Adipose knockout of H-ferritin improves energy metabolism in mice

Objective Ferritin, the principal iron storage protein, is essential to iron homeostasis. How iron homeostasis affects the adipose tissue is not well understood. We investigated the role of ferritin heavy chain in adipocytes in energy metabolism. Methods We generated adipocyte-specific ferritin heavy chain (Fth, also known as Fth1) knockout mice, herein referred to as FthAKO. These mice were analyzed for iron homeostasis, oxidative stress, mitochondrial biogenesis and activity, adaptive thermogenesis, insulin sensitivity, and metabolic measurements. Mouse embryonic fibroblasts and primary mouse adipocytes were used for in vitro experiments. Results In FthAKO mice, the adipose iron homeostasis was disrupted, accompanied by elevated expression of adipokines, dramatically induced heme oxygenase 1(Hmox1) expression, and a notable decrease in the mitochondrial ROS level. Cytosolic ROS elevation in the adipose tissue of FthAKO mice was very mild, and we only observed this in the brown adipose tissue (BAT) but not in the white adipose tissue (WAT). FthAKO mice presented an altered metabolic profile and showed increased insulin sensitivity, glucose tolerance, and improved adaptive thermogenesis. Interestingly, loss of ferritin resulted in enhanced mitochondrial respiration capacity and a preference for lipid metabolism. Conclusions These findings indicate that ferritin in adipocytes is indispensable to intracellular iron homeostasis and regulates systemic lipid and glucose metabolism.


INTRODUCTION
Iron is an essential micronutrient involved in various physiological processes.It promotes energy production as an essential component of many enzymes and plays a critical role in cell differentiation and proliferation.Meanwhile, excessive iron can generate reactive oxygen species (ROS) via the Fenton reaction and even lead to ferroptosis, which is the programmed cell death characterized by iron-dependent accumulation of lipid peroxidation.Therefore, iron concentration is strictly regulated in all species.Recent studies have suggested a relationship between iron, energy metabolism [1], and thermogenic functions [2e4].However, the exact connection has never been clarified.Some data are even incongruent with each other: some work reported that iron increases insulin resistance [5,6], but meanwhile, transferrin receptor gene (Tfrc) deletion in mouse adipocytes results in decreased iron in adipocytes, accompanied by insulin insensitivity, increased adiposity, and defective thermogenesis [4].There seems not to be a simple or direct relationship between them.One important reason for this confusion might be that the already complex connection between iron and energy metabolism is further exacerbated by the fact that both iron and glucose/fat metabolisms are subject to systemic regulation.For example, fat metabolism may be regulated not only by iron in the adipocyte but also by cross-talks with the liver [7], CNS [8], and gut [9,10], where iron all exert effects.A hypothetical scenario to illustrate this is that, on the one hand, iron outside of the adipocytes could act in such a way that directly stimulates fat synthesis [11] or diet intake [12]; however, on the other hand, iron in the adipocyte could influence oppositely, i.e., increase fat utilization [3,13].The multiple facets of systemic energy regulation by iron thus make conclusions from different studies sometimes hard to reconcile.To simplify matters, focusing on one aspect of iron and energy metabolism in a particular tissue appears necessary.Ferritin consists of two types of subunits: ferritin heavy chain (FTH) and ferritin light chain (FTL).As the major iron-storage molecule, ferritin serves a vital function in regulating free iron levels and reducing ROS production.Both the increased free iron level and the oxidative stress could induce ferritin expression [14,15].Clinically, serum ferritin is a well-known marker of body iron levels and inflammation.It is also correlated to several serum adipokines, including insulin [16], adiponectin [17], leptin [12], resistin [18], and fatty acid binding protein 4 (FABP4) [19], while some of the mechanisms are not fully understood.There is a strong correlation between ferritin and energy metabolism.Fth germline deletion in mice results in embryonic lethality at a very early stage [20].Inducible systemic Fth gene knockout in adult mice also results in a quick death after induction, accompanied by complete loss of adaptive thermogenesis [21].The broad inactivation of Fth and tissue-specific Fth targeting, as reported in other studies [22e24], all created altered iron homeostasis.Interestingly, while conditional knockout of Fth in several tissues produced some physiological consequences, such as deletion in macrophages resulting in protective effects on obesity and diabetes [22], none of these lead to severe consequences under normal conditions [23,24].One hypothesis posits that secreted ferritin may have an iron regulating and iron transport function locally in some tissues [25e27], compensating for the loss of ferritin in other tissues.It is also possible that the key tissues for FTH functions are hitherto not revealed, and the several tissues where Fth has been removed are not where Fth manifests its critical physiological roles.At the embryonic stage, Fth is early and robustly expressed in the heart and brown fat tissue [20].While Fth deletion in the heart mildly affects the mice, the role of FTH in adipose tissues is not reported.To further explore the FTH function and more precisely analyze the iron-fat connection, we report here the generation of adipocyte-specific Fth1 deletion (Fth AKO ) in the mouse.Adipocyte-specific Fth1 deletion does not lead to death or severe abnormality either.However, a notable increase in glucose tolerance, insulin sensitivity, and a minor increase in adaptive thermogenesis was observed.Concomitantly, an antioxidant-heme oxygenase 1 (Hmox1), mitochondrial respiratory ability, and expression of several adipokines were upregulated.This research contributes to understanding the complex interplay between ferritin and energy metabolism and identifying potential therapeutic targets of metabolic diseases.

Animals
Mice referred to as Fth AKO mice are conditional adipocyte Fth1 knockout (Adipoq-Cre; Fth fl/fl ) mice, generated by intercrossing Fth fl/fl mice (Jackson Laboratory, stock #018063) [28] with mice carrying the adiponectin promoter-cre transgene (Cyagen Biosciences, Suzhou, China) [29].Fth fl/fl littermate mice were used as controls.Male mice were used in this study.All mice were housed under specific pathogen-free conditions, 20 Ce26 C, in a 12-h/12-h light/dark cycle.They were given a normal diet (Beijing Keao Xieli Feed Co., Ltd., China) or a vitamin E-supplemented diet (0.5 g of a-tocopherol [Coolaber, Beijing, China] added in 1 kg normal chow) and water ad libitum.All animal protocols were approved by the Institutional Animal Care and Use Committee of Shenzhen Institute of Advanced Technology.

Aconitase activity assay
Adipose tissues were dissected and washed with cold PBS.The fresh tissues were homogenized in cold Triton X-100 lysis buffer (20 mmol/l TriseHCl, pH 8.0, 137 nmol/l NaCl, 1 % Triton X-100, protease inhibitors) and centrifuged at 12,000Âg, 4 C to obtain clear tissue lysates.The protein concentration was measured with a BCA kit (Beyotime, Shanghai, China).60 mg of protein was added to 200 ml of the reaction buffer (50 mmol/l K 2 HPO 4 , 30 mmol/l citric acid, pH 7.4).The increment of absorbance at wavelength 240 nm within 5 min was detected by a microplate reader (Biotek Synergy H1, Santa Clara, CA, USA) as the relative aconitase activity.

Ferrozine iron assay
Detection of total iron based on the colorimetric reaction of ferrous and ferrozine was performed as described previously [30].Briefly, iron in tissue lysates or serum samples was released by heating and 37 % HCl treatment, reduced to Fe 2þ by adding ascorbic acid, and reacted with an excess amount of ferrozine.The absorbance of the iron ferrozine complex was measured at 562 nm.Iron concentration was calculated and normalized to protein concentration.

Isolation of mature adipocytes
Adipose tissues of mice were dissected, minced, and put into 1.5 ml centrifuge tubes. 1 ml collagenase type II buffer (collagenase 2 mg/ml, Ca 2þ 10 mmol/l, 2 % BSA) was added to each tube, incubated at 37 C for 1e2 h, and vortexed every 10 min.After digestion, the brown and white adipocyte suspension was passed through pre-wetted 70 mm and 200 mm sterile cell strainers respectively.The cell suspension was centrifuged at 50Âg for 5 min at room temperature.The digestion buffer under the floating adipocyte layer was removed with syringes and needles.The adipocytes were rinsed with medium and centrifuged again at 50Âg for 5 min.The excess medium was removed, and the adipocytes were left in the tubes for follow-up experiments.

Quantitative RT-PCR (qRT-PCR)
RNA was extracted from adipose tissues of mice using TRIzol reagent, then reverse-transcribed to cDNA using a cDNA Synthesis Kit (TRANSGEN, Beijing, China).qRT-PCR was performed using a qRT-PCR reaction kit (TRANSGEN) on a Bio-rad CFX connect detection system (Biorad, Hercules, CA, USA), using 4e6 independent RNA samples as biological replicates.The mRNA expression of the target genes was normalized to the expression of 18S rRNA or Arbp0, and the values were compared with the control group.The primer sequences are shown in Suppl.Table S1 and coverslips.The H&E staining procedure was the same as above, except that the sections were shortly stained with hematoxylin solution and eosin solution.The slides were scanned (3DHISTECH, Budapest, Hungary) and analyzed.

Rectal temperature measurement
The tested mouse was held in one hand, and the Vaseline-lubricated probe of a digital rectal thermometer (TH212, Zhongjiao, Cangzhou, China) was gently and fully inserted into the rectum.The temperature was monitored for about 1 min until a steady reading was recorded.

Cold exposure test
Just before sacrifice, 16-17 weeks-old mice were housed one per cage in a 4 C cold chamber for 24 h (9 a.m.e9 a.m.) with unlimited access to food and water.Rectal temperature was measured before and by the end of cold exposure sessions at the same time of the day.

Serum leptin
Serum leptin concentration was measured by a mouse leptin ELISA kit (ProteinTech, Wuhan, China) according to the manufacturer's instructions.
2.17.Statistical analysis qRT-PCR results were presented as the mean þSD.Other results were presented as the mean AE SD.Images were analyzed using ImageJ.Data analysis was performed using the GraphPad Prism 9 program.Statistical significance was determined by a two-tailed unpaired t-test for comparisons between two groups or by two-way ANOVA followed by Sidak's multiple comparisons unless specifically noted.Differences with a p-value <0.05 were considered statistically significant.Representation of the p-value was *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns: not significant (p > 0.05).

Targeted mutagenesis of Fth1 in the adipose tissue had little effect on the mice's appearance
To investigate the function of ferritin in the adipose tissue (AT), we generated mice with adipocyte-specific deletion of Fth1 (Fth AKO ) by crossing Fth fl/fl mice [28] with adipoq-cre mice [29].We found that Fth1 was successfully deleted in most AT by examining the Fth1 mRNA expression (Figure 1A).Consistent with the Fth1 mRNA results, we found that FTH protein expression was greatly reduced in the interscapular brown adipose tissue (iBAT) of Fth AKO mice as compared to that of control littermates (Fth fl/fl ) (Figure 1B,D).However, the FTH protein level in the inguinal white adipose tissue (iWAT) of Fth AKO mice was only moderately lower than the control.To be certain whether this was due to mixing in some other cell types in the iWAT, we generated primary adipocyte cultures from iWAT of Fth AKO mice and analyzed their Fth1 expression.The FTH protein level was dramatically reduced in the cultured primary adipocytes from iWAT of Fth AKO mice (Figure 1C,E).Therefore, Fth1 was efficiently removed in both iBAT and iWAT.It might be that Fth expression was relatively low in the white fat, making the observed Fth1 expression more vulnerable to interference from the other cell types.There was no difference between Fth AKO and the control in the body weight or AT sizes, including those of iBAT, iWAT, and epididymal adipose tissue (eWAT) (Figure 1F,I).Nevertheless, Fth AKO mice demonstrated a decrease in the adipocyte cell size of eWAT, which was not found in the iBAT or iWAT (Suppl.Fig.S1).

Iron homeostasis of adipose tissue was altered in
Fth AKO mice Ferritin is a critical intracellular iron storage protein.To examine possible adipose iron disrupted homeostasis in the Fth AKO mice, we first inspected the total iron level in the fat tissues.We observed that total tissue iron levels were significantly lower in iBAT and slightly lower in iWAT and eWAT, which was almost certainly due to a substantial reduction in the Fth1 expression (Figure 2A).Perls/DAB iron staining of iBAT also illustrated the same result (Figure 2B).In contrast, no significant change in total iron was observed in the liver, small intestine, or blood serum (Figure 2A,C).Transferrin saturation and total iron binding capacity were unaffected either (Figure 2D,E), showing that Fth knockout in the AT does not obviously affect system iron homeostasis.In WATs, total iron levels appeared less affected, likely due to the white fat's intrinsically low iron-storage property.The labile iron pool (LIP) of adipocytes isolated from the iBAT of Fth AKO mice was higher than that of controls (Figure 2F).However, we failed to detect significant expression change of transferrin receptor (Tfrc) and ferritin light chain (Ftl) (Figure 2G,J) at the mRNA level.In the iWAT of Fth AKO mice, the expression of ferroportin (Fpn) was higher than that of the control (Figure 2H), implying disrupted labile iron homeostasis.Aconitase activities depend on FeeS cluster formation, and iron insufficiency reduces the enzymatic activity.We found that the aconitase activity was much higher in the iBAT of Fth AKO (Figure 2J, Suppl.Fig.S2B), indicating an elevated FeeS cluster assembly, consistent with the previously observed increased LIP.Collectively, these results suggest that Fth1 knockout leads to increased available iron in iBAT and iWAT.
3.3.Mild oxidative stress without obvious oxidative damage was observed in the AT of Fth AKO mice Iron is the primary cellular ROS catalyzer via the Fenton's reaction.As LIP was upregulated in the AT, we next investigated whether ROS production was induced in iBAT and iWAT of Fth AKO mice.The expression of many oxidative stress-responding genes, including Nrf2 and its downstream targets, were not affected in iBAT, while peroxiredoxin-1 (Prdx1) and superoxide dismutase 2 (Sod2) were significantly increased in iWAT (Figure 3A,B).Intriguingly, Fth1 deletion dramatically elevated the expression of stress-responsive gene heme oxygenase-1 (Hmox1) in iBAT, iWAT, and eWAT (Figure 3A,B, Suppl.Fig.S2A).We thus directly measured the cytosolic ROS level in mature adipocytes isolated from iBAT and iWAT.A slight increase in ROS was only observed in the iBAT of Fth AKO mice (Figure 3C).This slight ROS elevation appeared to produce no significant change in the level of malondialdehyde (MDA), which is the product of lipid peroxidation (Figure 3E).The remarkable increment of Hmox1 prompted us to examine other factors that could induce its expression.Surprisingly, neither the heme level nor the expression of hypoxia-inducible factor-1a (Hif1a) had significantly increased in Fth-deficient ATs compared to the control (Figure 3F,G).We speculated that Fth1 deletion altered the LIP and caused only very mild oxidative stress, which was insufficient to bring about a broad and robust ROS response.Nevertheless, this low level of ROS was sufficient to stimulate the expression of the antioxidant genes Prdx1 and Sod2 in iWAT and Hmox1 in both iBAT and iWAT.We noticed that Hmox1 induction in our experiments was about 100-fold, whereas, in Fth1 global deletion mice, which was associated with strong ROS induction, Hmox1 induction reached more than 1000fold [21].Considering that oxidative stress could cause inflammation, we also examined the expression of inflammation markers and found no changes in the expression of Il-6 or Tnf-a in ATs (Figure 3H).We, in addition, monitored the mitochondrial ROS (mtROS) level.Interestingly, unlike cytosolic ROS, mtROS slightly declined in adipocytes from iBAT and iWAT of Fth AKO mice (Figure 3D).In conclusion, our results suggest that Fth ablation induced a slight cytosolic ROS elevation but mtROS reduction in the fat.No overt lipid oxidation or inflammation was observed, likely due to the mild nature of the oxidative stress and concomitant stimulated expression of several cytoprotective genes.

Fth deletion augmented mitochondrial biogenesis and respirational activity in iWAT, and enhanced body adaptive thermogenesis
It has been proven that Hmox1 gene expression promotes mitochondrial biogenesis and supports electron transport chain (ETC) activity via activation of Pgc1, Nrf1, and Nrf2 regulated pathways [32e34].We thus measured the expression levels of Pgc1a, Nrf1, Nrf2, and ETC genes in ATs.The expression of Pgc1a, Nrf1, or Nrf2 was not upregulated.However, the expression of ETC genes (Cox4 and Atp5a1) was significantly enhanced in the Fth-deficient iWAT (Figure 4A,B).To confirm the effect of Fth deletion on mitochondrial function, we checked the respiratory activity of mitochondria isolated from Fthdeleted mouse embryo fibroblasts (MEFs) by measuring the consumption of respiration substrate NADH.Mitochondria from Fthdeleted MEFs exhibited an increased NADH consumption rate compared to those from the control MEFs (Figure 4C).We further tested the activities of mitochondrial ETC complex I and II of mitochondria isolated from iBAT or iWAT adipocytes.In the iBAT of Fth AKO , complex II activity was significantly more robust, while complex I activity also trended higher, though not statistically significant (Figure 4D).We did not observe any difference in activity of complex I or II in the iWAT.
The mitochondrial function and content in ATs are parallel with the browning of white fat and its thermogenesis function.Therefore, we inspected the expression of thermogenic genes and white fat browning markers in iBAT and iWAT.Induced mRNA expression of white fat browning markers Car4 and Dio2 was observed in iWAT of Fth AKO mice, but no significant alteration was seen in iBAT (Figure 4E,F).We then asked whether Fth knockout in the ATs affects thermogenesis in mice.We challenged Fth AKO and control mice with cold stress to elucidate their adaptive thermogenic functions.Fth AKO mice maintained slightly higher body core temperatures compared to the control in the cold exposure test (Figure 4G).The cold-induced elevation of free fatty acids in the serum was greater in Fth AKO mice (Figure 4H), suggesting augmented lipolysis in ATs and more fuel consumption for heat production.We measured the IL-6 level in serum to determine if the inflammatory effect contributed to the higher body temperature in cold conditions.The result showed no difference in serum IL-6 level (Figure 4I).Altogether, the increased mitochondrial biogenesis, mitochondrial function, and lipolysis process in ATs supported Fth AKO mice to maintain a higher body core temperature in cold environments.
3.5.Fth AKO mice exhibited an altered energy substrate preference, accompanied by increased blood glucose tolerance and insulin sensitivity An altered metabolic profile under cold stress prompted us to explore more fully possible changes in lipid metabolism with Fth AKO mice under normal conditions.We therefore used qRT-PCR to analyze expression patterns of some representative adipogenic, lipogenic, and lipogenesis genes.We found that Fth deletion significantly upregulated the expression of fatty acid binding protein 4 gene (Fabp4) and leptin (Lep) and downregulated the expression of hormone-sensitive lipase (Hsl) in iBAT (Figure 5A,B).Meanwhile, resistin (Retn), adiponectin (AdipoQ), glucose transporter type 4 (Glut4), Fabp4, and Lep were significantly upregulated in Fthdeficient iWAT (Figure 5C,D).These results suggest that Fth deletion in ATs might lead to elevated fatty acid content in iBAT and iWAT [35] and activated lipogenesis in iWAT.In agreement with gene expression data, we found that serum leptin level was also notably enhanced in Fth AKO mice (Figure 5E).As leptin and FABP4 have opposite effects on mitochondrial fatty-acid b-oxidation (FAO) function [36], we also tested whether the FAO capacity was enhanced in Fth AKO mice.The expression of b-oxidation enzyme gene Cpt1a was elevated in the iWAT of Fth AKO mice (Figure 5F).Fat and glucose metabolisms are often interconnected.We next investigated the effect of adipocyte Fth deletion on glucose metabolism in mice.We subjected these mice to IPGTTs and IPITTs.Fth AKO mice exhibited significantly increased glucose tolerance and insulin sensitivity (Figure 6AeD).Furthermore, we found that Fth AKO mice had a lower average respiratory exchange ratio (RER) compared to control littermates, suggesting a shift of metabolic substrate from carbohydrates toward fats (Figure 6E).Additionally, we tested liver function markers serum AST and ALT, which were also related to insulin resistance [37].Both serum AST and ALT levels of Fth AKO mice were significantly lower than the control, though all were within the normal range (Figure 6F,G).
Elevated Lep, AdipoQ, and Glut4 expression levels in Fth AKO mice were consistent with their increased insulin sensitivity and glucose tolerance [38e40].Increased expression of Lep, AdipoQ, and altered metabolism substrate were in line with the enhanced FAO in Fthdeficient ATs [41,42].Taken together, our results indicated that Fth adipocyte knockout increased the expression of several lipogenic  genes in ATs, which was plausibly associated with increased glucose tolerance, insulin sensitivity, and altered energy substrate.
3.6.Vitamin E supplementation improved the glucose tolerance and insulin resistance of both Fth AKO and control mice and diminished their difference To verify whether the increment of glucose tolerance and insulin sensitivity in Fth AKO mice was related to oxidative stress, we fed both Fth AKO and control mice with a diet supplemented with an antioxidant, vitamin E (0.5 g a-tocopherol/kg of food), for three weeks and performed IPGTTs and IPITTs on them.Glucose tolerance and insulin sensitivity were dramatically increased in vitamin E-supplemented mice, both in Fth AKO and control mice, compared to Fth fl/fl fed with a normal diet (ND) (Figure 7AeD).More importantly, the difference in glucose tolerance and insulin sensitivity between Fth AKO and control mice was eliminated by vitamin E treatment.To confirm that excess vitamin E indeed reduced oxidative stress in AT, we examined the expression level of Hmox1.In iBAT and iWAT of vitamin E-supplemented mice, the expression of Hmox1 was significantly reduced (Figure 7E,F).Previous studies suggest that ROS can both enhance and reduce insulin sensitivity [43,44].Our results indicate that vitamin E supplementation reduced oxidative stress and promoted insulin sensitivity in both Fth AKO and control mice.Moreover, vitamin E attenuated the difference in insulin sensitivity between ND-fed Fth AKO and control mice, suggesting that the original difference is connected to oxidative stress.

DISCUSSION
The role of ferritin in influencing glucose and lipid metabolism in adipose tissue is not known.Here we report that ferritin in adipocytes plays an important role in maintaining adipocyte iron homeostasis and whole-body glucose metabolism.Loss of Fth in adipocytes reduced intracellular iron storage and slightly increased the labile iron level and oxidative stress, but dramatically induced the expression of Hmox1, concomitant with enhanced mitochondrial ETC gene expression and respirational capability.Fth AKO mice exhibited increased systemic glucose tolerance and insulin sensitivity, shifting substrate preference toward fats from carbohydrates.While some aspects of metabolic disruption were pronounced, we noticed others only partially changed.We attribute this to the mild shift of iron and ROS metabolism due to FTH loss in the adipocyte.It is possible that some genes, such as Hmox1, are more sensitive to the change than others.Ferritin knockout resulted in two major alterations.One is mildly elevated cytosolic iron and ROS, and another is strengthened mitochondrial functions.The latter might be the result of elevated labile iron because more iron available may enhance the formation of ETC complexes and, consequently, activity.Either one or a combination of both changes may be responsible for the observed improved energy metabolism phenotype.Increased ROS has been reported to be beneficial for adipose physiology [45].On the other hand, more robust mitochondria favor fatty acid oxidation or oxygen consumption, allowing more glucose conversion to fatty acids.Elevated cytosolic iron led to increased ROS in the cytosol.However, in the mitochondria, ROS was reduced.Studies report that they have opposite effects on mitochondrial function and animal lifespans [46,47].What has caused the decreased mtROS?We speculate that the increased mitochondrial respiring capability, enabled by increased FeeS cluster synthesis and enhanced ETC complex's activity, might have dampened the mtROS generation.Alternatively, it has been reported that cytosolic ROS-induced HO-1 could enter mitochondria and reduce mtROS [48].
In agreement with other Fth knockout or knockdown models, deletion of Fth in adipocytes reduces intracellular iron storage, increases cytosolic LIP, and oxidative stress via changes in the expression of some iron-responsive genes or activity [23,24,49,50], suggesting the iron storage function is closely tied with the protective function against oxidative stress.FTH ablation in this study does not cause severe iron deficiency or iron overload in ATs, which is likely due to the compensatory functions of other iron homeostasis genes as well as local iron exchange, facilitated by iron and ferritin exchange of adipocytes with local macrophage [27,51].Meanwhile, iron distribution in cellular compartments is perturbed in FTH-deficient AT.The increased aconitase activity suggests increased FeeS cluster assembly, implying sufficient iron supply, which is in line with the results of LIP measurements in iBAT.Consistently, activity assays of mitochondrial ETC complex I and II and NADH consumption assay suggest augmented mitochondrial ETC function and more active energy metabolism [52,53].
Our data shows that Fth deletion leads to statistically valid changes in cytosolic ROS, LIP, and aconitase activity only in iBAT and not in iWAT.A plausible explanation of such differential effects of the Fth gene knockout is the large difference in iron demand and iron flux between brown fat and white fat [54,55].Due to the energy metabolism and thermogenic function of iBAT, a large amount of iron is required; on the contrary, high iron flux is not essential in iWAT.
It is well established that iron in adipocyte tissues plays a crucial role in supporting mitochondrial biogenesis and regulating organismal thermogenesis [3,11,56].Studies have shown that iron is necessary for adipocyte differentiation and the thermogenic function [11,56,57].Iron deficiency can impair the formation of beige adipocytes and reduce the thermogenic ability of brown adipocytes [2,4].Adipocyte-specific deletion of transferrin receptor 1 (Tfrc) in adipocytes shows reduced iron content in AT, damaged mitochondrial function, and impaired brown fat thermogenesis [4,9].In this study, Fth deletion in AT increases the expression of some mitochondrial ETC genes and white fat browning marker genes, leading to enhanced heat-generating capacity and intensified adaptive thermogenesis.Adaptive thermogenesis is also improved in Fth AKO mice, as indicated by the higher body temperature under cold stress.However, the gene expression of Ucp1 is not induced compared to controls.We explain that Ucp1 expression does not directly lead to the increase in thermogenesis; Instead, it might be more activated in Fth1 knockout BAT [58].Further investigation is needed to elucidate the mechanism of increased adaptive thermogenesis in Fth AKO mice.Many previous Fth gene knockout studies have reported the phenotype of upregulated Hmox1 expression and attributed this to oxidative stress [22,59,60].Hmox1 responds to various stimuli, including but not limited to ROS, hypoxia, heme, free fatty acids, and heavy metals.Our results suggest that elevated Hmox1 expression under FTH deficiency is also induced by ROS rather than hypoxia or heme, though we could not rule out other possibilities.The observation that no appreciable lipid peroxidation damage is detected could be attributed to the fact that the very mild oxidative stress might be well protected by oxidative response genes such as Hmox1 induction to prevent and mitigate lipid oxidation damage.In other words, the upregulation of Hmox1 expression may compensate for the antioxidant function of ferritin.It is also possible that upregulation of Hmox1 expression occurs in AT macrophages, and accelerated heme recycling helps reduce ROS and damage in the adipocytes [61].The upregulated Hmox1 has been proven to improve mitochondrial biogenesis and ETC gene expression [32e34], consistent with our findings.
Our data suggests that the expression of several adipokines is upregulated in ATs of Fth AKO mice.The upregulated expression of AdipoQ may be related to increased mitochondrial function and Hmox1 [62].Increased mitochondrial biogenesis and function may be related to adiponectin synthesis [63].Despite one research that disagrees with a direct HO-1-adiponectin axis hypothesis [64], massive studies provide evidence of simultaneous upregulation of the two genes [65e67].In FTH-deleted iWAT, the qRT-PCR analyses demonstrate the activation of both browning and lipogenesis.Browning of WAT is often related to increased lipolysis and energy expenditure, which seems to conflict with lipogenesis.However, several studies report simultaneous up- [68,69] or down-regulated [70] browning and lipogenesis.The increased insulin sensitivity is aligned with upregulated lipogenesis genes and Lep in the iWAT of the Fth AKO mice.Meanwhile, the body weights and fat sizes show no difference between the Fth AKO and control mice, which demonstrates the net lipid gain is comparable.We speculate that lipogenesis, browning, and lipid turnover are enhanced and reach a new steady state in FTH-deleted iWAT.The observation of synchronously upregulated Fabp4 and Lep expression suggested a sign of hypoxia [71,72] or inflammation [73,74], but we did not see any increase in hypoxia or inflammation markers in our study.Further studies are needed to clarify the mechanisms.It has become a consensus that ROS causes insulin resistance.However, a growing number of studies suggest that mildly increased ROS could improve insulin sensitivity [44,75].In this study, we found no lipid peroxidation damage in Fth-deficient AT, and the cytosolic ROS increment is moderate.An appropriate increase in ROS is also beneficial for the thermogenic functions of mitochondria.Coupled with the protective effect of induced antioxidant genes in FTH-deleted AT, we believe that the slightly increased ROS may have beneficial effects on insulin sensitivity and thermogenesis.Vitamin E is a potent antioxidant and ROS scavenger.Up to now there is no consistent conclusion on the effect and mechanism of vitamin E on mice's insulin sensitivity [76e 79].Our data suggests that vitamin E robustly reduces oxidative stress in ATs and promotes insulin sensitivity in both Fth AKO and control mice.Additionally, vitamin E supplementation erases the difference in insulin sensitivity between Fth AKO and control mice, further supporting that the enhanced insulin sensitivity caused by Fth AKO is associated with oxidative stress.Both slight ROS elevation and ROS mitigation may promote insulin sensitivity: in Fth AKO , a slight ROS increase may underlie the observed insulin sensitivity and glucose tolerance, but in vitamin E-treated mice, ROS is both suppressed in Fth AKO and control mice, leading to a further increment of insulin sensitivity and glucose tolerance in them.Noteworthy is that Hmox1 expression is not completely normalized after vitamin E treatment.This raises the question of how Hmox1 was induced in Fth AKO , whether vitamin E completely suppressed the ROS or even more, and whether vitamin E had other mechanisms regulating insulin sensitivity in addition to neutralizing the ROS.
Studies have shown a cross-talk between iron and glucose metabolism [1].Iron may indirectly interfere with glucose metabolism and insulin resistance [80,81].Meanwhile, iron level affects the synthesis and the release of adipokines, including adiponectin and leptin, which have a strong influence on glucose metabolism and insulin resistance [17,82,83].As we observed a manifest increase in glucose tolerance and insulin sensitivity in addition to changes in the expression of several adipokines as the consequence of adipocyte-specific Fth1 knockout, we believe overall the level of free iron is well regulated and the beneficial consequences caused by Fth1 deletion, including mild ROS increment, Hmox1 induction, and mitochondrial function improvement, overrides possible detrimental consequences caused by slightly increased LIP.
In conclusion, our study provides valuable insights into the role of FTH in adipocytes in modulating glucose and lipid metabolism via influencing iron homeostasis and mitochondrial function.By eliminating Fth expression, we observed significant alterations in glucose tolerance, insulin sensitivity, and adaptive thermogenesis.Our findings suggest that ferritin may serve as a potential therapeutic target for obesity and diabetes.However, the mechanism of ferritin or iron regulating some of the adipokines remains unclear and warrants further investigation.

Figure 1 :
Figure 1: Fth AKO mice presented no apparent alteration in body weight or fat size.(A) qRT-PCR analysis of Fth expression in iBAT, iWAT, and eWAT of Fth AKO and Fth fl/fl mice (n ¼ 4, data pooled from two independent experiments).(B,D) Representative western blot images and quantifications of iBAT and iWAT from Fth AKO and Fth fl/fl mice (n ¼ 3e4, data pooled from two independent experiments).(C,E) Representative western blot images and quantifications of primary adipocytes derived from iWAT (n ¼ 4, data pooled from two independent experiments).(F) Body weights of Fth AKO and Fth fl/fl mice were measured each week for 14 weeks (n ¼ 5).(G) Fth AKO and Fth fl/fl mice were sacrificed at 20 weeks, and representative images of iBAT, iWAT, and eWAT were shown.(I) Weights of iBAT, iWAT, and eWAT dissected from Fth AKO and Fth fl/fl mice were measured (n ¼ 7, from two independent experiments).Data was analyzed by two-tailed unpaired t-tests.*p < 0.05, **p < 0.01, and ****p < 0.0001.

Figure 2 :
Figure 2: Iron homeostasis of the adipose tissue was altered in Fth AKO mice.(A) Total non-heme iron of the iWAT, iBAT, eWAT, liver, and small intestine in Fth AKO and Fth fl/fl mice (n ¼ 4e5, from two independent experiments).(B) Representative Perls/DAB iron staining (brown) and hematoxylin staining (blue) images of iBAT of Fth AKO and Fth fl/fl mice; scale bar, 100 mm.(CeE) Serum iron, total iron-binding capacity, and transferrin saturation levels of Fth AKO and Fth fl/fl mice (n ¼ 4).(F) Cytosolic LIP levels of mature adipocytes isolated from iBAT and iWAT were detected with the DCFH-DA probe and normalized by protein concentrations (n ¼ 4e5, from two independent experiments).(GeI) qRT-PCR analysis of Tfrc, Fpn, and Ftl expression in iBAT, iWAT, and eWAT of Fth AKO and Fth fl/fl mice (G: n ¼ 3e4; H: n ¼ 4e6; I: n ¼ 4e5, from two independent experiments).(J) Aconitase activity of iBAT of Fth AKO and Fth fl/fl mice (n ¼ 3e4, from two independent experiments).Data was analyzed by two-tailed unpaired t-tests.*p < 0.05, **p < 0.01, and ns: not significant (p > 0.05).(For interpretation of the references to color/colour in this figure legend, the reader is referred to the Web version of this article.)

Figure 3 :
Figure 3: Fth KO in the adipose tissue induced a mild cytosolic ROS elevation and mitochondrial ROS reduction but no overt lipid oxidation or inflammation.(A,B) qRT-PCR analysis of Hmox1, Nrf2, Nqo1, Prdx1, Sod1, and Sod2 expression in iBAT and iWAT of Fth AKO and Fth fl/fl mice (n ¼ 4e6, from three independent experiments).(C) Summary data of flow cytometry analysis of cytosolic ROS tested by DCFH-DA probe in iBAT and iWAT adipocytes from Fth AKO and Fth fl/fl mice (n ¼ 3, from three independent experiments).(D) Mitochondrial ROS level probed by MitoSOX in adipocytes isolated from the iBAT and iWAT of Fth AKO and Fth fl/fl mice, normalized to protein content (n ¼ 3 mice, from two independent experiments).(E) MDA levels in iBAT, iWAT, and eWAT of Fth AKO and Fth fl/fl mice, normalized to protein content (n ¼ 3e5, from two independent experiments).(F) Western blots of HIF1a expression in iBAT and iWAT from Fth AKO and Fth fl/fl mice.(n ¼ 4, from two independent experiments).(G) Heme concentration in iBAT and iWAT of Fth AKO and Fth fl/fl mice, normalized to protein content (n ¼ 3, from two independent experiments).(H) qRT-PCR analysis of Il-6 and Tnf-a expression in iBAT and iWAT of Fth AKO and Fth fl/fl mice (n ¼ 4, from two independent experiments).Data from Figure 3C was analyzed by paired t-tests.Other data was analyzed by two-tailed unpaired t-tests.*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Figure 4 :
Figure 4: Fth1 deletion in the adipose tissue caused upregulation of mitochondrial biogenesis and respiratory capacity and a tendency of iWAT browning.(A,B) qRT-PCR analysis of Pgc1a, Nrf1, Ndufs1, Sdhc, Cox4, Cox8b, and Atp5a1 expression in iBAT and iWAT of Fth AKO and Fth fl/fl mice (n ¼ 4e6, from two independent experiments).(C) NADH consumption assay of mitochondria isolated from Fth À/À and Fth fl/fl MEFs (n ¼ 5, from 3 independent experiments).(D) The activities of mitochondrial ETC complex I and II of mitochondria isolated from mature adipocytes of iBAT and iWAT (n ¼ 3e5, from two independent experiments).(E,F) qRT-PCR analysis of Ucp1, Car4, Dio2, Elovl3, Cidea, and Serca2 expression in iBAT and iWAT of Fth AKO and Fth fl/fl mice (E: n ¼ 4e6; F: n ¼ 3e6, from three independent experiments).(G) Rectal temperature of Fth AKO and Fth fl/fl mice housed at room temperature or 4 C (n ¼ 6, from three independent experiments).(H) Free fatty acid in mice serum before and after cold exposure test (n ¼ 4e5, from two independent experiments).(I) Serum Il-6 of Fth AKO and Fth fl/fl mice housed at 4 C (n ¼ 3).Statistical significance of the difference between Fth À/À and Fth fl/fl in Figure 4C was determined by two-way ANOVA followed by Sidak's multiple comparisons.Other data was analyzed by two-tailed unpaired t-tests.*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns: not significant (p > 0.05).

Figure 6 :
Figure 6: Fth AKO mice had increased glucose tolerance, insulin sensitivity, and altered energy substrate preference.(A,B) IPGTT and (C,D) IPITT results of Fth AKO and Fth fl/ fl mice and AUC calculated from them (n ¼ 7e9, from three independent experiments).(E) The respiratory exchange ratio (RER, VCO 2 /VO 2 ) of Fth AKO and Fth fl/fl mice calculated from VCO 2 and VO 2 monitored with metabolism cages over a 24-h period (n ¼ 5e6).(F,G) Serum AST and serum ALT concentrations in Fth AKO and Fth fl/fl mice (n ¼ 4e5, from two independent experiments.).Data from Figure 6A,C was analyzed by two-way ANOVA followed by Sidak's multiple comparisons.Other data was analyzed by two-tailed unpaired t-tests.*p < 0.05, **p < 0.01, and ***p < 0.001.
2.6.Labile iron pool (LIP) measurementMature adipocytes were prepared as described above.The number of cells was determined by a cell counter (Nexcelom, Cellometer Mini, The protein concentration of the samples determined by a BCA kit (Beyotime) was used for subsequent calculation.Malondialdehyde concentration was measured with an MDA assay kit (Beyotime) based on the colorimetric reaction of MDA and thiobarbituric acid producing a red product.2.12.Paraffin embedding, sectioning, Prussian blue and DAB staining, and H&E staining The dissected tissues were fixed with 4 % paraformaldehyde for 24 h at room temperature.After dehydration, they were paraffin-embedded and sectioned at 8 mm.Paraffin sections on slides were deparaffinized and rehydrated.Tissue sections were flooded with the Prussian blue staining buffer (0.15 g potassium hexacyanoferrate, 0.4 ml 37 % HCl, 7.1 ml H 2 O) for 60 min in the dark, then were washed in H 2 O four changes for 5 min each.Sections were then stained with DAB (diaminobenzidine) in the dark for about 40 min and washed in H 2 O four changes for 5 min each.They were stained shortly with hematoxylin solution and washed well in H .2.11.Malondialdehyde (MDA) assayAdipose tissues were homogenized in PBS and centrifuged at 10,000Âg, 4 C for 10 min.The fat cakes and the bottom debris were removed. 2 O.The slides were dehydrated in graded alcohols and cleared in xylene.They were air-dried and mounted with neutral balsam mounting medium (BBI Life Science, Shanghai, China) Lu: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing e original draft.Shanshan Guo: Methodology, Resources, Visualization.Jialin Zhao: Data curation, Methodology.Xiaoting Wang: Methodology.Bing Zhou: Conceptualization, Formal analysis, Funding acquisition, Project administration, Supervision, Writing e review & editing.