Development, Activation, and Therapeutic Potential of Thermogenic Adipocytes

Abstract

During the last decade, significant progress has been made in understanding adipocytes with a particular focus on thermogenic fat cells, which effectively convert chemical energy into heat in addition to their other metabolic functions. It has been increasingly recognized that different types and subtypes of adipocytes exist and the developmental origins of various types of fat cells are being intensively investigated. Previous work using immortalized fat cell lines has established an intricate transcriptional network that regulates adipocyte function. Recent work has illustrated how these key transcriptional components mediate thermogenic activation in fat cells. Last but not least, cumulative evidence supports an incontestable role of thermogenic fat in influencing systemic metabolism in humans. Here we summarize the exciting advancements in our understanding of thermogenic fat, discuss the advantages and limitations of the experimental tools currently available, and explore the future directions of this fast-evolving field.

Introduction

The rediscovery of thermogenic adipocytes in human adults introduced a fresh target to investigate and modulate in hopes of leveraging the balance of energy homeostasis towards expenditure, therefore possibly presenting new opportunities counteract human obesity. While pursuing the potential metabolic benefits of these thermogenic fat cells, researchers in this area have investigated many aspects of the basic biology of adipocytes and the last ten years has witnessed a riveting renaissance of all things fat [1].

Broadly speaking, chemical energy storing, unilocular white adipocytes and heat producing, multilocular thermogenic fat cells constitute the majority of the fat in mammals. Emerging evidence additionally suggests that adipocytes in the bone marrow are distinctive in many aspects from fat found in the rest of the body and are often considered in a separate category of its own [2]. Among all the thermogenic fat cells, what we now call classical brown fat cells arise from a skeletal muscle like origin and localize within a number of specific anatomical locations [3]. In comparison, the exact definition of the inducible beige adipocyte is continuously evolving as investigation moves forward. In this mini review, what we refer to as beige adipocytes can be loosely defined as the thermogenic fat cells residing within white adipose tissue that do not share the same developmental origin as the classical brown adipocytes.

Thermogenesis, similar to any other energy consuming process, has evolved to be tightly controlled so that no unnecessary energy loss can occur. Signaling cascades that regulate thermogenic fat activation have recently been thoroughly reviewed elsewhere [4]. Here we will focus our discussion on the developmental origin, transcriptional control, and therapeutic potential of both brown and beige adipocytes.

Developmental origins of thermogenic fat

It has been more than ten years since transcription co-component PR domain containing 16 (PRDM16) was identified as one of the key regulators for brown fat differentiation and a loss of function study revealed that instead of turning into white fat cells as expected, PRDM16 knockdown causes primary brown preadipocytes to differentiate into myocytes [3, 5]. Cell fate mapping experiments with skeletal muscle marker Myf5 driven cre mediated reporter expression uncovered the surprising shared lineage between skeletal muscle and interscapular brown fat [3]. This skeletal muscle-like origin for brown adipocytes has been validated by subsequent studies using other skeletal muscle markers [6–8]. The finding that the adrenergic-induced thermogenic cells within the subcutaneous inguinal fat depot arise from a different developmental origin than brown fat led to the study of clonal cell lines derived from this depot, which found that beige fat arises from a subpopulation of distinct progenitor cells [3, 9].

Since then, further evidence has emerged showing that cold exposure induces de novo adipogenesis of beige adipocytes. The “Adipo-Chaser” mouse, which pulse labels mature adipocytes with lacZ, found that a large population of non-lacZ labeled multilocular adipocytes (newly differentiated beige fat cells) emerge in the white fat depots after cold exposure [10]. The developmental origin(s) of these inducible beige fat cells have been intensively investigated. Using a Ucp1-TRAP system to specifically analyze polysomes from the Ucp1-expressing cells in white fat depots showed that these cells had enriched expression of multiple genes associated with smooth muscle [11]. Further investigation using a myosin heavy chain 11 (Myh11)-cre driven GFP reporter found that many UCP1+ cells in the inguinal depot arose from Myh11+ progenitor cells [11]. This potential smooth muscle like origin of beige fat cells was also implicated by fate mapping studies using reporter mice that label cells from the SMA lineage (smooth muscle actin, also called ACTA2, actin alpha 2) [8, 12]. Vascular smooth muscle cells and pericytes, generally referred to as mural cells, derive from the mesenchyme and give rise to vessels. Mural cells expressing platelet derived growth factor receptor β (PDGFRβ) have been shown to differentiate into beige adipocytes in the inguinal depot in response to long-term cold exposure [13]. Interestingly, capillary cells from human adipose tissue explants have been shown to be able to differentiate into thermogenic adipocytes, further suggesting that vascular progenitors and beige adipocytes may share a common origin [14]. In addition to studies of what may lead to the formation and activation of inducible beige adipocytes, recent work has started to inspect how the deactivation of beige fat may be regulated. A swift decrease of thermogenic activity was observed upon stimuli withdrawal in beige fat, in contrast to classical brown fat [15]. This is consistent with the notion that beige adipocytes with “white-like” morphology exist in an inactivated state in the absence of thermogenic stimuli and could constitute at least one of the potential mechanisms for “interconversion” and “transdifferentiation” observed in previous investigations [16, 17].

Transcriptional control of the activation of thermogenesis

A great deal of work has been done to characterize the transcriptional network responsible for thermogenic fat activation. Much of this regulation has emanated from interactions between peroxisome proliferator-activated receptor gamma (PPARγ) and two key transcriptional co-components: PRDM16, and peroxisome proliferator-activated receptor gamma coactivator 1 alpha (PGC1α) (Figure 1).

Figure 1. PRDM16 and PGC1α interact with PPARγ to activate the thermogenic program.

Both the transcription of Prdm16, regulated by PPARγ and EBF2, as well as the function of PRDM16 as it forms transcriptional complexes with PPARγ and C/EBPβ, are responsible for driving thermogenic activation. Similarly, the transcriptional complexes that PGC1α forms with PPARγ and IRF4 mediate activation of the thermogenic program.

PPARγ is the master regulator of adipocyte differentiation and function [1] and chronic PPARγ agonism by drugs such as the thiazolidinedione rosiglitazone during adipocyte development results in an increased capacity for thermogenesis in these cells [18–20]. While the mechanisms by which PPARγ promotes thermogenic development are continuously being investigated, a number of posttranslational modifications of PPARγ have been suggested to play a role in the upregulation of the thermogenic program during adipocyte differentiation [20–22]. PPARγ both drives the transcription of and works alongside PRDM16, a transcriptional co-component that plays a pivotal role in regulating brown fat cell fate and beige fat function [3, 5, 23, 24]. Unlike the drastic cell type switch from brown adipocytes to myotubes seen when PRDM16 is knocked down in culture, deletion of PRDM16 in the brown fat lineage by Myf5-cre in vivo compromised brown fat function during aging but did not affect brown fat in early development, likely due to compensation from PRDM3 [25]. Despite arising from a separate lineage that PRDM16 may not affect, inducible beige adipocytes are functionally responsive to this transcriptional co-component. Beiging due to rosiglitazone treatment has been shown to be at least partially mediated through PRDM16 stabilization [20]. Gain and loss of function of PRDM16 in adipocytes increases and decreases the abundance of thermogenic beige adipocytes in subcutaneous adipose tissue, respectively [23, 24]. It was shown that PRDM16 transcriptional activity works in part by making a complex with CCAAT/enhancer binding protein β (C/EBPβ) [26]. Accordingly, thermogenic fat function can be affected by micro RNAs that regulate C/EBPβ, such as mir-155 [27] or miR-196a [28].

The regulation of Prdm16 transcription plays a substantial role in the maintenance of the thermogenic program. Early b-cell factor 2 (EBF2) is a DNA binding protein that has been reported to direct PPARγ towards transcription of a brown adipocyte program, including upregulation of Prdm16 transcription [29]. Blnc1 is a long noncoding RNA that has been shown to interact with EBF2 to promote thermogenic adipocyte differentiation [30]. Blnc1, as well as the long noncoding RNA lnc-BATE1, were shown to upregulate thermogenesis via interactions with the structural protein heterogeneous nuclear ribonucleoprotein U (hnRNPU) [31, 32]. A screen to find transcription factors that induce thermogenic adipocyte differentiation identified ZBTB7B, which was shown to interact with Blnc1 and hnRNPU to upregulate the expression of thermogenic genes [32]. ZFP423, in addition to its role in regulating adipocyte commitment [33], directly binds to EBF2 and inhibits browning through suppression of Prdm16 transcription [34]. Other suppressors of Prdm16 transcription also inhibit thermogenesis, including miR-27b and miR-133 which both inhibit thermogenesis through targeting of the 3′-untranslated region (UTR) of Prdm16 [35–38]. A number of proteins activate or inhibit thermogenesis through interaction with PRDM16. Activators include euchromatic histone-lysine N-methyltransferase 1 (EHMT1) [25, 39], ZFP516 [40], and lysine-specific demethylase 1 (LSD1) [41, 42], which also interacts with ZFP516 to promote thermogenic gene transcription [43]. Repressors include transducin like enhancer of split 3 (TLE3), a transcription factor that promotes white adipogenesis through competition with PRDM16 to bind promoters of genes associated with lipid storage and thermogenesis [44, 45].

PGC1α was initially identified as a transcriptional cofactor interacting with PPARγ with enriched expression in brown fat compared to white fat [46]. Further research has confirmed that it plays a pivotal role in mitochondrial biogenesis and oxidative metabolism through interactions with various nuclear receptors [47]. PGC1α and related protein PGC1β have been shown to be crucial in the regulation of mitochondrial biogenesis in activated thermogenic fat [48–50]. Pgc1a transcription is induced by CREB and ATF2 [51, 52] and upregulation and stabilization of Pgc1a mRNA activates the thermogenic program. Interferon regulatory factor 4 (IRF4) regulates thermogenic gene transcription via direct interaction with PGC1α, and thermogenesis induced by PGC1α overexpression in fat is blunted in the absence of IRF4 [53]. Conversely, factors that inhibit PGC1α activity decrease thermogenesis. Twist Family BHLH Transcription Factor 1 (TWIST1) is a transcriptional regulator that interacts with PGC1α and suppresses the transcription of its target genes, possibly through a negative feedback mechanism [54]. Knockout of the related proteins RB and p107 were found to increase thermogenesis in adipocytes through alleviated repression of PGC1α [55, 56]. Similarly, RIP140 interacts with PGC1α, repressing transcription of thermogenic genes such as cell death-inducing DFFA-like effector A (Cidea) and Ucp1 [57, 58]. Finally, inhibition of miR-34a robustly increases browning of white adipocytes by enhancing SIRT1-mediated PGC1α deacetylation, resulting in an increase of thermogenic gene transcription [59].

Therapeutic potential of thermogenic fat

Better understanding of the development and activation of thermogenic adipocytes can help to build the foundation for potentially harnessing the metabolic benefits of these fat cells in humans. Since 2009, when multiple groups discovered activatable thermogenic fat in the supraclavicular and neck regions of humans [60–63], a great deal of research has been done to characterize human thermogenic fat. These initial ground-breaking discoveries were accompanied by both the excitement that these metabolically active cells may help to counteract human obesity as well as caution concerning whether the amount and activity of these cells in an average human could constitute a palpable influence on systemic metabolism. Much of the hard work since then has addressed these concerns and evaluated the therapeutic potential of thermogenic fat in various human studies (Table 1). It has been shown that in healthy individuals acute cold exposure can improve glucose homeostasis and insulin sensitivity [64] and that varying lengths of chronic cold exposure can increase insulin sensitivity [65] and decrease obesity [66]. It has additionally been shown that a 10 day cold exposure protocol can increase insulin sensitivity in obese diabetic patients [67]. Cold activates thermogenic fat through the central nervous system, which signals to target tissues through the β-adrenergic pathway [1]. After early mixed results [68–71], recent work has shown that the β3-adrenergic receptor agonist mirabegron can stimulate human thermogenic fat [72]. Ongoing efforts have revealed that other pathways, such as stimulation with capsinoids [66, 73], chenodeoxycholic acid (CDCA) [74], or glucocorticoids [75], can all promote the activity of human thermogenic fat.

Table 1.

Recent advancements in human thermogenic fat study

in vivo
Detection and measurement	PET/CT (18F-FDG, 18F-FTHA, 15O)	[60–63, 84–87]
	Nonionizing (Contrast enhanced ultrasound, Infrared thermography, MRI, Near infrared spectroscopy)	[86, 88, 90–94]
Stimulation of thermogenic fat	Cold exposure (Improves glucose homeostasis, Decreases body fat)	[64–67]
	β-adrenergic stimulation (Mirabegron, Ephedrine)	[70, 72]
	Other drugs (Capsinoids, CDCA, Glucocorticoids)	[66, 73–75]
in vitro
Supraclavicular and Neck fat	(Forskolin, BMP7, Cortisol, db-cAMP, Norepinephrine, FGF21, Isoproterenol)	[75–77, 95–98]
Subcutaneous fat	(Rosiglitazone, Mirabegron, BMP4/BMP7, Natriuretic peptides, Cinnamaldehyde)	[14, 101–104]
Multipotent adipose derived stem (hMADS) cells	(Rosiglitazone, Natriuretic peptides, ß-adrenergic receptor agonists, Arachidonic acid)	[103, 105–109]
iPS cells	(Lactate, Forskolin, cAMP, Isoproterenol, JAK inhibition)	[110–114]
Fetal stem cells	Fetal interscapular fat (Norepinephrine)	[116, 117]
	Fetal mesenchymal stem cells (Rosiglitazone)	[115]

To further investigate the therapeutic potential of these cells, studies have sought to develop tools to more effectively study their activation and function. Clonal cell lines were derived from human neck fat and studied using RNA sequencing and using a Ucp1 reporter system, identifying multiple novel markers for human thermogenic adipocytes [76, 77]. Murine systems that measure substrate uptake and thermogenic activity in vivo allow for investigation into fundamental questions regarding the physiological function of the activated thermogenic program. It has recently been reported that long chain fatty acids (LCFAs) are required for UCP1 function and can be used as a substrate for thermogenesis [78]. Injection of a luciferin-conjugate LCFA, which is cleaved by glutathione and activated upon uptake into cells, allows for real-time detection of β-adrenergic agonist induced fatty acid uptake increase in the brown fat of mice expressing luciferase, presenting an imaging approach that allows investigation of metabolic regulation and pathological alterations of thermogenic fat [79]. This method was applied to Ucp1-cre driven luciferase-expressing mice and directly demonstrated that lipokine 12, 13-diHOME promotes fatty acid uptake in brown fat [80]. A number of mice have additionally been generated with reporter constructs for Ucp1 expression, the ‘ThermoMouse’ which expresses luciferase under the control of the Ucp1 promoter region [81] and the ‘Ucp1-2A luciferase knock-in mouse’ which contains luciferase appended to the end of the Ucp1 gene [82] can both faithfully recapitulate Ucp1 regulation in vivo and can be used for future exploration to uncover regulatory signaling or drug compounds targeting thermogenic fat.

Investigation so far indicates that not only the activity level but also the actual amount of activatable thermogenic fat may vary from individual to individual [83], methods that can accurately measure activatable thermogenic fat can therefore be used to better personalize therapeutic strategies by identifying likely responders. Using radiolabeled substrates for thermogenesis as surrogates, studies have been carried out using positron emission tomography (PET) scans with either ¹⁸F-FDG or the fatty acid tracer ¹⁸F-FHTA, combined with computed tomography (CT) scanning to measure activated thermogenic fat [60–63, 84, 85]. Similarly, PET scans using ¹⁵O to measure oxygen consumption have also been adapted to measure the activity of thermogenic fat [86, 87]. Nonionizing techniques have also been applied to evaluate thermogenic fat content and activity in humans. Contrast enhanced ultrasound can be used to measure the increased blood flow to thermogenic fat tissue following cold exposure [88]. Similar to infrared imaging of mice [89], measurements of surface temperature in the supraclavicular region using infrared thermography can detect human thermogenic fat activity [90, 91]. Water-fat separated magnetic resonance imaging (MRI) was adapted to distinguish thermogenic and white fat based on contrasting tissue characteristics instead of measurement of thermogenic activity, by taking advantage of the fact that thermogenic fat is more densely vascularized than white fat [92, 93]. A recent study with a hybrid FDG-PET/MR scanner demonstrates that MRI when used in combination with PET may present a radiation-free alternative to CT [94]. Continuous efforts to develop accurate, non-invasive methods to detect and monitor thermogenic fat in humans are warranted to enable longitudinal studies in infants, children, and healthy adults in the future.

While thermogenic activity of supraclavicular and neck fat has been extensively studied in humans [75–77, 95–98], an increasing number of studies indicate that activatable thermogenic fat may very well exist elsewhere in the human body. Seasonal cold exposure has been shown to induce thermogenic gene expression in subcutaneous white adipose tissue [99] and long-term adrenergic stimulation as a result of burn injuries can thermogenically activate subcutaneous fat from multiple depots [100]. Studies of human subcutaneous adipocytes in culture have revealed that these cells respond to thermogenic stimuli, such as PPARγ agonist rosiglitazone [101], bone morphogenic proteins (BMP) 4 and 7 [102], natriuretic peptides [103], and the food compound cinnamaldehyde [104]. It is also worth noting that many human adipose cell lines have been used to investigate thermogenic regulation, including human multipotent adipose derived stem (hMADS) cells [103, 105–109], human induced pluripotent stem (iPS) cells [110–114], and a number of cell lines derived from fetal tissue [115–117]. It is unlikely that quantifiable changes in actual thermogenic capacity (e.g. respiration) can be measured in all of the above mentioned in vitro systems, given that the actual expression levels of thermogenic genes are fairly low in some of these cell lines. However, it is tempting to speculate that gene expression based assays could potentially be carried out in many commonly available human fat cell systems, at least as an initial proof-of-principle investigation.

Future directions

Most of the research done on both murine and human thermogenic fat thus far has focused on UCP1-mediated thermogenesis, however, recent work indicates that a deeper understanding of non-shivering thermogenesis and a broader definition of beige adipocytes are warranted. Creatine cycling, which enhances respiration in wild type mitochondria when there are limiting amounts of ADP, has been observed in adipocytes from UCP1 KO mice, suggesting that this constitutes a mechanism for UCP1 independent thermogenesis [118]. This creatine-dependent futile cycle was further confirmed using patch clamp analysis of mitochondria isolated from visceral adipocytes after thermogenic stimulation. This study found that the majority of epididymal mitochondria were UCP1 negative and that thermogenic capacity was achieved through creatine cycling in these abdominal UCP1-negative “beige” adipocytes [119]. Furthermore, mice with an adipocyte specific knockout of glycine amidinotransferase (GATM), the rate limiting enzyme in creatine synthesis, have an increased tendency towards obesity, demonstrating the systemic influence of adipose creatine-cycling [120]. UCP1-independent thermogenesis can also be mediated through N-acyl amino acids, endogenous uncouplers of mitochondrial respiration in brown and beige adipocytes [121]. Since the discovery of thermogenic fat in adult humans in 2009 there has been an enormous amount of research on the development, function, and therapeutic use of these cells. It is apparent now more than ever that many new targets are yet to be discovered that can help to materialize the potential of thermogenic adipocytes in influencing human physiology in health and disease.

Highlights.

• Different types of adipocytes arise from various developmental origins

• Thermogenic fat activation is regulated through a complex transcriptional network

• Stimulation of human thermogenic fat improves metabolic fitness