UMR 5018, UPS-CNRS, IFR31, CHU Rangueil, Toulouse, France

Correspondence to: L Casteilla, UMR 5018, UPS-CNRS, IFR31, CHU Rangueil, 1 avenue Jean Poulhès, 31403 Toulouse Cedex 4, France. E-mail: casteil@rangueil.inserm.fr

^aPresent address: U317, CHU Rangueil, 1 avenue Jean Poulhès, 31403 Toulouse Cedex, France.

BACKGROUND: Several indirect arguments agree with the existence of a brown preadipocyte distinct from a white one. Nevertheless, to date, no molecular marker has been available to directly in vivo demonstrate this hypothesis.

OBJECTIVE: The aim of this study was to find a gene expressed in brown preadipocyte but not in white and to use it as a molecular marker to analyse brown preadipocyte recruitment in different physiological and physiopathological situations.

METHOD: Differential display was performed on stromal-vascular and adipocyte fractions of white and brown adipose tissues in rat.

RESULTS: We identified a new gene, BUG, preferentially expressed in the stromal-vascular fraction of brown fat vs other adipose tissues fractions in adult rat. This RNA is also highly expressed in heart and, to a lesser extent, in other tissues such as kidney and brain. The BUG transcript is detected by in situ hybridization in putative preadipocytes within brown adipose tissue. Its level is transiently and specifically up-regulated during early stages of brown preadipocyte differentiation in a primary culture system, before the acquisition of late brown adipocyte phenotype. During development, BUG can be detected before the emergence of UCP-1 expression. In adult rats, BUG expression is inversely associated to brown adipose tissue (BAT) activation during cold exposure as well as in obese animals.

CONCLUSIONS: The pattern of BUG expression agrees with an early divergence between brown and white adipocyte lineages. It also reveals the existence of a pool of committed brown preadipocytes within BAT that are recruited during cold exposure. BUG expression is increased in obese animals, suggesting that an early defect in brown preadipocyte differentiation could account for impaired BAT function in genetically obese rats.

International Journal of Obesity (2001) 25, 1431-1441

preadipocyte; brown adipose tissue; lineage; development; cold exposure

Introduction

White and brown adipose tissues (WAT and BAT) assume quite opposite physiological functions. Their respective adipocytes have both lipolytic and lipogenic properties, but the main role of the white adipocyte is to store energy as triglycerides while the brown adipocyte is specialized in energy dissipation as heat. This basic difference is due to the presence in brown adipocytes mitochondria of a specific protein, UCP-1, which uncouples ATP production from the respiratory chain function.¹

Both types of adipocytes emerge from fibroblast-like precursors of mesenchymal origin.² These precursors are present within adipose tissues throughout lifetime.^3,4,5 They are located in the stromal-vascular fraction (SVF), which can be isolated after collagenase digestion of the extracellular matrix of the tissue. These precursors can be differentiated in primary cultures, but the successive steps leading to adipocyte phenotype acquisition have been mainly studied in preadipocyte cell lines immortalized from mouse.^6,7 Predipocytes can be characterized by the expression of early molecular markers, as A2COL6, pref-1 and FRP-2.^8,9,10 During differentiation into adipocytes, this expression declines and genes coding for late markers like GPDH, aP2 or leptin are switched on. For a long time, UCP-1 was the single gene known to be specifically expressed in the brown adipocyte lineage. Its expression is considered to be a late differentiation event. Recently, PPAR-gamma coactivator-1 (PGC-1) has been cloned from HIB 1B transformed brown adipocytes.¹¹ It has been shown to control mitochondrial activity and biogenesis and could play a fundamental role in BAT activation.¹² It thus provides a new tool to characterize mature brown adipocytes vs white ones. Nevertheless, the time-course of PGC-1 expression and its emergence during brown adipocyte differentiation has not yet been elucidated.

Different indirect observations support the hypothesis of an early divergence between white and brown adipocyte differentiation pathways. First, whatever the species, brown adipocytes generally emerge earlier and in different sites than white ones during development.¹³ Secondly, the recruitment of adipocyte precursors in both types of adipose tissue is partly under the control of the autonomic nervous system, but in opposite ways: in rats, preadipocyte proliferation and differentiation are increased in surgically sympathetic denervated WAT,¹⁴ whereas norepinephrine treatment strongly stimulates the development of preadipocytes in BAT.¹⁵ Last but not least, precursor cells isolated from BAT are able, after growth arrest, to express UCP-1 but this is not the case for precursor cells isolated from WAT and cultured in the same conditions.^16,17

Nevertheless, the existence of brown preadipocytes, with a specific gene expression pattern, has never been directly confirmed. The lack of molecular markers for this stage limits the knowledge on recruitment processes of adipocyte precursors in different physiological or pathophysiological situations.

Differential display is a powerful technique to identify new genes specifically expressed in a cell type. It has been successfully used to compare mRNAs patterns in 3T3-L1 preadipocytes and differentiated adipocytes, resulting in the cloning of genes whose expression is increased during the differentiation.^18,19 Applying this technique to typically WAT and BAT in rats, we cloned a cDNA fragment corresponding to a transcript preferentially expressed in brown adipocyte precursor.

Methods

Animals

Nine-week-old female Wistar and Zücker rats were purchased from Iffa-Credo (France). They were maintained on a 12 h light/dark cycle, at 20°C, and had free access to water and standard food (UAR A04, Villemoisson/Orge, France). To induce brown fat, rats were kept at 4°C for cold exposure or injected intraperitonally with isoproterenol (0.5 mg/kg/day; Sigma, St Louis, MO) or the 3-adrenergic agonist CL316243 (1 mg/kg/day), as indicated. For developmental study, female Wistar rats were mated with adult male rats, a couple per cage for one night, and gestation was confirmed by palpation 2 weeks later. The mean number of pups per litter was 9±1 (10 litters).

Adipose cells isolation

Adipose tissue fractions were obtained as previously described,^20,21 with minor modifications. Brown interscapular and white inguinal fat pads from four to six rats were carefully dissected to remove adhering tissue, capillaries and WAT surrounding the interscapular brown deposit. They were digested separately at 37°C in oxygenated Krebs solution (pH 7.4) containing 20 mM HEPES, 3 mM glucose and 3.5% free fatty acid BSA, with 2 mg collagenase type II (Sigma) per gram of tissue, for 45 min under slow agitation (WAT), or 15 mg collagenase type II (Sigma) per gram of tissue, for 20 min under maximal agitation (BAT). Undigested tissue was then removed by filtering the suspension and floating mature adipocytes were separated from the infranatant, ie the SVF, which contains precursor cells. These cells were pelleted by centrifugation (20 min, 4°C, 500 g).

Total RNA extraction and Northern blot analysis

Total RNAs were prepared from cell fractions or rats by the acidic guanidium isothiocyanate method.²² Twenty micrograms of total RNAs were size fractionated on denaturing formaldehyde (1.1%) agarose gel (1.2%), transferred onto a GeneScreen membrane (NEN, Boston, MA) and UV cross-linked. The homogeneity of RNA amounts present in each lane was controlled by methylene blue staining. After prehybridization (2 h at 42°C in phosphate buffer 0.1 M pH 6.5, formamide 45%, SSC 4´, SDS 0.1%, Denhardt 5´, salmon sperm DNA 75 µg/ml), the membrane was hybridized at 42°C for 16 h in the same buffer (except that Denhardt was 1´) containing the radiolabelled probe. The membrane was washed in SDS 0.1%/SSC 2´-0.1´ solutions between 42 and 65°C, before being exposed on a Kodak X-OMAT AR film at 80°C.

Probes

Fifty nanograms of probes were radiolabelled with 2.5 µCi ³²P-dCTP by the random priming method (Mega Prime DNA labelling System, Amersham, Arlington Heights, IL). Unincorporated dNTPs were removed using QIAquick Nucleotide Removal Kit (QIAGEN, Chatsworth, CA). The BUG probe was the cDNA fragment obtained by differential display with the following primers: 5^'-GTGAGTTAAT-3^' and 5^'-GG(T11)-3^'. The UCP-1 probe was obtained by PCR from position 261 to 1086 of the coding phase,²³ using sense 5^'-AGACATCATCACCTTCCC-3^' and antisense 5^'-CAGCTGTTCAAAGCACAC-3^' primers; the aP2 probe was a 525 bp PstI digest of the mouse aP2 cDNA,²⁴ and the 36B4 probe was a 750 bp PstI digest of the mouse 36B4 cDNA.²⁵

Differential display

Our protocol was derived from the method published by Liang and Pardee.²⁶ To eliminate chromosomal DNA contamination, 10 µg of total RNA extracted from cell fractions were first treated with 10 U of DNAse I (Boehringer Mannheim, Germany) for 15 min at 37°C in Tris 1 mM pH 7.5, MgCl₂ 0.25 mM, in the presence of 132 U of RNAse inhibitor (RNAGuard, Pharmacia-Biotech, Piscataway, NJ). Two hundred nanograms of these RNAs were incubated for 1 h at 42°C in a 20 µl reverse transcriptase reaction containing 50 mM Tris HCl pH 8.3, 75 mM KCl, 3 mM MgCl2, 10 mM DTT, 20 µM of each dNTP, 1.25 µM of the anchored primer (T11MN), 33 U RNAse inhibitor and 200 U RT SuperScript II RNAse H^- (Life Technology, Gaithersburg, MD). Negative controls of the reverse transcription reaction were performed in the same conditions but without reverse transcriptase. A 2 µl aliquot of these reactions was used for cDNA amplification in a 20 µl reaction mix containing 10 mM Tris pH 9, 50 mM KCl, 1.5 mM MgCl₂, 0.1% Triton, 0.2 µg/µl BSA, 2.5 µM of each dNTP, 2.5 µM of the anchored primer used for the reverse transcription, 2.5 µM of arbitrary decamer, 1 U of Taq DNA Polymerase (Appligene, Illkrich, France) and 10 µCi -³⁵S-dCTP. Following a 2 min denaturation at 96°C, the PCR steps consisted of 30 s at 94°C, 1 min 30 at 40°C, 1 min at 72°C, for 40 cycles, followed by 5 min at 72°C and a 4°C hold. Samples were dried under vacuum and 5 µl of non-denaturant loading dye (0.5 M EDTA, 10% glycerol and 0.09% xylene cyanol/bromophenol blue) were added to each sample. The PCR products were separated onto a non-denaturing 6% acrylamide/bisacrylamide gel in 1´TBE (Tris 90 mM, boric acid 90 mM and EDTA 50 mM pH 8). Samples were loaded after a 5 min denaturation step at 95°C and the gel was run at 70 W. It was then dried onto a Whatman paper and exposed 16 h on a Kodak BioMax MR film at 4°C.

Recovery, cloning and sequencing of the cDNA

The differential bands were excised from the differential display gel, and directly put in the same PCR reaction mix as described above, except for the concentrations of dNTPs (50 µM) and of the primers (0.2 µM). The PCR steps were exactly the same as for the differential display. The amplified fragments were purified using QIAquick PCR purification kit or QIAEX II Agarose Gel Extraction kit (both from QIAGEN). The purified cDNA fragments were cloned into the pGEM-T Easy vector (pGEM-T Vector System II, Promega, Madison, WI), according to the manufacturer's instructions. The inserts sequencing were carried out by the cycle-sequencing method using the ABI PRISM Ready Reaction Ampli Taq Dye Deoxy Terminator kit (Applied Biosystems, Foster City, CA) on an Applied 373 A sequencer. Homology searches were done in GeneBank EMBL DNA databases, using the BLAST search server at the National Center for Biotechnology Information.²⁷

Histological analysis

For in situ hybridization, the BUG cDNA-containing plasmid was linearized and used as template for antisense and sense digoxigenin (DIG)-labelled riboprobes synthesis according to manufacturer's instructions (DIG RNA Labelling kit, Boëhringer Mannheim, Germany). Rats were perfused in vivo with phosphate-buffered saline solution (PBS), 3.7% formaldehyde, tissues samples were taken, dehydrated and embedded in paraffin. Then 5 mm-thick sections were rehydrated through successive baths of ethanol and water. Cells from the brown SVF were attached to silanized slides by a cytospin and fixed in acetone at -20°C for 20 min. After 30 min in DEPC-treated PBS containing 0.1% active DEPC and an equilibration step of 15 min in SSC 5´, both types of slides were prehybridized for 2 h at 58°C in 50% formamide, SSC 5´, salmon sperm DNA 250 µg/ml. The hybridization was performed at 58°C for 16 h with 0.4 µg/ml DIG-labelled riboprobes in the same buffer (with salmon sperm DNA 40 g/ml). After incubation, slides were washed for 30 min in SSC 2´ (room temperature), 1 h in SSC 2´ at 65°C and 1 h in SSC 0.1´ at 65°C. DIG-labelled riboprobes were detected using the DIG wash and block buffer set from Boëhringer Mannheim, according to the manufacturer's instructions, with an anti-DIG-alkaline phosphatase conjugate (Boëhringer Mannheim, Germany) associated to an anti-sheep-alkaline phosphatase conjugate (Sigma). These antibodies were used at 1:250 and 1:200 dilutions, respectively. The same types of slides were stained with an anti-A2COL6 antibody (dilution 1:200), revealed by an anti-rabbit immunoglobulin coupled with alcaline phosphatase (dilution 1:25). Enzyme activity was visualized by means of the substrate 5-bromo-4-chloro-3-indolyl phosphate (BCIP) and the chromogen 4-nitro blue tetrazolium chloride (NBT), from Dako (Carpinteria, CA), before counterstaining with nuclear red.

Primary cultures

Preadipocytes were isolated from 10 five-week-old Wistar female rats (Iffa-Credo) and cultured as previously described.²⁸ After counting an aliquot stained with trypan blue, 100 mm culture dishes were inoculated with 450 000 cells. They were maintained at 37°C in a humidified 5% CO₂ atmosphere and rinsed twice with PBS, 4 h after inoculation. Twenty-four hours later, the cultured cells were placed in serum-free medium containing transferrin (1 µg/ml), T3 (3 nM) and insulin (21 nM for brown preadipocytes, 892 nM for white ones) (Sigma). Media were then changed every 2 days. After confluence, dexamethazone (Sigma) 100 nM (brown preadipocytes) or 33 nM (white preadipocytes) were added to this medium to induce the differentiation process. After washing with PBS and cell scrapping, total RNAs were extracted from plates by the acidic guanidium isothiocyanate method.²²

Statistical analysis

Northern blot autoradiograms were scanned with a SI densitometer (Molecular Dynamics, Sunnyvale, CA) and signals were quantified using ImageQuant software (Molecular Dynamics). All statistical analyses were done by unpaired t-test, with Prism software (GraphPad software, San Diego, CA).

Results

Cloning of BUG RNA, a newly identified transcript preferentially expressed in the stromal-vascular fraction of rat brown adipose tissue

Our differential display study concerned the RNA populations expressed in mature adipocytes and SVF isolated from the interscapular brown and the inguinal white fat pads. Only nine primer pairs were tested for this analysis. Each differential display lane yielded around 50 discrete bands ranging from 100 to 600 bp, allowing about 600 different RNA species to be screened. Most of bands were similar in size between the different fractions compared. A small number (about 2%) appeared in only one of the four lanes, and were absent in the negative control of the reverse transcription. Figure 1A shows the example of a few bands, indicated by small arrows, which were present in the white adipocytes fraction but also appeared in the corresponding control lane, thus being false differential bands issued from contaminating genomic DNA. Display of RNA was duplicated using samples independently obtained from different animals and only the reproducible bands were selected for further analysis.

According to these criteria, a total of 10 bands were differentially detected in one fraction or another, but testing the corresponding fragments of cDNA on Northern blots with total RNAs revealed that five of them could not detect any transcript and four were false positives, detecting non-differential transcripts. Only one cDNA fragment of 470 bp was differentially found in the brown SVF (large arrow on Figure 1A). Using this cDNA fragment as a probe in Northern analysis, we detected a major RNA species at about 1.4 kb. Two other bands much less intense could also be detected, the most often seen sizing 0.6 kb. As these bands varied in the same way, we only quantified the major band in the following Northern blottings. The levels observed confirmed that this major RNA species was abundant in cell fractions isolated from brown fat but faintly detected in white ones (Figure 1B). The maximal content was clearly localized in the SVF from BAT compared to mature brown adipocytes and was also higher in white SVF than in white adipocytes, although this last difference was not significant. When comparing both SVFs, the level in the brown one was about 7-fold higher than in the white one (Figure 1C). The absence of contaminating mature adipocytes in both SVF was confirmed by the absence of aP2 and UCP-1 mRNA, two late markers of adipocyte differentiation (Figure 1B).

This pattern indicated that this RNA species was preferentially expressed in brown vs WAT and more abundant in the SVF of this tissue, ie the fraction containing adipocytes precursors. The sequence of the corresponding cDNA fragment was determined and homologous sequences were searched in GenBankÔ database using the BLAST tool. This analysis did not reveal any significant homology to previously reported sequences. The gene corresponding to the transcript revealed by this cDNA fragment was thus termed BUG (brown unknown gene).*

The higher level of BUG transcript in BAT, compared to WAT, was confirmed in whole tissues. A representative Northern blot is shown in Figure 1D. The BUG transcript could be detected in different adult rat tissues. The highest expression was observed in BAT and heart. Nevertheless the level of expression in heart is significantly different from the one in BAT (79% of BAT level, P<0.05, data not shown). The lower expression was observed in WAT, liver, spleen and muscle. No signal could be detected in mouse or rabbit tissues (data not shown). No difference according to sex in BUG expression was observed (data not shown).

Genome walking was performed to sequence DNA fragments around the primarily cloned cDNA. A 213 bp fragment, next to our first sequence in the 3^' direction, was able to detect the BUG RNA on Northern blots. Its sequence was thus added to the one obtained by differential display and it revealed the existence of a poly-A tract inside the BUG cDNA sequence (Figure 2). None of the available programs could detect any consensus exon/intron limit within these 213 bp, and then the following DNA in 3^' is negative on Northern blots. Numerous 5^'- or 3^'-RACE-PCRs attempts were carried out, ending in nothing but artefactual cDNA fragments that failed to detect the BUG RNA when used as probes on Northern blots. We thus pursued genome walking: each new DNA fragment obtained was cloned, sequenced, submitted to databases and used as a probe on Northern blots. Within 14 kb obtained by this means, no DNA fragment could reveal BUG RNA. Having more than 470 bp, ie more than one-third of the entire BUG RNA sequence, we searched for ORFs. Figure 2 shows that there are numerous dispersed stop codons in all reading frames, resulting in a maximal length ORF of 242 bp.

Determination of the BUG expressing cell type in interscapular BAT

Interscapular BAT sections and brown SVF isolated cells were both used for in situ hybridization with sense and antisense riboprobes corresponding to BUG RNA and for immunostaining with an anti-A2COL6 antibody. Representative views are shown in Figure 2. Mature brown adipocytes, with their numerous lipid droplets, are clearly recognizable on these sections (Figure 3A-D). Cell nuclei are pink-coloured by the counterstaining. In situ hybridization experiments gave a positive purple signal in very small cells dispersed within the tissue with the antisense riboprobe (Figure 3A and C), but not with the sense one (Figure 3B). The positive cells on these sections, designated by yellow arrows, were round in shape and had the same size as nuclei of brown adipocytes. To avoid morphological deformations due to dehydration of the samples, separated cells from the brown SVF were attached to slides by a centrifugation process and subjected to the same hybridization experiment. Figure 3E clearly shows the localization of the staining in the cytoplasm of positive cells (yellow arrow), while negative cells keep a clear cytoplasm (red arrow). The stained cells have a high nucleo-cytoplasmic ratio. Immunostaining with the anti-A2COL6 antibody gave exactly the same kind of image, on interscapular BAT sections as well as on brown SVF cells (Figure 3D and F).

Changes in BUG transcript level during brown adipocyte in vitro differentiation

Precursor cells contained in the SVF isolated from interscapular BAT were differentiated in primary culture. They reached confluence (ie growth arrest, indicated as 'C') around 5 days after the beginning of the culture and then started to differentiate and to accumulate lipids. Seven days after confluence, they showed the same multilocular appearance as mature brown adipocytes in vivo. The differentiation process was confirmed by aP2 and UCP-1 mRNA emergence, reaching a maximum at the end of the culture (Figure 4A, upper panel). A 2.5-fold increase was observed in the BUG transcript content from confluence to 3 days after growth arrest (Figure 4A, upper panel). Then the level significantly decreased in differentiated cells (P<0.01), to values similar to those observed around confluence. A similar pattern was observed for A2COL6 mRNA (Figure 4A, upper panel). In white preadipocyte primary culture, no change in the very faint BUG expression was observed (Figure 4A, lower panel). As previously suspected by Northern blotting with mouse tissues, BUG RNA could not be detected in mouse HIB1B transformed brown preadipocyte cell line (data not shown).

Changes in BUG transcript level in BAT in physiological and pathophysiological situations

In rats, the interscapular depot development takes place during late foetal life, to be fully differentiated in the days following birth.^29,30 As described in the literature, we found that UCP-1 mRNA was induced at day 20 of the gestation and that its level suddenly increased at birth (Figure 4B). The BUG transcript appeared much earlier as it was already detectable at a low level in interscapular BAT of 18-day-old foetuses (Figure 4B). It first increased at day 20 of gestation (2-fold) and then 2 days after birth (2-fold). It is noteworthy that no variation in BUG level was observed in the liver (data not shown). One week after birth, both UCP-1 and BUG transcript levels were slightly diminished.

Cold exposure is known to stimulate thermogenesis and hyperplasia in interscapular BAT.³¹ Rats were exposed to a temperature of 4°C for 5 days. A 4-fold induction of UCP-1 mRNA occurred in the interscapular BAT after only 10 h of cold exposure, then the level fell and remained around 2-fold higher than in the control (Figure 5A, upper panel). Concerning BUG transcript level, a marked decrease from 10 h of cold exposure was observed, reaching a 1.6-fold reduction at day 5 (Figure 5A, lower panel). A significant effect of cold exposure on BUG expression was observed neither in the inguinal fat pad nor in the heart of these animals (data not shown).

In Figure 5B, the 2-fold increase in UCP-1 transcript level in interscapular BAT induced by 5 days of cold exposure was reproduced by daily injections of CL316243 or of isoproterenol, a 3- and non-specific adrenoreceptor agonist, respectively. In contrast, these treatments had no effect on BUG transcript level in interscapular BAT.

Obesity is associated with effects on BAT morphology, physiology and UCP-1 content opposite to those of cold exposure.³² Figure 5C shows a 3-fold reduction in UCP-1 mRNA level in interscapular BAT of obese Zücker rats (fa/fa) compared to lean ones (Fa/?). On the contrary, BUG transcript level was significantly higher in obese than in lean rats (1.6±0.2-fold increase, P<0.05).

Discussion

The differential display allows the direct comparison of RNA populations expressed in several samples of cells or tissues.²⁶ Tissue fractions separated from adipose tissues represented a challenge but were the best way to find molecular markers able to identify putative brown or white adipocytes precursors in vivo. The in vivo approach is relevant to investigate the existence of two distinct preadipocytes and their respective involvement in adipose tissue development. The choice of the most homogeneous white and brown adipose depots was important to work on adipocytes and precursors clearly belonging to one or the other adipocyte lineage. We had previously demonstrated that, in rats, the interscapular and the inguinal fat pads are two subcutaneous depots typically brown and white, respectively.³³ These two adipose tissues were thus chosen to apply the differential display technique on their respective SVF and mature adipocyte fractions.

One transcript (BUG) was confirmed to be differentially and highly expressed in the brown SVF by Northern blot experiment. It was also detected in mature brown adipocyte fraction, where this BUG residual presence could be due to a contamination by stromal-vascular cells during the separation of the fractions.

When cells of the brown SVF were used in primary culture and induced to differentiate into brown adipocytes, the BUG transcript was transiently induced in a differentiation-dependent manner during the first steps of the differentiation program leading to the brown adipocyte phenotype. This pattern of expression is similar to the one of A2COL6, considered as a marker of the early steps of brown and white adipocyte differentiation pathways.^8,34 Moreover, histology experiments demonstrated that both transcripts are expressed in vivo in cells that have the morphology of immature cells. The pattern of expression of BUG during brown adipocyte differentiation process is associated with the brown fat lineage specificity. Indeed, no similar change can be observed in primary culture of white SVF cells. BUG could therefore be considered as an early and transiently expressed marker of the brown differentiation pathway. Other genes are known to be either specifically expressed in preadipocytes vs mature adipocytes in both lineages (pref-1, FRP-2) or specific to brown fat cells in their late differentiation stages (PGC-1, UCP-1).^1,9,10,11 To our knowledge, BUG is the first gene identified that combines early expression with the brown lineage specificity, thus definitively establishing in vivo the existence of a brown preadipocyte different from a white one. The divergence between both lineages takes place before BUG expression.

During development, the interscapular BAT depot can be macroscopically visualized at 18 days of gestation in rodents.²⁹ UCP-1 mRNA appears at 19 days of gestation and suddenly increases at birth.³⁰ In comparison, BUG transcript was already present in foetuses before the detection of UCP-1, as in primary culture. This suggests that the divergence between white and brown preadipocytes takes place at the very beginning of BAT development. The increase of BUG RNA level during the perinatal period could correspond to the successive recruitment of brown precursors before birth and during the days following it, when the BAT reaches its full development.^35,36

In adults, it is well established that cold exposure and obesity represent opposite situations leading to activation or inhibition of BAT, respectively, as determined by changes in UCP-1 content.^32,37,38 Interestingly, BUG transcript level in BAT was inversely regulated in these two opposite situations, with a 60% decrease at 5 days of cold exposure, and a 60% increase in obese compared to lean Zücker rats. This regulation is specific to BAT. Indeed, change was observed neither in WAT nor in heart, in spite of the strong level of expression observed in this last tissue. The novelty of BUG resides in this regulation, compared to other known genes implicated in BAT activation (UCP-1, PGC-1), that are classically up-regulated during cold adaptation and down-regulated in obesity.^11,32 Precursor cells in BAT are known to undergo proliferation during the first week of cold exposure and give rise to new mature brown adipocytes between 4 and 16 days after the beginning of cold stimulation.^31,39,40 In similar conditions, the decrease in BUG transcript level implies that this RNA is not located in precursor cells induced to proliferate under this stimulus. Moreover, the observed variation is progressive compared to the UCP-1 one. This is not in favour of a rapid regulation of BUG gene by BAT inducers and agrees with the absence of effect of -adrenergic agonists. The decrease of BUG transcript level could be explained by the differentiation of latent brown preadipocytes strongly expressing it into mature adipocytes with reduced BUG expression. This implies that, in adults, a population of precursor cells in the interscapular BAT are blocked in a state of differentiation different from the white one and just preceding the terminal acquisition of brown adipocyte phenotype. Such cells could be massively recruited during the first days of cold exposure. These results reinforce the hypothesis of a recruitment of brown precursors in successive waves, previously described during rat cold exposure.³⁴ The higher expression of BUG transcript in interscapular BAT of obese compared to lean rats reveals a greater number of brown preadipocytes blocked in the BUG-expressing state. In this perspective, the deficiency of BAT in obese Zücker rats could be due to an arrest of brown precursor cells at this precise point of their evolution, before the initiation of the late differentiation program but after the commitment to the brown fat lineage. Thus, BUG function would allow an interpretation of the BAT inactivation in a genetic model of obesity.

Despite our great efforts, we failed to clone the entire cDNA and to definitively conclude about its putative function. Nevertheless, several points can be discussed. This transcript is not specific to brown fat and is highly expressed in heart and kidney. Similar profiles can be obtained with mitochondria transcript and could suggest a particular relationship between BUG and mitochondria. Nevertheless, the decline of BUG expression after cold exposure, during which mitochondriogenesis is strongly stimulated, does not agree with this.¹⁷ No opened reading frame could be detected in more than half of cDNA length. This suggests that the sequence we obtained corresponds to a non-coding sequence. This last hypothesis could explain the lack of cross-hybridization with BUG probe in mouse.

In conclusion, BUG is a new marker which features strongly suggest an expression in brown preadipocytes. The cloning of BUG demonstrates at the molecular level, in vivo and in vitro, the independence of brown vs white adipocyte lineage as soon as the preadipocyte stage. This study also reveals the existence of a pool of committed brown preadipocytes, which recruitment is impaired in genetically obese animals. Despite the absence of BUG complete sequence, which is currently under investigation, its discovery is giving insight into the differentiation status of brown adipocyte precursors within BAT.

Acknowledgements

We wish to thank N Truel and M Desbazeille for their competent help in Northern experiments. We also thank A Belliure for secretarial assistance, J Hillat for technical help and J-M Lherme from the zootechnical department. This work was supported in part by a grant from CNRS and the Région Midi-Pyrénées (France). Karine Moulin is a fellow of the French Ministère de l'Education Nationale, de la Recherche et de l'Industrie.

*The nucelotide sequence for the brown unknown gene (BUG) partiel cDNA has been deposited in the GenBank accession no. AF 247002.

1 Nicholls DG, Locke RM. Thermogenic mechanisms in brown fat. Physiol Rev 1984; 64: 1-64, MEDLINE

2 Napolitano L. The differentiation of white adipose cells: an electron microscope study. J Cell Biol 1963; 18: 663-679,

3 Van RL, Bayliss CE, Roncari DA. Cytological and enzymological characterization of adult human adipocyte precursors in culture. J Clin Invest 1976; 58: 699-704,

4 Van RL, Roncari DA. Isolation of fat cell precursors from adult rat adipose tissue. Cell Tissue Res 1977; 181: 197-203,

5 Djian P, Roncari DAK, Hollenberg CH. Influence of anatomic site and age on the replication and differentiation of rat adipocyte precursors in culture. J Clin Invest 1983; 72: 1200-1208,

6 Ailhaud G, Grimaldi P, Négrel R. Cellular and molecular aspects of adipose tissue development. A Rev Nutr 1992; 12: 207-233,

7 Klaus S. Functional differentiation of white and brown adipocytes. BioEssays 1996; 19: 1-9,

8 Ibrahimi A, Bertrand B, Bardon S, Amri E-Z, Grimaldi P, Ailhaud G, Dani C. Cloning of alpha-2 chain of type VI collagen and expression during mouse development. Biochem J 1993; 289: 141-147,

9 Smas CM, Sook Sul H. Pref-1, a protein containing EGF-like repeats, inhibits adipocyte differentiation. Cell 1993; 73: 725-734, MEDLINE

10 Hu E, Zhu Y, Fredrickson T, Barnes M, Kelsell D, Beeley L, Brooks D. Tissue restricted expression of two human Frzbs in preadipocytes and pancreas. Biochem Biophys Res Commun 1998; 247: 287-293,

11 Puigserver P, Wu Z, Won Park C, Graves R, Wright M, Spiegelman BM. A cold-inducible coactivator of nuclear receptors linked to adaptative thermogenesis. Cell 1998; 92: 829-839, MEDLINE

12 Wu Z, Puigserver P, Andersson U, Zhang C, Adelmant G, Mootha V, Troy A, Cinti S, Lowell B, Scarpulla RC, Spiegelman BM. Mechanisms controlling mitochondrial biogenesis and respiration through the thermogenic coactivator PGC-1. Cell 1999; 98: 115-124, MEDLINE

13 Ashwell M. Is there a continuous spectrum of the adipose tissues in animals and man? In: Vague J et al (eds). Metabolic complications of human obesities 1985; pp 265-274,

14 Cousin B, Casteilla L, Lafontan M, Ambid L, Langin D, Berthault M-F, Pénicaud L. Local sympathetic denervation of white adipose tissue in rats induces preadipocytes proliferation without noticeable changes in metabolism. Endocrinology 1993; 133: 2255-2262,

15 Géloën A, Collet AJ, Guay G, Bukowiecki LJ. Beta-adrenergic stimulation of brown adipocyte proliferation. Am J Physiol 1988; 254: C175-C182,

16 Rehnmark S, Nechad M, Herron D, Cannon B, Nedergaard J. Alpha- and beta-adrenergic induction of the expression of the uncoupling protein thermogenin in brown adipocytes differentiated in culture. J Biol Chem 1990; 265: 16464-16471,

17 Casteilla L, Nougues J, Reyne Y, Ricquier D. Differentiation of ovine brown adipocyte precursor cells in a chemically defined serum-free medium. Importance of glucocorticoids and age of animals. Eur J Biochem 1991; 198: 195-199,

18 Moldes M, Fève B, Pairault J. Molecular cloning of a major mRNA species in murine 3T3 adipocyte lineage. J Biol Chem 1999; 274: 9515-9523,

19 Hu E, Liang P, Spiegelman BM. AdipoQ is a novel adipose-specific gene dysregulated in obesity. J Biol Chem 1996; 271: 10697-10703, Article MEDLINE

20 Rodbell M. Metabolism of isolated fat cell. Effects of hormones on glucose metabolism and lipolysis. J Biol Chem 1964; 239: 375-380,

21 Parker J, Lane J, Axelrod L. Cooperation of adipocytes and endothelial cells required for catecholamine stimulation of PGI2 production by rat adipose tissue. Diabetes 1989; 38: 1123-1132,

22 Chomezynski P, Sacchi N. Single-step method of RNA isolation by acidic guanidium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987; 162: 156-159, Article MEDLINE

23 Bouillaud F, Weissenbach J, Ricquier D. Complete cDNA-derived amino acid sequence of rat brown fat uncoupling protein. J Biol Chem 1986; 261: 1487-1490, MEDLINE

24 Hunt CR, Ro JH, Dobson DE, Min HY, Spiegelman BM. Adipocyte P2 gene: developmental expression and omology of 5^'-flanking sequences among fat cell-specific genes. Proc Natl Acad Sci USA 1986; 83: 3786-3790, MEDLINE

25 Krowczynska AM, Coutts M, Makrides S, Brawerman G. The mouse homologue for human acidic ribosomal phosphoprotein PO: a highly conserved polypeptide that is under translational control. Nucleic Acids Res 1989; 17: 6408,

26 Liang P, Pardee AB. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 1992; 257: 967-971, MEDLINE

27 Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol 1990; 15: 403-410,

28 Klaus S, Cassard-Doulcier AM, Ricquier D. Development of Phodopus sungorus brown preadipocytes in primary cell culture: effect of an atypical beta-adrenergic agonist, insulin, and triiodothyronine on differentiation, mitochondrial development, and expression of the uncoupling protein UCP. J Cell Biol 1991; 115: 1783-1790,

29 Rowlatt U, Mrosovsky N, English A. A comparative survey of brown fat in the neck and axilla of mammals at birth. Biol Neonate 1971; 17: 53-83,

30 Giralt M, Martin I, Iglesias R, Vinas O, Villarroya F, Mampel T. Ontogeny and perinatal modulation of gene expression in rat brown adipose tissue. Unaltered iodothyronine 5^'-deiodinase activity is necessary for the response to environmental temperature at birth. Eur J Biochem 1990; 193: 297-302,

31 Bukowiecki L, Collet AJ, Follea N, Guay G, Jahjah L. Brown adipose tissue hyperplasia: a fundamental mechanism of adaptation to cold and hyperphagia. Am J Physiol 1982; 242: E353-E359,

32 Himms-Hagen J. Brown adipose tissue thermogenesis: interdisciplinary studies. FASEB 1990; 4: 2890-2898,

33 Cousin B, Cinti S, Morroni M, Raimbault S, Ricquier D, Pénicaud L, Casteilla L. Occurrence of brown adipocytes in rat white adipose tissue: molecular and morphological characterization. J Cell Sci 1992; 103: 931-942, MEDLINE

34 Cousin B, Bascands-Viguerie N, Kassis N, Nibbelink M, Ambid L, Casteilla L, Pénicaud L. Cellular changes during cold acclimatation in adipose tissues. J Cell Physiol 1996; 167: 285-289, Article MEDLINE

35 Skala J, Barnard T, Lindberg O. Changes in interscapular brown adipose tissue of the rat during perinatal and early post-natal development and after cold acclimatation-II. mitochondrial changes. Comp Biochem Physiol 1970; 33: 509-528,

36 Sundin U, Herron D, Cannon B. Brown fat thermoregulation in developing hamsters (Mesocricetus auratus): a GDP-binding study. Biol Neonate 1981; 39: 141-149,

37 Trayhurn P. Brown adipose tissue and energy balance. In: Trayhurn P, Nicholls DG (eds). Brown adipose tissue. Edward Arnold: London, 1986, 1-30.

38 Marette A, Géloën A, Collet A, Bukowiecki LJ. Defective metabolic effects of norepinephrine and insulin in obese Zucker rat brown adipose tissue. Am J Physiol 1990; 258: E320-E328,

39 Cameron IL, Smith RE. Cytological responses of brown fat tissue in cold-exposed rats. J Cell Biol 1964; 23: 89-100,

40 Bukowiecki LJ, Géloën A, Collet AJ. Proliferation and differentiation of brown adipocytes from intersticial cells during cold acclimatation. Am J Physiol 1986; 250: C880-C887,

Figure 1 Identification of BUG by differential display of stromal-vascular and adipocytes fractions (SVF and AF, respectively) of brown (BAT) and white (WAT) fat pads. (A) Autoradiogram of a display gel. Differential display was performed as described in experimental procedures using RNA from (1) brown SVF, (2) brown AF, (3) white SVF and (4) white AF. Lanes (1^-), (2^-) and (3^-) correspond to negative controls of reverse transcription on RNA from brown SVF, white SVF and white adipocytes, respectively. A large arrow indicates the differential band corresponding to BUG. False differential signals due to DNA contamination are indicated by small arrows. (B) Representative Northern blot from one experiment. The molecular size marker is indicated in kb. Total RNAs (20 µg per lane) were blotted and successively hybridized with probes corresponding to BUG, aP2 and UCP-1 cDNA. Methylene blue staining of 18S ribosomal RNA is shown. (C) Quantification of BUG transcript levels resulting from 6 independent experiments. The amount of BUG transcript in brown SVF was taken as 100%. (*P<0.05; ***P<0.001). (D) BUG transcript expression pattern in different tissues. Total RNAs (20 µg per lane), isolated from brown interscapular (B) and white inguinal (W) fat pads, muscle (M), heart (H), liver (L), spleen (S), brain (Br) and kidney (K), were blotted and hybridized with probe corresponding to BUG transcript. The upper panel corresponds to the autoradiogram of this Northern blot, the lower one shows the 18S rRNA content stained with methylene blue.

Figure 2 Nucleotide sequence of the rat extended BUG cDNA fragment. The primarily obtained cDNA sequence appears in bold characters, while the complementary sequence obtained from genomic sequencing is in normal letters. (a), (b) and (c) indicate the three possible reading frames. In each of them, putative ATG start codons (M) and stop codons (*) are indicated. All putative ORFs are shown, to the end of the extended sequence.

Figure 3 Localization of BUG transcript in interscapular brown adipose tissue and its stromal-vascular cells. Interscapular BAT sections were hybridized in situ with antisense (A) or sense (B) BUG riboprobes. A higher magnification view of the antisense hybridized section is shown in (C). (D) is an immunostaining of interscapular BAT section with an anti-A2COL6 antibody. Nuclei are counterstained with nuclear red. Yellow arrowheads designate stained cells in both cases. Isolated brown SVF cells were concentrated on slides by cytospin method and submitted to same in situ hybridization and immunostaining (E, F). On these views, yellow arrowheads show positively stained cells while red ones indicate non stained cells. Scale bars represent 30 µm in (A) and (B), 12 µm in (C) and (D) and 6 µm in (E) and (F).

Figure 4 Specific evolution of BUG transcript during brown preadipocytes differentiation in vitro and in vivo. (A) Changes of BUG transcript level in primary cultures of brown and white SVF of rats. Cells of the stromal-vascular fractions separated from interscapular brown and inguinal white fat pads of 10 rats were seeded and proliferated during 4-5 days, till confluence (day 'C'), followed by differentiation. Culture dishes were scraped at different times for total RNA extraction. Northern blots were performed and successively hybridized with probes corresponding to BUG, A2COL6, aP2, UCP-1 and 36B4 transcripts. Results of quantification are expressed as percentage of BUG or other transcripts levels, normalized to 36B4, and relative to final values observed for each culture. Results are the mean of at least three independent cultures. **Significantly different from the BUG (upper panel) or the A2COL6 (lower panel) maximal value at P<0.01. (B) Developmental pattern of UCP-1 and BUG transcripts levels in interscapular BAT during the perinatal period. On days 18, 19 and 20 of gestation, foetuses were delivered by rapid hysterectomy. Interscapular brown adipose tissues were removed from foetuses, newborns or 2-, 4- or 7-day-old suckling pups and frozen at -80°C for total RNA extraction. Each lane on the Northern blot shown corresponds to 20 g of total RNA obtained from all the pups of one litter, except for 18- and 19-day-old foetuses where 3 and 2 litters were pooled. Methylene blue staining of 18S ribosomal RNA is shown.

Figure 5 (A) Effects of cold exposure on BUG and UCP-1 transcripts levels. Rats were housed at 4°C, 2 per cage, for 5 days. After sacrifice at different times, brown interscapular fat pad was removed and frozen at -80°C for total RNA extraction. Northern blots were performed using 20 g of total RNA per lane and successively hybridized with probes corresponding to UCP-1 (upper panel) and BUG (lower panel) transcripts. RNA levels were quantified and expressed relative to RNA content in control animals (day 0, ie before cold exposure). Results are expressed as the mean obtained on at least 4 animals per point. (B) Effects of adrenergic stimulation on BUG and UCP-1 transcripts levels. 18 rats (6 per group) were injected daily intraperitonally for 5 days with NaCl (CTL group), CL316243 1 mg/kg/day (CL group) or isoproterenol 0.5 mg/kg/day (ISO group) and killed 24 h after the last injection. BAT was removed and directly frozen at -80°C for total RNA extraction. Northern blot was performed using 20 g of total RNA per lane and successively hybridized with probes corresponding to BUG and UCP-1 transcripts. RNA levels were quantified and expressed relative to RNA content in control group (100%). (C) Comparison of BUG transcript levels in interscapular fat pads of lean and obese Zücker rats. Total RNAs were isolated from interscapular brown adipose tissue from five lean (Fa/?) and 11 obese (fa/fa) 9-week-old female Zücker rats. Northern blots were performed using 20 µg of total RNA per lane and successively hybridized with probes corresponding to BUG and UCP-1 transcripts. RNA levels were quantified and are expressed relative to RNA content in lean group (100%). *Significantly different from the control value at P<0.05; **P<0.01; ***P<0.001.