Chaperone-mediated autophagy controls brown adipose tissue thermogenic activity

Abstract

Brown adipose tissue (BAT) protects against obesity, diabetes, and cardiovascular disease. During BAT activation, macroautophagy is inhibited, while chaperone-mediated autophagy (CMA) is induced, promoting thermogenic gene expression, adipokine release, oxidative activity, and lipolysis. Aging reduces BAT function and lowers levels of LAMP2A, the rate-limiting CMA component. Pharmacological CMA activation restores BAT activity in aged mice. To explore the CMA’s role in BAT, we generated LAMP2A-deficient brown adipocytes and found that CMA regulates proteins essential for thermogenesis and metabolism. Blocking CMA in BAT reduced energy expenditure, raised blood triglycerides, impaired secretion, and led to an increase of thermogenesis repressors. These findings show that CMA is essential for maintaining BAT function, especially during adaptive thermogenesis. By degrading repressors of thermogenesis, CMA supports BAT activity under cold or metabolic stress. This work highlights CMA as a key regulator of BAT plasticity and a promising therapeutic target for treating age-related metabolic disorders.

Chaperone-mediated autophagy regulates brown fat activation and is a potential target for treating metabolic disorders.

INTRODUCTION

Experimental studies and analysis of large human cohorts have shown that active brown adipose tissue (BAT) is associated with protection against obesity, type 2 diabetes, and cardiovascular disease, whereas its atrophy and inactivation are associated with obesity and aging (1, 2). The decline in BAT activity occurring in aging is considered to contribute to susceptibility to chronic metabolic and cardiovascular diseases at advanced ages (3). Since researchers rediscovered active BAT in humans several years ago, they have made great efforts to identify the mechanisms underlying its activation, inactivation, and age-related changes, hoping to design potential strategies to fight chronic metabolic and cardiovascular diseases.

BAT shows a marked plasticity in response to thermogenic stimuli. Thus, in conditions of thermogenic activation such as a cold environment, brown adipocytes remodel their cellular proteome to sustain thermogenesis, increasing the abundance of uncoupling protein 1 (UCP1) and other enzymes and proteins involved in mitochondrial oxidation (4). Moreover, sustained thermogenic stimulus leads ultimately to BAT hyperplasia, increasing the number of active brown adipocytes in BAT depots (5). BAT activation leads also to a specific pattern of secretion of regulatory molecules, the so-called brown adipokines or batokines, which act locally and also at a distance on other tissues leading to the systemic metabolic adaptation to enhanced thermogenic conditions (6). In aging, obesity or during adaptation to warm environments, BAT hypotrophies and brown adipocytes tend to “whiten,” reducing their cellular thermogenic equipment and acquiring a more “white adipose-type” secretory pattern (3). Similarly, the process of adipose tissue browning (e.g., appearance of thermogenic brown-like adipocytes, so-called beige adipocytes, within white adipose depots) involves proteome remodeling of adipocytes to acquire thermogenic biochemical machinery under conditions of high thermogenic requirement, as well as the reciprocal loss of thermogenic capacity when beige adipocytes reacquire a white adipocyte phenotype in conditions of suppressed thermogenic stimulus (7).

Macroautophagy, a process by which intracellular components are degraded within lysosomes after being sequestered in autophagosomes, is essential in maintenance of cellular homeostasis and in the tissue remodeling occurring in response to multiple environmental physiological or pathogenic stimuli. Macroautophagy has emerged as an important regulatory mechanism of brown/beige fat plasticity (8). Studies from several laboratories have shown that macroautophagy is inversely regulated in relation to adipose thermogenic activation (9, 10) and that macroautophagy induction is key for thermogenic deactivation, i.e., “whitening,” of adipose tissues (11). Chaperone-mediated autophagy (CMA) is a type of selective lysosomal protein degradation, in which proteins reach the lysosomal lumen through direct membrane translocation (12). The protein targets of CMA contain a KFERQ-like pentapeptide motif (13) that enables their lysosomal targeting by the heat shock protein of 70 kDa (HSC70) (14). Upon binding to the lysosome-associated membrane protein type 2A [(LAMP2A (L2A)] (15), the substrate is delivered into lysosomes by the cooperative actions of L2A and a luminal form of HSC70 (16). Levels of L2A at the lysosomal surface determine CMA activity, as binding of substrates to this receptor is the main limiting step in this pathway. L2A is one of the three alternative splicing-originated transcripts from the Lamp2 gene (17).

CMA is increasingly recognized as key in the control of multiple physiological processes such as regulation of lipid and glucose metabolism, proteostasis, cell cycle, circadian rhythm, cellular differentiation, and immune activation (12). Consequently, CMA failure has been shown to lead to neurodegeneration, retinal degeneration, lipodystrophy, atherosclerosis, reduced hematopoiesis, immunosenescence, and other conditions. Although recent data support that CMA is required for general adipogenesis (18), its role in thermogenic adipose tissue plasticity remains unknown. Hereby, we report the direct involvement of CMA activation as an essential component of the biological program of thermogenic activation of BAT.

RESULTS

CMA is induced in response to the thermogenic activation of BAT and the induction of browning in WAT

To explore the role of CMA in BAT thermogenic activation, we first analyzed changes in this autophagic pathway in response to a thermogenic challenge using an RNA sequencing (RNA-seq)–based transcriptome analysis of interscapular BAT (iBAT), inguinal white adipose tissue (iWAT) (representative of subcutaneous adipose tissue), epididymal WAT (eWAT; representative of visceral WAT), and liver from mice maintained at thermoneutrality or exposed to 4°C during 24 hours. We used the expression of 19 genes involved in CMA to calculate the recently developed CMA score, which predicts changes in CMA activity based on the expression values of these components of the CMA network (19). Expression levels of effectors and positive CMA regulators were added up, and the additive expression of the negative regulators was subtracted from this value, which was corrected by the sum of assigned weights to each component. Results indicated a significant increase in the CMA score of BAT and eWAT in response to the thermogenic challenge and a similar trend in inguinal WAT (iWAT), whereas the CMA score in the liver remained unchanged (Fig. 1A).

Fig. 1. Thermogenic activation induces CMA activity in BAT.

(A) Top: Heatmap for the expression of CMA network components in liver, BAT, iWAT, and eWAT from mice maintained at thermoneutral temperature (TN, 30°C) or under cold exposure (CE, 4°C, 24 hours). Bottom: CMA scores calculated from the gene expression shown in the heatmap. The number of replicates is 5 in each condition. (B) DimPlots of CMA score levels in snRNA-seq analysis of BAT adipocytes from mice at room temperature (21°C) or exposed to cold (8°C) according to the indicated value ranges (left); bars are means ± SEM for pseudobulk CMA score (right). (C) Left: Representative green fluorescence of room temperature (RT, 21°C) or cold-exposed BAT tissues (CE, 4°C, 24 hours) from KFERQ-Dendra mice. The number of replicates is 8 to 10 in each condition. Scale bar, 20 μm; right, higher magnification and quantification of red puncta as CMA activity. Results [(A) and (C)] are shown as means ± SEM, *P < 0.05 and **P < 0.01 in statistical comparisons relative to thermoneutrality (A) or room temperature (C). In (B), ***P < 0.001 according to the Wald test.

To confirm that CMA induction in BAT in response to cold exposure occurs specifically in brown adipocytes, we analyzed an available database of single-nuclei RNA-seq (snRNA-seq) of brown adipocytes from iBAT of mice at room temperature (21°C) or cold exposed (8°C) (20). Cells were clustered according to ranked CMA values, and results revealed a massive induction of CMA in brown adipocytes in response to cold (Fig. 1B).

To directly analyze the impact on CMA activity of the thermogenic activation of BAT, we next used a transgenic mouse model systemically expressing KFERQ-tagged photoswitchable fluorescent Dendra protein. Upon photoswitching, green Dendra fluoresces in red, and delivery of this fluorescent CMA reporter to lysosomes highlights them as red fluorescent puncta and allows for quantification of changes in CMA activity (21). We detected a significant increase in the number of red fluorescent puncta in BAT after cold exposure of mice, thus confirming that thermogenic activation leads to an increase of CMA activity in BAT (Fig. 1C).

Thermogenic activation induces LAMP2A expression in BAT

Since L2A, limiting CMA component, is a spliced variant of the Lamp2 gene, we next interrogated the BAT transcriptomic database for specific alternative splicing–related events. We found a reciprocal regulation of the Lamp2A transcript but down-regulation of Lamp2c mRNA, unrelated to CMA. Lamp2A was one of the only 36 genes of the whole transcriptome that showed such reciprocal change in alternative-splicing driven transcripts in response to cold in BAT (table S1).

To confirm the relationship between the extent of thermogenic activation, CMA activity, and Lamp2A expression in BAT, we directly assessed Lamp2A transcript levels in mouse BAT under conditions of thermogenic activation/deactivation. We found that the Lamp2A mRNA levels detected in inactive BAT from mice adapted to a thermoneutral environment increased in the mildly active BAT of mice at 21°C environment temperature and reached maximal levels when mice had been adapted to 4°C for 7 days (Fig. 2A). Deacclimation of cold-exposed mice to thermoneutral temperature, leading to repressed BAT activity, caused a reduction of Lamp2A mRNA levels (Fig. 2A). The effects of cold on BAT were confirmed for the L2A protein, whose abundance was significantly increased after exposure of mice to 4°C, in parallel with the induction of the canonical marker of thermogenesis UCP1 protein (Fig. 2B).

Fig. 2. Noradrenergic activation of BAT and in cultured brown adipocytes induces L2A expression.

(A) Relative transcript levels of Lamp2A in the BAT of mice maintained at thermoneutral temperature (TN, 30°C), room temperature (RT, 21°C), cold exposure (CE, 4°C, 21 days), and deacclimated back to thermoneutrality following previous cold exposure (CD, 7 days back to 30°C) (n = 3 to 4). (B) Left: Representative immunoblot images of L2A and Ucp1 in the BAT of mice at room temperature (RT, 21°C) or after cold exposure (CE, 4°C, 24 hours). Right: Quantification of L2A and Ucp1 protein levels (n = 3). (C) Top: Representative immunoblot images of L2A in the iWAT, eWAT, and liver of mice at room temperature (RT, 21°C) or after cold exposure (CE, 4°C, 24 hours). Bottom: Quantification of L2A protein levels (n = 3). (D) Relative transcript levels of Lamp2A in adipose tissues and liver from mice treated with CL316,243 (1 mg/kg per day, 1 week) (n = 3 to 6). (E) Left: Effects of 0.5 μM norepinephrine (NE) on relative Lamp2A mRNA expression in brown adipocytes. Right: Representative immunoblot and quantification of 0.5 μM NE effects on L2A protein levels in brown adipocytes (n = 3). (F) Left: Representative immunoblot image for the effect of 24-hour treatment with 1 mM dibutyryl-cAMP (cAMP) compared to control condition (CTRL) on L2A protein levels in brown adipocytes. Right: Quantification of L2A protein levels (n = 6). Results are shown as means ± SEM, *P < 0.05, **P < 0.01, and ***P < 0.001 in statistical comparisons relative to thermoneutrality or cold exposure (A), room temperature [(B) and (C)], vehicle-injected mice (D), time 0 of treatment (E), or relative control condition (F). AU, arbitrary units; rRNA, ribosomal RNA.

We also determined changes in L2A protein levels in response to cold-induced browning of WATs and found that cold exposure increased L2A levels in WAT, but this effect was only statistically significant in iWAT (Fig. 2C). Similarly, when using chronic treatment of mice with the β3-adrenoreceptor agonist CL316,243, as an additional approach to activating BAT and inducing WAT browning (3.2-fold and 8.1-fold induction of UCP1 expression in BAT and iWAT, respectively), we found significant transcriptional up-regulation of Lamp2A in BAT and iWAT (Fig. 2D).

To confirm the involvement of brown adipocytes in the CMA response of BAT to thermogenic activation, we treated brown adipocytes in culture with norepinephrine (NE), to mimic sympathetic activation. We found that the effects of thermogenic activation on Lamp2A expression were cell autonomous, as evidenced by the significant induction of the Lamp2A transcript and a notable increase in L2A protein levels in response to NE in brown adipocytes in culture (Fig. 2E). Notably, the induction of Lamp2A in response to NE was not secondary to the NE-induced repression of macroautophagy, known to occur in brown adipocytes (9), as Lamp2A gene was similarly expressed and induced by NE in macroautophagy-deficient (Atg7-knockdown) and control brown adipocytes (fig. S1). We found that treatment of cells with cyclic adenosine 3′,5′-monophosphate (cAMP), the main intracellular mediator of the thermogenic effects of NE in brown adipocytes, also increased significantly L2A protein levels (Fig. 2F)

Chemical enhancement of CMA in aged mice reactivates BAT

Aging is the physiological condition more widely resulting in spontaneous BAT inactivation. As part of a multiorgan study using the CMA reporter mouse model, we have recently found that CMA activity significantly decreases with age in BAT in male mice (22). To further investigate the mechanism behind the decline in CMA, here, we analyzed the status of the CMA transcriptional network and calculated the CMA score in BAT of mice of different ages. We found that from 5-month-old (adult male mice) to 15-month-old (aged male mice), there is a marked reduction in the transcriptional CMA score in BAT, which remained even somewhat lower in 23-month-old mice (Fig. 3A). This pronounced decrease by 15 months of age was confirmed by an independent analysis of the available Tabula Muris data (23) of BAT transcriptome at similar ages ( fig. S2). We also demonstrated a concordant decrease in L2A protein levels in BAT across these aging stages in mice (Fig. 3B).

Fig. 3. CMA activation prevents the decline in BAT activity associated with aging.

(A) Left: Heatmap for the expression of CMA network components in BAT from male mice at 5 (5 m), 15 (15 m), and 23 (23 m) months or age. Right: CMA scores calculated from the gene expression shown in the heatmap (n = 4 to 8). (B) Top: Representative immunoblot image of L2A protein levels in the BAT of mice at 5, 15, and 23 months of age. Bottom: Quantification of L2A protein levels (n = 3). PS, Ponceau staining. (C) Left: Representative histology images of BAT from 23-month-old male mice treated daily with CA77.1 for 5 months since they were 18 months old (old + CA77.1) or receiving only vehicle for the same time (old + vehicle). Right: Higher magnification and quantification of average lipid droplet size and the percentage of microscopy area occupied by lipid droplets (n = 4 to 5). LD, lipid droplet. (D) Left: Representative immunoblot images of L2A and Ucp1 protein levels in the BAT from mice treated with CA77.1 relative to vehicle-treated control mice. Right: Quantification of L2A and Ucp1 protein levels (n = 3 to 4). (E) Relative transcript levels of genes related to BAT thermogenic, metabolic, and secretory function (n = 3 to 4). Results are shown as means ± SEM, *P < 0.05, **P < 0.01. and ***P < 0.001 in statistical comparisons between different mouse ages [(A) and (B)] and vehicle-treated mice [(C) to (E)].

Considering that the age-related decline on CMA in BAT seems to be mostly transcriptionally driven, we next explored the possible impact of treatment of old mice with CA77.1, a recently developed small molecule capable of activating CMA through transcriptional up-regulation of the CMA network (19). We administered CA77.1 (30 mg/kg body weight) daily in the form of jelly pills to 18 months old for 5 months and, at 23 months of age, compared them with vehicle-treated age-matched mice. Results indicated that CA77.1 treatment in old mice modified the histological morphology in BAT leading to brown adipocytes with smaller lipid droplets relative to vehicle-treated mice, which is compatible with BAT activation (Fig. 3C). The CA77.1 treatment significantly increased L2A protein levels in BAT and also resulted in a trend to increased UCP1 protein levels (Fig. 3D). The pattern of transcript modifications in CA77.1-treated mice was consistent also with BAT activation. Although Ucp1 mRNA levels were not significantly modified, the transcript of Dio2, Cpt1b, Cox4i1, Cidea, and Bsg, known components of the biochemical thermogenic machinery (24), were significantly induced in CA77.1-treated mice (Fig. 3E). These findings support that pharmacological activation of CMA in old animals prevents age-related changes in BAT.

CMA activation enhances brown adipocyte thermogenic activity

To dissect the mechanisms behind BAT improvement upon pharmacological activation of CMA and directly analyze the effect of this intervention in adipocytes, we used brown adipocytes differentiated in culture and treated with the CMA activator CA77 (a variant of CA77.1 more suitable for in vitro studies) (25) under basal conditions or after cAMP activation. Under basal conditions, we only found discrete changes in the pattern of transcripts associated with brown adipocyte activity upon CA77 administration; however, under cAMP-treated conditions, CA77 globally induced an overactivation in the expression of genes related to thermogenesis and metabolic activation (Ucp1, Ppargc1a, Cpt1b, and Scl2a1), as well as in genes encoding brown adipokines known to be secreted by brown adipocytes in response to activation (Fgf21, Gdf15, IL6, and Cxcl14) (Fig. 4A). Concordantly, CA77 treatment resulted in a trend toward reduced expression of leptin, a marker of white-versus-brown phenotype (Fig. 4A). Analysis of UCP1 protein indicated a significant increase in response to CA77-mediated CMA up-regulation in basal and a substantial increase in cAMP-treated conditions (Fig. 4B). We next analyzed the functional consequences of those changes in gene expression and found that the release of brown adipokines (Fgf21, Gdf15, and IL6) was significantly enhanced by CA77 treatment (Fig. 4C). Conversely, CA77 repressed leptin secretion both in basal and cAMP-treated conditions (Fig. 4C). We also found that CA77 treatment caused a significant increase in mitochondrial oxygen consumption in cAMP-treated conditions (Fig. 4D). Extracellular acidification, a surrogate marker of lactate release to the medium, was also enhanced in response to CA77, indicating increased glycolysis (Fig. 4E). Up-regulation of CMA with CA77 also further enhanced the extracellular release of glycerol elicited by cAMP (Fig. 4F), thus indicating that CMA activation favors lipolysis, a key process in BAT thermogenic activation. Collectively, these data indicate that CMA activation causes a global enhancement of brown adipocyte thermogenic and metabolic activity, especially when they are activated with cAMP.

Fig. 4. CMA activation induces thermogenic, metabolic, and secretory activity in brown adipocytes.

Differentiated brown adipocytes were treated with 10 μM CA77 for 16 hours and subsequently treated with 1 mM dibutyryl-cAMP (cAMP). (A) Relative transcript levels of genes related to thermogenesis, metabolism, and secretory activity (n = 3). (B) Left: Representative immunoblot image of Ucp1 protein levels. Right: Quantification of Ucp1 protein levels (n = 3). (C) Measurement of brown adipokines and leptin levels in the cell culture medium (n = 5 to 6). (D) Mitochondrial oxygen consumption (n = 5). OCR, oxygen consumption rate; (E) extracellular acidification rate (ECAR) (n = 4 to 5); (F) glycerol levels in the brown adipocytes cell culture medium (n = 4 to 5). Results are shown as means ± SEM, *P < 0.05, **P < 0.01, and ***P < 0.001 in statistical comparisons between CA77-treated and untreated cells under basal or cAMP-stimulated conditions, #P < 0.05, ##P < 0.01, and ###P < 0.001 for the effect of cAMP in a given condition. ATP, adenosine triphosphate; DMSO, dimethyl sulfoxide.

L2A-KD in brown adipocytes represses their thermogenic activity

An independent approach, using a loss-of-function model based on L2A depletion, was undertaken to confirm the role of CMA in brown adipocytes. Differentiated brown adipocytes were transduced with lentiviral-driven short hairpin RNA (shRNA) for Lamp2A [L2A knockdown (L2A-KD)], which led to a strong reduction in L2A protein levels, even when cells were stimulated with cAMP (Fig. 5A). We confirmed that L2A-KD did not alter significantly the morphological differentiation of brown adipocytes (fig. S3). Assessment of transcript levels indicated that L2A depletion led to either no significant changes (Ucp1, Ppargc1a, Cpt1b, and Pparg) or substantial reduction (Slc2a1, Fgf21, Gdf15, IL6, Cxcl14, and Bmp8b) of several marker genes of cAMP-responsive metabolism and secretion. Transcript levels for leptin were significantly up-regulated in L2A-KD brown adipocytes both under basal and cAMP-treated conditions (Fig. 5B). Despite the discrete changes in Ucp1 transcription, we found that UCP1 protein levels were strongly decreased in response to L2A depletion in both basal and cAMP conditions (Fig. 5C). Concordantly, levels of components of the active brown adipocyte secretome (Fgf21 and Gdf15) in the culture medium were significantly reduced in response to L2A depletion, with the exception of IL6, which displayed higher extracellular levels. In agreement with the transcriptional data, basal and cAMP-induced secretion of leptin was significantly increased in L2A-KD brown adipocytes (Fig. 5D).

Fig. 5. Lamp2A gene knockdown represses thermogenic, metabolic, and secretory activity in brown adipocytes.

Brown adipocytes were transduced with a lentiviral vector driving a specific shRNA for Lamp2A (L2A-KD) or scrambled shRNA (control, CTRL) and subsequently differentiated and treated or not with 1 mM dibutyryl-cAMP (cAMP). (A) Left: Representative immunoblot image of L2A protein levels. Right: Quantification of L2A protein levels (n = 3). (B) Relative transcript levels of genes related to thermogenesis, metabolism, and secretory activity (n = 3). (C) Left: Representative immunoblot image of Ucp1 protein levels. Right: Quantification of Ucp1 protein levels (n = 3 to 4). (D) Measurement of brown adipokines and leptin levels in the cell culture medium (n = 3 to 4). (E) Mitochondrial oxygen consumption (n = 3 to 4); (F) extracellular acidification rate (n = 3 to 4); (G) glycerol levels in the brown adipocytes cell culture medium (n = 6). Results are shown as means ± SEM, *P < 0.05, **P < 0.01, and ***P < 0.001 in statistical comparisons between L2A-KD cells and controls under basal or cAMP-stimulated conditions, #P < 0.05, ##P < 0.01, and ###P < 0.001 for the effect of cAMP in a given condition.

Functional assays revealed a significant reduction in oxygen consumption (Fig. 5E) and extracellular acidification (Fig. 5F) under basal conditions, as well as a marked decreased in glycerol release to the medium (Fig. 5G) in L2A-KD brown adipocytes when compared to control cells. Together, these data indicate thermogenic and metabolic impairment in CMA-defective brown adipocytes. This pattern of findings closely mirrors, in a reciprocal manner, the effects found when CMA was activated using the CA77.1 compound.

Effects of CMA blockage on the proteome of brown adipocytes under basal and cAMP-stimulated conditions

To identify the consequences of CMA blockage in the proteome of brown adipocytes and the contribution of this pathway to proteome remodeling during thermogenesis in these cells, we next performed comparative quantitative proteomics of control or L2A-KD brown adipocytes treated or not with cAMP. We found that loss of CMA results in major changes in the basal proteome of brown adipocytes with 387 proteins showing higher abundance in the knockdown group and 290 detected at lower levels in L2A-KD cells compared to control (Fig. 6, A and B, and fig. S4A). Pathway enrichment analysis using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) identified an increase in L2A-KD cells of proteins related to protein folding and processing in the endoplasmic reticulum, actin cytoskeleton remodeling, and regulation of the endolysosomal system (Fig. 6C). Similar analysis in proteins with reduced abundance in L2A-KD cells identified proteins related with mitochondrial organization and function, especially those known to participate in thermogenesis, and proteins involved in lipid metabolism and peroxisome proliferator–activated receptor signaling (Fig. 6D). This is overall in concordance with our functional findings in CMA-deficient brown adipocytes. Added to these basal changes, we also found that upon cAMP treatment, L2A-KD cells failed to display some of the changes (increases and decreases) in levels of proteins noted in the control cells but displayed additional changes unique for the L2A-KD cells (Fig. 6, E and F, and fig. S4, B to D). We focused on the group of proteins whose levels decrease in control cells upon cAMP treatment as we hypothesized that this decrease could be in part due to increased degradation through CMA as part of the induction of the thermogenic program in brown adipocytes. Analysis of this group of proteins demonstrated enrichment of those involved in metabolism, especially of lipids, response to cytokines and insulin, and PI3K–adenosine triphosphate signaling pathways (Fig. 6E). Among them, the group of proteins that fail to decrease upon cAMP treatment if CMA is not functional included proteins involved in glucose and lipid metabolism, mitochondrial respiration and protein import, intracellular signaling pathways, as well as calcium dynamics and extracellular matrix organization (Fig. 6F). These findings confirm the existence of basal differences in the proteome of cells defective in CMA that could be responsible for their phenotypic differences and highlight the need of a functional CMA to attain some of the proteome remodeling occurring in brown adipocyte during thermogenic activation.

Fig. 6. CMA contributes to brown adipocyte proteome remodeling during thermogenesis.

Comparative quantitative proteomics of control (Ctrl) and L-2A knockdown (L2A-KD) brown adipocytes under basal conditions or upon treatment with dibutyryl-cAMP (cAMP). (A) Heatmap to illustrate the differences in protein levels between Ctrl and L2A-KD cells under basal and cAMP-treated conditions. The numbers denote major change types: (1) proteins decreasing, (3) increasing with cAMP in control but not L2A-KD cells, and (2) proteins unchanged in control but altered in L2A-KD cells. (B) Venn diagram representation of the number of proteins with significantly different levels between Ctrl and L2A-KD cells under basal conditions. (C) STRING analysis of proteins significantly more abundant (>1.2-fold) in L2A-KD cells compared to Ctrl under basal conditions. (D) STRING analysis of proteins significantly less abundant (<0.8-fold) in L2A-KD cells compared to Ctrl under basal conditions. (E) STRING analysis of proteins with decreased levels upon cAMP treatment in Ctrl cells. (F) Functional protein families that fail to decrease upon cAMP treatment in L2A-KD cells (potential CMA substrates). VEGFA-VEGFR2, Vascular endothelial growth factor A-Vascular endothelial growth factor receptor 2. (G) Left: Log₂ fold changes (log FC) in protein levels between L2A-KD cells and control after cAMP versus basal. Blue-labeled proteins are known BAT and thermogenesis repressors (26, 27, 36–42). Right: Log FC in protein levels between basal and cAMP-treated conditions upon L2A-KD cells against changes in control cells. (H) Protein levels of Pdgrfb and Aebp1 in LA-KD or control (CTRL) brown adipocytes. All Gene Ontology terms in (E) to (G) are statistically enriched with P < 0.001. Bars are shown as means ± SEM with three replicates in each condition, *P < 0.05 and **P < 0.01 in statistical comparisons between L2A-KD and controls under basal or cAMP-stimulated conditions, #P < 0.05 and ##P < 0.01 for the effect of cAMP in a given condition. ER, endoplasmic reticulum; PPAR, peroxisome proliferator–activated receptor; CREB1, cAMP response element–binding protein; PKA, cAMP-dependent protein kinase.

A close examination of the proteins detected at higher levels upon CMA blockage in brown adipocytes revealed accumulation of several proteins with previously described roles as repressors of BAT thermogenic activity (Fig. 6G). We validated our mass spectrometry findings by immunoblot analysis of the platelet-derived growth factor receptor factor-β (Pdgfrb), a known promoter of white-versus-brown phenotype (26), and the adipocyte enhancer binding protein-1 (Aebp1), a negative regulator of adaptive thermogenesis (27), in L2A-KD brown adipocytes. We confirmed accumulation of both proteins upon CMA blockage (Fig. 6H), which supports a role for CMA in timely degrading BAT thermogenic repressors to activate thermogenesis.

We performed a bioinformatic analysis to identify transcription factors (TFs) predicted to control the genes encoding the proteins up-regulated in response to L2A-KD. Most of the predicted TFs (15 up to 18) cluster in an interrelated regulatory network, according to annotated interactions (fig. S5A). This analysis predicted Aebp1, consistent with the observed increase in Aebp1 protein levels in L2A-KD brown adipocytes mentioned above. Among the other predicted TFs, several had previously uncharacterized roles in the regulation of BAT. Notably, however, a substantial number of the top-ranked TFs have been reported to exert repressive effects on BAT thermogenesis (fig. S5B) (see Discussion). Analysis of these TFs’ sequences demonstrated the presence of KFERQ-like motifs in >80% of them (13 of 16 TFs bear canonical motifs or motifs generated by single phosphorylation or acetylation, with the other 3 showing potential motifs generated by multiple posttranslational modifications or asparagine usage) (table S4). These findings support the possibility of these TFs also being CMA substrates and predict an increase in their protein levels as result of L2A-KD.

To directly evaluate the potential repressive role of TFs that accumulate upon CMA blockage, Aebp1 was selected as a representative candidate. Aebp1 expression in brown adipocytes was knocked down using small interfering RNA (siRNA) specifically targeting Aebp1, resulting in a significant reduction of both Aebp1 mRNA and Aebp1 protein levels under basal and cAMP-stimulated conditions (fig. S6). While Aebp1 knockdown had minimal impact on the basal expression of thermogenic marker transcripts, it significantly impaired the cAMP-induced expression of key thermogenic genes, including Ucp1, Ppargc-1a, Pparα, and others (fig. S6). In contrast, Aebp1 knockdown repressed the expression of the white-versus-brown marker gene leptin. These effects were accompanied by an up-regulation of UCP1 protein levels in response to Aebp1 silencing. Overall, the observed changes were inversely mirrored by those seen with L2A-KD and support the notion that CMA induction during thermogenic activation functions to suppress intracellular repressors of thermogenesis.

LAMP2A knockdown in vivo represses BAT activity and promotes accumulation of thermogenic repressors

To confirm the role of CMA in the control of BAT thermogenic activity in vivo, we invalidated L2A expression by local injection into iBAT of an adeno-associated viral vector-8 (AAV8) carrying Lamp2A shRNA (AAV8-Lamp2A-shRNA). Six weeks after the injection, we confirmed efficient knockdown, with L2A protein levels being 20% of control in the AAV-Lamp2A-shRNA–injected mice (Fig. 7A). There was no leaking of the AAV8-mediated local knockdown at iBAT as, for example, L2A expression in iWAT, eWAT, and liver was unaltered (fig. S7). L2A depletion in BAT did not modify overall body weight or food intake, although WAT depots weight trended higher (see table S2). Histological analysis of BAT revealed increased fat accumulation and increased lipid droplet size in L2A-KD mice (Fig. 7B). Oxygen consumption was significantly decreased in the BAT of these mice, indicating a reduction in energy expenditure (Fig. 7C). After acute cold exposure (4°C, 24 hours), mice with shRNA-mediated L2A depletion in BAT were tolerant to 24 hours cold and showed no changes in circulating glucose but showed significantly increased triglyceride levels relative to cold-exposed controls (Fig. 7D). In L2A-KD BAT, we found that Ucp1 protein trended to decrease, and Fgf21 protein levels were significantly reduced, indicating altered intracellular and secretory functions of BAT upon thermogenic activation (Fig. 7E). Reciprocally, Pdgfrb and Aebp1 proteins were increased in L2A-KD BAT (Fig. 7F). These findings suggest a scenario in which CMA activity is essential for BAT activity and energy expenditure through the selective degradation of key biological repressors of thermogenesis in the brown adipocyte.

Fig. 7. In vivo knockdown of L2A in BAT reduces energy expenditure, impairs thermogenic activity in BAT, and promotes accumulation of potential repressors of BAT activity in mice.

(A) Mice were locally injected at the iBAT with AAV8-shLamp2A (shRNA-Lamp2A) or control AAV8 (shRNA-scrambled). Lamp2A mRNA, representative immunoblot image, and quantification of L2A protein levels in BAT (n = 3 to 6). ITR, inverted terminal repeat. GFP, green fluorescent protein. (B) Top: Representative histology images of BAT. Bottom: Average lipid droplet size and percentage of microscopy area occupied by lipid droplets (n = 9). (C) Left: Representative oxygen consumption graph. Right: Average oxygen consumption during the day and night periods in mice (n = 5 to 6). AUC, area under the curve. (D) Blood glucose and triglyceride levels in L2A-KD and control mice after cold exposure (4°C, 24 hours) (n = 5 to 6). (E) Representative immunoblot images and quantification of Ucp1 protein levels (left) (n = 4) and Fgf21 protein levels (right, n = 5) in the BAT of L2A-KD and control mice after cold exposure (4°C, 24 hours). (F) Representative immunoblot images and quantification of Pdgfrb protein levels (top, n = 5) and Aebp1 (bottom, n = 5) in the BAT of L2A-KD and control mice after cold exposure (4°C, 24 hours). Results are shown as means ± SEM, *P < 0.05 and ***P < 0.001 in statistical comparisons between L2A-KD and control mice.

DISCUSSION

In this study, we identify CMA as a key process in BAT biology. We found that CMA is up-regulated in response to thermogenic activation and that blockage of CMA via Lamp2A depletion impairs BAT activity with subsequent systemic effects including reduced energy expenditure. We demonstrate that pharmacological activation of CMA in vivo is effective in increasing BAT activity in old mice.

A first aspect to highlight from our results is the recognition of a remarkable opposite behavior of CMA and macroautophagy in relation to BAT thermogenic activation. Although there are indications that macroautophagy may be required for adequate brown adipocyte differentiation (28), in differentiated brown adipocytes, macroautophagy is repressed in response to thermogenic activation, likely as a mechanism for preservation of the cellular mitochondrial equipment and other cellular machinery required to sustain maximum thermogenesis (9–11, 29, 30). Conversely, we found that CMA is markedly up-regulated in response to BAT thermogenic activation. According to our results, the regulation of expression of Lamp2A and other components of the CMA machinery in BAT is a component of the overall cold-induced thermogenic gene expression program. Cross-talk between macroautophagy and CMA has been reported previously in other systems (31, 32) but mostly as compensatory mechanism in response to failure of one of these pathways; however, BAT is the first example of autonomous responsiveness of CMA to thermogenic stimulus, not secondary to thermogenesis-induced macroautophagy inhibition but instead mediated directly by the induction of L2A expression in response to cAMP-mediated noradrenergic signaling.

Our gain- and loss-of-function experimental approaches revealed a marked requirement of active CMA for proper brown adipocyte thermogenic activity. The impact of CMA modulation in BAT in vivo and vitro spans the overall pattern of BAT functions related to BAT activation, from oxidative activity and metabolic adaptations, such as lipolysis and glycolysis, to the secretory pattern of brown adipokines typical of thermogenically active BAT. Accordingly, we found a marked impact of L2A depletion in the proteome of brown adipocytes, which becomes even more evident during activation of thermogenesis, when we identified a large set of proteins that are degraded by CMA under these conditions, including key thermogenic pathway inhibitors. There were also several components of metabolic pathways that are down-regulated in response to cAMP-mediated thermogenesis stimulus which failed to be reduced when CMA is impaired. Examination of proteins detected at higher levels in response to L2A depletion in brown adipocytes revealed accumulation of several known repressors of adipose thermogenic activity, all of them containing the KFERQ-like pentapeptide motif indicative of potential targeting by CMA. PDGFRB, which is down-regulated in response to cAMP in control brown adipocytes but accumulates in L2A-KD cells under cAMP stimulus, is a known inducer of white-versus-brown phenotype in cell differentiation (26). Under cAMP-mediated stimulus, L2A-KD cells also accumulate AEBP1, whose invalidation in mice leads to resistance to experimental obesity through enhanced energy expenditure (27). It is worth noting that AEBP1 protein is induced in response to cAMP in brown adipocytes; however, this induction does not preclude a repressive role in thermogenesis, as evidenced by the increased expression of cAMP-induced thermogenic genes following Aebp1 knockdown. Recent studies have indicated a similar behavior for proteins that repress BAT activity such as kininogen (33), sLR11 (a soluble relative of the low-density lipoprotein receptor) (34), and adenylyl cyclase 3 (35), attributing them a role as a rheostatic regulators of BAT to avoid excessive energy dissipation.

CMA blockage also induced the accumulation of SMAD2 and the Gja1-encoded connexin-43, both established repressors of BAT activity (36–38). COL12A1 has been recently reported to be negatively associated with BAT activity (39), whereas CD81 is expressed in beige but not in brown precursor cells before differentiation (40). Accumulation of C/EBPB peptides in our proteomic analysis will deserve further analysis particularly in relation to LIP (liver-enriched inhibitory protein), the form of C/EBPB with transcriptional repressive action upon C/EBPB targets (41, 42). Our proteomic analysis also unveiled the accumulation of proteins with no previous knowledge of relationship with BAT biology, such as Dpysil3, Mfge8, or Crabp1, which deserve further research to establish their role in the regulation of BAT activity mediated by CMA. In addition, in silico prediction of TFs regulating the genes encoding proteins up-regulated in L2A-KD brown adipocytes identified, alongside Aebp1, several TFs previously reported to repress brown adipocyte differentiation and/or thermogenic activation. For example, members of the PRRX family (PRRX1 and PRRX2) suppress adipogenesis and impair the differentiation of BAT-derived precursor cells into mature brown adipocytes (43). Similarly, GLI3 has been identified as a repressor of BAT development and activation via the Hedgehog signaling pathway (44, 45). Ablation of NFAT4 (Nuclear factor of activated T-cells 4) TF protects against obesity in association with BAT activation (46), and SHOX2 (Short-stature homeobox 2) is considered a marker of white-versus-brown adipocyte identity (47). The presence of KFERQ-like motifs in all these TFs make them putative CMA substrates, thus predicting that up-regulation of CMA during thermogenesis would lead to reduced cellular levels of these TFs and contribute to inactivate a transcriptional program enriched for repressors of brown adipocyte development and thermogenic activation.

Overall, these data are consistent with a scenario in which CMA up-regulation upon thermogenic activation is required for selective degradation of a set of proteins with a repressive role in thermogenesis and whose down-regulation via CMA is essential for adequate activation of BAT. We propose that whereas upon thermogenic activation, macroautophagy is inhibited to prevent a gross degradation of mitochondrial and other intracellular material required to maintain the cellular thermogenic machinery, the selectivity at the level of single proteins offered by CMA allows to down-regulate intracellular repressors thus allowing thermogenic activation.

The importance of CMA activity in BAT adaptive thermogenesis is expected to have significant consequences in whole body homeostasis. Our findings indicating that selective depletion of Lamp2A in BAT reduces energy expenditure are consistent with our previous findings indicating that whole body L2A-KO mice show a spontaneous increase in adiposity attributable to decreased energy expenditure (48). Increased triglyceridemia in mice with L2A depletion in BAT is also consistent with the key role of active BAT in draining triglycerides from circulation (49). Considering the druggability of CMA, as shown here with the experimental use of the CA77.1 CMA activator, the current findings may open prospects for pharmacological interventions to profit the healthy properties of active BAT in the context of metabolic diseases, and especially in aging, when preventing the decline of CMA could lead to metabolic improvements. It is important to note, however, the limitation that our current study was restricted to male mice, and further research will be necessary to confirm the role of CMA in BAT in females, given the sex-related differences in the thermogenic regulation of BAT (50) and the recently reported sex differences in CMA activity in this tissue (22).

MATERIALS AND METHODS

Mouse experiments

Experiments were conducted on 3-month-old C57Bl/6J male mice (Harlan Laboratories) unless otherwise stated. Mice were maintained at the standard animal facility temperature (21°C), at thermoneutrality (30°C) or under cold exposure (4°C), as stated in each experiment. Unless otherwise specified, all animals were maintained under a 12-hour dark/light cycle with ad libitum access to food and water. When indicated, mice were injected intraperitoneally of CL316,243 (1 mg/kg; 17499, Cayman Chemical) or saline once per day for 1 week. At the endpoint of noninvasive experimental procedures, mice were euthanized by decapitation, and blood and tissues were collected. Tissues were dissected, weighed, and frozen for further mRNA and protein analyses, or they were fixed and processed for microscopic analyses (see below). KFERQ-Dendra mice used to monitor the CMA activity were generated as described before (21). Male, 3- to 5-month old mice were used for image experiments where after dissection and before fixation, small BAT pieces were photoactivated (405/20-nm light-emitting diode array, Norlux) for 10 min using 50-mW/cm² light intensity, similarly to previous reports (51). The number of mice used for study was no larger than the estimated using G*Power 3.1.9.7 (52). Animals were randomly assigned to treatment groups, and no animal was excluded from the final analysis. Investigators were blinded to group allocation, collection of samples, and analyses. All experiments were performed in accordance with European Community Council directive 86/609/EEC, and the experiments and numbers of animals to be used were approved by the Institutional Animal Care and Use Committee of Albert Einstein College of Medicine (P: 00001436) and the Institutional Animal Care and Use Committee of the University of Barcelona (19/24 P2).

Pharmacological activation of CMA in old mice

Male, 18-month-old mice were randomly divided into groups receiving vehicle or 30 mg/kg body weight CA77.1 daily in the form of sucralose gelatin agar pills as previously described (19). Briefly, the jelly pellets were prepared by dissolving the final amount of compound per day in ethanol, which was then mixed with warm gelatin solution (100 mg/ml, 10 mg/ml sucralose in water) and poured into 24-well flat-bottom plates for solidification. Mice were housed in groups of two or three to mitigate stress and competition for the pellets and were trained for 1 week to consume vehicle pellets in the corner of each cage. Eating of the pellet was monitored until completion (average time, 2 min). Treatment was performed for 5 months, all animals were euthanized, and their BAT was dissected at 23 months of age.

Lamp2A knockdown in BAT

Male C57BL/6J mice (8 weeks old) underwent bilateral silencing of Lamp2A in iBAT following previously described procedures (53). Mice were anesthetized, iBAT exposed, and injected with AAV8-Lamp2A-shRNA, an adeno-associated virus serotype 8 (1 × 10¹¹ genome copies/50 μl per mouse, split in six independent injections) driving expression of a shRNA specifically targeting Lamp2A (54). Control mice received iBAT direct injections encoding a scrambled sequence AAV8 (scrambled-shRNA). The viral vector carried a green fluorescent protein reporter under the control of a cytomegalovirus promoter (Vector Biolabs, Malvern, PA). The titer for each virus was 10¹³ gene copies/ml. Five weeks after the injections, mice were analyzed for noninvasive procedures, and a subset was exposed to cold (4°C, 24 hours). Mice were later euthanized, and samples were obtained as described above. Blood glucose and triglyceride levels were measured using Accutrend Technology (Roche Diagnostics).

Determination of respiration parameters in mice

Mice were single-housed and acclimated in Promethion Core metabolic chambers (Sable Systems International), and data were recorded. Oxygen consumption (VO₂) and carbon dioxide expiration (VCO₂) were monitored every 32 min. From these data, the energy expenditure was automatically calculated from VO₂ measurements according to the manufacturer’s guidelines

Optical and fluorescence microscopy

For hematoxylin and eosin staining, tissue samples were fixed overnight in 4% formalin, paraffin-embedded, processed according to standard procedures, and observed under an optical microscope. Lipid droplet area quantifications used the Fiji ImageJ Adiposoft suit adapted to BAT (33, 55). For fluorescence imaging of BAT, tissue was fixed for 12 hours at 4°C in picric acid fixation buffer [2% formaldehyde and 0.2% picric acid in phosphate-buffered saline (PBS) (pH 7.0)] and then washed with 70% ethanol, followed by two washes in PBS. Tissue was immersed in 30% sucrose and then embedded in the optimal cutting temperature (OCT) for sectioning in a cryostat (Leica CM3050 S). After air drying for 30 min, sections were stored at −20°C until use. Following procedures previously described in (56), slices were mounted in 4′,6-diamidino-2-phenylindole–Fluoromount-G, and images were acquired in xyz planes with an Axiovert 200 fluorescence microscope (Carl Zeiss Ltd.), mounted with an ApoTome.2 slider.

RNA isolation and quantitative real-time PCR

RNA was extracted from tissues and cells using a NucleoSpin RNA kit (Macherey-Nagel), and the mRNA levels were determined by quantitative reverse transcription polymerase chain reaction (PCR), using the appropriate TaqMan (Applied Biosystems) or SybrGreen (Sigma-Aldrich) probes (tables S3 and S4). The mRNA level of each gene of interest was normalized to that of the appropriate housekeeping reference gene (Ppia or 18S for Taqman and Rps9 for SybrGreen) using the comparative (2^-ΔCt) method.

RNA-seq analysis

RNA-seq libraries were prepared following the SMARTseq2 protocol (57) with some modifications. Briefly, reverse transcription of the total RNA input material of 1 μg was performed using SuperScript II (Invitrogen) in the presence of oligo-dT30VN (1 μM; 5′-AAG CAG TGG TAT CAA CGC AGA GTA CT30VN-3′), template-switching oligonucleotides (1 μM), and betaine (1 M). The cDNA was amplified using the KAPA Hifi Hotstart ReadyMix (2×) (Roche) and 100 nM Illumina Sequencing PCR primer (5′-AAG CAG TGG TAT CAA CGC AGA GT-3′) with 8 cycles of PCR amplification. Following purification with Agencourt Ampure XP beads (1:1 ratio; Beckmann Coulter), the product size distribution and the quantity were assessed with a Bioanalyzer High Sensitivity DNA Kit (Agilent). The amplified cDNA (200 ng) was fragmented for 10 min at 55°C using Nextera XT (Illumina) and amplified for 12 cycles with indexed Nextera PCR primers. The library was purified twice with Agencourt Ampure XP beads (0.8:1 ratio) and quantified on a Bioanalyzer using a High Sensitivity DNA Kit. The libraries were sequenced on HiSeq2000 (Illumina, Centro Nacional de Análisis Genómico, Spain) in paired-end mode with a read length of 2 × 76 bp using the TruSeq SBS Kit v3-HS(Illumina) in a fraction of a sequencing flow cell lane, following the manufacturer’s protocol. Image analysis, base calling, and quality scoring of the run were processed using the manufacturer’s software Real Time Analysis (RTA 1.13.48) and followed by generation of FASTQ sequence files by CASAVA (Consensus Assessment of Sequence And Variation). Mouse RNA-seq reads were mapped against the Mus musculus reference genome (GRCm38) with STAR/2.5 (58) using ENCODE (Encyclopedia of DNA Elements) parameters. Genes and isoforms were quantified with RSEM (RNA-Seq by Expectation-Maximization)/2.3.0 (59) with default parameters using the gencode.M15 annotation. The original raw and processed data are available from the Gene Expression Omnibus, accession number GSE290875. Differential expression analysis was performed with the R package DESeq2/1.18 (60). The regularized log transformation of the counts was used for plotting. Genes with false discovery rate (FDR) < 5% and |fold change (FC)| > 1.5 were considered significantly differentially expressed. Heatmaps were drawn with the R package “ggplot2” using z-score normalization, and the principal components analysis (PCA) was done with the “prcomp” R function.

snRNA-seq data analysis

For analysis of snRNA-seq of brown adipocytes in iBAT from mice at room temperature (21°C) and under cold exposure (8°C, 2 days), available data at ArrayExpress, accession code E-MTAB-856219 (20), were used and analyzed following the code deposited in GitHub (https://github.com/IRCGP-Lab/Macrophage-heterogeneity-after-MI), with minor modifications. Bioinformatic processing of the snRNA-seq data was performed with the R package Seurat (v.3.2.0) (61). To exclude low-quality cells in snRNA-seq, we filtered cells with an expressed gene count and total RNA molecules count fewer than 2% or greater than 98%. In addition, cells in which more than 5% of reads corresponded to mitochondrial genes were removed. Data were log-normalized, and highly variable features were identified on the basis of a variance stabilizing transformation method. All datasets [room temperature (RT) and cold exposure (CE)] were then integrated using the canonical correlation analysis method, “Find Integration Anchors” and “Integrate Data” functions in Seurat. PCA was performed on the integrated datasets. On the basis of the top 50 principal components, graph-based clustering was performed using the shared nearest-neighbor modularity optimization with resolution set to 1.1, and cells resulted to be classified into 13 clusters. Clustering data was then applied followed by uniform manifold approximation and projection (UMAP) allowing the visualization of identified clusters in UMAP plots (62). Average log FC was used for comparison of Ucp1 expression levels between brown adipocytes in “room temperature” versus “cold exposure” condition to confirm thermogenic activation (fig. S8). Normalized data of genes used for CMA score calculations from each cell (19) were extracted through the LayerData function. CMA score was calculated from gene expression data, and a CMA-SCORE-level parameter was generated by establishing 7 CMA score ranges, from a minimum for values < −0.75 to a maximum for values > 0.5, in intervals of 0.25. CMA-SCORE-level was introduced in Seurat object as part of the metadata with the AddMetadata function. Then CMA-SCORE-level was plotted in the split maps with DimPlot function. Pseudobulk approximation (classifying every cell based on cluster-and-condition identity) was done, and CMA score values were summed up for 13 “samples” in each condition (13 clusters and 2 conditions: RT and CE). The statistic Wald test was then applied to confirm the differences on CMA induction between room temperature and cold exposure conditions (***P < 0.001).

Proteomic analysis

An aliquot corresponding to 30 μg of protein of each brown adipocyte cell culture sample (five samples per experimental group) was loaded in SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gel and run 1 cm to eliminate possible interferents. Then, proteins were in gel digested after reduction (20 mM dithiothreitol; 60 min, 60°C) and alkylation (55 mM iodoacetamide; 25°C, 30 min, in the dark). Proteins were digested with trypsin [sequencing grade modified trypsin, Promega (pH 8), 37°C]. Peptide mixtures were cleaned-up with a C18 tip (ZipTip) as per the manufacturer’s protocol. The dried-down peptide mixtures were analyzed in a Dionex Ultimate 3000 LC system coupled to an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific). The tryptic digests were resuspended in 3% acetonitrile (ACN) and 1% formic acid (FA) solution, and an aliquot of 600 ng per sample was injected for chromatographic separation (trap column: 300 μm inner diameter by 5 mm PepMap100, 5 μm, 100 Å, C18, Thermo Fisher Scientific, column: NanoEase MZ HSS T3 column, 75 μm by 250 mm, 1.8 μm, 100 Å, Waters). The gradient used for the elution was 3 min at 3% B followed by 3 to 35% B in 180 min (A: 0.1% FA; B: 100% ACN, 0.1% FA; flow rate: 250 nl/min). The column temperature is 40°C. Raw data were processed with MaxQuant software (v_1.6.6.0) (63). The spectra were searched using its built-in Andromeda search engine, against the SwissProt Mouse database (v_231102) including contaminants. The following parameters were used: fixed modifications: carbamidomethylation of cysteine; variable modifications: methionine oxidation and protein N terminus acetylation; enzyme: trypsin; maximum allowed missed cleavage: two. For label-free quantification (64), the minimum ratio count was set to 2, and both razor and unique peptides were used for quantitation. FDR was set to 1% for both protein and peptide spectrum match levels.

Label-free quantitative data were processed using Perseus open software (v_2.0.10.0). Perseus was used to obtain the curated protein dataset by removing proteins identified as contaminants, proteins identified only by site, and proteins identified from the redundant and reversed databases. In addition, data were filtered so missing values were excluded if three valid values were not present in at least one group. Results were filtered at 0.01 FDR (peptide and protein levels). For label-free quantification, match between runs option was enabled. Afterward, the “proteinGroups.txt” file was loaded in Prostar (v1.14) (65) using the intensity values for further statistical analysis. A global normalization of log₂-transformed intensities across samples was performed using the LOESS function. Missing values were imputed using the Structured Least Squares Algorithm (SLSA) (for partially observed values) and DetQuantile (for values missing on an entire condition). Differential analysis was done using the empirical Bayes statistics limma. Proteins with a P value of <0.05 and a log₂ ratio filling the criteria >1.2-fold or <0.8-fold change were defined as different in abundance. The FDR was estimated to be below 5% by Benjamini-Hochberg. Pathway analysis was performed using the Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems) and STRING database (https://string-db.org/). The presence of KFERQ-motifs in protein sequences was analyzed using the KFERQ-finder (V0.8) online tool https://rshine.einsteinmed.edu/ (13).

To perform in silico modeling of the TF regulatory network associated with proteins induced in L2A-KD adipocytes, gene lists encoding the up-regulated proteins in L2A-KD cells, under basal conditions and in response to cAMP, were extracted. TF enrichment analysis was carried out using ChIP-X Enrichment Analysis 3 (66).

Cell culture and reagents

Immortalized brown adipocytes from C57BL/6J mice were provided by A. M. Valverde (Instituto de Investigaciones Biomédicas, CSIC, Madrid). Cells were maintained in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum and 20 mM Hepes at 37°C and 7% CO₂. To differentiate, cells were plated in 12-well plates (100,000 cells per well), and 1 nM T3 and 20 nM insulin were added to the media until 70 to 80% confluence was reached. Then, 500 nM dexamethasone, 1 μM rosiglitazone, 125 μM indomethacin, and 500 μM 3-isobutyl-1-methylxanthine were added to the media less than 48 hours. After that, cells were cultured with media containing only T3 and insulin. Cells were cultured for 4 to 5 days more until they were totally differentiated. When indicated, differentiated brown adipocytes were treated with 0.5 μM NE or 1 mM dibutyryl cAMP at the times indicated for each experiment. Cell culture reagents were from Sigma-Aldrich unless otherwise indicated. L2A knockdown brown adipocyte cell lines were established transducing the cells with a shRNA lentiviral vector driving the reported sequence (54), under the phosphoglycerate kinase promoter or the control empty vector. When indicated, brown adipocytes were treated with 10 μM CA77 (25) for 16 hours and subsequently treated or not with cAMP for further 24 hours. Glycerol concentration to assess lipolysis in the media was determined with spectrophotometric methods (Sigma-Aldrich). Mouse-specific enzyme-linked immunosorbent assay assays were used for quantification of GDF15 (R&D Systems) and FGF21 (Biovendor). Interleukin-6 and leptin in cell culture medium were measured using a multiplex kit (MADCYMAG-72 K; Merck-Millipore, Billerica, MA) and Luminex 100 IS version 2 equipment (Luminex, Austin, TX). When indicated, brown adipocytes were transfected with either a Dicer-substrate small interference RNA (DsiRNA) duplex targeting Aebp1 or a nontargeting control duplex (Integrated DNA Technologies, Coralville, IA, USA) using Lipofectamine RNAiMAX and the TriFECTa kit (Thermo Fisher Scientific). Cells were transfected with a 10 nM combination of three Aebp1-targeting siRNA duplexes (duplexes 1, 2, and 3) to minimize off-target effects. Transfections were performed in Opti-MEM medium (Thermo Fisher Scientific) for 6 hours on days 4 and 6 of brown adipocyte differentiation. After the initial 6-hour incubation, the medium was supplemented with 2× growth medium at a 1:1 volume ratio and maintained until 24 hours posttransfection. At that point, the medium was replaced with fresh growth medium, and cAMP treatment was applied as previously described. Cells were harvested 24 hours later for RNA and protein analyses.

Immunoblot analysis

Tissue extracts were homogenized in lysis buffer [50 mM tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, and 0.1% SDS] containing a protease inhibitor cocktail (Roche) and phosphatase inhibitors (2 mM sodium orthovanadate, 1 mM sodium pyrophosphate, and 10 mM sodium fluoride). The total protein content was measured using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). The proteins were resolved by 12 or 15% SDS-PAGE and electrotransferred to Immobilon-P polyvinylidene difluoride membranes (GE Healthcare). The membranes were incubated with primary antibodies specific for L2A (51-2200, Life Technologies), UCP1 (ab10983, Abcam), FGF21 (sc-81946, Santa Cruz), AEBP1 (sc-271374, Santa Cruz), PDGFRB (MA5-15143, Invitrogen), β-tubulin (T8660, Sigma-Aldrich), and then with horseradish peroxidase (HRP)–conjugated anti-mouse immunoglobulin G (IgG; 1721011, Bio-Rad), or anti-rabbit IgG (ab6721, Abcam), as appropriate. Signals were detected using a chemiluminescence-HRP substrate (EMD Millipore). Densitometric analyses of digitalized images were performed using ImageJ software. Images were processed using Adobe Photoshop CS6 (Adobe Systems); brightness and contrast adjustments were applied uniformly across the entire image.

Oxygen consumption determination in brown adipocytes

Mitochondrial oxidative activity and extracellular acidification were measured in brown adipocytes using the standard Seahorse XF Cell Mito Stress Test protocol in a Seahorse XFe24 Analyzer (Agilent, Santa Clara, USA). Protein quantification was used to normalize the results.

Statistical analysis

Statistical significance was assessed using the two-tailed unpaired Student’s t test or one-way analysis of variance (ANOVA) followed by the Dunnett’s or Tukey’s post hoc tests or by two-way ANOVA, all of which were applied with the GraphPad statistical software (GraphPad Prism 8 Software). Welch’s correction was applied when unequal variances were detected by an F test. Statistical significance was set with an α value of P < 0.05, and the underlying assumptions for validity were assessed for all tests. Data are shown as means ± SEM.

Supplementary Materials

This PDF file includes:

Figs. S1 to S8

Tables S1 to S4