Importance of adipocyte browning in the evolution of endothermy

Abstract

Endothermy changes the relationship between organisms and their environment fundamentally, and it is therefore of major ecological and evolutionary significance. Endothermy is characterized by non-shivering thermogenesis, that is metabolic heat production in the absence of muscular activity. In many eutherian mammals, brown adipose tissue (BAT) is an evolutionary innovation that facilitates non-shivering heat production in mitochondria by uncoupling food-derived substrate oxidation from chemical energy (ATP) production. Consequently, energy turnover is accelerated resulting in increased heat release. The defining characteristics of BAT are high contents of mitochondria and vascularization, and the presence of uncoupling protein 1. Recent insights, however, reveal that a range of stimuli such as exercise, diet and the immune system can cause the browning of white adipocytes, thereby increasing energy expenditure and heat production even in the absence of BAT. Here, we review the molecular mechanisms that cause browning of white adipose tissue, and their potential contribution to thermoregulation. The significance for palaeophysiology lies in the presence of adipose tissue and the mechanisms that cause its browning and uncoupling in all amniotes. Hence, adipocytes may have played a role in the evolution of endothermy beyond the more specific evolution of BAT in eutherians.

This article is part of the theme issue ‘Vertebrate palaeophysiology’.

1. Introduction

Endothermy is characterized by high metabolic rates that permit sufficient heat production for thermoregulation, and it is likely to have evolved in therapsids during the Permian period [1]. Crucially, heat is produced by non-shivering processes [2], although muscular activity and shivering may contribute to heat production in the short term. Endotherms respond to variation in their thermal environment by adjusting metabolic rates and hence heat production, resulting in body temperatures that are often high and stable relative to environmental temperature fluctuations [3]. The advantages of endothermy lie in extending ecological niches and activity temporally into the cooler parts of the day and season, and spatially into cooler climates [4]. However, the high metabolic rates necessary for thermoregulation restrict endotherms to high-energy nutritional environments. Similarly, high rates of heat loss preclude endotherms from attaining the small body sizes that are typical of ectotherms [5].

Relatively few animals are endothermic, but endothermic animals can be hugely successful ecologically. Not surprisingly, the evolutionary processes that have led to endothermy have been debated extensively. Selection pressures that have led to endothermy are thought to fall within two categories: selection for increased activity or for increased body temperature [6–8]. Selection for increased body temperature alone is probably unlikely to have led to endothermy, because during the lifetime of an organism the increases would have been so incremental that their effect on physiological functions would have been negligible [6]. However, it may have been advantageous in association with increased activity, and together incrementally increasing activity and body temperature may have led to improved parental care and hence to endothermy [9–11].

Theories about purported selection pressures are difficult to test empirically. However, the activity theory has been tested quite elegantly using artificial selection experiments [12]. One of the major assumptions underlying the activity theory is that basal metabolic rates are mechanistically linked to maximal metabolic rates, so that if selection favoured increased activity fuelled by high metabolic rates, basal metabolic rates followed passively [13]. Selection for high metabolic rate should therefore also increase basal metabolic rate and thereby promote non-shivering thermogenesis. These predictions have been borne out in some artificial selection experiments, but not in all [8,14]. Importantly, basal metabolic rates are not necessarily linked mechanistically to maximal metabolic rates. Unlike maximal rates, basal metabolic rates are driven by energy (ATP) demands at rest from processes including ATPase activities and protein synthesis [15], and are typically not limited by maximal physiological capacities. In contrast, maximal metabolic rates are constrained by maximal capacities of mitochondria and the cardiovascular system.

A different way to understand the evolution of endothermy is to consider the evolution of the underlying enabling mechanisms. At the cellular level, heat production principally occurs by the collection of all exergonic processes. Mitochondria play a central role either by providing ATP for cellular work that will eventually lead to heat release or by uncoupling mitochondrial electron transport from ATP production [16]. At the mitochondrial level, electrons donated by food during substrate oxidation (carried by NADH and FADH₂) are used by protein complexes in the mitochondrial inner membrane to pump protons (H⁺) from the matrix into the intermembrane space. The resulting proton gradient (proton motive force) drives the rotary function of the mitochondrial F₀F₁ATPase to phosphorylate ADP to ATP [17]. The consumption of ATP by the cellular work will set the tone of how fast ADP is converted back to ATP. However, the energy conversion process is not perfectly efficient, and protons may also leak back across the mitochondrial inner membrane without ATP production, thereby accelerating catabolic processes irrespective of ATP homeostasis and increasing heat production. The mitochondrial proton leak thereby provides a basic mechanism for heat production [17]. Endotherms have greater proton leak than ectotherms [18] at least in isolated liver mitochondria. Non-shivering thermogenesis is tied to the capacity for increased mitochondrial proton leak in response to thermogenic demands. Eutherian mammals, in particular, have evolved brown adipose tissue (BAT), which is specialized for heat production by maximizing mitochondrial proton leak [19]. BAT is rich in mitochondria, which contain a specialized protein, uncoupling protein 1 (UCP1), that acts as an inducible proton channel to release the proton motive force as heat [20]. UCP1 activity and BAT heat production are activated by cold exposure via the sympathetic nervous system and noradrenaline stimulation of adrenergic receptors [19]. The UCP1 orthologue is present in fish and amphibians, but it has been lost in reptiles including birds and in several lineages of eutherian mammals [21–23]. The UCP1 orthologue, therefore, must have existed before the divergence of ray- and lobed-finned vertebrates. Phylogenetic analyses suggest that UCP1 is important for heat production only in eutherian mammals, and its function in other groups, notably also in endothermic monotremes and marsupials, is not clear [23]. In marsupials without any detectable UCP1 levels, there is evidence for cold-induced thyroid hormone levels that could impact metabolic rates and heat production [24]. In eutherian mammals, thyroid hormone interacts synergistically with noradrenaline to induce UCP1, and it is, therefore, an important regulator of BAT [25,26]. In addition to BAT, skeletal muscle is an important site for non-shivering thermogenesis, particularly for birds which do not possess BAT and UCP1 [27–29].

Interestingly, white adipose tissue (WAT) can undergo a process of ‘browning’. White and brown adipocytes have different stem cell origins: most BAT cells are derived from the same myf5-expressing progenitors as skeletal muscle cells, whereas WAT cells develop from white preadipocytes [30]. Nonetheless, some of the BAT characteristics can be induced in WAT, such as increased mitochondrial density and UCP1 expression [31]. Browning of murine WAT occurs in response to a number of stimuli including cold exposure, muscle activity and exercise, nutritional status and immune responses [32], although the process of WAT browning is as yet poorly understood in other species. This raises the question of whether ‘beige’ adipose tissue (i.e. remodelled WAT that has undergone browning) has played a role in the evolution of thermogenesis [33] and whether it predated the classical BAT thermogenesis. It may have played a part in heat generation, although, as is the case with BAT, significant heat generation is likely to be a derived characteristic. It may be that WAT remodelling played a broader role in energetics, for example, by supplying fuel for muscle activity and immune responses. As metabolic capacities increased, there may have been substantial release of heat in beige adipose tissue, particularly in cases where UCP1 was also expressed.

Here, we explore the potential role of WAT remodelling in the evolution of endothermy. First, we summarize mechanisms that mediate heat production at a cellular level. We then introduce browning and review underlying mechanisms that lead to browning of WAT, followed by an assessment of the consequences of WAT browning for energetics and heat production. We will use this background to discuss how browning may have been important in the evolution of endothermy, particularly in a phylogenetic context and with respect to the evolution of regulators that mediate browning.

2. Mechanisms of thermogenesis

Cellular heat production strictly depends on energy turnover rates in mitochondria, which are pacemakers for thermogenesis. Systemically, even anaerobic processes in one organ will eventually lead to mitochondrial flux in other organs. For example, escaping lactate during anaerobic glycolysis will be captured in the Cori cycle and fully combusted in the liver, at least in mammals [34]. Almost every catabolic process will ultimately feed high-energy electrons into mitochondrial electron transport and generate proton motive force. Thus, the dissipation of the mitochondrial proton motive force represents the master regulator of thermogenic output. Both the production of ATP and the mitochondrial proton leak dissipate the proton motive force, and an increase in cellular ATP consumption will directly increase cellular heat output. ATP consumption can be achieved either by cellular work, e.g. muscle contraction, or by apparently ‘futile’ cycles that consume ATP without achieving their primary outcome. A prominent ‘futile’ cycle occurs in the brain heater organ of scrombroid fish, where ATP is hydrolyzed by calcium pumping into the sarcoplasmatic reticulum from which it is subsequently released through the ryanodine receptor without muscle contraction [35]. This ‘futile’ cycle of calcium releases sufficient heat to warm the visionary system and adjacent brain areas to above surrounding water temperatures [35]. ‘Futile’ pumping of calcium may also occur in skeletal muscle of mammals to facilitate muscle non-shivering thermogenesis [28,29,36]. Another heat-producing process is lipid ‘futile’ cycling, defined as the simultaneous occurrence of lipolysis and lipogenesis to increase ATP turnover in adipocytes, which may occur in adipose tissue of mammals [37].

The canonical mechanism of mitochondrial heat production in mammals is based on the mitochondrial proton leak, which does not depend on cellular ATP turnover. Collapsing the proton gradient will directly produce heat by accelerating catabolic processes. The mitochondrial proton leak can be divided into two forms: the basal proton leak that is associated with basal metabolism and heat production of the cell, and the inducible proton leak that enables additional increases in heat production [17]. The basal proton leak is determined by the combination of integral proteins and membrane composition [38]. Basal proton leak contributes directly to cellular metabolism. Indeed, the magnitude of liver mitochondrial proton leak reflects allometric differences in specific heat production of endotherms [39,40]. The basal proton leak can be controlled by thyroid hormone, and hypothyroidism decreases the leak magnitude [41]. The adenine nucleotide translocase (ANT), belonging to the same family of mitochondrial anion carrier proteins as UCP1, appears to contribute to basal proton leak in a quantitative manner [38] and may be important for heat production in birds which lack UCP1 [27]. However, only eutherian UCP1 has been unequivocally accepted as a thermogenic mechanism via increased mitochondrial proton leak, which is activated by free fatty acids in (brown) adipose tissue mitochondria [42]. Note, however, that most data stem from laboratory mice and capacities of BAT and beige WAT in other species may differ substantially. Compared to UCP1, the function of other mitochondrial carriers, such as UCP2 and UCP3, is less clear. These proteins are less likely to contribute to heat production via proton leak and may be more important in regulating reactive oxygen species [43] or for the transport of metabolites [44].

3. Could browning be important in the evolution of endothermy?

Browning of white adipocytes is a multistage process, which comprises several mechanisms that evolved independently from each other (figure 1). Beige adipose tissue is implicated in increasing glucose uptake and lipid metabolism, and browning is now linked to changes in energy status resulting from cold exposure, nutrition, exercise and muscle activity, and immune responses [32]. Interestingly, the functions of both beige adipose tissue and BAT are under the control of AMP-activated protein kinase (AMPK) [45,46], which is an ancient cellular pathway that regulated energy status and energy metabolism in prokaryotes [47]. Hence, in evolutionary terms, the derived functions of beige adipose tissue and BAT were integrated within existing regulatory networks. The complexity of regular pathways increased over evolutionary times, particularly among vertebrates [48] (figure 1). However, it seems most parsimonious that the individual mechanisms that are now known to be important in endothermic regulation did not evolve as a result of selection for endothermy per se. Rather, their increasingly complex interactions made it more likely to result in metabolic phenotypes that resembled endothermy and which were advantageous under specific sets of environmental conditions. Endothermy is rare among animals, and many regulatory pathways underlying it (autonomic nervous system and thyroid hormone) are shared with ectotherms. Increasing complexity over evolutionary time would have increased the likelihood of endothermic phenotypes appearing, with a range of different manifestations [40,49]. While the appearance of BAT occurred as a result of a distinct evolutionary transition [50], browning of WAT occurred more gradually, and its evolutionary appearance and underlying enabling mechanisms are more widespread phylogenetically. Browning, therefore, may have played a greater role in inducing metabolic diversity over evolutionary time and producing intermediate metabolic phenotypes. Beige fat may have contributed to thermogenesis, particularly in the presence of UCP1. However, its broader role in responding to negative energy balance may have been more important in its evolution because the regulation of energy status, particularly via the AMPK pathway, is at the core of maintaining cellular homeostasis in all organisms [47,51]. It is conceivable that the metabolic role of beige fat could explain differential growth rates among vertebrates, which have been used to define dinosaurs as ‘mesotherms’ [52]. In addition to direct thermogenesis, beige fat could contribute to heat production by supplying substrates for ATP production, fuelling ‘futile’ cycling by ATPases [28].

Figure 1.

Summary of principal mechanisms underlying endothermy in a phylogenetic context. Browning of WAT is mediated by several independent mechanisms that have evolved at different times during the evolution of animals (dark shaded bars with white writing). The ‘thermogenic gene program’ that mediates both beige and BATs evolved in metazoans and comprises a range of transcription factors and nuclear receptors of which the PPARs, their co-activators (PGC-1) and vascular endothelial growth factors (VEGFs) are among the most important. Beige fat is also characterized by multilocular fat droplets that facilitate lipid metabolism. Similar to BAT, beige fat function may involve uncoupling of mitochondrial electron transport by UCP1, although this is not obligatory. Note that UCP1 had evolved before the divergence of lobe- and ray-finned vertebrates as it can be found in extant fish species, but its uncoupling function evolved at later stages in the common ancestor of eutherian mammals, where it was also lost in several groups including pigs. The more derived mechanisms (e.g. BAT) rely on ancient metabolic structures and regulators that were already present in or even before the last eukaryotic common ancestor (LECA), and which include mitochondria, AMP-activated protein kinase (AMPK), target of rapamycin (TOR) and adenine nuclear transporter (ANT). The autonomic nervous system appeared in early vertebrates and is now one of the most dominant regulators of BAT and beige fat. Similarly, internal thyroid hormone production occurred in early chordates and has now assumed broad regulatory roles in endothermic physiology. The sarco-/endoplasmic reticulum Ca²⁺-ATPase (SERCA) functions in muscle relaxation, but the characteristic of certain isoforms to undergo futile cycling, which uses ATP in the absence of calcium transport and can contribute to heat production, has been co-opted for endothermy notably in the heater organ of scombroid fish. Note that individual regulators are unlikely to have evolved for the ‘purpose’ of mediating endothermy per se, but their increasing complexity led to an endothermic state in which BAT and UCP1 are not obligatory (indicated by the horizontal (red) box). (Online version in colour.)

4. Browning phenotype

The conventional definition of WAT browning comprises multiple morphological and molecular changes (figure 2), mostly investigated in mouse models and human fat samples. Morphologically, unilocular fat droplets of fat-storing white adipocytes disappear, and multilocular beige adipose tissue appears, which is a sign of enhanced lipid mobilization [53]. During the browning process, white adipocytes are remodelled towards beige adipocytes [54] or are gradually replaced by new beige adipocytes within the WAT [55]. During browning, mitochondrial mass and angiogenesis increase to facilitate energy turnover [56]. Some of the molecular networks that are usually induced during activation of BAT thermogenesis can also be recapitulated during browning and may be labelled as ‘thermogenic gene programming’ [57]. The principal components of this thermogenic gene programming involve the nuclear receptor peroxisome proliferator-activated receptor γ (PPARγ) and its co-activator PPARγ co-activator 1α (PGC-1α) [57]. PGC-1α is induced by cold and exercise via AMPK [58], and it stimulates UCP1 activity [59]. PGC-1α activity is stimulated by thyroid hormone [60] and the sympathetic nervous system [61], and beyond its activation of UCP1 it also mediates mitochondrial biogenesis and increased metabolic capacity via PPARγ [62]. The crucial aspect in adipocyte remodelling is the activation of UCP1 gene expression, which strongly suggests the recruitment of UCP1-dependent thermogenesis, the canonical mechanism of thermogenesis [20,42]. Collectively, these multiple steps of remodelling have been termed browning. The relation of WAT browning to heat production is supported by mouse models of genetic UCP1 ablation and BAT reduction, which both lead to increased compensatory browning of inguinal fat [63,64]. Given the presence of normal, neonatal BAT in wild-type mice, it is still controversial how much beige adipose tissue really contributes to systemic thermoregulation. The metabolic impact of beige adipose tissue in modulating substrate preferences such as increased glucose and lipid uptake may play a more significant role for systemic metabolism than heat production per se [33]. In past years, the terminology ‘browning’, ‘beige’ and ‘brite’ adipose tissue has been frequently used for all forms of beige adipose tissue, but the browning phenotype may still not be completely defined. For example, browning was originally defined as UCP1-postive white adipocytes with partial BAT features [53] (figure 2). These BAT features, however, persist or are enhanced in the absence of UCP1 [64]. Thus, a more nuanced approach for the definition of browning is required, considering the involvement of several independent processes. Three major aspects contribute to adipose browning: the remodelling of morphology, of molecular networks, and the induction of UCP1 protein. Resolving the evolutionary timeline when these contributors of browning have been incorporated will greatly assist in getting new insights on the ecological, physiological and molecular significance of browning, and eventually on its role in the evolution of non-shivering thermogenesis.

Figure 2.

Comparison between BAT and browning of WAT. Both BAT and browning of WAT are stimulated by cold and increasing energy (ATP) scarcity. Browning of WAT is also induced by other processes such as inflammation and exercise (small black arrows indicate the direction of stimulus and wide arrows indicate shifts between increased and decreased browning). Molecular mediators and cellular effects of both types of tissue are similar, although beige (browned) WAT has lower or even absent UCP1 content and lower mitochondrial densities. BAT has a significant role in thermogenesis, but the role of beige WAT in thermogenesis is less clear. Browning of WAT may have functioned primarily in mobilizing fatty acids to supply muscle and other organs with energy; ultimately, this would also have resulted in increased heat release. (Online version in colour.)

5. Evolution of browning

Browned adipose tissue of eutherian mammals represents a sophisticated end product of several hundred million years of evolution, incorporating multilocularity, the coordinated induction of thermogenic genes and UCP1. Although all these characteristics are perfectly integrated into a system, it could be possible to dissect aspects of browning and sort them to distinct evolutionary events. Notably, the genetic ablation of UCP1 in mice does not prohibit the occurrence of multilocularity and thermogenic networks, and these characteristics even seem to be enhanced, therefore suggesting that UCP1 has been implemented at later stages [63,65]. Indeed, the earliest genetic targeting of UCP1 expression to adipose tissue occurred in marsupials [66], while there is no evidence for adipose UCP1 expression in fish and amphibians [21,67]. While the UCP1-ablated mouse represents an artificially constructed animal model, pigs, which are eutherian mammals that have lost UCP1 secondarily [68], show multilocularity in their adipocytes after birth [69] and during cold exposure [70]. Whether or not this occurs in other eutherian lineages that similarly lack UCP1 (xenarthrans, pangolins, paenungulates, cetaceans and equids; [71]) has yet to be examined. Indeed, cold-induced multilocularity of adipocytes is also found in birds [72]. Thus, signalling mechanisms leading to remodelling adipocytes towards browning presumably predated the incorporation of UCP1. In ectothermic vertebrates such as fish and amphibians, the situation is less clear. These vertebrates possess adipocytes and their lipid metabolism can be stimulated by adrenergic stimulation per se [73]. However, to the best of our knowledge, no reports on multilocular adipocytes exist to date for ectothermic vertebrates. The formation of cold-induced multilocular adipocytes in birds, however, strongly suggests that the ancestor of amniotes must have possessed this browning feature. This ability to remodel towards multilocularity and lipid mobilization should also be accompanied by a coordinating molecular network in non-mammalian species, which has not been investigated in ectothermic vertebrates so far. In rodents, several molecular factors have been identified that coordinate adipocyte remodelling such as perilipin, plin5 and Fsp27 [74]. Notably, cideA, classified as a typical thermogenic gene in mammalian adipose tissue, directly controls lipid droplet formation [75]. Lipid droplet remodelling promotes rapid lipid mobilization and oxidation. In addition to heat production mediated by UCP1 in beige adipocytes, the increased mobilization of free fatty acids as a result of lipid droplet remodelling that is characteristic of beige fat may also complement demands of other organs by providing fuel. It is therefore not surprising that exercise induces browning [32], because this may be a way to fuel skeletal muscle with lipids and other metabolites. Eventually, these paracrine and endocrine actions, serving to provide fuel for energy-demanding processes in other tissues, may have caused the release of energy as heat in beige adipocytes.

6. Browning signalling pathways

Understanding the role of adipose tissue browning will depend on the identification of the underlying controlling factors. All characteristics of browning can be induced by adrenergic stimulation in eutherian mammals, and it represents the classical signalling pathway of BAT thermogenesis in response to cold ambient temperatures [42]. In recent years, alternative pathways to this canonical pathway have been discovered that also allow control over browning by interorgan crosstalk. Three major classes of browning factors have been identified so far, either of peptidic, micro RNA or metabolite nature [30]. These factors are induced and released by inflammatory responses, exercise or nutritional status [32,76,77]. The underlying mechanisms, such as signalling cascades in the adipocytes, for example, are still unknown. There is increasing evidence from mice that immune cells impact beige adipose tissue thermogenesis [76], but the physiological significance is still controversial. For example, the idea of adipose tissue-resident macrophages secreting noradrenaline to activate thermogenic adipose tissue [78] has been disputed by others [79]. Similarly, the effects of exercise on WAT browning are evident [80], but the exact causal mechanisms remain enigmatic. The experimental evidence for irisin, an exercise-induced myokine of the FNDC5 gene [81], is not fully conclusive because of experimental issues [82] and the lack of substantial induction following exercise [83]. Additionally, small metabolites such as lactate and β-aminoisobutyric acid can act as browning agents [84,85]. Given the controversy in this non-canonical field of browning induction, broader studies not only focusing on mice but also including non-model organisms, such as evolutionary distantly related mammals and other vertebrates, may assist in casting light on the role of browning. In the closest relatives of modern eutherian mammals, the marsupials, adrenergic stimulation elicits only a minor thermogenic response that is not related to UCP1 expression levels as is the case in rodents [66]. Considering that marsupials diverged from their eutherian ancestors about 120 million years ago, some of the non-canonical endocrine signals may have preceded the canonical adrenergic activation known from eutherian brown and beige adipose tissue. Browning studies on marsupials and ectothermic vertebrates may be rewarding to understand the role of adipose tissue browning in the absence of the thermogenic BAT. In ectotherms, the browning process would not be confounded by endothermy per se, which may uncover the additional roles for activated beige adipose tissue in metabolism.

7. Theory and mechanisms of endothermy

The evolution of endothermy is most often discussed in the context of possible selection pressures that have resulted in endothermy. However, endothermy is a physiological state, and there is a certain philosophical disjoint between interpretations of evolutionary theories and the physiological mechanisms that support endothermy. In a physiological context, the ‘function’ of a trait such as enzyme activity or signalling molecule can be seen simply as the roles those molecules play in their pathways. However, in an evolutionary sense, there is a teleological (goal-directed or purposeful) dimension to the meaning of the function. The common logic is that traits are present because their effects were selected for in the past for the purpose of increasing the fitness of the organism. There are several problems with this formulation. Selectionist theories define a function with an exclusive reference to external causation (selective process), but pay little attention to internal causation in relation to organizational properties [86]. Recruiting physiological support for evolutionary arguments may be problematic because current physiological functions do not necessarily correlate with current or past fitness, or reflect the past selection. Traits may be co-opted for different functions [87,88] and genetic drift may be as or even more important than selection in directing evolutionary dynamics [89], so that selectionist arguments do not necessarily work well [90]. Endothermy is possible only with a complex regulatory system, but that regulatory system has not necessarily evolved for the ‘purpose’ of endothermy. For example, thyroid hormone signalling has not evolved to regulate BAT metabolism [91], but it is essential for that function at later stages [92]. It is therefore not necessarily a particular trait or its effect that represented a selective advantage but the increased regulatory complexity itself, because it increased the likelihood of endotherm-like phenotypes, which may have been selected for under a specific set of environmental circumstances.

The various theories explaining the evolution of endothermy attempt to understand these environmental circumstances, but in themselves, they cannot explain the traits that enabled endothermy or their evolution. It is informative, therefore, to consider both in parallel. Functional analysis of endothermy should distinguish between the current utility of a trait to the properties of the system (e.g. endothermic thermoregulation) and explanations of the historical processes that led to the presence of the trait [86,87]. Failure to distinguish between current utility and historical processes will lead to a trivialization of evolutionary processes and misinterpretations of ‘spandrels’ [93]. Moreover, once the current utility of traits is understood, the evolutionary history of those traits can also contribute to explaining the evolution of the system. Browning of WAT is a good example where underlying regulatory processes are most likely to have evolved independently, but ultimately their interaction led to a distinct physiological phenomenon. The current importance of beige fat for thermoregulation or energy expenditure is therefore unrelated to the selection pressures that have resulted in the evolution of the regulatory processes. Whether or not browning of WAT contributes substantially to thermogenesis is unresolved. However, the regulatory components underlying browning and their evolution led to the appearance of distinct metabolic phenotypes that may have contributed to endothermy. For example, the increased mobility of fatty acids from adipocytes that is characteristic of browning may have functioned primarily in supplying fuel substrates to muscle and other organs, thereby allowing increased metabolic capacity. The increased metabolic capacity would ultimately also lead to increased heat production, so that there is a link between browning and thermogenesis that is independent of BAT and which has not evolved as a result of selection for endothermy per se but ultimately contributed to it.

Data accessibility

This article has no additional data.

Authors' contributions

M.J. and F.S. contributed equally to the conceptualization and writing the manuscript.

Competing interests

We declare we have no competing interests.

Funding

M.J. is supported by the Swedish Research Council (grant no. 2018-03472) and F.S. was supported by Australian Research Council Discovery grant no. DP180103036.