Obesity and related diseases are a major cause of human morbidity and mortality and constitute a substantial economic burden for society. Effective treatment regimens are scarce, and new therapeutic targets are needed. Brown adipose tissue, an energy‐expending tissue that produces heat, represents a potential therapeutic target. Its presence is associated with low body mass index, low total adipose tissue content and a lower risk of type 2 diabetes mellitus. Knowledge about the development and function of thermogenic adipocytes in brown adipose tissue has increased substantially in the last decade. Important transcriptional regulators have been identified, and hormones able to modulate the thermogenic capacity of the tissue have been recognized. Intriguingly, it is now clear that humans, like rodents, possess two types of thermogenic adipocytes: the classical brown adipocytes found in the interscapular brown adipose organ and the so‐called beige adipocytes primarily found in subcutaneous white adipose tissue after adrenergic stimulation. The presence of two distinct types of energy‐expending adipocytes in humans is conceptually important because these cells might be stimulated and recruited by different signals, raising the possibility that they might be separate potential targets for therapeutic intervention. In this review, we will discuss important features of the energy‐expending brown adipose tissue and highlight those that may serve as potential targets for pharmacological intervention aimed at expanding the tissue and/or enhancing its function to counteract obesity.

Introduction In 1551, Swiss naturalist Konrad Gessner first described brown adipose tissue (BAT) in the interscapular region of marmots [Muris (Marmota) alpinus]. However, it was not until the early 20th century that the existence of BAT in humans was recognized 1, 2. At the time, the tissue was referred to as the interscapular gland in humans or the hibernating gland in smaller mammals. These early studies described the anatomical location, gross morphology and histomorphology of the highly vascularized and innervated tissue, which is largely composed of rounded or polygonal multilocular fat‐filled cells that have granular cytoplasm and centrally located nuclei. BAT was initially believed to function during hibernation, but its presence in nonhibernating animals and humans suggested other uses. Although Polimanti speculated about a thermoregulatory function in 1912, it was not until 1958 that Johansson concluded from a literature review that ‘brown fat, at least in some animals, appears to be important in the regulation of body temperature’ 3, 4. During the 1960s, several studies evaluated the tissue as a thermogenic organ 5. Silverman et al. 6 showed that during cold exposure of human newborns, the skin temperature of human newborns fell the least over the nape of the neck, the location coinciding with BAT. Dawkins and Scopes found that the increased oxygen consumption that occurred in infants subjected to cold was associated with elevated plasma glycerol levels but no accompanying rise in plasma free fatty acids (FFAs) 7. As BAT does not express glycerol kinase, these results indicated that lipolysis occurred locally in BAT, which used the released FFAs for heat production. Additionally, histological analysis of BAT from deceased neonates revealed that the tissue was filled with fat if the infants had been nursed at thermoneutrality, but that it was depleted from fat if they had been conventionally swaddled and nursed at room temperature (reviewed in 5). In the 1960s and early 1970s, several studies were conducted to determine the distribution of BAT in humans. The studies focused primarily on the presence of BAT in infants in whom the tissue was abundant and widely distributed 8-10. Heaton also presented data, suggesting that BAT gradually disappeared with increasing age, which supported the prevailing opinion that thermogenically active BAT was common in human infants but regressed with age, leaving little or no active BAT in adults 9. However, the view that adult humans virtually lack active BAT was challenged by unexpected findings in nuclear medicine about 20 years later. Specifically, during positron emission tomography (PET) with the tracer [18F]‐fluorodeoxyglucose (FDG) for cancer staging or surveillance, a confounding symmetrical tracer uptake was often found in the neck and shoulder area of patients 11. These areas were unrelated to the tumours, and computed tomography (CT) showed that they had features of adipose tissue. Uptake in these regions was less prevalent when patients were acclimatized to warm indoor temperatures prior to the scan, which led to the hypothesis that adult humans retain significant amounts of metabolically active BAT. Several studies were subsequently initiated to test the hypothesis, and in April 2009, three independent studies, all confirming the presence of metabolically active BAT in adult humans, were published 12-14. Since then, many studies related to BAT in humans have been published, and it is now well accepted that most adults have metabolically active BAT 15-17. Rodent studies suggest that activation and/or expansion of energy‐expending BAT is associated with a healthy metabolic phenotype 18-21, and increasing evidence suggests a similar connection in humans. For example, several studies have shown an inverse association between the presence of BAT and obesity and type 2 diabetes mellitus 13, 15, 22, 23. Hence, from having been an obscure gland connected with hibernation at the beginning of last century, BAT is now considered a potential target for therapy to treat obesity and obesity‐related diseases.

Brown adipose tissue – a thermogenic organ When rodents or humans experience cold, thermogenesis is initially mediated by shivering. During prolonged cold exposure, shivering gradually subsides, whilst nonshivering thermogenesis increases 24, 25. Animal studies have shown that this type of thermogenesis depends on BAT 26. Upon cold exposure, efferent signals from the hypothalamus that activate BAT thermogenesis are relayed to the tissue by sympathetic innervation (Fig. 1). Norepinephrine acts as the primary transmitter and activates β‐adrenergic signalling pathways within the brown adipocytes 27. The subsequent induction of lipolysis generates FFAs, the main substrate for heat production in BAT. The FFAs in turn activate uncoupling protein 1 (UCP1), the core protein of the ‘thermogenic engine’ 27, 28, which is only found in the inner mitochondrial membrane of brown adipocytes. Activated UCP1 uncouples oxidative phosphorylation from ATP regeneration by facilitating the reflux of protons across the inner mitochondrial membrane down the proton gradient, thus bypassing ATP synthase 29. It is this ‘short circuit’ of the proton gradient that generates heat. BAT can therefore be seen as capable of transforming energy stored as triglycerides into heat. The rich vascularization of the tissue is essential for supplying the tissue with oxygen and transporting the generated heat to the rest of the body. The importance of a rich blood supply for cold‐induced thermogenesis is illustrated by BAT perfusion more than doubling when human subjects are subjected to acute cold 30. Figure 1 Open in figure viewerPowerPoint Overview of cold‐induced activation of thermogenesis in brown adipose tissue (BAT). In response to cold exposure, afferent signals from cold receptors are integrated in the hypothalamus, which in turn activates BAT via the sympathetic nervous system (SNS). Norepinephrine (NE) stimulates β3‐adrenergic receptors (β3) on the surface of thermogenic adipocytes. The activation of the adrenergic signalling pathway induces lipolysis and expression of thermogenic genes in the adipocytes. Free fatty acids released during lipolysis activate UCP1 and serve as the main fuel for thermogenesis. By short‐circuiting the proton gradient built up over the inner mitochondrial membrane by the electron transport chain, the activated UCP1 uncouples the electromotive force from ATP synthesis by ATP synthase and heat is generated. Apart from the acute effect of increased adrenergic activity, prolonged activity due to persistent cold exposure has additional consequences for BAT, including increased amounts of UCP1, increased mitochondrial biogenesis and both hyperplasia and hypertrophy of the tissue 31, 32. Two important players in these events are peroxisome proliferator‐activated receptor γ (PPARγ) coactivator 1α (PGC‐1α) and type 2 iodothyronine deiodinase (DIO2), which are both up‐regulated and activated by adrenergic signalling 33, 34. PGC‐1α, a transcriptional coactivator of the nuclear receptor PPARγ, is an integral regulator of genes that are involved in mitochondrial biogenesis and oxidative metabolism 35. In addition, PGC‐1α activates transcription of the Ucp1 gene by coactivating nuclear receptors, such as PPARγ and the thyroid hormone receptor (THR), that are assembled on Ucp1‐regulating DNA elements 31. The enzyme DIO2 activates THR by generating its most active ligand triiodothyronine (T3) from thyroxine (T4) locally in the brown adipocytes 34. Hence, the thyroid and sympathetic systems act synergistically to increase the thermogenic potential of BAT by increasing its UCP1 content. The essential role of functional BAT for thermoregulation in small mammals such as rodents is undisputed and illustrated by several mouse models. Mice with reduced BAT mass due to transgenic expression of the cell toxic diphtheria toxin A‐chain in brown adipocytes are cold sensitive, as are Ucp1‐ablated mice 36, 37. Although multiple lines of evidence suggest that BAT is important for thermoregulation in human infants, less is known about its role in adults. However, it is clear that a large proportion of adults retain BAT that can be activated by cold 12, 17, 30. This fact together with BAT apparently being more active during the cold months of the year strongly suggests that the tissue is also involved in thermoregulation in adults 13, 22, 38, 39.

Two types of brown adipocytes Animal studies have shown that apart from the brown adipocytes found in classical BAT depots, brown‐like adipocytes occur within white adipose tissue (WAT) depots in response to chronic cold exposure or administration of β3‐adrenergic agonists or thiazolidinediones (TZDs), a group of PPARγ agonists 40-43. These cells, referred to as inducible brown adipocytes, brite (brown in white) or beige adipocytes, present with a phenotype similar to that of brown adipocytes in classical BAT depots, including multilocular lipid droplets, a high mitochondrial content and expression of Ucp1 and Pgc1α genes. It has also been confirmed that these cells can induce energy‐expending thermogenesis 44, 45. Importantly, despite their similarities, classical brown and beige adipocytes differ in their basal expression of Ucp1. Whilst classical brown adipocytes express Ucp1 at high levels in the basal state, beige adipocytes exhibit much lower basal expression. However, stimulation with β‐adrenergic agonists or activators of PPARγ can induce Ucp1 expression in beige adipocytes to levels comparable to those in classical brown adipocytes 45. Due to white and brown adipocytes sharing many features, the two cell types were until recently thought to stem from a common progenitor cell. However, using a lineage tracing approach, Seale et al. 46 showed that brown adipocytes and skeletal muscle cells, but not white adipocytes, descend from progenitor cells expressing Myf5, the gene encoding myogenic regulatory factor MYF‐5. This close developmental connection had previously been suggested by Atit et al. 47 who demonstrated that dorsal dermis, epaxial skeletal muscle and interscapular BAT (iBAT) derive from En1‐expressing cells of the central dermomyotome. Most importantly, it became apparent that brown and beige adipocytes were developmentally different cell types because, like white adipocytes, the beige adipocytes within the WAT of mice treated with a selective β3‐adrenergic agonist did not derive from Myf5‐expressing progenitors 46. Discrete and unique gene signatures of brown and beige adipocytes provide further evidence that the two entities are distinct cell types 45, 48. Recent studies have contributed important information about the development of beige adipocytes. Utilizing genetic lineage tracing techniques, Lee et al. 49 identified bipotent precursor cells expressing platelet‐derived growth factor receptor (PDGFR)α, Sca‐1 and CD34 in close proximity to blood vessels in WAT. When stimulated by a β3‐adrenergic agonist, these cells developed into beige adipocytes. If the animals were fed a high‐fat diet, however, the cells developed into white adipocytes. The supraclavicular BAT depot of adult humans was recently found to express a gene signature resembling that of beige adipocytes, raising the question of whether humans possess classical brown adipocytes at all 45, 48, 50. The old indication that human infants might have an iBAT depot, a depot known to contain classical brown adipocytes in rodents, led us to test the hypothesis that at least human infants, like other small mammals, have an anatomically distinguishable iBAT depot consisting of classical brown adipocytes8, 51. Postmortem magnetic resonance imaging of human infants indeed revealed tissue in the interscapular region with an intermediate‐fat fraction, as expected for BAT. The histomorphology of the sampled tissue closely resembled that of iBAT depots of rodents, presenting with densely packed multilocular and UCP1‐positive cells delineated from subcutaneous WAT by a layer of connective tissue. Gene expression analysis using previously described marker genes for classical brown and beige adipocytes revealed a gene signature more similar to that expected for classical brown adipocytes. Hence, it appears that at least human infants, like rodents, possess bona fide iBAT consisting of classical brown adipocytes, which suggests that humans actually have both types of thermogenic adipocytes. Additionally, two independent studies characterizing BAT from the neck region of adults support our findings and indicate that classical brown adipocytes are not restricted to infants but also exist in adults 52, 53. The presence of two distinct types of thermogenically competent and energy‐expending adipocytes in humans is conceptually important because they might be stimulated and recruited by different signals; they therefore represent separate potential targets for therapeutic intervention. Additionally, the finding in mice that genetic variability affects the development of beige adipocytes, but not classical brown adipocytes support differential regulation of the two cell types 54.

Important players in the development and function of BAT PR domain zinc finger protein 16 Prdm16, the gene encoding PR domain zinc finger protein 16 (PRDM16), is one of a few murine transcriptional components preferentially expressed in brown versus white adipocytes 55. Overexpression of PRDM16 in fibroblasts or white adipocyte precursors induces a full brown adipocyte gene programme and stimulates both mitochondrial biogenesis and uncoupled cellular respiration. In contrast, a reduction of cellular PRDM16 levels in brown preadipocytes causes a corresponding decrease in the expression of brown adipocyte‐associated genes. In line with this, transgenic overexpression of Prdm16 in WAT of mice increases the expression of brown adipocyte‐associated genes and boosts the number of brown‐like adipocytes (presumably beige adipocytes) in epididymal WAT (a classical WAT depot) after β3‐adrenergic stimulation. In both in vitro and in vivo experiments, the induction of the brown adipocyte phenotype is accompanied by reduced expression of white adipocyte‐associated markers. This dual function of PRDM16 as an inducer of brown adipocyte‐associated genes and a repressor of white adipocyte‐associated genes is mediated through interaction with PGC1α on promoters of brown adipocyte‐associated genes and with corepressing C‐terminal‐binding proteins 1 and 2 (CtBP‐1 and CtBP‐2) on promoters of white adipocyte‐associated genes, respectively 56. Although preferentially expressed in BAT as compared with WAT, Prdm16 was recently shown to also be selectively expressed in subcutaneous WAT depots in contrast to intra‐abdominal WAT depots 57. In addition, inguinal subcutaneous WAT of mice overexpressing Prdm16 in adipose tissue presented with a BAT‐like phenotype even without β3‐adrenergic agonist treatment. However, this alteration was not seen either in the epidydimal WAT depot of the transgenic mice or in the inguinal depot of wild‐type littermates. The importance of Prdm16 for the β3‐adrenergic‐mediated appearance of presumed beige adipocytes in WAT was further highlighted by Prdm16 heterozygous mice developing many fewer clusters of brown‐like adipocytes in their inguinal WAT compared to wild‐type littermates when subjected to a β3‐adrenergic agonist. PRDM16 is also required for the browning of WAT in response to TZDs, and it has been suggested that the effects of TZDs on browning are due in large measure to the stabilization and accumulation of PRDM16 58. Apart from its suggested role in beige adipocyte formation, PRDM16 acts as the molecular switch for directing Myf5‐positive progenitor cells into the brown adipocyte lineage, whilst repressing their development into skeletal muscle cells 46. Interactions between PRDM16 and both PPARγ and the transcription factor C/EBPβ are important in controlling this cell fate switch 46, 59. From the preceding discussion, it is clear that PRDM16 plays a vital role in the formation of both classical brown and beige adipocytes. Hence, this protein, often referred to as a master regulator of brown adipocytes, constitutes a very interesting target for pharmacologic intervention aimed at increasing the presence of thermogenically competent adipocytes. Pgc1α PGC1α was first identified as a protein that interacted with and coactivated PPARγ and THR in brown adipocytes 33. Its expression was shown to be dramatically increased in both BAT of cold‐exposed mice and in adipocytes subjected to a β‐adrenergic agonist. Forced expression of the protein in white adipocytes not only increased the expression of Ucp1 and genes encoding proteins of the respiratory chain but also the mitochondrial DNA content of the cells, suggesting increased mitochondriogenesis. Hence, PGC1α appears to be a factor that could induce a full thermogenic programme in response to adrenergic signalling triggered by a cold environment. In support of this view, PGC1α‐deficient mice have a severely blunted capacity for cold‐induced thermogenesis 60. In addition, PGC1α‐deficient brown adipocytes fail to induce expression of thermogenic genes in response to adrenergic signalling 61. Despite its importance for adaptive thermogenesis, PGC1α does not seem to be a master regulator of BAT formation per se as both iBAT and subcutaneous WAT depots normally containing clusters of beige adipocytes, present with a seemingly normal morphology in PGC1α‐deficient mice 62. The differentiation of brown preadipocytes lacking PGC1α also seems to be normal as they express brown adipocyte marker genes and develop into cells with a characteristic brown adipocyte morphology 61. Hence, PGC1α seems to be an essential regulator of adaptive thermogenesis but not an indispensable factor for determination of brown or beige adipocytes. PRDM16 and PGC1α – players at the crossroad of BAT formation and function Several transcriptional regulators affect BAT formation and function, and the effects of many of them are, at least in part, exerted through modulation of the cellular content or function of either PRDM16 or PGC1α. Euchromatic histone‐lysine N‐methyltransferase (EHMT1), a protein selectively expressed in brown adipocytes, was recently shown to be an integral component of the machinery that induces brown adipocyte cell fate 63. Deletion of Ehmt1 in brown adipocytes leads to the loss of BAT characteristics accompanied by the induction of a skeletal muscle gene programme. In addition, mice with adipose tissue‐specific deletion of Ehmt1 are cold sensitive due to a marked reduction in BAT‐mediated adaptive thermogenesis and fail to develop beige adipocytes in WAT depots in response to β3‐adrenergic stimulation. In contrast, overexpression of EHMT1 in brown adipocytes increases expression of thermogenic genes and raises the oxygen consumption rate. EHMT1 controls brown adipocyte cell fate and BAT thermogenesis by interacting with and stabilizing PRDM16. The NAD‐dependent protein deacetylase sirtuin‐1 (SIRT1) is another protein with the capacity to modulate adaptive thermogenesis. Increased activity of SIRT1 can mimic the action of TZDs and induce browning of subcutaneous WAT depots by facilitating the interaction between PPARγ and PRDM16 by deacetylating PPARγ 64. Several transcriptional regulators have been demonstrated to affect the development and function of thermogenic adipocytes, at least in part, by influencing the activity or cellular levels of PGC1α. One such protein is nuclear receptor‐interacting protein 1 (NRIP1, often referred to as RIP140), which binds directly to PGC1α and inhibits its activity 65. In agreement with this, NRIP1‐deficient mice attain clusters of beige adipocytes in their WAT depots, and NRIP1‐null cells express high levels of UCP1 and have elevated energy expenditure 66, 67. Steroid receptor coactivators (SRCs) belonging to the p160 family have also been shown to affect BAT function by modulating the activity of PGC1α. Whilst the expression levels of thermogenic genes are reduced and adaptive thermogenesis is impaired in Src1‐deficient mice, Src2‐deficient mice display increased expression levels of thermogenic genes and have increased capacity for adaptive thermogenesis 68. Picard et al. 68 showed that SRC‐1 stabilizes the interaction between PGC1α and PPARγ, whereas SRC‐2 inhibits this interaction by competing with SRC‐1 by forming a less active complex with PGC1α. SRC‐3 has also been shown to inhibit the activity of PGC1α 69. SRC‐3 appears to mediate its effects by increasing the expression of Kat2a (also called Gcn5), the gene encoding histone acetyltransferase KAT2A. This enzyme can acetylate PGC1α and thereby inhibit its activity 70. Hence, Src3‐deficient mice have increased expression of thermogenic genes and increased energy expenditure 69. Identification of proteins that influence the activity or cellular levels of PRDM16 and PGC1α is of great importance because such factors represent potential targets for pharmacological intervention aimed at expanding BAT and/or enhancing its function.

The importance of BAT as a metabolic regulator Given BAT's capability to dissipate chemical energy as heat, it is not hard to envision BAT playing a major role in the regulation of metabolism. In 1979, Rothwell and Stock discovered that overfeeding rats with a palatable diet induced increased thermogenesis in BAT 71. This discovery led to the concept of diet‐induced thermogenesis, meaning that activation of BAT in response to caloric excess could reduce metabolic efficiency and avoid or diminish obesity. Consistent with this view, early studies using surgical denervation or transgenic ablation of BAT by overexpression of diphtheria toxin A‐chain specifically in brown adipocytes demonstrated increased body weight and insulin resistance in treated mice 37, 72. As previously mentioned, ablation of the Ucp1 gene in mice leads to reduced cold‐induced thermogenesis. Surprisingly, such mice did not become obese when housed at room temperature 36. However, they did become obese when housed at thermoneutral temperature (29 °C), which suggests an important metabolic role of BAT 73. Although the idea that BAT, in addition to being a source of heat, plays a role in an innate defence against obesity is controversial, it is clear that an increased number or activity of thermogenic cells in the form of brown or beige adipocytes can counteract obesity and insulin resistance. This is illustrated in several mouse models in which the number of such cells has been artificially increased by genetic manipulation of BAT‐regulating genes. As an example, increasing the number of beige adipocytes in WAT by overexpressing UCP1 in the adipose tissue of a mouse strain genetically prone to obesity normalized the phenotype of the animals 74. In line with this, adipose tissue‐specific overexpression of FOXC2, a transcription factor thought to sensitize adrenergic signalling, increased the amount of beige adipocytes in WAT depots and protected mice against diet‐induced obesity, hypertriglyceridaemia and insulin resistance 19, 75. Importantly, the capability of rodents to recruit beige adipocytes within WAT depots upon cold exposure or adrenergic stimulation is subject to genetic variation and varies greatly between different mouse strains; the capacity is lowest in strains prone to obesity and insulin resistance. However, the iBAT depot does not show this genetic variation in rodents 41, 76, 77. Table 1 gives an overview of the metabolic implications of BAT in rodents. Table 1. Importance of brown adipose tissue for metabolism in rodents References Ablation of BAT causes obesity and insulin resistance. 37, 72 Thyroid hormones increase BAT differentiation and activity and are locally regulated by type 2 iodothyronine deiodinase. 87-89 Thermogenic adipocytes are present as classical brown adipocytes in iBAT and as beige adipocytes in WAT after adrenergic stimulation. 40-45 The ability to recruit beige adipocytes in WAT is associated with decreased obesity. 41, 76, 77 Expansion of BAT in transgenic mouse models leads to resistance to obesity and hypertriglyceridaemia. 19, 75 Upon cold exposure, BAT is activated within minutes and its energy demands are met by rapid lipolysis of the intracellular lipid stores 78. To replenish the intracellular triglycerides, BAT takes up FFAs released from the lipids of triglyceride‐rich lipoproteins (TRL) by lipoprotein lipase in the endothelium of its dense vasculature. BAT activation through short‐term cold exposure was recently shown to increase TRL metabolism by controlling vascular lipoprotein homeostasis in mice and ameliorating hyperlipidaemia 79. Recently, metformin was demonstrated to reduce plasma cholesterol and lipid levels in an animal model of human lipoprotein metabolism by increasing BAT activity, leading to increased triglyceride and VLDL uptake through BAT 80. Similar to the effects in exercising muscle, BAT activation leads to dramatically increased glucose uptake by cells in humans (Fig. 2) 30. Although evidence for a potential therapeutic effect in humans is still missing, data from rodents provide robust indications for amelioration of insulin resistance upon physiological or pharmacological stimulation of BAT 81-83. Importantly, several observational studies in humans indicate an association of increased BAT activity and insulin sensitivity as well as reduced obesity 13, 15, 17, 22, 23. Table 2 summarizes relevant findings on BAT in humans. Table 2. Relevance of brown adipose tissue in humans References BAT is present in a majority of adult humans. 12-17 BAT activity is increased in hyperthyroidism. 91 BAT activity can be increased by repeated mild cold exposure. 119 BAT activity is inversely associated with obesity and type 2 diabetes. 13, 15, 17, 22, 23 Severe obesity is associated with less BAT. 110, 111 BAT in adult humans is predominantly of the ‘beige’ type. 45, 48, 50-53 Outdoor temperature is inversely associated with BAT activity. 13, 22, 38, 39 Figure 2 Open in figure viewerPowerPoint 18F]‐fluorodeoxyglucose. The location of the supraclavicular BAT depots are indicated with arrows. (From Virtanen KA, Lidell ME, Orava J, Heglind M, Westergren R, Niemi T, Taittonen M, Laine J, Savisto NJ, Enerbäck S, Nuutila P. Functional brown adipose tissue in healthy adults. Cold‐induced glucose uptake in supraclavicular brown adipose tissue. Computed tomography (CT) and positron emission tomography (PET) images of a subject exposed to cold (PET cold) and room temperature (PET warm), respectively. Two hours prior to the ‘PET cold scans’, the subject was exposed to 17–19 °C temperature whilst wearing light clothing. During imaging one of the subject's feet was occasionally placed in ice water. PET was performed with the tracer [F]‐fluorodeoxyglucose. The location of the supraclavicular BAT depots are indicated with arrows. (From Virtanen KA, Lidell ME, Orava J, Heglind M, Westergren R, Niemi T, Taittonen M, Laine J, Savisto NJ, Enerbäck S, Nuutila P. Functional brown adipose tissue in healthy adults. New England Journal of Medicine 360:1520. Copyright © 2009 Massachusetts Medical Society. Reprinted with permission.) Reprinted with permission.) Taken together, several lines of evidence demonstrate major effects of BAT on metabolism both in animals and in humans. Moreover, BAT could be a potential target for therapeutic intervention to treat metabolic disease in humans.

Endocrine regulation of BAT Several hormones influence both the activity and expansion of BAT. Physiologically, increased activity of the sympathetic nervous system transmitted by norepinephrine acting primarily on β3‐adrenoreceptors activates BAT 27. Accordingly, the role of catecholamines in adrenergic stimulation of BAT activation is well established. Hypersecretion of catecholamines from pheochromocytoma in humans has been noted to lead to the transformation of WAT depots to a BAT phenotype, and an inverse association of abdominal obesity and plasma catecholamine levels in patients with pheochromocytoma has been described 84-86. Thyroid hormones play a pivotal role in the proper function of BAT by modulating adrenergic signalling, and it has long been known that hyperthyroidism increases resting metabolic rate. The hormones are predominantly secreted as T4, which is converted into the biologically more potent T3 in peripheral tissues by deiodinases. The expression of DIO2 is a characteristic feature of brown adipocytes. Upon noradrenergic stimulation, DIO2 expression in BAT increases dramatically, leading to a high local availability of T3 87. Stimulation of THRα1 leads to increased sensitivity of the tissue towards catecholamines, whilst THRβ is crucial for expression of UCP1 in BAT 88, 89. Additionally, expression of PGC1α, the master regulator of BAT function, is increased by T3 via a thyroid hormone response element in the Pgc1α promoter 90. Although these mechanisms are known from animal models, the effect of hyperthyroidism on human BAT was only recently demonstrated by FDG‐PET/CT and indirect calorimetry 91. The important interplay between thyroid hormones and catecholamines is further modulated by bile acids that increase the expression of DIO2 by activating the G‐protein‐coupled bile acid receptor TGR5 92. Specific activation of TGR5 also leads to increased insulin sensitivity in mice 93. In recent years, effects of several peptides on BAT function and development have been described, mostly in animal models. Bone morphogenetic proteins (BMPs) are a family of growth factors that were first described in the context of bone and cartilage development. However, they also play a major role in the morphogenesis of other tissues, and BMPs have recently been implicated in the development and growth of BAT. BMP7 activates brown adipogenesis by inducing early regulators such as PRDM16 and PGC1α. Adenovirus‐mediated overexpression of this cytokine leads to a significant increase in BAT mass and blunts weight gain 94. Subcutaneous infusion of BMP7 increases the amount of beige adipocytes in WAT depots and reduces obesity in mice kept at temperatures below thermoneutrality, indicating that the cold stimulus is still important for activation of expanded BAT depots 95. Another member of the BMP family, BMP8B, regulates BAT activity on both a central and a peripheral level 96. Even though the BAT morphology of BMP8B‐deficient mice appears to be normal, the thermogenic function of the tissue is impaired. Central administration of BMP8B increases sympathetic outflow towards BAT, whilst direct treatment of brown adipocytes with the peptide increases their lipolytic capacity in response to norepinephrine 96. Administration of fibroblast growth factor 21 (FGF21) increases body temperature and BAT thermogenesis in newborn mice and induces expression of thermogenic genes in vitro 97. Additionally, FGF21 seems to be secreted by BAT in an auto‐ or paracrine fashion in response to adrenergic activation upon cold exposure 98, 99. Recently, however, the BAT specificity of FGF21‐stimulated thermogenesis has been questioned 100. Recent evidence suggests that natriuretic peptides (NPs) have the capacity to induce browning of WAT depots 101. The levels of these peptides in plasma increase dramatically in response to cold exposure, and infusion of NPs leads to increased energy expenditure and expression of thermogenic genes in subcutaneous WAT in mice. Mechanistically the effects of NPs are mediated by the NP receptors via the cGMP second messenger pathway and protein kinase G (PKG). The actions of PKG overlap with those of the catecholamine‐stimulated protein kinase A (PKA): activation of hormone‐sensitive lipase and perilipin induces lipolysis and phosphorylation of p38 MAPK, which in turn triggers phosphorylation of PGC1α, finally leading to transcription of Ucp1 101. The signalling pathways of NPs and catecholamines thus seem to converge and act synergistically to activate BAT. Whilst this observation has to be corroborated in humans, it might provide a possible link between heart failure and related cachexia. Recently, Boström et al. 18 described irisin, a novel peptide from muscle that induced a brown adipocyte phenotype when applied to subcutaneous white adipocytes in culture. Adenovirus‐mediated overexpression of irisin in mice led to a beige phenotype in subcutaneous adipose tissue depots. Additionally, increased plasma levels of irisin were detected in humans after muscular exercise. Using the recombinant peptide, others were initially not able to reproduce these effects in vitro 102. However, a recent study by Zhang et al. 103 convincingly showed that recombinant irisin induces browning of adipocytes isolated from subcutaneous WAT of rats. Importantly, the recombinant peptide also induced browning of subcutaneous fat pads in vivo, and protected mice against weight gain and insulin resistance induced by a high‐fat diet. It has been questioned whether the beneficial effects of irisin observed in mice can be translated to humans or not as a mutation (ATG → ATA) present in the start codon of the human FNDC5 gene, that encodes the precursor of irisin, greatly reduces the translation efficiency into full‐length peptide 102. However, Lee et al. 104 recently confirmed circulating irisin in humans using a mass spectrometry approach. The authors also confirmed that exercise increases the serum levels of irisin. In addition, they demonstrated that cold exposure of healthy humans increased circulating irisin levels and that the secretion of the peptide correlated with shivering intensity. Irisin secretion might thus not necessarily be linked to muscular exercise per se but to repetitive small movements as in shivering. Importantly, Lee et al. 104 also showed that the recombinant peptide induced browning and thermogenesis in human neck adipocytes in vitro. Hence, irisin appears to link shivering thermogenesis to an increase in nonshivering thermogenesis. The major endocrine factors influencing BAT are listed in Table 3. Table 3. Key endocrine regulators of BAT function and development Norepinephrine Primary neurotransmitter responsible for acute activation of BAT as well as mitochondrial biogenesis, increased UCP1 expression and tissue hyperplasia. Thyroid hormones Increase local sensitivity of brown adipocytes towards catecholamines and expression of UCP1 and PGC1α; local availability of T3 increased by expression of DIO2 in response to adrenergic stimulation. BMP7/BMP8B Natriuretic peptides FGF21 Irisin These hormones have recently been shown to enhance the induction of beige adipocytes in WAT depots.

Increasing BAT energy expenditure as a therapeutic means Enhancing thermogenesis to facilitate weight reduction is not an entirely new concept. During the 1930s 2,4‐dinitrophenol (DNP), an artificial uncoupler of mitochondrial respiration, was widely popular as a drug for inducing weight loss. This small lipophilic molecule acts as a shuttle for protons across the inner mitochondrial membrane, thereby short‐circuiting the respiratory chain in a manner similar to UCP1, albeit in an unregulated fashion 105. Compared with UCP1, DNP does not selectively act on brown adipocyte mitochondria, and although it was effective in inducing weight loss, it was withdrawn from the market because of a narrow therapeutic range and severe side effects 106. Prior to the discovery of metabolically active BAT in human adults, several studies evaluated the use of β3‐adrenergic agonists to facilitate weight loss in humans. Although the selective β3‐adrenergic agonist L‐796568 acutely increased energy expenditure in obese men, it failed to induce weight loss or long‐term increased energy expenditure 107, 108. Administration of another β3‐adrenergic agonist, CL 316,243, initially led to increased insulin sensitivity and fat oxidation, but it also failed to alter energy expenditure after 8 weeks of use; this was related to an unexpected decline in the plasma levels of the drug, indicating that the metabolism of the drug might have accelerated over time 109. However, several other factors may explain why long‐term treatment with β3‐adrenergic agonists has been unsuccessful. Chronic stimulation of adrenergic receptors commonly leads to down‐regulation of the target receptors over the course of a few days, a phenomenon that is referred to as tachyphylaxis, which is well known in clinical application of catecholamine derivatives. Stimulation of BAT by exogenous catecholamine analogues might also reduce the endogenous activation of the tissue. Furthermore, the prevalence and activity of targetable BAT appears to be lower in obese persons 110, 111. It should be pointed out that the human studies were carried out before the presence of thermogenically active BAT in adult humans was demonstrated and the tissue's activity was thus not measured. Given the increased knowledge about BAT in humans and the much improved methods of its assessment, it is still worthwhile to pursue developing new β3‐adrenergic agonists with enhanced specificity, oral bioavailability and more constant plasma half‐life over time, as such drugs very well could give the desired effects in humans. Stimulation of BAT via the sympathetic nervous system is thought to contribute to the weight loss effects from appetite‐reducing drugs such as sibutramine or ephedrine 112-114. These centrally acting drugs can cause serious side effects such as cardiac arrhythmias or hypertension, however, and have therefore been withdrawn from the market. A more selective way to increase BAT thermogenesis might be to target local conversion of T4 to T3 by stimulation of the G‐protein‐coupled bile acid receptor TGR5, thereby inducing expression of DIO2, or direct targeting of THRβ with selective agonists 115-117. Pharmacologic interventions that activate and expand BAT would provide a very attractive means for weight reduction, especially in individuals unable to exercise. However, physiological interventions such as intermittent cold stimuli or a reduction of ambient temperature could also provide a safe and affordable alternative, analogous to increasing energy expenditure through muscular exercise. In this respect, it is important to notice that human BAT is activated by mild cold stimuli such as ambient temperatures around 16 °C or by placing a hand in cold water at around 20 °C 14, 118. Accordingly, recent preclinical research demonstrated an increase in BAT activity and cold‐induced thermogenesis as well as a reduced amount of body fat after 6 weeks of repeated short‐term mild cold exposure (17 °C for 2 h) in healthy men with normal weight 119. Whether these effects translate to reduced obesity or increased insulin sensitivity needs to be evaluated in future studies.

Conclusion From the literature, it is clear that expansion or increased activity of BAT in rodents is associated with a metabolically healthy phenotype. Despite this intriguing association in rodents, BAT was until recently not seen as a potential target for anti‐obesity drugs in humans because adults, the primary age group that would use such drugs, were not believed to possess significant amounts of metabolically active BAT. Hence, BAT was perceived as a conceptually interesting target without promise for therapeutic use. With the discovery of metabolically active BAT in adults in 2009, however, the tissue has become the subject of intense research. Over the last 5 years, increasing numbers of BAT‐related studies have been published that confirm that the association between BAT and a metabolically healthy phenotype also holds true for humans. During the last decade, our knowledge about factors influencing BAT recruitment and function in rodents has increased substantially. Hopefully, this new knowledge can be extrapolated to humans and provide a foundation for studies that will identify and evaluate potential molecular drug targets in the coming years.

Acknowledgements The work was supported by grants from the Swedish Research Council (2012‐1652 and 2010‐3281), the Knut and Alice Wallenberg Foundation, the Sahlgrenska University Hospital (LUA‐ALF), the European Union (HEALTH‐F2‐2011‐278373; DIABAT), the IngaBritt and Arne Lundgren Foundation, the Söderberg Foundation and the King Gustaf V and Queen Victoria Freemason Foundation.

Conflict of interest statement Sven Enerbäck is shareholder and consultant to Ember Therapeutics.