Of mice and men: novel insights regarding constitutive and recruitable brown adipocytes

Abstract

Recently, there has been great attention given to the possibility of combating obesity by targeting brown fat activity or increasing differentiation of brown adipocytes in white fat depots through a process termed ‘browning'. Sympathetic innervation of brown and white adipose tissues provides adrenergic input that drives thermogenesis and regulates fatty acid metabolism, as well as stimulating adipogenesis of recruitable brown adipocyte tissue (rBAT, also known as beige or brite) in white fat. Other factors acting in an endocrine or autocrine/paracrine manner in adipose tissue may also stimulate browning. There have been significant recent advances in understanding the mechanisms of increasing adipose tissue energy expenditure, as well as how brown adipocytes appear in white fat depots, including via de novo adipogenesis from tissue precursor cells. In this article, we integrate this new knowledge with a historical perspective on the discovery of ‘browning'. We also provide an overview of constitutive BAT vs rBAT in mouse and human.

Obesity represents a major risk factor for the development of several of our most common medical conditions, including type 2 diabetes mellitus, dyslipidemia, non-alcoholic fatty liver, cardiovascular disease and even some cancers.¹ Increased adiposity is the main characteristic of obesity; however, not all fat depots are created solely for energy storage.² Adipocytes found in white adipose tissue (WAT) contain a single large lipid droplet and have well-characterized roles in fuel storage and immune-endocrine functions. By contrast, brown adipose tissue (BAT) is specialized for energy expenditure. Adipocytes in BAT contain many small, multilocular lipid droplets, and are tightly packed with mitochondria. In addition, BAT is highly vascularized and densely innervated by the sympathetic nervous system (SNS). BAT uniquely expresses uncoupling protein 1 (UCP1), which is localized to the inner mitochondrial membrane, and acts to uncouple oxidative phosphorylation from ATP production, resulting in the electron gradient being dissipated as heat in a process termed thermogenesis.^{3, 4, 5, 6} In response to cold, or other stimuli such as diet or activation of β3-adrenergic receptors (ADRB3), thermogenesis is activated as a result of increased sympathetic input to BAT.^{7, 8}

In addition to thermogenesis, recent studies have demonstrated that BAT is involved in triglyceride clearance⁹ and glucose disposal,¹⁰ and is a source of adipokines (which we call ‘BATokines' for BAT adipokines⁸), including FGF21,^{11, 12} Irisin/FNDC5^{(refs. 13, 14)} and interleukin-6^{(ref. 15)}; (reviewed in Villarroya et al.¹⁶). Following the rediscovery of functionally active BAT in adult humans,^{17, 18, 19, 20, 21, 22, 23} brown fat has become an exciting area for obesity research. Given BAT's immense capacity for energy expenditure and its newly recognized effects on fatty acid and glucose metabolism, there is great hope that BAT's energetic capacity may be tapped by various medical interventions as a means to increase whole-body energy expenditure and reduce adiposity.^{8, 24, 25} Undoubtedly, improved knowledge about the regulation of brown fat formation, activation and communication with the central nervous system (CNS) are required to make such therapeutic approaches conceivable.

Although interscapular BAT is the major brown fat depot in mice and is constitutively expressed (constitutive BAT, cBAT), multilocular and UCP1-positive brown fat cells can be found in different anatomical locations as well. These ‘inducible' or ‘recruitable' brown fat cells (recruitable brown adipose tissue, rBAT),²⁶ also known as ‘beige'²⁷ or ‘brite'²⁸ adipocytes, are found to be highly enriched in WAT and skeletal muscle in obesity-resistant strains of mice.^{29, 30, 31} Physiological stimuli, such as cold exposure and sympathetic activation, are also known to induce brown adipogenesis in white fat depots.³² This phenomenon, known as ‘browning', has been described in WAT, especially in the subcutaneous depot and after cold exposure or CNS manipulations that increased sympathetic outflow to these white fat depots.^{33, 34} Similarly, when mice are given chronic daily injections of CL 316,243, an ADBR3 agonist, subcutaneous WAT also undergoes ‘browning' and body temperature rises.³⁵ In addition to SNS input, cBAT sends sensory nerve afferents to the CNS, which also are involved in the regulation of thermogenesis.³⁶

Browning and cBAT vs rBAT: historical perspectives and current findings

The process of browning of white fat was first described in the 1984 FEBS paper by Young et al.,³⁷ who wrote ‘…during a preliminary study of the effects of cold acclimation on female BALB/c mice, we noticed some brown areas in the perigenital fat pad. Since this observation was contrary to the popular view of distinct white and brown fat regions, we decided to set up this study so that morphological and biochemical criteria could be used to differentiate between white and brown adipocytes…' These brown areas were close to blood vessels and had UCP1 (assessed by radioimmunoassay). The cells were centrally nucleated with multilocular lipid droplets and a high concentration of mitochondria (measured by electron microscopy), and the tissue displayed high activities of mitochondrial enzymes.³⁷ This finding was confirmed in 1986 using a cat model,³⁸ and was followed up by several other observations,^{39, 40, 41} including studies in animals after treatment with ADBR3 agonists.^{42, 43}

More recently, browning has been described as being induced by numerous factors,^{27, 44} including cardiac-derived natriuretic peptides,⁴⁵ action of central and peripheral SIRT1,^{46, 47} central brain-derived neurotrophic factor action, as well as brain-derived neurotrophic factor induced by animals being housed in an enriched environment,⁴⁸ central and possibly peripheral orexin action,⁴⁹ muscle-derived irisin,¹³ heart-derived natriuretic peptides,⁴⁵ liver- and BAT-derived FGF21^{50, 51} and bone morphogenetic proteins (BMPs)\transforming growth factor beta (TGFβ),^{35, 52, 53, 54, 55} among others. Other triggers shown to induce browning are as follows: overexpression of perilipin in WAT,⁵⁶ addition of a PGC1α adenovirus to human white preadipocytes,⁵⁷ prolactin receptor knockout,⁵⁸ exercise,⁵⁹ knockdown of neuropeptide Y in the dorsomedial hypothalamus⁶⁰ and vascular endothelial growth factor-A overexpression in WAT.⁶¹ New signaling pathways have been implicated in the process of brown adipogenesis in WAT, including retinaldehyde dehydrogenase,⁶² 4E-BP1,⁶³ Rb,⁶⁴ RIP140,⁶⁵ LiverXRa,⁶⁶ FoxC2,⁶⁷ TIF2,⁶⁸ p107,⁶⁹ TNFαR⁷⁰ and others. MicroRNAs (miRs) have also been linked with the control of brown adipogenesis, such as mir-193b-365,⁷¹ mir-196a,⁷² mir-133,⁷³ miR-106b-93^{(ref. 74)} and mir-155.⁷⁵ Likely, these miRs have a role in regulating gene expression of transcription factors involved in controlling the brown adipocyte fate during cellular differentiation (reviewed in Trajkovski and Lodish⁷⁶), and could also represent an aspect of epigenetic regulation of brown adipogenesis.

Whether the browning observed in the studies outlined above is the result of de novo adipogenesis of precursor cells into brown adipocytes and whether mature white adipocytes may transdifferentiate directly into mature brown adipocytes are theories still up for debate. Transdifferentiation is characterized by the presence of paucilocular UCP1+ adipocytes appearing in white fat,^{77, 78, 79} and has been described as early as 1966 by Hull and Segall.⁸⁰ They suggested that white and brown adipocytes are essentially two forms of the same tissue, potentially representing a metabolic flexibility to either produce heat or store lipids, depending on the needs of the body.⁸⁰ In support of this notion, UCP1-positive brown adipocytes were found in inguinal WAT between 10 and 21 days of age in mice maintained at room temperature, but these cells disappeared by 60 days of age.⁸¹ Interestingly, these cells could re-emanate upon cold exposure or could be suppressed by undernutrition from birth to 21 days post natal, suggesting a high degree of plasticity. A recent lineage-tracing study purported to reveal that mature adipocytes are capable of making a bi-directional switch: from mature white to mature brown adipocytes during cold exposure, and back again during re-warming.⁸² However, the cells pictured retain some multilocular morphology and low leptin expression even after re-warming; thus, the cell type with apparent plasticity in response to changes in environmental temperature may represent a distinct ‘recruitable' cell that does not, in fact, return to a truly white, unilocular adipocyte state with low levels of UCP1 and few mitochondria. Furthermore, it is interesting that these transdifferentiating cells occur in pockets, which may represent their close proximity to both sympathetic innervation and vascular supply, in order to mediate their flexibility in gene expression and thermogenic potential. It is certainly likely that sympathetic neurites undergo their own plasticity in response to numerous bouts of cold exposure, and may innervate and activate a slightly different group of precursor cells with each bout.

On the other hand, numerous studies have demonstrated that the appearance of brown adipocytes in WAT depots is the result of de novo adipogenesis. Precursor cells expressing stem cell antigen-1 and platelet-derived growth factor receptor-α in the abdominal fat pad similarly exhibit the dual potential to differentiate into either rBAT or WAT.^{35, 83} Interestingly, the stem cell antigen-1-positive cells isolated from cBAT and WAT express distinct molecular signatures and respond differently to inductive cues.³⁵ Of the PDGFRα cells in WAT, a CD24+ population was identified that loses expression of CD24, as it more fully commits to the adipocyte lineage.⁸⁴ Subtypes of precursor cells that can differentiate into brown adipocytes from white fat depots have been identified, including those expressing CD137, Tmem26 and Tbx1,⁸⁵ but these cells have not been directly monitored in vivo as they undergo adipogenesis. A recent publication utilizing the AdipoChaser mouse, a doxycycline-inducible, mature adipocyte-specific tracing system, developed to answer the question of whether rBAT arises from a lineage distinct from white adipocytes.⁸⁶ The AdipoChaser mouse was pulse-chased to indelibly label mature adipocytes, enabling the discovery that most of the rBATs appear in subcutaneous WAT in response to cold or treatment with β3-adrenergic agonists as the result of de novo adipogenesis, rather than preexisting white adipocytes.

Brown adipogenesis also occurs in situations with a reduction in autophagy in Myf5+ precursor cells (with a deletion of Atg7, a key gene for autophagy), which then increases rBAT.⁸⁷ This is intriguing, given that autophagy is a situation responsive to nutrient status and is involved in lipid metabolism (reviewed in Christian et al.⁸⁸). Targeting Myf5+ adipocyte precursors is significant, given the findings in mice that the Myf5+ lineage contributes to the development of the interscapular, or cBAT, whereas Myf5− cells largely comprise white fat depots and the rBAT compartment.⁸⁹ Deletion of the type 1 bone morphogenetic protein receptor BMPR1A from Myf5+ cells results in a severe paucity of cBAT with a resulting compensation of rBAT in WAT depots.⁹⁰ As a result, the knockout mice could maintain proper body temperature as adults and are resistant to high-fat diet-induced obesity. These data highlight the existence of a physiological system for thermoregulation and energy homeostasis by modulating total BAT-mediated thermogenic capacity. Mechanistically, this is likely due to feedback from the impaired cBAT to brain, and a resulting increase of SNS input to WAT in order to drive brown adipogenesis and rBAT development. Alternatively, the cross talk between cBAT and rBAT could be mediated by secreted factors presumably produced by residual cBAT in the Myf5-BMPR1A knockout mice. The understanding of fat depot cross talk is in its infancy, although it is known that sympathetic denervation of one WAT depot will influence norepinephrine turnover in the intact depots of WAT and BAT, but will not affect depot weights.⁹¹ Likely, this occurs via CNS integration of sensory afferents from adipose depots.⁹² More research is needed to better understand how adipose depots communicate with each other via CNS and SNS intermediaries.

Interestingly, ablation of the PI3K signaling inhibitor phosphatase and tensin homolog in Myf5+ precursors leads to a remodeling of adipose tissues in the body with increased adiposity of WAT and BAT, especially in the neck and shoulder regions.⁹³ As a result, the cBAT in the interscapular region was more lipid laden. This study also revealed that a subset of cells in white fat depots, such as interscapular WAT and retroperitoneal WAT, in fact comprises a mix of Myf5− and Myf5+ precursors.⁹³ Given the difficulty in comparing WAT and BAT depots across different studies, owing to variations in nomenclature and anatomical dissections, as well as differences conferred by mouse strain (reviewed in Yadav and Rane⁹⁴), it will likely be important for the field to develop a recognized characterization of mouse fat depots. Such an advance would accommodate our growing knowledge about the differences among white fat depots and their relative response to browning, including new insights stemming from lineage-tracing studies (reviewed in Sanchez-Gurmaches and Guertin⁹⁵).

Several new concepts have arisen in the study of rBAT, including whether the brown adipocytes interspersed between muscle fibers also contribute to anti-obesity effects,^{31, 35} and whether preadipocytes directly sense cold temperature and thereby turn on a thermogenic program or differentiate into brown adipocytes. A recent study demonstrated that 3T3-F442A cells exposed to cold temperatures in vitro exhibit a mild increase in UCP1 gene expression.⁹⁶ It is interesting to note that single-celled organisms such as the bacterium Escherichia coli can directly sense and respond to environmental temperature.⁹⁷ Such behavior suggests the possibility that CNS–SNS-independent pathways may exist for cell autonomous temperature sensation.

Human vs mouse BAT

To date, there have been few studies in humans to identify non-cold-temperature means of activating BAT or increasing its mass, but a multitude of rodent studies have revealed many potential options for this purpose. Although the function of BAT appears similar between rodents and humans, making mice a useful model species, there are a few important distinctions. Rodents, such as laboratory mice, have a different distribution of BAT vs humans (Figure 1), which in mice includes a large, discrete interscapular pad that is similar to what is observed in human babies.^{98, 99} In adult humans, BAT is mainly clustered around the neck, clavicle and spinal cord.^{18, 19, 21} Mice contain other smaller depots of BAT, such as around the kidneys, cervical spine and heart,¹⁰⁰ as well as the rBAT found in white fat and skeletal muscle.¹⁰¹ Humans also possess various small BAT depots including around organs such as kidney and heart, and in other subcutaneous depots such as under the arms.^{98, 102} Whether or not humans can boost their available BAT mass was previously unclear, but two recent studies have demonstrated that cold acclimation in humans, after repeated daily cold exposure, elevates BAT volume and activity, and increases energy expenditure.^{103, 104}

Figure 1.

Mouse vs human BAT. Anatomical and other similarities, and differences between mouse and human BAT.

In additional, mice are often housed at a room temperature that is not thermoneutrality for them, forcing them to undergo thermogenesis at room temperature in order to maintain their body temperature. Such an adaptation renders their BAT chronically activated. Humans, on the other hand, live at thermoneutrality in addition to heating their homes and wearing clothing, which likely maintains BAT at a lower level of activation.

Another difference between rodent and human BAT has to do with the second population of brown adipocytes, the inducible or rBAT (brite or beige) as described above, which can be found in white fat depots or in the muscle of mice. These cells develop from a different lineage than the cBAT found in the interscapular region (reviewed in Townsend and Tseng⁸), and are recruitable in the sense that SNS stimulation (in the form of β-adrenergic agonists or cold exposure) or activation of signaling pathways (such as those reviewed in the sections above) can induce the presence of these brown adipocytes. It is believed that these cells can then heighten the energy expenditure of the whole animal. On the other hand, human BAT is now believed to be comprised of either cBAT (as is found in human babies' interscapular region⁹⁹ and the neck,¹⁰⁵ or supraclavicular region¹⁰⁶ of adults), or human BAT comprises rBAT (as is found in the clavicular region⁸⁵ and the retroperitoneal, intra-abdominal and other regions.¹⁰⁷) The constitutive and recruitable types of brown adipocytes display a different gene expression signature,^{85, 100} but whether there is a functional difference in addition to a different lineage for these two types of adipocytes remains to be determined.

BAT mass in rodents is ~0.4–1% of body weight (Townsend and Tseng, unpublished dissection data, and Geisler¹⁰⁸), whereas in humans it is estimated at ~0.02% of body weight,¹⁰⁸ and thus humans have relatively less BAT than rodents. Small mammals may utilize BAT to oxidize up to 90% of daily fuel intake; BAT has been shown to account for the majority uptake of ingested glucose and the most cold-stimulated triglyceride clearance.^{109, 110} Lipogenesis in BAT may account for 40% of the total in cold-exposed rats.¹¹¹ In humans, it has been estimated that a few grams of BAT may increase daily energy expenditure from 6 to 20%.^{112, 113} Regardless of the differences in the amount or anatomical locations of brown fat between rodents and humans, both human and rodent BAT possesses great capacity for thermogenesis, and serves as an important site for glucose and fatty acid metabolism (described above and Ouellet et al.¹¹⁴).

Although adipose tissues express numerous subtypes of the adrenergic receptor, ADBR3 has a more limited distribution in mice and is most highly expressed in BAT and WAT, thereby making it the most likely adrenergic receptor isoform to mediate sympathetic effects on energy expenditure in murine adipose tissues.¹¹⁵ On the other hand, humans have more widespread expression of ADBR3, including in adipose tissues, urinary bladder, smooth muscle and gut.¹¹⁶ In humans, the blockade of β-adrenergic receptors by propranolol before cold exposure does not inhibit cold-induced thermogenesis, suggesting that skeletal muscle uncoupling downstream of the β2-receptor may be the culprit.¹¹⁷ In mice, the triple β-receptor knockout is cold intolerant and obese.^{118, 119} It was recently demonstrated that the β1-receptor in mice mediates most of the cold- and diet-induced thermogenesis,¹²⁰ further supporting the involvement of multiple adrenergic receptors in regulating adipose tissue energy expenditure in response to catecholamine release from sympathetic nerve terminals.

In rodents, ADRB3 agonists effectively activate BAT, leading to weight loss and improved insulin sensitivity.^{43, 121, 122} However, the effect of these compounds in humans appears negligible, which has been attributed to reduced ligand-binding ability or bioavailability to human ADRB3.^{123, 124} (After submission of the original manuscript, a new study demonstrated that mirabegron, a highly specific β3-adrenergic receptor agonist, can stimulate human brown fat thermogenesis (Cypess et al., Cell Metab 2015; 21: 33–38)) Furthermore, the cross-reactivity of these agonists to β1 or β2 adrenergic receptors may lead to unwanted cardiovascular side effects in humans.

Concluding remarks and future perspectives

New knowledge about the presence and activity of human BAT represents an exciting opportunity to exploit physiological pathways for increasing thermogenesis and energy expenditure as a potential therapeutic target for human obesity. Using mouse models, several novel pathways regulating BAT and WAT energy expenditure have been identified. The ability to utilize human subjects to obtain a better understanding of BAT function and activity is also improving. Considering the similarities and differences between mouse and human BAT, together these studies should provide important translational work to enable development of new therapeutics targeting BAT for the treatment of obesity and other metabolic diseases. The field is moving toward identification of cell types and molecular pathways mediating the development of rBAT in WAT depots. Such studies could determine whether these cells are identical in thermogenic function to cBAT,¹²⁵ or whether different fat depots give rise to different subtypes of rBAT cells through mechanisms such as de novo adipogenesis or transdifferentiation. Finally, it remains to be resolved whether increasing rBAT will alone suffice to increase energy expenditure and restore metabolic health to obese humans, but given the data outlined here, the prospect is a promising one.