Microglia Regulate Sleep and Thermoregulatory Stability

Abstract

Sleep fragmentation disrupts physiological homeostasis, including thermoregulation, metabolism, and immune signaling; however, cellular mechanisms that stabilize core body temperature and sleep architecture during disrupted sleep remain poorly understood. Microglia are the resident immune cells of the central nervous system and play a central role in integrating neuroimmune signaling with hypothalamic regulation of thermoregulation and sleep-wake control. We tested the hypothesis that microglia are required to maintain both thermoregulatory and sleep stability during physiological sleep and when sleep is disrupted.

Male and female C57BL/6J mice (n = 128) were implanted with telemetry devices for continuous core body temperature monitoring and maintained on either a control diet or diet containing the CSF1R-inhibitor PLX5622 (PLX) to pharmacologically deplete microglia. Body temperature and sleep-wake behavior were assessed across a two-week depletion period and during a subsequent 7-day sleep fragmentation paradigm. Hierarchical models were used to quantify the effects of microglial depletion (via diet), sex, and time-of-day (light/dark period).

Microglial depletion produced sex- and diet-dependent alterations in both thermoregulation and sleep measures. PLX increased body temperature in females early during the depletion period, after which body temperature normalized and subsequently declined to pre-PLX baseline values. In contrast, PLX minimally influenced outcome metrics in males. During sleep fragmentation, microglial depletion induced time-of-day-specific hypothermia in males and females during the dark and light periods, respectively. Notably, microglial depletion dissociated components of sleep in a sex-dependent manner: Total sleep time increased in females, whereas males exhibited reduced sleep-wake transitions, suggesting increased sleep consolidation and therefore distinct regulation of sleep quantity versus stability. Across experimental conditions, body temperature was a strong nonlinear predictor of sleep, with higher temperatures (>35°C) associated with substantially reduced total sleep time (5–10 min reduction for every 1°C increase), particularly during the dark (active) period.

These findings collectively suggest that microglia contribute to thermoregulatory processes and sleep stability in a sex-dependent manner, supporting a role for microglia in coordinating physiological homeostasis. Together, our results support a neuroimmune framework in which microglia may help coordinate thermoregulation and sleep, as microglial depletion was associated with dysregulation of both processes.

Introduction

Sleep is a fundamental physiological process that influences nearly every biological system and is essential for maintaining homeostasis across species. In mammals, the sleep-wake cycle consists of alternating periods of non-rapid eye movement (NREM) and rapid eye movement (REM) sleep, coordinated by hypothalamic and brainstem circuits that regulate arousal, metabolism, and temperature [1]. NREM sleep, particularly its slow-wave component, is associated with synaptic downscaling, energy restoration, and immune modulation, whereas REM sleep supports memory consolidation and emotional regulation [2–5]. Chronic sleep restriction is increasingly prevalent in modern societies, with approximately 30% of adults sleeping fewer than seven hours per night, and up to one-third reporting symptoms of insomnia at some point during their lifetime [6, 7]. Insufficient sleep contributes to cognitive deficits, impaired thermoregulation, and heightened inflammatory activity, highlighting its broad physiological and public health relevance [8, 9].

Importantly, insufficient sleep and sleep disturbances in clinical contexts often manifest not solely as changes in total sleep time, but also as fragmentation, defined as repeated transitions between sleep and wake states that destabilize sleep continuity [10–13]. Unlike total sleep deprivation, sleep fragmentation permits intermittent sleep while repeatedly interrupting consolidation, producing physiological stress and inflammatory activation [8, 14]. Mounting evidence suggests that sleep continuity, rather than duration alone, is a critical determinant of physiological resilience [15–17]. Importantly, sleep fragmentation represents a distinct and clinically relevant stressor that may differentially engage neuroimmune and thermoregulatory systems compared to total sleep loss.

Core body temperature is a tightly regulated physiological variable that reflects the balance between heat production and dissipation. In mammals, core temperature follows a circadian rhythm, typically reaching its lowest point during rest and rising during active periods [18]. Regulatory interactions between core body temperature rhythms and sleep have long been known [19–21]. Sleep and thermoregulation are tightly linked yet dissociable physiological processes that are particularly important for the normal transition from wakefulness to sleep [22–24]. Specifically, a reduction in core temperature associated with peripheral vasodilation precedes and facilitates the transition to sleep [23]. Across all mammalian species, this decline in core body temperature reliably precedes sleep onset, supporting a functional role for thermoregulatory mechanisms in sleep initiation. Conversely, experimental warming of the skin accelerates sleep onset and deepens NREM sleep, whereas cooler environments delay sleep initiation and increase arousal [25, 26]. Recently, a polysomnography study demonstrated that dynamically adjusting mattress temperature in real-time, according to sleep stage, significantly improved sleep efficiency and increased total sleep time, which highlights the practical importance of thermoregulatory adjustment in sleep quality [27].

Thermoregulation interacts closely with metabolic and immune systems but is also tightly integrated with sleep-wake control [28]. Within the hypothalamus, particularly the preoptic area, distinct populations of temperature-sensitive neurons contribute to this coordination [29]. Warm-sensitive neurons are predominantly sleep-promoting and facilitate NREM sleep, whereas cold-sensitive neurons are largely wake-promoting and support arousal [29–31]. This organization links thermal state to sleep-wake regulation at the circuit level, particularly during the transition from wakefulness to sleep. Despite this well-established relationship, the cellular mechanisms that coordinate thermoregulation and sleep, especially under conditions of sleep fragmentation, remain poorly defined.

Microglia are the resident immune cells of the brain, which serve as central regulators of neuroinflammation and are increasingly recognized as modulators of sleep [32, 33]. Pro-inflammatory cytokines, such as interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6), promote NREM sleep, whereas anti-inflammatory cytokines like IL-4 and IL-10 exert opposing effects [34–36]. These cytokines are largely produced or regulated by reactive microglia, positioning them as key intermediaries between immune signaling and sleep-wake regulation. Beyond sleep, cytokines influence body temperature, appetite, and energy balance [37], highlighting their broader role in coordinating physiological homeostasis. Given their central role in cytokine production and hypothalamic signaling, microglia may integrate immune activity with thermoregulatory and sleep circuits.

Pharmacological inhibition of colony-stimulating factor 1 receptor (CSF1R) using PLX5622 enables selective depletion of microglia [38–40], which provides a causal approach for testing hypotheses about the role of microglia in physiological processes. Although sleep, thermoregulation, and immune activity are closely linked, microglial contributions to this coordination during sleep disruption are not well understood. Most prior work has focused on total sleep deprivation, with fewer studies examining sleep fragmentation, which preserves sleep but disrupts consolidation. This pattern may uniquely challenge thermoregulatory and inflammatory homeostasis, yet how microglia influence these responses during acute sleep fragmentation remains unclear. Moreover, previous studies have largely focused on changes in sleep quantity following microglial depletion, whereas considerably less is known about whether microglia coordinate multiple components of physiological homeostasis, including thermoregulation and sleep stability, particularly across biological sex and time-of-day.

Human data suggest that adaptive thermoregulation can improve sleep architecture [41], but the cellular mechanisms, particularly involving microglia, remain unexplored. To address this knowledge gap, we employed continuous core body temperature telemetry in mice subjected to a defined experimental paradigm that included acclimation, surgical implantation of temperature probes, pharmacological microglial depletion, and controlled sleep fragmentation. This longitudinal design enabled simultaneous assessment of thermoregulation, sleep quantity, and sleep stability under both physiological conditions and during chronic sleep fragmentation in male and female mice. Rather than focusing solely on total sleep time, we also examined sleep-wake transitions and modeled the relationship between core body temperature and sleep to determine how microglial depletion influences the coordination of these interconnected physiological processes. This approach extends previous work by evaluating whether microglia contribute to integrated physiological homeostasis rather than sleep behavior alone. Based on the established roles of microglia in neuroimmune signaling and physiological homeostasis, we hypothesized that microglia are required to maintain thermoregulatory stability during acute sleep fragmentation. Specifically, we postulated that depletion of microglia would impair thermoregulatory stability during sleep fragmentation, resulting in exaggerated thermal fluctuations and altered homeostatic responses compared to microglia-intact controls.

Methods

Rigor

All animal studies were conducted in accordance with the guidelines set by the Institutional Animal Care and Use Committee (IACUC) at the University of Colorado Boulder (IACUC protocol: 2819) and the National Institutes of Health guidelines for the care and use of laboratory animals. Studies were designed and reported in accordance with the Animal Research: Reporting In Vivo Experiments guidelines [42]. To account for sex as a relevant biological variable, both male and female mice were included in the studies, and sex was included as a categorical variable in all statistical models. All analyses were performed by investigators blinded to experimental groups. Randomization procedures were implemented prior to study initiation, and predefined exclusion criteria were established before data collection began.

Animals

Adult (3–5-month-old) male and female C57BL/6J mice from an in-house breeding colony were used for all experiments (n total = 128; males n = 64; females n = 64). In-house breeder pairs were obtained from Jackson Laboratories (Bar Harbor, ME). Mice were group-housed with littermates until the initiation of the study protocol. Mice had ad libitum access to food and water throughout the study. Food intake and body weight were monitored regularly. Female mice were naturally cycling, and to limit handling stress and unwanted sleep disruption, estrous stage was not determined or controlled for as a variable. All procedures were performed between zeitgeber time (ZT) 4 and 6 to control for circadian influences on experimental outcomes. Accordingly, all data collection was aligned to begin at ZT 6, and each 24-hour cycle was segmented into three periods: the initial light period (ZT 6–11), the dark period (ZT 12–23), and the subsequent light period (ZT 0–5) [39].

For sleep protocols, all mice were singly housed and maintained on a 12-hour light:dark cycle (200 lux, cool white, fluorescent light) at an ambient temperature of 25°C ± 1°C, with controlled humidity. Due to the open top configuration of the cages, male and female cohorts were run at separate times to prevent pheromonal influences on sleep behavior. All mice were acclimated to non-invasive piezoelectric sleep cages for a minimum of 7 days prior to data collection to minimize novelty-induced alterations in sleep-wake behavior.

Prior to study initiation, mice were randomly assigned to an experimental group (PLX, control; Figure 1). The temperature implant surgery was performed under anesthesia with isoflurane, with depth of anesthesia monitored by toe pinch reflex and respiratory rate throughout the procedure. Exclusion criteria were predetermined: mice that lost >20% of their body weight or had visible signs of pain were humanely euthanized and excluded from the study (n = 5).

Figure 1. Experimental design for assessing the role of microglia in thermoregulation and sleep during sleep fragmentation.

Adult male and female C57BL/6J mice (n = 128) were acclimated to piezoelectric sleep recording cages for 7 days under standard housing conditions while maintained on a normal diet. Mice then underwent surgical implantation of intraperitoneal temperature transmitters (Anipill^®) for continuous core body temperature monitoring, followed by a 7-day postoperative recovery period with continued temperature recording. After recovery, mice were assigned to either a control AIN-76A diet or a PLX5622-formulated diet (1200 mg/kg) for 14 days to achieve pharmacological microglial depletion via CSF1R inhibition. During this period, continuous body temperature monitoring and noninvasive sleep-wake measurements were collected. Following the depletion period, mice underwent 7 consecutive days of automated sleep fragmentation using a motorized sweeper bar (2-min intervals) while remaining on their assigned diet. Core body temperature was continuously recorded throughout all experimental paradigms.

The group sizes used in statistical analyses were n = 28 control diet females, n = 31 control diet males, n = 32 PLX diet females, and n = 32 PLX diet males. Final sample sizes reflect exclusion of animals meeting predefined humane endpoints and did not differ significantly between groups.

Core Body Temperature Measures

Mice were anesthetized with 5% isoflurane in 100% oxygen for 5 minutes, and baseline body weights were collected. A small patch of fur was shaved from the lower quadrant of the abdomen to expose the peritoneal area. The mice were transferred to a nose cone with a continuous delivery of 2.5% isoflurane in 100% oxygen. The skin was aseptically prepared using alternating betadine and ethanol scrubs. A small incision was made to access the peritoneal cavity, and a sterilized Anipill^® temperature recording transmitter (Anipill^® System, BodyCAP, Hérouville Saint-Clair, France) was inserted into the intraperitoneal space. Care was taken to avoid manipulating visceral organs. The peritoneum was closed using absorbable sutures, and the external skin was closed with surgical staples. Postoperatively, mice received extended-release buprenorphine for analgesia, followed by subcutaneous sterile saline for hydration. Antibiotic ointment (bacitracin) was applied, and the mouse was returned to a heated holding cage. Mice were monitored for pain and ambulation before returning to their home cage. Mice were monitored daily for 3 days postoperatively to assess wound healing, body weight, and general health status. Skin staples were removed 7 days after surgery following confirmation of complete wound closure.

For temperature monitoring and data collection, the Anipill^® capsules continuously transmitted core body temperature data to an Anilogger^® Monitor placed within 1 meter of the animals. Data were recorded in real-time and stored in the internal memory of the capsule for synchronization when the mice were within communication range of the monitor. The sampling interval was set to 15-minute resolution, providing sufficient temporal resolution to capture sensitive temperature fluctuations. For data processing, temperature data were exported from the Anilogger^® Monitor using the Anilogger^® Manager Software and saved as comma-separated value (CSV) files for further processing. All temperature traces were visually inspected for signal artifacts prior to statistical analysis, and any missing or corrupted data points were documented and excluded according to predefined criteria. For statistical analyses and graphical representation, four consecutive 15-minute measurements were averaged to generate hourly mean core body temperature values.

Plexxikon (PLX) Administration

Pharmacological inhibition of CSF1R using PLX5622 enables selective depletion of microglia [38–40]. After a 7-day recovery period from the temperature implant surgery, mice were placed on Control AIN-76A rodent chow (AIN-76A rodent chow, Research Diets, Inc., New Brunswick, NJ) or diet with PLX5622 (1200 mg/kg, Plexxikon; herein referred to as PLX) formulated in AIN-76A rodent chow for the remainder of the study. The 1200 mg/kg dose was selected based on prior studies that demonstrated robust (>90%) central microglial depletion [38–40, 43].

Piezoelectric sleep recording parameters

Sleep-wake behavior was determined using a noninvasive piezoelectric cage system (Signal Solutions, Lexington, KY, USA), as previously described [44–48]. Briefly, each cage has an open bottom that allows the mouse to be placed directly on a cage floor with bedding over a Polyvinylidene Difluoride (PVDF) sensor. Regular low-amplitude (~3 Hz) rhythmic signals correspond to sleep, whereas higher-amplitude, irregular signals indicate wakefulness [49]. Signals were analyzed at 2-second resolution and classified into sleep or wake states using SleepStats Data Explorer Version 4 (Signal Solutions), which calculates a sleep-wake decision statistic and automated decision threshold for each cage. Hourly sleep metrics were derived using a rolling average of the percentage of recording time spent asleep and total minutes slept per hour [50].

Sleep-wake transitions were quantified using a standardized transition-analysis pipeline for piezoelectric systems [51]. Sleep-wake decision statistic data exported at 2-second resolution were processed using an automated threshold-based classification algorithm to identify state changes between consecutive epochs. Transitions were defined as threshold crossings between sleep and wake states and were summed within hourly and period intervals for statistical analysis. This automated approach enables high-throughput, reproducible quantification of sleep fragmentation without manual threshold adjustment or subjective scoring [51].

Manual Sleep Fragmentation

All mice underwent seven consecutive days of sleep fragmentation. Sleep was fragmented by housing mice in automated sleep fragmentation cages (Lafayette Instrument Company, Model 80391, Lafayette, IN) that contain a motorized horizontal sweeper bar. This sweeper bar traverses the cage floor at 2-min intervals, a frequency selected to repeatedly interrupt sleep [14, 52]. This paradigm disrupts sleep continuity while permitting short sleep bouts, thereby modeling chronic sleep fragmentation rather than total sleep deprivation. The sleep fragmentation protocol was conducted continuously across the light and dark periods. Mice remained on their experimental diet (PLX or control) throughout the fragmentation protocol. Cages were inspected twice a day to ensure proper sweeper-bar function and to confirm normal grooming, feeding, and ambulation behaviors.

Longitudinal Body Weight Measurements

Mice were weighed longitudinally to monitor overall health, recovery from surgical procedures, and tolerability of experimental manipulations. Body weights were recorded prior to temperature device implantation (baseline), on post-operative days 1, 2, 3, and 7, and again at the initiation of the sleep fragmentation protocol.

Statistical Analyses

We used hierarchical generalized linear mixed models to reveal differences in sleep and body temperature between diet groups and sexes over time [53, 54]. We fit all models using the packages glmmTMB and brms in the R statistical computing environment [55, 56]. Because outcome measures of interest were positive continuous variables (body temperature) and overdispersed counts (total sleep [number of minutes] and number of transitions), we specified Gaussian and negative-binomial response distributions, respectively, in the corresponding models [57, 58]. Statistical inference for generalized linear mixed-effects models, planned pairwise comparisons were performed using estimated marginal means. The experimental unit for all inferential analyses was the individual animal, with repeated observations within animals accounted for by random intercepts.

Body temperatures were measured hourly during two experimental paradigms, microglial depletion (PLX diet administration) only and microglial depletion concurrent with sleep fragmentation. We fit Gaussian mixed-effects models that included a four-way fixed effects interaction among diet, sex, a nonlinear effect of time (ZT) modeled via basis splines, and day [44, 45]. We included random intercepts for individual mice to account for the repeated measurements across hours and days [44, 45]. For the planned comparisons reported in the Results, estimated marginal means were used to compare diet groups and sexes within each experimental day and light/dark interval

Total minutes slept and the number of transitions between arousal states were calculated daily during PLX diet administration. We fit negative-binomial mixed-effects models to each outcome that included a four-way fixed effects interaction among diet, sex, light/dark period, and day. We included random intercepts for individual mice to account for repeated measurements within animals across light/dark periods and days.

To investigate whether body temperature predicted sleep, we fit period-specific (light and dark periods) Bayesian negative-binomial mixed-effects models to the total hourly minutes slept (bounded between 0 and 60 minutes per hour). Models included a three-way fixed effects interaction among hourly body temperature sex, and diet. These predictive models also included crossed random intercepts for individual animals and day to account for repeated measurements within animals and clustering within days. The complexity of these predictive models necessitated fitting them in a Bayesian framework, for which we specified Normal(0,1) priors for population-level effects (i.e., fixed effects parameters) and half-Cauchy(0,2) priors for group-level effects (i.e., random intercepts parameters) and variance components (i.e., shape/scale parameters) [59]. Posterior comparisons were conducted to estimate the posterior probabilities, given the observed data and specified priors, of diet-dependent differences in the temperature-sleep relationship within each sex and light/dark period.

Figures display raw observations from individual animals together with model-derived conditional estimates and 95% confidence intervals. Sample sizes for each experimental group are provided in the corresponding figure legends.

Results

Microglial depletion differentially alters body temperature and total sleep during week 1 of depletion

Mice were administered control or PLX diets for a two-week microglial depletion period. During the first week, microglial depletion produced time-of-day-dependent increases in body temperature, with effects that were more pronounced in females (Figure 2; Supplementary Figure 1). In females, microglial depletion significantly increased body temperature beginning within the first day of depletion, particularly during the dark period (Day 1: p = 0.0060), and these increases persisted across multiple days (Days 2–6: all p ≤ 0.0075). Elevated temperature in females also extended into the second light period (ZT 0–5), with significant effects observed on Day 2 (p < 0.0001) and Day 3 (p = 0.0299). In contrast, males exhibited a more limited and transient response, with a modest increase in temperature observed only during the first light period (ZT 6–11) on Day 1 (p = 0.0085), and no consistent effects across subsequent days or light/dark periods. Consistent with these findings, baseline sex differences in temperature were evident during the light period, with males exhibiting lower temperature than females across both light periods (e.g., Day 1 Light1: p = 0.0104; Light2: p = 0.0133).

Figure 2: Longitudinal body temperature and sleep during microglial depletion.

Core body temperature (top panels) and total sleep (bottom panels; min/h) across the light-dark cycle in male and female mice fed either control diet or PLX5622 during the microglial depletion period. Panels are organized by day: (A) Day 1, (B) Day 3, (C) Day 7, and (D) Day 14. Across all days, both body temperature and sleep exhibit robust time-of-day patterning, with higher temperature and reduced sleep during the dark period, and lower temperature with increased sleep during the light period. Differences between control and PLX-treated groups, as well as sex differences, vary across days and time-of-day, as indicated by statistical annotations. Each raw data point represents the hourly value from an individual animal at the indicated zeitgeber time (ZT). Solid lines show model-derived conditional estimates, and shaded bands indicate 95% confidence intervals. Gray vertical bands denote the dark, or active, period. Axis limits were selected to display the complete range of individual observations and were standardized across comparable panels to facilitate direct comparison among days, sexes, and treatment groups. Sample sizes were: n = 28 control diet females, n = 31 control diet males, n = 32 PLX diet females, and n = 32 PLX diet males. Statistical annotations indicate planned model-based comparisons at the corresponding time-of-day interval: *diet effect, p < 0.05; #sex effect, p < 0.05.

During the first week of PLX administration, microglial depletion produced only limited effects on total sleep. In females, microglial depletion resulted in less sleep compared to controls on Day 2 (p = 0.0217), whereas no significant effects of microglial depletion were observed on subsequent days or in males across either the light or dark periods (all remaining p > 0.05). Sex differences in total sleep were also minimal during week 1, with only a transient difference observed on Day 4 (p = 0.0416), reflecting slightly greater sleep in males compared to females.

Temperature effects attenuate and reveal microglia-dependent sex differences in week 2 of depletion

During the second week of microglial depletion, the initial temperature increases observed in females were no longer present, indicating an attenuation of the early thermogenic response (Figure 2; Supplementary Figure 2). Instead, a reversal in the direction of effect emerged, with PLX-treated females exhibiting reduced body temperature during the second light period (ZT 0–5), reaching significance on Day 12 (p = 0.0129). No consistent temperature differences were observed in males during week 2 across the light or dark periods (all p > 0.05). Across week 2, robust sex differences in temperature persisted during the light period, with males consistently exhibiting lower temperature than females indicating a stable, intrinsic sex difference in thermoregulation during the rest period. In contrast, sex differences in the dark period remained minimal under control conditions, but emerged under PLX treatment, with significant differences observed on Day 10 (p = 0.0159), Day 11 (p = 0.0315), Day 12 (p = 0.0218), and Day 14 (p = 0.00171).

In contrast to the limited effects observed during week 1, microglial depletion produced a robust and sex-specific increase in total sleep during week 2 that was confined to females (Figure 2; Supplementary Figure 2). In females, PLX significantly increased total sleep across both light and dark periods throughout week 2. In the dark period, significant increases were observed on Day 8 (p = 0.0312), Day 9 (p = 0.0348), Day 10 (p = 0.0349), Day 12 (p = 0.0236), Day 13 (p = 0.0180), and Day 14 (p = 0.0318). In the light period, microglial depletion also significantly increased total sleep on Day 8 (p = 0.0338), Day 9 (p = 0.0320), Day 10 (p = 0.0486), Day 11 (p = 0.0308), Day 12 (p = 0.0313), Day 13 (p = 0.0395), and Day 14 (p = 0.0355). These effects corresponded to a large increase in total sleep in PLX-treated females relative to controls. In contrast, no significant effects of PLX on total sleep were observed in males in either light or dark periods at any time point (all p > 0.05). During week 2, a consistent sex difference in total sleep also emerged, with males exhibiting significantly greater total sleep than females across multiple days, including Day 8 (p = 0.00905), Day 9 (p = 0.0114), Day 10 (p = 0.0262), Day 11 (p = 0.0360), Day 12 (p = 0.0168), Day 13 (p = 0.0110), and Day 14 (p = 0.0112).

Microglial depletion reduces sleep-wake transitions in males

Sleep-wake transitions were analyzed as an indicator of sleep fragmentation, with a higher number of transitions reflecting more fragmented sleep. Microglial depletion produced a sex- and time-of-day-specific reduction in sleep-wake transitions that was confined to males and most prominent during the dark period (Figure 3; Supplementary Figure 3). In males, PLX significantly reduced transitions in the dark period beginning late in week 1 and persisting throughout week 2, including Day 3 (p = 0.031), Day 4 (p = 0.049), Day 6 (p = 0.000344), and Day 7 (p = 0.0003), with stronger and more consistent effects observed across week 2 (Day 8: p < 0.0001; Day 9: p < 0.0001; Day 10: p = 0.0008; Day 11: p = 0.0003; Day 12: p < 0.0001; Day 13: p = 0.0018; Day 14: p = 0.0016).

Figure 3: Microglial depletion reduces sleep-wake transitions in male mice.

Number of sleep-wake transitions during the light and dark periods in male and female mice fed either control diet or PLX during the two-week microglial depletion period. Microglial depletion produced a sex- and time-of-day-specific reduction in sleep-wake transitions that was confined to males and most prominent during the dark period. In males, PLX significantly reduced transitions beginning late in week 1 and persisting throughout week 2, with consistent reductions observed across multiple days of the dark period. In contrast, no significant effects of PLX on transitions were observed in females at any time point. Effects during the light period were limited and inconsistent, with only isolated differences observed late in week 2. Across both treatment conditions, robust sex differences in transitions were present during the light period, with males exhibiting fewer transitions than females across all days. In contrast, no baseline sex differences were observed in the dark period under control conditions, but sex differences emerged during week 2 under PLX treatment. Each raw data point represents the hourly value from an individual animal at the indicated zeitgeber time (ZT). Solid lines show model-derived conditional estimates, and shaded bands indicate 95% confidence intervals. Gray vertical bands denote the dark, or active, period. Axis limits were selected to display the complete range of individual observations and were standardized across comparable panels to facilitate direct comparison among days, sexes, and treatment groups. Sample sizes were: n = 28 control diet females, n = 31 control diet males, n = 32 PLX diet females, and n = 32 PLX diet males. Statistical annotations indicate planned model-based comparisons at the corresponding time-of-day interval: *diet effect, p < 0.05; #sex effect, p < 0.05.

These reductions corresponded to an approximate 15–20% decrease in sleep-wake transitions in PLX-treated males relative to controls. In contrast, PLX effects in the light period were limited and inconsistent, reaching significance only on Day 13 (p = 0.018) and Day 14 (p = 0.036), with no sustained pattern across earlier days. No significant PLX effects on transitions were observed in females in either light or dark periods at any time point (all p > 0.05).

Sex differences in sleep-wake transitions are modified by microglial depletion

A robust and consistent sex difference in sleep-wake transitions was observed during the light period, with males exhibiting significantly fewer transitions than females across all days and in both treatment conditions (Figure 3; Supplementary Figure 3). In control animals, this sex difference was present on every day, including Day 1 (p = 0.0019), Day 2 (p = 0.0026), and Days 3–14 (all p < 0.0001). A similar but even stronger pattern was observed in PLX-treated animals, with significant sex differences across all days (Day 1: p < 0.001; Days 2–14: all p < 0.001). In contrast, no significant sex differences were observed in the dark period under control conditions (all p > 0.05), indicating the absence of a baseline sex effect during the active period. However, PLX treatment introduced a sex-specific divergence in the dark period during week 2, with significant differences emerging on Day 8 (p = 0.00113), Day 9 (p < 0.001), Day 10 (p < 0.001), Day 11 (p < 0.001), Day 12 (p < 0.001), Day 13 (p < 0.001), and Day 14 (p < 0.001).

Microglial depletion alters thermoregulation in a sex- and time-of-day-dependent manner during sleep fragmentation

Mice were subjected to seven days of sleep fragmentation. During this period, microglial depletion produced robust sex- and time-of-day-dependent alterations in body temperature, revealing a critical role for microglia in maintaining thermoregulatory stability under physiological stress (Figure 4; Supplementary Figure 4). In males, microglial depletion resulted in a consistent reduction in body temperature during the dark (active) period, with significant decreases observed on Day 2 (p = 0.0460), Day 3 (p = 0.0102), Day 4 (p = 0.0116), Day 5 (p = 1.82 × 10⁻⁴), Day 6 (p = 0.00549), and Day 7 (p = 0.0166).

Figure 4: Longitudinal body temperature during sleep fragmentation.

Body temperature across the light-dark cycle in male and female mice fed either control diet or PLX during seven days of sleep fragmentation. Panels are organized by day: (A) Day 1, (B) Day 3, (C) Day 5, and (D) Day 7. Across days, body temperature exhibits clear time-of-day patterning, with higher temperatures during the dark period and lower temperatures during the light period. Differences between diet groups and between sexes are observed in a time- and day-dependent manner, as indicated by statistical annotations Each raw data point represents the hourly value from an individual animal at the indicated zeitgeber time (ZT). Solid lines show model-derived conditional estimates, and shaded bands indicate 95% confidence intervals. Gray vertical bands denote the dark, or active, period. Axis limits were selected to display the complete range of individual observations and were standardized across comparable panels to facilitate direct comparison among days, sexes, and treatment groups. Sample sizes were: n = 28 control diet females, n = 31 control diet males, n = 32 PLX diet females, and n = 32 PLX diet males. Statistical annotations indicate planned model-based comparisons at the corresponding time-of-day interval: *diet effect, p < 0.05; #sex effect, p < 0.05.

In contrast, females exhibited reductions in body temperature that were primarily confined to the second light period (ZT 0–5), with significant decreases observed on Day 1 (p = 0.0341), Day 2 (p = 0.0151), Day 5 (p = 0.0436), and Day 6 (p = 0.0198). A more limited effect was observed in the dark period in females, reaching significance only on Day 5 (p = 0.0407).

Sleep fragmentation reveals time-of-day-dependent sex differences in thermoregulation

Sleep fragmentation also revealed dynamic and time-of-day-dependent sex differences in body temperature (Figure 4; Supplementary Figure 4). During the dark period, males exhibited higher body temperature than females under both control and PLX conditions early in the sleep fragmentation paradigm (Day 1: p = 0.0394; Day 2: p = 0.0261; Day 3: p = 0.00738), indicating a reversal of the baseline pattern observed during the light period. In contrast, during the light period, males consistently exhibited lower body temperature than females across multiple days (e.g., Day 5 Light1: p < 0.001; Day 6 Light1: p < 0.001; Day 5 Light2: p < 0.0001; Day 6 Light2: p < 0.0001).

Body temperature predicts total sleep

To determine whether body temperature predicts sleep, we modeled hourly total sleep as a function of body temperature using Bayesian hierarchical negative-binomial models. A statistically significant nonlinear relationship between body temperature and sleep was estimated during week 1 of the microglial depletion period (Figure 5). Across both light and dark periods, higher body temperatures were associated with reduced sleep, with a pronounced threshold-like effect such that sleep declined steeply as temperature exceeded approximately 35°C. Posterior probabilities supported an inverse relationship between body temperature and sleep during week 1 (posterior probability, P = 0.99), providing very strong statistical support for this effect.

Figure 5: Body temperature nonlinearly predicts total sleep.

Hourly total sleep (minutes) is plotted as a function of hourly body temperature during the microglial depletion period in female and male mice maintained on either control or PLX5622 (PLX) diet. Panels are separated by light period (A) and dark period (B), with sex-specific conditional effects shown for females and males. Across all groups, a significant nonlinear inverse relationship between body temperature and sleep was observed, whereby lower temperatures were associated with increased sleep, and higher temperatures were associated with a steep decline in sleep, particularly above ~35°C. During the light period, PLX-treated mice exhibited a stronger negative relationship between body temperature and sleep compared to controls, with this effect more pronounced in males than females. In contrast, during the dark period, the relationship between temperature and sleep was similarly robust across diet groups and sexes, indicating minimal influence of microglial depletion on the temperature-sleep relationship during the active period. Raw data points (gray dots) are shown in the background, with model-estimated posterior conditional effects (solid lines) and 95% credible intervals (highest posterior density intervals; shaded bands) overlaid. Estimates reflect population-level effects derived from Bayesian hierarchical negative-binomial models that accounted for repeated measures across animals and days.

During the light period, the magnitude of the temperature-sleep relationship differed by both sex and diet. In females maintained on the control diet, body temperatures between 35–38°C were associated with an approximate 5.82-minute reduction in hourly sleep for each 1°C increase in body temperature, whereas PLX-treated females exhibited a stronger decline of approximately 6.57 minutes per 1°C increase. Strong evidence existed for a diet-dependent effect in females (posterior probability of diet difference: P(Δ[Control vs PLX] ≠ 0) = 0.98). In males, the relationship was more pronounced, with control diet males exhibiting an approximate 8.31-minute reduction in hourly sleep per 1°C increase in temperature, and PLX-treated males exhibiting an approximate 9.46-minute reduction. A diet-dependent effect in males was also strongly supported (P(Δ[Control vs PLX] ≠ 0) = 0.99). Collectively, these findings indicate that during the light period, microglial depletion strengthened the inverse relationship between body temperature and sleep, with effects that were more pronounced in males than females.

In contrast, during the dark period, the inverse relationship between body temperature and sleep was uniformly strong across sexes and diet groups but was unaffected by diet. In females, body temperatures between 35–38°C were associated with an approximate 10.36-minute reduction in hourly sleep per 1°C increase under control conditions and a 10.32-minute reduction under PLX treatment. A diet-dependent effect in females during the dark period was unsupported (P(Δ[Control vs PLX] ≠ 0) = 0.11). Similarly, in males, hourly sleep declined by approximately 10.51 minutes per 1°C increase in temperature under control conditions and by 10.31 minutes under PLX treatment, with negligible evidence for a diet-dependent effect (P(Δ[Control vs PLX] ≠ 0) = 0.32). Overall, the inverse relationship between body temperature and sleep was substantially stronger during the dark period (mean Δ_35–38°C = −10.38) than during the light period (mean Δ_35–38°C= −7.54), indicating that elevated body temperature more strongly constrained sleep during the active period.

A second Bayesian model was fitted using data from week 2 of the microglial depletion period (Supplementary Figure 5). Consistent with week 1 findings, a strong nonlinear, inverse relationship between body temperature and sleep existed across both light and dark periods, with higher temperatures associated with reduced sleep and a steep decline in sleep above approximately 35°C. However, unlike week 1, there was minimal evidence during week 2 that microglial depletion altered the relationship between body temperature and sleep within either sex or light/dark period. During the light period, posterior probabilities supporting a diet-dependent effect were low in both females (P(Δ[Control vs PLX] ≠ 0) = 0.53) and males (P(Δ[Control vs PLX] ≠ 0) = 0.86). Similarly, during the dark period, posterior probabilities remained below the predefined threshold for evidence of a diet effect in females (P(Δ[Control vs PLX] ≠ 0) = 0.78) and males (P(Δ[Control vs PLX] ≠ 0) = 0.65).

Discussion

Results of our study demonstrate that microglial depletion differentially alters sleep architecture and thermoregulation in a sex- and time-of-day-dependent manner, thereby dissociating sleep quantity from sleep stability and revealing a strong nonlinear relationship between body temperature and sleep. These findings suggest that microglia may act as critical regulators of coordinated sleep and thermoregulatory homeostasis, particularly under conditions of disrupted sleep. Whereas previous studies of PLX5622-mediated microglial depletion have primarily focused on sleep quantity and conventional measures of sleep architecture, the present study extends this literature by demonstrating that microglial depletion also differentially influences thermoregulatory homeostasis, sleep stability, and the relationship between body temperature and sleep in a sex- and time-of-day-dependent manner. These findings further indicate that sex differences in total sleep may become more pronounced over time and are largely independent of PLX treatment; although, microglial depletion selectively increases sleep in females. In contrast, whereas sex differences in sleep-wake transitions are intrinsically present during the light period, microglial depletion selectively unmasks or amplifies these differences during the dark period, highlighting a time-of-day-dependent role for microglia in regulating sleep stability.

Microglia Differentially Regulate Sleep Quantity and Sleep Stability

A central finding of this study is the dissociation between sleep quantity and sleep stability. Microglial depletion produces a striking sex-specific divergence in sleep architecture; in males, depletion selectively reduces sleep-wake transitions without altering total sleep, whereas in females, depletion increases total sleep without affecting sleep-wake transitions. Notably, this pattern builds on our prior work that demonstrated inherent sex differences in sleep architecture under baseline conditions, when female mice exhibited less total sleep [45], and increased sleep-wake transitions, compared to males [51], which indicates greater intrinsic sleep fragmentation. Together, these findings indicate that microglia may not uniformly regulate sleep but may instead differentially control distinct components of sleep architecture in a sex-dependent manner. By distinguishing effects on sleep quantity from sleep stability, these findings extend previous work and suggest that sleep fragmentation provides complementary physiological information beyond measures of total sleep time alone.

Importantly, this interpretation is consistent with our prior work in male mice, wherein microglial depletion during the first week of PLX exposure produced a transient increase in sleep bout length, which suggests more consolidated sleep [39]. This pattern aligns with the present findings, whereby males on PLX diet exhibited reduced sleep-wake transitions indicative of increased sleep consolidation. This dissociation is particularly important given sleep fragmentation and total sleep are often treated as interchangeable metrics, despite emerging evidence that they reflect separable physiological mechanisms [60]. These findings reinforce the concept that microglia may regulate sleep continuity rather than total sleep time, with effects that may only secondarily influence sleep quantity under specific conditions.

Mechanistically, reduced sleep-wake transitions in males may indicate that microglia contribute to regulating network instability and state switching, particularly during the active (dark) period when arousal systems are highly engaged. This pattern is consistent with a model in which microglia may modulate the excitability of arousal-promoting circuits, including hypothalamic hypocretin (orexin) circuits, which are critical for maintaining wakefulness and regulating state transitions [61, 62]. Under baseline conditions, microglial signaling may promote state transitions; therefore, depletion may stabilize these circuits, reducing fragmentation without increasing total sleep time. Consistent with this interpretation, our prior work demonstrated that under inflammatory conditions following traumatic brain injury, mice exhibit increased sleep-wake transitions, indicating heightened sleep fragmentation [44]. In contrast, in females, increased total sleep may suggest that microglia preferentially regulate sleep drive or homeostatic sleep processes, potentially through neuroimmune signaling pathways, including cytokine-mediated regulation of sleep propensity [36, 63]. Given well-established sex differences in microglial development, activation, and inflammatory signaling, these findings support a model in which microglia exert sex-dependent control over distinct components of sleep homeostasis [64, 65].

Microglia Coordinate Thermoregulatory Homeostasis During Sleep Fragmentation

Microglial depletion also produces dynamic, sex- and time-of-day-dependent alterations in thermoregulation, revealing an additional physiological axis through which microglia influence sleep. Although previous studies have linked microglia to sleep regulation, comparatively little is known about their role in coordinating thermoregulation during sleep disruption. Our findings identify thermoregulatory homeostasis as an additional physiological process influenced by microglial depletion, broadening the functional consequences of microglia beyond sleep behavior alone. During baseline conditions, depletion increases body temperature in females early in the depletion period, whereas males show minimal changes, consistent with known sexual dimorphism in microglial function and inflammatory responsiveness [59, 64, 66–68]. Over time, these effects attenuate and reverse, which may suggest adaptive or compensatory mechanisms in thermoregulatory control. Notably, early during PLX exposure, baseline sex differences in thermoregulation are preserved, indicating that initial microglial depletion primarily modulates overall temperature without disrupting intrinsic sex-dependent regulatory patterns. These findings suggest that under baseline conditions, microglia may act to buffer or constrain sex differences in thermoregulation, with their depletion revealing latent sex-specific divergence. However, during sleep fragmentation, microglial depletion reveals a distinct and robust thermoregulatory phenotype whereby males exhibit sustained hypothermia during the dark period and females show reductions primarily during the late light period. Thus, microglia may be required to maintain thermoregulatory stability under conditions of sustained sleep disruption, likely through modulation of hypothalamic circuits that integrate immune, thermal, and behavioral signals [65]. Consistent with this interpretation, recent work demonstrated that microglia are required for the increase in slow-wave sleep induced by elevated ambient temperature [33], supporting the concept that microglia contribute to coupling thermal state with sleep regulation.

Importantly, these findings suggest that sleep fragmentation may reorganize thermoregulatory sex differences in a time-of-day-dependent manner. Although baseline sex differences are primarily observed during the light period, sleep fragmentation and microglial depletion unmask or amplify sex differences during the dark period. This time-of-day-dependent reversal suggests that thermoregulation may be dynamically regulated across behavioral states and that microglia may contribute to coordinating these responses across circadian and physiological contexts. The emergence of sustained hypothermic responses in PLX-treated animals, combined with time-of-day-specific reversals in sex differences, may further indicate that microglia are essential for maintaining stable thermoregulatory control under conditions of sleep fragmentation.

Microglia Integrate Temperature–Sleep Coupling

The integration of thermoregulatory and sleep findings is further supported by the strong nonlinear relationship between body temperature and sleep that was estimated by our predictive models. Elevated body temperature was associated with reduced sleep, with the steepest sleep reduction above a temperature threshold of approximately 35°C; this is consistent with extensive evidence that thermoregulation is a critical determinant of sleep propensity [20, 69]. This threshold-like relationship suggests that temperature actively constrains sleep rather than passively reflects it. Importantly, the magnitude of this relationship differed across light and dark periods and was further modulated by sex and microglial depletion. During the light period, PLX-treated mice exhibited a stronger inverse relationship between body temperature and sleep compared to control animals, with these effects more pronounced in males than females. Posterior comparisons further indicated strong evidence that microglial depletion amplified the negative relationship between body temperature and sleep during the light period, such that PLX-treated mice slept less than controls as temperature increased. Although this effect was observed in both sexes, the magnitude of the relationship was nearly two-fold greater in males than females, suggesting sex-dependent sensitivity of temperature-sleep coupling following microglial depletion.

In contrast, during the dark period, the relationship between elevated temperature and reduced sleep was uniformly strong across sexes and diet groups, suggesting that the influence of body temperature on sleep is greatest during the active period and less sensitive to microglial depletion under these conditions. Notably, the diet-dependent amplification of the temperature-sleep relationship that was present during the light period appeared to diminish over time: Prolonged PLX exposure during week 2 no longer produced strong evidence for sex- or period-specific diet effects despite the persistence of a robust inverse relationship between body temperature and sleep. These findings may suggest adaptive or compensatory responses emerge following sustained microglial depletion, resulting in stabilization of the temperature-sleep relationship over time. Within this framework, the effects of microglial depletion can be interpreted as coordinated alterations in both temperature and sleep. In females, increased total sleep coincides with reduced body temperature, whereas in males, reduced transitions and hypothermia during sleep disruption suggest that microglia may regulate sleep stability, in part, through thermoregulatory pathways.

Potential Neuroimmune Mechanisms Linking Microglia, Sleep, and Thermoregulation

A potential mechanism underlying these findings involves microglial cytokine signaling. Pro-inflammatory cytokines such as interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) are well-established regulators of sleep and are known to act within hypothalamic circuits to influence both sleep and thermoregulation [36, 63]. Microglia are a primary source of these signals in the central nervous system, particularly under conditions of physiological stress [39, 70]. Although cytokine signaling was not directly evaluated in the present study, disruption of microglial signaling through PLX-mediated depletion has been shown in previous studies following inflammatory challenge [40, 43], and could thereby be contributing to the observed changes in sleep and thermoregulation. Future studies incorporating central and peripheral immune profiling during sleep fragmentation will be necessary to determine whether inflammatory signaling mediates the physiological changes observed following microglial depletion.

Collectively, these findings support a unifying model in which microglia may act as central regulators of physiological homeostasis, coordinating interactions among sleep architecture, thermoregulation, and neural circuit stability. This interpretation is consistent with emerging frameworks proposing hypothalamic microglia function as active regulators of physiological homeostasis rather than solely inflammatory responders [65]. Microglia do not exert uniform effects on sleep; rather, they appear to differentially regulate sleep quantity and stability in a sex-specific and time-of-day-dependent manner. Under baseline conditions, microglia may modulate circadian temperature set points and sleep patterns, whereas under conditions of sleep disruption, microglia become essential for maintaining stability across both thermoregulatory and sleep systems. The dissociation between temperature, total sleep, and sleep fragmentation suggests that microglia may independently regulate multiple components of physiological function, including sleep drive, network stability, and energy balance. If true, these findings position microglia as key integrators of neuroimmune and physiological processes that enable adaptive responses to environmental and physiological stressors and highlight microglial signaling as a potential therapeutic target for sleep and circadian disturbances.

Methodological Considerations

Several methodological considerations should be considered when interpreting our results. First, we did not assess sleep stage (NREM, REM), and the noninvasive piezoelectric monitoring does not capture spectral features, such as delta power. Consequently, our findings should be interpreted as reflecting changes in sleep-wake behavior and continuity rather than definitive alterations in sleep-stage architecture or electrophysiological features of sleep. However, this noninvasive approach enables continuous, longitudinal assessment of sleep-wake behavior in large cohorts without the confounding effects of surgical implantation and tethering required for electroencephalography/electromyography (EEG/EMG) (i.e., sleep data are obtained noninvasively). As such, our findings emphasize robust, behaviorally relevant measures of sleep quantity and stability that are well-suited for detecting systems-level physiological changes. Future studies incorporating EEG/EMG will be important for determining whether microglial depletion differentially affects sleep-stage specific amplitudes or oscillatory dynamics. Second, although our analyses focused on time-of-day effects, we did not explicitly quantify circadian rhythmicity metrics under free-running conditions (e.g., amplitude, phase, or intra-daily variability), which may also be regulated by microglia. However, by structuring analyses across defined ZT windows, we were able to capture functionally meaningful, sleep-dependent effects while minimizing model complexity. Future work incorporating formal circadian analyses will be critical for determining whether microglia contribute to endogenous rhythm generation or stability.

We also note that implantation of temperature telemetry devices may induce a localized inflammatory response that could influence both thermoregulation and sleep. However, all animals underwent identical surgical procedures and sufficient recovery periods, ensuring that any residual effects were consistent across groups. Thus, the observed differences are unlikely to reflect surgical artifacts but instead represent the interaction between microglial depletion and physiological regulation. Finally, microglial depletion was achieved using PLX5622, a CSF1R inhibitor that produces global depletion of microglia. This approach is widely used and highly effective, but CSF1R inhibition may have off-target effects, including potential impacts on peripheral macrophage populations or other CSF1R-expressing cells [40, 43]. As such, the present findings should be interpreted in the context of global microglial depletion. Future studies using cell-type-specific or region-specific approaches will be important for isolating the precise microglial populations and circuits responsible for regulating sleep and thermoregulation. Finally, although microglia are well-established regulators of neuroimmune signaling, we did not directly quantify central or peripheral inflammatory mediators following sleep fragmentation or PLX5622 treatment. Consequently, the molecular pathways linking microglial depletion to the observed physiological phenotypes remain to be determined and represent an important direction for future investigation.

Conclusions

In summary, our results suggest that microglia are likely required for maintaining thermoregulatory stability during sleep fragmentation and reveal sex-specific roles for microglia in coordinating temperature and sleep homeostasis. Microglial depletion disrupts both thermoregulation and sleep-wake behavior, particularly under conditions of sleep fragmentation, and uncovers a strong inverse relationship between core body temperature and sleep. These results suggest that microglia contribute to coordinating neuroimmune, thermoregulatory, and sleep processes and identify microglial signaling as a potential therapeutic target for mitigating physiological dysregulation associated with sleep disturbance.

These findings may have direct clinical relevance for human populations exposed to chronic sleep fragmentation, including individuals with insomnia, shift work schedules, obstructive sleep apnea, and neurological injury. Disruptions in thermoregulation and sleep are hallmark features of these conditions and are often accompanied by heightened inflammatory signaling. The identification of microglia as a central mediator linking these processes provides a mechanistic framework for understanding how sleep disruption contributes to systemic and neurological dysfunction. Although the mechanisms remain to be fully established, our findings provide a framework for understanding how microglia may contribute to the physiological consequences of sleep disruption. Furthermore, interventions targeting neuroimmune signaling or optimizing thermoregulatory conditions may improve sleep quality and physiological resilience. More broadly, this work highlights sleep continuity and thermal homeostasis as fundamental, interdependent components of physiological health.

Supplementary Material

1

Supplementary Figure 1: Longitudinal body temperature and total sleep during week 1 of microglial depletion. Core body temperature (top panels) and total sleep (bottom panels; min/h) across the light-dark cycle in male and female mice fed control or PLX5622 diet during the first week of microglial depletion. Both measures exhibited robust time-of-day patterning across days, with higher temperature and reduced sleep during the dark period and lower temperature with increased sleep during the light period. During week 1, PLX5622 primarily increased body temperature in females, particularly during the dark period, whereas effects on total sleep were minimal. Raw data points are overlaid with conditional effects estimates (solid lines) and 95% confidence intervals (shaded bands). Gray vertical bands indicate the dark period. ZT = zeitgeber time. *p < 0.05 (diet effect); #p < 0.05 (sex effect).

2

Supplementary Figure 2: Longitudinal body temperature and total sleep during week 2 of microglial depletion. Core body temperature (top panels) and total sleep (bottom panels; min/h) across the light-dark cycle in male and female mice fed control or PLX5622 diet during the second week of microglial depletion. Both measures continued to exhibit robust time-of-day patterning across days. During week 2, early PLX-induced increases in body temperature attenuated, whereas PLX-treated females exhibited increased total sleep across both light and dark periods. Raw data points are overlaid with conditional effects estimates (solid lines) and 95% confidence intervals (shaded bands). Gray vertical bands indicate the dark period. ZT = zeitgeber time. *p < 0.05 (diet effect); #p < 0.05 (sex effect).

3

Supplementary Figure 3: Sleep-wake transitions during microglial depletion. Number of sleep-wake transitions across the light-dark cycle in male and female mice fed control or PLX5622 diet. Raw data (points) are overlaid with model-estimated means (lines) ± 95% confidence intervals (shaded). *p < 0.05 (diet effect); #p < 0.05 (sex effect).

4

Supplementary Figure 4: Longitudinal body temperature during sleep fragmentation. Body temperature across the light–dark cycle in male and female mice fed control or PLX5622 diet during sleep fragmentation. Raw data (points) are overlaid with model-estimated means (lines) ± 95% confidence intervals (shaded). Gray bands indicate the dark period. *p < 0.05 (diet effect); #p < 0.05 (sex effect). ZT = zeitgeber time.

5

Supplementary Figure 5: Body temperature nonlinearly predicts total sleep during week 2 of microglial depletion. Hourly total sleep (minutes) is plotted as a function of hourly body temperature in female and male mice maintained on either control or PLX5622 (PLX) diet. Panels are separated by light period (A) and dark period (B). Across all groups, higher body temperatures were associated with reduced sleep, with a steep decline in sleep above ~35°C. Unlike week 1, minimal diet-dependent differences were observed during week 2 across either sex or light/dark period. Raw data points (gray dots) are overlaid with model-predicted conditional effects (solid lines) and 95% credible intervals (shaded bands). Estimates reflect Bayesian hierarchical negative-binomial models adjusted for repeated measures across animals and days.

Highlights.

• Microglia regulate sleep stability and thermoregulatory homeostasis

• Microglial depletion alters sleep and body temperature by sex and time-of-day

• Sleep fragmentation unmasks microglia-dependent thermoregulatory instability

• Body temperature nonlinearly predicts sleep across light and dark periods

• Microglia modulate temperature-dependent regulation of sleep

Data availability

Data will be made available on request.