Brown adipose tissue during puberty and with aging

Abstract

It was previously assumed that brown adipose tissue (BAT) is present in humans only for a short period following birth, the time in which mechanisms of generating heat by way of shivering are not yet developed. Although BAT is maximally recruited in early infancy, findings in recent years have led to a new consensus that metabolically active BAT remains present in most children and many adult humans. Evidence to date supports a slow and steady decline in BAT activity throughout life, with the exception of an intriguing spike in the prevalence and volume of BAT around the time of puberty that remains poorly understood. Because BAT activity is more commonly observed in individuals with a lower body mass index, an association seen in both adult and pediatric populations, there is the exciting possibility that BAT is protective against childhood and adult obesity. Indeed, the function and metabolic relevance of human BAT is currently an area of vigorous research. The goal of this review is to summarize what is currently known about changes that occur in BAT during various stages of life, with a particular emphasis on puberty and aging.

Key words::

• Aging
• beige fat
• browning
• metabolism

Key messages

• Throughout human life, there is a gradual loss of brown adipocytes throughout the body.

• Puberty, particularly later stages, is associated with transient increases in brown adipose tissue (BAT) that are presently poorly understood but coincident with gains in skeletal muscle.

• Age-related changes in BAT correlate with signs of metabolic deterioration, suggesting BAT may be able to provide mechanistic insight into aging and associated metabolic disease.

Adipose tissue is no longer viewed as a simple storage depot, but is instead appreciated as a highly dynamic and metabolically active tissue that plays an essential role in co-ordinated regulation of energy balance. The amount of fat found in the human body is astoundingly broad, ranging from 2% to 70% of total body weight (1), and varies significantly not only across individuals but also within one individual's life-span. This wide range in adiposity largely reflects quantitative differences in white adipose tissue (WAT), the most predominant form of fat. A related subtype of fat called brown adipose tissue (BAT) also varies between people and throughout life, but generally contrary to variations in WAT.

WAT and BAT share a unique capacity for processing lipid, but in the context of metabolism they serve opposing functions. WAT is comprised of white adipocytes specialized for storing large amounts of triglyceride, generally in the form of one large lipid droplet, and secreting various endocrine factors. In contrast, BAT contains cells uniquely designed for lipid oxidation as a means of generating heat. Brown adipocytes have abundant mitochondria and multiple smaller lipid droplets, a form of storage that facilitates lipid–lipase interactions, and are defined by the production of uncoupling protein-1 (UCP1), a mitochondrial pore that, upon stimulation, uncouples adenosine triphosphate generation from oxidative phosphorylation, thereby allowing energy to be dissipated as heat. Generating heat in this way, or non-shivering thermogenesis, is subject to complex co-ordinated regulation in response to cold or diet (reviewed by Cannon and Nedergaard (2)), and involves factors and pathways that include sympathetic innervation, catecholamines, thyroid hormones, growth factors, and natriuretic peptides (3).

It was formerly thought that human BAT is present only during infancy and then disappears shortly after birth as alternative ways of generating heat (e.g. shivering) are established and non-shivering thermogenesis is no longer required for survival (4). However, metabolically active BAT has now been conclusively shown in adult humans (5–9). BAT can be broadly viewed as a diverse set of fat deposits that contain both brown and white adipocytes, the proportion of which differs based on anatomical location (10). More discrete depots of BAT are found in the cervical-supraclavicular region, or between the shoulder-blades, as well as in cervical, axillary, mediastinal, paravertebral, perirenal, and peri-aortic regions (11). However, brown adipocytes are also found interspersed within deposits of WAT, often referred to as beige, brite (brown within white), or recruitable BAT, based on an ability to be activated by cold or sympathetic/adrenergic stimuli (12,13). Anatomically distinct brown adipocytes are all functionally thermogenic (14), but animal data suggest ‘canonical’ BAT shares a precursor with skeletal muscle, whereas beige fat does not (15). For BAT found in the neck region of adult humans, there is currently evidence of both brown and beige features (16–18).

Underscoring the relevance of BAT to humans throughout life, studies suggest BAT activity correlates with favorable metabolic features (19) and is inversely related to body mass index (BMI) (5,6,8,9,20). Similar associations have also been observed in pediatric populations (21), suggesting the intriguing possibility that BAT is protective against childhood and adult obesity and related metabolic dysfunction. While many questions remain regarding the metabolic relevance of human BAT, the aim of this review is to summarize what is currently known about the evolution of BAT throughout the life-cycle, with a particular focus on how BAT changes with puberty and during aging.

Neonates

Human BAT is formed during gestation and maximally recruited at birth, and the transition from fetal to neonatal life represents a rapid and remarkable shift from a quiescent thermoregulatory state to one of maximal rate of heat production (22). Studies have shown that the most metabolically active form of BAT in human infants is the large bilateral supraclavicular depot (23). Indeed, defining features of this depot include extremely rapid activation and a very high capacity for heat production (24). Evidence from autopsy studies suggests additional more deeply situated BAT in the mediastinal and abdominal regions (25), as well as in the axillary and deep cervical regions (26) of human infants. BAT can account for as much as 5% of total body weight at birth (27), and is a critical means of thermoregulation, as newborn humans do not yet possess the ability to shiver.

Highlighting the degree to which thermogenesis is activated in newborn infants, pronounced depletion of BAT lipid stores is apparent within just a few days of life (25). Further, when newborn infants are exposed to cooler temperatures, BAT likely mediates a rapid doubling of oxygen consumption rates (28). The capacity for stimulated BAT thermogenesis is highly dependent on UCP1 content, and BAT from pre-term and stillborn infants has significantly less UCP1 than older infants, consistent with premature infants having reduced thermogenic capacity and stillborn infants not having been exposed to cold temperatures (26). BAT activity remains extremely high for the first 8 weeks of life, and subsequently declines into childhood (29).

Childhood and adolescence

As with infancy, adolescent BAT is located predominantly in the supraclavicular region (21), but whereas BAT is present in virtually every infant, metabolically active BAT is detectable at room temperature in 20%–75% of children and adolescents, with a similar frequency observed in young boys and young girls (21,30–32). While this highlights an age-dependent decline in BAT activity that begins very early in life (Figure 1), emerging evidence suggests there may be a transient increase in the prevalence and volume of BAT around the time of puberty (21,30,32), adding provocative support to BAT having a function outside of being a substitute for undeveloped shivering mechanisms.

Figure 1. Human BAT incidence as a function of age. Detection of metabolically active BAT in the neck region of humans is illustrated with incidence (y-axis) as a function of age (x-axis). Non-stimulated BAT (dashed line) and cold-stimulated BAT incidence (solid line) are summarized as general trends based on data derived from studies discussed in the text.

Initial evidence of a puberty-associated increase in BAT activity was reported by Gelfand and colleagues in 2005. These investigators retrospectively examined BAT incidence using [¹⁸F]2-fluoro-2-deoxyglucose (FDG)-based positron emission tomography (FDG-PET) scans performed in pediatric cancer patients and found that children over the age of 10 years had a higher incidence of BAT compared to those under the age of 10 years (30). In a similar retrospective study by Drubach et al., analysis of FDG-PET scans from oncology patients between the ages of 1 and 21 years revealed an increase in BAT incidence from childhood to adolescence, with the highest incidence of BAT observed around the age of 13 years (21). These studies collectively support a putative increase in adolescent BAT activity around the time of puberty. Because both studies analyzed BAT in subjects at room temperature, observations may suggest that puberty positively modulates basal BAT activity or reduces the degree of cold required to activate BAT.

In 2012, Gilsanz and colleagues analyzed BAT at different stages of sexual development and found that pediatric BAT prevalence and volume are highly dependent on age and level of sexual maturity, and, conversely, independent of BMI, sex, or season. In this retrospective study, BAT was examined in disease-free patients who underwent FDG-PET combined with X-ray computed tomography (PET/CT) as a follow-up for a previous malignancy (32). Although the presence or absence of malignant disease reportedly has no effect on the ability to detect BAT by PET/CT (33–35), it is worth noting that this study detected BAT in 59% of the tumor-free patients (32), whereas in 2011 a prevalence of 42% was observed using scans from many of the same pediatric patients at a time in which tumors were present (31). Within the overall incidence of 59%, Gilsanz et al. found that prevalence in pre-pubertal children (Tanner stage 1) was only 15%, whereas in older subjects (Tanner stages 2–5) it was 75%. They report a positive association between BAT incidence and degree of sexual development in all five Tanner stages of puberty, but find the relationship is most predominant during Tanner stages 4–5, or late puberty. This study similarly showed that the volume of BAT increases with age and degree of sexual development and particularly during Tanner stages 4–5 (32). Taken together, emerging correlative evidence supports the intriguing possibility of a pubertal increase in BAT activity at room temperature (Figure 1).

From retrospective FDG-PET/CT studies done by Gilsanz et al. in 2011, it was reported that BAT-positive boys and girls have on average 30%–50% greater muscle volume compared to pediatric patients without identifiable BAT (31). The authors extended these findings in 2012, showing skeletal muscle and BAT display correlative increases in volume throughout puberty, with both preferentially increasing later during sexual development (Tanner stages 4–5) (32). In another study, investigators found that pediatric BAT volume is positively associated with appendicular bone mass, independent of other bone determinants such as height, weight, and gender, in disease-free patients with detectable BAT (36). Collectively, these findings suggest there is a transient increase in BAT activity during adolescence that is coincident with sexual maturation and associated musculoskeletal development (32).

Although findings to date are based exclusively on retrospective analyses of clinically warranted FDG-PET/CT scans, and therefore lack supporting functional data in human or relevant animal models, synchronized growth of BAT and skeletal muscle during puberty is consistent with reports that these two tissues are developmentally related (15). Further, many systemic hormones that promote pubertal musculoskeletal development, such as growth hormone, gonadal sex steroids, insulin, or insulin-like growth factor 1 (IGF-1) (37), may also favor growth of BAT (38–40).

Another interesting possibility is that BAT and skeletal muscle modulate the activity of one another more directly. As described recently by Ponrartana et al., circumstances associated with counterintuitive integrity of musculoskeletal development share a defining feature of considerable BAT abundance. Such contexts include the absence of mechanical loading in relatively motionless infants that simultaneously progress through a period of rapid musculoskeletal development, and hibernating mammals that maintain muscle mass while remaining completely immobile for nearly 7 months (41). These observations imply more direct lines of communication between BAT and the musculoskeletal system.

Indeed, there is accumulating evidence that BAT may have an endocrine function analogous to that of WAT (42). Recent findings suggest fibroblast growth factor 21 is a cold-stimulated human brown adipokine (43,44), and additional BAT-derived factors, such as IL-6 (45) and IGF-1 (46), are implicated in amelioration of diabetes that results from BAT transplantation in rodents. BAT could therefore theoretically have an anabolic effect on skeletal muscle by secreting IGF-1 or other relevant factors.

Alternatively, the musculoskeletal system might release factors that modulate the development of BAT. Circulating factors secreted by skeletal muscle include irisin, which has been shown to positively regulate brown adipogenesis (47). There is presently very little known about irisin during adolescence, but one report suggests circulating concentrations of irisin are higher in young girls compared to young boys and negatively associated with fasting blood glucose levels (48). It would be interesting in future studies to determine whether circulating concentrations of irisin vary according to stage of sexual maturity.

Although there are presently more questions than answers regarding adolescent BAT, in part because invasive aspects of detection have precluded the possibility of prospective study to date, recent observations suggest puberty may favorably influence BAT activity. The observation that occurrence of BAT during adolescence is inversely associated with fat accumulation during subsequent years (49) supports metabolic relevance, but whether the putative increases in BAT observed during this period meaningfully alter energy expenditure or other metabolic indices is unknown. As less invasive and more sensitive imaging techniques emerge, a more detailed understanding of this transient re-emergence of BAT coincident with puberty should be revealed. This will be important, as a better understanding of mechanisms that govern endogenous BAT reactivation could facilitate efforts to reactivate BAT pharmaceutically as a strategy for treating obesity and metabolic disease.

Adulthood and aging

The anatomical distribution of BAT in adults is similar to that observed in adolescents (21), with the most prominent and metabolically active depot in the cervical-supraclavicular region and smaller depots located in the axillary, mediastinal, paravertebral, epicardial, and abdominal areas (reviewed in (24)). The transition from adolescence into adulthood is associated with reduced ability to detect unstimulated, basal BAT activity, as studies collectively suggest ∼5% of adults have detectable BAT under thermoneutral conditions, whereas BAT is detected in 15%–50% of patients under the age of 18 under similar conditions (Table I) (50–52). As with pediatric populations, the presence of unstimulated adult BAT is associated with favorable metabolic features (19).

Table I. Brown adipose tissue prevalence in pediatric and adult populations.

	Study (reference)	Type	Cohort age (years)/diagnoses	n	Detection (temperature)	Prevalence
Pediatric	Drubach et al., 2011 (21)	Retrospective	5–21/malignancy	172	FDG-PET (21°C)	44%
	Gelfand et al., 2005 (30)	Retrospective	1–22/malignancy	118	FDG-PET (RT)	46%
	Gilsanz et al., 2011 (31)	Retrospective	6–20/malignancy	71	FDG-PET/CT (22°C)	42%
	Gilsanz et al., 2012 (32)	Retrospective	4–20/previous malignancy	73	FDG-PET/CT (22°C)	59%
	Zukotynski et al., 2010 (53)	Retrospective	< 22/malignancy	99	FDG-PET (21°C)	27%
				103	FDG-PET (24°C)	9%
Adult	Au-Yong et al., 2009 (33)	Retrospective	13–88/cancer staging	3614	FDG-PET/CT (RT)	5%
	Cohade et al., 2003 (50)	Retrospective	1–18/cancer staging	21	FDG-PET/CT (RT)	24%
			19–93/cancer staging	884	FDG-PET/CT (RT)	6%
	Cypess et al., 2009 (5)	Retrospective	30–80/various diagnoses	1972	FDG-PET/CT (RT)	5%
	Lee et al., 2010 (34)	Retrospective	18–87/cancer staging	2934	FDG-PET/CT (21°C)	9%
	Persichetti et al., 2013 (35)	Retrospective	2–94/cancer staging	6454	FDG-PET/CT (22°C)	3%
	Stefan et al., 2009 (58)	Retrospective	2–94/cancer staging	3604	FDG-PET/CT (22°C)	3%
	Truong et al., 2004 (51)	Retrospective	5–17/cancer staging	8	FDG-PET/CT (RT)	50%
			18–90/cancer staging	837	FDG-PET/CT (RT)	1%
	Yeung et al., 2003 (52)	Retrospective	2–17/cancer staging	26	FDG-PET/CT (RT)	15%
			18–88/cancer staging	837	FDG-PET/CT (RT)	2%
	Zingaretti et al., 2009 (8)	Prospective	18–82/thyroid disease	35	UCP1 content (RT)	33%
	Saito et al., 2009 (9)	Prospective	23–35/healthy volunteers	31	FDG-PET/CT (19°C, ice)	52%
			38–65/healthy volunteers	24	FDG-PET/CT (19°C, ice)	8%
	van Marken Lichtenbelt et al., 2009 (6)	Prospective	18–32/healthy volunteers	24	FDG-PET/CT (16°C)	96%
				3	FDG-PET/CT (22°C)	0%
	Virtanen et al., 2009 (7)	Prospective	20–50/healthy volunteers	5	FDG-PET/CT (18°C, ice)	100%
	Yoneshiro et al., 2011 (59)	Prospective	20–49/healthy volunteers	147	FDG-PET/CT (19°C, ice)	45%
			50–73/healthy volunteers	15	FDG-PET/CT (19°C, ice)	7%

CT = computed tomography; FDG = ¹⁸F-Fluorodeoxyglucose; PET = positron emission tomography; RT = room temperature.

Because BAT is activated by cold temperatures, the true prevalence of BAT cannot be derived from retrospective analysis of FDG-PET/CT scans performed at room temperature. Cold-stimulation studies have not been performed in children, but evidence suggests prevalence of cold-stimulated BAT in adults could be as high as ∼100% (7), at least for individuals under the age of 40 years (6). Interestingly, BAT prevalence observed in pediatric patients at 24°C (53) approaches the range observed in adults at 21°C (Table I). More detailed studies are needed, but this may suggest that, at a given temperature, a higher degree of BAT activation occurs in children compared to adults.

In addition to attenuated activity, the capacity for heat production in supraclavicular BAT, as well as the rapidity with which it can be activated, declines into adulthood, perhaps reflecting BAT that becomes increasingly ‘beige’ as a function of age (54). Indeed, autopsy studies performed in 1972 showed that the widest distribution and greatest activation of brown fat occurs in the first decade of life and that many, if not all, anatomical fat deposits subsequently lose brown-like characteristics over the next eight decades (55). Consistent with the primary function of BAT being to generate heat, these studies suggest peripheral depots (e.g. interscapular) are the first to lose brown fat with age, whereas more deeply situated depots, such as those surrounding the kidneys or aorta, retain brown-like features as far as the eighth decade of life. Interestingly, loss of BAT may plateau around the sixth decade of life, followed by a subsequent decline into later decades (55), which may help explain why elderly individuals lose the ability to regulate body temperature effectively (56) and are often unable to tolerate cold temperatures (57).

Findings from a number of more recent studies using FDG-PET/CT to visualize BAT in live subjects (Table I) further support a marked decline in BAT activity with age (5,9,20,33,35,58,59). In 2009, Cypess and colleagues retrospectively analyzed ∼2000 adult scans that were clinically warranted for various reasons and found that non-stimulated BAT could be detected in individuals under the age of 50 years three times more often than in individuals over the age of 64 years (5). Similar analyses using FDG-PET/CT scans performed for cancer staging showed that, within BAT-positive cohorts, age is associated with reduced intensity of BAT and fewer BAT-positive depots, with evidence of a particularly, steep decline in the later years (33,35). Studies also suggest age-associated loss of non-stimulated BAT activity may be more pronounced in males (20) and overweight individuals (5). Although a variety of uncontrolled factors such as clothing and climate might affect the ability to detect BAT activity at room temperature, general trends for non-stimulated BAT incidence as a function of age are summarized in Figure 1 (dashed line).

There is also evidence that cold-stimulated BAT activity is attenuated with age (9,59). Saito and colleagues used FDG-PET/CT to visualize BAT in healthy volunteers under warm and cold conditions and found cold-activated BAT could be detected in 52% of younger subjects (aged 23–35 years) compared to only 8% of older subjects (aged 38–65 years) (9). Similarly, Yoneshiro et al. observed cold-induced activity of supraclavicular BAT in more than half of subjects in their 20s, but only 7% of subjects in their 50s and 60s (Table I) (59). These studies collectively suggest cold-stimulated BAT activity may drop precipitously around middle age and become undetectable with advancing age. Cold-stimulated BAT incidence as a function of age is summarized in Figure 1 (solid line), with caveats that include limited or absent cold-stimulation studies performed in elderly and pediatric subjects, respectively.

Importantly, because FDG-PET/CT studies quantify only BAT that can be activated, findings to date support age-related deficits in BAT activity, which may or may not reflect changes in mass. Functional deficits are consistent with findings from rodents, as the interscapular BAT depot in rats becomes enlarged with age, likely due to white adipocyte infiltration, but this is coincident with a significant decline in UCP1 activity (60) and loss of function (61). A striking age-dependent loss of UCP1 is also observed in rodent subcutaneous WAT (62), consistent with the morphological evidence derived from human autopsy studies that looked at fat deposits throughout the body (55). Likewise, fat found in human bone marrow, sometimes referred to as yellow adipose tissue (YAT) based on an increased number of mitochondria and other BAT-like features, loses BAT-like characteristics with age (63). It therefore seems that many if not all forms of BAT are gradually transitioned towards WAT with increasing age.

Underlying mechanisms that promote loss of BAT with age remain elusive. Although mechanisms are likely to be complex and synergistic, a few of the pathways that have been implicated are discussed briefly below. Data regarding human BAT are currently limited and largely correlative. While caution should be used before extrapolating findings to humans, drawing on the animal literature will be incorporated in an effort to provide what are likely to be relevant mechanistic clues.

Because diminished proliferative capacity and cellular senescence often accompany aging, age-related loss of BAT might occur as a result of changes to the brown adipocyte precursor pools. Evidence currently supports the presence of precursors during aging, at least for some period of time, as brown adipocyte progenitors have been successfully isolated from donors as old as 78 years (64). However, the possibility of functional deficits cannot be ruled out. A detailed characterization of human brown adipocyte-specific progenitor function during aging remains an important future study, but age-dependent declines in the capacity of white adipocyte precursors to proliferate and differentiate are well documented (reviewed in (65)). Further, rodent BAT displays a diminished proliferative response with age, and, while mechanisms are not fully elucidated, this has been attributed to altered availability of trophic factors (66).

Indeed, there are a number of age-related changes in the somatotropic and gonadotropic axes that could directly or indirectly lead to BAT dysfunction. Aging is associated with a decline in the production of growth hormone, estrogens, and androgens, all of which may positively modulate activity of BAT (38–40). Moreover, a significant correlation between thermogenic parameters of BAT during aging and serum levels of sex steroid hormones has been shown in rodents (67). Changes in concentrations of estrogens might be particularly relevant, as high expression of estrogen receptor in human fetal BAT (68) suggests tissue-level regulation.

Based on reports suggesting glucocorticoids have negative effects (69) whereas estrogens and androgens may have positive effects on BAT (38,39), Nedergaard and Cannon hypothesized an ‘endocrine switch’ whereby a decrease in sex hormones in late adulthood leads to disinhibited actions of glucocorticoids and, consequently, loss of BAT activity (70). Additional hormones implicated in the beneficial actions of estrogens on BAT include thyroid hormone (67) and leptin, based on evidence that estradiol is required for hypothalamic actions of leptin (71) that include regulation of BAT (72). The importance of testosterone to BAT function is less clear. Androgen-deficient mice are obese, and while there is evidence of decreased UCP1 expression (39) there are also reports that BAT is unaltered in these mice (73), as well as orchidectomized rats (74). Further, when testosterone is administered to male mice, BAT mass, protein content, and mitochondrial mass are reportedly unchanged, suggesting testosterone modulates body weight without altering BAT thermogenesis (75).

Thyroid hormone is an appreciated regulator of thermogenesis (76), and human aging is associated with a decrease in serum triiodothyronine (77) and reduced conversion of inactive thyroxine (T4) to active triiodothyronine (T3) (78), a reaction that requires type II iodothyronine deiodinase (DIO2). Likewise, thermogenic parameters in rodent BAT correlate with circulating levels of T3 during aging (67), and age-dependent loss of DIO2 expression is evident in murine WAT (62). Observations such as regulation of DIO2 activity by estrogen (79) suggest significant cross-talk between the thyroid and gonadotropic axes in modulating BAT activity, which could mean that age-dependent alterations to these two systems are collectively more detrimental.

Finally, one of the most critical regulatory systems controlling the activity of BAT is the sympathetic nervous system (SNS). The SNS controls local catecholamine concentrations, and catecholamines in turn act on adrenergic receptors located on the surface of adipocytes to modulate brown adipogenesis and thermogenesis (2). A decline in SNS input and/or impaired downstream signaling could therefore result in a marked decrease in BAT activity. Current evidence argues against an age-dependent loss of SNS stimulation, as a number of studies show that basal as well as stress-induced plasma noradrenaline levels are increased, not decreased, with age in humans (80,81). Likewise, neural recordings done in rodents suggest sympathetic signaling is elevated with age (82), and findings from BAT norepinephrine turnover rate studies do not support blunted sympathetic stimulation to BAT as a cause of declining thermogenic capacity with age (83).

What seems clear from rodent studies is that, although neural signals are intact, thermogenesis increases to a smaller degree with a given sympathetic stimulus with increasing age (84,85). This may be due in part to a decrease in adrenergic receptor density in interscapular BAT, as this is observed in rats with age (86). Age-related down-regulation of adrenergic receptor expression is also observed in rodent subcutaneous WAT, in conjunction with pronounced loss of UCP1 expression and additional intracellular changes, such as enhanced expression of enzymes that degrade catecholamines, that collectively support an age-dependent loss of local adrenergic tone (62).

Consistent with lower basal adrenergic tone, humans with low room temperature BAT activity can still recruit BAT in the cold (87). Similarly, older rodents display a marked loss of beige fat under normal conditions but maintain the ability to recruit beige fat in response exogenous adrenergic stimuli (62). These findings collectively implicate age-related resistance to actions of catecholamines as a contributing factor in associated loss of BAT activity. However, there are also findings that suggest non-SNS pathways may play a major role in BAT activation, including lack of BAT activation in humans in response to pharmacologic activation of the SNS (88) and evidence that fat cells can directly sense cold temperature and activate thermogenesis (89).

In the end, aging is associated with a ubiquitous decline in BAT activity, but details remain obscure, and particularly those specific to loss of human BAT activity. A few candidate mechanisms have been discussed, but it is likely that a myriad of changes coincident with aging may collectively promote a definitive loss of BAT function over time. Because loss of BAT is associated with various pathologies that include metabolic dysfunction and aberrant temperature regulation, a more thorough understanding of mechanisms that facilitate loss as a consequence of age could be beneficial to future efforts to prolong human health span. It will also be critical to elucidate whether there is a point in which BAT is no longer dormant, but instead absent, as humans age, so that age groups most likely to benefit from future therapies that target BAT can be identified.

BAT and age-related metabolic disease

There is a clear correlation between risk factors for metabolic disease (i.e. BMI, age) and reduced BAT activity (5,6,8,9,20), and loss of metabolically active BAT is associated with accumulation of body fat that occurs specifically during aging (59). However, while WAT and BAT are consistently inversely related (5,6,8,9), whether decreases in BAT cause subsequent adiposity or increases in WAT lead to dysfunctional BAT remains less clear.

In the former scenario, a decline in BAT activity that is otherwise programmed into aging could reduce energy expenditure and thereby provide a basis for over-accumulation of WAT. Evidence for a programmed loss of brown adipocytes is derived from rodent caloric restriction studies that show weight-gain is not required for age-associated loss of subcutaneous beige fat (62). The notion that activation of human BAT may increase energy expenditure is supported by extrapolated calculations (7), as well as more recent studies showing that activation of human BAT via cold exposure is associated with measurable increases in energy expenditure (87,90,91) and reduced adiposity (87). However, there are also other reports suggesting that cold-simulated thermogenesis in human BAT accounts for as little as 15 kcal/day, which argues little contribution of BAT to whole-body energy expenditure (92).

Alternatively, it is possible that an increase in WAT precedes an involution of BAT. Increased WAT stores provide more insulation, which might dictate a decrease in heat-generating mechanisms such as non-shivering thermogenesis. Further, because loss of BAT activity is associated with the infiltration of white adipocytes, ectopic accumulation of WAT within BAT could itself lead to dysfunction, as detrimental effects of ectopic lipid are well-documented in other metabolic tissues such as liver and skeletal muscle (93). Supporting the importance of changes in WAT, a reduction in WAT accumulation through caloric restriction can attenuate age-dependent loss of mitochondrial function in canonical rodent BAT (61).

Importantly, BAT activity is favorably associated with a variety of metabolic parameters outside of BMI such as insulin secretion (94), glucoregulation (35,95), lipoprotein metabolism (96), triglyceride clearance (97), and circulating concentrations of HDL (19), all of which are conversely unfavorably associated with aging. Indeed, accumulating evidence suggests BAT is protective against metabolic disease. In mice, subcutaneous beige fat decreases dramatically with age, and this decline is associated with age-dependent loss of glucose homeostasis (62,98). Similarly, obese humans with diabetes have lower levels of UCP1 in subcutaneous WAT than similarly obese but diabetes-free individuals (99), suggesting subcutaneous beige fat is protective against obesity-associated metabolic disease. More recently it was shown that transplanting donor BAT into the visceral cavity of recipient mice prevents impaired glucose tolerance from developing in response to high-fat diet (45). Remarkably, protection was also observed with autonomous BAT transplantation, which may support the design of therapeutics that promote browning of WAT as a means of treating age-associated metabolic disease. Whether detrimental changes in BAT function that accompany aging are directly responsible for associated systemic metabolic dysfunction remains unclear, but a more thorough understanding of the metabolic relevance of BAT is likely to provide insight into underlying mechanisms that link aging and associated metabolic dysfunction.

Concluding remarks

Contrary to previous belief, BAT does not simply involute following birth, as it has now been unequivocally observed in both children and adults. BAT is found in similar anatomical locations, and activity correlates with favorable metabolic features at all stages of life. Although BAT activity generally declines with age, there is one exception around the time of puberty (21). The transient spike in BAT activity during adolescence is poorly understood, but may provide a unique opportunity to study endogenous mechanisms that regulate BAT development. Natural reactivation of BAT also highlights plasticity that adds significant support to its relevance as a therapeutic target. It would be interesting in future studies to learn whether this reactivation is specific to depots in the neck and shoulder region or if this effect extends to brown adipocytes in other areas, such as subcutaneous WAT.

Considerable changes in BAT biology are observed during puberty and with aging, albeit in opposing directions. Likewise, these two contexts are associated with a number of opposing endocrine fluctuations and, in particular, contrary changes in the production of sex hormones. These observations collectively implicate a regulatory role for sex hormones in changes in BAT throughout the life-cycle. Based on current evidence, it is possible that steroid hormones such as estrogen act directly on BAT and/or mediate effects indirectly through other hormones (e.g. leptin), or other tissues, such as skeletal muscle.

During puberty, the volume of BAT and skeletal muscle increase in parallel, and, conversely, a decline in the function of both BAT and skeletal muscle is observed with advancing age. Skeletal muscle produces factors such as irisin that can positively influence BAT development, and BAT produces factors such as IGF-1 that can favor growth of skeletal muscle; interestingly, in both contexts the secreted factors are linked favorably with glucose homeostasis (46,48). However, the degree to which age-related changes in BAT are mechanistically linked to changes related to the size and/or function of skeletal muscle (or vice versa), and how this might relate to insulin sensitivity, which often precipitously declines with age, are presently unknown but warrant future study.

Finally, although great strides have been made, the presence of BAT is likely still underestimated due to detection limitations. FDG-PET/CT is most commonly employed, but this method detects only metabolically active BAT (100), which may or may not reflect the total amount of BAT present. While a decline in the activity of BAT with age is clear, an important unanswered question from a therapeutic point of view is if and when BAT becomes completely absent with age. As detection methods improve, understanding the precise timing of the steep decline in BAT activity in later years and whether or not a small fraction of BAT remains dormant with advancing age will be important questions to address in future studies. Likewise, the question of whether or not BAT can be reactivated in elderly individuals as it can in younger adults (87,91), and the degree to which this could be used to treat age-associated metabolic disease, will need to be determined in the future.

Declaration of interest: The authors report no conflicts of interest.