Conventional ambient temperature housing impacts central oxytocin levels in mice

Abstract

Neuroscience depends heavily on research done in mice (Mus musculus), yet the field has done little to address the chronic cold stress mice perpetually face during conventional “room temperature” housing (20°C–22°C). Contributions from other biomedical fields, such as immunology, oncology, and metabolic physiology, have shown that housing mice at room temperature substantially impacts broad and fundamental aspects of murine biology in ways that negatively affect the translational value of the research derived from these animals. Prairie voles (Microtus ochragaster) are an alternative small rodent model for neuroscience that are adapted for cold weather and better tolerate the ambient temperature of conventional housing. Here, we examined the effect of 3 days of housing at one of three ambient temperature conditions: 20°C, 25°C, or 30°C on oxytocin and vasopressin immunoreactivity within the paraventricular nucleus of the hypothalamus in both mice and voles. We found that increases in ambient temperature above 20°C led to a 32% reduction in oxytocin immunoreactivity in mice, while having no effect in voles. Vasopressin was unaffected in either species. Since oxytocin is a pleiotropic neuropeptide, responsible for regulating a number of homeostatic, emotional, and social circuits, this work calls into question whether findings from mice housed at 20°C can be reliably translated to humans living in thermoneutral conditions. This finding should spur further neuroscience research to reconcile how the assumptions of conventional housing have shaped murine neurobiology.

1 |. INTRODUCTION

Animal models are an invaluable tool in neuroscience research and form the basis for the successful translation of findings from lab bench to hospital bedside. As small mammals that grow and reproduce quickly, rodents generally offer several advantages for the laboratory setting. Among all available animal models, there is none more popular than the mouse, Mus musculus. For example, among all NIH grants issued 2013–2024, 19.9% contained either of the terms “mouse” or “mice” in the text of their abstract, while only 2.4% of abstracts contained either “rat” or “rats” (via NIHExPORTER). Publications featuring mice overtook those featuring rats around 2012 in neuroscience,¹ ~10 years after having done so across the wider biomedical research literature.² Contemporary neuroscience papers featuring mice perform better than those featuring rats, garnering more citations and appearing in more high-impact journals.³

Originally beginning outside of neuroscience in fields related to obesity,^4–7 cancer,^8,9 sleep,^10,11 and immunology,^9,12 research has been slowly building to appreciate the impact of thermoregulation on mouse physiology.^2,13 This sentiment, that it is important to consider how the thermal conditions of the conventional vivarium housing experience impact mice’s physiology, is now beginning to permeate neuroscience also. It is easy for human researchers to overlook the importance of thermoregulation, as humans have essentially mastered the thermal environment, engineering the indoors such that clothed humans can take for granted the fact that we are nearly always in thermoneutral conditions.¹⁴ As such, little to no energy is generally expended for humans to stay warm. Rodents, being smaller while having relatively larger surface volume and metabolic rates than humans, are more sensitive to the effects of ambient temperature.¹³ Ambient temperature is one of the largest determinants of energy expenditure in mice,¹⁵ and the basal metabolic rate and food intake of mice housed at 20°C are both ~50% higher than those housed at a thermoneutral 30°C.^2,4,6,16 Specifically, 20°C housing increases spontaneous locomotor activity, the thermic effect of digestion, and cold-induced thermogenesis.¹⁷ Tumor formation, growth rate, and metastasis are all accelerated in conventionally housed mice compared with mice kept at thermoneutrality.⁹ Ambient temperature also has profound effects on the pharmacokinetics and toxicology of many drugs.¹³ Under conventional housing conditions, lab mice face chronic cold stress that substantially impacts broad and fundamental aspects of their physiology in ways that negatively impact the translational value of the research derived from them.¹⁸

Voles of the Microtus genus represent a small rodent alternative to mice for neuroscience, albeit a far less popular one. Since 2000, the combination of the terms “vole and brain” is outnumbered in PubMed search results by the terms “mouse and brain” by >40×. Among voles, prairie voles (Microtus ochrogaster) are the most well-studied animal model in neuroscience, owing to their social monogamy and suite of related prosocial behaviors.¹⁹ However, voles also appear to have a series of adaptations that render them more resilient to housing at conventional “room temperature” 20°C. Voles are originally adapted to boreal habitats, with thick coats of fur,^20,21 short tails that reduce heat loss relative to rats or mice,²² and small ears, from which the genus takes its name.

Mice experience conventional vivarium housing as a chronic cold stress because they only experience “thermoneutrality” between 29°C and 33°C depending on the circadian phase,^8,17,23 meaning at conventional “room temperature” housing (20°C–22°C), mice must expend energy to maintain body heat. Voles, meanwhile, have a thermoneutral range ~25°C–30°C that extends down to 20°C when nesting is considered.^24,25 Unfortunately, nesting and group housing do little to ameliorate the chronic cold stress faced by mice.²⁶ We recently observed that prairie voles show only ~8% change in resting heart rate in response to changes in ambient temperature between 18°C and 30°C.²⁷ This contrasts with reported values for mice of >50% between 22°C and 30°C.^28,29

Up to now, neuroscience has only just begun to investigate the consequences of the chronic cold stress borne by mice at conventional 20°C housing. Chronic thermoneutral housing reduces hippocampal Aβ levels in a mouse model of Alzheimer’s disease.³⁰ Meanwhile, the acute effects of ambient temperature are better documented, including activating a circuit originating from the paraventricular nucleus of the hypothalamus (PVN) that projects to the anteroventral and periventricular portions of the medial preoptic area to control temperature-dependent changes in feeding behavior.³¹ The hypothalamic PVN is a valuable starting point for these investigations, as it coordinates a wide range of homeostatic, metabolic, and thermoregulatory functions. The broader neurobiological impact of the chronic cold stress posed by conventional housing remains to be fully explored.

Here, we focused on two neuropeptides that have been the targets of neuroscience research using both mice and voles: oxytocin (OXT) and arginine vasopressin (AVP). These two neuropeptides are studied for their roles in regulating body temperature,^32–35 as well as social behavior,³⁶ fear,³⁷ autonomic regulation,³⁸ and appetite/metabolism.³⁹ The major source nucleus for both neuropeptides is the PVN, which is why we chose to focus on that region in the present study using immunohistochemistry to quantify immunoreactivity (-ir) for each neuropeptide. Since both neuropeptides control thermoregulation, we sought to investigate how each would respond to a prolonged (3-day) change in ambient temperature. We hypothesized that (1) mice would show more sensitivity than voles in neuropeptide expression to changes in ambient temperature, owing to the latter’s adaptations for colder climates, and (2) females would show more sensitivity than males in both species, owing to the latter’s larger body size. As expected, there were large species differences in OXT-ir and AVP-ir, with voles displaying far greater quantities and having greater variability as well. Since the purpose of this study was not to replicate already well-established species differences, but to examine how each species responded to changes in ambient temperature, we were less interested in comparing the level of neuropeptide -ir between species than in the species’ sensitivity to ambient temperature.

2 |. MATERIALS AND METHODS

2.1 |. Subjects

This study utilized 60 laboratory-bred prairie voles (30:30 Male:Female) and 60 mice (28:32 Male:Female) total. Mice were either CFW (n = 16) purchased from Charles River Breeding Laboratories (Massachusetts) or C57BL/6 (n = 44) generously donated by the Duncan and Lachke laboratories at the University of Delaware. These C57BL/6 animals were either wildtype or heterozygote genetic modifications that were unrelated to thermoregulation. The two strains were randomly and evenly distributed across ambient temperature conditions. All subjects were given ad libitum access to food (standard chow for voles: LabDiet Rabbit Diet HF 5326, standard chow for mice: LabDiet Prolab RMH 3000) and housed on a 12:12 light:dark cycle. On post-natal day (PND) 21, subjects of both species were weaned and housed in groups of 2–4 with same-sex litter mates. All subjects were experimentally naive. Voles averaged PND 120 ± 4.06 and mice averaged PND 108 ± 4.66 at the time of testing. All methods in this study were approved by the University of Delaware Institutional Laboratory Animal Care and Use Committee and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

2.2 |. Temperature

Subjects were housed in rooms set to an ambient temperature of either 20°C, 25°C, or 30°C for 3 days. All subjects remained in their home cage with standard nesting and same-sex cage mates throughout testing. Mouse and vole cages were identical in size and bedding. Ambient temperature was adjusted using the Building Automated Systems via the thermostat. Subjects and room temperature were monitored each day. See Figure 1 for the experimental timeline.

FIGURE 1.

Ambient temperature experimental timeline. Subjects were housed in rooms set to an ambient temperature of either 20°C, 25°C, or 30°C for 3 days. Subjects were sacrificed immediately after completion of temperature exposure. Voles averaged PND 120 ± 4.06 and mice averaged PND 108 ± 4.66 at the time of testing with the full range of all subjects’ PND being 57–172. Created in BioRender. Rogers, K. (2025) https://BioRender.com/a78ic2z.

2.3 |. Tissue extraction and immunohistochemistry

Sacrifice occurred immediately after the 3 days of exposure to one of the three ambient temperatures. Mice and voles were sacrificed concurrently at 11:00–13:00, in the middle of their light phase. Isoflurane was administered to subjects for ~5 min, and then subjects were cervically dislocated. Brains were placed free-floating in 4% paraformaldehyde for 24 h after extraction, and then switched to a 25% sucrose and 0.01% sodium azide solution for storage at 4°C for at least 24 h. Fixed brain tissue was sectioned at 30 μm using a cryostat microtome from medial prefrontal cortex to ventral hippocampus, then stored in cryoprotectant at −20°C.

All immunohistochemical analysis procedures were performed at room temperature unless otherwise noted. Tissue was rinsed in 0.05 M potassium phosphate buffered saline (KPBS) for 5 min three consecutive times, and then incubated in 1% hydrogen peroxide for 30 min. Tissue was then rinsed in potassium phosphate buffered saline (KPBS) for 5 min three consecutive times. After this, tissue was incubated in primary antibodies (either OXT 1:24,000 dilution or AVP 1:16,000 dilution in KPBS + 0.4% Triton-X, MilliporeSigma antibodies #MAB5296 and AB1565, respectively) for 60 min and then incubated for 60 h at 4°C. Sections were then rinsed for 5 min three times in KPBS and then incubated in 1% goat serum (Thermo Fisher Scientific) + KPBS +0.4% Triton-X for 30 min at room temperature. The tissue was then incubated in biotinylated goat anti-mouse/rabbit IgG secondary antibody (1:600 dilution in KPBS + 0.4% Triton-X, Vector Laboratories) for 60 min, and then rinsed for 5 min five times in KPBS. Next, tissue was incubated in avidin–biotin peroxidase complex (Vector Labs; Vectastain Elite ABC Universal kit; 45 μL A, 45 μL B per 10 mL KPBS + 0.4% Triton-X) for 60 min, and then rinsed in KPBS for 5 min three times. The tissue was then rinsed in 0.175 M sodium acetate for 5 min three times and then placed in a nickel-diaminobenzene solution (1 tablet diaminobenzene, 50 mL sodium acetate, 1.25 g nickel sulfate per dish). Hydrogen peroxide (41.5 μL per 50 mL) was added immediately before placement in the nickel-diaminobenzene solution for 6 min. Finally, tissue was rinsed in sodium acetate and then KPBS, each for 5 min three times.

Sections were mounted onto superfrost plus microscope slides (Thermo Fisher Scientific), then dehydrated in ascending ethanol concentrations and Histoclear, and then cover-slipped. Slides were scanned at 4× magnification using an AmScope microscope. OXT-ir and AVP-ir images, specifically of the PVN of the hypothalamus (anterior to posterior), were quantified in ImageJ⁴⁰ using a standardized threshold above background (0–170 px OXT, 0–110 px AVP), that is, non-specific staining, to measure the total area of immunoreactive tissue, which represents the sum total of both cell bodies and fibers. Because disambiguation between magnocellular and parvocellular neurons is not possible in voles, the total neuropeptide-ir was analyzed for both species. Anterior–posterior positions were assigned according to reference images, nos. 23–27, in the prairie vole brain atlas.⁴¹

2.3.1 |. Statistical analysis

This was a somewhat complex dataset due to the simultaneous consideration of species, ambient temperature, sex, and anterior–posterior position across the PVN. Additionally, there were large species differences in neuropeptide-ir that also meant large species differences in the variability of these measures (heteroscedasticity). Traditionally, neuroanatomical studies like this one would focus on a single anterior–posterior position (typically what is referred to here as position no. 25, where neuropeptide-ir in the PVN is highest), but we sought to consider anterior–posterior position as a covariate in our analyses, necessitating a repeated-measures design.

Because 72% of usable samples from mice had little to no OXT in the posterior-most section of the PVN (compared with 26% of usable samples from voles), we excluded the posterior-most section from subsequent analyses, leaving 4 usable positions (nos. 23–26). Similarly high rates of AVP-ir absence were noted for both the anterior-most and posterior-most sections, leaving 2 usable middle positions (nos. 24–25). We then excluded outliers for each species, at each anterior–posterior position, based on a criteria of 1.5× the interquartile range. For OXT-ir, 7% of all samples were removed, ranging from 3.5% of samples from voles at 30°C to 10.1% of samples from voles at 20°C. For AVP-ir, 6.8% of all samples were removed, ranging from 3.1% of samples from mice at 30°C to 9.3% of samples from mice at 20°C. The removal of outliers resulted in one animal (a mouse in the 30°C condition) no longer contributing any data to the OXT-ir measure; no animals were completely lost from the AVP-ir measure. Since some tissue was lost during processing, brain tissue sample sizes were reduced further. For OXT-ir, the final sample sizes ranged from n = 16 (mice at 25°C) to n = 20 (mice at 20°C). For AVP-ir, the final sample sizes ranged from n = 15 (voles at 25°C) to n = 19 (mice at 20°C). All results were square-root transformed to reduce skewness.

All statistical analyses were carried out in R. Because voles are already known to show far greater OXT-ir than mice, we analyzed each species separately. We used a linear mixed effects model with repeated measures to compare ambient temperature and sex as fixed effects, and subject and anterior–posterior position as random effects, using the lmer() function from the lme4 library,⁴² followed by calling the anova() function on the resulting model (type III sum of squares) to produce conventionally understood p-values for main effects and interactions. When an effect of ambient temperature was found, post hoc analyses were performed for each anterior–posterior position, and when an effect of ambient temperature was found for a specific anterior–posterior position, further post hoc testing was performed via post hoc testing done with Tukey’s. The α for all of the above testing was set at 0.05. To control for multiple comparisons, we subjected the p-values to False Discovery Rate correction.

3 |. RESULTS

OXT-ir results are shown in Figure 2. For OXT-ir, there was a significant main effect of ambient temperature on OXT-ir in mice (F(2,257) = 11.74, p < .001, Figure 2A). This difference represented a 32% reduction in OXT-ir from 20°C to 25°C (d = 0.69) and a 33% reduction from 20°C to 30°C (d = 0.75). Specifically, at anterior–posterior position no. 25, there was a significant main effect of ambient temperature (F(2,86) = 5.46, p < .006), such that mice at 20°C had greater OXT-ir than mice at 25°C or 30°C (p = .045 and .007, respectively, Figure 2B). At anterior–posterior position no. 26, there was another significant main effect of ambient temperature (F(2,69) = 3.25, p = .045), such that mice at 25°C had greater OXT-ir than mice at 30°C (p = .044); however, this did not survive correction for multiple comparisons. There was no effect of sex on OXT-ir in mice, nor was there an effect of either ambient temperature or sex on OXT-ir in voles. AVP-ir results are shown in Figure 3. There were no effects observed for sex or ambient temperature in either voles or mice.

FIGURE 2.

Oxytocin immunoreactivity. (A) Total OXT-ir across the anterior–posterior extent of the PVN for both voles and mice housed at either 20°C, 25°C, or 30°C. Mice (but not voles) showed decreased total OXT-ir when housed above 20°C (*p = .005). For the sake of clarity, each species is presented with its own y-axis scale. (B) OXT-ir at individual positions across the anterior–posterior axis. Voles consistently had greater OXT-ir than mice at all positions and temperatures. Mice (but not voles) showed decreased OXT-ir at position no. 25 when housed above 20°C (p < .045) and mice at 30°C showed less OXT-ir than mice at 25°C at position no. 26 (*p = .045). Error bars within each violin plot indicate the range of the mean ± standard error. (C) Representative photomicrographs of OXT-ir for both species at each ambient temperature.

FIGURE 3.

Vasopressin immunoreactivity. (A) Total AVP-ir across the anterior–posterior extent of the PVN for both voles and mice housed at either 20°C, 25°C, or 30°C. There was no effect of temperature on total AVP-ir for either mice or voles. For the sake of clarity, each species is presented with its own y-axis scale. (B) AVP-ir at individual positions across the anterior–posterior axis. Voles consistently had greater AVP-ir than mice at all positions and temperatures. There were no effects of sex or ambient temperature. Error bars within each violin plot indicate the range of the mean ± standard error. (C) Representative photomicrographs of AVP-ir for both species at each ambient temperature.

4 |. DISCUSSION

The chief finding of this work is that compared with those housed at 20°C, mice housed at warmer ambient temperatures produce 32%–33% less OXT-ir within the PVN of the hypothalamus. This effect was not observed for AVP or for either neuropeptide in voles. These results support our primary hypothesis that mice would be sensitive to the effects of ambient temperature and voles would not, likely owing to the species’ differing natural histories. We observed no linearity in OXT-ir response to temperature in mice, as OXT levels at 25°C were no different than those at 30°C, suggesting that there may be a threshold between 20°C and 25°C that affects murine OXT. We can only speculate whether voles would show a similar increase in OXT-ir if the ambient temperature were lowered further to generate an analogous cold stress for that species. Our secondary hypothesis was not supported, however; we found no evidence to suggest that females’ OXT or AVP were affected by ambient temperature more so than males. These findings extend the lack of sex differences in AVP-ir and OXT-ir previously noted in the PVN of rodents housed at conventional ambient temperature⁴³ to now also include rodents housed in warmed ambient temperatures.

The changes in OXT levels seen in mice suggest a range of possible consequences. Central OXT levels respond to conditions of chronic stress, such as when socially isolated prairie voles increase OXT-ir in the PVN.⁴⁴ When central OXT levels are experimentally reduced via knockdown, multiple aspects of social behavior are affected.⁴⁵ Thus, the reduction in PVN OXT-ir seen here likely has functional consequences that await further study. For example, assessing the effect of housing temperature on the expression of the receptors for these neuropeptides.

Previously, either 6 h or 5 days of exposure to 4°C was found to upregulate the expression of the OXT receptor, Oxtr, (but not OXT itself) throughout the whole brain of male mice when comparing to mice housed at 25°C.⁴⁶ In that same study, Oxt gene expression was upregulated in bone and downregulated in both testes and brown adipose tissue. The present study builds on those findings by narrowing in on the PVN, focusing the range of ambient temperatures to conventional room temperature (20°C–22°C), including a thermoneutral comparison, including females, and measuring protein rather than mRNA. Our finding that the murine OXT system is impacted by conventional housing conditions has both broad implications for research derived from mice as well as the pleiotropic aspects of homeostatic, emotional, and behavioral functions that OXT subsumes.⁴⁷ Here, we examined a single anatomical measure of the OXT and AVP systems. Receptor expression, firing rates, innervation patterns, and other neural properties might all also be sensitive to ambient temperature. Likewise, OXT and AVP are far from the only thermoregulators. Estrogen, leptin, and PACAP each control thermoregulation—are they likewise sensitive to ambient temperature?

We observed that housing at 20°C leads mice to increase OXT production in the PVN. We suspect OXT was impacted here because of its role in thermoregulation.^32,33,48 OXT acts to enhance adaptive thermogenesis, both browning adipose tissue and stimulating brown adipose tissue.⁴⁹ However, OXT also regulates metabolism, reproduction, as well as fundamental processes like mineral balance and bone density. Given that 20°C housing leads mice to increases in food consumption, blood pressure, and heart rate of 40%–60%,^4,29 and that mice experience this effect transgenerationally, it is therefore possible that a wide swath of murine neurobiology is impacted by conventional housing conditions in ways that could negatively impact translational value. For example, OXT is broadly anti-inflammatory,⁵⁰ and systemically affecting resting inflammatory tone would have widespread consequences on brain and behavior.

Acute bouts of thermoneutrality activate both OXT and AVP neurons in the PVN, and this appears to modulate murine maternal behavior.⁵¹ Thus, murine maternal care could also be affected by conventional housing. Given the far-reaching developmental consequences of differential maternal behavior,⁵² this suggests that large portions of mouse neuroscience research may have been conducted on a background of altered developmental programming. Variation in maternal behavior developmentally fine-tunes stress reactivity,^53,54 sensitivity to addiction,⁵⁵ and pain tolerance⁵⁶ among many other aspects of behavior. In the context of the developmental programming of metabolism, young rodents are especially vulnerable to cold ambient temperatures. For example, rearing rat pups at 22°C leads to developmental programming of food intake, thermoregulation, adiposity, and sympathetic innervation even after transfer to 28°C housing.⁵⁷ Here, the prairie vole has the added benefit of biparental care, where fathers routinely huddle their litters, providing added warmth. Mouse social behavior may also be impacted by ambient temperature, not just indirectly through changes in OXT, which is a powerful regulator of social behavior, or through maternal care, but also because mice presumably experience cage-mate conspecifics as the only source of warmth available to them.

Despite the chronic stress present for mice at 30°C, some argue it is the optimal ambient temperature to house mice to produce a human-like metabolic rate⁵⁸; however, others using similar reasoning have arrived at a value of 25°C as the optimal ambient temperature.⁵⁹ There is no simple remedy to the chronic cold stress mice experience at 20°C. Individually warmed cages would address the issue without compromising animal care staff comfort and safety. However, this is an expensive solution and one that makes behavioral testing more difficult. Reorienting all of neuroscience to switch from mice to focus on voles or some other species is even more onerous. Rather, it seems important to come to terms with the scope of the impact of conventional housing on the neurobiology of the mouse and then pursue the most severely affected subdomains. Unfortunately, we expect a wide range of murine neurobiology to be affected, given the severity of metabolic disturbance and the breadth of OXT’s functions.

This study is not without limitations. We did not assess behavioral thermoregulation, meaning we do not know whether mice or voles huddled more at cooler ambient temperatures, which could have either contributed to or followed from the changes in OXT seen here. Future work in this area should consider huddling as well as nest construction and circadian phase. Including body weights as a measure would have been informative regardless of whether 3 days of a changed ambient temperature were enough to produce a difference. The results presented here found an effect of temperature in mice, but not in voles; however, we were unable to conduct direct species comparisons in the response to ambient temperature due to statistical constraints. Future work should also examine the longer, developmental consequences of ambient temperature by rearing mice from birth (or ideally even pregestationally) at different temperatures. We also hope that future work will broaden out the scope of neurobiological measures to include other aspects of OXT (e.g., fiber innervation, receptor density) as well as other thermoregulatory brain regions, such as the preoptic area⁶⁰ and ventromedial hypothalamus,⁶¹ as well as other neurotransmitter systems. Most of all, we hope these findings spur future research to grapple with the impact of the chronic cold stress faced by mice in neuroscience labs across the world.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.