Sex and stress paradigm influence behavioural, neuroimmune, and microbiome responses in pubertal CD-1 mice Authors: Michaela Dworsky-Fried, Umar Haris Iqbal, Stéphane Bronner, Jérémie Auger, Olivier Mathieu, Noémie Auclair-Ouellet, Jacky Liang, Nafissa Ismail Published: 2026 ABSTRACT AbstractPuberty is a critical developmental period marked by heightened vulnerability to stress. Exposure to pubertal stress can have lasting effects on behaviour, immune function, and the gut–brain axis. The NLRP3 inflammasome and gut microbiome are key regulators of neuroinflammation, yet their roles across distinct pubertal stress paradigms remain unclear. This study compared behavioural, immune, and microbial outcomes in 100 CD-1 mice following chronic unpredictable stress (CUS; 14 or 28 days) or 7-day sleep disruption (SD) in pubertal males and females. Behavioural testing revealed stressor- and sex-specific effects: females spent more time in the open arms of the elevated plus maze, SD reduced open-arm exploration, 14-day CUS reduced centre activity in the open field, and SD increased immobility in the forced swim test. Plasma cytokines showed divergent patterns, with IL-22 elevated across stress conditions and IL-17A highest following 28-day CUS and in females overall. In the prefrontal cortex, SD males exhibited increased NLRP3 and CASP1 mRNA expression, whereas 28-day CUS females showed reduced NFκB1, and NFκB2 expression was elevated in stress-exposed males. In the hippocampus, 28-day CUS females demonstrated coordinated reductions in NFκB1, NLRP3, and CASP1 expression. Microbiome analyses indicated that SD and 14-day CUS produced the most substantial alterations, characterized by reductions in short-chain fatty acid–producing taxa (Ligilactobacillus, Acetatifactor, Christensenella) and enrichment of stress-associated genera (Flintibacter, Oscillospiraceae). Together, these findings indicate that stressor type and sex differentially shape behavioural, neuroimmune, and microbial profiles during puberty, with sleep disruption exerting particularly strong effects on behavioural and gut–immune regulation.Keywords: Chronic stress, Neuroinflammation, Puberty, NLRP3 inflammasome, Gut microbiome, Gut-brain axis Introduction Puberty represents a dynamic intersection between brain maturation and stress regulation, during which neuroendocrine systems undergo profound changes. During this period, the brain undergoes substantial maturation, including synaptic pruning and myelination, driven in part by rising gonadal hormones following activation of the hypothalamic-pituitary-gonadal (HPG) axis (Larsen and Luna, 2018). These hormones interact with the hypothalamic–pituitary–adrenal (HPA) axis, the central regulator of stress, whose activity is initiated in the paraventricular nucleus (PVN) of the hypothalamus and modulated by the prefrontal cortex (PFC) and amygdala (AMY; Bains et al., 2015; Smith and Vale, 2006). PVN activation leads to corticotropin-releasing hormone and arginine vasopressin release, stimulating adrenocorticotropic hormone (ACTH) from the pituitary and glucocorticoid release (cortisol in humans, corticosterone in rodents) from the adrenal cortex (Stephens and Wand, 2012). While acute glucocorticoid elevations promote adaptive responses, chronic stress disrupts PFC, AMY, and hippocampal (HIPP) function, increasing the risk for psychopathology (de Kloet et al., 2005). In mice, a particularly stress-sensitive stage occurs around six weeks of age (∼ postnatal day 42), when stress exposure yields pronounced and enduring effects on neural and behavioural outcomes (Ismail et al., 2011). Pubertal gonadal hormones shape stress-sensitive brain regions and contribute to sex differences in stress responsivity. Specifically, estrogen enhances ACTH sensitivity and elevates corticosterone, while testosterone suppresses HPA axis activation via reduced vasopressin signalling (Figueiredo et al., 2007; Goel et al., 2014). These mechanisms may underlie the increased stress vulnerability observed in females, especially in the context of hormonal fluctuations (Holmes et al., 2003). Through interactions with neurotrophic factors, gonadal steroids also promote circuit refinement via synaptic pruning and myelination, reinforcing long-term changes in stress regulation (Hensch, 2005; Sisk and Zehr, 2005; Spear, 2013). Acute stress refers to a brief, single exposure to a stressor, whereas chronic stress involves repeated or prolonged exposure to a variety of stressors over an extended period (Baum, O'Keeffe and Davidson, 1990). Acute stress is typically adaptive, but prolonged exposure to stress or chronic stress can negatively impair brain structure and function and lead to maladaptive behavioural outcomes (Sheth et al., 2017). Acute stress activates the HPA axis, triggering a transient glucocorticoid response that typically suppresses immune activity. In contrast, chronic stress results in sustained activation of both the HPA axis and the immune system, promoting systemic inflammation and neuroinflammation. These prolonged immune responses contribute to more persistent anxiety- and depression-like behaviours in mice, reflecting the cumulative impact of chronic stress on brain function and emotional regulation (McCormick and Green, 2013; Papadopoulou et al., 2015; Tran and Gellner, 2023). The NOD-, LRR-, and pyrin domain-containing protein 3 (NLRP3) inflammasome plays a central role in stress-induced neuroinflammation by regulating microglial activation. Microglia shift between pro-inflammatory (M1) and anti-inflammatory (M2) phenotypes in response to peripheral cytokines crossing the blood-brain barrier (Tang and Le, 2016). Chronic stress promotes sustained M1 polarization, contributing to anxiety- and depression-like behaviours (Bisht et al., 2018). This shift is driven in part by NLRP3 inflammasome activation, which involves the NLRP3 sensor, ASC adaptor, and caspase-1, facilitating the maturation of interleukin 1β (IL-1β) and interleukin 18 (IL-18; Franklin et al., 2018; Kelley et al., 2019). NLRP3 also upregulates indoleamine 2,3-dioxygenase (IDO), linking neuroinflammation to the kynurenine pathway and depression-like behaviour (Jeon et al., 2017). Pro-inflammatory cytokines like interleukin 17A (IL-17A) exacerbate NLRP3 activity through IL-17R-mediated nuclear factor kappa B (NF-κB) signalling, a key step in inflammasome priming (Xu and Cao, 2010). In contrast, IL-22 binds to its receptor and activates IRF3 phosphorylation, inhibiting NF-κB and attenuating inflammation (Guo et al., 2022; Tarassishin et al., 2011). Thus, the balance between pro- and anti-inflammatory cytokines modulates NLRP3 activation and may underlie inflammation-associated emotional dysregulation. Preclinical models of chronic stress replicate behavioural, endocrine, and physiological changes that mirror features of human psychiatric disorders (Radley and Herman, 2023), offering a valuable framework for studying underlying mechanisms. One widely used paradigm is chronic unpredictable stress (CUS), also referred to as chronic variable or mild stress. In this model, rodents are exposed to different stressors once or twice daily over several weeks. Stressors include environmental changes, social isolation, mild restraint, and water/food deprivation, delivered in an unpredictable sequence to simulate real-life stress (Markov and Novosadova, 2022). However, inconsistencies in the implementation of CUS paradigms, including stressor type, duration, and frequency, can lead to variable outcomes across experiments, even when protocols are rigorously implemented. Sleep disruption (SD) represents an alternative chronic stress model that induces robust behavioural and physiological disturbances. Chronic alterations to sleep patterns elevate pro-inflammatory cytokines (Rohleder et al., 2012) and induce anxiety- and depression-like behaviours (Banks and Dinges, 2007; Gonzalez-Castañeda et al., 2016; Murack et al., 2021). For instance, mice deprived of sleep for 72 h show heightened anxiety-like behaviours (Silva et al., 2004), while adolescent mice exhibit greater depression-like responses following eight days of SD compared to adults (Murack et al., 2021). These findings emphasize the role of sleep disturbances in stress-related pathology and support improving sleep quality as a preventive strategy for psychiatric conditions (Meerlo et al., 2008). Despite its translational relevance, sleep disruption primarily targets sleep–wake regulation and does not capture the broader range of environmental and psychological stressors present in chronic stress. It also may trigger compensatory processes, such as circadian adaptation, which can attenuate long-term effects. At the same time, methodological variability may introduce confounds unrelated to sleep disruption itself. Emerging research highlights the gut-brain-axis (GBA) as a critical pathway linking chronic stress to mental health outcomes, particularly during puberty. The GBA is a bidirectional communication system between the gut microbiota and the central nervous system, influencing behaviour, emotional regulation, and neuroendocrine function (Dinan and Cryan, 2017; Mayer et al., 2022). Chronic stress can induce gut dysbiosis, an imbalance in the gut microbial community, by altering bacterial composition and diversity. Such disruptions often involve reductions in beneficial taxa such as Lactobacillus and Bifidobacterium alongside increases in potentially pathogenic bacteria like Enterobacteriaceae (Chen et al., 2024; Karl et al., 2018). These microbial shifts can compromise intestinal barrier integrity, increase permeability and allow for the translocation of pro-inflammatory molecules such as lipopolysaccharides (LPS) into systemic circulation, which in turn promotes systemic inflammation and neuroinflammatory processes associated with anxiety and depression (Bowers et al., 2020; Dantzer, 2017; Gao et al., 2019). Notably, LPS exposure can disturb the gut microbiome and provoke behavioural changes in a sex- and age-dependent manner, with pubertal mice, particularly males, exhibiting greater microbiota depletion and heightened anxiety-like behaviours than adults (Cai et al., 2016; Murray et al., 2019; Sharma et al., 2018). Together, these findings underscore the intricate interplay between gut dysbiosis, immune activation, and neurobehavioral outcomes, suggesting that the gut microbiome may be a promising target for mitigating stress-related psychopathology. While it is well established that chronic stress can activate the NLRP3 inflammasome and promote neuroinflammation, it remains unclear which chronic stress paradigm most strongly engages this pathway. Given stress's complex and variable nature, identifying specific models that reliably activate the NLRP3 inflammasome both centrally and peripherally is critical to advancing our understanding of stress-related pathology. Compared to SD, CUS exposes animals to a diverse and unpredictable sequence of stressors that may more robustly engage multiple physiological stress pathways, leading to sustained glucocorticoid elevations, heightened immune activation, and broader behavioural effects. Furthermore, as the gut-brain axis plays an increasingly recognized role in modulating inflammation and behaviour, it is essential to characterize each stressor's impact on the gut microbiota. The present study aims to compare the effects of two ecologically relevant stress paradigms, CUS and 7-day SD on NLRP3 inflammasome activation, their impact on the gut microbiome and behavioural outcomes related to anxiety and depression. We hypothesize that the 28-day CUS paradigm will induce stronger NLRP3 inflammasome activation, greater microbial disruption, and increase anxiety- and depression-like behaviours compared to 14-day CUS exposure or SD, and that effects will vary in a sex-dependent manner. Methods and materials Animals One hundred (50 male and 50 female) CD-1 mice were obtained from Charles River Laboratories (St-Constant, Quebec, Canada) at three weeks of age. Mice were housed in groups of 2-3 in sex-specific housing rooms that were kept at constant temperature (24 ± 2 °C), relative humidity (40 ± 5%), and on a reversed light cycle (14h:10h light/dark) with lights off at 10 a.m. All mice were housed in polycarbonate Lexan housing cages (17 cm wide × 28 cm long × 12 cm high) that were bedded with Teklad Corn Cob bedding (Harlan Laboratories, Inc., Madison, WI, USA) and enriched with one square piece of Nestlet (Ancare Corp., Bellmore, NY, USA) and a cardboard refuge hut (Ketchum Manufacturing, Inc., Brockville, ON, Canada). Food (Harlan Laboratories, Inc., Madison, WI, US, T2018 – Global 18% rodent) and water were available ad libitum. All observational tests were completed during the dark phase under dim red light unless specified. Experimental procedures Each cage of mice was randomly assigned to either undergo 28 days of CUS, 14 days of CUS, 7 days of SD, or be left undisturbed. Two control groups were age-matched to their corresponding CUS conditions: a pubertal control group (matched to the 14-day CUS and 7-day SD groups) was euthanized during late puberty, while an adult control group (matched to the 28-day CUS group) reached adulthood by the time of euthanasia. This distinction was necessary because mice in the 28-day CUS group began stress exposure at four weeks of age and were therefore older at endpoint than those in the 14-day CUS group. Mice in both control groups were handled daily to match the interaction received by stressed animals but were not exposed to any CUS or SD stressors. Twenty-four hours following the completion of the stress paradigm, all mice underwent behavioural testing using the elevated plus maze (EPM), open field test (OFT), and forced swim test (FST). Twelve hours after the end of behavioural testing, mice were euthanized with Euthanyl (sodium pentobarbital; 120 mg/kg, intraperitoneal injection; Bimeda-MTC Animal Health Inc., Cambridge, Ontario, Canada). Brain, blood, and fecal samples were collected to assess the effects of CUS. All procedures were approved by the University of Ottawa’s Animal Care Committee and complied with the ARRIVE guidelines. Stress treatments CUS protocols Mice were exposed to CUS beginning at either four or five weeks of age to ensure that the six-week peak pubertal period was encompassed. The 14-day CUS group began stress exposure at five weeks of age, while the 28-day CUS group started at four weeks of age. Each day, mice were subjected to either a combination of two short stressors or one long stressor, alternating daily. The short stressors included in the CUS protocol were cage shaking (10 min at 150 rpm), cage tilt (2 h at 45°), tail pinch (1 min), empty cage (45 min), warm air flow (10 min), and restraint (45 min). The long stressors included water restriction (5 h), food restriction (5 h), wet bedding (10 h), and social isolation (10 h). The complete schedule of daily stressors is provided in Supplementary Table S1. 7-Day SD At six weeks of age, mice assigned to the SD condition were moved to a separate room at the beginning of the rest phase for seven consecutive days and were prevented from sleeping for the first 4 h. To maintain alertness and induce arousal in the mice, gentle stimulation was applied by brushing the nape and back with a paintbrush. This stimulation was given whenever a decrease in locomotion, drooping eyelids, and huddling was observed. After 4 h of manual sleep disruption, the mice were returned to their housing rooms and left undisturbed until the next day. The shorter duration of SD compared to CUS was chosen to avoid the severe physiological consequences associated with prolonged sleep disruption in rodents, which include weight loss and increased mortality risk (Sang et al., 2023). This procedure followed the same protocol previously implemented in our laboratory to induce acute sleep disruption (Murack et al., 2021). Behavioural tests of anxiety- and depression-like behaviour Elevated plus maze The EPM consisted of four branching arms: two open arms (66.5 cm × 6.5 cm) and two closed arms (66.5 cm × 6.5 cm x 16 cm). Each subject was placed in the hub of the apparatus which was elevated 82.5 cm from the floor. During the test, mice were placed in the centre facing one of the open arms and left to explore the maze for 5 min. The EthoVision XT tracking software (Version 18; Noldus Information Technology Inc., Leesburg, VA, USA) was used to measure the amount of time spent in each portion of the maze throughout the 5-min trial. An increased amount of time spent in the closed arms of the maze is linked to greater anxiety-like behaviour (Roy and Chapillon, 2004). Open field test The OFT procedure involved placing mice in an arena (90 cm × 90 cm) for 5 min and monitoring locomotion using EthoVision XT 18 tracking software. The software defined a centre/inner zone (30 cm × 30 cm) and a peripheral/outer zone. An increased amount of time spent in the peripheral zone is linked to greater anxiety (Prut and Belzung, 2003). Forced swim test The FST involved placing mice individually in a 4 L glass cylinder (15 cm × 30 cm [diameter x height]) containing 3 L of water at 24 °C ± 2 °C for 5 min. Behaviour was recorded using the Panasonic WV-CP284 camera (Kadoma, Japan). Videos were then scored manually by two raters blind to treatment conditions using ODLog software (Version 2.7.2; Macropod Software, Yarraville, Australia). The total time spent climbing (vertical movements along the wall of the cylinder), swimming (active displacement), and immobile (absence of movement or floating) was extracted from ODLog and compared between raters, yielding inter-rater reliability within 25%. The results were calculated as the average of the two raters’ scores. Increased time spent immobile was an indicator of behavioural despair and depression (Yankelevitch-Yahav et al., 2015). After testing, mice were gently dried with a towel and placed in recovery cages (30 °C) for 5 min before being returned to their home cage. Plasma extraction Twelve hours following the end of behavioural testing, mice were anesthetized with Euthanyl (sodium pentobarbital; 120 mg/kg, intraperitoneal injection; Bimeda-MTC Animal Health Inc., Cambridge, Ontario, Canada) and assessed for motor reflexes by gently pinching their feet. Once no motor reflexes were detected, trunk blood was collected by decapitation and placed into Microvette CB 300 K2E capillary blood collection tubes (Sarstedt AG & Co. KG, Cat # 16.444.100; Nümbrecht, Germany) coated with the anticoagulant ethylenediaminetetraacetic acid. Tubes were inverted several times to ensure proper mixing with the anticoagulant and kept at 4 °C until processing. Within 3 h of collection, plasma was separated by centrifugation (1000 × g for 15 min at 20 °C) and stored in aliquots at −80 °C. Cytokine measurement Plasma concentrations of IL-17A, IL-22, and IL-1β were measured with a multiplex bead-based Luminex immunoassay using commercially available kits (Millipore-Sigma, Cat # MTH17MAG-47K; Burlington, MA, USA) according to the manufacturer’s instructions. Thirty microlitres (30 μL) of each sample was run in duplicate on the same plate to minimize intra-assay variability. Acceptable intra-assay coefficients of variation (CVs) were set at <15% in accordance with manufacturer guidelines. Each plate contained one pooled sample to monitor inter-assay variation. Cytokine concentrations were calculated from median fluorescence intensity (MFI) values using a 5-parameter logistic (5-PL) curve in xPONENT software. The MAGPIX system (DiaSorin, Saluggia, Italy) was used to measure final cytokine concentrations. For IL-17A, the lower limit of detection (LLOD) was determined from the standard curve and background readings in xPONENT as 28.9 pg/mL; values below this threshold were replaced with the LLOD for statistical analyses. Samples with concentrations outside the standard curve range were extrapolated by the software and retained in analyses. Fecal Sample extraction and genomic DNA extraction Fecal samples were collected at the time of euthanasia to examine the effects of the stress treatment and sex differences on gut microbiota. Two to three fecal pellets were extracted from the descending colon and placed in a 1.5 mL microcentrifuge tube (PROGENE, Cat. #87-B150-C; Ultident Scientific, Saint-Laurent, Québec, Canada), then immediately stored at −80 °C until DNA extraction. A FastDNA spin kit (MP Biomedicals, Cat # 116540600; Solon, Ohio, USA) was used to extract DNA. Fecal samples were homogenized with two mechanical lysis cycles in a cell lysis solution and lysing matrix using a Mixer Miller 400 at 25 Hz for 40 s with 5 min cooling on ice between the two cycles. Negative (no fecal matter) and positive (ZymoBIOMICS Microbial Community Standard, Zymo Research, Cat #D6300; Irvine, California, USA) controls were included for extraction and subsequent PCR reactions. DNA was isolated and purified with the FastDNA Kit (MP Biomedicals, Cat # 116540600; Solon, Ohio, USA) according to the manufacturer’s protocol. DNA extraction was assessed with the Qubit Fluorometer (Invitrogen, Cat #Q33238; Carlsbad, California, USA) using the Qubit HS DNA Assay Kit (Invitrogen, Cat #Q33230; Carlsbad, California, USA), in which 1 μL of sample was mixed with 199 μL of assay buffer and 1 μL of dye working solution. Samples were normalized to 20 ng/μL, and aliquots were prepared and stored at −80 °C until 16S rRNA amplicon sequencing. 16S rRNA amplicon sequencing of fecal DNA Mouse fecal gDNA samples were processed for amplicon sequencing of the 16S rRNA V3-V4 regions. Fifty ng of each extracted gDNA sample was amplified using 1X KAPA HiFi HotStart ReadyMix (Roche, Cat # KK2802; Basel, Switzerland) and 200 nM primers (forward 5′-CCTACGGGNGGCWGCAG-3′ and reverse 5′-GACTACHVGGGTATCTAATCC-3′) in a 25 μL reaction volume (Klindworth et al., 2013). Cycling conditions for the amplicon PCR were as follows: initial denaturation at 95 °C for 3 min followed by 25 cycles of 95 °C for 30 s, primer annealing at 55 °C for 30 s and primer extension at 72 °C for 30 s before a 5-min final extension at 72 °C. Commercially available mock communities (Cedarlane, Cat # MSA-1002; Burlington, Ontario, Canada; 20 strain even mix; genomic material) were used as sequencing controls and water as no template amplification control. PCR products were visualized on a 2% agarose precast E-Gel stained with SYBR Safe DNA gel stain (Invitrogen, Cat #G72080; Carlsbad, California, USA). Amplicons were purified with Agencourt AMPure XP beads (Beckman Coulter, Cat # A63881; Brea, California, USA) following Illumina 16S Metagenomic sequencing library preparation protocol. A second round of amplification using 5 μL of the first PCR reaction as template, 2.5 μL each of Nextera XT V2 primers sets C and D (Illumina, Cat # FC-131-2003 and FC-131-2004; San Diego, California, USA) and 1X KAPA HiFi ReadyMix was performed in 25 μL reactions with the same cycling conditions as the previous PCR except that only 8 cycles were performed. PCR reactions were again purified with AMPure beads before individual fluorescent quantification by Quant-iT PicoGreen dsDNA assay (Life Technologies, Cat #P7589; Carlsbad, California, USA). Volumes corresponding to 200 ng of each purified Index PCR reaction were pooled using an EpMotion 5075 liquid handling robot (Eppendorf), and 5 μL of this pool was quantified in triplicate with Qubit Broad Range assay (ThermoScientific, Cat #Q32853; Waltham, Massachusetts) following the manufacturer’s instructions. Pooled libraries were quality-controlled by running a High Sensitivity D1000 TapeStation assay (Agilent, Cat # 5067-5584/5585; Santa Clara, California, USA). Final library was denatured in 0.2N NaOH following Illumina’s protocol, sequenced at 8 pM with 10% PhiX (Illumina, Cat # FC-110-3001; San Diego, California, USA) in a MiSeq Reagent 600 cycles v3 Kit cartridge (Illumina, Cat # MS-102-3003; San Diego, California, USA) and processed for 2x 300 cycles paired-end sequencing on an Illumina MiSeq instrument. Brain tissue extraction Following blood collection, brains were extracted and flash-frozen in liquid nitrogen and stored at −80 °C until processing. The brain tissue was sliced with a LEICA CM1950 cryostat at 300 μm and carefully mounted onto microscope slides. This preparation allowed for precise tissue punching of specific brain regions in subsequent analyses. Tissue from the PFC, HIPP, AMY, and PFC were collected using a 1 mm diameter biopsy punch following the guidelines from The Mouse Brain Atlas in Stereotaxic Coordinates (Paxinos and Franklin, 2019). Tissue punches were placed into RNA-free tubes and stored at −80 °C until RNA extraction. However, due to a high proportion of compromised AMY samples during tissue punching, this region was excluded from subsequent RT-qPCR analyses. mRNA extraction and cDNA synthesis Total RNA was extracted from HIPP and PFC tissue using the PureLink RNA Mini Kit (Thermo-Fisher Scientific, Cat. #12183020; Waltham, Massachusetts, USA) according to the manufacturer’s instructions. Extracted RNA was incubated with genomic DNA (gDNA) wipeout buffer to remove contaminating genomic DNA prior to cDNA synthesis. gDNA contamination was further controlled for at multiple levels. All qPCR primers were designed to span exon–exon junctions to minimize amplification of residual gDNA. In addition, no reverse transcription (–RT) controls were included to verify the efficiency of genomic DNA removal. Specifically, one negative RT control was randomly selected from every 24 samples and assessed by qPCR; no amplification was detected, confirming the absence of detectable gDNA contamination. Complementary DNA (cDNA) was subsequently synthesized using the QuantiTect Reverse Transcription Kit (QIAGEN, Cat. #205311; Hilden, Germany), which includes an integrated genomic DNA removal step before reverse transcription. cDNA was synthesized with the QuantiTect Reverse Transcription kit (QIAGEN, Cat # 205311; Hilden, Germany). The products of the cDNA synthesis step were used in subsequent real-time quantitative PCR. Real-time quantitative polymerase chain reaction (RT-qPCR) Relative gene expression was assessed using the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, Cat # 1725274; Hercules, California, USA) in triplicate of 10 μL reactions on the Bio-Rad CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Primer efficiency was determined using the slope between RNA quantity and cycle thresholds with CFX Maestro software (Bio-Rad). All primer pairs achieved reaction efficiency between 90 and 110%. β-actin and GAPDH were used as a housekeeping gene for all samples and did not change significantly across experimental conditions. For each reaction, the quantitative threshold amplification cycle number (Cq) was determined, and log transformed as previously described (Taylor et al., 2019). Primers were ordered from Integrated DNA Technologies (IDT), and sequences are displayed in Table 1. All primers were diluted to a final concentration of 0.3-0.5 μM for the real-time PCR reaction. Relative gene expression was calculated using Bio-Rad’s Ultimate qPCR Guide workflow up to Step 8. Briefly, Cq values for each target were normalized to the geometric mean of reference genes (β-actin and GAPDH) and converted to relative quantity (RQ) using the 2−ΔCq2 method. The resulting normalized fold-change values were used for statistical analyses and plotting. No log transformation was applied to the values presented in the figures. Table 1Primer used for gene expression analyses.Table 1Target GeneForwardSequenceReverseSequenceAnnealing Temperatureβ-actinGAACCCTAAGGCCAACCGTGGGTACGACCAGAGGCATACAGG67 °CGAPDHTCAACAGCAACTCCCACTCTCCACCACCCTGTTGCTGTA62 °CCASP1CTCGTACACGTCTTGCCCTCCCTCTTTCACCATCTCCAGAGC62 °CNLRP3TTATTTGTACCCAAGGCTGCTATCGGGCTTAGGTCCACACAGAAAG60 °CNFκB1GCTATGTGTGTGAAGGCCCATGCAAATTTTGACCTGTGGGT62 °CNFκB2AATGTGAATGCCCGGACCTTCTTCAGAGTCGGAGTCGCTC64 °C Primer used for gene expression analyses. Quantitative western blot analysis Total protein from the dissected PFC, HIPP, and AMY was extracted using radioimmunoprecipitation assay (RIPA) buffer (50 mM sodium chloride, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulphate (SDS), 50 mM Tris-HCl, pH 8), containing the protease inhibitors PhosSTOP™ (MilliporeSigma, Cat # 04001, Burlington, Massachusetts, USA) and cOmplete™ ULTRA Tablets, EDTA-free (MilliporeSigma, Cat # 05001; Burlington, Massachusetts, USA). The brain homogenates were incubated on ice for 10 min, then centrifuged at 4 °C at 19,000 xg for 20 min. The supernatants were collected to assay total protein concentrations using the Pierce™ Bicinchoninic Acid Assay (BCA) protein assay kit (Thermo Fisher Scientific, Cat # 23227; Waltham, Massachusetts, USA). A total of 10 μg of protein was mixed with a loading buffer containing 1% 2-mercaptoethanol (BioShop Canada Inc., Cat # MER002.100; Burlington, Ontario, Canada). Samples were electrophoresed on a 10% polyacrylamide gel using the Mini-PROTEAN® 3 Dodeca™ Cell system (Bio-Rad, Cat # 165-4100; Hercules, California, USA). A pooled protein sample was loaded onto each gel to account for inter-gel variation. The separated proteins were then transferred onto 0.2 μm nitrocellulose membranes using the Trans-Blot® Electrophoretic Transfer Cell system (Bio-Rad, Cat # 170-3910; Hercules, California, USA). Post transfer, membranes were dried for 15 min at 37 °C and then rehydrated using phosphate-buffered saline (PBS). The membranes were stained with Revert™ 700 Total Protein Stain (Licor, Cat # 926-11021; Lincoln, Nebraska, USA) for 5 min, washed twice in water containing 6.7% acetic acid and 30% methanol, and then imaged using Odyssey® software (version 3.0; LI-COR Biosciences, Lincoln, Nebraska, USA). These images were used for protein loading normalization instead of running a housekeeping protein (Eaton et al., 2013; Fosang and Colbran, 2015). The nitrocellulose membranes containing brain samples were then blocked for 1 h at room temperature in a solution containing 5% skim milk in PBS. Membranes were then incubated overnight (17-18 h at 4 °C) in either 5% milk in TBST (20 mM Tris base, 137 mM NaCl, 0.1% Tween 20) or 3% bovine serum albumin (BSA)/TBST, depending on the primary antibody used. Primary antibodies were diluted in the blocking solution and included: rabbit anti-NFκB (1/1000; Cell Signalling, Cat # 8242, Danvers, Massachusetts, USA), rabbit anti-Caspase 1 (1/2000; Proteintech, Cat # 22915–1-AP, Rosemont, Illinois, USA), and rabbit anti-NLRP3 (1/1000; Invitrogen, Cat # MA5-32255; Carlsbad, California, USA). Following primary antibody incubation, all membranes underwent 3 × 10-min washes with TBST and then incubated for 1 h at room temperature with goat anti-rabbit IRDye® 800CW secondary antibodies (LiCor Biosciences, Cat # 926-32211; Lincoln, Nebraska, USA) in a 5% skim milk/TBST solution. Following another set of 3 × 10-min washes with TBST, membranes were imaged, and protein bands intensity were quantified using Li-Cor Biosciences Odyssey® software. The intensity was normalized using the loaded protein measurement from the gel image. Finally, to control for inter-gel variation, the ratio of the normalized protein intensity between each sample and their intra-gel pooled sample was calculated and presented as a fold change (mean of duplicate ± SEM), which was used for relative protein abundance comparison between groups. Statistical analyses Statistical analyses were completed using IBM® SPSS® Statistics (version 29.0.1.0). Cases that exceeded the 1.5 interquartile range following boxplot analyses (behaviour tests, multiplex, RT-qPCR data) were considered statistical outliers and were winsorized to the next outer-most score within the 1.5 interquartile range (Hastings et al., 1947). For all measures, a 2 x 5 analysis of variance (ANOVA) was performed to examine the effects of sex and stress treatment, and their interaction on the dependent variables. When appropriate, statistically significant interaction or main effects were followed by pairwise comparisons with Bonferroni corrections. Measures of effect sizes were estimated using partial eta-squared (ηp2). Statistical significance was set to p < 0.05. In the absence of significant interaction effects, data were collapsed across the non-significant factor for graphical clarity. Significant interaction effects were visualized by stratifying both variables within the same figure. To analyze the data from the 16S rRNA sequencing, machine learning classifiers were implemented in QIIME 2 to assess whether gut microbiota composition could reliably distinguish between experimental groups. Due to structural constraints of the multivariate models and to ensure appropriate developmental comparisons, microbial analyses were conducted using age-matched groupings (pubertal controls vs. 14-day CUS and 7-day SD; adult controls vs. 28-day CUS) rather than pooling all five conditions into a single model. Classification accuracy was calculated and compared to a no-information rate (chance = 0.50) to evaluate group-level separability. Taxa contributing most strongly to classification performance were visualized using pirate plots to illustrate shifts in relative abundance across conditions. Results Anxiety- and depression-like behaviour Elevated plus maze In the EPM, both sex and stress paradigm influenced time spent in the open arms, with significant main effects of sex, F(1, 90) = 19.38, p < 0.001, η2 = 0.18, and stress, F(4, 90) = 2.52, p = 0.046, η2 = 0.10. Pairwise comparisons indicated that mice exposed to sleep disruption spent significantly less time in the open arms compared to those exposed to 14-day CUS (MD = −25.20, SE = 8.41, p = 0.035; Fig. 1A). In addition, females spent more time in the open arms than males, regardless of stress condition (MD = 23.42, SE = 5.32, p < 0.001; Fig. 1B). Effects of stress paradigm and sex on anxiety- and depression-like behavioural Measures (A) Mice exposed to 7-day SD spent less time in the open arms than those subjected to 14-day CUS. (B) In the EPM, females spent more time in the open arms than males, regardless of stress intervention. (C) In the OFT, 14-day CUS reduced center exploration compared with adult control and 28-day CUS groups, while sleep disruption decreased center time relative to adult controls. (D) In the FST, 7-day SD increased immobility compared with pubertal and adult controls, and 14 and 28-day CUS mice. Bars represent mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate significant pairwise differences (Bonferroni-corrected). Open field test Centre exploration in the OFT was also significantly affected by stress (F(4, 90) = 8.67, p < 0.001, η2 = 0.28). Post hoc analyses revealed that adult controls spent more time in the centre than the 14-day CUS (MD = 31.78, SE = 6.07, p < 0.001) and 7-day sleep disruption groups (MD = 24.99, SE = 6.07, p < 0.001), while 28-day CUS mice spent more time in the centre than 14-day CUS (MD = 22.30, SE = 6.07, p = 0.003; Fig. 1C). Forced swim test Stress significantly influenced immobility duration in the FST (F(4, 88) = 6.06, p < 0.001, η2 = 0.22). Mice exposed to sleep disruption displayed greater immobility compared to pubertal controls (MD = 38.51, SE = 11.09, p = 0.008), adult controls (MD = 40.76, SE = 10.80, p = 0.003), 14-day CUS (MD = 47.81, SE = 10.80, p < 0.001), and 28-day CUS (MD = 37.30, SE = 10.80, p = 0.009 (Fig. 1D)). Peripheral inflammatory cytokine levels IL-22 Plasma IL-22 concentrations were modulated by stress (F(4, 86) = 4.97, p = 0.001, η2 = 0.19). Post hoc tests demonstrated that pubertal controls had significantly lower IL-22 than mice exposed to 14-day CUS (p = 0.002), 28-day CUS (p = 0.031), and 7-day SD (p = 0.007; Fig. 2A). Plasma concentrations of IL-22 and IL-17A across stress conditions and sex (A) IL22 concentrations were significantly increased following 14-day CUS, 28-day CUS, and 7-day SD compared to pubertal controls. (B) IL17A concentrations were significantly elevated following 28-day CUS compared to pubertal controls, adult controls, and 7-day SD. (C) Females exhibited elevated levels of IL17A relative to males. Bars represent mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate significant pairwise differences (Bonferroni-corrected). IL-17A For IL-17A, significant main effects of both stress and sex were observed (F(4, 83) = 6.82, p < 0.001, η2 = 0.25, and F(1, 83) = 6.68, p = 0.011, η2 = 0.08, respectively). Pairwise comparisons revealed that IL-17A concentrations were higher in the 28-day CUS group compared to pubertal controls (p < 0.001), adult controls (p = 0.039), and 7-day SD (p < 0.001; Fig. 2B). Regardless of stress conditions, females exhibited significantly higher IL-17A concentrations than males overall (p = 0.011; Fig. 2C). Regarding IL-17A measurement, several samples yielded concentrations below the assay’s LLOD (28.9 pg/mL) and were replaced with the LLOD, rather than raw readings at or below the detection threshold, for statistical analyses. Samples with concentrations outside the standard curve range were extrapolated by the Luminex software and included in analyses. IL-1β All IL-1β values were below the assay’s lower limit of detection (LLOD) and were extrapolated by the Luminex software and, therefore, excluded from statistical analyses. Fecal microbiota composition Microbiome-level changes following stress were evaluated using classification analyses in QIIME 2. Models distinguished treatment groups from controls with varying accuracy across stress paradigms and sexes (Table 2), with violin plots of relative abundances shown in Supplementary Figs. S1–6. Weighted UniFrac-based Principal Coordinate Analyses (PCoA) demonstrated treatment-related clustering of fecal microbiota (Fig. 3), with separate plots provided for each comparison: 7-day SD vs. pubertal control, 14-day CUS vs. pubertal control, and 28-day CUS vs. adult control. Whereas the PCoA plots capture overall group-level differences, the classification analyses revealed sex-specific microbial shifts, described in the following subsections. Table 2Differentially abundant taxa in chronic stress and sleep disruption comparisons by sex.Table 2ComparisonSexClassification AccuracyTaxa ↓ in Stress Group (Enriched in Controls)Function/RelevanceTaxa ↑ in Stress GroupFunction/RelevanceComp 1:7-Day SD vs. Pubertal ControlFemale0.82Tyzzerella, Ligilactobacillus, MediterraneibacterSCFA production; barrier integrity; microbial stabilityFlintibacter, Oscillospira,Mucispirillum, Eubacterium,BacteroidesDisrupted SCFA metabolism; barrier disruption; pro-inflammatory potentialMale0.82Akkermansia, Tyzzerella, Ligilactobacillus, ParaprevotellaSCFA production; barrier integrity; immune regulationFlintibacter, OscillospiraceaeDisrupted SCFA metabolism; dysbiosis; pro-inflammatory potentialComp 2:14-Day CUS vs. Pubertal ControlFemale0.63Tyzzerella, Phocaicola, Holdemania,Acetatifactor, LactobacillusSCFA production; immune regulation; barrier integrityFlintibacter, Oscillospira, Paludicola, Caproicipacter, Clostridia (class)Disrupted SCFA metabolism; barrier disruption; pro-inflammatory potentialMale0.70Caproicipacter, Acetatifactor, Oscillospira, Ligilactobacillus, Bacteroides, Acetatifactor,SCFA production; barrier integrity; microbial stabilityFlintibacter, Paludicola,Clostridia (class)Disrupted SCFA metabolism; pro-inflammatory potentialComp 3:28-Day CUS vs. Adult ControlFemale0.61Staphylococcus, Akkermansia, Christensenella, Longibaculum, Erysipelatoclostrid, Adlercreutzia,Ligilactobacillus,Barrier integrity; anti-inflammatory potential; microbial stabilityKineothrix, Clostridium,Dysbiosis; disrupted SCFA metabolism; pro-inflammatory potentialMale0.58Christensenella, Acetatifactor, Tyzzerella,StaphylococcusSCFA production; microbial stability; GBA regulationNone—Note. Functional roles of taxa derived from published literature (e.g., Derrien et al., 2017; López-García et al., 2022; Smith and Vale, 2006). Full references are listed in the reference section. Differentially abundant taxa in chronic stress and sleep disruption comparisons by sex. Note. Functional roles of taxa derived from published literature (e.g., Derrien et al., 2017; López-García et al., 2022; Smith and Vale, 2006). Full references are listed in the reference section. Principal coordinate analysis of fecal microbiota composition A) 7-day SD vs. pubertal control, (B) 14-day CUS vs. pubertal control, (C) 28-day CUS vs. adult control. Each point represents the fecal microbiota of an individual mouse, coloured by treatment group. Ellipses indicate 95% confidence intervals around group centroids. Clustering reflects treatment-related differences in microbial community structure (weighted UniFrac). Comparison 1: 7-day SD vs. pubertal control In this comparison, classification accuracy was high in both females and males (0.82). Among females, the control group showed greater relative abundance of Tyzzerella, Paraprevotella, Ligilactobacillus, and Mediterraneibacter, whereas Flintibacter, Oscillospira, Mucispirillum, Eubacterium, and Bacteroides were enriched following SD. In males, controls were distinguished by higher levels of Akkermansia, Tyzzerella, Ligilactobacillus, and Paraprevotella, while Flintibacter and Oscillospiraceae were more abundant in the SD group. Comparison 2: 14-day CUS vs. pubertal control For the 14-day CUS vs. pubertal control comparison, classification accuracy was moderate (females: 0.63; males: 0.70). Female controls had a higher abundance of Tyzzerella, Phocaicola, Holdemania, Acetatifactor, and Lactobacillus, while Flintibacter, Oscillospira, Paludicola, Caproicipacter, and Clostridia (class) were increased in the CUS group. Male mice exposed to CUS displayed greater levels of Flintibacter, Paludicola, and Clostridia (class), whereas controls showed enrichment of Caproicipacter, Acetatifactor, Oscillospira, Ligilactobacillus, Bacteroides, and Erysipelatoclostrid. Comparison 3: 28-day CUS vs. adult control By contrast, the 28-day CUS vs. adult control comparison yielded slightly lower classification accuracy (females: 0.61; males: 0.58). In females, controls were characterized by increased levels of Staphylococcus, Akkermansia, Christensenella, Longibaculum, Erysipelatoclostrid, Adlercreutzia, and Ligilactobacillus, while Kineothrix and Clostridium were elevated under CUS. In males, Christensenella, Acetatifactor, Tyzzerella, and Staphylococcus were more abundant in controls, with no genera significantly enriched in the CUS group. Across all comparisons, stress exposure was consistently associated with reduced abundance of short-chain fatty acid (SCFA)-producing and barrier integrity-supporting taxa, including Ligilactobacillus, Acetatifactor, Akkermansia, and Christensenella. In contrast, stress groups showed enrichment of taxa such as Flintibacter, Clostridia, and Oscillospiraceae, which were recurrently detected across conditions. These shifts were observed in both sexes, although specific taxa varied by stressor type and duration. Neuroinflammatory gene expression Prefrontal cortex Analyses revealed distinct patterns of stress- and sex-related differences in prefrontal cortex gene expression. NLRP3 expression differed significantly by sex and stress, with a sex × stress interaction (F(4, 40) = 4.78, p = 0.003, η2 = 0.32). Male mice exposed to 7-day sleep disruption exhibited significantly higher NLRP3 expression compared to their female counterparts (p < 0.001), pubertal controls (p < 0.001), adult controls (p < 0.001), 14-day CUS (p < 0.008), and 28-day CUS (p = 0.002; (Fig. 4A). Sex- and stress-dependent alterations in inflammasome gene expression in the prefrontal cortex (A) In males, 7-day SD induced significantly higher NLRP3 expression compared to all other male groups and their female counterparts, whereas no stress-related differences were observed among females. (B) CASP1 expression was elevated in males relative to females following 14-day CUS and 7-day SD. Within males, 7-day SD expression was higher than both pubertal and adult control groups, and 28-day CUS. (C) NFκB1 expression was higher in males than females under 28-day CUS and 7-day SD, and females in the adult control group had greater expression than females in the 28-day CUS group. (D) NFκB2 expression was significant higher in males compared to females under 14-day CUS and 7-day SD. Males in the 7-day SD had greater expression than males in both pubertal and adult controls. Bars represent mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate significant pairwise differences (Bonferroni-corrected). Expression values are presented as normalized fold change calculated as 2−ΔCq2 and normalized to the geometric mean of reference genes (β-actin and GAPDH; Bio-Rad Step 8 workflow). For CASP1, analysis of variance indicated a significant sex × stress interaction (F(4, 40) = 4.55, p = 0.004, η2 = 0.31). Post hoc tests showed that males exposed to 14-day CUS and 7-day SD exhibited significantly higher CASP1 expression relative to females within the same groups. Further, males subjected to sleep disruption had increased expression compared to male pubertal controls (p = 0.006), adult controls (p = 0.008), and 28-day CUS (p = 0.022; Fig. 4B). A significant sex × stress interaction also emerged for NFκB1 mRNA expression (F(4, 40) = 3.67, p = 0.012, η2 = 0.27). Follow-up comparisons revealed that male mice in the 28-day CUS and 7-day SD groups showed significantly higher levels than their female counterparts (p = 0.009 and p = 0.013, respectively). Additionally, within the female mice, those in the adult control group exhibited significantly higher NFκB1 expression than those in the 28-day CUS group (p = 0.008; (Fig. 4C). Finally, NFκB2 mRNA expression showed a sex × stress interaction (F(4, 40) = 3.94, p = 0.009, η2 = 0.28), where males exhibited significantly higher NFκB2 levels than their female counterparts in the 14-day CUS (p = 0.041) and 7-day SD groups (p < 0.001). Moreover, sleep disrupted males displayed increased NFκB2 in the PFC relative to pubertal (p = 0.017) and adult controls (p = 0.027; Fig. 4D). Hippocampus Within the hippocampus, NLRP3 mRNA expression showed a significant sex × stress interaction (F(4, 40) = 3.11, p = 0.025, η2 = 0.24). Post hoc comparisons revealed that male mice in the 28 day-CUS had higher NLRP3 than females in the same group (p = 0.002). Among females, mice in the 14-day CUS and 7-day SD groups displayed significantly higher NLRP3 expression than mice in the 28-day CUS group (p = 0.011 and p < 0.001, respectively; Fig. 5A). Sex- and stress-dependent differences in inflammasome gene expression in the hippocampus (A) NLRP3 expression was significantly lower in females than their male 28-day CUS counterparts, as well as females in the 14-day CUS and 7-day SD groups. (B) CASP1 expression was significantly higher in adult control than the 28-day CUS group. (C) Females in the 28-day CUS group showed significantly decreased NFκB1 expression than female pubertal and adult controls, 14-day CUS, 7-day SD, and their male counterparts. (D) NFκB2 expression did not significantly differ across stress groups or between sexes. Bars represent mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001 indicate significant pairwise differences (Bonferroni-corrected). Expression values are presented as normalized fold change calculated as 2−ΔCq2 and normalized to the geometric mean of reference genes (β-actin and GAPDH; Bio-Rad Step 8 workflow). For CASP1, significant stress effects were evident (F(4, 40) = 2.78, p = 0.040, η2 = 0.22). Follow-up tests indicated that adult controls had significantly higher CASP1 expression compared to 28-day CUS (p = 0.032; Fig. 5B). For NFκB1 expression in the hippocampus, the sex × stress interaction was significant (F(4, 40) = 15.29, p < 0.001, η2 = 0.605). Male mice in the 28-day CUS displayed elevated NFκB1 levels relative to the females in their respective group (p < 0.001). Post hoc analyses further showed that female mice exposed to 28-day CUS expressed significantly lower NFκB1 than female pubertal controls (p = 0.050), adult controls (p = 0.023), 14-day CUS (p < 0.001), and 7-day SD (p < 0.001; Fig. 5C). By contrast, NFκB2 expression was unaffected by sex, stress, or their interaction, with no significant effects observed (all p > 0.05; Fig. 5D). Western blot Prefrontal cortex No significant main effects or interactions were observed for NLRP3, CASP1, or NFκB expression in the prefrontal cortex (Fig. 6A–C; p > 0.05). Protein abundance of inflammasome-related markers in the prefrontal cortex (A–C) NLRP3, CASP1, and NFκB protein abundance in the prefrontal cortex did not significantly differ across stress conditions or between sexes. Bars represent mean ± SEM. Hippocampus No significant main effects or interactions were observed for NLRP3 or NFκB protein abundance (Fig. 7A–C). In contrast, a significant sex × stress interaction was present for CASP1 protein abundance in the hippocampus (F(4, 38) = 3.29, p = 0.021, η2 = 0.257). Follow-up comparisons revealed that female mice in the adult control group showed significantly higher CASP1 levels than their male counterparts (p = 0.005). Further, among female mice, those in the adult control group had significantly higher expression than the pubertal controls (p = 0.042; Fig. 7B). Protein abundance of inflammasome-related markers in the prefrontal cortex (A–C) Hippocampal NLRP3 and NFκB protein abundance did not significantly differ across stress conditions or between sexes (A, C), whereas CASP1 protein abundance exhibited a significant sex × stress interaction, such that adult control females displayed higher CASP1 levels than adult control males and pubertal control females (B). Bars represent mean ± SEM. Amygdala No significant main effects or interactions were observed for NLRP3, CASP1, or NFκB expression in the amygdala (Fig. 8A–C; p > 0.05). Protein abundance of inflammasome-related markers in the amygdala (A–C) NLRP3, CASP1, and NFκB protein abundance in the amygdala did not significantly differ across stress conditions or between sexes. Bars represent mean ± SEM. Discussion This study investigated how distinct chronic stress paradigms influence NLRP3 activation and gut–brain axis function during puberty, revealing stress- and sex-specific effects across behavioural, immune, and microbial domains. Sleep disruption produced the most pronounced depressive-like phenotype and reduced exploratory behaviour, whereas 14-day CUS heightened anxiety-like responses that appeared to attenuate following 28 days of exposure. Prolonged CUS selectively elevated IL-17A, while IL-22 increased across all stress conditions, suggesting engagement of both pro-inflammatory and regulatory immune pathways. In the brain, inflammasome-related transcription varied by region and sex: sleep disruption upregulated NLRP3 and CASP1 in the male prefrontal cortex, whereas 28-day CUS was associated with coordinated reductions in NF-κB and inflammasome-related transcripts in females, particularly in the hippocampus. These molecular changes occurred alongside stress-associated shifts in gut microbial composition, including depletion of short-chain fatty acid–producing, barrier-supporting genera and enrichment of inflammation-linked taxa, with distinct patterns depending on stressor type and sex. Together, these findings demonstrate that the nature and duration of stress exposure critically shape outcomes of the gut–brain axis during puberty. Behaviourally, exploratory activity in anxiety-related contexts was influenced by both sex and stress exposure. In the EPM, females spent more time in the open arms than males. Sex differences in open-arm exploration are frequently reported in rodents, with females often displaying greater exploratory behaviour in this task, particularly during adolescence and young adulthood (Palanza, 2001). These differences are generally interpreted as reflecting sex-specific patterns of exploration and risk assessment rather than overall differences in stress susceptibility. Sleep disruption further reduced open-arm exploration compared to 14-day CUS, and in the FST, it produced markedly greater immobility across sexes, indicating a robust depressive-like phenotype. These findings suggest that SD exerts more substantial anxiogenic and depressogenic effects than moderate-duration CUS. Notably, contrary to our hypothesis, SD was not less effective than 28-day CUS; rather, it produced even greater behavioural alterations. This heightened impact may reflect the unique effects of sleep disruption on circadian regulation, HPA axis activity, and neuroimmune signalling (Balbo et al., 2010), which may drive a distinct and dysregulated stress response relative to the CUS models. In the OFT, stress exposure significantly altered centre-zone behaviour. Adult controls spent more time in the centre than both 14-day CUS and 7-day SD mice, consistent with increased anxiety-like behaviour following shorter stress exposures. In contrast, 28-day CUS mice spent more time in the centre than 14-day CUS mice and did not differ from adult controls, suggesting attenuation of the anxiogenic effect with prolonged stress exposure. This pattern contradicts our initial hypothesis that extended CUS would exacerbate anxiety-like behaviour. Instead, it points to non-linear stress effects, whereby anxiety-like responses are more pronounced following shorter exposures (14-day CUS, 7-day SD) but diminish with longer durations. Such dynamics may reflect habituation, compensatory adaptation, or stress-induced shifts in exploratory states (Cameron and Schoenfeld, 2018). However, because the 14-day and 28-day CUS groups were assessed at different developmental endpoints, differences between these groups cannot be attributed solely to stress duration and may also reflect developmental maturation. At the peripheral level, cytokine profiles showed distinct stress- and sex-dependent patterns. Plasma IL-17A concentrations were significantly elevated following 28 days of CUS, irrespective of sex. In addition, females exhibited overall higher IL-17A levels than males across conditions. Together, these findings align with evidence of enhanced pro-inflammatory responsivity following prolonged stress exposure and female-biased immune activation (Bekhbat and Neigh, 2018; Goel et al., 2014). In contrast, IL-22 was elevated across all stress conditions relative to pubertal controls, suggesting that even brief or moderate stress exposure is sufficient to upregulate this cytokine. Given its dual role in mucosal immunity and epithelial repair, IL-22 elevation may reflect an early immunoregulatory response to stress-induced intestinal barrier disruption (Guo et al., 2022; Tarassishin et al., 2011). Notably, IL-22 is often induced downstream of Th17-related signalling yet can exert barrier-protective effects, suggesting that its elevation represents a compensatory response to stress-related inflammatory activation. Together, these peripheral cytokine patterns suggest that prolonged stress preferentially enhances pro-inflammatory IL-17A signalling, whereas both acute and chronic stressors engage IL-22–mediated regulatory pathways. These divergent immune signatures may parallel the region- and sex-specific neuroimmune alterations observed in the brain. Unlike the peripheral cytokines, central inflammasome-related transcription varied by sex and stressor, indicating region-specific regulation. In the PFC, males exposed to 7-day SD showed robust increases in NLRP3 and CASP1, indicating that short-term sleep loss strongly triggers inflammasome activation in males. In the PFC, females exposed to 28-day CUS showed reduced NFκB1 expression compared to adult controls, suggesting that prolonged stress dampens canonical NF-κB signalling in a sex-specific manner. I. In contrast, NFκB2, a component of the non-canonical NF-κB pathway (Shih et al., 2011), exhibited a sex × stress interaction, with males showing higher expression than females following 14-day CUS and 7-day sleep disruption. Additionally, NFκB2 expression was elevated in sleep-disrupted males relative to both pubertal and adult controls, suggesting enhanced engagement of non-canonical NF-κB signalling under specific stress conditions. In the hippocampus, a distinct regulatory pattern emerged. NLRP3 was lower in 28-day CUS females compared with 14-day CUS and 7-day SD, suggesting that prolonged unpredictable stress suppresses inflammasome priming in the female hippocampus, in contrast to elevations with shorter CUS or sleep loss. CASP1 was decreased in mice in the 28-day CUS groups compared with adult controls, regardless of sex, reinforcing a CUS-associated reduction in downstream inflammasome signalling. Notably, NFκB1 was lowest in 28-day CUS females relative to all other groups, including adult controls, highlighting a female-specific suppression of canonical NF-κB transcription in this region. NFκB2 showed no hippocampal effects. Our results demonstrating female-specific reductions in NLRP3/CASP1/NFκB1 following 28-day CUS diverge from the typical pro-inflammatory increases reported. Previous work has shown that hippocampal NLRP3, CASP1, and NF-κB are often elevated following chronic unpredictable mild stress (CUMS) or sleep loss (e.g., Du et al., 2016; Sahin et al., 2016) with some studies reporting enduring increases even after 28 days of CUMS (Wang et al., 2024). In contrast, our findings highlight a female-specific suppression of these pathways at 28-day CUS, suggesting that chronic stress may lead to either persistent activation or compensatory down-regulation depending on stress duration, sex, and experimental context. Although most protein-level analyses did not reveal significant stress-related effects, a significant sex × stress interaction was observed for hippocampal CASP1 abundance. Specifically, adult control females exhibited higher CASP1 levels than both adult control males and pubertal control females, indicating a developmental and sex-dependent difference independent of stress exposure. In contrast, no significant main effects or interactions were detected for NLRP3 or NF-κB protein abundance. Thus, while RT-qPCR identified several transcriptional alterations, these changes did not consistently translate into corresponding protein-level differences following stress exposure. Several factors may explain this partial dissociation between mRNA and protein. Protein abundance is inherently more variable than transcript levels and reflects not only transcription but also translational regulation, post-translational modification, and protein degradation, which may obscure subtle effects (Liu et al., 2016). In addition, Western blotting measures total protein across heterogeneous cell populations within a brain region, potentially masking changes confined to specific cell types such as microglia. Prior work demonstrates that correlations between mRNA and protein expression are often modest (Vogel and Marcotte, 2012), making such discrepancies unsurprising. Future studies incorporating more sensitive and cell-specific approaches, such as immunohistochemistry to localize expression or assays targeting activated protein forms (e.g., cleaved CASP1 p20 or NF-κB nuclear translocation), may better capture functional inflammasome activity than measures of total protein abundance. These neuroimmune patterns occurred alongside shifts in the gut microbiome, another key component of stress-related physiology. Across comparisons, stress was generally associated with depletion of SCFA-producing and barrier integrity-supporting genera (e.g., Ligilactobacillus, Acetatifactor, Akkermansia, Christensenella) and enrichment of taxa previously linked to inflammation and dysbiosis (e.g., Flintibacter, Oscillospiraceae, Clostridia). These changes were robustly detected in both males and females, but with sex-dependent patterns in the taxa most affected. Under 7-day SD, females showed reductions in Ligilactobacillus, while males displayed decreases in both Akkermansia and Ligilactobacillus. These taxa are strongly implicated in mucosal barrier maintenance and SCFA production, suggesting that acute sleep disruption compromises barrier-supporting microbes. In the 14-day CUS comparison, females exhibited enrichment of stress-associated genera such as Flintibacter and Oscillospira alongside reductions in Lactobacillus and Acetatifactor, while males showed a loss of Ligilactobacillus and Bacteroides. By contrast, 28-day CUS was associated with subtler microbial differences. Female controls were enriched with Akkermansia and Christensenella, and male controls showed higher levels of Christensenella and Acetatifactor compared to stressed counterparts. Collectively, the observed microbial changes imply that stress exposure may weaken gut barrier integrity and SCFA-mediated signalling, providing a mechanistic link to systemic inflammation and neuroimmune activation (Parada Venegas et al., 2019; Siddiqui and Cresci, 2021). Notably, taxa such as Ligilactobacillus, Acetatifactor, and Christensenella are central to SCFA production and epithelial barrier maintenance, and their depletion under SD and CUS suggests compromised barrier function. The role of Akkermansia is more complex, as it is often associated with improved metabolic health (Gao et al., 2025) However, elevated levels have also been reported in neurodegenerative and inflammatory conditions (Ma et al., 2025), suggesting that its influence on neuroimmune outcomes depends on the physiological context. Thus, sex- and stressor-specific reductions in SCFA producers in combination with the rise of stress-associated taxa such as Flintibacter and Oscillospira, may compromise the GBA and facilitate neuroimmune dysregulation under chronic stress. Together, these findings demonstrate that both stress paradigm and biological sex are critical determinants of behavioural, neuroimmune, and microbiome responses during puberty in mice. The differential impact of CUS and SD underscores the multifaceted nature of stress-related pathophysiology (Bekhbat and Neigh, 2018; Jašarević et al., 2016; McCormick and Green, 2013). Specifically, CUS predominantly elicited anxiety-like behaviours and upregulated neuroimmune gene expression, particularly in females, suggesting a heightened vulnerability of the female brain to prolonged exposure to stress. In contrast, SD exerted more profound effects on depressive-like behaviours and induced pronounced disruptions in gut microbial composition, with distinct sex-dependent patterns. These findings suggest that acute physiological stressors, such as sleep disruption, may more directly perturb gut homeostasis and systemic inflammation pathways, potentially contributing to the development of mood-related disorders through the gut-brain axis. Moreover, the observed sex-dependent responses align with existing literature, indicating that males and females differ in their neuroimmune and microbiome reactivity to stress, likely driven by interactions between sex hormones, immune function, and the gut microbiome composition during critical developmental windows (Bekhbat and Neigh, 2018; Goel et al., 2014; Jašarević et al., 2016; Freimer et al., 2022). These differences emphasize that stress susceptibility is not uniform across sexes or stress paradigms, but rather, shaped by a complex interplay of biological and environmental factors. Collectively, these stressor-specific and sex-dependent effects highlight the necessity of selecting appropriate and contextually relevant models when probing distinct dimensions of stress vulnerability. Relying on a single stress model overlooks the distinct biological pathways activated by different stressors and may obscure critical sex-specific vulnerabilities. By incorporating both sex and stressor type as central variables, future research can better elucidate the mechanisms underlying individual differences in stress-related disorders and inform the development of more targeted and effective interventions. Limitations and future directions Several limitations of the current study should be acknowledged. First, all molecular and microbiome analyses were conducted at a single post-stress time point, which restricts our ability to capture the temporal dynamics of stress-induced neuroimmune and microbial responses. As a result, we cannot conclude how these outcomes may change over time following stress exposure. Second, while behavioural, molecular, and microbiome outcomes were assessed, this study did not explore whether individual variations in behaviour corresponded to changes in neuroinflammation or gut microbiota composition. Without such correlational analyses, the direct relationship between stress-induced behavioural phenotypes and underlying biological alterations remains unclear. Third, although gene expression analyses indicated the activation of neuroimmune pathways, microglial activation states were not directly assessed. Therefore, we cannot confirm whether the observed changes in gene expression reflect microglial polarization or the involvement of other cell types. Another limitation is that stress duration and developmental stage cannot be fully disentangled in the current design, as the different stress paradigms resulted in assessments at distinct developmental endpoints. Although age-matched control groups were included to account for developmental differences, comparisons across stress durations should be interpreted cautiously, as both stress exposure and developmental maturation may have contributed to the observed outcomes. Lastly, circulating levels of gonadal hormones were not measured, leaving uncertain whether the observed sex differences were directly influenced by hormonal fluctuations during puberty. Future studies should investigate stress-induced neuroimmune and microbiome changes across multiple time points to better understand the trajectory of these responses. Correlating individual behavioural phenotypes with neuroimmune markers and microbial composition would further illuminate the biological mechanisms driving stress-related behaviours. Additionally, assessing microglial activation through cell-specific markers and quantifying circulating sex hormones would provide valuable insights into the cellular and endocrine factors underlying the observed sex differences. Conclusion This study demonstrates that both stress paradigm and biological sex are critical determinants of behavioural, neuroimmune, and microbiome responses during puberty. By revealing distinct profiles of pathology under CUS and SD conditions, our findings highlight the value of sex-informed approaches to modelling stress in adolescence. These results contribute to a growing literature recognizing puberty as a sensitive window of vulnerability, shaped by immune and microbial mechanisms, and suggest that targeting neuroimmune and GBA pathways may represent a promising avenue for mitigating stress-related outcomes. CRediT authorship contribution statement Michaela Dworsky-Fried: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. Umar Haris Iqbal: Conceptualization, Methodology, Writing – review & editing. Stéphane Bronner: Conceptualization, Methodology, Resources, Writing – review & editing. Jérémie Auger: Formal analysis, Visualization, Writing – review & editing. Olivier Mathieu: Data curation, Formal analysis, Writing – review & editing. Noémie Auclair-Ouellet: Writing – review & editing. Jacky Liang: Methodology, Project administration. Nafissa Ismail: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – review & editing. Data availability Data will be made available on request. Funding This work was supported by a Discovery grant from the Natural Sciences and Engineering Research Council of Canada (2020-04302) to NI and by a Mitacs Accelerate Grant in partnership with Lallemand Health Solutions. Declaration of competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments The authors would like to thank all the members of the NISE Lab and the ACVS staff at the University of Ottawa for their assistance with this project. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbih.2026.101341. Supplementary data The following is the Supplementary data to this article. Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.