) is important for energy expenditure and is mediated by tissue-specific uncoupling protein 1 (UCP-1), which is known to mediate brown adipocyte-specific nonshivering thermogenesis and energy dissipation (Lowell et al., 1993;Porter, 2006). This process modifies whole-body metabolism, insulin sensitivity, longevity, and susceptibility to weight gain (Cypess et al., 2009;Ortega-Molina et al., 2012). Because considerable amounts of BAT have been found using fluorodeoxyglucose-positron emission tomographycomputed tomography (FDG-PET-CT) in adult humans (van Marken Lichtenbelt et al., 2009;Cypess et al., 2009;Saito et al., 2009;Virtanen and Al., 2009), BAT has received renewed attention. Some human and animal studies have been conducted on activated BAT over the past few decades. BAT is activated under conditions of cold-and diet-induced thermogenesis and pharmacological treatment; however, none of those studies have indicated any safe, stable, and reliable treatments for obesity (Marlatt and Ravussin, 2017).
Mitochondria are densely distributed in the brown adipocytes and provide cellular fuel through the conversion of nutrients into adenosine triphosphate (ATP) energy in the cell through oxidative phosphorylation (OXPHOS) (Gaspari et al., 2004). Mitochondrial activity and metabolite production can influence mesenchymal stem cells (MSCs), i.e., preadipocytes renewal and fate (Zhang et al., 2018;Bahat and Gross, 2019). In addition, during adipogenic cell differentiation and activation, mitochondrial DNA (mtDNA) content, mitochondrial biogenesis, respiratory enzyme complexes, oxygen consumption, and intracellular ATP content are elevated (Li et al., 2017). Thus, intervening in mitochondrial dynamics by targeting brown adipocyte differentiation and activation, thereby facilitating lipid consumption by heat production, could constitute a promising anti-obesity treatment.
Mitochondrial transcription factor A (TFAM) was firstly cloned as a transcription factor for mtDNA by Fisher and Clayton (Fisher and Clayton, 1988). TFAM was then identified as a major component of the

RESULTS hTFAM overexpression in BAT enhances general metabolism and promotes anti-obesity effects
We initially measured hTFAM expression in BAT and WAT. The expression of hTFAM which is controlled by the chicken b-actin promoter, was expressed mainly in brown adipocytes ( Figure 1A), which is consistent with their origin from the Myf5 positive skeletal myogenic lineage (Seale et al., 2008). The detected proteins were confirmed by purified hTFAM and mouse (mTFAM) proteins. hTFAM protein expression was approximately 1.33 fold higher than endogenous mTFAM expression in homozygous (TgTg) mice (Figures 1B and 1C). Therefore, the total amount of TFAM (hTFAM 1.33; mTFAM 1.0) was 2.33 higher compared to the physiological condition. Furthermore, hTFAM expression was approximately 2.1 times higher in TgTg mice than in heterozygous (Tg) mice ( Figure 1D).
Next, we determined the characteristics of hTFAM TgTg mice under a normal chow diet (NCD) for 24 weeks. BW was significantly lower in the TgTg mice compared to the WT as was the weight of inguinal (i)-and epididymal (e)-WAT. On the other hand, the weight of the BAT normalized with the size of the tibia was significantly decreased to 63.5% compared to the WT, however, the degree of decline was more gradual compared to that in the i-and e-WAT (Figures 1E and S1) ( Table 1). Histological analysis showed that TgTg promoted a reduction in the adipocyte area, which is consistent with the decreasing weight of i-and e-WAT in TgTg mice ( Figures 1F and 1G). In the case of Tg, no anti-obesity effect was observed. The h-and m-TFAM protein levels were normalized by the levels of HSP60. The quantitative results are expressed as the mean percentage of the level in the Tg G standard error (SE) (n = 3/group). Total protein extracts from eÀ and i-WAT and BAT probed with a polyclonal antiserum against h-and m-TFAM. (E) Photographs show representative conditions of e À and i-WAT, liver, and BAT from NCD-fed WT, Tg, and TgTg mice. (F) Effect of hTFAM overexpression on e-adipocyte size. Pictures are representative of each sample stained with hematoxylin and eosin and examined by optical microscopy. Scale bar represents 100 mm. (G) Frequency distribution of the e-adipocyte area in NCD-fed WT, Tg, and TgTg mice. **p < 0.01, versus Tg mice. Two-tailed unpaired Student's t test in (C) and (D).

OPEN ACCESS
iScience 25, 104889, September 16, 2022 3 iScience Article Eight-week-old mice were then fed a high fat diet (HFD) for 8 weeks. TgTg mice had extremely low i-and e-WAT accumulation compared with the WT mice ( Figure 2A). The weight of i-and e-WAT was lower in TgTg mice, however, the ratio of BAT to the size of the tibia showed no difference between WT and TgTg mice (Table 2). Moreover, TgTg mice showed no weight change over the 8 weeks, even though the amount of intake was almost the same as that in the WT group ( Figures 2B and S2F). In addition, insulin resistance and glucose tolerance improved significantly in TgTg mice ( Figures 2C-2E). Histological analysis showed lipid accumulation in BAT in mice on the HFD in the WT group; however, the lipid accumulation did not occur in the BAT of the TgTg mice ( Figure 2F).
Analysis of the metabolic rate showed that oxygen consumption rates (VO 2 ) and carbon dioxide production rates (VCO 2 ) were increased in TgTg mice, both during the day and night (Figures 2G and 2H). Furthermore, the respiratory exchange ratio (RER) was significantly lower in TgTg mice (Figures 2G and 2H), which is consistent with the acceleration of lipid metabolism because of increasing brown adipocyte markers in both i-and e-WAT . On the other hand, heat production was increased through BW normalization, however, there was no significant change without BW normalization. Nevertheless, the deep body temperature (rectal temperature) tended to increase in TgTg ( Figure S3J), indicating that hTFAM overexpression may contribute to facilitate heat production.
The daily production of dopamine, noradrenalin, and adrenalin were examined by urine collection and adjusted by the total urine volume and BW. Noradrenaline and adrenaline production was significantly increased in TgTg mice ( Figure 2I). These circulating catecholamines could mediate the inter-organ network between BAT and i-and e-WAT.
The ratio of each limb muscle weight to total BW ( Figure S2A) showed that the soleus muscle, a Type I fiber rich in mitochondria and capillaries conducting OXPHOS to generate energy (Schiaffino and Reggiani, 2011), was bigger in TgTg mice, and the gastrocnemius (thigh) muscle, a Type II fiber with low mitochondrial content (Schiaffino and Reggiani, 2011), was smaller in the TgTg mice compared to the WT. In contrast, the concentrations of mitochondrial OXPHOS proteins in Type II fibers were significantly higher in complex I, III, and V fibers ( Figures S2D and S2E), whereas almost no change was observed in Type I fibers ( Figures S2B and S2C).
Collectively, these results indicated that hTFAM overexpression in BAT significantly increased metabolic activity while maintaining initial weight and avoiding lipid accumulation in systemic organs under HFD feeding conditions.
hTFAM promotes BAT mitochondrial OXPHOS, resulting in both activation of brown adipocytes and browning of white adipocytes To examine the effect of hTFAM overexpression on mitochondrial enzymes involved in mitochondrial OXPHOS in BAT, mRNA and protein expression of each complex was analyzed. qPCR analysis showed that the expression in TgTg was significantly higher than in the WT ( Figure 3A). Although the protein expression of the voltage-dependent anion channel (VDAC) showed no difference between WT and TgTg, complexes II-V were significantly increased in TgTg (Figures 3B and 3C), indicating that mitochondrial activity in brown adipocytes was accelerated by hTFAM overexpression. In parallel to increasing mitochondrial activity, UCP-1 protein expression increased significantly, as did deep body temperature in TgTg  Figure 3D).
Of note, mitochondrial OXPHOS mRNA expression increased significantly in the remote i-and e-WAT (S Figures 3A and 3B), although hTFAM was not overexpressed in these tissues ( Figure 1A). Furthermore, the transcription factor that white adipocytes need for browning was significantly upregulated ( Figures S3C and S3F). With the increased expression of the transcription factors, the mRNA levels of beige-specific genes were also elevated (Figures 3G and 3H). Consistent with the upregulated browning and large decrease in the tissue weight of i-and e-WAT, the mRNA (Figures S3C and S3F) and protein (Figures S3D and S3E and S3G and S3H) levels of UCP-1 were significantly increased.
In this way, appropriately promoted mitochondrial function in brown adipocytes through hTFAM overexpression, induced both thermogenesis and brown adipocyte differentiation/activation-related mRNA and proteins, and WAT browning maker mRNA and proteins, which was reflected in the higher deep body temperature ( Figure S3J) as well as in the weight loss ( Figure 2B).

Effects of hTFAM overexpressed primary brown adipocyte transplantation in HFD fed WT mice
To understand the anti-obesity effect of hTFAM-overexpressing TgTg brown adipocytes alone, adipocytes originating in hTFAM-overexpressing mice were transplanted into WT mice. First, we determined whether the harvested brown preadipocytes differentiated into mature brown adipocytes after incubation through induction and maintenance media ( Figures S4A and S4B). Brown adipocytes ( Figures S4C and S4D) kept in the Matrigel were injected beside the BAT in recipient WT mice. Because the transplanted cells grafted well, which was confirmed by hematoxylin and eosin (HE) staining and immunohistochemistry analysis (Figures 4A and 4B), we fed recipient mice a HFD for 8 weeks. As expected, BW was significantly lower in TgTg transplanted mice (TgTg-t) compared to WT transplanted mice (WT-t) under the same food intake condition (Figure 4C) (Table 3), with significant improvement in insulin resistance and glucose tolerance ( Figures 4D-4F). The weight of i-and e-WAT was also significantly decreased in TgTg-t mice, however, the real BAT weight and the ratio of BAT to the size of the tibia showed no difference between TgTg-t and WT-t mice (Table 3).
Next, we investigated mitochondrial OXPHOS in the BAT. The OXOPHOS mRNA expressions were significantly increased ( Figure 4G) in parallel with Ucp-1, Prdm16, and C/ebp-1b mRNA and UCP-1 protein expression (Figures 4H and 4I and 4J) in endogenous TgTg-t BAT. Although the i-and e-WAT are distant from the brown adipocyte-transplanted site, the weight of the tissues was also significantly decreased. In addition, both i-and e-WAT mitochondrial OXPHOS mRNA expressions were significantly increased in TgTg-t mice ( Figures S4E and S4F). Furthermore, the transcription factors needed for the browning of white adipocytes were increased ( Figures S4G and S4I) and, as a result, most of the browning markers in both iand e-WAT, including UCP-1 protein expression, were significantly increased ( Figures S4H and S4J) ( Figures S5A and S5B), which is consistent with the reduction in WAT weight and upregulation of mitochondrial activity. Moreover, histological analysis showed that TgTg promoted a reduction in the adipocyte area in both the BAT and WAT ( Figure S5C).
These results suggested that the transplanted active brown adipocytes possess anti-obesity effects by activating brown adipocytes in the recipient. They also remotely induced a qualitative shift from white to beige iScience Article adipocytes. These observations indicated that obesity resistance in TgTg-t is directly attributable to activated hTFAM overexpressing brown adipocytes.

TgTg brown adipocyte differentiation and activation by humoral factors in an autocrine or paracrine manner
To reveal the mechanism of obesity resistance and the remote effects on low-hTFAM-expressing adipose tissues in TgTg-t, we observed the change in characteristics of the primary brown preadipocyte culture over time without stimuli for differentiation. Notably, when we did not change the medium for a week, the primary preadipocytes originating in TgTg, differentiated into mature adipocytes, identifiable by multiple lipid droplets, which did not occur in the WT ( Figure 5A). Next, we co-cultured the WT and TgTg preadipocytes by separating them with a 0.6 mm filter or non-permeable sheet ( Figure 5B). The WT cells partly differentiated and grew multiple lipid droplets similar to the TgTg cells after 7 days of seeding the cells, however, the completely separated WT cells showed almost no differentiation ( Figure 5C). We also examined brown adipocyte differentiation-and activity-related gene expression. The 0.6 mm filtered WT cells showed significantly higher expression of Pgc-1a, Prdm16, and C/ebp-1b, compared to isolated WT cells ( Figure 5D). We then observed the WT cell phenotype after adding cell culture supernatants of WT or TgTg.
Although we observed that a small number of WT cells differentiated with WT supernatants, more cells showed significant differentiation in the TgTg supernatant group ( Figure 5E). These data indicated that TgTg cells secrete substances that induce adipocyte differentiation and activation in an autocrine or paracrine manner.

Enhanced EV released in the TgTg brown adipocytes accelerate differentiation and activity
To explore the cause of enhanced auto-activation in the TgTg brown adipocytes, we performed a co-culture study in which WT and TgTg cells were separated by 0.6 mm or 0.03 mm pore size filters. Our results found that WT adipocytes exhibited almost no cell differentiation when the well was separated by 0.03 mm ( Figure 6A). Meanwhile, cells separated by the 0.6 mm filter did exhibit differentiation. Consistent with the cell phenotype, 0.03 mm filtered WT cells did not show brown adipocyte differentiation and activation, but 0.6mm filtered WT cells had a tendency to control TgTg preadipocytes, as shown by mRNA expression of Ucp-1, Pgc-1a, Prdm16, Pparg, and C/ebp-1b ( Figure 6B).
The finding that adipocyte differentiation and activation were blocked by the 0.03 mm filter placed the focus on EVs, especially exosomes, with a diameter of 0.03-0.15 mm (30-150 nm). Transmission electron microscopy (TEM) revealed that multivesicular bodies (MVBs) in TgTg cells were much more abundant than in WT cells ( Figure 6C). We then counted the EVs derived from the WT and TgTg cells and analyzed them using an ExoCounter. Using anti-CD9 Ab (disc) and anti-CD63 (beads), CD9/CD63 double-positive EVs were detected in cells of both genotypes, with higher concentrations seen in TgTg cells. Moreover, using anti-CD9 Ab (disc and beads), significantly more EVs occurred in TgTg cells ( Figure 6D). The diameter of the detected EVs, analyzed by NanoSight, was between 50-150 nm, consistent with exosome size ( Figures S7A and S7B).
We examined nucleotide production in cells of both genotypes. All nucleotides were significantly overexpressed in TgTg cells, suggesting a higher energy metabolism state in TgTg than in WT cells ( Figure S8). Antimycin A, an inhibitor of complex III, strongly suppressed the expression of Ucp-1 and Pgc-1a (Figure 6E), which suggests that the enhanced OXOPHOS activity caused by hTFAM-expression is essential for the activation. iScience Article GW4869, which inhibits the ceramide-mediated inward budding of MVBs and therefore the release of mature exosomes (Essandoh et al., 2015), and nexinhib20, an inhibitor of the interaction between Rab27a and its effector, are known to cause the downregulation of MVB docking to the plasma membrane and exosome secretion (Johnson et al., 2016;Song et al., 2019). Both GW4869 ( Figure 6F) and nexinhib20 ( Figure 6G) showed decreased expressions of brown adipocyte activation marker genes. The inhibitory effect of nexinhib20 was much stronger than GW4869. Moreover, we confirmed the direct effect of EVs released from WT and TgTg adipocytes on adipocyte activity. We found that the EVs from both genotypes increased the expression of activation marker genes ( Figure 6H). Pgc-1a mRNA expression especially, was increased in a dose dependent manner. Furthermore, Ucp-1 mRNA expression showed partial dose dependency. In addition, we evaluated the contents of the EVs from both WT and TgTg adipocytes using secretome analysis. There was no major difference in the exosome content, which contribute to brown adipocyte differentiation and activation (Table S1), suggesting that EVs per se are essentially the same between WT and TgTg. Therefore, the secretion capacity of EVs rather than their content may be the decisive factor in brown adipocyte activity regulation.
Finally, to confirm whether the EVs are responsible for the WAT browning, we used the co-culture (TgTg brown adipocytes and differentiated 3T3-L1 cells) with same time course as brown adipocyte study. We found that the browning marker gene expression, Ucp-1, Cd137, and Eva1, are increased in co-culture 3T3-L1 cells ( Figures S5D and S5E). This means that EVs may influence, at least in part, white adipocyte thermogenesis and its characteristics.
Collectively, these results suggest that a higher mitochondrial function in hTFAM-overexpressing brown adipocytes facilitates MVB traffic and EV release, resulting in the activation of brown adipocytes. Moreover, the beneficial effects of EVs occurred in a dose-dependent manner.

DISCUSSION
Obesity is associated with reduced BAT function, as evidenced by lipid accumulation and mitochondrial dysfunction and loss (Hung et al., 2014;Shimizu et al., 2014). It has also been reported that the downregulation of the OXPHOS system in WAT is correlated with the level of obesity (De Pauw et al., 2009;Kaaman et al., 2007;Heinonen et al., 2015). In the present study, we showed that hTFAM overexpression in the BAT of mice exhibited marked resistance against age-and diet-induced obesity, produced higher energy expenditure, improved insulin secretory function and resistance, and increased mitochondrial function. We also found that hTFAM-overexpressing brown adipocytes release more EVs which work through an autocrine or paracrine manner, resulting in a virtuous cycle, thereby maintaining BAT activity and exhibiting anti-obesity effects.
Several TFAM expression-modified animal model studies involving metabolism have been reported. There are reports showing that TFAM deletion in adipose tissue (Vernochet et al., 2012) and overexpression in skeletal muscle (Koh et al., 2019) increased mitochondrial function through remodeling the OXPHOS complexes heterogeneously, exhibited anti-obesity effects, and improved general metabolic improvement. Contrary to the findings described in the report above (Vernochet et al., 2012). TFAM downregulation in adipose tissue led to adipocyte death and whitening of BAT (Vernochet et al., 2014), with decreased expression and enzymatic activity of the OXPHOS system because TFAM is essential for mtDNA transcription and replication. In the present study, hTFAM overexpression in BAT (TgTg) resulted in increased mitochondrial function through the comprehensive remodeling of the OXPHOS complexes ( Figures 3A-3C), however, there was no effect in the hTFAM heterozygous (Tg) mice in terms of mitochondrial function iScience Article and obesity resistance (Figures 2A and 2B and 3A-3C), presenting a phenotype similar to that of the WT. It has been reported that a mild increase in TFAM levels ($2-fold) leads to dramatic functional effects on transcription and mtDNA replication (Farge et al., 2014), whereas forced overexpression of TFAM results in mtDNA depletion and reduced enzymatic activity (Farge et al., 2014;Kukat et al., 2015;Ylikallio et al., 2010). As seen in our results, the BAT in TgTg mice had approximately 1.33 times higher TFAM expression than that of endogenous mTFAM (the total expression of TFAM was 2.33 fold higher compared to the physiological condition, adding hTFAM [1.33] and mTFAM [1.0]). A 2.33-fold TFAM expression may be appropriate to elicit higher mitochondrial activity leading to BAT activity, phenotypic obesity resistance, and general metabolic improvement.
In our hTFAM transgenic mice, hTFAM is highly expressed in the heart and skeletal muscle (Ikeuchi et al., 2005), but barely detected in other major organs, such as the lungs, liver, and kidneys. Because brown adipocytes develop from a Myf5 positive skeletal myogenic lineage (Seale et al., 2008), the mice also exhibited strong hTFAM expression in BAT but very low levels in WAT ( Figure 1A). Furthermore, in this mouse model, OXPHOS expression in slow-twitch muscles, such as the soleus muscle, which is rich in mitochondria and capillaries (Schiaffino and Reggiani, 2011), did not significantly differ between WT and TgTg mice ( Figures S2B and S2C), whereas in fast-twitch muscles, such as the extensor digitorum longus, with low mitochondrial content (Schiaffino and Reggiani, 2011), complex I, III, and V showed increased expression ( Figures S2D and S2E).
To eliminate all metabolic effects in skeletal muscle, we performed an ex vivo cellular experiment (Min et al., 2016;White et al., 2019). We extracted brown adipocytes from mice and re-implanted them into donors. We then evaluated the independent anti-obesity ability of TFAM-overexpression in brown adipocytes. We first determined the appropriate brown adipocyte number for transplantation; 1310 6 cells, a relatively smaller number compared to previous reports (Min et al., 2016;Kishida et al., 2015), were decided as suitable for graft survival and optimal weight loss. The solidified Matrigel in the circumference of BAT did not elicit immunoreaction, i.e., there was no accumulation of immune cells, in any time course, between 24 h and 2 weeks after transplantation ( Figure 4A). We also confirmed that the transplanted adipocytes had matured during the second week, as exemplified by multiple lipid droplets and an integrating well in the surrounding tissue ( Figure 4B).
TgTg mice showed significant weight loss on the HFD, mainly via i-and e-WAT loss (Table 2), despite TFAM not being overexpressed in WAT ( Figure 1A). However, the gene expression of OXPHOS and beige adipocyte markers were increased in WAT, which is similar to the phenomenon induced by sustained thermogenic activation (Villarroya et al., 2017). Compared to BAT, mitochondria in WAT are fewer in number, smaller in size, contain fewer cristae, and have lower expression levels of fatty acid b-oxidation-related enzymes and acyl CoA dehydrogenase, resulting in triglyceride storage (Cedikova et al., 2016). However, the ''brown-like'' cells (also called ''beige'' cells) within white adipose depots are not derived from the Myf-5 lineage (Seale et al., 2011;Wu et al., 2012), but possess high mitochondrial content and OXPHOS capacity (Ohyama et al., 2016). The conversion of white into beige adipocytes is typically induced by certain environmental cues, such as b2 or 3-adrenergic receptor stimulus by catecholamine, chronic cold exposure, exercise, or long-term treatment with a PPARg agonist (Vosselman et al., 2012;Yamaga et al., 2008;Kuji et al., 2008;Harms and Seale, 2013). In the present study, the absolute daily catecholamine production did not iScience Article differ between WT and TgTg mice, even though the BW was significantly lower in TgTg mice, indicating that the catecholamine levels per unit volume of adipose tissue were significantly higher in TgTg ( Figure 2I).
Recently, an inter-tissue communication mechanism mediated by EVs containing nicotinamide phosphoribosyl transferase (NAMPT) has been suggested. NAMPT is a key biosynthetic enzyme of NAD + that induces activation of sirtuins. NAMPT-containing EVs secreted from both white and brown adipocytes internalize into primary hypothalamic neurons, resulting in the activation of the sympathetic nervous system (Imai, 2016;Yoshida et al., 2019;Tokizane and Imai, 2021). In this article, we showed that EVs from brown adipocytes enhanced the differentiation and activation of WT cells ( Figure 6H). In addition, we showed that hTFAM-overexpressing brown adipocytes secrete more EVs than the WT cells ( Figure 6D). These reports and our results clearly link an important key concept explaining the mechanism sustaining high catecholamine production levels and activation of brown adipocytes. It is however still unclear whether it is the same EVs and intra-vesicular factors responsible for the two phenomena mentioned above.
The fact that hTFAM-overexpressing brown adipocytes accelerate EVs secretion prompts us to question the relationship between brown adipocyte activation and remote effects on i-and eÀ WAT browning ( Figures 3G and 3H) ( Figures S3C-S3H) ( Figures S4G-S4J). During the co-culture study, TgTg brown adipocyte and differentiated 3T3-L1 cells showed that only some browning marker gene expression was increased ( Figures S5D and S5E). Nevertheless, these data indicated that EVs partially contribute to thermogenesis even in white adipocytes and affect the white adipocyte characteristics. Therefore, it is expected that iand eÀ WAT browning may go through the process with other factors such as catecholamine in vivo.
The overexpression of all nucleotides in TgTg cells, ( Figure S8) and the inhibitory effects of antimycin A on the expression of Ucp-1 and Pgc-1a ( Figure 6E) suggests that the higher exosome secretion and enhanced differentiation depend on the higher OXOPHOS activity caused by hTFAM-expression. Moreover, both GW4869 and nexinhib20 suppress the expression of brown adipocyte activation marker genes ( Figures 6F and 6G). Of note, the inhibitory effect of nexinhib20 was much stronger than that of GW4869. Because Rab-GTPase is associated with vesicles and the inner side of the plasma membrane and participates in intracellular trafficking of vesicles (Jordens et al., 2005;Blanc and Vidal, 2018), increased GTP levels in hTFAM brown adipocytes might enhance the switch from Rab to the active state on binding of GTP (Blanc and Vidal, 2018), resulting in EV secretion.
In addition, it is reported that EVs released by BAT contain miR-99b accumulated in hepatocytes and reduced hepatic FGF21 levels, resulting in decreased circulating FDF21 (Thomou et al., 2017). In this study, serum FGF21 levels in both TgTg and TgTg-t were significantly lower than those in the control ( Figure S3I) ( Figure S6). Together with more EVs released from hTFAM-overexpressing brown adipocytes, this is consistent with the mechanism by which exosomal miRNAs derived from BAT regulate whole-body metabolism (Thomou et al., 2017). In the in vitro study, we determined that primary brown adipocytes from TgTg mice have potent differentiation and activation abilities without a preadipogenic medium ( Figure 5A).
The 0.6 mm-filtered co-culture showed that the differentiation-related genes in WT adipocytes were significantly increased compared to those in the completely separated culture ( Figure 5D). In addition, WT adipocytes cultured in the conditioned medium originating in the TgTg adipocyte culture increased differentiation compared to the regular medium ( Figure 5E). These data led us to the hypothesis that an autocrine or paracrine mechanism accelerates brown adipocyte differentiation and activity. Because the size of the candidate substance is over 30 nm ( Figures 6A and 6B), we speculate that exosomes, which are a class  iScience Article of EVs that range between 30 to 150 nm in size and are formed from MVBs along the endocytic pathway (Quan and Kuang, 2020), play an important role in regulating brown adipocytes. This speculation is supported by the exosome inhibition study with GW4869 ( Figure 6F) and nexinhib20 ( Figure 6G).
The activation of adipocytes undergoes a subsequent multi-step activation of the transcription factor network, which affects cell phenotype, determination, differentiation, and activation (Kajimura, 2015;Seale, 2015). Moreover, only EVs from TgTg increased the brown adipocyte activity, but the EVs from the WT did not. However, adding the concentrated EVs, from both WT and TgTg, to WT preadiopocytes showed increasing levels of Ucp-1 and Pgc-1a expression in a dose-dependent manner ( Figure 6H). This result is consistent with the actual measurement value of the number of EVs, which is 2-to 4-fold in TgTg compared to the WT ( Figure 6D). During the secretome analysis, we did not see a major difference of between exosome content which contribute to brown adipocyte differentiation and activation (Table S1). Thus, it is the EV secretion capacity rather than EV content which may be the decisive factor in regulating brown adipocyte activity.
To date, BAT has been known to secrete many bioactive factors that activate BAT itself under thermogenic stimuli (Villarroya et al., 2013). The present study demonstrated that the EVs secreted by hTFAM-overexpressing brown adipocytes could permanently maintain the brown adipocyte phenotype as an active state even without stimuli, resulting in potent obesity resistance, locally, or at a distance. This study gives a good rationale for the stimulation of EV secretion from brown adipocytes for anti-obesity treatment. However, the direct effect between enhanced EV release and the beneficial effect on the metabolic phenotypes remain to be elucidated. As a next stage, we suggest further research into establishing the stable system to collect a large number of EVs to analyze the direct effects of EVs in vivo.
In this study, we discovered that the differentiation and activation of brown adipocyte are proceeded by EVs secreted from brown adipocyte in autocrine/paracrine and dose-dependent manners. In conclusion, our study demonstrated that the hTFAM overexpression in brown adipocytes activates adipocytes and increases mitochondrial functions via enhanced exosome secretion, which may contribute to obesity resistance.

Limitations of the study
In the present study, we reveal that hTFAM TgTg has strong anti-obesity phenotype. Also we advocated that hTFAM overexpression in the brown preadipocytes accelerate mitochondrial function and EVs secretion, then facilitate self-preadipocyte differentiation as an anti-obesity mechanism. However, it has not yet been verified if the factor that influences on this phenotype is due to TFAM's gain-of-function or loss-offunction. Therefore, it is required further verification with comparing the phenotype of this TgTg mice and another model, e.g. brown adipocyte specific conditional hTFAM knock-in mouse using ROSA 26 locus, to eliminate the possibility of off-target effects due to insertion of transgene construct, as a future study.
Another limitation to this study is that the number of brown adipocytes that could be used for transplantation was restricted. The weight loss effect was mild in the TgTg-t model compared to TgTg during HFD feeding. One reason for this difference is an insufficient number of transplanted cells because BAT in WT mice has at least 1310 7 brown adipocyte based on the size of brown cell pellet. Despite trying to transplant 1310 7 cells, the effect of BW loss was worse. Transplantation with a higher number of cells, therefore, may cause adverse effects possibly because of increased unknown artificial factors. We also found that it raised the immune reaction. Given that significant BW loss is obtained even with a transplantation of 1310 6 cells, hTFAM-overexpressing brown adipocytes partially play an important role against obesity. Moreover, we do not rule out the possibility that factors other than brown adipocytes also work against obesity in TgTg mice. Therefore, to eliminate these limitations, more in vivo studies are required to verify whether a BAT specific hTFAM-overexpressing model shows the same degree of obesity resistance as TgTg mice.

SUPPLEMENTAL INFORMATION
Supplemental information can be found online at https://doi.org/10.1016/j.isci.2022.104889. Figure 6. Enhanced EV release in TgTg brown adipocytes increases differentiation and activity (A) Differentiated positive control TgTg preadipocytes and aspects of WT preadipocytes incubated in 0.03 mm or 0.6 mm filtered co-culture well. Number of mature cells was counted, and the results were expressed as the mean percentage of the number in control WT adipocytes incubated in the isolation culture GSE. *p < 0.05, versus 0.6 mm co-culture WT. (B) Thermogenic and brown adipocyte differentiation/activation-related gene expression in BAT, as assessed by RT-PCR. The mRNA levels were normalized to the levels of Hprt, and the results are expressed as the mean percentage of the level in control WT adipocytes incubated in the isolation culture GSE (n = 10/group). *p < 0.05, ***p < 0.001, versus control WT. (C) Molecular features of WT and TgTg brown adipocytes, assessed by electron microscopy. Scale bars represent 1 mm and 500 nm. High-power view of TgTg shows the MVB (boxed area). (D) Quantification of EVs derived from WT or TgTg brown adipocytes with the ExoCounter using anti-CD63 (left) or anti-CD9 (right) Ab-coated discs and beads conjugated with Ab against CD9 (n = 4/group). ### p < 0.001 versus WT. Inhibitory effect of exosomes on TgTg brown adipocytes incubated with or without (E) Antimycin A, (F) GW4869, or (G) nexinhib20. The expression of genes related to brown adipocyte activation was assessed by RT-PCR (n = 3/group) (E), (n = 4-8/group) (F, G). *p < 0.05, ***p < 0.001 versus Antimycin A (À), GW4869(À), or Nexinhib(À). (H) Brown adipocyte activation-related gene expression in adipocytes cultured with EVs purified from WT and TgTg adipocytes, assessed by RT-PCR (n = 5/ group). ## p < 0.01, ### p < 0.001 versus Exo(À). Two-tailed unpaired Student's t test in (D), (F) and (G). One-way ANOVA with Tukey's multiple comparisons test in (A), (B), (E), and (H). Bar graphs represent mean G SE.

Materials availability
This study did not generate new unique reagents.
Data and code availability d All data reported in this paper will be shared by the lead contact on request.
d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact on request.

Mouse models
We bred and used human TFAM (hTFAM) transgenic mice originally generated in our laboratory by Ikeuchi et al. (2005). hTFAM cDNA was inserted into the unique EcoRI site between the CAG (modified chicken iScience Article b-actin promoter with CMV-IE enhancer) promoter and 3 0 -flanking sequence of the rabbit b-globin gene of the pCAGGS expression vector and used to generate transgenic mice. The pronuclei of fertilized eggs from hyperovulated C57BL/6 mice were microinjected with this DNA construct. The presence of the TFAM transgenes was confirmed by polymerase chain reaction (PCR) before the experiments. Four independent founder lines were identified and mated to C57BL/6 wildtype (WT) mice to generate pure C57BL/6 genetic background hemizygous/homozygous hTFAM transgenic (Tg and TgTg, respectively) and non-transgenic WT offspring. The animals were allowed free access to water and a standard diet (SD, CE-2; 343 kcal/100 g, 12.6% energy as fat; CLEA Japan, Inc.). Eight-week-old male mice were fed a high-fat diet (HFD) (HFD 32 provided 507.6 Kcal/100 g [32.0% energy as fat; CLEA Japan, Inc.]) for 8 weeks. At the end of the experiment, the animals were sacrificed by isoflurane anesthesia inhalation after 16 h of fasting. Mice were housed 2-5 per cage at standard housing conditions (22 C, 12h light/dark cycle, relative humidity of 45-65 rH). All mice were kept under specific pathogen-free conditions in the animal facility at Kyushu University. All protocols were approved by the Committee on the Ethics of Animal Experiments at the Graduate School of Medical Sciences, Kyushu University.

Cell culture Isolation of primary Brown preadipocytes
Primary brown preadipocytes were isolated as previously described (Aune et al., 2013). Briefly, adipose tissue was dissected from euthanized 6-8-week-old mice. Digestion medium (DMEM/F12 containing 10% FBS, P/S, 2.4 U/mL Dispase II, 0.2 W/V % Collagenase D and 10mM CaCl2: 1 mL/g tissue) was added to the tissue, and it was then minced into small pieces. A brownish pellet was obtained after digestion at 37 C with constant agitation at 200 rpm for 40 min. The cell suspension was filtered using a cell strainer (100 mm Falcon cell strainer) (Becton Dickinson, Lincoln Park, NJ, USA), and cells were plated on collagen I-coated dishes. Cells were washed twice with PBS, and fresh medium was added 1-2 h after plating the cells to remove red blood cells. All cells were grown at 37 C in a 5% CO 2 humid atmosphere.

Transplantation
Primary brown preadipocytes (passage 4-6, 1 3 10 6 cells) from WT or TgTg mice were induced to differentiation after 48 h of incubation in induction medium (DMEM/F12 [10% FBS and P/S], 5 mg/mL insulin, 1 nM Triiodothyronine [T3], 125 mM indomethacin, 2 mg/mL dexamethasone, 0.5 mM isobutylmethylxanthine [IBMX], 0.5 mM rosiglitazone). Those in vitro differentiated (also called ''activated'') brown preadipocytes were suspended in Matrigel (Corning, NY, USA). Eight-week-old male mice were sub-dermally injected with the cell suspension on the dorsal body surface above the BAT under anesthesia. After transplantation, mice were fed an HFD for 8 weeks. At the end of the experiment, the animals were sacrificed by the inhalation of isoflurane after 16 h of fasting.

Respiratory metabolism
The metabolic rate was measured by indirect calorimetric analysis in WT, Tg, and TgTg mice using an opencircuit calorimeter (Oxymax; Columbus Instruments, Columbus, OH, USA). Sixteen-week-old mice, after 8 weeks on a HFD, were housed in individual chambers. The temperature was maintained at 22 C, with an airflow of 0.5 L/min. The animals were allowed free access to water and a standard diet. Mice were acclimatized to the chambers for 24 h before monitoring. They were then monitored for 72 h VO 2 and VCO 2 were measured every 10 min using an electrochemical O 2 analyzer and CO 2 sensor (Oxymax), and the RER was calculated as VCO 2 /VO 2 (volume of CO 2 produced per volume of O 2 consumed [mL/Kg/h]).

RNA extraction and quantitative RT-PCR
Total RNA was extracted from frozen adipose tissue and adipocyte samples using the Isogen reagent (Nippon Gene, Tokyo, Japan), according to the manufacturer's instructions. Extracted RNA (1 mg) was converted into single-stranded cDNA using a PrimeScriptä RT Reagent Kit (Takara Bio Inc., Tokyo, Japan For total protein extract and western blot analysis, hTFAM, mouse TFAM, UCP-1, PGC-1a, OXPHOS, inguinal white adipose tissue (i-WAT), epididymal white adipose tissue (e-WAT), and BAT were homogenized in RIPA buffer (Nakarai, Kyoto, Japan), and centrifuged for 10minat 13,000 rpm. Protein concentrations were determined using a Pierce TM BCA Protein Assay Kit (Thermo Fisher Scientific); 20 mg protein/ lane was separated discontinuously on sodium dodecyl sulfate-polyacrylamide gels (10%) and transferred to PVDF 0.45mm membranes (Millipore, USA). After blocking nonspecific binding sites, membranes were incubated overnight at 4 C with primary antibodies, followed by HRP-conjugated secondary antibodies. The primary and secondary antibodies are listed in the key resources table. Chemi-Lumi One Super (Nakarai, Kyoto, Japan) was used for the detection.
Measurement of blood glucose, intraperitoneal glucose tolerance test (IPGTT), and intraperitoneal insulin tolerance test (IPITT) Blood samples were obtained from the tail vein of the mice. Plasma glucose and insulin concentrations were determined using the glucose oxidase method and an enzyme-linked immunosorbent assay (ELISA; Morinaga Institute of Biological Science, Yokohama, Japan), respectively. After 16 h of fasting, glucose tolerance was assessed by an IPGTT, and a glucose bolus (5 mg/kg BW) was injected intraperitoneally. For the IPITT, mice were intraperitoneally injected with human biosynthetic insulin (0.5 U/kg BW) (Novo Nordisk, NJ, USA).

Morphometric study
BAT, iWAT, and eWAT were stained with hematoxylin and eosin (HE). Mouse adipose tissues were obtained after 8 weeks of a normal chow diet (NCD) or HFD conditions. To calculate the adipocyte area, the sections were coded and analyzed by a blinded observer. In each animal from the four experimental groups, 2,000 or more adipocytes in 24 randomly selected fields at 200-fold magnification using fluorescent microscopy (Model BZ-9000, Keyence, Osaka, Japan) were examined and averaged for morphometric analysis. Immunostaining for hTFAM in BAT was performed as previously described (Ikeuchi et al., 2005). Briefly, the BAT were fixed in 10% formaldehyde and embedded in paraffin. Paraffin sections were cut at 3 mm and deparaffinized. After inactivation of endogenous peroxidase with 10% H 2 O 2 in methanol for 20 min at room temperature, the sections were preincubated for 30 min with 1% bovine serum albumin in phosphate-buffered saline (PBS). The samples were then incubated with anti-hTFAM affinity purified antibody (4 mg/mL) overnight at 4 C, washed in PBS, and probed with anti-mouse IgG antibody labeled with peroxidase (Histofine Simple Stain MAX PO(M)) for 30 min at room temperature. The peroxidase was then visualized with diaminobenzidine. The primary antibody was replaced with mouse IgG as a negative control.
BAT, including the Matrigel part, was briefly fixed in 10% formaldehyde and embedded in paraffin. Paraffin sections were cut to 3 mm thickness and deparaffinized. After inactivation of endogenous peroxidase with 10% H 2 O 2 in methanol for 20 min at room temperature, the sections were preincubated for 30 min with 1% bovine serum albumin in phosphate-buffered saline (PBS). The samples were then incubated with anti-human TFAM mouse monoclonal antibody (4 mg/mL) overnight at 4 C, washed in PBS, and probed with antimouse IgG antibody labeled with peroxidase (Histofine Simple Stain MAX PO(M), Nichirei Biosciences Inc., Tokyo, Japan) for 30 min at room temperature. The peroxidase sites were visualized with diaminobenzidine (Nichirei Biosciences Inc., Tokyo, Japan).

Urine analysis
A 24-h urine sample was collected, and an acidic urine analysis was conducted with 6N HCl using metabolic cages. Urinary dopamine, noradrenalin, and adrenalin were analyzed by high-performance liquid chromatography (HPLC). The urine was then mixed and centrifuged at 75003 g for 5 min before being stored at À80 C for analysis.

Nucleotide purification
Nucleotides were extracted from WT and TgTg preadipocytes using 2% trichloroacetic acid (TCA). The sample solution was incubated on ice for 3 min and then centrifuged at maximum rpm for 5minat 4 C. Then, 0.1% trifluoroacetic acid (TFA) in 50% acetonitrile (ACN) was added to the supernatants, and the mixture was transferred to a MonoSpin TiOspin column (GL Sciences Inc., Tokyo, Japan) and centrifuged at 50003g for 2 min. Next, 0.1% TFA in 80% ACN was added, the solution was centrifuged at 50003g for ll OPEN ACCESS iScience 25, 104889, September 16, 2022 23 iScience Article 2 min, 2.5% 50 mL of aqueous ammonia solution was added, and it was centrifuged at 10003g for 1 min. The purified sample was collected with an equivalent volume of ACN and then injected into a liquid chromatography-mass spectrometer (LC/MS).

Medium transplantation study
WT cells were cultured until confluent in a 6-well plate covered with collagen I. The cells were then washed with PBS and the supernatants from TgTg cells culture were added, which was collected 7 days after confluent condition. After transplantation, the differentiated cells, which could be detected via the apparent lipid droplets, were counted until day 7.

Co-culture study
We evaluated the contribution of EVs to adipocyte differentiation using a horizontally connected co-culture system that allowed for the simultaneous observation of both culture vessels. The same numbers of WT and TgTg cells were seeded into the UniWellsä Horizontal Co-Culture Plate (FUJIFILM Wako Chemicals USA. Corporation), separated by a 0.03 or 0.6 mm pore size filter. Both cells were collected 10 d after incubation at 37 C, and mRNA was extracted. We also performed this with the co-culture cells, 4 days after the establishment of the confluent condition, and with the matured 3T3-L1 cells, 7 days from the beginning of differentiation. 3T3-L1 cells were seeded and differentiated on the UniWellsä Horizontal Co-Culture Plate before being connected to the TgTg cell plate.

Electron microscopy
The samples were fixed with 2% paraformaldehyde (PFA) and 2% glutaraldehyde (GA), post-fixed with 2% osmium tetroxide and embedded in resin. The polymerized resins were ultra-thin sectioned at 70 nm using an ultramicrotome. They were stained with 2% uranyl acetate and further stained with a lead stain solution. The grids were observed using a transmission electron microscope (TEM; JEM-1400Plus; JEOL Ltd., Tokyo, Japan) at an acceleration voltage of 100 kV. Digital images were captured using a CCD camera (EM-14830RUBY2; JEOL Ltd., Tokyo, Japan).

Exosome isolation
Cells were cultured until confluent in 100 mm dishes, then washed with PBS and added to the medium supplemented with EV depleted FBS. Vesicle depletion in FBS was performed via ultracentrifugation at 100,000 3 g at 4 C for 16 hours. The cell number in both WT and TgTg were approximately 1310 6 per 100mm dish. Cell culture conditioned media were harvested 48 h after changing medium and centrifuged at 12,000 rpm for 1 h at 4 C to remove contaminating apoptotic bodies and cell debris. The supernatants were transferred to a clean tube and centrifuged again at 34,500 rpm for 2 h at 4 C to pellet the exosomes. The supernatants were carefully removed, and exosome-containing pellets were resuspended in 200 mL of PBS and pooled.

Nanoparticle tracking analysis (NTA)
NTA measurements were performed using a NanoSight NS300 Instrument (NanoSight, Amesbury, UK). After resuspending exosome pellets in 200 mL of PBS, samples were diluted 1000-fold with PBS before measurement. Particles in the laser beam underwent Brownian motion, and a video of the particle movements was recorded. NTA 3.2 software was used to analyze the video and to determine particle size distribution. Twenty-five frames per second were recorded for each sample at appropriate dilutions at the ''frames processed'' settings of 1,500. The detection threshold was set at ''5 Multi,'' and at least 1,000 tracks were analyzed for each video.

Exosome count
The exosomes were counted as described previously (Kabe et al., 2018). The optical disc was attached to a removal plate containing 16 wells for sample injection. Each well was coated with 5 mg/L anti-CD9 or anti-CD63 antibody in a carbonate-bicarbonate buffer (pH9.6) overnight at 4 C. After removing the carbonatebicarbonate buffer and washing with 0.05% Tween 20 in PBS (PBS-T), blocking solution (0.1% casein in PBS-T) was added and incubated for 30 min at 37 C. Next, 50 mL of the sample solution was added to each well and incubated for 2 h at 37 C followed by washing with PBS-T. Then, approximately 1 mg of anti-CD9 or anti-CD63 antibody-attached beads in blocking solution was added to each well and incubated for 90 min at 37 C. Each well was washed with PBS-T, followed by PBS-T and deionized water.

OPEN ACCESS
iScience Article