Human brown adipose tissue: Classical brown rather than brite/beige?

What is the topic of this review? It has been suggested that human brown adipose tissue (BAT) is more similar to the brite/beige adipose tissue of mice than to classical BAT of mice. The basis of this is discussed in relationship to the physiological conditions of standard experimental mice. What advances does it highlight? We highlight that, provided mouse adipose tissues are examined under physiological conditions closer to those prevalent for most humans, the gene expression profile of mouse classical BAT is more similar to that of human BAT than is the profile of mouse brite/beige adipose tissue. Human BAT is therefore not different in nature from classical mouse BAT.

F I G U R E 1 Initial (but potentially misleading) comparison of tissue appearance between human brown adipose tissue (BAT) versus brown and brite/beige adipose tissues from 'standard' mice. The BAT observed in mice living in standard conditions (i.e. young, chow-fed mice, 20 • C) has small cells that are multilocular [contain many small lipid droplets (yellow)] and a dense mitochondrial network (black), where large amounts of UCP1 are found (red dots). This is morphologically different from the appearance of human BAT, which mainly has larger cells that are often unilocular (contain one large lipid droplet) and contain few mitochondria; the tissue also contains some clusters of cells that look more similar to BAT in mice and also contain UCP1. Human BAT thus appears to be similar morphologically to the brite/beige adipose tissue (inguinal subcutaneous adipose tissue) observed in standard mice, where some multilocular cells are also found amidst mainly large unilocular cells; some of the multilocular cells in certain mouse strains, e.g. 129Sv, may contain Ucp1 protein, albeit in small amounts that mouse classical BAT is more akin to human BAT than is mouse brite/beige adipose tissue.

BROWN ADIPOSE TISSUE IN HUMANS
Until 2007, the general view was that, in humans, the presence of active BAT was limited to neonates and that it atrophied and disappeared during late childhood. A few reports of brown fat in adult humans were published. However, these reports concerned rather special physiological conditions (outdoor workers in cold Finnish winters) or pathological conditions (patients suffering from phaeochromocytoma or hibernoma) (reviewed by Lean, 1989). Brown adipose tissue could, therefore, not be assigned any relevance for normal adult human metabolism. However, in 2007, based on a compilation of analyses of positron emission tomography scans obtained earlier from patients undergoing investigations of fluorodeoxyglucose uptake for the identification of cancer metastases (e.g. Hany et al., 2002), it was established that even adult humans possessed what would seem to be active BAT (Nedergaard, Bengtsson, & Cannon, 2007), at least when defined as adipose depots that showed high glucose uptake when the patients felt cold (Christensen, Clark, & Morton, 2006). Some of these depots had also been verified to express

New Findings
• What is the topic of this review?
It has been suggested that human brown adipose tissue (BAT) is more similar to the brite/beige adipose tissue of mice than to classical BAT of mice. The basis of this is discussed in relationship to the physiological conditions of standard experimental mice.
• What advances does it highlight?
We highlight that, provided mouse adipose tissues are examined under physiological conditions closer to those prevalent for most humans, the gene expression profile of mouse classical BAT is more similar to that of human BAT than is the profile of mouse brite/beige adipose tissue.
Human BAT is therefore not different in nature from classical mouse BAT.
However, when the morphology of the human BAT was examined microscopically, it was observed -initially perhaps disappointinglythat it did not resemble BAT as we knew it from small mammals. It did contain UCP1, but the UCP1 was found only in rather small amounts in only some of the cells (in cell clusters), and a large fraction of the tissue presented with large adipocytes containing large fat droplets, perhaps only a single droplet (Zingaretti et al., 2009). Thus, morphologically, the human tissue did not seem to be similar to classical BAT ( Figure 1, human BAT).

DIFFERENT ADIPOSE TISSUE DEPOTS
However, there are many different depots of adipose tissue in the mouse; at least some 13 depots can easily be identified anatomically (de Jong, Larsson, Cannon, & Nedergaard, 2015;Zhang et al., 2018). (brown-like in white) adipocytes was introduced (Petrovic et al., 2010).
Subsequently, in an elegant molecular study (Wu et al., 2012), it was observed that a series of genes were well expressed in immortalized

INITIAL ANALYSIS OF ADIPOSE TISSUE IDENTITY ACCORDING TO MARKER GENES
To distinguish between classical BAT and brite/beige adipose tissues, it was thus possible to suggest some genes (the molecular function of which is not relevant for this discussion) as marker genes for each tissue (Sharp et al., 2012;Wu et al., 2012). As examples, Eva1 may be considered a marker gene for brown (its expression is about fivefold higher in brown than in brite/beige adipose tissue collected from mice living in standard animal housing conditions), whereas Tmem26 may be considered a brite/beige marker, because its expression is >10-fold higher in brite/beige than in brown adipose tissue in these conditions (Sharp et al., 2012;Wu et al., 2012;de Jong et al., 2019). Indeed, when we examined the expression pattern of these proposed 'marker' genes in human BAT versus brown or inguinal (brite/beige) adipose tissue from standard mice in a principal components analysis, we could confirm that the closest mouse equivalent to human BAT is the inguinal brite/beige tissue rather than the classical BAT ( Figure 2). Thus, both from the morphological observations ( Figure 1) and from the marker gene expression analysis ( Figure 2) performed with "standard" mice, it would seem that BAT in humans should be considered brite/beige ( Figure 1).

WHAT IS A "STANDARD" MOUSE?
From a physiological point of view, a 'standard mouse' , as used for most experimental research, is a mouse that is in a physiological state that is not very similar to that of adult humans. Firstly, the mice normally used are very young. Metabolic studies are often initiated when mice are 6-8 weeks of age, i.e. mice that have entered puberty and are thus comparable to human teenagers. The mice indeed continue to grow, in total protein content etc., to ≥12 weeks of age. Secondly, the mice are fed a 'chow' diet, i.e. a nutritionally well-balanced diet, but one that is dry and not very tasty (at least not to human taste). Most humans in industrialized countries are presently exposed to diets that are tasty, even hedonically so. Thus, we eat not only to obtain calories and nutrients but also because it gives us pleasure. Thirdly, the mice are housed in cages kept at a temperature of ∼20 • C. For mice, this is a cold temperature. Mice exhibit their lowest metabolism (their basal metabolic rate) at ∼30 • C (their thermoneutral temperature). When they are living at 20 • C, they have to almost double their metabolism (food intake etc.) to maintain body temperature (e.g. Fischer, Cannon, & Nedergaard, 2018). For humans, we would have to be constantly (day and night) exposed naked to 10 • C to induce an increase in metabolism (Erikson, Krog, Andersen, & Scholander, 1956) and food intake similar to that which the mice experience continuously in standard animal house conditions. Thus, from a translational point of view, 'standard' mice might be said to be rather special.
(There are evidently many other factors that distinguish experimental mouse life from normal human life: a low degree of environmental novelty, different degrees of social interaction, different daily rhythm, etc., but for metabolic studies, we would consider those mentioned here to be the most important ones.)

THE PHYSIOLOGICALLY HUMANIZED MOUSE
The main translational target for metabolic studies performed in mice is obviously not human teenagers living naked at 10 • C and being fed dry food pellets. Thus, the question may be raised: are young, coldstressed, chow-fed mice really good models for most adult humans that are middle-aged, living primarily close to thermoneutrality and eating hedonically attractive, rather high-fat and tasty meals?
We would instead suggest that mature (>6 months old) mice fed a tasty high-fat/high-sugar diet (e.g. 45 energy % lipid, plus 35 energy % carbohydrate, of which half is sucrose, the standard 45 % high fat diet F I G U R E 2 Initial (but potentially misleading) comparison of gene expression pattern between human brown adipose tissue (BAT) versus brown and brite/beige adipose tissues from 'standard' mice. The morphological similarity between human brown and mouse brite/beige adipose tissue ( Figure 1) is accompanied by molecular similarities in the expression pattern of genes that have been suggested to be 'markers' for brown versus brite/beige adipose tissues (Sharp et al., 2012;Wu et al., 2012). Indeed, as seen in this principal components analysis (PCA), the general expression pattern of these markers would clearly indicate that human BAT is more close to mouse brite/beige than to mouse BAT. The analysis shown here is based on data from the study by de Jong et al. (2019). The human BAT samples (n = 8) were obtained from thermogenically confirmed depots in the supraclavicular region (available as data sets in the European Nucleotide Archive, with the accession number PRJEB20634), and interscapular BAT (IBAT) and inguinal white adipose tissue (ingWAT) samples were obtained from 'standard' mice (young, chow-fed mice, at 20 • C; n = 3; available as data sets in Array Express, with the accession number E-MTAB-7561). The samples were analysed with RNA-Seq. The PCA of the suggested marker genes (brown: Fbxo31, Eva1/Mpzl2 and Ebf3; and brite/beige: Cd137, Tbx1 and Tmem26) was performed using normalized counts per million reads (cpm) values [we have not included Zic1 as a brown marker; although we initially implied it as a marker (Timmons et al., 2007), we found in later studies that its expression was restricted to anterior adipose tissues, regardless of adipose tissue 'colour' ]. The highlighted ellipsoidal area for each tissue corresponds to a confidence interval of 67%. The numbers in parentheses on the axes represent the proportion of data variance explained by each principal component (PC); to obtain uniform representation of variance over the graph surface, the axes were adjusted according to the percentage of variance explained by each of the components. Note that, as discussed in the main text, the pattern obtained here is influenced by the physiological state of the mice examined here (contrast Figure 5) from Research Diets) and living at mouse thermoneutrality (∼30 • C) should be better metabolic models for adult humans than are the 'standard' mice commonly used in metabolic studies. We will refer to these mice as physiologically humanized.
Exposure to these conditions will evidently affect the general state of the mice. Just as middle-aged humans generally are less lean than teenagers, the physiologically humanized mice are heavier than standard mice, weighing ∼50 g, versus 30 g for a standard mouse, with a total lipid content of 20 versus 3 g. However, despite their obesity, these mice are not markedly insulin resistant, at least as estimated from an insulin tolerance test (Abreu-Vieira et al., 2015;de Jong et al., 2019).

DO HUMANS LIVE AT THERMONEUTRALITY ?
Most people will agree that humans normally eat food that is different from mouse chow and that most humans are not young teenagers, but a more controversial issue is whether most humans live most of the time at thermoneutrality.
Of course, this question entails several methodological problems.
Examinations of (nearly) naked humans imply that human thermoneutrality is rather close to that of mice, i.e. a little lower than 30 • C (Hill, Muhich, & Humphries, 2013). Evidently, this is not an environmental temperature that most humans are exposed to most of the time. However, we are not naked most of the time either, and there is good reason to believe that we generally dress so that we minimize our heat loss in an attempt to maintain thermal comfort. Studies that directly examine this are, however, scarce and would, in any case, not have been performed on larger cohorts of people. Thus, the issue has to be approached in an indirect way.
One indirect way to approach this is to consider the definition of thermoneutrality, i.e. the temperature zone where metabolic rate is lowest. However, metabolic rate, or daily mean energy expenditure (MEE), consists of two components. There is the true basal metabolic rate (BMR) plus what may be referred to as 'physical activity' (although this might not be the best term for this factor, because it includes not only physical activity but also components such as diet-induced thermogenesis, but for the present we can ignore this distinction). The so-called 'physical activity level' (PAL) for a given individual is the ratio between the mean energy expenditure and the basal metabolic rate: PAL = MEE/BMR (Figure 3).
In humans, it is straightforward to measure the BMR by asking the subject to remain still, lying down, being at thermal comfort (with clothes and blankets) and not recently having eaten and to measure metabolic rate for ∼30 min.

F I G U R E 3
Identification of the mouse metabolic state most similar to the human metabolic state. In humans, basal metabolic rate (BMR) can be determined by direct measurements, and the mean energy expenditure (MEE) can be determined through the double-labelled water technique. The double-labelled water technique allows the individuals to be free living, in this respect including living at any temperature they choose and being dressed in any way they choose. The extra energy thus used above the BMR in normal 'free-living' conditions may be referred to as the 'physical activity' (PA). The ratio between the MEE and the BMR is referred to as the 'physical activity level' (PAL) and has been measured to be ∼1.7 for most individuals. In mice, the BMR can be approximated as being the lowest continuous metabolic rate observed in metabolic chambers at thermoneutrality, and the MEE can be obtained from continuous measurements in such chambers. For mice at thermoneutrality (30 • C), the ratio between these values (PAL) is 1.7 (Westerterp, 2018). Thus, free-living humans and mice at thermoneutrality have identical PALs. This indicates that these mice are good metabolic models for humans. The correspondence between the human free-living PAL and the mouse PAL does not necessarily demonstrate that humans live essentially at thermoneutrality, but we would consider this an implication. Mice living at the standard animal house temperature have a much higher MEE (nearly double that of the mice at 30 • C), owing to a combination of 'normal' physical activity and cold-induced thermogenesis (CIT). They may thus be considered less relevant metabolic models for humans. (Principal summary of data from Fischer et al., 2018;Fischer, Cannon, & Nedergaard, 2019;Fischer, Csikasz, von Essen, Cannon, & Nedergaard, 2016;Westerterp, 2018; picture created with BioRender.com) To obtain the MEE in humans is not as easy. It is possible to confine the subjects for a prolonged time (>24 h) in metabolic chambers, but this probably lowers spontaneous physical activity. A better alternative is to measure metabolism in 'free-living subjects' , being at whatever temperature they prefer and being clothed however they wish and otherwise doing whatever they want. It is possible to do this by using the 'double labelled water method' to measure metabolism.
Basal metabolic rate and MEE have been measured in parallel in rather large cohorts of humans (in northern temperate zones and with a Western lifestyle). Although a rather wide spread of values for PAL are obtained from different individuals and different populations, the mean value obtained is ∼1.7 (Westerterp, 2018). This means that during normal life, adult humans expend (only) ∼70% extra energy on 'activity' , on top of their BMR. It can therefore be argued that a relevant translational mouse model for metabolism should, likewise, expend not more than ∼70% above its BMR for all activities, including heat produced to counteract heat loss in the cold (Figure 3).
The determination of PAL in the mouse also has methodological difficulties. One of these is that it is clearly not possible to ask a mouse to remain still, not to have eaten and to ensure that it is in thermal comfort. What can be done instead is to assume that the mouse in its daily life might be in such conditions spontaneously during certain periods: being inactive and not having eaten recently. Indeed, by using high-time-resolution indirect calorimetry in mice exposed to what are thermoneutral ambient temperatures for mice (∼30 • C), it is possible to identify periods of ∼10 min in duration when the energy expenditure of the mice is at its lowest (Fischer et al., 2018). Such periods tend to occur towards the end of the light phase of the mice (Fischer et al., 2018;Keijer, Li, & Speakman, 2019), indicating that they are not random inconsistencies in measurement but represent biological events. It is thus possible to consider the metabolic rate during these periods as a reasonable proxy of the BMR of the mouse. During prolonged measurements of metabolic rates in indirect calorimeters, a value for the MEE at thermoneutrality can also be obtained (note that this value is sometimes incorrectly referred to as the BMR of the mouse, but, exactly as in humans, the total energy utilization is the sum of the BMR plus the extra energy expenditure of physical activity etc.). Thus, for mice maintained at thermoneutrality, a value for PAL can be obtained and, as a mean, this value is 1.7 (Fischer et al., 2018). This indicates that mice living at thermoneutrality have a total metabolic rate 70% above their BMR; this is exactly the mean metabolism of humans in 'free-living conditions' . Therefore, mice at thermoneutrality may be considered good metabolic models of humans. Note particularly that at any temperature below thermoneutrality, the metabolic rate (and consequently, the PAL) is higher. Thus, less translationally relevant PAL values are obtained (Figure 3). Given these criteria, housing experimental mice in 'standard' conditions might not be optimal for translationally intended metabolic research.
Thus, although these indirect measurements do not demonstrate in themselves that adult humans normally attempt to live in thermoneutral conditions (although we would think that we do), they do demonstrate that mice at thermoneutrality are more physiologically humanized metabolically than mice at any other ambient temperature ( Figure 3).

MORPHOLOGICAL APPEARANCE OF BROWN FAT IN PHYSIOLOGICALLY HUMANIZED MICE
Given this understanding of the metabolic characteristics of physiologically humanized mice, it would appear relevant to study these mice to address the question of brown and brite/beige adipose tissue similarities in mice and humans. Visually, the BAT of mice living in physiologically humanized conditions (∼9 months old, fed a high-fat diet for ≥6 months and living at 30 • C) contrasts sharply with that of BAT of 'standard' mice (only ∼8-10 weeks old, fed chow and living at 20 • C). The physiologically humanized mice have a rather whitishlooking 'brown' adipose tissue, with large, rather lipid-filled cells, most of which are unilocular. Although UCP1 is present, it is found in only certain cells, and these cells tend to be in clusters of multilocular cells within the tissue (de Jong et al., 2019). In these respects, the mouse classical BAT depots now appear very similar to the BAT observed in humans ( Figure 4). A question would be to what extent is this also reflected in the gene expression profile.

DISTINCT GENE EXPRESSION IN ADIPOSE TISSUES OF PHYSIOLOGICALLY HUMANIZED MICE
As would be expected, the gene expression level not only of Ucp1 but also of several other genes normally considered to be part of the thermogenic programme (Pgc1a, Cidea and Elovl3) in adipose tissues is lower in physiologically humanized mice than in standard mice. This is the case in both classical BAT and in the inguinal brite/beige tissue (although the expression of these genes in the inguinal depots is generally at least 10-100 times lower than in classical BAT; de Jong et al., 2019).
Indeed, not only genes directly related to thermogenesis but a very broad array of genes have different expression levels when adipose tissues from mice living in standard or physiologically F I G U R E 4 When human brown adipose tissue (BAT) is compared with mouse brown and brite/beige adipose tissues from physiologically humanized mice, the result is substantially different from a comparison made with standard mice (as in Figures 1 and 2). In physiologically humanized mice, the BAT has a morphological appearance similar to that of human BAT, with clusters of multilocular cells containing many mitochondria being found among unilocular cells. In these conditions, the brite/beige adipose tissue has completely lost its similarity to human BAT; it is now fully unilocular, as is 'normal' white adipose tissue, and UCP1 protein can no longer be detected

A DIRECT COMPARISON BETWEEN HUMAN AND MOUSE ADIPOSE TISSUE TRANSCRIPTOMES
For a general approach to the question of the identity of human BAT versus mouse brite/beige or classical BAT in the different physiological conditions, it should be possible to compare the entire transcriptome between these tissues. This is possible because the transcriptome of F I G U R E 5 Comparison of gene expression patterns between human brown adipose tissue (BAT) versus brown and brite/beige adipose tissues from physiologically humanized mice. The morphological similarity between human BAT and classical BAT from humanized mice (Figure 4) is accompanied by molecular similarities in the expression pattern of the genes that have been suggested to be 'markers' for brown versus brite/beige adipose tissues (Sharp et al., 2012;Wu et al., 2012). The analysis of the data is similar to that described for Figure 2 except that interscapular BAT (IBAT) and inguinal white adipose tissue (ingWAT) from physiologically humanized mice (n = 5) are analysed here. The issue of the relevance of markers and the nature of human BAT has also been discussed elsewhere (Cypess et al., 2013;de Jong et al., 2015;Jespersen et al., 2013;Lidell et al., 2013;Nedergaard & Cannon, 2013;Sanchez-Gurmaches, Hung, & Guertin, 2016;Walden, Hansen, Timmons, Cannon, & Nedergaard, 2012) human thermogenically verified BAT has been characterized (Perdikari et al., 2018).
However, a simple, direct comparison of all the relevant transcriptomes yields the unsurprising but not very elucidating conclusion that humans and mice are different species! Thus, correspondence between the adipose tissues in the two species cannot be established (de Jong et al., 2019). It transpires that the number of genes that characterize the different adipose tissues is low compared with the entire transcriptome of the tissues, a problem that has been encountered earlier (Breschi et al., 2016;Lin et al., 2014). What is perhaps somewhat unexpected is that these mouse versus human cells apparently in general carry transcriptional signatures that identify them as the respective species; however, an analysis of the molecular background of these species signatures is outside the scope of the present discussion.

A REMARKABLE SHIFT IN 'MARKER' GENE IMPLICATIONS: MOUSE CLASSICAL BROWN ADIPOSE TISSUE IS NOW SIMILAR TO HUMAN BROWN ADIPOSE TISSUE
Given that a comparison between the total transcriptomes does not answer the question of the nature of human versus mouse adipose tissues, other strategies must be used. One of them is to revert to the 'marker' genes discussed earlier (Figure 2) for brite/beige versus brown adipose tissue. In standard mice, the expression pattern of these marker genes clearly indicated that mouse brite/beige adipose tissue was the adipose tissue most similar to human BAT ( Figure 2). However, when the expression pattern of these marker genes is compared between thermogenically verified human BAT and inguinal and BAT from physiologically humanized mice, a rather clear but surprisingly A similar conclusion may be reached if adipose tissue-defining genes (as determined by the BATLAS gene list (Perdikari et al., 2018)) and genes related to thermogenic potential (as determined by the ProFAT computational tool (Cheng et al., 2018)) are instead compared (de Jong et al., 2019). Notably, the expression level of Ucp1 is very similar between BAT from physiologically humanized mice and true human BAT, whereas the Ucp1 expression level in brite/beige adipose tissue even from standard mice is some 100-fold lower.
Thus, despite the molecular challenge that lies in understanding why some cells in white adipose tissue may express Ucp1 (i.e. become brite/beige) and the possibilities implied for vastly increasing Ucp1 gene expression in these depots, it should be understood that the study of brite/beige adipose tissue is not a study of the mouse equivalent to human BAT. For translationally relevant studies of mouse adipose tissue to understand and affect human BAT, classical BAT (in physiologically humanized mice) should be the preferred tissue.

DOES BROWN ADIPOSE TISSUE IN PHYSIOLOGICALLY HUMANIZED MICE RETAIN ITS THERMOGENIC COMPETENCE?
The BAT encountered in physiologically humanized mice looks very atrophied, being lipid filled and with few Ucp1-expressing cells, etc. It may therefore be questioned whether this tissue still retains its ability to become recruited. In particular, it may be asked whether it can be induced to attain UCP1 protein levels of the same magnitude as those observed in mice that have been 'directly' acclimated to cold.
The answer to this question would seem to be yes (in agreement with in silico predictions from ProFAT; Cheng et al., 2018). When the physiologically humanized mice are exposed to cold for an extended time (∼1 month), the total amount of UCP1 found in the tissue is at least at the same level as it is in cold-acclimated standard mice (standard mice that are transferred directly to the cold without first having been physiologically humanized). The tissue has, however, a somewhat different appearance, with larger lipid droplets (de Jong et al., 2019).
However, very remarkably, but again in agreement with in silico results from ProFAT, in the physiologically humanized mice that are chronically cold exposed, the brite/beige adipose tissue loses its ability to become thermogenic, and UCP1 protein is no longer found in the tissue, even in the cold-acclimated state. The reason for this loss of thermogenic capacity has been examined in cell cultures derived from young and older mice. Brown adipocytes isolated from classical BAT depots fully retained the ability to respond to noradrenaline stimulation with a large increase in Ucp1 gene expression even when they were isolated from older mice. However, the brite/beige adipocytes isolated from older mice had completely lost this ability (which was present in brite/beige adipocytes isolated from young mice; de Jong et al., 2019).
Thus, it may be envisaged that human BAT, although it appears atrophied, can still be recruited fully by chronic cold exposure (or perhaps by pharmacological means), even in middle-aged humans living at thermoneutrality and eating palatable food. This view is supported by the outcome of direct studies of human brown adipogenesis (Lee, Swarbrick, Zhao, & Ho, 2011).

IS BROWN ADIPOSE TISSUE OF METABOLIC SIGNIFICANCE IN ADULT HUMANS?
The implication above that human BAT can be recruited by chronic cold, despite its atrophied appearance, might be considered to be of cursory interest. However, a main issue under debate is, of course, whether human BAT could be able to influence energy balance in humans. Can the small and apparently atrophied BAT encountered in adult humans affect the development of obesity in humans?
At present, there is no clear answer to this question. However, it should be remembered that the presence or absence of BAT (UCP1) in mice would seem to affect energy balance in mice living at thermoneutrality, where the tissue appears as atrophied as it does in humans. Thus, when they are exposed to a high-fat diet, mice lacking the thermogenic capacity of BAT have generally been observed in different laboratories to become somewhat more obese, more quickly than wild-type mice. Concordantly, exposure to the high-fat diet is paralleled by a small, but apparently metabolically significant, increase in total UCP1 protein in wild-type mice (Feldmann, Golozoubova, Cannon, & Nedergaard, 2009;Luijten, Feldmann, von Essen, Cannon, & Nedergaard, 2019;Rowland, Maurya, Bal, Kozak, & Periasamy, 2016;von Essen, Lindsund, Cannon, & Nedergaard, 2017;Winn et al., 2017).
It is thus still possible that human BAT, which is of the same nature as mouse BAT, might be able to have an ameliorating effect on the development of obesity in humans living in 'physiologically humanized' conditions, i.e. being middle-aged, living in thermoneutral conditions (owing to housing and clothes) and constantly exposed to unlimited amounts of tasty and calorie-dense foods.