Occurrence of brown adipocytes in rat white adipose tissue ... · fat pads of rat analyzed by Northern blotting experiments. A 20 µ g sample of total RNA extracted from different

Introduction

Two functionally different types of adipocytes are knownin mammals: white adipocytes, which store energy astriglycerides and release it according to organism needs,and brown adipocytes, which dissipate energy as heat. Untilrecently, brown and white adipocytes were considered tobe mainly located in distinct depots, i.e. brown (BAT) andwhite (WAT) adipose tissue, respectively (Himms-Hagen,1990; Klaus et al. 1991b). These pads are recognized bytheir different morphological structure and distribution(Napolitano, 1963; Afzelius, 1970; Néchad, 1986; Cinti,1992). BAT plays an important role in the regulation ofbody temperature in hibernating as well as in small andnewborn mammals. It has also been shown to play a rolein diet-induced thermogenesis in small rodents, i.e. therelease of excess food energy as heat (Rothwell and Stock,

1985; Trayhurn, 1986; Himms-Hagen, 1990; Klaus et al.1991b).

In BAT, all differentiated brown adipocytes express anuncoupling protein (UCP). This mitochondrial carrier,unique to these cells, uncouples mitochondrial ATP syn-thesis from the respiratory chain activity and is responsiblefor heat production by brown adipocytes (Nicholls andLocke, 1984; Klaus et al. 1991b). Based on these consid-erations, the presence of UCP is the unique feature that dis-tinguishes brown from white adipocytes. The expression ofUCP is strongly regulated at the transcriptional level by cat-echolamines (Ricquier et al. 1986; Rehnmark et al. 1990).Moreover, triiodothyronine and insulin also participate inthis regulation (Silva, 1988; Geloën and Trayhurn, 1990;Klaus et al. 1991a).

In adult rodents, the distribution of fat depots betweenBAT and WAT has been well documented but most of

931Journal of Cell Science 103, 931-942 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

Brown adipocytes are thermogenic cells which play animportant role in energy balance. Their thermogenicactivity is due to the presence of a mitochondrial uncou-pling protein (UCP). Until recently, it was admitted thatin rodents brown adipocytes were mainly located in clas-sical brown adipose tissue (BAT). In the present study,we have investigated the presence of UCP protein ormRNA in white adipose tissue (WAT) of rats. Usingpolymerase chain reaction or Northern blot hybridiz-ation, UCP mRNA was detected in mesenteric, epidydi-mal, retroperitoneal, inguinal and particularly in peri-ovarian adipose depots. The uncoupling protein wasdetected by Western blotting in mitochondria from peri-ovarian adipose tissue. When rats were submitted tocold or to treatment with a -adrenoceptor agonist, UCPexpression was increased in this tissue as in typicalbrown fat. Moreover, the expression was decreased inobese fa/fa rats compared to lean controls. Morpholog-

ical studies showed that periovarian adipose tissue ofrats kept at 24°C contained cells with numerous typicalBAT mitochondria with or without multilocular lipiddroplets. Immunocytochemistry confirmed that multi-locular cells expressed mitochondrial UCP. Further-more, the number of brown adipocytes and the densityof mitochondrial cristae increased in parallel with expo-sure to cold. These results demonstrate that adipocytesexpressing UCP are present in adipose deposits consid-ered as white fat. They suggest the existence of a con-tinuum in rodents between BAT and WAT, and a greatplasticity between adipose tissue phenotypes. The phys-iological importance of brown adipocytes in WAT andthe regulation of UCP expression remain open ques-tions.

Key words: adipocytes, thermogenic cells, mitochondria.

Summary

Occurrence of brown adipocytes in rat white adipose tissue: molecular

and morphological characterization

B. COUSIN1,2, S. CINTI3, M. MORRONI3, S. RAIMBAULT2, D. RICQUIER2, L. PÉNICAUD1

and L. CASTEILLA2,*

1Laboratory of Physiopathology of Nutrition, CNRS URA 307, University of Paris VII, Paris, France2CEREMOD, CNRS, 9 rue Hetzel, 92190 Meudon, France3Institute of Normal Human Morphology, School of Medicine, University of Ancona, Ancona, Italy

*Author for correspondence

932

these results were based on histological data (Afzelius,1970; Néchad, 1986). Unfortunately, these criteria aresomewhat insufficient; in only a few investigations anti-bodies against UCP or UCP cDNA were used. Whetherbrown adipocytes are present among white fat and whatare the relationships between both adipose cellular typesremain open questions. These questions are major pointsof interest in understanding the development and regula-tion of body adipose mass.

Thus, we have investigated the presence of UCP mRNAin several fat deposits considered as white adipose tissue inrat using Northern blotting or PCR experiments combinedwith histological and functional studies characterizing cellsexpressing UCP.

Materials and methods

AnimalsMale and female Wistar rats (8 to 11 week-old), lean (Fa/Fa) andobese (fa/fa) female Zucker rats, were housed in animal quartersin which the temperature was maintained at 24°C with food andwater ad libitum, and 12 h light per day. Some rats were cagedindividually and kept at a constant cold temperature (4°C) for 24hours, 3 or 10 days. Another group was treated with a β3-adreno-ceptor agonist (BRL 26830A, 10 mg/kg, i.p.), and animals werekilled 3 hours after the injection. Animals were killed by cervicaldislocation. Whole separate fat pads were rapidly removed andimmediately frozen in liquid nitrogen. For histological analysis,rats were anaesthetized with chloral hydrate (400 mg/kg) and

transaortically perfused with a 2% paraformaldehyde-saline solu-tion for 2 min.

Northern blot analysisAdipose tissue samples from animals were powdered in liquidnitrogen and a sample (0.3 to 1 g) from each rat was used toextract total RNA using the guanidine thiocyanate technique(Chomczynski and Sacchi, 1987). RNA concentration was deter-mined by absorbance at 260 nm, and RNAs were stored in water+ diethyl pyrocarbonate (0.02%) at −80°C until used. In order todetect the presence of UCP mRNA in different pads consideredas white adipose tissue, Northern blotting experiments were per-formed according to the method of Sambrook et al. (1989) with20 µg of total RNA extracted from these pads (Fig. 1A). Two µgof total RNA from interscapular BAT (IBAT, which will be con-sidered as typical brown fat) was used as control in all experi-ments. After blotting, nylon membranes were colored with meth-ylene blue to detect DNA contamination and verify the quality ofloading and transfer (Sambrook et al. 1989). Blots were succes-sively hybridized with several probes, labelled with 32P using aMegaprime labelling system kit (Amersham, Bucks, UK).Hybridizations were made with a pUCP 36 insert, containing thewhole cDNA for rat UCP mRNA (Bouillaud et al. 1985), with apBR 325 ST41 plasmid containing the total mouse mitochondrialgenome (Bibb et al. 1981) and other probes used as controls. Afterstringent washings, the blots were then exposed for 4 to 48 hoursat −80°C with intensifiying screens. Quantification was performedby scanning densitometry.

Mitochondrial preparation and Western blot analysisMitochondria were isolated from white and brown adipose tissues

B. Cousin and others

Fig. 1. Atypical expression of UCP mRNA in whitefat pads of rat analyzed by Northern blottingexperiments. A 20 µg sample of total RNA extractedfrom different adipose deposits or 2 µg of BAT totalRNA were analyzed by Northern blotting. The nylonmembranes were hybridized with 32P-labelled UCPcDNA or a 32P-labelled plasmid containing themitochondrial DNA. (A) Rats were maintained instandard conditions at 24°C. The fat pads wereinterscapular brown adipose tissue (BAT),periovarian (PO), retroperitoneal (RP), mesenteric(M) and inguinal deposits (I) of three rats. (B) Ratswere cold exposed for 24 hours to 4°C before killingand the same adipose deposits were analyzed.

933Atypical localization of brown adipocytes

as described by Klaus et al. (1991a). Proteins were quantifiedusing Bradford’s (1976) technique.

Immunodetection of UCP in homogenates and mitochondriawas performed as previously described (Klaus et al. 1991a).

Reverse transcription and polymerase chain reaction Both cDNA synthesis and polymerase chain reaction (PCR) wereperformed in the same buffer (Tris-HCl, 20 mM, pH 8.4; KCl, 50mM; MgCl2, 2.5 mM; 0.1 mg/ml nuclease-free BSA) as describedpreviously (Ricquier et al. 1992). Sense primer OL3 matched withUCP cDNA (EMBL data bank: RNUCPG) from position 589 toposition 608 (GTGAAGGTCAGAATGCAAGC), antisenseprimer OL4 matched from position 785 to position 766 (AGGGC-CCCCTTCATGAGGTC).

Light microscopyPeriovarian white adipose tissue was removed and immediatelyfixed overnight with 3% paraformaldehyde in 0.1 M phosphatebuffer, pH 7.4, dehydrated in ethanol and paraffin embedded. Eachpad was embedded at the same orientation and 30 sections (3 µmthick) were serially sectioned every 200 µm and stained withhaematoxylin and eosin.

Electron microscopyPeriovarian adipose tissue was isolated and sectioned in fragmentsof about 1 mm3. These pieces were immersed in a fixative con-sisting of 2% glutaraldehyde and 2% formaldehyde in 0.1 M phos-phate buffer, pH 7.4, at 4°C for 3 hours. Specimens were thenpostfixed in 1% OsO4, dehydrated in ethanol and embedded in anEpon-Araldite mixture. Semithin sections (2 µm) were stainedwith toluidine blue; thin sections were obtained with a ReichertUltracut E, stained with lead citrate and examined with a PhilipsCM10 or Zeiss 902 electron microscope.

MorphometryNumber of multilocular cells per section andmultilocular cell density

Each section of the periovarian pad was projected on a screen bya projecting light microscope Leitz Tele-Promar 500 at a finalmagnification of ×150. At this magnification it was easy to dis-tinguish between unilocular (i.e. showing absolute prevalence ofone, large lipid droplet and a thin rim of cytoplasm) and multi-

locular (i.e. showing several small lipid droplets in a granulouseosinophilic cytoplam) cells. The latter were counted directly. Thetotal area of the tissue in each section was measured using a SEM-IPS Kontron image analyzer (Germany) with a Panasonic CCTVcamera.

The maximum diameter of unilocular cells and the number ofcapillaries per unilocular cell were determined at the light micro-scopic level using eyepieces with an inbuilt calibrator, on resin-embedded material: four samples were selected for each animalstudied (four sections × three or four animals). Thick sections (2µm) were cut from each block and stained with toluidine blue. Inevery section we measured the diameter of the ten largest cells;therefore in total, for each experimental group, we measured thediameters of 106 cells. The 60 cells having the smallest diam-eters were eliminated and we calculated the mean diameter for theremaining 100 cells (mean maximal diameter) (Loncar et al.1988). In the same sections, the ratio of capillaries to unilocularadipocytes has been measured. For these two measurements onlysections composed of unilocular cells were used.

Mitochondrial area and mitochondria cristae permitochondrion

A total of 300 mitochondria, well preserved at a final magnifica-tion of ×17500, were measured for each experimental group. Thetotal area and the total length of the cristae of each mitochondrionwere plotted on the graphic tablet of the analyzer and automati-cally measured. All data were subjected to statistical analysis usingStudent’s t-test, for differences among the three experimentalgroups.

ImmunocytochemistryWe used sections of the tissue prepared for light microscopy asdescribed above. The 3 µm sections were dewaxed and washedtwice with ethanol before the endogenous peroxidase was blockedin 0.3% hydrogen peroxide in methanol for 30 min. A sheep anti-rat UCP serum was used at a dilution of 1:2500 by the ABCmethod (Hsu et al. 1981). In controls, non-immune purified sheepIgG (Sigma Chem., Saint Louis, USA) was substituted for the pri-mary antiserum.

Immunoelectron microscopySome fragments of periovarian adipose tissue, fixed in

Fig. 2. Atypical expression of UCPmRNA in white fat pads of ratanalyzed by PCR experiments. PCRanalysis was performed on 300 ng oftotal RNA with (lanes +) or without(lanes −) reverse transcriptase in thebuffer. The different fat pads wereinguinal (lanes 1), epididymal (lanes2), periovarian (lanes 3), mesenteric(lanes 4) and retroperitoneal (lanes5) adipose tissues of male (lanes *)and female rats. 100 ng (lane 6.1),50 ng (lane 6.2) and 10 ng (lane 6.3)of total IBAT RNA were used ascontrol. We used 300 ng of totalliver RNA (lanes 7) as negativecontrol and the plasmid containingthe UCP cDNA (lane 0) as positivecontrol. The size of the mRNA wasdetermined by using a kb ladder(MW).

934

paraformaldehyde, 3% in phosphate buffer, were embedded in LRWhite Resin (London Resin Company Ltd, England) and heat-cured at 52°C. Ultrathin sections were collected on 300-meshnickel grids and placed on droplets of bovine serum albumin(BSA, fraction V, Sigma Chem.) in 0.1 M phosphate bufferedsaline (PBS, pH 7.4) containing 1% Tween 20 (Sigma Chem.) for20 min at room temperature. After removing excess reagent with-out washing, the grids were incubated with 20 µl of a sheep anti-rat UCP antibody at a dilution of 1:10 in PBS/BSA + 1% Tween20 overnight at 4°C in a humid chamber. Grids were then washedby flotation on several droplets of PBS/BSA (1%) and succes-sively exposed to Protein A-colloidal gold complex (pAg, 10 mmparticle size, Sigma Chem.), diluted 1:25 in 1% PBS/BSA + 1%Tween 20, for 3 hours, at room temperature in a humid chamber.This step was followed by successive washes in PBS, postfix-ation in 1% glutaraldehyde in PBS for 5 min and a final wash indistilled water. The grids were then observed under a PhilipsCM10 electron microscope. In controls, non-immune purifiedsheep IgG (Sigma Chem.) was substituted for the primary anti-serum.

Results

Molecular and functional studiesAtypical UCP mRNA expression

The presence of UCP mRNA was investigated by North-ern blotting (Fig. 1A) or PCR experiments (Fig. 2). Asshown in Fig. 1A, using Northern blotting of 20 µg of totalRNA, no UCP mRNA could be detected in inguinal andmesenteric adipose deposits from control rats even afteroverexposure of the autoradiographs (not shown). However,in periovarian adipose tissue and, to a lesser extent, inretroperitoneal fat pad, two signals were detected. Thesesignals correspond to the sizes identified in IBAT (1.8 and1.5 kb) and the shorter signal was the more abundant, asin IBAT RNA. These results demonstrated that UCP mRNAwas expressed in other sites than those described as typi-cal BAT.

Twenty-four hours cold-acclimatation led to an increasein UCP mRNA in BAT, as well as in periovarian andretroperitoneal fat pads. Again no signal could be detectedin inguinal and mesenteric adipose tissue (Fig. 1B). Todetermine the level of mitochondrial transcription, which ishigh in brown adipocytes, we also hybridized these blotswith mitochondrial DNA. As shown in Fig. 1, the mito-chondrial transcript level was higher in depots whichexpressed UCP mRNA, except in inguinal fat pad. Greatindividual variability was noticed and it was higher inretroperitoneal than in periovarian pads. When the sameblots were hybridized with probes detecting lipoproteinlipase, Glut 4, fatty acid synthase or protein disulfide iso-merase mRNAs, such variability was not found in WAT(data not shown), leading to the conclusion that the vari-ability was specific to the UCP mRNA expression level.

To investigate the faint expression of UCP mRNA in epi-didymal, mesenteric and inguinal deposits, we performedPCR experiments on 300 ng of total RNA extracted fromdifferent white fat pads and liver. To test the efficiency ofthe amplification and to quantify the reaction, the same pro-cedure was carried out using 10, 50 or 100 ng of total RNAfrom IBAT. In the presence of reverse polymerase, a bandof expected size (196 bp) was detected in all white adipose

tissues studied, but not in liver. As shown in Fig. 2, nosignal was obtained when the reverse polymerase was omit-ted, except in mesenteric adipose tissue from a male rat. Inthis case, the faint signal at 287 bp is probably due to ampli-fication of genomic DNA: actually, the primers, OL3 andOL4, that were used match exons III and IV, respectively,and a size of 287 bp can be obtained by amplification ofthe end of exon III (117 bp), all of intron III (91 bp) andthe beginning of exon IV (79 bp) (Bouillaud et al. 1988).

Since the highest level of UCP mRNA expression wasfound in periovarian fat pad, studies to characterize betterthe cell type expressing UCP were undertaken on this pad.

Immunological detection of UCP protein Using anti-UCP antibodies, the presence of UCP could befaintly detected in homogenates from periovarian fat pads(lane 3, Fig. 3A) in contrast to homogenates from IBAT(lane 1, Fig. 3A). UCP was enriched in mitochondrial frac-tions (lane 4 compared to lane 3, Fig. 3A) and its level wasincreased after cold adaptation of rats for 3 days (lanes 2,3, 4 versus 5, 6, 7, Fig. 3B). However, UCP levels werestill much lower than in mitochondria isolated from BAT(lane 2, Fig. 3A; lane 1, Fig. 3B).

Regulation of UCP in periovarian fat pad: effect of cold,β-agonist and obesity

In IBAT, acute exposure to cold led to a rapid increase inthermogenic response and UCP mRNA levels, whichreturned to near control values after 10 days of cold expo-sure (Reichling et al. 1987). The same phenomenon wasobserved in periovarian fat (Fig. 1A and data not shown).

Treatment with a β3-adrenoceptor agonist (BRL 26830A,10 mg/kg) increased UCP mRNA levels considerably inperiovarian adipose tissue. The relative increase was evenhigher than it was in IBAT (Fig. 4). These results indicatethat the regulation of UCP mRNA expression by cold andβ3-adrenoceptor agonist are qualitatively similar in whiteadipose deposit and in IBAT.

B. Cousin and others

Fig. 3. Detection, by immunoblotting, of UCP in mitochondria ofperiovarian fat pads and increase during cold exposure.(A) Immunoblotting of UCP was performed on homogenates(lanes 1, 3) or mitochondria (lanes 2, 4), isolated from IBAT(lanes 1, 2) and periovarian fat pads (lanes 3, 4). 5 and 10 µg ofproteins were loaded per lane for IBAT and periovarian fat,respectively. (B) Mitochondria were isolated from IBAT (lane 1)and periovarian tissue of control (lanes 5 to 7) and animalsexposed to cold for 3 days (lanes 2 to 4). 10 µg of proteins wasloaded on each lane.

935Atypical localization of brown adipocytes

A defect in thermoregulation is believed to be involvedin the development of obesity (Himms-Hagen, 1990; Roth-well and Stock, 1979; Trayhurn, 1986). Indeed, a decreasedUCP mRNA concentration was observed in periovarian adi-pose tissue of obese (fa/fa) versus lean (Fa/Fa) Zucker rats,as in IBAT (Fig. 5).

Morphological studiesMacroscopically, the periovarian adipose tissue lookedwhite and was quite indistinguishable from classical whiteadipose tissue both in control rats (24°C, group A, n = 4)and in cold-exposed animals (3 days at 4°C, group B, n =4; 10 days at 4°C, group C, n = 3). To characterize betterthe cells expressing UCP in periovarian fat, histologicalanalysis was then undertaken.

Cell phenotypes in control ratsLight microscopy revealed the presence of multilocularcells in periovarian adipose tissue. Most multilocular cellswere isolated among unilocular cells. The number of mul-tilocular adipocytes varied considerably from one sectionto the other of the same fat pad (data not shown). In elec-tron microscopic analysis, histological data showed thepresence of at least five types of cells: poorly differentiated

cells with numerous atypical WAT-like mitochondria (Fig.6A), weakly differentiated multilocular cells with numer-ous typical mitochondria (young brown adipocytes) (Figs6B, 8A) found in proximity to capillary walls, well differ-entiated multilocular cells (brown adipocytes) (Fig. 8A),unilocular cells (white adipocytes) (Fig. 6A,B) and uniloc-ular cells with numerous mitochondria showing a mor-phology between BAT and WAT (Fig. 6B). It was note-worthy that the cell diameter was very variable, dependingon the differentiation state.

Using immunocytochemistry, multilocular cells werefound positive for UCP (Fig. 7). By contrast, unilocularcells were always negative for UCP. Multilocular cells werenegative when non-immune purified sheep antiserum wassubstituted for the primary antiserum (data not shown).

Effect of cold on morphological parametersIn periovarian pads of rats submitted to 3 or 10 days ofcold exposure (groups B and C, respectively), both themean number per section and the mean cellular density ofmultilocular cells increased when compared with controlgroup (A) (Table 1). The maximal diameter of unilocular

Fig. 4. Adrenergic control of the UCP mRNA level in IBAT andperiovarian fat pads. Rats were injected with either physiologicalserum (control rats) or BRL 26830A (10 mg/1 kg). Three hourslater, animals were killed and IBAT (A) and periovarian fat pads(B) were collected. Total RNA was extracted and analyzed byNorthern blotting (A, 2 µg; B, 20 µg). The membrane wasautoradiographed for 3 days.

Fig. 5. Decrease of UCP mRNA expression in periovarian fatpads of obese Zucker rats compared to lean rats. RNA wasextracted from IBAT (A) and periovarian fat pads (B) of 8-week-old obese and lean Zucker rats. After electrophoresis and blottingonto membrane, the nylon blots were hybridized to detect UCPmRNA (A, 20 µg; B, 5 µg) and autoradiographed for 3 days.

Table 1. Influenceof temperature on morphological parameters (± s.e.m.) in the periovarian adipose tissue

20°C 4°C for 3 days 4°C for 10 days

Mean maximal unilocular cell diameter (µm) 74.21±7.27 57.28±2 56.67±2.78Capillaries per unilocular cell (no.) 0.13±0.03 0.19±0.03 0.19±0.04Mean number of multilocular cells per section (no.) 9.625±1.99 97.3±20.28 376.63±85.07*Mean density of multilocular cells (no./mm2) 0.112±0.02 0.589±0.122 5.612±1.175*Mean mitochondrial area (µm2) 0.534±0.02 0.574±0.027 0.591±0.031Mitochondrial cristae per mitochondrion (µm/µm2) 5.551±0.24 7.145±0.28* 7.573±0.267

*P<0.05 (t-test Newman-Keuls).

936 B. Cousin and others

Fig. 6. Periovarian adipose tissue (group A: control animals). (A) Poorly differentiated cell with numerous mitochondria, in proximity toa capillary (c) and three unilocular adipocytes (U). Inset: enlargement of the framed area showing that the mitochondria are not typical ofthe mitochondria usually present in brown adipocytes. (B) Young brown adipocyte among two capillaries (c) and three unilocular cells(U). The unilocular cell on the upper left shows mitochondria with morphology transitional between typical (BAT-like) and atypical(WAT-like) mitochondria. Inset: enlargement of the framed area showing typical mitochondria.

937Atypical localization of brown adipocytes

cells and the number of capillaries per unilocular adipocytewere not significantly altered in groups B and C. Mito-chondrial area of groups submitted to cold exposure (B andC) did not change significantly, but the density of mito-chondrial cristae increased (Table 1).

In rats exposed for 3 days to cold (group B), as in con-trol rats (group A), most cells exhibiting the typical mor-phology of brown adipocytes were isolated among theunilocular cells. In group B, most of the weakly differen-tiated cells surrounding capillaries were young brownadipocytes with BAT-like mitochondria (Fig. 8A). We alsoobserved cells with BAT-like mitochondria, that had a por-tion of their cytoplasm partially surrounding a capillary wall(Fig. 8B).

In samples from group C (10 days at 4°C), we foundwell-differentiated brown adipocytes containing numerouslarge mitochondria packed with cristae (Fig. 9) and joinedby numerous gap junctions (Fig. 10). Young brownadipocytes were easily found in close apposition to capil-laries. Parenchymal nerve fibres were present only in areashaving groups of brown adipocytes from an animal of groupC (Fig. 9).

All multilocular cells from the three groups were posi-tive for UCP (Fig. 7). In the sections from rats exposed tocold (groups B and C), the reaction looked more intensethan in the section from control rats (group A). Cells fromthe three groups were processed for immunoelectronmicroscopy. In group C, numerous UCP-gold particles werepresent in the well-differentiated mitochondria of the multi-locular cells (Fig. 9). Young brown fat cell showed a lessintense, but still positive, reaction in the mitochondria (Fig.11). Interestingly, WAT-like mitochondria of the uni-locular cells showed a few UCP-gold particles (Fig. 11).

Discussion

The main result of this work was the detection of UCP inrat fat pads that were considered until now to be WAT.UCP is expressed in cells morphologically identical to typ-ical brown adipocytes found in IBAT.

UCP is not expressed exclusively in areas described astypical brown adipose tissueUtilization of antibodies and/or cDNAs led several authorsto conclude that UCP is unique to BAT (Cannon et al. 1982;Klaus et al. 1991b). The only exhaustive study of UCPexpression in different fat pads of animals kept at normaltemperature was made in ruminant species (Casteilla et al.1987; Vatnick et al. 1987; Casteilla et al. 1989; Soppela etal. 1991). To our knowledge, no such study has been per-formed in adult rodents not exposed to cold and no PCRanalysis has been performed on any species. The presentdata show clearly that UCP expression is not specific toareas described as typical BAT, but can also be found infat pads classically considered to be white. Thus, Northernanalysis and/or PCR experiments showed the presence ofUCP in periovarian, retroperitoneal and to a lesser extentepididymal, mesenteric and inguinal fat pads.

The cells expressing UCP in these deposits have the bio-

Fig. 7. Periovarian adipose tissue. (A) Animal kept at 20°C; (B)animal kept at 4°C for 3 days; (C) animal kept at 4°C for 10 days.Immunocytochemical staining with UCP antibodies (ABCmethod). Multilocular cells are positive for UCP.

938 B. Cousin and others

Fig. 8. Periovarian adipose tissue (group B: animal kept at 4°C for 3 days ). (A) Brown adipocytes at different levels of differentiation,surrounding a capillary (C). Inset: enlargment of the framed area showing numerous typical mitochondria. (B) Poorly differentiated cellwith a portion of its cytoplasm (arrowheads) partially surrounding the capillary (C) wall. The cell is surrounded by a distinct basal lamina(BL) and contains typical mitochondria.

939Atypical localization of brown adipocytes

chemical and morphological characteristics of brownadipocytes. Among the biochemical characteristics, theincrease in UCP level in these tissues after cold exposureor β-adrenoceptor agonist treatment favours the presence oftrue brown adipocytes. This was confirmed in periovarianadipose tissue of control rats by immunohistologicalstudies. When animals were cold-adapted, the number ofmultilocular cells and the density of mitochondrial cristaeincreased. These results are in agreement with those foundfor inguinal (Loncar, 1991) and parametrial adipose tissue(Young et al. 1984) of mice.

Physiological relevanceTaking into account the total RNA content of IBAT andWAT, and the results of Northern analysis, periovarian andretroperitoneal adipose depots are equivalent to one hun-dredth and one thousandth of IBAT, respectively, regard-

ing UCP expression. Although this amount of UCP mRNAin WAT is relatively small in comparison with that foundin BAT, one cannot exclude the possibility that there issome physiological relevance. Indeed, it should be notedthat under cold stress or β-adrenoceptor agonist treatment,the enhanced expression of UCP was more marked in peri-ovarian adipose tissue (7-fold) than in IBAT (3-fold). Theseobservations are in agreement with the thermogenic role ofbrown adipocytes. In addition to non-shivering thermogen-esis and arousal from hibernation, brown fat thermogene-sis may also participate in the regulation of energy balance.BAT thermogenesis is defective in genetic and experimen-tal models of obesity (Himms-Hagen, 1990; Trayhurn,1986). UCP reinduction in dogs treated with a β3-adreno-ceptor agonist is concomitant with a decreased weight gain(Champigny et al. 1991). In the present work, UCP mRNAwas expressed less in periovarian adipose tissue of obese

Fig. 9. Periovarian adipose tissue (group C: animal kept at 4°C for 10 days). Cytoplasm of a brown adipocyte showing typicalmitochondria packed with cristae. Right bottom inset: same type of cell tested with antibodies to UCP protein linked to 10 nm goldparticles (pAg method). The presence of numerous gold particles on the mitochondria demonstrates the presence of UCP in this organelle.Left upper inset: a parenchymal nerve fibre with numerous vesicles.

940

Zucker rats than in lean rats. This is similar to results foundfor IBAT of obese rats (Ricquier et al. 1986).

Thus, brown adipocytes from periovarian fat pad seemto behave like brown adipocytes from typical brown fat. Aspecific function for these brown fat cells related to theirlocation cannot be ruled out. On the other hand, the greatindividual variability of UCP expression in periovarian adi-pose tissue, which persists after cold exposure or β3-ago-nist treatment, cannot be explained. The possibility of hor-monal regulation linked to the sexual cycle is underinvestigation.

Tissue and cell plasticityThe present data demonstrate the great plasticity of adiposetissues in rodents but the mechanism underlying the reor-ganization of periovarian fat pad upon exposure to cold isstill difficult to explain. Adipose tissue plasticity hasalready been observed in developing hamsters (Smalley,1970), cats (Loncar and Afzelius, 1989) and ruminants(Gemell et al. 1972; Casteilla et al. 1987; Vatnick et al.1987; Casteilla et al. 1989; Soppela et al. 1991). In thesespecies, studies have suggested that rapidly after birth all

B. Cousin and others

Fig. 10. A gap junction between two brown adipocytes in theperiovarian adipose tissue of animal exposed in cold for 10 days(group C).

Fig. 11. Periovarian adipose tissue (group C). Immunoelectron microscopy of UCP antibodies (pAg method). A brown adipocyteprecursor with poorly differentiated mitochondria. Upper right inset: enlargement of the framed area showing the UCP-gold particles onthe mitochondria. Lower left inset: atypical mitochondrion of a unilocular cell showing only a few UCP-gold particles (arrows).

941Atypical localization of brown adipocytes

deposits loose BAT characteristics and seem to be trans-formed into white-like adipose tissues. On the other hand,WAT of adult humans and animals can be reconverted toBAT in particular conditions (Ricquier et al. 1982; Younget al. 1984; Loncar et al. 1988; Champigny et al. 1991;Cinti et al. 1991). Taken together, these results suggest apossible interconversion at the cellular level between brownand white-like adipocytes.

It is rather difficult to decide whether there has been anew recruitment of preadipocytes to the tissue or a real con-version of adipocyte phenotypes. Cells exhibiting the ultra-structure of brown adipocyte precursors as defined byBarnard and Skala (1970) and Néchad and Barnard (1979)are present in periovarian adipose tissue of rats kept at20°C. On the other hand, the presence of unilocular cellswith a morphology between BAT and WAT, or with mito-chondria faintly positive for UCP, suggests that someunilocular cells are “masked” brown adipocytes whichcould be “unmasked” under physiological or pharmacolog-ical stimuli. This is reminiscent of the histological obser-vations made by Gemmel et al. (1972) and of the proposalthat there is convertible adipose tissue, made by Loncar(Loncar et al. 1986; Loncar, 1991). Innervation, which hasbeen observed in BAT-like arrangements of the tissue fromcold-adapted rats, could be a major determinant of theseprocesses. Nevertheless, more quantitative data are neces-sary in order to draw reliable conclusions.

A new classification for adipose tissue: analogy withmuscle The present study suggests a new classification for adiposetissues. Ashwell proposed the existence of a continuumbetween BAT and WAT among different types of mam-mals (Ashwell, 1985; Ashwell et al. 1990). Our results arein favor of this hypothesis and suggest that this continuumof adipose tissue from typical BAT to typical WAT existsin the same species. In rodents, the adipose tissues form-ing the extremities of this spectrum would have a popula-tion of adipocytes with a major phenotype and the inter-mediates would be made up of mixed populations, likeperiovarian adipose tissue. Furthermore, the data suggestthat transformation during cold exposure or in obese ratscould take place along this spectrum. According to theirplace along this spectrum, adipose tissues could participatein different functions and in metabolic regulation.

This suggests an analogy between muscle and adiposetissue. Like adipose tissue, muscle is a heterogeneous tissuecomposed of a mixture of red and white fibres the propor-tions of which vary from one muscle to another. Red fibresand brown adipocytes, as opposed to white fibres and whiteadipocytes, are highly oxidative cells. In rodents, skeletalmuscles, as well as adipose tissues, seem to be very adapt-able and to respond to different situations by changing theirstructure and metabolism according to new metabolicdemands. As in adipose tissue, in a given muscle the pro-portions of each type of cells vary, according to the phys-iological (Salmon and Henriksson, 1981), and/or patholog-ical situation (Lillioja et al. 1987; Torgan et al. 1989).

In summary, these data demonstrate the presence ofbrown adipocytes in numerous white fat pads of rodents.This suggests that adipocytes with thermogenic properties

may be more generally distributed than previously thought.Whether this is the case in other species, particularly inhumans, remains to be determined.

The authors thank C. Zingaretti, E. Ceresi and V. Carboni fortheir excellent technical assistance, and Dr J. Girard for his sup-port. We thank Drs Odette Champigny and Susi Klaus for theirhelpful comments and corrections on this manuscript. B. Cousinwas a fellow of the French Ministry of Research and Technology.This work was made possible by grants from the Centre Nationalde la Recherche Scientifique, Fondation pour la Recherche Médi-cale.

References

Afzelius, B. A. (1970). Brown adipose tissue: Its gross anatomy, histol-ogy and cytology. In Brown Adipose Tissue (ed. O. Lindberg), pp. 1-31. New York: American Elsevier.

Ashwell, M. (1985). Is there a continuous spectrum of the adipose tissuesin animals and man? In Metabolic Complications of Human Obesities(ed. J. Vague et al.), pp. 265-274. New York: Elsevier Science Pub-lishers B. V.

Ashwell, M., Stirling, D., Freeman, S. and Holloway, B. (1990). Trans-formations within the continuous spectrum of the adipose tissues. InRecent Advances in Obesity Research: V (ed. E. M. Berry et al.), pp.160-166. London: J. Libbey.

Barnard, T. and Skala, J. (1970). The development of brown adiposetissue. In Brown Adipose Tissue (ed. O. Lindberg), pp. 33-72. NewYork: American Elsevier.

Bibb, M., Van Etten, R., Wright, C., Walberg, M. and Clayton, D.(1981). Sequence and gene organization of mouse mitochondrial DNA.Cell 26, 167-178.

Bouillaud, F., Raimbault, S., and Ricquier, D. (1988) The gene for ratuncoupling protein: complete sequence, structure of primary transcriptand evolutionary relationship between exons. Biochem. Biophys. Res.Commun. 157, 783-792.

Bouillaud, F., Ricquier, D., Thibault, J. and Weissenbach, J. (1985).Molecular approach to thermogenesis in brown adipose tissue. cDNAcloning of the mitochondrial uncoupling protein. Proc. Nat. Acad. Sci.U.S.A. 82, 445-448.

Bradford, M. M. (1976). A rapid and sensitive method for the quantita-tion of microgram quantities of protein utilizing the principle of pro-tein-dye binding. Anal. Biochem. 72, 248-254.

Bukowiecki, L. J., Geloen, A. and Collet, A. J. (1986). Proliferation anddifferentiation of brown adipocytes from interstitial cells during coldexposure. Amer. J. Physiol. 250, C880-C887.

Cannon, B., Hedin, A. and Nedergaard, J. (1982). Exclusive occurrenceof thermogenin antigen in brown adipose tissue. FEBS Lett. 150, 129-132.

Casteilla, L., Champigny, O., Bouillaud, F., Robelin, J. and Ricquier,D. (1989). Sequential changes in the expression of mitochondrial pro-tein mRNA during the development of brown adipose tissue in bovineand ovine species. Biochem. J. 257, 665-671.

Casteilla, L., Forest, C., Robelin, J., Ricquier, D., Lombet, A. and Ail-haud, G. (1987). Characterization of mitochondrial uncoupling pro-tein in bovine fetus and newborn calf. Disappearance in lamb duringaging. Amer. J. Physiol. 252, E627-E636.

Champigny, O., Ricquier, D., Blondel, O., Mayers, R. M., Briscoe, M.G. and Holloway, B. R. (1991). β3-adrenergic receptor stimulationrestores message and expression of brown-fat mitochondrial uncou-pling protein in adult dogs. Proc. Nat. Acad. Sci. U.S.A. 88, 10774-10777.

Chomczynski, P. and Sacchi, N. (1987). Single-step method of RNA iso-lation by acid guanidinium thiocyanate-phenol-chloroform extraction.Anal. Biochem. 162, 156-159.

Cinti, S. (1992). Morphological and functional aspects of brown adiposetissue. In The Obese Child, vol. 2 (ed. P. L. Giorgi et al.), pp. 125-132. Basel, Karger.

Cinti, S., Morroni, M. and Carboni, V. (1991). Interscapular brown adi-pose tissue of old rats after cold acclimatation and cafeteria diet: amorphometric study. Int. J. Obes. 15 (Suppl. 1), 64.

942

Geloën, A. and Trayhurn, P. (1990). Regulation of the level of uncou-pling protein in brown adipose tissue by insulin requires the media-tion of the sympathetic nervous system. FEBS Lett. 267, 265-267.

Gemmell, R. T., Bell, A. W. and Alexander, G. (1972). Morphology ofadipose cells in lambs at birth and during subsequent transition ofbrown to white adipose tissue in cold and in warm conditions. Amer.J. Anat. 133, 143-164.

Himms-Hagen, J. (1990). Brown adipose tissue thermogenesis: interdis-ciplinary studies. FASEB J. 4, 2890-2898.

Hsu, S. M., Raine, L., and Fanger, H. (1981). Use of avidin-biotin-per-oxydase complex (ABC) in immunoperoxydase technic: a comparisonbetween ABC and unlabelled antibody (PAP procedures). J. His -tochem. Cytochem. 29, 577-580.

Klaus, S., Cassard-Doulcier, A. M. and Ricquier, D. (1991a). Devel-opment of Phodopus Sungorus brown preadipocytes in primary cellculture: Effect of an atypical beta-adrenergic agonist, insulin, and tri-iodothyronine on differentiation, mitochondrial development, andexpression of the uncoupling protein UCP. J. Cell Biol. 115, 1783-1790.

Klaus, S., Casteilla, L., Bouillaud, F. and Ricquier, D. (1991b). Theuncoupling protein UCP: a membraneous mitochondrial ion carrierexclusively expressed in brown adipose tissue. Int. J. Biochem. 23,791-801.

Lillioja, S., Young, A. A., Culter, C. L., Ivy, J. L., Abbott, W. G. H.,Zawadzki, J. K., Yki-Jarvinen, H., Christin, L., Secomb, T. W. andBogardus, C. (1987). Skeletal muscle capillary density and fiber typeare possible determinants of in vivo insulin resistance in man. J. Clin.Invest. 80, 415-424.

Loncar, D. (1991). Convertible adipose tissue in mice. Cell Tiss. Res. 266,149-161.

Loncar, D. and Afzelius, B. A. (1989). Ontogenetical changes in adiposetissue of the cat: convertible adipose tissue. J. Ultrastruct. Res. 102,9-23.

Loncar, D., Afzelius, B. A. and Cannon, B. (1988). Epididymal whiteadipose tissue after cold stress in rats. I. Nonmitochondrial changes.J. Ultrastruct. Res. 101, 199-209.

Loncar, D., Bedrica, L., Mayer, J., Cannon, B., Nedergaard, J.,Afzelius, B. A. and Svajger, A. (1986). The effect of intermittent coldtreatment on the adipose tissue of the cat. Apparent transformationfrom white to brown adipose tissue. J. Ultrastruct. Mol. Struct. Res.97, 119-129.

Napolitano, L. (1963). The differentiation of white adipose cells. J. CellBiol. 18, 663-679.

Néchad, M. and Barnard, T. (1979). Development of the interscapularbrown adipose tissue in the hamster. Two pathways of adipocyte differ-entiation and the development of the sympathetic innervation. Biol.Cell 36, 43-53.

Néchad, M. (1986). Structure and development of brown adipose tissue.In Brown Adipose Tissue (ed. P. Trayhurn and D. G. Nicholls), pp.1-30. London: E. Arnold.

Nicholls, D. G. and Locke, R. M. (1984). Thermogenic mechanisms inbrown fat. Physiol. Rev. 64, 1-64.

Reichling, S., Ridley, R.G., Patel, R. H., Harley, C. V., and Freeman

K. B. (1987) Loss of brown adipose tissue uncoupling protein mRNAon deacclimation of cold-exposed rats. Biochem. Biophys. Res.Commun. 142, 696-701.

Rehnmark, S., Néchad, M., Herron, D., Cannon, B. and Nedergaard,J. (1990). α and β-adrenergic induction of the expression of the uncou-pling protein thermogenin in brown adipocytes differentiated in cul-ture. J. Biol. Chem. 265, 16464-16471.

Ricquier, D., Bouillaud, F., Toumelin, P., Mory, G., Bazin, R., Arch,J. and Pénicaud, L. (1986). Expression of the uncoupling proteinmRNA in thermogenic or weakly thermogenic brown adipose tissue.Evidence for a rapid β-adrenoceptor-mediated and transcriptionallyregulated step during activation of thermogenesis. J. Biol. Chem. 261,13905-13910.

Ricquier, D., Néchad, M. and Mory, G. (1982). Ultrastructural and bio-chemical characterization of human brown adipose tissue in pheochro-mocytoma. J. Clin. Endocrinol. Metab. 54, 803-807.

Ricquier, D., Raimbault, S., Champigny, O., Miroux, B. and Bouil-laud, F. (1992). The uncoupling protein is not expressed in rat liver.FEBS Lett. 303, 103-106.

Rothwell, N. J. and Stock, M. J. (1979). A role for brown adipose tissuein diet-induced thermogenesis. Nature 281, 31-35.

Rothwell, N. J. and Stock, M. J. (1985). Biological distribution and sig-nificance of brown adipose tissue. Comp. Biochem. Physiol. 82A, 745-751.

Salmon, S. and Henriksson, J. (1981). The adaptative response to skele-tal muscle to increased used. Muscle Nerve 4, 94-105.

Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). In MolecularCloning. A Laboratory Manual (ed. C. Nolan et al.), New York: C. S.H.

Silva, J. E. (1988). Full expression of uncoupling protein gene requiresthe concurrence of norepinephrine and triiodothyronine. Mol.Endocrinol. 2, 706-713.

Smalley, R. L. (1970). Changes in composition and metabolism duringadipose tissue development. In Brown Adipose Tissue (ed. O. Lind-berg), pp. 73-95. New York: American Elsevier.

Soppela, P., Nieminen, M., Saarela, S., Keith, J. S., Morrison, J. N.,McFarlane, F. and Trayhurn, P. (1991). Brown fat-specific mito-chondrial uncoupling protein in adipose tissues of newborn reindeer.Amer. J. Physiol. 260, R1229-R1234.

Torgan, C. E., Brozinick, J. T., Kastello, G. M. and Ivy, J. L. (1989).Muscle morphological and biochemical adaptation to training in obeseZucker rats. J. Appl. Physiol. 67, 1807-1813.

Trayhurn, P. (1986). Brown adipose tissue and energy balance. In BrownAdipose Tissue (ed. P. Trayhurn and D. G. Nicholls), pp. 299-338.London: E. Arnold.

Vatnick, I., Tyzbir, R. S., Welch, J. G. and Hooper, A. P. (1987).Regression of brown adipose tissue mitochondrial function and struc-ture in neonatal goats. Amer. J. Physiol. 252, E391-E395.

Young, P., Arch, J. R. S. and Ashwell, M. (1984). Brown adipose tissuein the parametrial fat pad of the mouse. FEBS Lett. 167, 10-14.

(Received 16 June 1992 - Accepted 3 September 1992)

B. Cousin and others