Introduction

Interleukin-1, a major mediator of inflammation, also performs numerous functions related to host defence by regulating not only the immune system, but also the neural and endocrine systems. Accordingly, IL-1 is produced by a wide variety of cells, including monocytes, macrophages, epithelial cells, endothelial cells and glial cells. Moreover, IL-1 receptors (IL-1Rs) are found in a wide range of immune, neural and endocrine cells, reflecting the pleiotropic activities of this molecule [1]. Leptin, which is released from adipocytes, exerts an inhibitory feedback effect on fat masses by acting on hypothalamic nuclei, which express the cognate signal-transducing leptin receptor ObRb [2]. IL-1 production, which is induced by leptin, is involved in the leptin-induced suppression of feeding [3]. Another member of the IL-1 family, IL-1 receptor antagonist (IL1RA) binds to IL-1 receptors without exerting agonistic activity. This protein is considered to be a negative regulator of IL-1 signalling, participating in the complex regulation of IL-1 activity. Interestingly, serum IL1RA levels are seven times higher in obese human patients than in non-obese study participants [4]. In addition, a large quantity of IL1RA is secreted from adipose tissue, although the biological significance of this phenomenon is not well understood [5]. IL-1 activates sympathetic neural circuits; moreover, i.v. and intracerebroventricular administration of IL-1 activates efferent sympathetic nerve discharge [6–8]. Using Il1 −/− and Il1ra (also known as Il1rn)-deficient (Il1ra −/−) mice, we demonstrated that IL-1 has a physiological role in feeding behaviour and energy metabolism [9]. Il1ra −/− mice, which have a defect in lipid accumulation in adipose tissue, are lean and have reduced serum insulin levels [9, 10]. Recently, Garcia et al. demonstrated that Il1r1 −/− mice developed maturity-onset obesity, beginning to deviate from the weight of wild-type (WT) mice at 5 to 6 months of age [11].

A balance between energy intake and expenditure is important for the maintenance of body weight and normal physiology. Energy expenditure represents the sum of the total energy needed for maintaining normal cell and organ function, metabolism, physical activity and facultative (adaptive) thermogenesis [12]. The sympathetic nervous system (SNS) is thought to be critical for the prevention of obesity associated with a high-fat diet (HFD) [13]. Activation of the SNS induces uncoupling protein 1 (UCP-1) production and activates diet-induced thermogenesis in brown adipose tissue (BAT) [14, 15]. BAT is thought to be involved in the control of body temperature and body weight via cold- and diet-induced thermogenesis, whereas white adipose tissue (WAT) stores and releases energy. WAT is directly innervated by the SNS, as has been established for BAT [16]. Mice lacking BAT due to the production of diphtheria toxin in this tissue are obese and insulin-resistant [17], indicating the importance of BAT in the control of body weight. Administration of β-adrenergic receptor (AR) agonists increases the metabolic rate [18–20], and mice lacking all three β-AR genes become markedly obese when reared on a HFD [21]. In contrast, transgenic overexpression of human β1-AR (also known as ADRB1) in WAT resulted in lean mice and an abundance of brown adipocytes in WAT [22]. Brown and white adipocytes are usually located in distinct depots and can be distinguished morphologically [23]; brown adipocytes contain multilocular lipid vacuoles and numerous mitochondria, whereas white adipocytes have a unilocular lipid vacuole and few mitochondria. UCP-1 is specifically produced in BAT; this protein uncouples oxidative phosphorylation from electron transport, resulting in the dissipation of energy as heat [24, 25]. The emergence of these ectopic cells in the WAT of rats [26] and of mice [27] was found to be induced by cold acclimatisation or by the administration of selective β3-AR agonists [27, 28]. Interestingly, the emergence of brown fat cells in white fat depots was associated with a lean phenotype in several transgenic mouse models, including β1-AR transgenic mice [22], FOXC2 transgenic mice [29] and Tif2 −/− (also known as Ncoa2 −/−) mice [30].

In this study, we further characterised the metabolic phenotype of Il1ra −/− mice reared on a HFD. We found that Il1ra −/− mice are resistant to obesity associated with decreased respiratory quotient (RQ) (i.e. RQ = carbon dioxide production (\(\dot V{\text{CO}}_{\text{2}} \))/oxygen uptake (\(\dot V{\text{O}}_{\text{2}} \)), indicating that the ratio of fat:carbohydrate metabolism in Il1ra −/− mice is greater than in controls. Histological analysis of BAT suggests that this tissue is hyper-activated in Il1ra −/− mice. Gene expression and electron microscopy analyses of inguinal WAT in Il1ra −/− mice revealed that this tissue had BAT-like characteristics. The urinary norepinephrine concentration was significantly higher in Il1ra −/− mice than in WT mice, indicating that the former had increased sympathetic tone. Moreover, heart rate and body temperature in Il1ra −/− mice were significantly increased compared with WT mice, which is also consistent with the idea that Il1ra −/− mice have increased sympathetic tone. Our results suggest that activation of BAT and BAT-like changes of WAT in Il1ra −/− mice are responsible for the decreased RQ and resistance to HFD-induced obesity.

Methods

Animals

Il1ra −/− mice were produced as described previously [31]. These mice were backcrossed to C57BL/6J mice for nine generations. After weaning, 4-week-old mice were housed individually. Age-matched male littermates or adult (9–16 weeks of age) male mice obtained from a breeder (SLC, Shizuoka, Japan) and housed at our facility were used for each experiment. Mice were kept under specific pathogen-free conditions in environmentally controlled clean rooms at the Center for Experimental Medicine, Institute of Medical Science, University of Tokyo. Animals were housed at an ambient temperature of 24°C under a daily 12 h light–dark cycle (lights on 08:00–20:00 hours) on a normal-chow diet (5.1% of the total energy from fat; total energy 17.5 kJ/g). All experiments were performed according to the institutional ethical guidelines for animal experimentation and safety guidelines for gene manipulation experiments.

Diet study and metabolic measurements

At 4 weeks of age, male Il1ra −/− and WT mice were placed on and had free access to a high-carbohydrate HFD (50% of total energy from fat; total energy 22.1 kJ/g). Animals were then studied for the next 12 weeks. Total body weight was measured weekly for 12 weeks starting at 4 weeks of age. Blood samples were collected after a 6 h fast at 4, 8 and 12 weeks of age. The mice were killed at 16 weeks of age. Food intake was measured as described previously [32].

Indirect calorimetry

Whole-body O2 consumption and CO2 production were measured in a respiration chamber that measured 140 × 80 × 90 mm in size and was ventilated with fresh air at a rate of 200 ml/min. The differences in the concentrations of O2 and CO2 between inflow and outflow air were measured with a differential O2 analyser (LC700E; Toray, Tokyo, Japan) and two CO2 sensors (GMW22D; Vaisala, Helsinki, Finland). Each mouse was placed in the chamber for 23 h; to avoid the influence of emotional thermogenic responses to cage-exchange stress, the data recorded during the first hour were not analysed. The results were then corrected for metabolic body mass (g 0.75).

Measurement of heart rate and body temperature

A telemetric recording system was used to simultaneously record locomotor activity, heart rate and body temperature from conscious and unrestrained mice as previously described [33, 34]. Briefly, a telemetric transmitter for locomotor activity, electrocardiogram and body temperature was implanted under anaesthesia. The mice were housed in individual cages and placed on a signal-receiving board in a light-proof chamber. Heart rate and body temperature were continuously recorded every 5min by a Data Quest analysing system (Data Sciences International, St Paul, MN, USA).

Measurement of blood, serum and urinary parameters

Blood glucose levels were measured using the glucose oxidase method (Terumo, Tokyo, Japan), whereas serum triacylglycerol and NEFA levels were examined using colorimetric assays (triacylglycerol-E and NEFA-C tests, respectively; Wako, Osaka, Japan). Serum insulin and leptin levels were both measured with enzyme-linked immunosorbent assays (Seikagaku, Tokyo, Japan) and radioimmunoassays (Eiken, Tokyo, Japan). At 12 weeks of age, Il1ra −/− mice and WT mice were placed in metabolism cages that provided free access to tap water and food. Urinary catecholamine excretion was determined by reverse-phase HPLC (SRL, Tokyo, Japan) and corrected by urine creatinine concentration (SRL, Tokyo, Japan). Glucose tolerance and insulin tolerance tests were performed as described in the Electronic supplementary material (ESM).

Electron microscopy

Tissues were dissected from the mice under deep anaesthesia and immediately fixed in a solution containing 2.5% (wt/vol.) glutaraldehyde and 2% (wt/vol.) formaldehyde in 0.1 mol/l sodium phosphate buffer (pH 7.4) for 2 h at room temperature. After fixation, tissues were rinsed and post-fixed in 2% (wt/vol.) osmium tetroxide in the same buffer solution on ice. The samples were then washed, dehydrated in a graded series of ethanol and embedded in Epon 812 resin mixture (TAAB, Aldermaston, UK). Semi-thin sections (approximately 0.7µm thick) were cut on a an Ultracut N ultramicrotome (Reichert Depew, NY, USA), stained with 0.2% Toluidine Blue and examined under a microscope (Eclipse E600; Nikon, Tokyo, Japan). Ultra-thin sections were cut, stained with uranyl acetate and lead citrate, and examined with an electron microscope (H-7500; Hitachi, Tokyo, Japan).

Statistical analysis

All values are represented as the means ± SEM. Differences between the body weight curves and food intake were evaluated using repeated-measures ANOVA, in which factor 1 was the between-groups factor and factor 2 was the within-subject factor (different ages). Differences between the glucose and insulin level were evaluated using two-way ANOVA followed by Tukey’s tests. Comparisons of the two groups were analysed using Student’s t tests. In all analyses, a two-tailed probability of less than 5% (p<0.05) was considered to be statistically significant.

Results

Il1ra−/− mice are protected from diet-induced obesity

To better understand the metabolic phenotype of Il1ra −/− mice, we monitored diet-induced changes in these mice. On a HFD, Il1ra −/− mice remained lean compared with WT mice (Fig. 1a), whereas HFD-fed WT control mice became obese compared with mice kept on a standard diet (Fig. 1a). The average weights of the epididymal and inguinal fat pads and interscapular brown fat pads of Il1ra −/− mice on a HFD were significantly lower than those of WT mice under the same feeding conditions, whereas the relative weights of liver, soleus muscle and spleen were proportional to the decreased body weight (Fig. 1b) of Il1ra −/− mice.

Fig. 1
figure 1

Growth curves, glucose levels and insulin levels in WT and Il1ra −/− mice on a HFD. a Growth curves in WT mice (littermates; n = 11, white diamonds, normal chow; n = 10, black diamonds, HFD) and Il1ra −/− mice (n = 9, white squares, normal chow; n = 8, black squares, HFD) that had free access to normal chow or HFD. b BAT, epididymal WAT (eWAT), inguinal WAT (iWAT), liver, soleus and spleen weight of 12-week-old Il1ra −/− mice (n = 6) under HFD. Data are expressed as a relative value compared with WT mice (littermates; n = 4). Blood glucose (c), serum insulin (d) and serum leptin levels (e) after 6 h fast in 16-week-old WT (littermates) and Il1ra −/− mice (KO) with free access to normal chow or HFD. Data are expressed as means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. BW, body weight

Full size image

Altered glucose homeostasis and enhanced insulin sensitivity in Il1ra−/− mice

We also compared glucose homeostasis in WT and Il1ra −/− mice. The fasting blood glucose levels in Il1ra −/− mice on a HFD were lower than those of the WT control mice (Fig. 1c), whereas normal chow-fed WT and Il1ra −/− mice did not show significant differences in their fasting or their fed blood glucose levels (Fig. 1c and [9]). Consistent with these data, Il1ra −/− mice under 20 weeks of age had a considerably enhanced ability to clear glucose from the peripheral circulation during i.p. glucose tolerance tests (ESM Fig. 1a). To determine whether increased insulin sensitivity could account for the improved glucose tolerance, we measured serum insulin levels and performed insulin tolerance tests in vivo. The average serum insulin level decreased by 78% in HFD-fed Il1ra −/− mice compared with WT siblings (Fig. 1d). Likewise, Il1ra −/− mice under 20 weeks of age on a HFD showed a significantly greater decrease than WT control mice in their blood glucose level during insulin tolerance tests (ESM Fig. 1b). Increased glucose and insulin tolerance were also observed in Il1ra −/− mice under 9 weeks of age (data not shown). Thus, insulin sensitivity was enhanced in HFD-fed Il1ra −/− mice compared with WT mice.

Serum leptin levels in HFD-fed Il1ra −/− mice were significantly lower than those of WT mice, consistent with the decreased fat mass in the former (WT mice 4941 ± 896 pg/ml; Il1ra −/− mice 907 ± 161 pg/ml; p<0.05) (Fig. 1e). The levels of circulating triacylglycerol were significantly decreased in HFD-fed Il1ra −/− mice compared with WT mice (WT mice 8.82 ± 0.55 mmol/l; Il1ra −/− mice 6.99±0.55 mmol/l; p < 0.05), while the levels of circulating NEFA were not significantly different (WT mice 0.52 ± 0.03 mEq/l; Il1ra −/− mice 0.55±0.03 mEq/l; p = 0.335), consistent with increased lipid metabolism in Il1ra −/− mice.

Increased energy expenditure and decreased respiratory quotient in Il1ra−/− mice

The markedly reduced weight of HFD-fed Il1ra −/− mice suggests that they may have either reduced food intake and/or increased energy expenditure [12]. Food intake per mouse was significantly decreased in Il1ra −/− mice, while food intake normalised to body weight was not significantly different between WT and Il1ra −/− mice that were fed a HFD (ESM Fig. 2). Total daily energy expenditure in HFD-fed WT and Il1ra −/− mice was assessed by measuring \(\dot V{\text{O}}_{\text{2}} \) and \(\dot V{\text{CO}}_{\text{2}} \), and activity was assessed by beam breaks. Daily levels in Il1ra −/− mice were significantly lower (Fig. 2a,b). On the other hand, daily energy expenditure (normalised for body weight) was significantly increased by 16% in Il1ra −/− mice compared with that of WT mice (Fig. 2c). Interestingly, RQ during the light period was significantly lower than in WT mice, whereas in the dark period RQ was not significantly different (Fig. 2d). Thus, the ratio of fat:carbohydrate metabolism in Il1ra −/− mice is greater than in WT mice, and they expend more energy than WT mice. Furthermore, the body temperature was significantly increased in HFD-fed Il1ra −/− mice compared with that of WT mice (Fig. 2e). These results suggest that Il1ra −/− mice have increased metabolic rates.

Fig. 2
figure 2

Activity and metabolic rate of WT and Il1ra −/− mice on a HFD. a Activity was measured as beam break counts during 24 h in WT (from breeder) (n = 6, white diamonds) and Il1ra −/− mice (n = 6, black squares). Data are expressed as means. b The bars represent the mean activity level over 1 day (LP + DP), light period (LP) or dark period (DP) in WT (from breeder; n = 6, white bars) and Il1ra −/− mice (n = 6, black bars). c, d Oxygen consumption relative to body mass (c) and RQ (d) measured over 1 day, light period or dark period in WT (littermates; n = 6, white bars) and Il1ra −/− mice (n = 5, black bars). e Body temperature was measured during 24 h on HFD in WT (from breeder; n = 6, white bars) and Il1ra −/− mice (n = 6, black bars). Data are expressed as means±SEM (b–e). *p < 0.05, **p < 0.01, ***p < 0.001

Full size image

Induction of Ucp1 mRNA expression in the inguinal WAT of Il1ra−/− mice

We next assayed the expression of Ucp1 in WAT, BAT and soleus muscle to survey the possible sites of energy expenditure. The expression level of Ucp1 mRNA in the inguinal WAT of Il1ra −/− mice was significantly higher than in WT mice (Fig. 3a), whereas levels of Ucp1 in epididymal WAT and soleus muscles were not significantly different (data not shown) [9]. We also found that the expression of genes encoding rate-limiting enzymes of mitochondrial fatty acid oxidation, such as Mcad (also known as Acadm) and Mcpt1, was increased in inguinal WAT from Il1ra −/− mice, suggesting that the increased level of oxidation in the inguinal WAT of Il1ra −/− mice probably resulted from an increase in the number of metabolically active mitochondria. The expression level of Ucp1 mRNA in the BAT was relatively high in Il1ra −/− mice compared with that in WT mice, although it was not statistically significant (Fig. 3b). The expression of Mcad and Cox4 (also known as Cox4i1) was significantly induced in Il1ra −/− BAT (Fig. 3b).

Fig. 3
figure 3

Real-time PCR quantification of Ucp1, Mcad, Mcpt1, Cox4 and Cyt-c (also known as Cycs) mRNA expression in BAT and inguinal WAT from WT and Il1ra −/− mice on a HFD. Inguinal WAT (a) and BAT (b) from WT (littermates) or Il1ra −/− mice were isolated and the level of mRNA expression as indicated was normalised using Rps3 values in Il1ra −/− mice (n = 6, black bars) and are shown as a ratio compared with WT mice (n = 6, white bars). Data are expressed as means±SEM. *p < 0.05

Full size image

The morphology of brown adipocytes in the inguinal WAT of Il1ra−/− mice

Histological examinations showed that the lipid vacuoles in the BAT of the Il1ra −/− mice were markedly smaller than those of WT mice (Fig. 4a,d), suggesting that the increased activity of BAT also contributes to increased RQ and metabolic rates. Moreover, the mass of the inguinal WAT was smaller in HFD-fed Il1ra −/− mice than in HFD-fed WT mice and some of the adipocytes from the Il1ra −/− mice exhibited a multilocular phenotype (Fig. 4c,f). Analysis of inguinal WAT using transmission electron microscopy showed a sharp increase in the number of mitochondria in the Il1ra −/− mice (Fig. 4g,i) compared with WT mice (Fig. 4h). While the sizes of the epididymal WAT were similar in HFD-fed Il1ra −/− mice and HFD-fed WT mice, there were some multilocular adipocytes in epididymal WAT in HFD-fed Il1ra −/− mice (Fig. 4b,e). The appearance of BAT-like cells was not observed at all in WAT depots in WT mice.

Fig. 4
figure 4

Histological analysis of adipose tissue in WT and Il1ra −/− mice on HFD. Histology of BAT (a, d), epididymal WAT (b, e) and inguinal WAT (c, f) from WT (littermates) mice (a–c) or Il1ra −/− mice (d–f) (magnification ×20) on a HFD. Electron micrographs of adipocytes in inguinal WAT (g) from 12-week-old Il1ra −/− mice or WT mice (h). Inset (g) is shown in higher magnification (×7,000) (i). Note the abundance of large mitochondria that are rich in cristae (well differentiated ‘brown’ mitochondria) and the juxtaposition of mitochondria to small lipid vacuoles. Scale bars, 50 µm (a–f), 2.5 µm (g, h)

Full size image

Hyperactivation of sympathetic nervous system in Il1ra−/− mice

We next analysed catecholamine metabolism as a measure of sympathetic tone in Il1ra −/− mice. There was no significant difference between Il1ra −/− mice and WT mice in the daily urine volume (data not shown). The urinary epinephrine and norepinephrine concentration normalised to urinary creatinine was significantly higher in Il1ra −/− mice than in WT mice (Fig. 5a,b). These results suggest that Il1ra −/− mice have increased sympathetic tone. Furthermore, overall, light- and dark-phase values of heart rate in Il1ra −/− mice were significantly higher than in WT mice (Fig. 5c). These results are consistent with the notion that Il1ra −/− mice have SNS hyperactivity.

Fig. 5
figure 5

Urinary excretion of catecholamines and heart rate in WT and Il1ra −/− mice on a HFD. Urine epinephrine (a) and norepinephrine (b) levels expressed as nmol/µmol creatinine in WT (from breeder; n = 6, white bars) and Il1ra −/− mice (n = 6, black bars). Urine samples (24 h) were collected and assayed for catecholamines. c Heart rate was measured during 24 h on HFD in WT (from breeder; n = 6, white bars) and Il1ra −/− mice (n = 6, black bars). DP, dark period; LP, light period; LP+DP, 1 day. Data are expressed as means ± SEM. *p < 0.05. bpm, beats per min

Full size image

Discussion

In this report, we analysed Il1ra −/− mice on a HFD and found that they are resistant to HFD-induced obesity, which is consistent with previous reports [9, 10]. The SNS is activated in these mice, resulting in the activation of BAT and emergence of multilocular BAT-like cells in WAT depots.

Il1ra −/− mice on a HFD had an increased level of energy expenditure normalised to body weight compared with WT mice (Fig. 2). Whereas Somm et al. observed an augmented level of energy expenditure in Il1ra −/− mice on a normal diet [10], we only detected a significant increase in energy expenditure on a HFD; the discrepancy is probably due to differences in the diet or housing conditions. As locomotor activity was decreased in Il1ra −/− mice under HFD (Fig. 2b, c), basal energy expenditure should increase. We found that HFD-fed Il1ra −/− mice had lower RQ than WT mice, indicating that the Il1ra −/− mice had an increased fat:carbohydrate oxidation ratio. The result is consistent with the observation that BAT-like cells appeared in WAT depots, because β-oxidation of fatty acids is preferentially used in BAT to produce energy. Our results are also consistent with the recent findings by Garcia et al., which demonstrated that Il1r1 −/− mice developed maturity-onset obesity with an increased RQ, indicating that Il1r1 −/− mice had a decreased fat:carbohydrate oxidation ratio [11]. Taken together, these results strongly suggest that IL-1 primarily affects fat:carbohydrate oxidation ratio.

Interestingly, some multilocular BAT-like cells appeared in the inguinal WAT depots of Il1ra −/− mice, whereas no BAT-like cells were observed in WT mice. We have never found any transformations of WAT into BAT in the Il1ra −/− mice under normal diet conditions [9]. Histologically, the BAT-like cells resembled brown adipocytes, and inguinal WAT depots from Il1ra −/− mice had increased levels of Ucp-1 (Figs 3 and 4). There were large differences in the numbers of BAT-like cells that appeared in WAT depots between each animal and between each depot in the same animal. BAT-like cells also appeared in some epididymal WAT depots in Il1ra −/− mice. However, we never found BAT-like cells in WT mice. Although it is possible that BAT-like cells in the WAT depots contribute to the increased RQ and increased energy expenditure, we should interpret these findings with caution, as the basal metabolism of WAT is very low.

Urinary norepinephrine and epinephrine excretion were significantly increased in Il1ra −/− mice compared with WT mice (Fig. 5), suggesting that sympathetic tone was increased in the Il1ra −/− mice. Consistent with the idea that Il1ra −/− mice have increased sympathetic tone, Il1ra −/− mice had increased heart rate and body temperature (Figs 2e and 5c). The key question is: How might Il1ra deficiency result in increased sympathetic tone? It was previously demonstrated that IL-1 acts as an activator of sympathetic neural circuits, and that i.v. and intracerebroventricular administration of IL-1 activates efferent sympathetic nerve discharge [6–8]. Since the balance between IL-1 and IL1RA determines the net IL-1 signalling, Il1ra deficiency in the brain may contribute to the increased activity of SNS. Consistent with this notion, we have previously demonstrated that levels of Il1β and Cox-2 (also known as Ptgs2) were increased in Il1ra −/− mice compared with WT mice [35], suggesting that excess IL-1 signalling is enhanced in the brain. Furthermore, levels of Il6, Tnfα (also known as Tnfa) and Il1β in WAT were decreased (though not significantly) in Il1ra −/− mice, indicating that inflammatory changes in response to HFD are relatively low in Il1ra −/− mice, as opposed to the possible role of IL-1 signalling in HFD-induced inflammatory change in WAT [35, 36]. Thus, it seems unlikely that excess IL-1 signalling in the periphery is responsible for the lean phenotype.

IL-1 has been recognised as a so-called endogenous pyrogen and is the signal responsible for fever. Considering the fact that Il1ra −/− mice on a HFD had increased body temperature, it is possible that the primary effect of Il1ra deficiency is to increase body temperature. However, we have previously demonstrated that Il1ra −/− Il-6 −/− mice on a normal diet show a lean phenotype comparable to Il1ra −/− mice [32]. Since IL-6 is essential for the IL-1-induced febrile response [37, 38], it is not possible that the lean phenotype could be explained solely by the action of IL-1 as a pyrogen. The activation of SNS and conversion of WAT into BAT may be involved in the elevated body temperature in Il1ra −/− mice.

We showed that serum leptin levels in Il1ra −/− mice on a HFD were significantly lower compared with WT mice, although food intake was similar between WT and mutant mice. These observations suggest that leptin sensitivity, which plays an important role in protecting against diet-induced obesity by suppressing appetite [39], is increased in Il1ra −/− mice. In this regard, it is interesting that leptin signalling in the CNS regulates energy expenditure via activation of the SNS, leading to BAT-like differentiation of WAT [40].

Because animals subjected to chronic β3 adrenergic stimulation show an increased number of multilocular BAT-like cells and decreased WAT stores [18], it is likely that the increased sympathetic tone in Il1ra −/− mice is responsible for the increased metabolic activity in BAT and the appearance of BAT-like cells in inguinal WAT of the Il1ra −/− mice (Figs 4 and 5). BAT is present throughout the lives of rodents, but in primates it disappears soon after birth; there are no BAT depots in adult humans. Nevertheless, variable quantities of brown adipocytes have been detected in typical WAT depots in humans [41]. Hence, brown adipocytes in WAT depots may play an important role in preventing obesity via a thermogenic mechanism, both in humans and in mice. While we found an increased Ucp1 expression in inguinal WAT, further studies will be required in order to determine whether the increase in Ucp1 represents an increase in brown adipocytes within WAT or a WAT-to-BAT conversion or clonal expansion of bona fide BAT depots.

It was recently demonstrated that Gnasxl m+P− mice are lean as a result of increased lipid mobilisation and oxidation in adipose tissue due to increased sympathetic tone [42]. A similar metabolic profile was also observed in mouse models with increased β-adrenergic/G protein α subunit/cAMP signalling in adipocytes, such as adipose-specific overproduction of the forkhead transcription factor FOXC2 [29], and in mice in which the C/ebpα (also known as Cebpa) gene was replaced with the C/ebpβ (also known as Cebpb) gene [43]. Taken together, these results indicate that β-adrenergic/G protein α subunit/cAMP signalling in adipocytes is an important regulator of whole-body homeostasis. Recently, several inflammatory cytokine-deficient mice, including Il6 −/− [44], Il1r1 −/− [11], Csf2 −/− [45], Il18 −/− [46, 47] and Il18r1 −/− [47] mice, were shown to develop maturity-onset obesity, suggesting the involvement of age-related changes in the SNS in the phenotypes of these cytokine-deficient mice. It is possible that inflammatory cytokines are protective against age-related declines in β-adrenergic responsiveness.

Recently, it was demonstrated that recombinant IL1RA (Anakinra) is beneficial in animal models [48] of diabetes and diabetic patients [49], suggesting that excess IL-1 signalling is responsible for the beta cell damage in diabetes. Consistent with this, IL-1 produced in beta cells contributes to glucotoxicity of beta cells [50]. In light of these observations, we histologically analysed islets of HFD-fed mice to examine possible effects of HFD on islets. However, HFD feeding for 16 weeks under our experimental conditions did not affect beta cell mass either in WT or in Il1ra −/− mice. Islets in Il1ra −/− mice were histologically normal and had normal hormonal content (O. Hashimoto, D. Chida, Y. Iwakura, unpublished data). Thus, the effect of Il1ra deficiency under HFD conditions is observed mostly in the SNS, and not in the pancreatic islets, in Il1ra −/− mice.

Taken together, these observations suggest that chronically increased sympathetic tone in Il1ra −/− mice is responsible for the activation of BAT and the emergence of multilocular BAT-like cells in WAT. Together with low levels of insulin secretion, these changes may contribute to the resistance to HFD-induced obesity observed in these animals. Our findings may suggest novel approaches for developing therapeutics for pathogenic obesity.