The confounding effects of sub-thermoneutral housing temperatures on aerobic exercise-induced adaptations in mouse subcutaneous white adipose tissue

Abstract

Mice are the most commonly used model organism for human biology, and failure to acknowledge fundamental differences in thermal biology between these species has confounded the study of adipose tissue metabolism in mice and its translational relevance to humans. Here, using exercise biochemistry as an example, we highlight the subtle yet detrimental effects sub-thermoneutral housing temperatures can have on the study of adipose tissue metabolism in mice. We encourage academics and publishers to consider ambient housing temperature as a key determinant in the methodological conception and reporting of all research on rodent white adipose tissue metabolism.

1. Sub-thermoneutral housing temperatures confound preclinical metabolic research

In response to a thermal challenge that threatens to lower core body temperature, mice engage a number of energetically inexpensive behavioural and autonomic heat gain responses that exist to minimize endogenous heat loss to the environment while also conserving energy. Such responses include huddling which reduces the surface area from which heat can radiate, piloerection which traps an insulating layer of still air adjacent to the body, and vasoconstriction which reduces perfusion to peripheral tissues [1]. Given the 320-fold difference in the mass of a mouse compared with an adult human, mice have a greater surface area to volume ratio than humans and a smaller absolute heat storage capacity, meaning that mice are more closely coupled to their thermal environment than humans. Passive heat gain responses alone are insufficient to offset this heat loss and defend core body temperature; thus mice rely heavily on skeletal muscle shivering and then brown adipose tissue non-shivering thermogenesis for endogenous heat production during a chronic thermal stress [1]. Since it is standard practice for mouse vivariums to be kept at an ambient temperature that is within a human researcher's thermal comfort zone (i.e. approximately 20–24°C), and because mice do not share the same thermal comfort zone as humans [1], mice housed at ambient room temperatures must maintain a rate of metabolic heat production that is much higher than that observed in humans in order to defend a constant internal core temperature [1]. Therefore, it can be reasonably assumed that the majority of mice used for preclinical metabolic research are chronically exposed to a moderate thermal stress.

One of the most influential proponents for recognizing the importance of the effects of ambient temperature on rodent metabolism was C. J. Gordon, who first outlined his ideas on the topic in 1990 [2]. These ideas have been reiterated numerous times by Reitman [3,4], Speakman [5,6], Cannon & Nedergaard [7,8], and the like, all modern-day advocates for the appropriate thermoneutral housing of rodents and thus the correct study of rodent metabolism. Thermoneutrality, defined as the range of ambient temperatures within which metabolic rate is maintained, is not a novel concept [1,9,10]; however, its inclusion in the study of both healthy [11] and diseased [12,13] models of preclinical metabolic research is growing given the increasing awareness of the effects ambient temperature can elicit on metabolic, immunologic and cardiovascular homeostasis [14]. Although the range of ambient temperatures defining the thermoneutral zone has not been agreed upon [3,5,6,8], likely because a number of factors including but not limited to strain, body mass, age and acclimation, there is no refuting that mice housed at thermoneutrality offer a more accurate model of the human metabolic phenotype compared with mice housed at room temperature. It should also be mentioned that other determinants, in addition to ambient temperature, can affect the rodent thermal environment. These factors must also be considered, and include age, strain, size and number of mice per cage, and the amount and type of bedding per cage.

Increasing evidence supports the notion that mice housed at sub-thermoneutral temperatures exhibit a metabolic phenotype that is quite different from mice housed at thermoneutrality [11,15–22], and the confounding effects of sub-thermoneutral housing temperatures are perhaps best highlighted by studies attempting to genetically phenocopy human metabolic diseases in mice, such as the early work on uncoupling protein 1 (i.e. UCP1) knockout mice. In the initial studies characterizing this knockout model [23,24], it was expected that upon a high-fat diet challenge the UCP1 knockout mice would exhibit improved metabolic efficiency; however, these mice instead exhibited a paradoxical resistance to high-fat diet-induced obesity. Only when these same mice were nutritionally challenged under thermoneutral conditions [25] did it become apparent that UCP1 knockout mice do indeed exhibit the expected phenotype of improved metabolic efficiency characterized by increased adiposity and weight gain. Similar examples exist for preclinical models of atherosclerosis [26,27] and hypothyroidism [28], where the expected metabolic phenotype of the knockout model failed to mimic that of the human disease when mice were housed at a sub-thermoneutral temperatures, but not at thermoneutrality. In all of these cases sub-thermoneutral housing temperatures unintentionally confounded key metabolic endpoints masking the true metabolic phenotype of the model. This clearly highlights the subtle yet important effects sub-thermoneutral housing temperatures can have on preclinical metabolic research.

2. Aerobic exercise-induced browning, or lack thereof, in subcutaneous white adipose tissue at thermoneutrality

One of the timeliest examples of sub-thermoneutral housing temperatures confounding preclinical metabolic research comes from the exercise biochemistry literature and concerns the notion that aerobic exercise training induces the browning of white adipose tissue (i.e. WAT). We have extensively reviewed the history and mechanisms underlying aerobic exercise-induced browning elsewhere [29], but in brief, ‘browning’ is the process whereby UCP1 + brown-like adipocytes proliferate within canonically non-thermogenic WAT depots in response to repeated adrenergic stressors. This phenomenon is almost entirely exclusive to rodents and is uniquely localized to their inguinal subcutaneous WAT (i.e. iWAT) depot.

The aerobic exercise-induced browning of WAT is a highly reproducible phenomenon in rodents [30–36] but it has not been faithfully recapitulated in the human literature [37–41]. As previously mentioned, we have reviewed the mechanisms underlying the aerobic exercise-induced browning of WAT [29] and contend that any stimulus inducing adrenergic stress (i.e. aerobic exercise or ambient cold exposure) is sufficient to induce the browning of WAT in mice. In line with this, we suspected that the reports of aerobic exercise-induced browning in rodents were incongruous with the human observations because mice were being exposed to a chronic adrenergic stressor (i.e. sub-thermoneutral housing temperatures). To test this hypothesis, we housed male C57BL/6 mice at ambient vivarium room temperature (i.e. 22°C) or at thermoneutrality (i.e. 29°C) and either gave them access to a voluntary running wheel for six weeks or not [11]. In an additional experiment, we also subjected male C57BL/6 mice to an acute exhaustive bout of treadmill running. In this way, we could test the effects of ambient housing temperature on both the acute and chronic metabolic adaptations to aerobic exercise in WAT. Confirming our hypothesis, housing temperature influenced the biochemical adaptations to acute and chronic exercise in WAT such that markers of browning were absent from the iWAT depot in mice housed at thermoneutrality but not room temperature, indicating that what previous publications [30–36] reported to be aerobic exercise-induced browning in rodent iWAT was most likely an artefact caused by the interaction of chronic thermal stress and training. Importantly, markers of oxidative metabolism were still increased in the iWAT depot of mice following aerobic exercise training, an observation consistent with recent human work also demonstrating that oxidative, but not thermogenic, markers are upregulated in abdominal subcutaneous WAT following aerobic exercise training in both men and women [39,41]. Thus, the primary implication of our work [11] is that aerobic exercise-induced browning is likely an artefact of poorly controlled experimental conditions and as such previously published studies on aerobic exercise and adipose tissue metabolism have been confounded by sub-thermoneutral housing temperatures. Moreover, our work extends prior observations [42] that question the physiological relevance of browning in iWAT, independent of aerobic exercise (figure 1).

Figure 1.

Aerobic exercise training in mice housed at sub-thermoneutral, but not thermoneutral, ambient temperatures leads to the browning of white adipose tissue. At sub-thermoneutral ambient temperature (i.e. room temperature), VWR causes the browning of iWAT, morphologically characterized by an increase in multilocular adipocytes and biochemically characterized by increased UCP1-dependent thermogenesis. However, when this chronic thermal stress is alleviated by housing mice at thermoneutrality, these ‘training adaptations’ are no longer apparent.

Richard's group recently demonstrated that at thermoneutrality the browning of iWAT is dissociated from any measurable increase in the oxidative activity of the tissue, suggesting that the browning of WAT is not sufficient to increase thermogenesis [42]. Through bilaterally denervating the brown adipose tissue depots of mice that were subjected to a short-term sub-thermoneutral cold stress, and measuring oxidative activity in iWAT in vivo, these authors were able to determine whether the cold-induced browning of iWAT was sufficient to increase the thermogenic activity of the tissue, independent of the confounding effects of the brain favouring the recruitment and activation of brown adipose tissue thermogenesis [43,44]. Despite the ability of ambient cold to stimulate browning in iWAT, this adaptation was dissociated from any measurable increase in the oxidative activity of the inguinal depot assessed via dynamic uptake of [¹⁸F]FDG and [¹¹C]acetate in vivo, strongly questioning the physiological relevance of the browning of WAT.

3. The contributions of inguinal white adipose tissue to the metabolic health benefits of aerobic exercise training are not physiologically relevant

Aerobic exercise training can induce favourable adaptations in iWAT [31,33,35,36,45,46], and these adaptations mediate improvements in whole-body glucose homeostasis following aerobic exercise training [35,47,48]. However, this notion is inconsistent with accumulating evidence from our group [11,36] and others [42] demonstrating that iWAT has a negligible role, if any, on whole-body metabolism.

The notion that iWAT mediates the beneficial effects of aerobic exercise training was initially established when Goodyear's group ‘transplanted’ approximately 850 mg of iWAT, more than double what is found in a sedentary chow-fed male mouse at thermoneutrality [11], from exercise-trained mice into sedentary recipients and observed favourable alterations in glucose homeostasis. More specifically, after demonstrating that 11 days of voluntary wheel running was sufficient to drive oxidative and thermogenic adaptations in iWAT of mice housed at sub-thermoneutral temperatures, the inguinal depots from exercise-trained donor mice were then ‘transplanted’ into the abdominal cavity of sedentary recipient mice and transient peripheral improvements in glucose tolerance and insulin sensitivity were observed [35]. Importantly, these improvements were absent when the inguinal depots of sedentary mice were transplanted into recipient mice, suggesting that the peripheral improvements in glucose homeostasis were due, at least in part, to aerobic exercise-induced adaptations present within the transplanted iWAT.

Although the exercise-induced adaptations in iWAT were shown to be sufficient for the beneficial improvements in whole-body glucose metabolism [35], Peppler et al. have shown more recently that these adaptations are not required for the metabolic health benefits of aerobic exercise training [36]. Having replicated the findings of Stanford and colleagues [35], that 11 days of voluntary wheel running is in fact sufficient to drive the induction of oxidative and thermogenic adaptations in iWAT of mice housed at sub-thermoneutral temperatures, Peppler et al. [36] then employed a surgical ablation approach performing lipectomy or sham surgeries on mice to remove their iWAT depots. Interestingly, when the sham and lipectomized mice were then given access to voluntary running wheels for 11 days both groups exhibited similar exercise-induced changes in body mass, carbohydrate utilization, as well as glucose and insulin tolerance. Moreover, these adaptations in the lipectomized group were not explained by compensatory alterations in other tissues, effectively highlighting that aerobic exercise-induced adaptations in iWAT are dispensable for the whole-body metabolic benefits of aerobic exercise [36].

Despite this fact, recent work from Goodyear's group [47,48] perpetuates the idea that iWAT mediates the beneficial metabolic effects of aerobic exercise via the secretion of adipokines such as transforming growth factor-β2 (i.e. TGFβ2) and cysteine-rich secretory protein 1 (i.e. Crisp1). This group observed increased mRNA expression, protein content and serum concentrations of TGFβ2 in iWAT following 11 days of voluntary wheel running in mice, and through a lactate dependent muscle–adipose tissue crosstalk mechanism demonstrated that aerobic exercise-induced increases in TGFβ2 improved glucose tolerance in vivo [47]. These experiments were conducted in mice housed at sub-thermoneutral temperatures and the authors noted that, in contrast with the majority of the literature [11,49–51], mice housed at thermoneutrality ran approximately 42% less than those at room temperature. We failed to observe an aerobic exercise training-induced increase in serum TGFβ2 concentrations in mice housed at thermoneutrality [11] and, interestingly, moderate-intensity aerobic exercise training in human participants also failed to increase serum concentrations of TGFβ2 [47], supporting our argument that sub-thermoneutral housing confounds the adaptive response to aerobic exercise training in rodent WAT. Similarly, the same group recently identified Crisp1 as a novel androgen-mediated sex-specific aerobic exercise-inducible adipokine that regulates browning and glucose metabolism in differentiated white adipocytes and 3T3-L1 adipocytes, respectively [48]. Again, this work was conducted at sub-thermoneutral temperatures and needs to be validated at thermoneutrality to establish its translational relevance. Nevertheless, should these results prove reproducible at thermoneutrality their physiological relevance would not be clear, as we have shown that aerobic exercise training improves glucose tolerance independent of the browning of iWAT in mice housed at thermoneutrality [11] and, most importantly, that iWAT is dispensable for the beneficial metabolic effects of aerobic exercise training in mice [36].

Despite compelling evidence that iWAT is dispensable for the beneficial metabolic effects of aerobic exercise training in mice, most of the aforementioned studies are only pertinent to male WAT metabolism, and seeing as the browning of WAT occurs in a sexually dimorphic and depot-specific manner [52], this conclusion cannot be applied to females with confidence until further work is done to assess these sex-specific differences. To this end, a recent study conducted at thermoneutrality by Clart et al. reported that in ovariectomized mice bred to lack the oestrogen receptor beta DNA-binding domain, aerobic exercise training failed to increase oxidative metabolism and markers of thermogenesis in the browning susceptible perigonadal WAT depot, relative to trained ovariectomized wildtype littermate controls [53]. Interestingly, there were no differences in whole-body energy expenditure between genotypes despite the lack of perigonadal WAT browning reported in the oestrogen receptor signalling knockout mice. This lack of aerobic exercise-induced browning occurred concomitant to increases in surrogate markers of WAT and whole-body insulin resistance, leading the authors to conclude that genomic oestrogen receptor signalling causally links the browning of perigonadal WAT to insulin sensitivity. Importantly, ovariectomized-induced insulin resistance is exacerbated in UCP1 knockout but not wild-type mice, consistent with the notion that UCP1 plays a metabolically protective role in the absence of ovarian hormone production [54]. Thus, it is of critical importance to determine whether aerobic exercise training can cause WAT browning in females as this could potentially combat the development of insulin resistance that is commonly observed during menopause [55,56].

4. Conclusion

Mice have become the most widely used model organism for human biology [57] and fundamental differences in thermal biology between these species need to be acknowledged if human metabolism is to be accurately phenocopied in mice. Using exercise biochemistry as an example, we have highlighted the confounding effects sub-thermoneutral housing temperatures can have on adipose tissue metabolism in mice. Going forward, we urge those studying adipose tissue metabolism to consider ambient housing temperature as a key determinant in the methodological conception of their studies [11] and expect academic journals to update author guidelines to reflect this. Understandably, there are sufficient reasons for housing rodents at sub-thermoneutral temperatures; however, if researchers are studying rodent white adipose tissue metabolism then their rodents ought to be housed at thermoneutrality, and if researchers are not housing their rodents at thermoneutrality, then the data must be viewed with caution.

Data accessibility

This article has no additional data.

Authors' contributions

G.L.M. and D.C.W. wrote and edited this manuscript together.

Competing interests

We declare we have no competing interest.

Funding

This work was funded by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant to D.C.W., who was a Tier II Canada Research Chair in Lipids, Metabolism, and Health; G.L.M. was supported by an NSERC Canada Graduate Scholarship.