Abstract
Uncoupling protein 1 (UCP1), the mammalian thermogenic mitochondrial protein, is found only in brown adipocytes, but its expression by immunohistochemistry is not homogeneous. Here we present evidence that the non-homogeneous pattern of immunostaining for UCP1 (referred to as the “Harlequin phenomenon”) is particularly evident after acute and chronic cold (4C) stimulus and after administration of a specific β3-adrenoceptor agonist (CL316,243). Accordingly, mRNA in situ expression confirmed the UCP1 non-homogeneous pattern of gene activation under conditions of adrenergic stimulus. Furthermore, morphometric analysis of immunogold-stained thin sections showed that UCP1-gold particle density was different among neighboring brown adipocytes with mitochondria of the same size and density. When the adrenergic stimulus was reduced in warm-acclimated animals (28C), UCP1 protein and mRNA expression was reduced and consequently the Harlequin phenomenon was barely visible. These data suggest the existence of an alternative and controlled functional recruitment of brown adipocytes in acute adrenergically stressed animals, possibly to avoid heat and metabolic damage in thermogenically active cells. Of note, the heat shock protein heme oxygenase 1 (HO1) is heterogeneously expressed in adrenergically stimulated brown adipose tissue and, specifically, cells expressing strong immunoreactivity for UCP1 also strongly express HO1.
B
Genetic ablation of BAT induces obesity (Lowell et al. 1993), whereas genetic ablation of UCP1 does not (Enerback et al. 1997). UCP1 knockout mice appear to compensate by increasing proton leak in their muscle mitochondria, but not via any known uncoupling protein (Monemdjou et al. 2000). A better knowledge of UCP1 expression and regulation could be very important for the study of pharmacological strategies for obesity treatment.
Other proteins, called UCP2 and UCP3, with high homology to the UCP1, have been discovered (Boss et al. 1997; Fleury et al. 1997; Vidal-Puig et al. 1997). UCP2 and −3 are found in BAT as well as in many other mammalian tissues, but UCP1 is exclusively expressed in BAT (Boss et al. 1997; Fleury et al. 1997; Vidal-Puig et al. 1997).
Before the identification of UCP2 and UCP3 proteins, non-uniform staining of brown adipocytes has been reported by an immunohistochemical study with a rabbit antiserum against rat UCP, suggesting cellular heterogeneity of brown adipocytes in BAT (Cadrin et al. 1985).
To better understand this phenomenon, we studied by IHC the expression of UCP1 protein in animals exposed to different thermogenic stimuli using a polyclonal antibody specific for UCP1 that did not crossreact with UCP2 and UCP3 (Pecqueur et al. 2000). Furthermore, we quantified the mitochondrial UCP1 expression at the ultrastructural level (by immunogold staining) to compare the ultrastructure of cells expressing different amounts of the protein.
To verify if this non-homogeneous pattern of expression is due to a real non-homogeneous expression of the gene, we studied the in situ expression of the UCP1 mRNA in serial sections obtained from the same animals. Our results indicate that the phenomenon is due to a true non-homogeneous UCP1 gene activation and protein expression, more evident and diffuse in tissues obtained under conditions of acute adrenergic stimulus.
We proposed that the Harlequin phenomenon is the consequence of an alternative functional recruitment, particularly well evident in conditions of major adrenergic stress (i.e., when maximal heat production is required from each single adipocyte). One possible explanation may be the need to avoid metabolic and heat-shock damage in individual cells.
Materials and Methods
Animals and Tissues
Male 6-week-old Sprague-Dawley rats were acclimated to cold (4C for 20 hr, acute stimulus; 4C for 2 weeks, chronic stimulus) and warmth (28C for 2 weeks). The specific β3-adrenoceptor pharmacological stimulus consisted of CL316,243 (disodium (R, R)-5-[2–[[2-(3-chlorophenyl)-2-hydroxyethyl]-amino]-propyl]-1,3-benzodioxazole-2,2-dicarboxylate) in saline solution administered by osmotic minipumps (Alzet 2001; Alza, Palo Alto, CA) implanted SC (daily dose 1 mg/Kg−1) for 7 days in adult male rats 22 weeks old; control animals received only saline solution. Three animals for each experimental condition were used. All rats were purchased from Harlan-Italy (Milan, Italy) and housed in plastic cages with free access to chow and water in a light-dark cycle of 12/12 hours. Animal care complied with institutional guidelines.
Before cold and warmth exposure and during the pharmacological treatment, animals were maintained at room temperature (20C). Under anesthesia with 0.4% xylazine-0.1% ketamine/500 g body weight, rats were transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. Interscapular brown adipose tissue (BAT), liver, kidney, spleen, skeletal muscle, and rat fetal liver were dissected, dehydrated, and paraffin-embedded. These tissues were used for tests of antibody specificity.
Immunohistochemistry
Serial (3-μm) sections were either stained with hematoxylin and eosin or processed for IHC with a polyclonal anti-rat UCP1 antibody raised in sheep (Ricquier et al. 1983) and, only for the acute cold stimulus group, with a polyclonal anti-rat inducible isoform of heme oxygenase 1 (HO1) antibody raised in rabbit (StressGen Biotechnologies; Victoria, BC, Canada). Immunohistochemical demonstration of UCP1 and HO1 was performed in de-waxed sections according to the ABC method (Hsu et al. 1981): 0.3% hydrogen peroxidase in methanol for 30 min to block endogenous peroxidase; normal rabbit (UCP1) and normal goat (HO1) serum diluted 1:75 in PBS for 20 min to reduce nonspecific background staining; polyclonal sheep anti-rat UCP1 1:12,000 and rabbit anti-rat HO1 1:200 in PBS overnight at 4C; biotinylated rabbit anti-sheep (UCP1) IgG and goat anti-rabbit (HO1) IgG (Vector Laboratories; Burlingame, CA) 1:200 for 30 min at RT; ABC complex (Vectastain ABC kit; Vector) for 60 min at RT; enzymatic development of peroxidase using 0.05% diaminobenzidine hydrochloride (Sigma; St Louis, MO) and 0.02% H2O2 in 0.05 M Tris buffer, pH 7.6, for 5 min in a dark room. UCP1 specificity was tested at the same dilution on tissues expressing the highest levels of UCP2 (skeletal muscle, kidney, spleen, fetal liver) (Boss et al. 1997; Fleury et al. 1997; Hodny et al. 1998) and UCP3 (skeletal muscle) (Vidal-Puig et al. 1997). The specificity of the UCP1 antibody employed in this study has been recently confirmed (Pecqueur et al. 2000). All the tissues were negative. HO1 specificity was tested on spleen (positive control; Braggins et al. 1986) from the same animals. Sections were then dehydrated in alcohol, cleared in xylol and mounted in Entellan (Merck; Darmstadt, Germany).
Electron Microscopy
Small tissue fragments were fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M PB, pH 7.4, and postfixed in 1% OsO4, dehydrated in ethanol, and embedded in an Epon-Araldite mixture (Epon: Multilab Supplies, Fetcham, UK; Araldite: Fluka Chemie, Buchs, Switzerland). Semi-thin sections (2 μm) were stained with toluidine blue; thin sections were obtained with a MT-X ultratome (RCM; Tucson, AZ), stained with lead citrate, and examined with a Philips CM10 transmission electron microscope (Philips; Eindhoven, Netherlands).
Immunoelectron Microscopy, Morphometry, and Statistical Analysis
These analyses were performed only in BAT of animals treated with the β3-adrenergic agonist (CL316,243) or stimulated by acute cold, conditions in which the non-homogeneous pattern of immunopositivity was clearly evident (a clearer Harlequin pattern).
After perfusion, small fragments were fixed in 4% paraformaldehyde in 0.1 M PB, pH 7.4, and embedded in LR White Resin (Multilab). All incubation steps were performed on ultrathin sections mounted on nickel grids, floating on drops of reagent. Grids were incubated with (a) 1% BSA (Sigma) in PBS for 30 min to reduce background staining, (b) primary antibody, sheep anti-rat UCP1, diluted 1:400 in PBS containing 1% BSA and 0.2% Tween-20 (Sigma) over-night at 4C, (c) PBS, (d) secondary antibody conjugated to gold particles, donkey anti-sheep IgG-5-nm gold (British Biocell; Cardiff, UK) diluted 1:40 in PBS for 30 min at RT. After several washes in PBS and distilled water, grids were counterstained with uranyl acetate and observed with a Philips CM10 transmission electron microscope.
For morphometric analysis, 20 adjacent cells were randomly selected for each animal of the two groups, in total 60 cells for experimental group. Image analysis of electron micrographs was performed at a final magnification of ×11, 270 with a Kontron KS100 image analyzer (Kontron Elektronics; Eching, Germany), a semiautomatic software that enabled us, by outlining profiles, to obtain data about mitochondrial area, mitochondrial density (mitochondrial number/cell cytoplasmic area, nucleus and lipid droplets excluded), and gold particle density per mitochondrion. Our statistical analysis was performed comparing only the 15 less positive cells of an experimental group with the 15 most positive cells of the same group. All results are presented as mean ± SEM. Group means were compared by two-way ANOVA.
In Situ Hybridization
For in situ hybridization (ISH) tests, animals were perfused in 4% paraformaldehyde diluted in 1 ×PBS; the same tissues indicated above (animal and tissue section) were dissected and paraffin-embedded with a different procedure: dehydration in alcohol (ETOH 70%, 85%, 95%, absolute), clarification in xylene (twice for 30 min), paraffin:xylol (equal volume) for 45 min at 56C, and three times in paraffin (at the third passage the tissue was oriented in the paraffin block). Sections of 3 μm were cut and mounted on poly-L-lysine-treated slides to avoid detachment of the sections during the hybridization procedure. The probe (~800-bp UCP1 cDNA fragment inserted into a polylinker site of pTZ-19R vector in opposite orientation to synthesize ISH riboprobe antisense and sense) was previously labeled by an RNA Labeling kit (Roche; Monza-Mi, Italy) accordingly to the manufacturer's protocol. The transcription mixture included 1 μg of linearized template cDNA, ATP, GTP, and CTP at 1 mM each, UTP 0.65 mM and 0.35 mM DIG-UTP, RNase inhibitor (1 U/μl of transcription mix), and T7 or SP6 RNA polymerase (1 U/μl of transcription mix). Transcription was performed for at least 2 hr at 37C. The template cDNAs were then digested by RNase-free DNase (30 min at 37C) and all reactions were stopped by addition of 1 mM EDTA. The riboprobes were then purified through precipitation by 4 M LiCl and 100% ETOH overnight at —20C or 2 hr at —70C. The DIG incorporation into the probes was controlled by dot-spots, according to a published method (Non-radioactive In Situ Hybridization 1996). Sections were rehydrated through successive baths of xylene, ETOH (100%, 95%, 85%, 75%, 50%, 30%), H2O, diethylpyrocarbonate (DEPC)-treated (twice for 5 min each), 1 × PBS solution (twice for 5 min each), 1 × PBS solution with 0.1% active DEPC for 15 min, and 5 × SSC (NaCl 0.75 M, Na citrate 0.075 M). Sections were then prehybridized for 2 hr at 55C in the hybridization mix (5 × SSC, 50% form-amide, 40 μg/ml denatured salmon sperm DNA). The probe was denatured for 5 min at 80C and added to the hybridization mix (400 ng/ml). The hybridization reaction was carried out at 55C overnight in a humid chamber. After the incubation, sections were washed for 30 min in 2 × SSC (RT), 1 hr in 2 × SSC (58C), 1 hr in 0.1 × SSC (58C), and equilibrated for 5 min in Buffer 1 (Tris-HCl 100 mM and NaCl 150 mM, pH 7.5). The sections were then incubated for 2 hr at RT with alkaline phosphatase-coupled anti-digoxigenin antibody (Roche) diluted 1:500 in Buffer 1 containing 0.5% blocking reagent (Roche). Excess antibody was removed by two 15-min washes in Buffer 1 and sections were equilibrated for 5 min in Buffer 2 (Tris-HCl 100 mM, NaCl 100 mM, and MgCl2 50 mM, pH 9.5). Color development was performed at RT overnight in Buffer 2 containing NBT/BCIP (Roche). Staining was stopped by a 10-min wash in Tris-EDTA (10–1 mM, pH 8.0). Sections were stained with hematoxylin and mounted with glycerol-gelatin, then observed by light microscopy. No positive signal was observed in liver, kidney, and skeletal muscle using the antisense UCP1 riboprobe and in BAT using the sense UCP1 riboprobe.
Results
Antibody and Riboprobe Tests
The results described here were obtained with specific anti-rat UCP1 and HO1 antibodies and rat UCP1 mRNA antisense riboprobes. No crossreactions with UCP2 (in liver, kidney, skeletal muscle, and fetal liver) and UCP3 (skeletal muscle) were observed by these techniques. For details see Materials and Methods.
Acute Cold (4C) Stress
By light microscopy, most brown adipocytes showed eosinophilic cytoplasm, apparently lacking in lipid droplets. Electron microscopy revealed a cytoplasm rich in well-developed mitochondria and few small lipid droplets in all multilocular cells. Brown adipocytes showed a uniform distribution of cellular organelles and particularly of mitochondria (Figure 1A). UCP1 was expressed at the IHC level with different intensities of positivity in adjacent brown adipocytes. Adipocytes strongly positive, nearly completely negative, or weakly positive were clearly observed. This IHC pattern was previously defined as the “Harlequin effect” in a preliminary report by our team (Cinti et al. 1997) (Figure 2A). In different lobules of BAT of the same or different animals, the number of strongly positive cells was widely variable. Nevertheless, the Harlequin pattern was detectable in each lobule of each acute cold-stressed animal. Wide areas of BAT lobules were completely negative. UCP1 mRNA expression had a positivity pattern very similar to that of the protein IHC localization described above. Strongly positive cells were found close to negative ones, and some cells also showed nuclear positivity, with or without cytoplasmic staining (Figure 2B).
Chronic Cold (4C) Stress
BAT of chronically stressed animals was hypertrophic. Brown adipocytes showed many lipid droplets easily
Representative electron micrographs of interscapular brown adipocytes of adult rats acutely exposed to cold, 4C (
Representative IHC (UCP1) and ISH (UCP1 mRNA) of interscapular brown adipose tissue of adult rats acutely exposed to cold (20 hr at 4C;
visible by light microscopy and many mitochondria with well-developed cristae (not shown). UCP1 was weakly expressed at IHC level by all brown adipocytes of BAT, and only some lobules showed the Harlequin effect (Figure 2C). No completely negative areas of BAT were detectable. In situ hybridization revealed a UCP1 mRNA expression pattern similar to the IHC positivity: weak and homogeneous positivity and small groups of adipocytes strongly positive were detectable (Figure 2D).
β3 (CL316,243)-agonist-treated Animals
After 7 days of CL316,243 treatment, all brown adipocytes of BAT showed the same morphology. By both light and electron microscopy, multilocular cells were very similar to multilocular cells of cold-acclimated rats. The Harlequin phenomenon was detectable in wide areas of BAT lobules, showing a pattern similar to that described for acute cold-stressed rats (Figure 3).
Animals Acclimated at 20C
BAT showed, by light microscopy, the same morphology observed in brown adipocytes of chronic cold-stressed rats. Many mitochondria were visible in all brown adipocytes, but number, dimension, and density of cristae were lower than in multilocular adipocytes of the experimental groups described above
Representative serial sections of interscapular brown adipose tissue of rats treated with the β3-agonist CL316,243 for 7 days, stained by IHC (UCP1;
(data not shown). UCP1 protein expression was weak, and wide areas of BAT were completely negative. Some lobules showed a weak Harlequin effect, although multilocular cells with a high intensity of immunopositivity were present. UCP1 mRNA was homogeneous and weak in all multilocular brown adipocytes (data not shown).
Animals Acclimated at 28C
The environmental temperature of 28C is very close to thermoneutrality for rats. As a consequence, inter-scapular brown adipocyte morphology was greatly modified in these animals, mainly by a higher lipid accumulation at the cytoplasmic level. Lipid droplets were larger than in multilocular cells of other conditions and, in many cells, lipid accumulation was unilocular. Mitochondria were numerous in all multilocular adipocytes but showed a lower density of cristae than those observed in cold or β3-adrenoceptor-stimulated animals (Figure 1B). UCP1 immuno-staining was weak and the Harlequin phenomenon was not visible because strongly positive adipocytes were absent (Figure 2E). UCP1 mRNA expression was not detectable (Figure 2F).
UCP1 Protein and mRNA Expression on Serial Sections of Brown Adipocytes
Expression of UCP1 protein and mRNA was studied in serial sections only in BAT with a clear and diffuse Harlequin pattern (i.e., acute cold-stressed and CL316, 243-treated animals). Strongly immunoreactive cells stained positive for UCP1 protein and for the UCP1
IHC on serial sections of BAT of an acute cold-stressed rat stained with UCP1 antibodies (
mRNA (Figures 3A and 3B), but brown adipocytes strongly positive for UCP1 protein and negative for the UCP1 mRNA, as well as negative for both of them, were also found (Figures 3C and 3D). Multilocular cells positive for the UCP1 mRNA and negative for UCP1 protein were not observed. In BAT serial sections of acute cold-stressed animals, we observed that the same areas of the tissue containing cells positive for UCP1 also contained HO1-positive cells (Figures 4A and 4B). It was also possible to observe that both proteins were contained in the same serial-sectioned cells (Figures 4E and 4F).
Immunoelectron Microscopy and Morphometry
UCP1 protein expression obtained with the immunogold staining (IGS) method was performed only in the experimental conditions in which the Harlequin pattern was clear and diffuse in the tissue (acute cold and CL316,243) because only a restricted area of the tissue can be examined with this technique. Indeed, this technique aimed to compare the ultrastructure of adjacent adipocytes with different expression of UCP1.
In acute cold and in CL 316,243 experimental conditions, the numbers of UCP1/gold particles in mitochondria of brown adipocytes were significantly different (p<0.001) in adjacent adipocytes (Figures 5 and 6). Comparison of the ultrastructure between cells with significantly different UCP1 expression showed no differences in mitochondrial numerical density and area (Figure 6).
Discussion
Before the discovery of uncoupling protein variants, a heterogeneous expression of UCP in brown adipocytes of stimulated animals was previously described (Cadrin et al. 1985), suggesting the presence of cell heterogeneity in BAT. In this article we demonstrate that this phenomenon, named by our team “Harlequin” in a preliminary report (Cinti et al. 1997), is still detectable using a polyclonal antibody specific for UCP1 and not crossreacting, in IHC tests with other cell types known to contain UCP2 and UCP3 (Boss et al. 1997; Fleury et al. 1997; Vidal-Puig et al. 1997), and that it is particularly evident and diffuse in BAT of adrenergically stimulated animals. The specificity of the anti-UCP1 antibody employed in this study has been recently confirmed (Pecqueur et al. 2000).
Representative immunoelectron microscopic images (IGS method) of two adjacent brown adipocytes of interscapular brown adipose tissue from a rat treated with the β3-agonist CL316,243 for 7 days. UCP1/5-nm gold particles are localized exclusively in the mitochondria. The density of the gold particles is clearly higher in the mitochondria of the cell at the top of the figure. Bar = 0.32 μm. (
An increase in UCP1 mRNA and protein in adrenergically stimulated BAT is well known (Bouillaud et al. 1984; Ricquier et al. 1986; Jacobsson et al. 1986, 1987, 1994). Our data, showing the overlapping distribution pattern observed for UCP1 mRNA and protein, led us to postulate that the Harlequin phenomenon could be the result of an alternative UCP1 gene activation and protein expression in neighboring brown adipocytes. The most common finding in serial sections of cells positive for the protein and mRNA, or only for the protein, is in keeping with the longer lifespan of the protein and the tight correlation of UCP1 and its messenger production in BAT (Nedergaard and Cannon 1998). Furthermore, it has been shown that cold selectively destabilizes the mRNA for UCP1 (Jacobsson et al. 1987). The large areas of BAT negative for both UCP1 and its mRNA in acute-stressed animals suggests the presence of different cell types not expressing this protein, but the homogeneous morphology of the parenchymal cells observed by electron microscopy and the diffuse UCP1 expression in chronically stimulated animals argue against this hypothesis.
Immunoelectron microscopy revealed that the different immunoreactivity is due to a different content of mitochondrial UCP1 protein in mitochondria of adjacent brown adipocytes showing similar ultrastructural morphology, i.e., with mitochondrial size and density not significantly different. The low resolution limit of the IGS method does not allow verification of the hypothesis that the different density of UCP1-gold particles in mitochondria is associated with a different development of cristae.
The absence or reduction of the Harlequin phenomenon in less stimulated animals (20C and 28C) shows a tight correlation with the level of activation of the UCP1 gene in brown adipocytes studied by the ISH technique. The distribution of parenchymal nerve fibers in IBAT of small mammals is dependent on the environmental temperature (De Matteis et al. 1998), and the norepinephrine concentration in this tissue is variable in different lobules (Foster et al. 1982).
However, the Harlequin phenomenon is not due to these factors. In fact, it is also clearly evident in pharmacologically treated animals in which the adrenergic stimulus is independent of the distribution of nerve fibers. The β3-adrenoceptor distribution at the parenchymal level was never studied, and we cannot exclude that a non-homogeneous distribution or downregulation of this receptor could be responsible, at least in part, for the phenomenon. The reduction of the heterogeneous positivity pattern in chronic cold-stimulated animals could be due to the hyperplasia of brown adipocytes that occurs in this condition (Morroni et al. 1995; Himms-Hagen and Ricquier 1998) and suggests that an increased number of brown adipocytes would not require a maximal activation of each adipocyte. A possible relationship between cells highly positive for UCP1 and proximity to blood vessels can be excluded on the basis of the distribution of capillaries among brown adipocytes. In fact, every single cell is reached by at least one capillary in the brown adipose tissue (Néchad 1986).
When brown adipocytes are stimulated by norepinephrine, the rapid induction of UCP1 uncouples oxidative phosphorylation with a concomitant increase in metabolic rate and fat utilization. Increased free radical generation due to increased fatty acid oxidation, relative hypoxemia, and heat production subjects brown adipocytes to metabolic stress. As a consequence, stimulated brown adipocytes produce different heat-shock proteins (HSP; Bouillaud et al. 1984; Matz et al. 1995; Beattie et al. 2000) and, specifically, the HO1 (also known as HSP32) (Giordano et al. 2000). HO1 expression increases in BAT of acute cold-stressed rats and, accordingly, norepinephrine upregulates its production in cultured brown adipocytes. Immunohistochemistry with anti-HO1 antibodies performed on brown fat of cold-exposed rats produces a Harlequin pattern of staining very similar to that observed for UCP1 and UCP1 mRNA. Furthermore, the presence of brown adipocytes expressing both UCP1 and HO1 proteins in activated BAT suggests a simultaneous and/or induced activation of the two genes: HO1 protein and its byproducts could be involved in the cell
Morphometry of immunoelectron microscopic experiments. The mean UCP1/gold particle mitochondrial density of the 15 cells with lower density (low-density dots) were compared with the mean UCP1/gold particle mitochondrial density of the 15 cells with the highest density (high-density dots) on a total of 60 examined cells for each experimental group. The difference between the two groups of cells is highly significant (p<0.001) (upper panel). The mitochondrial area and the mitochondrial density in the cytoplasm of the same cells are not significantly different (middle and bottom panels).
and tissue protection mechanisms triggered by heat produced by activated brown adipocytes. Of note, some nuclei of brown adipocytes also contain HO1, and their number increases after cold exposure (Giordano et al. 2000).
Taken together, our observations suggest that the immunohistochemical “Harlequin” pattern is a true functional phenomenon. This alternative functional recruitment of noradrenergic-stimulated brown adipocytes could be a protective mechanism to avoid metabolic- and heat production-dependent damage in activated cells, mainly under conditions of acute nor-adrenergic stress that probably require maximal heat production for each brown adipocyte. Further studies are need to confirm this hypothesis.
Footnotes
Acknowledgements
Supported by grants from Ministero dell'Università e della Ricerca Scientifica e Tecnologia 2001 (to SC) and by the University of Ancona.
We are grateful to Dr Kurt Steiner (Wyeth Ayerst) for the supply of CL316,243.