“Humanizing” mouse environments: humidity, diurnal cycles and thermoneutrality

Abstract

Thermoneutral housing has been shown to promote more accurate and robust development of several pathologies in mice. Raising animal housing temperatures a few degrees may create a relatively straightforward opportunity to improve translatability of mouse models. In this commentary, we discuss the changes of physiology induced in mice housed at thermoneutrality, and review techniques for measuring systemic thermogenesis, specifically those affecting storage and mobilization of lipids in adipose depots. Environmental cues are a component of the information integrated by the brain to calculate food consumption and calorie deposition. We show that relative humidity is one of those cues, inducing a rapid sensory response that is converted to a more chronic susceptibility to obesity. Given high inter-institutional variability in the regulation of relative humidity, study reproducibility may be improved by consideration of this factor. We evaluate a “humanized” environmental cycling protocol, where mice sleep in warm temperature housing, and are cool during the wake cycle. We show that this protocol suppresses adaptation to cool exposure, with consequence for adipose-associated lipid storage. To evaluate systemic cues in mice housed at thermoneutral temperatures, we characterized the circulating lipidome, and show that sera are highly depleted in some HDL-associated phospholipids, specifically phospholipids containing the essential fatty acid, 18:2 linoleic acid, and its derivative, arachidonic acid (20:4) and related ether-phospholipids. Given the role of these fatty acids in inflammatory responses, we propose they may underlie the differences in disease progression observed at thermoneutrality.

1. INTRODUCTION

1.1. The systemic thermogenic reaction that maintains mammalian body temperature is energetically expensive.

Depending on a variety of environmental factors, mammals lose body heat at all temperatures below their “thermoneutral” temperature. This loss is countered by heat production, generally fueled by lipids. Assuming that heat production is supported mostly by uncoupled mitochondria in brown and beige adipose depots, the assay of energy expenditure (EE) by respirometry reveals the proportion of energetic budget allocated to heat production. Mice placed in metabolic chambers and subjected to a gradual decrease in environmental temperature show an almost 2-fold increase in oxygen consumption (VO2) between thermoneutral housing (~29°C) (1) and room temperature (~21°C) (Fig. 1A).

Fig. 1. Thermoneutrality and assay of energetic demands.

A. Assaying thermoneutrality. Thermoneutral temperature is typically measured as the minimum energy expenditure (measured by respirometry) for a range of environmental temperature exposures for a given mammal. At this temperature, the mammal is assumed to have zero allocation of energy to thermogenesis. In this example, at the beginning of their sleep cycle (7 am), BALB/cJ female mice housed in metabolic chambers at thermoneutrality were exposed to seven 1°C hourly decreases of temperature to room temperature. B. Contrasting environmental thermoneutral temperatures for mouse and human. The temperatures used in this study include room temperature (21–24°C / 70–75°F), which is thermoneutral for clothed humans, and 29–31°C / 84–88°F, which is thermoneutral for the majority of mice with hair. This probably reflects their different surface area/volume ratio. C. Thermogenesis involves paracrine interactions of several adipose depots (scWAT, subcutaneous white adipose tissue or iWAT, inguinal WAT; BAT, brown adipose tissue), other tissues as indicated, and the hypothalamus (also discussed by (39)). D. Hair as a heat regulator. Mouse hair serves the same function as human clothes, reducing the demand for thermogenesis. BALB/cJ mice housed (as for Fig. 1A) were either shaved before assay, or not; mouse cohorts n>6.

1.2. Thermoneutral temperatures are different for mouse and human.

Clothed humans have designed their environment around “room temperature” because this is comfortable and thermoneutral. Van Marken Lichtenbelt and colleagues showed that EE plateaued between 19.2 C and 26.6°C for a group of young white males, mostly naked and lean (2). The upper critical temperature for this group was difficult to determine, presumably because of effective heat loss strategies (such as sweating). Therefore, at 24°C, clothed, humans are thermoneutral, and BAT is de-activated (3). Mice on the other hand are thermoneutral at higher ambient temperatures, approximately 29°C/85°F, the exact temperature depending on size, age, developmental conditioning, sex and pelt, amongst other factors. One of the contributions to higher heat loss in mice is the ~10-fold higher surface area/volume (Fig. 1B). In other words, most mice housed under standard environmental conditions spend their lives with the thermogenic program on.

1.3. Heat production is a systemic endocrine process.

There is a profound difference between life spent with the thermogenic reaction on or off, because many tissues are influenced by the systemic factors that induce, fuel and regulate thermogenesis, including β-adrenergic effectors, growth factors and lipids. The two best-known heat-producing adipose depots are brown adipose tissues (BAT), and beige adipose depots. In mice, these latter are often observed as the inguinal subcutaneous white adipose depot, one of the 10 subcutaneous white adipose fat pads (scWAT) of mice (Fig. 1C). Under hypothalamic control, β-adrenergic effectors induce these two adipose types to produce heat towards a homeostatic body temperature set-point. Activated thermogenic adipose depots are highly endocrine organs, producing a secretome of bioactive lipids and cytokines (4). Heat production consumes lipids, either mobilized from β-adrenergic-sensitive storage depots (scWAT) (5), or extracted from food in intestine (6). By a process of cold-regulated, receptor-mediated uptake, lipids are taken up, remodeled in intracellular lipid droplet reserves, and delivered to uncoupled mitochondria. As the pre-eminent metabolic coordinator, liver is involved in directing lipid trafficking, and the hypothalamus orchestrates body temperature homeostasis (7,8). However, the interaction is not limited to just these tissues, since many cell types express β-adrenergic receptors, or are modified by changing lipid flux, or express receptors for growth factors with effects on metabolism, such as insulin, Fgf21, leptin and adiponectin.

1.4. Skin is the main barrier to heat loss, and alterations in skin properties have profound effects on the set point for thermogenesis.

The main barrier to heat loss is skin. This barrier is finely tuned and highly regulated, releasing heat through evaporative cooling, convection and radiation (9,10). Any change that affects the thermodynamic properties of mouse pelt, such as fur density or permeability, also affects energy expenditure and global metabolism (11,12). For example, if mice are shaved, energy expenditure goes up and thermogenesis is hyper-activated: even the basal metabolic rate is increased (Fig. 1D).

1.5. Different animal species, or individuals within that species, can have different thermogenic strategies.

Various organs involved in the thermogenic circuit are diagrammed in Fig. 2. The best understood thermogenic strategy relies on an uncoupling protein-1 (UCP1)-dependent short-circuit of the mitochondrial membrane potential reaction in BAT (13). This reaction relies on the transcriptional induction of the UCP1 gene/protein and allosteric modulation of UCP1 function. Alongside UCP1, there are a series of other lipid metabolic genes also induced, such as Scd1 and Elovl3, albeit with less clearly defined roles for thermogenesis (14,15). Together, the expression of these genes is used as a genetic signature of thermogenic activation of mouse BAT. However, although mice have highly developed BAT, other species have not, for example pig (16). Imaging of cool-activated BAT using uptake of a radio-dense glucose (18F-fluorodeoxyglucose) by PET suggests that some humans activate BAT, others do not (17–19). To date, the reason for this is unknown but we could speculate that this reflects individual metabolic specialization, genetic variation, seasonal acclimation, relative exposure to cool temperatures or developmental conditioning. Note that it is more difficult to observe FDG accumulation in BAT from obese individuals (20).

Fig. 2. Various thermogenic strategies are available.

To maintain body temperature homeostasis, mice deploy both acute and chronic thermogenic strategies. BAT is rapidly activated to provide acute control (within minutes), whereas subcutaneous white adipose (scWAT) adapts to defend the average environmental challenge. Heat is produced mostly by non-UCP1 dependent futile cycles (non-canonical thermogenesis, ncT) in scWAT and muscle, perhaps other tissues. Together with dietary sources, scWAT is a prominent source of fatty acids for UCP1-dependent BAT combustion. Liver coordinates mobilization of glucose (from glycogen), de novo lipogenesis and VLDL release, and secretion of signaling cytokines such as Fgf21. Skin is a highly homeostatic organ, which responds to conserve heat when environmental temperatures decrease, reducing the demand for heat production. Ultimately, the hypothalamus determines the total heat production strategy.

For species and human individuals who do not contain specialized BAT, alternative mechanisms are activated to compensate (21,22). Under the blanket term of “UCP1-independent” thermogenesis (non-canonical thermogenesis; ncT), several mechanisms rely on the acceleration of futile biochemical reactions to make heat (23–26). For example, futile biochemical cycles associated with maintenance of Ca²⁺ gradients in the ER of muscle can be key to cold tolerance (27). In UCP1−/− mice, ncT is enough to maintain body temperature and allows the survival of mice at sub-thermoneutral temperatures. The activation of ncT relies on a circulating hormone, Fgf21, associated both with fasting and thermogenesis (28). Interestingly, when fed high fat diet, UCP1−/− mice are naturally resistant to diet induced obesity, implying that the ncT strategy for heat production promotes leanness, actively suppressing obesity. Thus, only when these reactions are turned off, at thermoneutrality, do mice have a positive energy balance and become obese (28–32).

1.6. Assay of thermogenesis.

Although indirect, many tissues and processes implicated in the thermogenic process can be scored as follows: 1) Direct visualization of BAT activation using thermography (forward-looking infrared; FLIR). Although this assay is only useful for hairless mice, the highest body surface temperature measured is typically quantitative and proportional to ambient temperature (Fig. 3A,B). This technique is non-invasive, and able to provide longitudinal data, if mice remain calm and unstressed. However, although surface temperature is increased by BAT activation, surface temperature is decreased by insulation in the skin, and also by evaporative cooling, which have opposite impact on heat loss, making interpretation of physiological state challenging (10). 2) Measurement of energy expenditure (EE). Although total energy expenditure (EE) can be related to ambient temperature (Fig. 3C), EE is a complex sum of processes, many adaptable to diet and environment. For example, body temperature may be raised in cool temperatures, generating excess energy expenditure (Fig. 3D). Therefore, neither the rate of thermogenesis, or component parts of the equation such as insulation can be accurately deduced from this assay alone. For example, the conclusions of the classic work of Scholander which attempted to deduce relative insulation from oxygen consumption measurements of different animals are unlikely to be correct (33). 3) Histologic assay of lipid reservoirs. Perhaps most robust, the lipid droplet content of BAT is a useful indicator of active thermogenesis (Fig. 3E), although this comes with certain assumptions. Principal amongst these is that an increased rate of thermogenesis will lead to a decreased size of the lipid droplet reservoir: this relies on our incomplete understanding of the rate determining steps which govern lipid uptake, remodeling, lipid droplet assembly and mobilization. 4) Transcriptome changes. Transcriptional changes occur as a component of the UCP1-dependent response. Indeed, UCP1 mRNA is typically >100x induced between thermoneutral mice and mice exposed to cold stress. Note that the presence of UCP1 mRNA is not quantitatively related to the production of heat (34), and indeed heat production has never been directly measured.

Fig. 3. Physiological response to room temperature.

The specific experimental conditions used for the mouse data shown here are as described by Kasza et al (10). A. Assay of thermogenic activation by FLIR. Assay of hairless SKH1 mice with a thermographic camera illustrates the activation of BAT (here 10 mins after transfer from a thermoneutral room to room temperature). This technique is non-invasive and can be used to generate longitudinal data, if mice remain calm and unstressed. B. At steady state, the highest surface temperature (skin over BAT) is proportional to the relative degree of activation of thermogenesis. However, surface temperature is a complex output: it is increased by the temperature of the underlying tissue, decreased by skin insulation (conserving energy), and decreased by evaporative cooling (draining energy), leading to paradoxes of interpretation. Any wet surface, such as eye, is especially vulnerable to this issue. C. Energy expenditure, measured in metabolic cages, reflects the activation of mitochondrial consumption of O2, which increases with decreased environmental temperature. D. Body temperature (BT) may change in mice during activation of thermogenesis, depending upon age, sex, pre-conditioning (in say cold temperatures), age, stress, strain, or latest meal. In this case, activation of thermogenesis results in BT elevation from 37°C for mice housed at thermoneutrality to 38.2°C in mice housed at room temperature. E. Assay of lipid droplet reserves. There is a proportional depletion of lipid reserves in BAT, conveniently assayed by image analysis of H&E-stained sections (data shown derived from mice analyzed for E-H). This assay can be confirmed by assay of molecular markers, such as Western blotting for UCP1 (10).

1.7. Assay of UCP1-independent thermogenesis.

The processes included in ncT depend upon the acceleration of pre-existing processes, relying upon futile cycles of consumption and re-synthesis, or uptake and release (35,36). These processes could be entirely biochemically controlled, and under-represented by assay of transcriptional activation. Kozak and colleagues showed that despite metabolic activation and increased O₂ consumption in muscle after cold exposure, there was little change of gene expression (22). This study associated the activation of scWAT with activation of the metabolic checkpoint protein, AMPK, phospholamban (a regulator of Ca²⁺ transport; PLN), and induction of 2-deiodinase (DIO2) mRNA. Many current studies rely instead upon the induction of a “diluted” BAT-associated browning signature. Thus, qPCR assay of scWAT typically reads out >10,000x less induction of thermogenesis signature genes compared to corresponding BAT (Elovl3, UCP1, Dio2, FASN, CideA) (28). The contribution of ncT to total heat generation is difficult to deduce: for example, since these reactions are not directly related to mitochondrial uncoupling, the stoichiometry of the oxygen consumption measured by respirometry has not been defined. This may explain why there are paradoxical gaps in energy expenditure assay results.

1.8. Mammalian brains perform complex integrated calculations to determine how much to eat.

Food consumption increases to provide energy for thermogenesis (Fig. 4A, B); on the other hand, mice can titrate their consumption of calorie dense (high fat) food pellets to eat fewer pellets than for regular chow (see example compare food intake (grams) with calorie intake adjusted for high fat content in Fig. 4C). The brain, via the coordinating (unconscious) hypothalamus, is responsible for balancing various inputs to come up with an appetite suitable to maintain a lean body composition (37,38). This calculation appears to be imperfect; for these examples, raising the environmental temperature from 20°C to 23°C (both temperatures in the average room temperature range of institutional vivaria) is enough to increase body weight for chow-fed mice by almost 1g (Fig. 4A). The energetic “pie” is diagrammed in Fig. 4D, to illustrate all the different, and modifiable processes fueled by calories. The prioritization of calorie delivery with respect to thermogenic demand and food calorie density is obviously complex (39). Nonetheless, the programming rules that govern this complex brain calculation are important, because they are widely considered to underlie the inadvertent imbalances that result in the development of obesity (38,40–42).

Fig. 4. Mice recalculate the fate of food calories with altered energy demands.

A. Vivaria room temperatures. The range of “room temperature” in general use changes calorie intake and utilization. BALB/cJ females (3 months old, n=8) continuously housed in environmental chambers, and adjusted to cool (20°C) and warmer (23°C) environments for 2 weeks show reduced food consumption but higher body weights in the warmer rooms. B, C. Mice recalculate food consumption to match diet and environment. Daily food intake of female BALB/cJ mice was increased by >50% in room temperature (21°C), compared to thermoneutral housing (29°C; B). C. Fed with either chow or a high fat diet, mice consume less of the more energy-dense high fat diet (grams) to arrive at the same daily calorie consumption (kcals) (n>6). D. Process-specific, adaptable energy expenditure. Food consumption responds to energy expenditure by different tissues, and upon demand. This scheme illustrates modifiers of energy expenditure, including basal metabolic rate, anabolism, thermogenesis and other “on-demand” activities such as exercise, immune response, complex brain functions and digestion. This is a complex supply chain, with feedback effectors from consumer tissues targeting hypothalamus and other tissues, to provide a specific unconscious demand for food. Depending on the priority of signals, calorie accumulation (obesity) and other homeostatic processes may be set at different levels.

1.9. De-activating the thermogenic circuit at thermoneutrality.

At thermoneutrality, many diverse aspects of mouse physiology are affected (Fig. 2). The activated BAT secretome is suppressed (4,43), immune cell function is activated (44–46), lipid metabolism is reorganized by liver and gut (47), the skin barrier is more permeable (our data, unpublished), and UCP1-dependent and - independent metabolism in accessory adipose depots and muscle is deactivated (28). Diverse cues from brain modify food consumption, vasoactivity, heartbeat and compensatory, tissue-specific signals. Various studies have shown that de-activating thermogenesis leads to dramatic differences in disease processes, including those regulated by metabolism and/or β-adrenergic sensitive immune cells, such as cancer, obesity, atherosclerosis, steatosis and infectious disease transmission (44,45,48–52). For example, myeloid derived suppressor cells (MDSCs) and CD8(+) effector T cell function are inhibited when thermogenesis is activated, leading to increased rates of tumor growth and metastasis (44,45,49). Likewise, differences in drug toxicity that are often attributed to co-housing stressors are more likely influenced by thermogenic status (53).

When mice (or rodents in general) are housed at thermoneutral temperatures, diseases tend to resemble their human counterparts better (54,55). For example, Divanovic and colleagues showed increased inflammation and progression of steatosis in mice housed at thermoneutrality (48). This study revealed that thermoneutral housing exacerbated nonalcoholic fatty liver disease in males and promoted the development of this pathology in females, which was absent in females housed at room temperature. Housing at thermoneutrality was found to lower cortisone levels, decrease intestinal barrier function, and augment pro-inflammatory immune response in response to high fat feeding.

1.10. Addressing gaps in our understanding of environmental impact on mouse physiology.

Here we characterize environmental factors that affect thermogenesis, which in turn affect systemic metabolism and disease progression. Our previous study identified evaporative cooling as a major energy drain for mice (10): when skins are more permeable, thermogenesis is induced to maintain body temperature, and energy expenditure increased. Since humid environments suppress evaporation, here, we evaluate the role of humidity in determining the rate of thermogenesis. We also draw attention to the typical “humanized” exposure to environmental temperature, which is to sleep wrapped up in bed, and be exposed to cooler temperatures (for example, outside) in the daytime. This contrasts with the constant temperature housing typically experienced by rodents in vivaria, albeit modified by the number of mice in a cage, enrichment devices such as nest-building materials or behaviors (socialization, tail tucking, exercising). As a means to understand the differences between mice housed at thermoneutrality versus room temperature, we characterize the circulating lipidome, and reveal systemic differences in lipids and lipoproteins.

2. MATERIALS AND METHODS

2.1. Ethical Approval. Mice.

These studies were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Experimental protocols were approved by the University of Wisconsin School of Medicine and Public Health Animal Care and Use Committee. The number of mice used to perform this study was minimized, and every effort was made to reduce the chance of pain or suffering. Method of euthanasia was CO₂ asphyxiation, as per guidelines. All authors understand the ethical principles and confirm that this work complies with the animal ethics checklist.

2.2. Mice.

Mouse strains and sexes are specified in the results and figure legends: SKH1-Elite (abbreviated SKH1 mice; Charles River; stock#477; Crl:SKH1-Hrhr) are an immunocompetent but hairless outbred strain of mice, and BALB/cJ (Jackson labs cat#00651). Unless otherwise specified, animals were housed at constant temperature (19-23°C) in 12 h light/dark cycles with free access to water and standard chow (Harlan Teklad Global Diet 2018), or high fat diet (HFD; Harlan Teklad diet#TD06415, 45% calories from fat). The use of metabolic cages was described in Kasza et al (56), and body temperature was measured with a rectal probe. Environmental responses were measured for singly housed mice in environmental chambers (Memmert HPP750 or Caron 7350).

2.3. FLIR imaging.

To measure surface temperatures by infrared thermography, we used a hand-held FLIR T360 camera (FLIR Systems, Oregon). Pin drops in the software of the FLIR camera were used to record surface temperatures and quantified using FLIR Tools Advanced Thermal Analysis and Reporting software; each photograph is internally and externally calibrated to show actual temperatures (10).

2.4. Histological Analysis.

Skin, BAT, perigonadal WAT (pgWAT) and the inguinal (mammary) fat pads (scWAT) were dissected for histological processing as described in Kasza et al (5). Briefly, paraformaldehyde-fixed, paraffin-embedded samples were H&E stained and assayed as follows: 1) dWAT thickness: 6 images of H&E-stained, non-anagen fields of skins (equivalent to ≥4500 μm linear dWAT) were assayed by image analysis (dividing total area by length). 2) Assay of BAT lipid stores: Lipid droplets were identified in gray scale images using circularity (0.1-1.0) and diameter (0.1-50 μm) thresholds in 6 independent fields of BAT (>1200 cells) and quantified using the open-source Fiji image processing package (https://loci.wisc.edu/software/fiji); data is expressed as average lipid droplet size (μm²). 3) Perigonadal (pg) WAT adipocyte size assay: 3-6 images across the length of the fat pad were scored for adipocyte size (measured as number of adipocytes/area scored) in each of 4-6 mice (total adipocytes scored >1500/mouse).

2.5. LC/QTOF-MS mass spectrometric analysis of lipids in sera.

Lipidomic analysis of sera was performed according to Simcox et al (57,58); briefly, 40 μl aliquots of sera were combined with 250 μls PBS and 225 μls ice-cold MeOH containing internal standards (Avanti Splash cat#3307-07), and homogenized. Samples were then mixed with 750 μls of ice-cold MTBE (methyl tert-butyl ether), rehomogenized, and separated by centrifugation (17,000g for 5 mins/4°C). The upper phase was transferred to a new tube, lyophilized and resuspended in 150 μls of isopropanol. Lipids were analyzed by UHPLC/MS/MS in positive and negative ion modes, at a dilution suitable to eliminate saturation issues. Extracts were separated on an Agilent 1260 Infinity II UHPLC system using an Acquity BEH C18 column (Waters 186009453; 1.7 μm 2.1 × 100 mm) maintained at 50°C with VanGuard BEH C18 precolumn (Waters 18003975), using the chromatography gradients described (58). The UHPLC system was connected to an Agilent 6546 Q-TOF MS dual AJS ESI mass spectrometer and run in both positive and negative modes as described. Samples were injected in a random order and scanned between 100 and 1,500 m/z. Tandem MS was performed at a fixed collision energy of 25 V. The injection volume was 2 μl for positive mode and 5 μl for negative mode.

Lipidomic data processing.

Methods for data processing were described by Jain et al (58). Briefly, MS/MS data were analyzed using Agilent MassHunter Qualitative Analysis and LipidAnnotator for lipid identification (59). Accuracy of retention times was checked by reference to internal standards. Data was imported into Agilent Profinder for lipid identification and peak integration (using sera-specific libraries). Data were analyzed using MetaboAnalyst free-ware (https://www.metaboanalyst.ca), Microsoft Excel and GraphPad Prism8 software (https://www.graphpad.com/scientific-software/prism).

2.6. FPLC analysis of lipoprotein fractions.

Size-exclusion chromatography of mouse plasma was performed on an AKTA Fast Protein Liquid Chromatograph (FPLC; Amersham Pharmacia Biotech). Equivalent volumes of plasma from each group of mice were pooled, totaling 350 μl. Plasma was diluted in PBS (pH 7.4, with 0.02% sodium azide) to a total sample volume of 1000 μl, injected into the FPLC instrument and fractionated through a Superose 6 10/300 GL followed in tandem with a Superdex 200 10/300 GL column. Forty-eight 0.5 ml fractions, containing separated lipoprotein classes in 10mM PBS, pH 7.4, were and analyzed for cholesterol and triglycerides using reagents from Pointe Scientific (respectively cat#G7521 and #T7532).

2.7. Statistical Analysis.

The statistical tests appropriate to each analysis are indicated in Figure legends. To test for normal or lognormal distribution of sample values we used the Anderson-Darling test, outliers were identified using ROUT method. Box-and-whiskers graphs show median values with 10-90 percentile whiskers; other data are expressed as mean ± standard deviation, unless specifically stated.

3. RESULTS

3.1. Humidity is not typically controlled in vivaria.

For the sake of reproducibility, accuracy and ethics, rodent caging is highly specified within limits of air quality, temperature, animal density and enrichment (52). However, mice housed in compliant caging can still experience a large range of conditions that affect heat loss, including air temperature, air speed (for high density caging), whether the cage is covered, the number of mice cohoused, and the presence of nesting materials and enrichment igloos (Fig. 5A). For example, the rate of air flow typically used in vivaria has been identified from study of the effect of air flow on mouse physiology (52). However, relative humidity (RH) has not been considered, beyond basic regulatory guidelines proposing that RH fall between 30-70%, notably to control diet consistency and mold at the high humidity limit and skin and lung irritation at the low humidity limit.

Fig. 5. Humidity can be highly unregulated in vivaria.

A. Thermogenic activation in mouse subjects. The average demand for thermogenesis is modified by a range of environmental factors, including the specific “room temperature” (RT) selected, the air speed chosen to eliminate accumulation of toxic metabolites (typically ammonia), whether the cage is covered, the number of mice cohoused and whether they are provided with igloos. When air is injected from outside environments, it may or may not be modified for humidity. B. Relative humidity. Water content increases quadratically with temperature over the range of outdoor temperatures typical for the mid-west (blue bar on x axis). Without humidification, the maximum water content of cold air from outside on a winter’s day is only 10% of maximum water content of inside air (shown as the grey arrow), thus the airflow is drying and cooling. The air water content with respect to actual temperature is expressed as the relative humidity (RH) and external humidity varies widely along with temperature (the “weather”). C. Relative humidity of a vivarium at the University Wisconsin-Madison. Daily means of yearlong RH are plotted, together with the seasons, average RH and standard deviation. D, E. Similar data for vivarium units from Jackson Labs (Maine, D and California, E).

Relative humidity describes the water-carrying content of air at different temperatures. There is a non-linear relationship of water content with air temperature, which can conspire with outdoor weather patterns to generate extreme swings of environment for caged mice subjected to outside air sources (Fig. 5B). For example, at 21°C, the maximum water carrying capacity of air is 20 g/m³: incoming air on a typical winter day in Wisconsin has a maximum capacity of only 2 g/m³, and the relative humidity experienced by the mice in our vivarium can drop to only 13% (Fig. 5C). This is well outside the recommended range, and indeed mice in this University of Wisconsin vivarium spend >60% of their time outside of this range. Interior relative humidity is closely correlated with the seasons, which probably explains why mouse colonies can sense and react to the time of year (for example for breeding purposes), regardless of our attempts to over-ride that and provide year-long consistency.

Even with less extreme weather conditions than the mid-west, a survey of vivaria around the US showed there were high fluctuations in relative humidity. These included UC Davis (California central valley, data courtesy of Dr. Jon Ramsey) and Emory University (Atlanta Georgia, data courtesy of Drs. Michael Huerkamp and Ray Dingledine).

The Jackson laboratory breeding facilities show much better humidity regulation than many individual institutions; year-long data shows that the room air for Maine-and California-based vivarium facilities maintain 35-60% RH more than 85% of the time in Maine, and all of the time in California (courtesy of the Jackson Lab; Fig. 5D, E). This is maintained by active humidification of zonal air input when winter-time humidity drops below 45%, and by condensation of excess water by chilled water coils in the summer. More advice on modifying or designing physical plant for air conditioning in vivaria are available direct from the Jackson Lab.

3.2. In cool conditions, does humidity matter?

From a human point of view, relative humidity is most often considered in the context of high temperatures, when high relative humidity increases the “feels-like” temperature and makes body cooling difficult (60). However, below thermoneutral temperatures, the impact of altered humidity is generally unknown. Previously, we showed that water lost by evaporation through mouse skins provides a significant energy drain, where high rates of water loss correlate with resistance to diet-induced obesity (10). This water loss is associated with the natural permeability of skin, thus described as “insensible” cooling; this is not related to sweating controlled by the autonomic nervous system (61). High relative humidity is predicted to suppress evaporation and therefore lower the demand for thermogenesis.

We tested the impact of different levels of relative humidity on thermogenesis and the consumption of food and water to assess mouse responses. Mice were moved from 40% RH, 20°C, which is an average and typical housing condition, to low (15%) or high (80%) RH / 20°C (both RH levels observed in our vivarium). Using food intake as a metric, mice reacted rapidly to altered RH (Fig. 6A). During a 1-2 day sensory window, mice exposed to low humidity consumed more calories, as predicted by the energy drain imposed by increased evaporative cooling (Fig. 6B). Mice fed high fat diet (HFD) over-eat during this window. Surprisingly, mice exposed to either environmental or dietary perturbation return to their normal food consumption after 2 days, suggesting that an adaptive response was activated.

Fig. 6. Mice sense altered humidity.

A, B. 3-month old BALBc/J female mice were housed in controlled environmental chambers for the times indicated, and transitioned from the average pre-condition (40% RH, 20°C) to either high (80% RH) or low (15% RH) at the same (cool) room temperature. Some cohorts were fed a high fat diet at the time of the transition, as indicated. C, D. Water consumption and skin permeability (TEWL) is also shown.

We hypothesized that mice may quickly reduce water loss in low humidity to account for this adaptation, however mice stimulated to drink more water by a move to low humidity continued to drink more even after the sensory period (by >40%, Fig. 6C). Skin permeability (measured as transepidermal water loss, TEWL) did not change during this window (Fig. 6D), although skin permeability did increase more slowly, after 2 weeks of exposure to high humidity. We conclude that suppression of water loss (and heat) is unlikely to explain the adaptation to contrasting RH.

To assess the thermogenic impact of high and low RH, we measured the lipid droplet reservoir in BAT. Immediately after transfer to 80% RH, and as predicted, mice showed significantly less thermogenic activation of BAT (scored as a larger lipid reservoir; Fig. 7A). However, BAT from mice housed at different RH were indistinguishable after the sensory period (2 weeks after exposure). Thus RH does not work as a straightforward suppressor of evaporation, to relieve the energy debt.

Fig. 7. High humidity is obesogenic.

A-C. Tissues from BALBc/J female mice (3–4 months old), housed at high and low humidity (per Fig. 5), were H&E stained for image analysis (A, BAT; B, skin, dWAT thickness; C, pgWAT); quantified on right hand side (n>6). D. BALBc/J female mice were fed high fat diet for 2 weeks at high and low humidity, and body weights measured (n>8).

Interestingly, the dermal white adipose thickened in response to both high and low RH (Fig. 7B). Although we expected the cool stress of low RH to induce thickening of dWAT (reported by Kasza et al 2014 (56)), we were surprised by the appearance of the same response in high RH. To test whether this reflected increased adiposity in general, we measured adipocyte size in perigonadal adipose: after 2 weeks, the average adipocyte size was larger for mice housed in 80% RH (Fig. 7C). This suggested that high humidity could be obesogenic. We therefore fed mice housed at high RH with a high fat diet and found that they gained more weight (3.92 ±1.17g compared to 0.78 ± 0.52g) than mice housed at low RH (Fig. 7D). Taken together, our data suggests that high RH is detected by mice and induces a reorganization of the cues that allocate fat to different depots, promoting obesity.

3.3. Environmental cycling restricts the adaptive responses to cool temperatures.

Vivaria maintain constant environmental temperatures, yet the majority of human subjects choose to sleep under blankets and be active in the day in cooler environments. We are frequently asked about the general impact of average climate temperature on human body composition and obesity, which can be almost irrelevant for human subjects who choose to minimize their exposure to outside climates. However, this does raise an important question, which cue, warm sleep or cool exposure, is dominant in the determination of thermal strategy?

To mimic a common human diurnal environmental exposure, we divided the housing cycle into a warm phase (thermoneutral, 31°C/88°F) and a cool phase (cool room temperature, 19°C/66°F; Fig. 8A), and moved mice between these environments, exposed to 12 hours of each. For comparison, cohorts of mice were housed continuously in cool or warm environments for 2 weeks.

Fig. 8. Temperature of sleep phase dominates thermogenic strategy.

A. Scheme of experiment. BALB/cJ female mice (n>6) were either housed at thermoneutrality (31°C), or at (cool) room temperature (19°C), or were cycled between these housing temperatures. These cycling cohorts spent identical time in each environment, but the “warm sleep” cohort was moved from 19°C to 31°C at 7 am (daytime sleep cycle) and back again at 7 pm. The cycling for the cool sleep group was inverted. B. Assays of thermogenic depots. BAT, dWAT thickness and pgWAT adipocyte size were measured 2 weeks after initiating the diurnal cycling of temperature.

The lipid stores of BAT were highly depleted in mice from the cool environment, compared to the thermoneutral warm environment (Fig. 8B). For mice exposed to a circadian cycle of temperatures, BAT from mice sleeping warm was less depleted than in mice sleeping cool. Since BAT lipid reservoirs are highly dynamic (depleting in minutes), we also assessed dWAT thickness, as a less labile marker of average thermogenic response. dWAT from mice with a warm sleep cycle was as thin as dWAT from mice housed warm continuously, confirming that mice sleeping warm do not show a cold signature. Interestingly, a warm sleep phase reduces pgWAT lipid load, where a cool sleep phase increases it, and we conclude that a cool sleep phase induces a redistribution of lipid into pgWAT and dWAT, where the latter hyper-insulates mice and exacerbates calorie conservation. We speculate that mice sleeping warm do not engage the adaptive thermogenic strategies that are associated with continuous cool exposure.

3.4. Are there changes in the supply of circulating lipids in mice housed at thermoneutrality?

Besides systemic changes in immune cell function, and pulses of β-adrenergic effectors, what other systemic changes exist to explain the diverse changes of physiology at thermoneutrality? Since lipid mobilization and delivery has a prominent role in thermogenesis, we tested whether steady-state lipids in sera differ significantly in mice housed at either 29°C (thermoneutral) or 20°C (room temperature).

Untargeted analysis of sera-based lipidomes identified 374 lipids (combined positive and negative modes), including (in positive mode) 9 acyl carnitines (ACar), 10 cholesteryl esters (CE), 5 ceramides (Cer), 10 di-acylglycerols (DG), 2 hexaceramides (HexCer), 41 lysophosphatidyl cholines (LPC), one lysophosphatidyl ethanolamine (LPE), 46 phosphatidyl cholines (PC), 3 phosphatidyl ethanolamines (PE), 1 phospatidylglycerol (PG), 24 sphingomyelins (SM) and 67 triglycerides (TG), and (in negative mode) 17 ceramides, 5 oxidized ether phosphatidylcholines (EtherOxPC), 33 ether phosphatidylcholines (EtherPC), 20 free fatty acids (FA), 4 fatty acyl esters of hydroxy fatty acids (FAHFA), 7 hexaceramides, 3 oxidized PCs (OxPCs), 29 phosphatidylethanolamines (PE), 10 phosphatidylglycerol (PG), and 23 phosphoinositides (PI).

Unsupervised clustering provided a robust separation of sera from male mice at the two housing temperatures (the top 25 differentially expressed lipids (DELs), as rated by their statistical significance, are shown as Fig. 9A). A highly characteristic signature emerged, summarized as the widespread depletion of specific phospholipid species in sera from mice at thermoneutrality (Fig. 9A, B). Even the most abundant phospholipids, including phosphatidylcholine (PC) 18:0 18:2, PC 18:0 20:4 and PC 16:0 20:4 were significantly depleted (Fig. 10A). Most significantly, all PC species with 18:2 or 20:4 acyl chains were depleted. Calculation of total acyl chain abundance for PCs (Fig. 10B) showed that sera from thermoneutral mice showed a 45% reduction of linoleic acid (18:2) and 55% reduction of arachidonic acid (20:4). These fatty acids are related, since arachidonic acid is generated from the essential dietary fatty acid, linoleic acid.

Fig. 9. A large group of phosphatidylcholine lipids are reduced in sera of mice housed at thermoneutrality.

BALB/cJ male mice (12 weeks old) were equilibrated to either room temperature (21°C) or thermoneutrality (29°C). Sera were analyzed by mass spectrometry, and 374 lipids identified from both positive and negative modes. A. Heatmap of 25 most significantly differentially expressed lipids, showing their consistent depletion in sera from mice housed at thermoneutrality. B. Volcano plot, to illustrate these changes in the context of other unchanging lipids, together with their statistical significance. Data were log2 transformed prior to fold change comparison and 2-sided t-tests were applied. An FDR correction was applied for multiple comparisons where q<0.15 was considered significant.

Fig. 10. The cluster of phosphatidylcholine lipids depleted in mice housed at thermoneutrality contain 18:2 or 20:4 acyl groups and include some of the most abundant species.

A. Abundance waterfall plot. Differences of abundance of each PC species is shown for male BALB/cJ mice, housed either at TMN or RT (n=4). PC 18:0 18:2, is reduced by nearly 50%, together with significant depletion of every other PC lipid conjugated with 18:2 and 20:4 acyl chains. B. Acyl chain abundance. Several acyl chains are significantly depleted (18:2, 18:0, 20:4 and 22:6) and arachidonic acid (20:4) is the most affected (55% depleted in PCs in sera from mice in thermoneutral environments). ** p<0.005; * p<0.05 (2-sided t tests).

Females also respond to environmental temperature, though response is more muted. Using the EtherPCs and FA subclasses of lipids, a heatmap illustrates the differential response of males and females to TMN and RT (Fig. 11A). Unsupervised clustering shows a robust separation of sexes and treatments. A subclass of EtherPCs is induced only in males at RT (including EtherPC 16:0e 18:2; others are induced in both females and males at RT (including EtherPC 18:0e 20:4; Fig. 11B). Depletion of free fatty acids is consistent for sera from both males and females housed at room temperature; however, depletion is less dramatic in females, illustrated for oleic acid (FA 18:1) and linoleic acid (FA 18:2) (Fig. 11C).

Fig. 11. Both males and females show depleted levels of circulating ether-phospholipids and increased free fatty acids.

A. Pattern of changes of ether phosphatidylcholines and free fatty acids. Unsupervised clustering of the relative abundance of entire EtherPC and FA classes show robust differences for specific species for both males and females housed at thermoneutrality or room temperature. Males are distinguished from females at both temperatures. B,C. Specific changes. Examples of two EtherPC species (EtherPC 16:0e 18:2 and EtherPC 18:0e 20:4) and for two FA species (FA 18:1, oleic acid and FA 18:2 linoleic acid) shows the relative magnitude of differences for male and female mice housed at TMN or RT. **** p<10⁻⁵; *** p<10⁻⁴; **p<10⁻³; * p<0.01.

Given the selective and striking signature of phospholipid depletion in sera of mice housed at thermoneutrality (Fig. 12A), together with the fact that phospholipids are highly enriched in high-density lipoprotein particles (HDL), we quantified the major classes of lipoprotein particles, triglyceride rich particles (TRL; including dietary and liver-derived VLDL and LDL) and HDL. Fast Protein Liquid Chromatography (FPLC) fractionation of sera indicated that HDL is depleted in sera from thermoneutral mice (Fig. 12B), most notably for males but also for females. The ratio of HDL/triglyceride enriched lipoproteins (TRLs) is reduced from 4.1 at room temperature to 2.4 at thermoneutrality for male mice, indicating a re-balancing of this ratio in favor of TRL-enriched lipoproteins.

Fig. 12. When mice are housed at thermoneutrality, the HDL-rich lipoprotein profile is decreased.

A. Total abundance of lipid classes in sera from mice housed at thermoneutrality. Abundance calculated as the sum of AUC scores from mass spectrometry were calculated for each of the 4 main classes of lipids, cholesteryl esters (CEs), triglycerides (TGs), phosphatidylcholines (PCs) and lysophosphatidyl cholines (LPCs). Statistical significance ** p<0.001; * p<0.01. B. FPLC analysis. FPLC of pooled sera from males and females (n>6) shows that the ratio of high-density lipoprotein particles (HDL) to triglyceride-rich lipoproteins (TRL) is decreased at thermoneutrality (shown right), from 4.1 to 2.4 for males and from 4.1 to 3.0 for females.

4. DISCUSSION

4.1. An overlooked environmental factor: relative humidity.

We have drawn attention to environmental factors that could modify experimental results to create “institution-specific” patterns of obesogenesis and thermogenesis. Specifically, we tested high and low humidity for their impact on mice in standard (room temperature) housing. We illustrated the wide range of relative humidity observed day to day, and season to season, in rodent vivaria. Within 1-2 days of the transition to a different humidity, mice ate more (low humidity) or less (high humidity) and showed an increased demand for heat production in low humidity (anticipated from the increased rate of evaporative cooling in low versus high humidity conditions). This was observed for both females (shown) and males.

More surprising than this result was the rapid “normalization” of mice, which re-adjusted to a standard food consumption after 2 days of exposure to the altered RH. Water consumption continued to be higher for mice housed in low humidity, illustrating that mice continued to lose water after the sensory period, and were subjected to higher levels of evaporative cooling. Vice versa, mice housed in high humidity showed increased insulation (dWAT thickness) and increased deposition of lipid into storage depots (pgWAT; Fig. 7C). This tendency was exacerbated when mice were fed high fat diet.

We propose that humidity is sensed by skin, relaying as one of the factors integrated by the hypothalamus to compute body composition, shifting the program of calorie consumption and deposition. This is supported by the results of an older study of dry- versus damp-cold exposure in men, which concluded that dry cold conditions initiated an autonomic response to counteract body temperature depression, where damp cold did not (62). The cold conditions used for men in this study resemble the thermal challenge typically encountered by mice in room temperature. The authors pointed to published studies showing increased hydration of skin in high humidity, and proposed that this suppressed the reaction of temperature sensor nerves via dilution of a temperature gradient. A review of “cold wet” conditions for humans solicited by the Department of Defense in Canada aimed to explain why cold wet is perceived as colder by humans (including soldiers) and also described a series of studies which implicated skin as the mediator (63). We propose that extremes of damp or dry cold interact with mouse pelt to affect calorie storage and utilization strategy, but in unpredictable ways: thus high humidity at room temperature cannot be used to suppress evaporative cooling to relieve the thermogenic draw.

A study drawn from several US institutions (including others abroad), showed that institutional factors were an important controller of obesogenesis for identical cohorts of mice placed on identical high fat diets (64). Thus, cohorts of C57BL/6 male mice housed at UC Davis and Yale gained almost 10 grams more than equivalent cohorts at UMass and Vanderbilt after 12 weeks of high fat diet feeding (whereas on chow diets body weights were unaffected by vivarium). We suggest that for institutional vivaria limited to using imperfect HVAC installations, studies should include data on relative humidity measurements to assist with record-keeping and reproducibility issues. Study outcomes within a research group are often disturbingly different from season to season, and we suggest that changes of relative humidity may provide an explanation. This reiterates policy changes recommended by publications across several decades (52).

4.2. Humanizing environmental exposures.

One of the main differences between environments for humans and experimental mice is their relative control of sleeping temperature. This raises several questions for rodent housing. What is the dominant environmental temperature for humans? Can the benefits associated with cool exposure and BAT activation be preserved by preventing acclimation? Does the amount of time animals spend warm every day change outcomes?

Using a diurnal temperature cycling strategy, we found that skins and BAT depots of mice sleeping in a warm sleeping environment trend towards those of mice continuously housed warm, thus BAT was deactivated and dWAT stayed thin. On the other hand, pgWAT adipose depots were as depleted as those from mice housed cool, suggesting that calories were withdrawn from the storage depots.

We speculate that a warm sleep phase prevents full adaptation to room temperature. Chronic exposure to thermogenic cues changes heat production strategies: mice defective in typical thermogenic reactions (UCP1 or Sdc1 mutants) show hyper-activated non-canonical thermogenesis (28,56). When raised from birth at room temperature, these mice defend their body temperature upon cold challenge even better than wild type mice: however, raised at thermoneutrality, with no preconditioning, cold shock is lethal. Indeed, developmental exposures change the tool-box available for thermogenesis: for example, perivascular cells are recruited to the beiging response when mice are exposed to (noxious) cold during the first 2 months of life (65) and differentiation of BAT adipocytes is promoted by cold exposure (66). Other reactions that depend upon badrenoceptor agonists are known to be primed by prior cold exposure (52): for example, if rats are housed at 4°C for 7 days prior to evaluation (at 24°C), the cardiotoxicity of isoprenaline increases by 1,000x in males, and 10,000x in females (53).

From a practical point of view, housing mice with nesting materials and igloos is therefore likely to produce quite different results from housing them without enrichment; indeed we predict that nests could approximate continuous thermoneutral housing. For investigators without access to environmental chambers, this may provide a practical solution to studying disease models.

4.3. Altered lipidome in sera of mice housed at thermoneutrality.

The “βadrenergic environment” associated with active thermogenesis is created by βadrenergic effectors and circulating cytokines (46,67–69). These stimuli are designed to move lipids from adipose depots or dietary sources in gut to the tissues making heat (6,47,51,70,71). However, changes in circulating lipids continue for as long as thermogenesis is activated and are predicted to have systemic impact on physiology.

We therefore predicted that circulating lipids would be affected in mice housed at thermoneutrality. We found an intriguing signature which has functional implications: Sera were depleted of specific phospholipid species, including some of the most abundant phosphatidylcholine species, for example PC 18:0 18:2 and PC 18:0 20:4. Indeed, all depleted phospholipid species contained either linoleic (ω6 18:2) or arachidonic acid (20:4). Thus at thermoneutrality, the incorporation of the diet-derived, essential fatty acid linoleic acid into phospholipids is reduced. Instead, the circulating levels of linoleic acid, and other free fatty acids increases: free fatty acids are typically regarded as signaling lipids, acting through G-protein coupled receptors (72). Since most circulating phospholipids are synthesized by liver, we suspect that a hepatic enzyme is modulated to prevent loading of dietary 18:2 and 20:4 onto phospholipids.

There is a relationship between the 18:2 and 20:4 acyl groups: dietary linoleic acid is the precursor to arachidonic acid (20:4), which is considered to be key to inflammatory reactions (73). Thus arachidonate is typically stored at the sn-2 position on the glycerol backbone of phospholipids, and requires release by phospholipase A2 for further metabolism to eicosanoids, prostaglandins and other inflammatory effectors (74). Elevated free fatty acids in sera have been associated with obesity, due to their interaction with the antilipolytic action of insulin, however they have never been studied for their impact on insulin signaling in mice housed at thermoneutrality.

Furthermore, the ether-phospholipids related to these depleted PC and PE species are likewise depleted. These lipids are peroxisome-derived glycerophospholipids in which the hydrocarbon chain at the sn-1 position of the glycerol backbone is attached by an ether bond instead of an ester bond (75). Although the ether variants are biochemically related to the common PC and PE versions, they have highly distinctive functions: thus they are bioactive for the assembly of membrane microdomains species, and are emerging as signaling effectors for several distinct immune cell types (75). Indeed, these species are highly enriched in specific immune cell types, such as neutrophils, and are vital to endothelial cell function (76): Our data offers no insight into the cellular origin of ether phospholipid changes.

4.4. Thermoneutral housing reduces the predominance of HDL in the mouse lipoprotein profile.

Given that phospholipids are highly enriched on HDL particles, we suspected that the pattern of lipoproteins could be affected by housing temperature. Although in general, the lipid species in sera of mice are a good mimic of human lipids (77), the fact that the lipoprotein profile of mice is so highly enriched in HDL compared to humans has compromised their use in studies of diseases arising from aberrant lipoprotein targeting and metabolism, such as atherosclerosis. HDL is broadly considered to be anti-atherogenic, taking up excess cholesterol for elimination. Indeed, unless mice are fed high fat diet or have genetic mutations in key lipoproteins, they tend to be resistant to these diseases.

Typically, the ratio of low-density lipoprotein particles to high density lipoproteins is used to measure hyperlipidemia in humans: it correlates with metabolic diseases such as atherosclerosis and liver steatosis (78). Our results reveal that the HDL concentrations typical of mice housed at room temperature (77) are modified by thermoneutral housing, so that the ratio of HDL/triglyceride rich lipoproteins is reduced. This trends towards a more “humanized” pattern.

Two groups have shown that male ApoE−/− C57BL/6 mice were more susceptible to atherosclerosis when housed at thermoneutrality (47,79): this is consistent with a reduced clearance and consumption of circulating dietary lipids when thermogenesis is turned off (80). Western diet feeding promoted the accumulation of circulating lipoproteins, including LDL: however, this study, and another from Berbee et al, did not note a significant difference in HDL/TRL ratio for chow-fed mice (51). Another study of atherosclerotic disease in mice suggested that cold exposure promoted atherosclerosis, but this group used an extreme comparison of 8-week exposure to either thermoneutrality or 4°C (81): these mice were therefore almost adipose-free and highly stressed, perhaps with little human parallel comparison. Consistent with our data, lipidomic analysis showed specific changes in HDL-associated phosphatidylcholine lipids in sera from atherosclerotic mice housed either at thermoneutrality or 4°C, or indeed for lean humans exposed to cold (82). This was associated with decreased systemic cholesterol flux through the HDL compartment in the less cold-stressed condition.

What are the phospholipids depleted of acyl groups 18:2 or 20:4 in the coats of HDL particles? The metabolism of tracylglycerol reserves is known to change depending on acyl group content (73,83); for example, saturated acyl groups are more resistant to lipolysis than poly-unsaturated acyl groups. However, the role of poly-unsaturated acyl groups in the phospholipid coats of HDL is not well understood, although incorporation of PC 16:0 18:1 into recombinant HDL affects cholesterol efflux and markers of inflammation (84). Worthmann and colleagues showed that endothelial lipase promotes extraction and hydrolysis of lipids from TRLs to promote the production of HDL during cholesterol efflux (71). Perhaps cold-activated endothelial cells hold the key to the signature we define here.

5. CONCLUSIONS

We have illustrated changes that occur in the size and thermogenic function of adipose depots in response to aspects of mouse environments which are not typically taken into account. Given the momentum towards humanizing mouse environments for the more faithful recapitulation of human diseases, it will be important to understand all aspects of this body-environment interface. One of these factors is relative humidity: We encourage investigators to find out how, or whether, relative humidity is regulated in their institutional vivaria. We found that the range of relative humidity observed in vivaria has different effects on the heat retention properties of skin, BAT activation and relative fat distribution patterns, and is sufficient to change the susceptibility of mice to obesogenesis when fed high fat diets.

Are humans in warm climates more susceptible to obesity? As a first step to untangling this complicated question, we showed that a cycling warm/cool environment controls adipose distribution, where a warm sleep environment (we propose is a humanized sleep environment) prevents the homeostatic adaptive processes that are induced to conserve heat in cool conditions. Therefore, it becomes important to understand exactly what threshold of cool exposure will trip the signal to adapt to cool exposure, switching into a calorie-conserving mode. In practical terms, criteria such as the number of co-housed mice, addition of igloos and nesting materials, and rate of air flow, should be recorded to assist with reproducibility between experiments and institutions.

We have reviewed current methods for measuring the alterations of size and function of adipose depots that occur in response to different environmental conditions. For any given mammal, there are many different tools available to control heat flow; the heat-producing tissue most able to respond acutely might be BAT, but depending upon genetics, developmental exposure and average environmental exposure, other paradigms can operate, including heat conservation and non-canonical thermogenesis. One of the most important gaps in our knowledge is that actual heat production by different tissues and adipose depots has never been measured. Although O₂ consumption is increased by the activation of non-canonical heat producing futile cycles, these reactions are not all based in mitochondria, therefore the stoichiometry of heat production/O₂ consumption is unknown, and the contribution of activated non-canonical thermogenesis is not accurately measured. This is especially important given that chronic activation of non-BAT dependent thermogenesis is linked to lean-ness when fed high fat diets.

Environmental responses, as with many/most metabolic responses, are highly dimorphic with regard to males and females. Our work has shown that skins from females reacts to suppress the impact of environment to promote homeostasis: however the potential for skin as a mediator of dimorphic responses is unknown. Our results suggest that sensory reactions in skin could modulate hypothalamic cues that govern the metabolic response.

We have revealed some signature changes in circulating lipoproteins from mice housed at thermoneutrality. These include the depletion of a cluster of phosphatidylcholines associated with HDL lipoprotein particles. The specific species affected include those with the essential fatty acid, linoleic acid (18:2) and its derivative inflammatory precursor fatty acid, arachidonate (20:4). Thermoneutral environment housing suppresses the levels of circulating ether phosphatidylcholines, therefore it could be important to assess average environmental exposure in human studies. Mice housed at thermoneutrality show a more humanized, lower HDL: TRL ratio than mice housed at room temperature, and the distinct properties of HDL phospholipid coats observed under these conditions are likely to be functionally important to the development of disease models, such as the inflammatory changes observed in nonalcoholic hepatic steatosis.

Abbreviations:

dWAT

dermal white adipose tissue

BAT

brown adipose tissue

scWAT

subcutaneous white adipose tissue

vWAT

visceral white adipose tissue

ncT

non-canonical thermogenesis

EE

energy expenditure