Mitochondrial Hormesis links nutrient restriction to improved metabolism in fat cell

Nutrient restriction leads to mitonuclear OxPHOS imbalance concomitantly to mitochondrial remodeling in white and beige adipose cells

White (WAT) and brown (BAT) adipose tissue display a diverse mitochondrial OxPHOS equipment, which specifically reflects the cellular and tissue functionality.

Accordingly, in BAT crude mitochondria from 1, 3 and 6 months old mice, we observed higher OxPHOS protein levels than epididymal WAT (eWAT) (Fig. 1A and Fig. S1A). In order to compare the transcriptional level of mitochondrial ^mtDNA- and nuclear ⁿDNA-encoded OxPHOS genes in white and brown adipose depots, we performed an OxPHOS gene expression array. As showed in Fig. 1B and S1B, both eWAT and BAT revealed higher expression of ^mtDNA- than ^mtDNA-encoded OxPHOS mRNA. However, the transcriptional efficiency of BAT (Fig. 1C and Fig. S1C) and T37i brown adipocytes (Fig. S1D) was mainly shifted toward ^mtDNA-encoded OxPHOS genes. Given that mitonuclear imbalance is involved in the adaptive response to metabolic requirements [18-19], we analyzed whether it could occur in adipocytes upon nutrient restriction (NR). As showed in Fig. 2A and Fig. S1E, NR leads to mitonucle ar protein imbalance in 3T3-L1 adipocytes, as evidenced by the increased ratio between the ^mtDNA-and the ⁿDNA-encoded OxPHOS proteins. We also revealed a significant increase in the mRNA ratios between ^mtDNA-encoded MTCO1 and ⁿDNA-encoded COX4b (Fig. 2B). The analyses of other OxPHOS gene expression ratios revealed the same trend (Fig. S1F). The results obtained by the calculation of the OxPHOS gene expression ratio revealed increased mitochondrial transcription efficiency in white adipose depots (Fig. S1G). Conversely, T37i adipocytes had not increased ratios between MTCO1 and ⁿDNA-encoded NDUFB8 proteins (Fig. 2A) as well as MTCO1 and Cox4b mRNAs (Fig. 2B). Furthermore, also beige adipose cells up-regulated the ratio between ^mtDNA-and ⁿDNA-encoded proteins and mRNAs after NR (Fig. 2C and 2D). The disturbance of the delicate mitonuclear balance triggers the induction of ⁿDNA-encoded mitochondrial stress factors such as antioxidant enzyme SOD2 and protease ClpP [16, 20]. As expected, we found that NR increased SOD2 and ClpP levels in 3T3-L1 and X9 but not in T37i adipocytes, where on the contrary SOD2 and ClpP levels were diminished (Fig. 2E and 2F). Similarly to immortalized adipocytes, primary adipocytes displayed an increased ratio between ^mtDNA-and ⁿDNA-encoded genes (Fig 2G) as well as increased ClpP mRNA levels (Fig 2H). Primary adipocytes derived from BAT did not show this trend (Fig 2G and 2H). Finally, we investigated the mitochondrial adaptive response in adipose tissues of 20 h fasted mice. An increased ratio between ^mtDNA-encoded MTCO2 and ⁿDNA-encoded SDHA proteins as well as an accumulation of SOD2 (Fig 2I) and ClpP (Fig 2J) was disclosed in iWAT but not in BAT crude mitochondrial fractions. Expectedly, 20 h fasting led to an increased mRNA ratio between ^mtDNA-and ⁿDNA-encoded OxPHOS genes in eWAT (Fig 2K) with respect to ad libitum fed mice.

Mitochondria undergo continuous cycles of selective fusion and fission, and nutrients as well as hormones efficiently affect these processes [21]. Although NR increases oxidative metabolism in white adipocytes [4], its effect on adipocyte mitochondrial dynamic is still unknown. Through morphometric analysis of confocal microscopy images we observed that mitochondria of 3T3-L1 adipocytes appeared higher in number but smaller in per imeter and area, implying the occurrence of mitochondrial fragmentation (Fig 3A). In line with this hypothesis, 3T3-L1 adipocytes accumulated Drp1 and Fis1 in total protein extracts (Fig 3B) as well as in crude mitochondrial fractions (Fig 3C), while the canonical marker of mitochondrial fusion OPA1 underwent a progressive reduction (Fig 3B). The mitochondrial dynamic observed in 3T3-L1 cells was not accompanied by variations in total levels of the mitochondrial protein translocase TOMM20 (Fig 3B). Nutrient refill with complete medium (CM) was able to restore the basal level of Drp1 and Fis1 localized at mitochondrial level in 3T3-L1 adipocytes (Fig 3C). Differently, no significant changes in mitochondrial fragmentation markers were detected in T37i adipocytes subjected to NR (Fig S1H). By confocal microscopy analysis we revealed a higher co-localization degree (Pearson coefficient, r = 0.96) between Drp1 and TOMM20 in 3T3-L1 adipocytes upon NR (Fig 3D) and increased mitochondrial fission was also observed in starved X9 adipocytes (Fig 3B). These results were recapitulated in crude mitochondria isolated from 20 h fasted mice in which we detected an increased level of Drp1 in eWAT mitochondria (Fig 3E). Next we quantified mitochondrial amount and membrane potential (ΔΨM) by staining 3T3-L1 and X9 adipocytes with MitoTracker Green (MTG) and MitoTracker Red (MTR), respectively. Interestingly, NR increased the degree of MTG positive 3T3-L1 (Fig 3F) and X9 adipocytes (Fig 3I). An increase of ΔΨM was evidenced, although at lesser extent with respect to mitochondrial number (Fig 3G and 3J). Such ΔΨM increase was apparent, as established by normalizing MTR fluorescence (sensitive both ΔΨM and mitochondrial number) with MTG value, which is sensitive only to mitochondrial number. In particular, as reported in Fig 3H and 3K, the calculation of Red-to-Green ratio (MTR/MTG) evidenced that starved adipocytes undergo mitochondrial depolarization. The same results were obtained by staining mitochondria with the cardiolipin fluorescent probe nonyl acridine orange (indicative of mitochondrial number) and tetramethyl rhodamine (sensitive to ΔΨM) (Fig S2A). In line with these data, primary beige adipocytes displayed a reduction in co-localization points between MTG and MTR after NR (Fig 3L). Mitochondrial changes were also accompanied by a diminished O₂ consumption along with a drop of ATP level in 3T3-L1 white adipocytes (Fig 3M and 3N). To verify whether such variations were due to the deprivation of nutrients, we refill cells with CM. Under this condition, we found restoration of O₂ consumption without a concomitant recovery of ATP (Fig 3M and 3N), suggesting the occurrence of an uncoupling process (Fig 3O), accompanied by improved OxPHOS machinery (Fig 2).

Starvation induces UCP1 independently of hormonal signaling cascade

Augmented mitochondrial fragmentation and uncoupling have been identified as key features that accompany the browning process [22]. As reported in Fig 4A, the expression of brown fat-related markers were increased in 3T3-L1 adipocytes upon NR to levels comparable to those reached upon treatment with the β-adrenoreceptor agonist isoproterenol. Concomitantly, UCP1 protein levels increased both during NR and isoproterenol treatment (Fig 4B). Furthermore, the newly synthetized UCP1 resulted localized into mitochondria (Fig S2B). To more specifically link the development of a brown-like phenotype to NR, we cultured starved 3T3-L1 adipocytes with CM (re-feed). As demonstrated in Fig 4C, nutrient refill was effective in reverting the up-regulation of PGC-1α, UCP1 and Cidea. In accordance with the above-described data, X9 adipocytes quickly responded to NR by up-regulating (Fig 4E and 4F). Conversely, we revealed a down-brown-related markers (Fig 4B and 4D). UCP1 was regulation of UCP1 level both in immortalized T37i also induced in primary beige and white adipocytes (Fig 4B) and primary brown adipocytes (Fig 4G).

Generally β-adrenoreceptor stimulation culminates in UCP1 expression and lipolysis in adipose cells by the activation of cAMP/PKA pathway [22]. To study the involvement of cAMP/PKA signaling cascade during NR, we checked the phosphorylation state of hormone sensitive lipase (HSL) on PKA-target serine (i.e. ser⁶⁶⁰). Differently from isoproterenol stimulation, NR did not induce changes in phospho-active levels of HSL on ser⁶⁶⁰ (HSLpS660) (Fig 4B), suggesting that 3T3-L1 and X9 adipocytes develop a brown-like phenotype independently of adrenergic stimulation. We next analyzed UCP1 levels in mitochondria isolated from tissues of ad libitum fed or fasted mice. As showed in Fig 4H, 20 h fasting was efficient in promoting UCP1 up-regulation only in white and beige adipose depots. However, given that in an in vivo system, the occurrence of noradrenaline modulation could affect the UCP1 level in adipose tissue, we intraperitoneally injected a selective β3 adrenoreceptor antagonist (SR59230A) prior to fasting. Notably, although SR59230A was effective in limiting fat mass loss during fasting (Fig S2C), it did not completely restrain UCP1 increase in iWAT (Fig 4I and 4J).

^mtROS trigger mitochondrial responses in white and beige adipocytes

Mitochondrial ROS (^mtROS) production appears to be a tightly regulated process that is essential for tuning the magnitude or effectiveness of the mitochondrial adaptive response [23]. NR is able to impinge ^mtROS that are key mediators of the signal transduction pathway culminating in the metabolic adaptation of 3T3-L1 adipocytes [4]. Similarly to 3T3-L1, we found an increased ^mtROS production also in X9 cells, while T37i adipocytes were unresponsive (Fig 5A). Interestingly, the lack of oxidative phenomena in brown adipocytes observed after ^mtROS-generating treatments (i.e. fasting) could be linked to basal higher mitochondria antioxidant defensive proteins (SOD2 and UCP1) in BAT with respect to WAT (Fig S3B). Successively, to dissect the role of ^mtROS in orchestrating mitochondrial stress response, we treated 3T3-L1 and X9 adipocytes with the antioxidant N-acetyl cysteine (NAC) prior to NR. As expected, NAC was able to prevent ^mtROS production both in 3T3-L1 and X9 adipocytes (Fig S3C) and it was effective in limiting SOD2 and UCP1 proteins up-regulation (Fig 5B). These results were recapitulated by overexpressing SOD2 [SOD2(+)] in 3T3-L1 adipocytes (Fig S3D). Furthermore, antioxidant-boosting strategies (NAC treatment or SOD2 overexpression) were also effective in limiting Drp1 induction (Fig 5B and S3D) and mitochondrial fragmentation (Fig 5C). Our data describe the ability of ^mtROS also in governing mitochondrial adaptive responses to NR in white and beige adipocytes. Indeed, NAC treatment prevented the alteration of the ratio between ^mtDNA-and ⁿDNA-encoded imbalance both in 3T3-L1 (Fig 5D) and X9 (Fig 5E and 5F) adipose cells. NAC was also able to prevent UCP1 and ClpP mRNA elevation in 3T3-L1 (Fig 5G) and X9 adipocytes (Fig 5H). Mitochondrial depolarization in 3T3-L1 adipocytes was also dampened (Fig 5I) in line with unchanged UCP1 levels.

ⁿFoxO1 mediates the retrograde communication enforcing mitochondrial adaptive responses in white and beige adipocytes

We previously showed that ^mtROS produced during NR in 3T3-L1 adipocytes are key mediators in forcing FoxO1 into nuclear compartment [4]. In line with this assumption, NR elicited the migration of FoxO1 into nuclei both in white and beige fat cells and NAC sup-plementation was able to restrain FoxO1 redistribution (Fig 6A and 6B). In the opposite manner, brown fat cells displayed a diminished ⁿFoxO1 level upon NR (Fig 6A and 6B). Nuclear FoxO1 is the up-stream mediator of transcription of metabolic genes conferring resistance to NR [17, 24]. Furthermore, nuclear form of FoxO1 induces PGC-1α, UCP1, SOD2 [25] as well as ATGL gene transcription [4]. To better understand whether FoxO1 was implicated in the retrograde communication of starved adipocytes, we initially down-regulated it in 3T3-L1 adipocytes (iFoxO1). As described in Fig 6C and 6D, an impaired up-regulation of brown fat-related genes in white iFoxO1 adipocytes, was observed. Concomitantly, we found that upon NR iFoxO1 adipocytes had lower level of mitochondrial fragmentation marker Drp1 than controls (Fig 6D). Similarly, down-regulation of FoxO1 in X9 beige adipocytes resulted in decreased level of Drp1, SOD2 and UCP1 (Fig 6E).

In white adipocytes, mitochondrial complex I inhibition by non-toxic amount of rotenone induces ^mtROS production prompting a pseudo-starvation response [4]. In line with our previous findings, rotenone impinges FoxO1 into nuclei of white adipocytes (Fig S4A) and this event is associated with SOD2 and UCP1 induction (Fig S4B and S4C). Differently, in T37i brown fat cells we did not observe the same responses (Fig S4A and S4B).

To finally implicate nuclear redistribution of FoxO1 in the uncoupling adaptive response of white adipocytes, we transfected 3T3-L1 adipocytes with a form of FoxO1 that is retained within the cytoplasm (cyt-FoxO1, GFP-tagged) or with a constitutively active form of FoxO1 (Myc-tagged) that is specifically localized within nuclei (nuc-FoxO1) (Fig S4D). Western blot analysis indicated that 3T3-L1 cells overexpressing nuc-FoxO1 up-regulate brown-related mitochondrial markers such as UCP1, Drp1 and SOD2 proteins (Fig 6F and Fig 6G). In line with the enhancement of the mitochondrial functionality coupled with the UCP1 induction, we detected enhanced oxygen consumption (Fig 6H) and increased uncoupling ratio (Fig 6I) exclusively in adipocytes overexpressing nuc-FoxO1. Reduced ΔΨM in nuc-FoxO1 cells further corroborates the occurrence of a brown fat-like function in white adipocytes (Fig 6J).

Animals and treatments

We conducted all mouse experimentations in accordance with accepted standard of humane animal care and with the approval by relevant national (Ministry of Health) and local (Institutional Animal Care and Use Committee, Tor Vergata University) committees. CD1 and C57BL/6 adult male mice were purchased from Harlan Laboratories S.r.l. (Urbino, Italy).

For in vivo experiment, twelve CD1 mice were randomly divided into two age-matched (4 months old) groups: ad libitum fed (Ctr) by standard diet and 20 h fasted (NR). Two C57BL/6 male mice per group were sacrificed to compare mitochondrial OxPHOS proteins in mitochondrial-derived eWAT and BAT at different ages (1, 3 and 6 months old). Adult C57BL/6 mice were randomly divided into three age-matched (4 months old) groups: ad libitum fed (Ctr) by standard diet; 20 h fasted and SR59230A (Santa Cruz Biotechnology, Heidelberg, Germany) treated mice. In this period each NR mouse had free access to water and was kept on a 12:12 h light:dark cycle. For treatment with the selective β₃-adrenoreceptor antagonist, mice were intraperitoneally (ip) injected with 10 mg/kg of SR59230A dissolved in a physiological saline solution. Both ad libitum and NR group was ip injected with saline solution. SR59230A was injected 6 h prior and at the beginning of fasting experiment. Mice were sacrificed by cervical dislocation, and tissue samples were adequately collected. For total protein and mRNA extraction, epididymal (eWAT), inguinal (iWAT) and brown (BAT) adipose tissue were explanted and immediately processed.

Cell lines, differentiation, primary cell isolation, transfections and treatments

3T3-L1 cell lines were purchased from ATCC (Manassas, VA, USA). T37i cell line was gently provided by Prof. Marc Lombes (Inserm U693, Paris, France) and X9 cells were gently provided by Prof. Bruce Spiegelman (Harvard Medical School, Boston, MA, USA). NR experiments were carried out as previously described [4, 17]. 3T3-L1-white adipocytes were cultured as previously described [4]. T37i cells were cultured and differentiated as described by Nakae et al., 2008 [40]. Immortalized X9-beige cell line was cultured and differentiated as reported by Ye et al. [9]. All experiments were performed in fully differentiated adipocytes (day 8).

For isolation of primary adipocytes, the stromal-vascular fraction of the epididymal (eWAT), inguinal (iWAT) and brown (BAT) fat pads of 5 weeks old male C57BL/6 mice were prepared and differentiated for 8 days, as reported by Fisher at al., 2012 [41].

Fully differentiated 3T3-L1-white or X9-beige adipocytes were transfected with FoxO1 (Santa Cruz Biotechnology) or scramble siRNAs (Santa Cruz Biotechnology) by using DeliverX Plus kit (Affymetrix, Santa Clara, CA, USA). Alternatively, they were transfected with SOD2 cDNA pcDNA3.1 plasmid or pCMV-DsRed plasmid (Clontech Lab, Mountain View, CA, USA) by using Turbofect Transfection Reagent (Thermo Scientific, Rockford, IL, USA). Adipocytes were subjected to NR 48 h after transfection. Plasmid nuc-FoxO1 pCMV5 (Addgene #12143) and cyt-FoxO1 (7KQ mutant, gently provided by Prof. Accilli D., Dept. of Medicine, Columbia University, New York, NY) were used to transfect 3T3-L1 cells.

Nutrient refill (re-feed) was performed by culturing starved adipocytes in complete culture medium (CM) for 1 h. Isoproterenol (Sigma-Aldrich) and N-acetyl cysteine (NAC) (Sigma-Aldrich) were dissolved in PBS and added in culture medium at final concentration of 10 μM and 2 mM, respectively. NAC was added 1 h prior to NR and maintained throughout the experiment. Rotenone (Sigma-Aldrich) was solubilized in DMSO and added in culture medium at final concentration of 0.1 μM.

Western blotting and OxPHOS protein ratio

Western blotting analysis was performed as previously described [17]. OxPHOS protein ratio was evaluated by Western blot calculating the stoichiometric ratio between nuclear-encoded and mitochondrial-encoded proteins after densitometric analysis of the immunoreactive bands. Protein concentration was determined by the method of Lowry.

RT-qPCR and OxPHOS gene expression analysis

RT-qPCR analysis was carried out as previously described [4]. To calculate OxPHOS gene expression ratio, nuclear-encoded OxPHOS mRNA levels were compared to mitochondrial-encoded OxPHOS mRNA. The relative mRNA levels were determined by using the 2-^ΔΔCt method and were normalized to β-actin. ΔCt values were used to generate Heat Map by Excel software.

Nuclear, cytoplasmic and mitochondrial fractioning

Nuclear and cytosolic fractions were obtained from cells and mice tissues using a commercially available NE-PER® extraction kit (Thermo Scientific). Mitochondria from differentiated adipocytes and eWAT were obtained as described by Wieckowski et al. [42].

Confocal microscopy

Olympus Fluoview 1000 confocal laser scanning system was used for colocalization experiments. The 488 nm laser was used to excite MTG and AlexaFluor 488-conjugated secondary antibody (Life Technologies). The 543 nm laser was used for detection of MTR and AlexaFluor 568 conjugated secondary antibody (Life Technologies). A 63X objective was used for all images. Nuclei were stained with Hoechst 33342 (10 μg/ml). Mitochondrial morphology analysis was performed by using Mitophagy Macro Plugin (ImageJ Software), and mitochondria area, perimeter and circularity were reported.

Cell respiration and ATP production

Oxygen consumption rate was determined in intact 3T3-L1-white and X9-beige adipocytes. Briefly, at the end of differentiation, cells were either starved for 4 h or cultured in complete culture medium. Real-time oxygen consumption measurement was determined at 30 °C, for 6 min by Oxygraph Plus oxygen electrode system (Hansatech Instruments Ltd, Norfolk, UK). Nutrient refill was performed by resuspending starved adipocytes in complete culture medium. Oxygen consumption was normalized for cell number. After cell respiration analysis, ATP level was detected by using ATP Bioluminescence assay kit (Roche Diagnostics) and values were normalized for cell number. To evaluate cellular uncoupling, we conventionally normalized oxygen consumption for ATP content.

Cytofluorimetric analysis of mitochondrial membrane potential, mass and ROS

Mitochondrial ROS, transmembrane potential and number were detected by incubating cells with the fluorescent probe MitoSOX RED (5 μM), MitoTracker Red CMX ROS (250 nM), Tetramethylrhodamine Ethyl Ester Perchlorate (TMRE, 20 nM), MitoTracker Green (250 nM) or Nonyl Acridine Orange (NAO, 10 nM) (Life Technologies) for 30 min at 37 °C. Subsequently, cells were collected and used for cytofluorimetric analyses by FACScalibur instrument (Beckton and Dickinson, San José, CA, USA). Changes in mitochondrial membrane potential were expressed as Red-to-Green ratio [43].

Statistical analysis

The results are presented as means ± S.D. Statistical evaluation was conducted by ANOVA, followed by the post Student-Newman-Keuls, by using GraphPad Prism 5 Software (GraphPad Software).