Biochemical Journal

Research article

Pref-1 in brown adipose tissue: specific involvement in brown adipocyte differentiation and regulatory role of C/EBPδ

Jordi Armengol, Josep A. Villena, Elayne Hondares, María C. Carmona, Hei Sook Sul, Roser Iglesias, Marta Giralt, Francesc Villarroya

Abstract

Pref-1 (pre-adipocyte factor-1) is known to play a central role in regulating white adipocyte differentiation, but the role of Pref-1 in BAT (brown adipose tissue) has not been analysed. In the present study we found that Pref-1 expression is high in fetal BAT and declines progressively after birth. However, Pref-1-null mice showed unaltered fetal development of BAT, but exhibited signs of over-activation of BAT thermogenesis in the post-natal period. In C/EBP (CCAAT/enhancer-binding protein) α-null mice, a rodent model of impaired fetal BAT differentiation, Pref-1 was dramatically overexpressed, in association with reduced expression of the Ucp1 (uncoupling protein 1) gene, a BAT-specific marker of thermogenic differentiation. In brown adipocyte cell culture models, Pref-1 was mostly expressed in pre-adipocytes and declined with brown adipocyte differentiation. The transcription factor C/EBPδ activated the Pref-1 gene transcription in brown adipocytes, through binding to the proximal promoter region. Accordingly, siRNA (small interfering RNA)-induced C/EBPδ knockdown led to reduced Pref-1 gene expression. This effect is consistent with the observed overexpression of C/EBPδ in C/EBPα-null BAT and high expression of C/EBPδ in brown pre-adipocytes. Dexamethasone treatment of brown pre-adipocytes suppressed Pref-1 down-regulation occurring throughout the brown adipocyte differentiation process, increased the expression of C/EBPδ and strongly impaired expression of the thermogenic markers UCP1 and PGC-1α [PPARγ (peroxisome-proliferator-activated receptor γ) co-activator-α]. However, it did not alter normal fat accumulation or expression of non-BAT-specific genes. Collectively, these results specifically implicate Pref-1 in controlling the thermogenic gene expression program in BAT, and identify C/EBPδ as a novel transcriptional regulator of Pref-1 gene expression that may be related to the specific role of glucocorticoids in BAT differentiation.

  • brown adipose tissue
  • CCAAT/enhancer-binding protein δ (C/EBPδ)
  • glucocorticoid
  • Pref-1
  • thermogenesis

INTRODUCTION

BAT (brown adipose tissue) is the main site of adaptive non-shivering thermogenesis and allows small mammals as well as neonates to adapt to cold environments. Moreover, BAT thermogenesis contributes to energy expenditure in response to overfeeding, and BAT thermogenic activity protects against obesity [1,2]. In humans, BAT has traditionally been considered physiologically relevant only in the neonatal period [1]. However, a previous study demonstrated the presence of significant amounts of active BAT in adult humans and showed an association of BAT with lowered body mass index [3].

The cell type specialized in non-shivering thermogenesis is the brown adipocyte, which contains a large number of mitochondria with a high oxidative capacity. Brown adipocyte mitochondria are naturally uncoupled due to the presence of UCP (uncoupling protein) 1, a protein exclusively expressed in brown adipocytes that confers on BAT mitochondria their capacity to produce heat. Brown adipocytes also contain a large number of lipid droplets capable of providing the fuel for oxidation and heat production [1]. Despite the widely recognized opposite physiological roles of BAT and WAT (white adipose tissue) (energy dissipation and energy accumulation respectively), the high intracellular lipid content of both brown and white adipocytes has led to the traditional concept that these two cell types differentiate from closely related precursor cells. However, recent research indicates that the brown adipocyte differentiation lineage is substantially different from the white adipocyte lineage and, in fact, brown adipocytes share mesenchymal precursor cells that closely correspond to the myogenic lineage of differentiation [4]. Several transcription factors, including PPARγ (peroxisome proliferator-activated receptor γ), play similar roles in BAT and WAT, essentially promoting lipid accumulation, whereas other transcription factors and co-regulators, such as PGC-1α (PPARγ co-activator-α) or PRDM16 (PR-domain-containing 16), act at distinct stages of the commitment of mesenchymal cells, driving specific differentiation towards a brown adipocyte phenotype [5]. Moreover, transcription factors such as C/EBP (CCAAT/enhancer binding protein) β are specifically involved in driving common precursors of brown and myogenic lineages toward the brown adipocyte differentiation pathway [6].

Pref-1 (pre-adipocyte factor-1), also known as Dlk-1 (delta-like-1 homologue) and FA-1 (fetal antigen-1), is an EGF-repeat-containing transmembrane protein. The biologically active soluble form of Pref-1 is produced by cleavage of the extracellular domain of Pref-1 by TACE [TNFα (tumour necrosis factor α)-converting enzyme] [7]. Pref-1 is highly expressed in embryonic tissues and is associated with the undifferentiated status of multiple cell types of the mesenchymal lineage. Pref-1 is highly expressed in white pre-adipocytes, and its expression is strongly down-regulated in association with the initiation of the white adipogenic differentiation process [8]. In fact, Pref-1 is known to play a pivotal role in inhibiting the differentiation of pre-adipocytes into white adipocytes both in vitro and in vivo [811] by mechanisms involving MAPK (mitogen-activated protein kinase)/ERK (extracellular-signal-regulated kinase) signalling, induction of the Sox (SRY-box-containing) gene and subsequent inhibition of C/EBPβ and C/EBPδ [12]. Some studies also suggest that Pref-1 could exert its action on adipocyte differentiation through modulation of Notch pathways [13]. The effects of Pref-1 on white adipocyte differentiation may be distinctly different depending on the extent of commitment of precursor cells to the adipogenic pathway. For instance, in C3H10T1/2c mesenchymal cells, Pref-1 does not exert the powerful anti-adipogenic action that it displays in 3T3-L1 pre-adipocytes or mouse embryonic fibroblasts [14].

The biological role and regulation of Pref-1 in BAT has not yet been directly addressed. In the present study, we analysed the regulation of Pref-1 expression during brown fat development and differentiation, and assessed the effects of manipulating Pref-1 levels on BAT and brown adipocyte differentiation. We established that Pref-1 is preferentially involved in controlling the specific thermogenic program of BAT, and also identified the transcription factor C/EBPδ as a novel transcriptional regulator of Pref-1 gene expression.

MATERIALS AND METHODS

Animals

The care and use of mice were in accordance with the European Community Council Directive 86/609/EEC, and were approved by the Comitè Ètic d'Experimentació Animal of the University of Barcelona. Mice were housed in same-sex groups of three animals per cage, in a room with constant temperature (21°C) and relative humidity (30–40%), and a 12h/12h light/dark cycle. Food pellets (A03, Panlab) and water were available ad libitum. Swiss Webster mice were used for studies of development changes. The day of pregnancy was determined by the presence of a vaginal plug (day 0). For studies in fetuses, Caesarean sections of pregnant mice were performed at the indicated days of gestation. Neonates (0 h to 21 days after birth) were killed by decapitation. Interscapular BAT was harvested, immediately frozen in liquid nitrogen and stored at −80°C. For studies using Pref-1-null mice [9] and C/EBPα-null mice [15], heterozygous female mice were mated with heterozygous males. Studies in Pref-1-null neonates and C/EBPα-null neonates were performed in pups 24 h after birth and 4 h after birth respectively. Mice were killed by decapitation. Interscapular BAT, liver and heart were harvested, and, where indicated, an aliquot was immediately frozen in liquid nitrogen and stored at −80°C until RNA or proteins were isolated, and a separate aliquot was fixed for TEM (transmission electron microscopy) analysis. Wild-type, heterozygous and homozygous mice were obtained from the same litter in each experiment. Where possible, samples from two or three pups were pooled for each experimental condition. At least three different litters were analysed independently for each developmental age. Transgenic mice that express C/EBPα under the control of the albumin enhancer/promoter were generated as described previously [16]. This line was bred with the C/EBPα-null strain to generate mice that express C/EBPα exclusively in the liver. Pups were studied on days 2 or 7 after birth and genotyped as described previously [16]. Transgenic wild-type (TG+, C/EBPα+/+) and homozygous (TG+, C/EBPα−/−) mice were obtained from the same litter in each experiment, and at least three different litters were analysed independently for each postnatal age.

TEM and stereological analysis

BAT samples were fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) and postfixed in 1% OsO4 and 0.8% FeCNK in phosphate buffer. After dehydration in a graded acetone series, tissue samples were embedded in Spurr resin. Ultrathin sections were stained with uranyl acetate and lead citrate and examined with a Hitachi H600AB transmission electron microscope at 75 kV. A total of two to three mice were analysed for each developmental age and genotype. For stereological quantification, the proportion of lipid with respect to cell volume was estimated using the volume density method [17], and mitochondrial inner membrane surface density was calculated employing a vertical sections method [18,19]. At least six distinct images from each mouse at each developmental age and genotype were analysed.

Cell culture

Differentiation of brown adipocytes in primary culture was performed as described previously [20]. Precursor cells were isolated from the interscapular, cervical and axillary depots of BAT from 21-day-old mice. Experiments were performed on day 8 or 9 of culture (a point at which 90% of cells were considered to be differentiated on the basis of lipid accumulation and brown adipocyte morphology under standard conditions of culture), or at earlier stages of differentiation, as indicated. For studies on the effects of corticosteroids, pre-adipocytes were exposed to 30 nM dexamethasone (Sigma) from culture day 4 onwards. Mouse HIB-1B cells [21] were grown in DMEM:F12 (1:1 mixture of Dulbecco's minimal essential medium/Ham's F12, Gibco) containing 10% (v/v) FBS (fetal bovine serum), 100 units/ml penicillin G and 100 μg/ml streptomycin. Differentiation was induced by growing HIB-1B cells in DMEM:F12 supplemented with 20 nM insulin and 1 nM T3 (3,3′,5-tri-iodothyronine). Once cells had reached 100% confluence, they were placed in DMEM:F12 supplemented with 20 nM insulin, 1 nM T3, 0.5 mM IBMX (3-isobutyl-1-methylxanthine), 0.5 μM hydrocortisone and 0.125 mM indomethacin for 24 h. HIB-1B cells were then maintained for 5–7 days in DMEM:F12 containing 5% (v/v) FBS, 20 nM insulin and 1 nM T3 to allow acquisition of a differentiated morphology. The cell diferentiation state was determined by assessing cytoplasmic fat accumulation using intracellular triglyceride staining with Oil Red O (Sigma).

RNA isolation, Northern blotting and quantitative real-time PCR analyses

Total RNA was extracted using the RNeasy Mini Kit (Qiagen). For Northern blot analysis, 10–15 μg of total RNA was denatured, electrophoresed on 1.5% formaldehyde/agarose gels and transferred on to positively charged nylon membranes (N+ Boehringer Mannheim). Equal loading of gels was confirmed by ethidium bromide staining and hybridization with an 18S rRNA probe. Prehybridization and hybridization were carried out as described previously [22]. Autoradiographs were quantified by densitometric analysis (Phoretics, Millipore). mRNA expression was quantitatively analysed using TaqMan real-time RT (reverse transcription)–PCR. The RT reaction was performed using 0.5 μg of RNA in a 25 μl volume containing 1.25 units/μl MultiScribe™ Reverse Transcriptase, 0.4 units/μl RNase inhibitor, 500 μM dNTP mixture, 2.5 μM random hexamers, 1× RT buffer and 5.5mM MgCl2 solution (all from Applied Biosystems). TaqMan Real-Time PCR reactions were performed using TaqMan Universal PCR Master Mix and the following standardized (‘Assay-on-Demand’) gene expression primers/probes: C/EBPα (Mm00514283_s1), C/EBPβ (Mm00843434_s1), C/EBPδ (Mm00786711_s1), UCP1 (Mm00494069_m1), aP2 (adipose protein 2)/FABP4 (fatty acid binding protein 4) (Mm00445880_m1), Pref-1 (Mm00494477_m1) and PGC-1α (Mm00447183_m1). COII (cytochrome c oxidase subunit 2) was detected using the custom-designed primers 5′-CAAACCACTTTCACCGCTACAC-3′ (forward) and 5′-GGACGATGGGCATGAAACTGT-3′ (reverse) and the FAM (6-carboxyfluorescein)-labelled probe 5′-AAATCTGTGGAGCAAACC-3′. The amount of mRNA for the gene of interest in each sample was normalized to that of the housekeeping reference 18S rRNA (Hs99999901, Applied Biosystems). Samples were run in duplicate on the ABI/Prism 7700HT Sequence Detection System (Applied Biosystems).

Immunoblot analysis

Samples of cell homogenates were separated by SDS/PAGE on 12.8% gels and transferred on to PVDF membranes (Immobilon-P, Millipore). Membranes were then incubated with primary antibodies against C/EBPα (sc-61X, 1:1000 dilution; Santa Cruz Biotechnology), C/EBPβ (sc-150, 1:300; Santa Cruz Biotechnology), C/EBPδ (sc-151X, 1:1000 dilution; Santa Cruz Biotechnology) or β-actin (AC-15, 1:10000 dilution; Sigma). β-Actin detection was used to ensure equal protein loading. Bound antibodies were detected using horseradish peroxidase-coupled anti-mouse (170-6516, 1:3000 dilution; Bio-Rad) or anti-rabbit (sc-2004, 1:3000 dilution; Santa Cruz Biotechnology) secondary antibodies and an ECL (enhanced chemiluminescence) detection kit (GE Healthcare). When required, quantitative analysis was performed by densitometry (Phoretics ID software).

Transient transfection assays

Transfection assays were carried out in HIB-1B cells at 50% confluence using FuGene6 Transfection Reagent (Roche) according to the manufacturer's instructions. Each experimental point was assayed in triplicate in a six-well plate. Cells were transfected with plasmids (1.5 μg) containing promoter–reporter constructs in which expression of the firefly (Photinus pyralis) luciferase gene was driven by variable-length promoter regions of the rat Pref-1 gene [−2.5KPref-1–Luc, −412Pref-1–Luc and −42Pref-1–Luc; gifts from Dr Hiroshi Takemori (Osaka University Medical School, Osaka, Japan)], as described previously [23]. Empty pGL3 vector (1.5 μg, Promega) was used as a negative control. In each transfection, cells were co-transfected with phRL-TK sea pansy (Renilla reniformis) luciferase expression vector (0.5 ng, Promega) as a control for transfection efficiency. Where indicated, cells were co-transfected with 0.5 μg of plasmid vectors driving the expression of C/EBPα (pMSV-C/EBPα) and C/EBPδ (pMEX-C/EBPδ), kindly provided by Professor Steven McKnight (University of Texas Southwestern Medical Center, Dallas, TX, U.S.A.) and Dr Peter Johnson (Laboratory of Protein Dynamics and Signaling, NCI-Frederick, Frederick, MD, U.S.A.) respectively. Expression vectors for the active (LAP) C/EBPβ isoform (pCMV-LAP) and truncated inhibitory (LIP) C/EBPβ isoform (pCMV-LIP) were a gift from Professor Ueli Schibler (Department of Molecular Biology, University of Geneva, Geneva, Switzerland). Cells were incubated for 48 h after transfection. Firefly luciferase and Renilla luciferase activities were measured in a Turner Designs Luminometer (Model TD20/20) using the Dual Luciferase Reporter Assay System kit (Promega) as described by the manufacturer. Luciferase activity elicited by Pref-1 promoter constructs was expressed relative to the Renilla luciferase activity to normalize for variations in transfection efficiency.

C/EBPδ-knockdown assays

For knockdown of C/EBPδ expression, 40 pmol/plate siRNA (small interfering RNA) duplex specific for mouse C/EBPδ (sc-37723, Santa Cruz Biotechnology) and scrambled control siRNA were transfected into proliferating HIB-1B cells using transfection reagent (sc-29528, Santa Cruz Biotechnology) and following the manufacturer's instructions. After 48 h, cells were harvested and C/EBPδ protein (immunoblot) and Pref-1 mRNA and aP2 mRNA (quantitative RT–PCR) were measured.

ChIP (chromatin immunoprecipitation) assay

ChIPs were performed as described previously [24] with a few modifications. For ChIPs using tissue, brown fat from fetuses at term was dissected and minced to a homogeneous consistency. Samples of BAT or HIB-1B cells were fixed for 10 min with 1% formaldehyde. Formaldehyde was then quenched by incubating with 125 mM glycine for 5 min. Cells were washed with PBS, and immediately scraped off the dishes. In both cases (tissue and cells) samples were harvest by centrifugation at 11000 g for 2 min. Nuclei were isolated by means of incubation with WB1 buffer (10 mM Tris/HCl, pH 8.0, 0.25% Triton X-100, 1 mM EDTA and 0.5 mM EGTA) for 10 min, centrifugation at 11000 g for 2 min and incubation with WB2 (10 mM Tris/HCl pH 8.0, 200 mM NaCl, 1 mM EDTA and 0.5 mM EGTA) for 10 min. After a final centrifugation at 11000 g for 2 min, the pellet was resuspended in SDS lysis buffer (15 mM Tris/HCl, pH 8.0, 10 mM EDTA and 1% SDS) and incubated for 10 min. Lysates were then sonicated on ice using a VC50 50 W sonicator (Sonics & Materials) to obtain 400–2000 bp chromatin fragments, and immunoprecipitated using an antibody against C/EBPδ (sc-151X, Santa Cruz Biotechnology) by incubating overnight at 4°C using Dynabeads M-280 Sheep anti-Rabbit IgG (Invitrogen). Pre-immune serum was used as a negative control to confirm the specificity of the assay. Thereafter, magnetic beads were washed once with low-salt buffer (20 mM Tris/HCl, pH 8.0, 0.1% SDS, 1% Triton X-100, 2 mM EDTA and 150 mM NaCl) and then twice with TE (Tris/EDTA) buffer (10 mM Tris/HCl, pH 8.0, and 1 mM EDTA). After elution with 1% SDS in 0.1 M NaHCO3 for 15 min, the DNA–protein cross-links were reversed by addition of NaCl to a final concentration of 200 mM and incubation at 65°C for 5 h. Finally, after adjusting the volume to achieve final EDTA and Tris/HCl concentrations of 10 mM and 40 mM respectively, proteins were digested by adding 40 μg/ml Proteinase K and incubating at 45°C for 1 h. The DNA was extracted with phenol/chloroform and resuspended in 20 μl of TE buffer. PCR was performed using 2 μl of DNA, 2.5 units of Taq DNA Polymerase, 1× PCR buffer, 1× Q-Solution, 200 μM dNTP mixture, 1.5 mM MgCl2 and 0.5 μM of each primer in a final volume of 25 μl. Amplification products were resolved on 2% agarose gels and visualized by ethidium bromide staining. The −381/−37 fragment of mouse Pref-1 was amplified with the primer pair 5′-GCGCGGGACTCCAGCCCTAAGT-3′ and 5′-GCGGTGCAGGGGCTGCTCCGGG-3′. A primer set that amplified a +6438/+6825 fragment of Ucp3 (5′-CATAGGCAGCAAAGGAAC-3′ and 5′-CTATATGGTTTACACAGC-3′) was used as a control for Pref-1 enrichment. When required, quantitative analysis was performed by densitometry (Phoretics ID software).

Statistical analysis

Results are expressed as means±S.E.M. Differences between means were analysed using Student's t test; P<0.05 was considered to be statistically significant.

RESULTS

Developmental regulation of Pref-1 gene expression in BAT

The analysis of Pref-1 mRNA levels by quantitative RT–PCR (Figure 1A) or Northern blotting (Figure 1B) revealed that Pref-1 was highly expressed in fetal BAT. The highest levels occurred on fetal day 16 and decreased somewhat in the late fetal period, prior to birth. The level of Pref-1 mRNA progressively decreased after birth, becoming almost undetectable in brown fat from 21-day-old mice (Figure 1A) and adult BAT, even using highly sensitive RT–PCR methods. In contrast, Pref-1 mRNA was readily detected by conventional Northern blotting in BAT from fetuses and young neonatal mice (Figure 1B). The pattern of Pref-1 mRNA expression during BAT differentiation/development was compared with that of UCP1, a specific marker of brown adipocytes. Consistent with previous reports [25,26], we found that Ucp1 mRNA expression was initiated in late fetal life and continued to increase in early neonatal life, peaking on day 2 after birth, before gradually declining in adulthood.

Figure 1 Expression of the Pref-1 gene in developing BAT

(A) Pref-1 mRNA expression during BAT development was measured in parallel with mRNA expression of UCP1, a differentiation marker. Each point corresponds to the mean of mRNA expression of at least three independent experiments for each time point of development, expressed as percentages of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). (B) Representative Northern blot of Pref-1 mRNA changes during BAT development. Liver and lung samples from adult mice were included as negative controls.

Pref-1 is required for an appropriate development of the thermogenic gene expression program in neonatal BAT

In order to determine the role of Pref-1 in BAT development, we analysed the impact of targeted deletion of the Pref-1 gene in brown fat at distinct stages of development. In the late fetal period, brown adipocytes from Pref-1-null mice appeared normal in terms of lipid droplet accumulation and mitochondrial morphology (Figure 2A and Table 1) However, at 2 days after birth, brown adipocytes from Pref-1-null mice showed significantly less accumulation of lipid droplets, although mitochondria appeared well developed, as in wild-type mice (Table 1). In young (21-day-old) mice, lipid droplets in Pref-1-null BAT were smaller than those in wild-type brown fat (Figure 2A) and occupied less cellular space (Table 1). The mean surface of inner mitochondrial membrane with respect to the surface of external mitochondria, an index of mitochondrial maturation in BAT [19,27], was not significantly altered in Pref-1-null mice (Table 1). A gene expression analysis revealed that marker genes of specific brown fat thermogenesis were abnormally overinduced in brown fat from Pref-1-null mice after birth: in newborns PGC-1α was overexpressed, whereas both PGC-1α and UCP1 were overexpressed in 21-day-old mice (Figure 2B). A similar trend was observed for 5′-deiodinase, another marker gene of BAT thermogenic activation (results not shown). In contrast, markers of overall adipogenesis, which do not distinguish between brown and white adipocyte phenotypes, such as aP2, were unaltered at the distinct stages of brown fat development analysed.

View this table:
Table 1 Stereological analysis of lipid content and mitochondrial inner membrane surface ratio in BAT from wild-type and Pref-1-null mice

Values are means±S.E.M. for at least six samples. **P<0.01 and ***P<0.001 indicate statistically significantly differences between Pref-1-null and wild-type mice; #P<0.05 and ##P<0.01 indicate statistical significance for comparisons with respect to fetuses; and +++P<0.001 indicates statistical significance for comparisons with respect to newborns. SIMM/SEMM, ratio of surface area of inner mitochondrial membrane to surface area of external mitochondrial membrane.

Figure 2 Effects of Pref-1 gene ablation in developing BAT

Analysis of BAT from homozygous Pref-1-null mice (Pref-1 KO) and wild-type (WT) littermates at fetal day 18 (fetus), and 2 days (Newborn) and 21 days after birth. (A) TEM analysis. Two or three mice were analysed for each genotype and developmental day. Scale bars: 2 μm (upper panels) and 0.4 μm (lower panels). N, nucleus; m, mitochondrion; L, lipid droplet. (B) Each bar shows the mean±S.E.M. of the relative mRNA level for three independent experiments from each time point of development and genotype, expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). Statistical significance of the comparisons between WT and Pref-1-null samples is shown as *P<0.05.

Pref-1 mRNA expression in BAT from C/EBPα-null fetuses and neonates

In order to further analyse the role of Pref-1 in BAT development, we studied C/EBPα-null mice, a model of specific impairment in BAT differentiation during fetal development [15,27]. As shown in Figure 3(A) and Table 2, brown adipocytes from C/EBPα-null mice showed defective differentiation, characterized by reduced accumulation of lipids and impaired mitochondrial maturation. Consistent with this, expression of the Ucp1 gene was significantly impaired in brown fat from C/EBPα-null fetuses and neonates (Figure 3B, left-hand panel). Brown fat from mice lacking C/EBPα showed a dramatic induction of Pref-1 mRNA levels, both in early fetal life (day 17) or in the immediate postnatal period (Figure 3B, middle and right-hand panels).

View this table:
Table 2 Stereological analysis of lipid content and mitochondrial inner membrane surface ratio in BAT from wild-type and C/EBPα-null fetuses

Values are means±S.E.M. for at least six samples. ***P<0.001 indicates a statistically significant difference between C/EBPα-null and wild-type mice; #P<0.05 and ##P<0.01 indicate statistical significance for the comparison between newborns and fetuses at day 17. SIMM/SEMM, ratio of surface area of inner mitochondrial membrane with respect to surface area of external mitochondrial membrane.

Figure 3 Pref-1 mRNA expression in BAT from C/EBPα-null mice

(A) TEM analysis. Scale bars: 8 μm (fetus), and 2 μm, (newborn and day 21). N, nucleus; m, mitochondrion; L, lipid droplet. (B) Left-hand and middle panels: Ucp1 mRNA and Pref-1 mRNA levels in interscapular BAT from homozygous C/EBPα-null mice (KO) and wild-type (WT) littermates in fetal day 17 and newborn mice. Each bar shows the mean±S.E.M. of the relative mRNA level for three independent experiments from each time point of development and genotype, expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). Right-hand panel: representative Northern blot of Pref-1 mRNA expression. Statistical significance of the comparisons between WT and C/EBPα-null samples are shown as *P<0.05, and those between newborns and fetuses is shown as #P<0.05.

Pref-1 mRNA expression in BAT during the post-natal period in mice lacking expression of C/EBPα in BAT

C/EBPα-null mice cannot be studied after birth because the impairment in liver gluconeogenesis caused by the lack of C/EBPα leads to mortality in the postnatal period [15]. To overcome this limitation, we used C/EBPα-null mice incorporating liver-specific transgenic expression of C/EBPα (TG+, C/EBPα−/− mice) which therefore have normalized glucose homoeostasis [16]. An analysis of the expression of Pref-1 in BAT from neonatal TG+, C/EBPα−/− mice showed that on both days 2 and 7 after birth, Pref-1 mRNA levels were still overexpressed in these mice relative to wild-type transgenic (TG+, C/EBPα+/+) mice (Figure 4). In fact, Pref-1 mRNA levels in BAT declined progressively after birth in TG+, C/EBPα−/− mice and, in 7-day-old transgenic pups, Pref-1 mRNA levels were less than 5% of the overexpressed levels in C/EBPα-null fetuses. A parallel assessment of UCP1 indicated that the level of UCP1 was normalized by postnatal day 7, but not yet by day 2 (results not shown).

Figure 4 Pref-1 mRNA expression in BAT from trangenic C/EBPα-null mice expressing C/EBPα in liver

Pref-1 mRNA in interscapular BAT from transgenic homozygous C/EBPα-null mice (Tg+, C/EBPα-KO) and wild-type (Tg+, WT) at neonatal days 2 and 7. Each bar shows the mean±S.E.M. of the relative mRNA level from three independent experiments for each time point of development and genotype, expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). Statistical significance of the comparisons between wild-type (Tg+, C/EBPα+/+) and knockout (Tg+, C/EBPα−/−) samples are shown as *P<0.05.

Changes in the expression of C/EBP transcription factors during BAT development: effects of Cebpa gene invalidation

The pattern of expression of members of the C/EBP family of transcription factors during BAT development was determined (Figure 5). C/EBPα and C/EBPβ expression followed a similar pattern: low in fetuses and beginning to rise after birth, reaching high levels at neonatal days 15 and 21. C/EBPδ expression showed a quite different pattern of developmental regulation. Although Cebpd mRNA levels were also low in early fetuses, they rose dramatically a few days before birth and attained the highest levels at term and in 1-day-old neonates. This was followed by a rapid decay, such that within a few days after birth and through day 21 Cebpd mRNA levels were again low. In accordance with a previous report [27], we confirmed that the expression of C/EBPβ and C/EBPδ proteins were overinduced in brown fat from C/EBPα-null neonates compared with wild-type neonates. The active LAP form of C/EBPβ was induced 3-fold, the inhibitory LIP form of C/EBPβ protein was induced 1.7-fold and C/EBPδ was induced 3.7-fold. Moreover, the expression of C/EBPδ, but not C/EBPβ, remained slightly increased in BAT from 7-day-old TG+, C/EBPα−/− mice (2.2-fold induction), also in agreement with previous reports [16].

Figure 5 Expression of C/EBP genes in developing BAT

Changes in Cebpa (C/EBPα), Cebpb (C/EBPβ) and Cebpd (C/EBPδ) mRNA expression. Each point corresponds to the mean for at least three independent experiments at each time point of development, expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units).

Pref-1 mRNA expression in primary brown adipocytes differentiating in culture

The mouse model studies above suggested that Pref-1 regulation is modulated by C/EBPα and/or by effects secondary to Cebpa gene ablation (e.g. overexpression of other C/EBPs). To gain insight into the relationship between Pref-1 expression and C/EBPs in BAT differentiation, we undertook further studies using a cell culture model of brown adipocyte differentiation. For primary cultures, non-differentiated precursor cells were obtained from mouse BAT and cultured under conditions that lead to brown adipocyte differentiation. Day 3 cultured cells had not yet differentiated, showing a fibroblast-like appearance and no lipid accumulation (Figure 6A). At this stage of culture, expression of brown adipocyte differentiation markers (Ucp1, Co2 or aP2) was low, whereas Pref-1 mRNA levels were high. As differentiation progressed, cells acquired a brown adipocyte morphology, characterized by lipid accumulation and increased expression of differentiation markers, and Pref-1 mRNA was dramatically down-regulated (Figure 6A).

Figure 6 Expression of Pref-1, marker genes of brown adipocyte differentiation markers and C/EBPs in brown adipocytes differentiating in primary culture

(A) Pref-1, Co2 (COII), aP2 and Ucp1 mRNA levels (left-hand panel), and representative micrographs of cultures of non-differentiated brown pre-adipocytes (day 3) and differentiated brown adipocytes (day 7) (right-hand panel). (B) Cebpa (C/EBPα), Cebpb (C/EBPβ) and Cebpd (C/EBPδ) mRNA levels. Each point corresponds to the mean for three experiments from independent cell cultures expressed as a percentage of the mean value at the time of maximal expression of each gene, which was set to 100 (arbitrary units). (C) Immunoblot analysis of C/EBPα, C/EBPβ and C/EBPδ proteins at 4 and 7 days of culture. Molecular mass in kDa is indicated on the right-hand side of each blot.

In this experimental setting, Cebpa and Cebpb mRNA expression increased as cells differentiated, more abruptly for Cebpa and progressively for Cebpb. In contrast, Cebpd expression was highest in non-differentiated cells (days 3 and 5 of culture) and decreased markedly with brown adipocyte differentiation (Figure 6B). These changes at the mRNA level were confirmed at the protein level (Figure 6C). Thus, Cebpa gene-encoded proteins (specifically the 42 kDa form) were higher in differentiated brown adipocytes (day 7) than in non-differentiated brown adipocytes (day 4). The same was observed for the C/EBPβ proteins LAP and LIP, although in fact the increase in LIP expression (measured as band density in immunoblots) was more marked in differentiated brown adipocytes. However, the opposite was observed for C/EBPδ, which was dramatically down-regulated in differentiated brown adipocytes (Figure 6C). These results indicated that, during differentiation of brown adipocytes, the expression profile of Pref-1 is inversely related to C/EBPα and C/EBPβ, but positively correlated with that of C/EBPδ.

In primary cultures of brown adipocytes, proliferation arrest and differentiation occur within a short period of time (between day 3/4 and day 7 of culture) making it impossible to distinguish events that occur in association with each process. To circumvent this limitation, we used HIB-1B cells as a second model of brown adipocyte differentiation. These cells can be studied during active proliferation and upon reaching confluence and ceasing proliferation, but in the absence of differentiation. They can also be studied several days after the induction of differentiation of confluent cells when they have acquired morphological (lipid accumulation) and molecular (induction of Ucp1 mRNA expression by noradrenaline) features of differentiated brown adipocyte [21,22]. Using this model, we found that C/EBPα proteins were significantly expressed only in differentiated HIB-1B brown adipocytes, and C/EBPβ LAP and LIP proteins were also mostly expressed in differentiated cells. In contrast, C/EBPδ expression was high in non-differentiated HIB-1B cells, both proliferating and confluent, and was reduced in association with differentiation (Figure 7A). Pref-1 mRNA levels were lower in differentiated HIB-1B brown adipocytes than in non-differentiated cells, either proliferating or confluent (Figure 7B), consistent with the positive association between Pref-1 and C/EBPδ expression observed in primary brown adipocyte cultures. These results indicate that the expression of C/EBPδ, in contrast with C/EBPα and C/EBPβ, is negatively associated with differentiation and positively associated with Pref-1 expression in the HIB-1B brown adipocyte cell model.

Figure 7 Changes in C/EBP proteins across the three stages of H1B-1B cell differentiation

(A) Representative immunoblot of three independent assays of protein extracts from proliferating, confluent and differentiated HIB-1B cells (see the Materials and methods section). Estimated sizes of C/EBP proteins in kDa are indicated on the right-hand side. CRM, cross-reactive material. (B) Bars show mean Pref-1 mRNA levels±S.E.M. for at least three independent experiments performed in triplicate. Statistically significant differences compared with proliferating cells are shown as **P<0.01.

Transcriptional regulation of the Pref-1 gene by C/EBPs: transcriptional activation by C/EBPδ

The previous results prompted us to investigate the direct effects of C/EBPs on Pref-1 gene expression. To this end, we transiently transfected HIB-1B with a Pref-1 gene promoter–luciferase reporter construct and studied the effects of co-transfection of expression vectors for individual C/EBP proteins (Figure 8). As shown in Figure 8(A), whereas C/EBPα did not significantly affect Pref-1 promoter activity, the inhibitory LIP form of C/EBPβ reduced Pref-1 promoter activity and the activating LAP form significantly induced it. However, the strongest induction of the Pref-1 gene promoter was observed with co-transfection of an expression vector for C/EBPδ. An in silico analysis of the Pref-1 gene promoter sequence revealed the presence of several potential C/EBP-binding sites in the proximal promoter region, between −412 and −47. Deletion analysis showed that most of the responsiveness to C/EBPδ was lost in a short construct containing only the −47 region, thus indicating that C/EBPδ responsiveness lies in this −412/−47 region (Figure 8B). Moreover, a longer construct, containing −2.5 kb of the Pref-1 promoter did not exhibit enhanced responsiveness to C/EBPδ relative to that of the −412 construct (results not shown). Thus it is unlikely that other sites in the 5′ region of the promoter upsteam of −412 were significantly involved in mediating the C/EBP responsiveness of the Pref-1 gene.

Figure 8 Effects of C/EBPs on Pref-1 promoter activity

(A) HIB-1B cells were co-transfected with −412-Pref-1–Luc (luciferase) and C/EBPα, LAP, LIP or C/EBPδ expression vectors, or empty vector (control). (B) HIB-1B cells were transfected with −412-Pref-1–Luc or the deletion-mutant form −47-Pref-1–Luc, and co-transfected with a C/EBPδ expression vector. Results are means±S.E.M. for at least three independent experiments performed in triplicate. Statistical significance of differences with respect to controls (C) is shown as *P<0.05, and that for the comparison of the activity of the two promoter constructs at a similar co-transfection condition is shown as #P<0.05.

In order to ascertain whether the activating role of C/EBPδ on the Pref-1 gene transcription occurs in vivo and affects endogenous Pref-1 gene expression, a loss-of-function approach was followed through siRNA-mediated knockdown of C/EBPδ. By these means, a siRNA reduction in C/EBPδ protein expression was achieved in HIB-1B cells, and this resulted in a significant and specific reduction in Pref-1 mRNA, but not in aP2 mRNA levels (Figure 9), thus confirming the involvement of C/EBPδ in Pref-1 gene regulation.

Figure 9 Effects of C/EBPδ knockdown on Pref-1 gene expression

C/EBPδ protein (A), and Pref-1 and aP2 mRNA levels (B) in HIB-1B cells previously treated with siRNA for C/EBPδ or scrambled siRNA [control (Ctrl)]. An example of a C/EBPδ immunoblot is shown in the right-hand panel in (A). Results are means±S.E.M. for at least three independent experiments performed in triplicate. Statistical significance of differences with respect to controls is shown as *P<0.05 and **P<0.01.

ChIP assays reveal C/EBPδ binding to the endogenous Pref-1 gene promoter

Next, we studied whether C/EBPδ effectively binds to the proximal promoter region of the Pref-1 gene using ChIP assays. Because Pref-1 mRNA is highly expressed in BAT from fetuses at term, a time that corresponds to the point in brown fat development when C/EBPδ expression is highest (see Figure 5), we used BAT from fetal day 19 mice for these assays, and we compared the results with those obtained using BAT from 21-day-old mice, a condition of almost negligible expression of the Pref-1 gene. As shown in Figure 10(A), the fragment of DNA corresponding to the 5′ region of the Pref-1 gene was specifically enriched when chromatin from fetal BAT was immunoprecipitated with a C/EBPδ-specific antibody, but not when 21-day-old BAT was analysed. Identical enrichment due to anti-C/EBPδ antibody was observed in HIB-1B cells (Figure 10B).

Figure 10 ChIP analysis of the binding of C/EBPδ to the endogenous Pref-1 gene promoter in BAT and HIB-1B cells

Anti-C/EBPδ antibody or pre-immune serum was used to immunoprecipitate protein–DNA complexes in late fetal BAT and 21-day-old BAT (A), and in HIB-1B cells (B). Representative images of PCR amplification are shown; arrowheads indicate the 530 bp Pref-1 gene PCR product and the 400 bp Ucp3 gene PCR product (Control). Bars show means±S.E.M. for four independent experiments as arbitrary units of fold enrichment with respect to pre-immune serum after densitometric scanning of PCR products. Statistical significance of differences between anti-C/EBPδ antibody and pre-immune incubation is shown as *P<0.05.

Effects of dexamethasone on Pref-1, C/EBPδ expression and differentiation of brown adipocytes

The finding that C/EBPδ was a direct activator of Pref-1 gene transcription in brown adipocytes led us to investigate the role of glucocorticoids, considering that C/EBPδ is considered a direct target of dexamethasone, at least in the context of white adipocyte differentiation [28]. Exposure of brown adipocyte precursors to dexamethasone during the differentiation process completely prevented the decay in Pref-1 mRNA expression (Figure 11A). However, the appearance of brown adipocyte morphology was completely preserved (Figure 11B). Dexamethasone caused a significant increase in Cebpd expression, an effect that was associated with dramatically impaired induction of the specific brown adipocyte markers Ucp1 and Ppargc1a (encoding PGC-1α) (Figure 11C). In accordance with the morphological observations, the marker gene of overall adipocyte differentiation, aP2, was unaltered.

Figure 11 Effects of dexamethasone on the differentiation of brown adipocytes, and the expression of Pref-1 and C/EBPδ

(A) Pref-1 mRNA expression during the differentiation of brown adipocytes in primary culture exposed to 30 nM dexamethasone (Dexa) from day 4 onwards. Each point corresponds to the mean±S.E.M. for four experiments from independent cell cultures expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). (B) Representative micrographs of brown adipocytes on day 8 of culture, grown in the absence (control) or presence of 30 nM dexamethasone. (C) mRNA expression of C/EBPδ and marker genes of brown adipocyte differentiation in pre-adipocytes (day 5) and differentiated (day 8) brown adipocytes. Each bar is the mean±S.E.M. of the relative mRNA level for at least four independent experiments for each condition, expressed as a percentage of the mean value at the time of maximal expression, which was set to 100 (arbitrary units). Statistically significant differences due to dexamethasone at each time of culture are shown as *P<0.05, **P<0.01, ***P<0.001 and those between day 5 and day 8 are shown as #P<0.05.

DISCUSSION

During development of rodents, BAT develops much earlier than WAT. In fact, brown adipocytes are considerably differentiated at birth, a time when WAT is practically absent. BAT activation and differentiation progress after birth, achieving maximal activity and differentiation in neonates and young pups [1,25,26]. During this period time, we observed that Pref-1 expression in BAT was progressively down-regulated, from very high levels in fetuses to practically undetectable levels several weeks after birth. This would be consistent with a repressive role of Pref-1 on brown adipocyte differentiation, similar to what is known to occur in white adipocytes. However, the fact that mitochondriogenesis, lipid accumulation and expression of marker genes of thermogenesis in BAT from Pref-1-null fetuses were normal suggests that, during fetal life, brown adipocyte differentiation in the absence of environmental thermal stress is not responsive to changes in Pref-1 levels. In contrast, Pref-1-null mice showed clear signs of thermogenic over-activation in BAT from neonates and young pups, including overexpression of thermogenic marker genes and evidence of enhanced mobilization of lipids previously accumulated during fetal development. These findings indicated that Pref-1 could have a specific role in the acquisition of the thermogenic pattern of gene expression in BAT, without significantly impacting other features of the brown adipocyte phenotype. Although the specific role of Pref-1 in BAT had not previously been directly addressed, the present findings are consistent with observations in some rodent models of lipodystrophy in which BAT depots show enhanced levels of Pref-1 in association with a specific loss of thermogenic properties but enhanced fat accumulation [29]. Supporting this notion, expression of the thermogenic gene Ucp1 is strongly impaired in BAT from transgenic mice overexpressing Pref-1 in both WAT and BAT, whereas other genes related to fat accumulation, such as aP2, are only mildly reduced in BAT [10].

These observations prompted us to further analyse the role and regulation of Pref-1 in the process of BAT development and differentiation using C/EBPα-null mice, a well established model of inhibition of BAT differentiation in the perinatal period [27]. These analyses revealed a dramatic induction of Pref-1 gene expression in association with impaired expression of Ucp1 and delayed acquisition of the specific features of thermogenic activation of BAT. The high levels of Pref-1 mRNA expression in the BAT from C/EBPα-null mice could potentially reflect relief from a repressive effect of C/EBPα on the Pref-1 gene by the targeted disruption of the Cebpa gene. This possibility would be compatible with the hypothesis that sustained high levels of C/EBPα are required for maintaining low Pref-1 gene expression in the differentiated brown adipocyte. However, this hypothesis is inconsistent with the time-course of Pref-1 and Cebpa gene expression throughout brown adipocyte differentiation, observed in the cell culture models reported in the present study. In these models, C/EBPα induction is a late phenomenon in the differentiation of brown adipocytes, whereas down-regulation of the Pref-1 mRNA occurs earlier. Notably, the analysis of the transcriptional regulation of the Pref-1 gene revealed that C/EBPα did not have a significant impact on Pref-1 gene promoter transcription, whereas another subtype of the C/EBP family, C/EBPδ, had a powerful inductive effect. Moreover, experimentally induced reduction of C/EBPδ led to a specific reduction in Pref-1 gene expression. The identification of C/EBPδ as a novel and powerful factor in the control of Pref-1 gene transcription strongly suggests that the overexpression of C/EBPδ expression that takes place in the BAT of C/EBPα-null mice ([27] and the present study) could elicit the persistently high levels of Pref-1 mRNA observed in BAT from these mice.

The positive action of C/EBPδ on the Pref-1 gene suggests a specific role for this transcription factor in brown adipocyte differentiation. In white adipocytes, C/EBPδ is induced in the first stages of the differentiation process, a period when Pref-1 expression is undergoing down-regulation [30]. The pattern of expression of C/EBP subtypes during brown adipocyte differentiation reported in the present study, both ‘in vivo’ (late fetal development) and in the two cell culture models analysed (primary cultures of murine brown pre-adipocytes and the HIB-1B brown adipocyte cell line) indicated that C/EBPδ is highly expressed in the brown pre-adipocyte stage, in association with high levels of Pref-1. Given the repressive role of Pref-1 on thermogenic gene expression which could be inferred from results obtained with Pref-1-null mice, we hypothesized that C/EBPδ could play an inhibitory role in thermogenic gene expression in brown adipocytes. This potential role of C/EBPδ could be related to the differential effects of glucocorticoids on brown adipocyte differentiation, especially compared with white adipocytes. Dexamethasone is a powerful inducer of Cebpd gene expression [28] as well as a repressor of Pref-1 gene expression [31] in white pre-adipocytes, events that are considered to favour white adipocyte differentiation. In contrast, glucocorticoids are known to repress the expression of BAT-specific thermogenic genes in brown adipocytes [32], without altering other features common to white adipocyte differentiation such as lipid accumulation. In fact, it has been proposed that, in some cell models, glucocorticoids specifically block brown adipocyte differentiation of precursor cells, driving their differentiation pattern towards a white-adipocyte-like phenotype [33]. Our investigation of the action of dexamethasone in brown adipocyte differentiation and the C/EBPδ–Pref-1 axis established that dexamethasone blunted the Pref-1 down-regulation that occurred with normal brown fat differentiation. These actions were associated with an induction of C/EBPδ expression, maintained acquisition of adipocyte morphology and induction of non-BAT-specific genes on the one hand, and dramatic repression of thermogenic genes such as Ucp1 or Ppargc1a on the other hand. These findings indicate that maintenance of high Pref-1 levels is not compatible with the acquisition of the brown-adipocyte-specific pattern of thermogenic gene expression. Again, these observations in cell culture models are fully consistent with the over-induction of thermogenic genes observed in BAT from Pref-1-null mice, and with previous reports of the association of high Pref-1 with impaired expression of UCP1 but near-normal fat accumulation in BAT in distinct rodent models [10,29].

In summary, we report in the present study that Pref-1 is an important negative regulator of the specific thermogenic pattern of gene expression in BAT. C/EBPδ is a powerful inducer of Pref-1 gene transcription, and its induction by glucocorticoids is likely to provide an indirect mechanism for glucocorticoid induction of the Pref-1 gene in brown pre-adipocytes. The C/EBPδ-mediated induction of Pref-1 by glucocorticoids is thus expected to contribute to the specific inhibitory action of glucorticoids on BAT thermogenic gene expression. The specific role and regulation of Pref-1 in BAT shown here further highlight the distinct processes controlling differentiation of the white and brown adipocyte cell lineages, findings that could inform efforts to modulate BAT activity and differentiation as a means to promote energy expenditure, decrease white fat accumulation and alleviate adiposity in humans.

AUTHOR CONTRIBUTION

Jordi Armengol, Josep Villena, Elayne Hondares and María Carmona performed the experiments and analysed the data. Hei Sook Sul, Roser Iglesias, Marta Giralt and Francesc Villarroya designed the research and analysed the data. Jordi Armengol, Josep Villena and Francesc Villarroya co-wrote the paper. Francesc Villarroya supervised the research.

FUNDING

This study was supported by the Ministerio de Ciencia e Innovación, Spain [grant numbers SAF2008-01896 and SAF2011-23636].

Acknowledgments

We thank Hiroshi Takemori (Osaka University Medical School, Osaka, Japan) for Pref-1 promoter plasmid constructs, and Steven McKnight (University of Texas Southwestern Medical Center, Dallas, TX, U.S.A.), Peter Johnson (Laboratory of Protein Dynamics and Signaling, NCI-Frederick, Frederick, MD, U.S.A.) and Ueli Schibler (Department of Molecular Biology, University of Geneva, Geneva, Switzerland) for plasmid expression vectors. We also acknowledge Gretchen Darlington (Baylor College of Medicine, Houston, TX, U.S.A.) for providing C/EBPα-null mice.

Abbreviations: aP2, adipose protein 2; BAT, brown adipose tissue; C/EBP, CCAAT/enhancer-binding protein; ChIP, chromatin immunoprecipitation; COII, cytochrome c oxidase subunit 2; DMEM:F12, 1:1 mixture of Dulbecco's minimal essential medium/Ham's F12; FBS, fetal bovine serum; PPARγ, peroxisome-proliferator-activated receptor γ; PGC-1α, PPARγ co-activator-α; Pref-1, pre-adipocyte factor-1; RT, reverse transcription; siRNA, small interfering RNA; T3, 3,3′,5-tri-iodothyronine; TE buffer,, Tris/EDTA buffer (10 mM Tris/HCl, pH 8.0, and 1 mM EDTA); TEM, transmission electron microscopy; UCP, uncoupling protein; WAT, white adipose tissue

References

View Abstract