Using thermal preference to understand mouse thermal biology

Abstract

The ability to tightly regulate and maintain a warm core body temperature (Tb) is a defining characteristic of homeotherms. Moving to a thermally preferred place is an energetically efficient first-line thermoregulatory mechanism. We optimized a mouse binary thermal preference assay, selecting floor plate temperatures of 38°C vs 25°C for males and 39°C vs 30°C for females as having no baseline preference. Warm preference was increased in females, by fasting, and with age. Genetic manipulations that increased warm preference included ablation of uncoupling protein 1 (Ucp1) or of all four adenosine receptors (Adora1, Adora2a, Adora2b, Adora3). Ablation of estrogen receptor α (Esr1) reduced the preference for the warmer side in females but had no effect on preference in males. Mice treated with the β₃ adrenergic receptor agonist CL-316243 or the adenosine receptor 3 agonist MRS5698 sought the cooler surface. Thermal preference assays complement the measurement of Tb, providing information about whether a change in Tb reflects a new target or setpoint Tb caused by the intervention, or if the mouse is trying to maintain an unchanged Tb. Thermal preference can be more sensitive than baseline Tb for detecting changes in thermal physiology.

Graphical Abstract

New & Noteworthy

A thermal preference assay optimized for each sex demonstrated preference differences depending on physiologic state, genetic manipulation, and drug treatment. Thermal preference assays provide information that is complementary to measurement of core body temperature. Thermal preference can be more sensitive to perturbations than baseline body temperature and informs the understanding of how genetic, pharmacologic, and physiologic interventions alter thermoregulatory drive.

INTRODUCTION

The ability to tightly regulate and maintain a warm body temperature (Tb) is a defining characteristic of homeotherms. Regardless of the external environment, mammalian homeotherms defend a relatively constant Tb, typically ~36-38°C [1]. Maintenance of Tb requires heat conservation and generation in the cold while heat dissipation is needed in a hot environment and when body heat generation is excessive. Very small mammals have very high levels of heat loss, with the mass-specific heat loss from a mouse ~14 times greater than from a human [2, 3]. Indeed, singly housed mice at room temperature devote ~1/3 of their food intake to heat generation (cold induced thermogenesis) [4]. Thus, moving to a thermally preferred ambient temperature (Ta) is an energetically efficient first-line thermoregulatory mechanism for defending Tb.

Thermal preference assays can be grouped by whether they are based on a continuous thermal gradient or a binary choice between two temperatures. In the former, the animal’s position on a thermal gradient is measured, either as a physical position (distance) or by the floor temperature of that position. Some of the considerations when implementing the assay include apparatus design (length, width, whether the gradient is linear), habituation protocol, duration of measurement, measurement details (time course vs single point; observer vs beam break vs video), air temperature, phase (light vs dark), whether the mouse is allowed to sleep, etc. [5–8]. In binary choice assays the mouse can only access two surfaces with specified temperatures [5, 9–12]. Many of the assay variables relevant to gradient assays also apply to binary assays. Additionally, the binary choice apparatus can be used to study response kinetics to a change in plate temperature or the lag time before the mouse removes itself from an undesired or nociceptive temperature.

Here we explored thermal preference in detail. We chose a binary rather than gradient assay since previous experience suggested that it allows higher throughput and appears to be more robust, while providing similar information content. We generated and used the optimized assays to probe the effect of sex, drugs, genetic mutations, and other modulators of energy homeostasis on thermal preference.

MATERIALS AND METHODS

Mice

Esr1^−/− (estrogen receptor α, ERα KO) mice were provided by Dr. Constance Noguchi [13, 14]. Brs3 null mice (Jax# 32580) and mice null for all four adenosine receptors (Adora1^−/−;Adora2a^−/−;Adora2b^−/−;Adora3^−/−, QKO, Jax# 034539) were reported previously [15, 16]. Sln^−/− mice were provided by Dr. Gopal Babu, New Jersey Medical School, Rutgers [17]. C57BL/6J and Ucp1^−/− (Jax# 003124) [18] mice were purchased from Jax Laboratory. Ucp1^−/−;Sln^−/− (DKO) mice were generated by crossing Ucp1^−/− and Sln^−/− mice. Animal procedures were approved by the National Institute of Diabetes and Digestive and Kidney Diseases Animal Care and Use Committee and carried out in accordance with the National Institutes of Health’s Guidelines for the Care and Use of Laboratory Animals. Mice were maintained under controlled temperature (~22°C) and photoperiod conditions (12/12 h light/dark cycle, lights on at 0600) with chow (LabDiet, cat# 5018, Animal Specialties and Provisions, PA, metabolizable energy 3.05 kcal/g food) and water available ad libitum. Estrous cycle stage was ascertained by examination of vaginal lavage fluid obtained immediately after the thermal preference assay and scored as described [19].

Temperature Choice Test

Temperature choice testing was done using a thermal place preference test instrument, TPPT, that was purchased from Bioseb, Pinellas Park, FL. To optimize sensitivity when the direction of a possible preference is unknown, the assay was designed so that the plate temperatures were equally preferred at baseline. With equal preferences one can think of the warm and cool plates as being ‘equidistant’ from the preferred temperature and thus equally comfortable (or equally aversive). Plate temperatures of 38°C and 25°C were selected for C57BL/6J male mice, after testing 30°C/20°C, 33°C/25°C, 35°C/25°C, 38°C/25°C, and 40°C/25°C in preliminary experiments. Plate temperatures of 39°C and 30°C were selected for female mice, after preliminary experiments testing 38°C/25°C, 38°C/28°C, 38°C/30°C, 39°C/30°C, and 40°C/30°C. Keeping plate temperatures closer to the most preferred temperature minimizes potential discomfort/ aversiveness and no obvious aversive behaviors were observed. When there is a known direction for an expected change, improved assay sensitivity is achieved by using plate temperature pairs with a baseline preference of ~20% for the expected preferred temperature.

During the assay, two mice were tested simultaneously, one in each lane of a divided plastic chamber requiring each mouse to choose between metal plates whose temperatures were controlled by the experimenter. Mouse movement was video recorded for 15 min, starting when the second mouse was placed in the apparatus. Three to four hours later the mice were retested in the other lane, with the plate temperatures reversed. Thus, each mouse experienced both sides of the apparatus. The mice could not see each other and did not appear to interact. The apparatus was cleaned after every test. No attempt was made to mask or otherwise modify possible olfactory influences. Locomotion distance and time on each plate were calculated by the manufacturer’s software, averaging the two runs for each mouse. Experiments were performed at 0930-1600 (light phase) at a room temperature of ~22°C. The ambient temperature at 2.3 cm above the plate at 38°C was 29.2°C.

Body Temperature and Activity Telemetry

Animals were anesthetized and G2 E-mitters (Starr Life Sciences, Oakmont, PA) were implanted intraperitonially. Tb and activity were continuously measured by telemetry, with 1-min sampling intervals and data collection with VitalView software (Starr Life Sciences). Separately within the light vs dark phases, we used the mean Tb of all intervals with no physical activity to estimate the resting/sleeping Tb (“resting Tb”) and the Tb 95^th percentile to estimate the active/awake Tb. Tb measurements in the Ucp1/Sln and ERα mice were obtained on different days than the temperature preference assays.

Indirect Calorimetry

A “home cage” Comprehensive Lab Animal Monitoring System (CLAMS-HC using Oxymax v5.52, Columbus Instruments, Columbus, OH) was used to measure Tb, TEE (total energy expenditure), RER (respiratory exchange ratio, ratio of CO₂ production to O₂ consumption), food intake, and physical activity by beam break simultaneously in mice implanted with E-mitters as described [20]. Experiments were performed at ~22°C.

Drugs

Compounds and vehicles were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise stated. MRS5698, 2-(3,4-difluorophenylethynyl)-N⁶-(3-chlorobenzyl)-(N)-methanocarbaadenosine-5’-methyluronamide [21], was provided by Dr. Kenneth Jacobson, NIDDK. MRS5698 was dissolved in 4:1 (v/v) PEG300/warm Solutol HS15, then an equal volume of 10% 2-hydroxypropyl-β-cyclodextrin in water was added during vortexing.

Statistical Analyses

T tests and ANOVA followed by Tukey’s post hoc tests were performed as indicated in figure legends using Prism (11.0.0, GraphPad Software LLC). Significance was ascribed at P < 0.05. General linear regression was performed using JMP (17.0.0, JMP Statistical Discovery LLC).

RESULTS

Optimizing a Mouse Thermal Preference Assay: Females Like It Warmer

We designed a robust thermal preference assay with no baseline preference (see Methods). For male mice, this was satisfied by plate temperatures of 38°C and 25°C (49 ±2% of the time was spent at 38°C). Female mice have a warmer Tb than males and other differences in thermal physiology [20, 22, 23]. In the preference assay conditions where the males showed no preference, the females preferred the warmer area (74 ±2% of the time at 38°C, P<0.0001 vs males, Figure 1A). Physical activity during the assay was the same in male and female mice (Figure 1B).

Figure 1.

Male and female mice have different thermal preferences. (A) Percentage of time on the 38°C plate. Male mice choose equally between 38°C and 25°C, while females prefer the warmer temperature. Data were pooled from 13 cohorts, consisting of wild type mice and mutant mice without a temperature preference phenotype. Data are mean ±SEM, n=81 males, n=48 females, unpaired t-test. (B) Locomotion of the same mice in A. (C, D) Percentage of time on 38°C or 39°C plate. Comparing (38°C vs 25°C) to (39°C vs 30°C) assay conditions in the same cohorts of female and male mice. Data are mean ±SEM, n=14/group, paired t-test. (E) Preference for 39°C during the estrous cycle (proestrus n=7, estrus n=19, metestrus n=13, diestrus n=42). (F) Locomotion of same mice in E. There were no significant differences by stage in E,F by one way ANOVA followed by Tukey’s post hoc tests. Locomotion data for C and D are in Supplemental Figure S4A, B.

We also desired female-optimized assay conditions with no baseline preference. Using plate temperatures of 39°C vs 30°C, females showed minimal preference (59 ±5% of time at 39°C, Figure 1C), while the males avoided the warmer side (31 ±3% at 39°C, Figure 1D).

Female mice have slightly increased physical activity and Tb during proestrus and estrus [22], but we are not aware of reports of thermal preference during the estrous cycle. We did not detect any difference by estrous stage (proestrus, estrus, metestrus, diestrus) in thermal preference or physical activity during the assay (Figure 1E,F).

Generally, mice actively explored the environment for ~5 min before settling down; the higher initial activity attenuated when the mice had been tested multiple times. We observed no temperature preference differences between acclimated and non-acclimated mice in the 15-minute assay (Supplemental Figure S1A–D). Physical activity level did not affect thermal preference in males, but higher activity correlated with a cooler thermal preference in females (Figure 2A,B).

Figure 2.

Effect of activity, age, and body weight on temperature preference. (A-F) Female (red) and male (blue) regression of time on the 38°C side vs physical activity, age, and body weight, as indicated. Data were pooled from multiple cohorts, including wild type and mutants without a temperature preference phenotype (same mice in figure 1A,B). Regression lines and 95% confidence bands are shown. (G) Results from a general linear model that includes the effects of activity, age, and body weight on temperature preference.

Mice Prefer a Warmer Temperature with Age

Thermal physiology changes with age in humans [24] and mice [25, 26]. Preference for the warm plate increased with increasing age in female and male mice (Figure 2C,D), consistent with previous reports [27, 28]. In females there was a positive correlation between body weight and warm preference, while in males body weight was negatively correlated (Figure 2E,F). In general linear models, the preference correlations with activity, age, and body weight were similar to the simple regressions. There were no significant interactions between the effects of activity, age, and body weight so no interaction terms were included in the final model (Figure 2G).

Thermal Physiology of Mutant Mice

Many genes contribute to sensing both the environment and the internal milieu and to achieving the target body temperature (e.g., [8, 29, 30]) so we investigated the contribution of some of these genes to thermal preference. UCP1 uncouples mitochondrial respiration and causes brown adipose tissue (BAT) heat production [31] and SLN is a regulator of muscle Ca²⁺ futile cycling, which contributes to heat generation [32]. Continuous Tb measurement in both female and male littermate control, Ucp1^−/−, Sln^−/−, and Ucp1^−/−;Sln^−/− double null (DKO) mice at 22°C did not reveal clear effects of genotype on Tb or physical activity, with Tb during the light and dark phases and resting and active states similar in all groups (Supplemental Figure S2A–F, S3A–F, Supplemental Table S1). There was also no effect of genotype on fasting Tb, with comparable time below 34°C and Tb during the last 4h of fasting (Supplemental Figure S2G, H, S3G, H).

In the temperature preference assay the Ucp1^−/− mice preferred a warmer temperature than control mice (females p=0.06, males p=0.007) (Figure 3A, C). In contrast, the Tb of the Ucp1^−/− mice (active, light phase, the conditions of the preference assay) was not different from that of controls (Figure 3B, D). However, loss of Sln did not affect thermal preference and there was no interaction between Ucp1 and Sln genotype status, indicating that the loss of Sln did not amplify or mask the phenotype due to loss of Ucp1.

Figure 3.

Thermal preference of mutant mice. (A, C) Percentage of time on the 38°C side of wildtype (WT), Sln^−/−, Ucp1^−/− and Ucp1^−/−;Sln^−/− double null (DKO), in females and males. (B, D) Active body temperature (Tb) in light phase measured by telemetry in same genotypes of mice as in A and C. (E) Percentage of time on the 39°C plate of female Adora QKO mice and controls. (F) Active light phase Tb measured by telemetry of Adora QKO female mice and controls. (G) Percentage of time on the 38°C plate of male Adora QKO mice and controls. (H) Percentage of time on the 38°C plate in Brs3 null mice of the indicated sex. Data are mean ±SEM, n=8-15/group in A, n=3-6/group in B, n=7-14/group in C, n=5-6/group in D, n=6-9/group in E and G, n=5/group in F, and n= 6-8/group in H. A, 2-way ANOVA p values: Ucp1 0.061, Sln 0.74, interaction 0.87. C, 2-way ANOVA p values: Ucp1 0.0065, Sln 0.29, interaction 0.45. E-H, unpaired t-test. Locomotion data for A,C,E,G and H are in Supplemental Figure S4C–G.

Adenosine often functions as a paracrine danger signal. QKO mice, null for all four adenosine receptors and thus lacking adenosine signaling, preferred the warmer side (Figure 3E, G). Again, these mice did not demonstrate a Tb phenotype ([16], Figure 3F, Table S1), but did have a temperature preference phenotype. Bombesin receptor subtype 3 (BRS3) contributes to the regulation of food intake, energy expenditure, and Tb. Brs3 null mice have a lower resting Tb but no change in Tb during physical activity [29]. Brs3 null mice showed no change in temperature preference (Figure 3H).

Fasted Mice Seek a Warmer Surface

Mice at 22°C respond to fasting initially with increased physical activity and later with a reduction in Tb. By 16 h of fasting the increased activity had passed and both female and male mice preferred the warmer side (Figure 4A, B), allowing Tb maintenance with reduced energy expenditure for cold-induced thermogenesis.

Figure 4.

Fasted mice prefer a warmer plate temperature. (A, B) Effect of fasting on percentage of time on the warm side. C57BL/6J mice were fasted from 5:00pm until finished with thermal preference assays, which started at ~9:30am the next day. Locomotion data are in Supplemental Figure S4H, I. Data are mean ±SEM, n=14/group, paired t-test.

Drug Effects on Thermal Preference of Mice

CL-316243 is a β₃ adrenergic receptor agonist that produces thermogenesis in BAT and lipolysis in white adipose tissue [33]. At 1 h after CL-316243 dosing there is a large, stable increase in energy expenditure and thus heat load [34] and mice adopt a splayed out posture in a warm environment. At this time after dosing the mice showed a thermal preference for the cooler side (Figure 5A).

Figure 5.

Treatment with a β₃ adrenergic agonist (CL-316243) or adenosine A₃ receptor agonist (MRS5698) reduced warm plate preference. Percentage of time on the 38°C plate in female mice treated with 0.1 mg/kg i.p. CL-316243 (A) or 3 mg/kg i.p. MRS5698 (B). Locomotion data are in Supplemental Figure S4J, K. Data are mean ±SEM, n=13/group, crossover design, paired t-test.

MRS5698 is an adenosine receptor 3 (A₃AR) agonist that causes hypothermia. Immediately after dosing mice reduce their Tb and choose a cooler place in a thermal gradient [8]. At 5 min after treatment with MRS5698 mice sought the cooler surface (Figure 5B).

Estrogen Receptor α Contributes to the Sex Difference in Thermal Preference

Estrogens and estrogen receptor alpha (ERα) modulate multiple components of thermoregulation [35, 36] so we examined thermal preference in mice lacking ERα. In females, loss of ERα reduced the preference for the warm plate (Figure 6A), while no change was observed in ERα KO males (Figure 6B). After ERα loss, there was no warm preference observed with fasting (Figure 6C, D).

Figure 6.

Thermal preference in ERα KO mice. Percentage of time in 39°C (vs 30°C) in female (A), and 38°C (vs 25°C) in male (B) ERα KO and littermate control mice. N=14-18/group. Percentage of time in 39°C (vs 30°C) in fed vs fasted female (C), and in 38°C (vs 25°C) in fed vs fasted male (D) ERα KO and littermate control mice. Locomotion data are in Supplemental Figure S4L–O. Data are mean ±SEM, n=8-9/group female, n=12-18/group male. C, 2-way ANOVA p values: genotype <0.0001; fed/fasted 0.0048; interaction 0.032. D, 2-way ANOVA p values: genotype 0.10; fed/fasted 0.54; interaction 0.060.

We characterized thermal physiology in the ERα KO mice. Compared to control females, ERα KO female mice had a lower Tb (dark and light phases) and were less physically active (dark and trending in light phases) (Supplemental Figure S5A, B). Both the resting and active Tb were reduced (Figure S5G,H). The TEE and RER were similar in ERα KO and controls (after accounting for body weight differences, Figure S5C–F). When we examined male ERα KO mice, Tb (resting and active), physical activity, TEE (accounting for body weight), and RER were not significantly different from the wild type littermate controls (Supplemental Figure S6A–H). No difference in the Tb response to fasting was observed either in female or male ERα KO (Figure S5I, J, Figure S6I, J). In addition, we did not observe differences in food intake in either female or male ERα KO mice (data not shown). In summary, loss of ERα reduced the Tb in female mice, with no effect in males.

DISCUSSION

We designed a mouse thermal preference assay with no baseline preference, optimizing sensitivity when the direction of a perturbation/manipulation is unknown. Environmental conditions (light phase, room temperature of ~22°C) were selected for experimental ease. Plate temperatures were chosen to be clearly different from each other while being non-aversive and non-extreme. Plate temperatures of 38°C vs 25°C for males and 39°C vs 30°C for females satisfied these conditions. Measurement of thermal preference is a quick and robust assay that does not require an environmental chamber for manipulating ambient temperature. The assay has an additional desirable feature, that neither prior habituation nor run-in times are required. Our standard assay was to take the average preference from two 15-minute periods measured ~3-4 h apart.

The thermal preference assay relies on the animal’s feet contacting metal plates, which conduct heat more effectively than the surfaces of a natural den or laboratory cage. As a result, the behavior reflects thermal sensation by the feet, while the rest of the body is exposed to temperatures closer to that of ambient air. Thus, the thermal preference assay complements studies that manipulate a uniform environmental temperature, such as in experiments studying thermoneutrality [3, 4].

Thermal preference data supplement the information gained from measurement of Tb. There are two classes of explanation for a change in thermal preference. In one class, the mouse is presumably at its target (or setpoint) Tb. An example of this is the Ucp1^−/− mouse, which has a normal Tb at room temperature that falls upon acute cold exposure [18]. The Ucp1^−/− warm preference is consistent with the desired Tb being reached more efficiently, requiring less cold-induced thermogenesis. Treatment with a β₃ adrenergic agonist at room temperature can cause a small Tb increase due to the increased heat production[37]. The cool preference upon treatment reflects the benefit of increased heat dissipation. These examples demonstrate that thermal preference can be a more sensitive probe than baseline Tb at an unstressed ambient temperature for detecting changes in thermal physiology.

A different class of changes in thermal preference is consistent with a model in which the defended Tb may be altered. For example, shortly after treatment with an A₃AR agonist mice sought the cooler surface. This confirms thermal gradient data [8] and supports the idea that the drug lowers the target Tb and the mouse is choosing an environment to facilitate that. Here the timing of the preference assay relative to a treatment is crucial. Other uses of the thermal preference assay could include studying thermal sensation and how it links to thermal effectors such as shivering, vasoconstriction, thermogenesis, and posture, and investigation of motivational state and cost-benefit integration as related to thermal biology.

Sex influences Tb physiology. Some examples of Tb sex dimorphisms include a higher Tb in females [22], Tb differences during single vs group housing [20], and the length and duration of torpor bouts during fasting [20, 23]. Responses to genetic manipulations and drug treatments can also demonstrate sex-dimorphic patterns [38–41]. Like Tb, thermal preference also differs between the sexes. Female mice have a warmer preference than males [41] and in postpartum female mice, thermal preference was altered due to reduction in the number of ERα-expressing neurons in the preoptic area neurons [42]. The higher Tb of females likely underlies their warmer thermal preference: analogous to the Ucp1^−/− mice, a warmer environment requires less thermogenesis to reach a higher Tb. The warmer preference requires expression of ERα, providing an underlying molecular basis.

A limitation of our study is that all experiments were conducted during the light phase. Mouse Tb is ~0.7°C lower in the light photoperiod [43]. We have not investigated the dark phase, or light/dark and dark/light transition periods. Additionally, mouse Tb is ~0.9°C higher during active/awake (vs resting/sleep) intervals, independent of light/dark phase [43]. However, this is not expected to impact the preference assay since the handling required and the short assay duration mean that the mice will be active/awake. A major input to thermal preference is whether the Tb is at its setpoint, so it is a limitation that Tb was not measured during the preference assay and that Tb setpoint was not directly measured in this study.

Thermal preference assays provide insights that complement the measurement of Tb. Indeed, we show examples of changes in thermal preference when differences in Tb are not detected.