ABSTRACT

Bipolar disorder (BD) is associated with mood dysregulation and neurobiological abnormalities such as oxidative stress, neuroinflammation, and neurodegeneration. While physical exercise shows promise in mental health, the mechanistic effects of strength training in BD remain poorly understood. This study aimed to investigate the impact of an 8‐week strength training protocol on behavioral, oxidative, and cellular alterations in a validated rat model of BD induced by ouabain. Adult male Wistar rats were randomly assigned to sedentary or exercised groups. After the training period, animals underwent surgery for cannula implantation. Following recovery, they received either ouabain or artificial cerebrospinal fluid. Behavioral assessments were conducted during the manic‐ (Day 7) and depressive‐like (Day 14) phases, and tissue samples were collected on Day 18 post‐injection. Neurochemical assays and immunohistochemical analyses were performed in the cerebral cortex and hippocampus. Ouabain induced manic‐ and depressive‐like behaviors, cognitive impairments along with oxidative imbalance, increased NF‐κB activation, astrogliosis, and neuronal degeneration. Notably, prior strength training prevented these behavioral disturbances and significantly attenuated oxidative stress, neuroinflammation, and cell death. Physical exercise normalized antioxidant enzyme activities, reduced reactive species accumulation, prevented NF‐κB activation, and decreased GFAP and Fluoro Jade‐C labeling. Correlation analyses revealed significant associations among oxidative stress, inflammation, neurodegeneration, and cognitive impairment. These findings demonstrate, for the first time, that structured strength training exerts neuroprotective effects in a BD model by modulating redox homeostasis, inflammatory signaling, and neuronal integrity. Strength training emerges as a promising, low‐cost, and mechanistically grounded adjunctive strategy in BD management.

Abstract

Schematic representation of the neuroprotective effects of strength training in an ouabain‐induced model of bipolar disorder. The intracerebroventricular injection of ouabain induces behavioral alterations (manic‐ and depressive‐like behaviors and cognitive deficits), increases oxidative stress (↑reactive species, ↓antioxidant defenses), promotes astrogliosis, and leads to neuronal death. In contrast, strength training attenuates these pathological changes, preventing behavioral deficits, restoring redox balance, reducing glial reactivity, and preserving neuronal integrity. Red elements represent ouabain‐induced effects, while green elements indicate effects of exercise. OUA, ouabain.

AbbreviationsaCSFartificial cerebrospinal fluidAKTprotein kinase BANOVAanalysis of varianceAPanteroposteriorATPadenosine triphosphateBCAbicinchoninic acidBDbipolar disorderBSAbovine serum albuminBWbody weightCA1/CA3cornu ammonis 1/cornu ammonis 3 (hippocampal subregions)CATcatalaseCDNB1‐chloro‐2,4‐dinitrobenzeneCOX‐2cyclooxygenase‐2CREBcAMP response element‐binding proteinDAPI4′,6‐diamidino‐2‐phenylindoleDCF/DCFHdichlorofluorescein/dichlorodihydrofluoresceinDGdentate gyrusDNAdeoxyribonucleic acidDTNB5,5′‐dithiobis(2‐nitrobenzoic acid)ERKextracellular signal‐regulated kinaseFJCfluoro‐jade CGFAPglial fibrillary acidic proteinGPxglutathione peroxidaseGRglutathione reductaseGSHreduced glutathioneGSSGoxidized glutathioneGSTglutathione S‐transferaseHNE/4‐HNE4‐Hydrynonenali.c.v.intracerebroventriculariNOSinducible nitric oxide synthasei.p.intraperitonealJjoulesLLlaterolateralLPtAlateral parietal association cortexMPtAmedial parietal association cortexmTORmechanistic target of rapamycinNADPHnicotinamide adenine dinucleotide phosphateNF‐κBnuclear factor kappa BNMDAN‐methyl‐D‐aspartateNORnovel object recognitionNP‐SHnon‐protein thiolsOFTopen field testOUAouabainp70s6Kribosomal protein S6 kinase 1PFAparaformaldehydePI3Kphosphoinositide 3‐kinasep‐NF‐κBphosphorylated nuclear factor kappa BRIPAradioimmunoprecipitation assay bufferROSreactive oxygen speciesRSDretrosplenial dysgranular cortexRSGcretrosplenial granular cortexSEMstandard error of the meanSODsuperoxide dismutaseVventral

1 Introduction

Bipolar disorder (BD) is a chronic psychiatric condition characterized by recurrent episodes of mania and depression, affecting approximately 1%–2% of the global population (Grande et al. 2016). Despite the availability of pharmacological and psychotherapeutic treatments, a substantial proportion of patients fail to achieve full remission, experience frequent relapses, or suffer from treatment‐resistant symptoms (Shuy et al. 2024). These challenges highlight the need to better understand the neurobiological basics of BD and to explore complementary or novel therapeutic strategies that can enhance clinical outcomes. Animal models, such as the ouabain‐induced model of BD, which replicate key behavioral and neurochemical features of the disorder, are valuable tools for investigating underlying mechanisms and testing potential interventions (Valvassori et al. 2019).

In addition to its profound psychological and social impact, several studies have implicated mitochondrial dysfunction and the overproduction of reactive oxygen species (ROS) in the pathophysiology of BD (Kageyama et al. 2025). These oxidative processes can damage critical cellular components, including DNA, proteins, and membrane lipids, ultimately compromising neuronal function and viability and indicating a state of persistent oxidative stress (Arat‐Çelik et al. 2024). Moreover, mitochondrial abnormalities—such as impaired respiratory chain activity, altered mitochondrial DNA, and reduced ATP production—have been consistently observed in BD patients, particularly in brain regions involved in mood regulation (Bodenstein et al. 2019; Gonzalez 2021). This mitochondrial dysfunction not only exacerbates oxidative stress but also affects calcium homeostasis and apoptotic signaling, increasing neuronal vulnerability (Baev et al. 2022). Postmortem and neuroimaging studies further support the presence of structural and functional brain alterations in BD, including reduced gray matter volume and impaired connectivity, which may be partially attributed to cumulative oxidative and metabolic damage (Cyrino et al. 2021; Ganzola and Duchesne 2017; Jones et al. 2025).

Physical exercise has emerged as a promising non‐pharmacological strategy for improving mental health across a range of neuropsychiatric disorders, including depression, anxiety, and schizophrenia (Firth et al. 2016; Harvey et al. 2018; McDowell et al. 2018). It is known to exert beneficial effects on mood regulation, cognitive performance, neuroplasticity, and stress resilience, in part by modulating inflammatory pathways, oxidative stress, and neurotrophic signaling (Ben Ezzdine et al. 2025; Radak et al. 2016). In the context of BD, however, the literature remains limited, with most studies emphasizing aerobic exercise or general physical activity levels, while structured exercise interventions—particularly strength training—have been still poorly explored (Dos Santos et al. 2025), with a lack of mechanistic insights. Strength exercise holds unique therapeutic potential due to its effects on molecular pathways associated with neuronal survival and cognitive function (Radak et al. 2016). In this regard, a recent study by (Maidana et al. 2025) was the first to demonstrate that a structured strength training attenuated BD‐like behavioral alterations and significantly reduced cognitive impairments in an animal model of BD. These effects were accompanied by the activation of key neuroplasticity‐related signaling pathways, including PI3K/AKT/mTOR/p70S6K and ERK/CREB, suggesting a mechanistic basis for the observed improvements. These findings offer new insights into the neurobiological effects of strength training and support its investigation as a potential adjunctive strategy in the management of BD (Maidana et al. 2025).

Considering the above, no previous study has evaluated the effect of strength training on oxidative stress and neuronal cell death in brain regions critically affected in BD. Therefore, this study aimed to investigate the effects of a structured strength training protocol on behavioral outcomes, oxidative damage, antioxidant defense, and neuronal viability in rats subjected to the ouabain‐induced model of BD.

2 Experimental Procedures

2.1 Animals

Adult male Wistar rats (n = 32; RRID:RRRC_00697), aged 60 days, were randomly assigned to four experimental groups and housed in groups of three per cage under standardized environmental conditions (temperature: 22°C; relative humidity: 45%–55%) with a 12‐h light/dark cycle. Food and water were provided ad libitum. The animals were obtained from the Animal Facility of the Federal University of Pampa (UNIPAMPA). All experimental procedures strictly adhered to the guidelines outlined in the Guide for the Care and Use of Laboratory Animals (8th edition, National Institutes of Health) and received prior approval from the institutional ethics committee (protocol no. 016/2023). Every effort was made to minimize both the number of animals used and their potential distress.

2.2 Experimental Design

Animals were allocated into four experimental groups: Sedentary + vehicle; Exercise + vehicle; Sedentary + OUA; and Exercise + OUA. A total of 32 animals (n = 8 per group) were included in the study. All animals completed the full experimental protocol and were allocated to the experimental groups by simple alternation to ensure balanced group sizes. The experimental timeline is illustrated in Figure 1. On the final day of the 8‐week ladder‐based strength training protocol, all animals underwent a progressive workload test to confirm that the physical training had elicited adaptations consistent with strength training, specifically an enhanced capacity to sustain increased overload. Stereotaxic surgery was performed on the first day following the conclusion of the strength training protocol. From the first to the sixth day post‐training, animals in the “Exercise” groups continued climbing sessions on the ladder apparatus without additional overload to prevent detraining effects. On Day 1 of the post‐training, animals received an intracerebroventricular (i.c.v.) injection of ouabain (OUA). Behavioral assessments were conducted on Days 7 and 14 following OUA administration, corresponding respectively to the expected manifestation of manic‐like and depressive‐like behaviors. Recognition memory test was performed on Days 16 and 17, respectively. Following the conclusion of the behavioral evaluations, animals were euthanized by decapitation, and the cerebral cortex and hippocampi were rapidly dissected and collected for subsequent neurochemical analyses. All animals underwent behavioral testing and were subsequently perfused, with no exclusions at any time point. No pre‐established inclusion or exclusion criteria were defined, and no animals died or were excluded during the experiments. No formal test for outliers was conducted, and no data points were excluded from the analyses. Due to tissue availability, cost considerations, and prior experimental evidence indicating sufficient effect sizes, randomly selected subsets of samples were used for specific analyses: four animals per group for immunohistochemistry and six animals per group for biochemical determinations. Sample selection was performed by simple alternation and conducted by an experimenter blinded to treatment and experimental groups.

2.2.1 Sample Size Determination

No a priori sample size calculation was performed. Sample sizes were determined based on pilot experiments and previously published studies using adult male Wistar rats and similar experimental designs and outcome measures (Maidana et al. 2025; Souza‐Pereira et al. 2025). Based on this evidence, group sizes ranging from 4 to 8 animals per group were selected. The adequacy of these sample sizes is supported by the consistent detection of statistically significant effects across multiple cortical and hippocampal regions using two‐way ANOVA, indicating that biologically meaningful differences were detectable with the chosen group sizes.

2.3 Strength Training Protocol—Vertical Ladder

During the first 3 days of the training protocol, animals underwent a familiarization phase on an 80° inclined vertical ladder apparatus (110 cm × 10 cm) equipped with a black housing chamber (10 cm height × 10 cm width × 15 cm depth) positioned at the top, as described by (Hornberger Jr. and Farrar 2004). Familiarization consisted of three consecutive climbing trials per day over 3 days, performed without any additional load. Following this phase, animals were subjected to the strength training protocol for the remaining duration, completing 8 weeks of training at a frequency of five sessions per week. Each training session comprised five sets of five climbs, with a 1‐min rest interval in the black chamber between sets. On the first day of training with workload, an initial load equivalent to 25% of the animal's body weight (BW) was affixed to the proximal region of the tail using adhesive tape. If the animal completed all five sets without failure, the load was increased by 5% BW on the following day. In cases of failure, the same load was maintained for the subsequent session. Progressive overload adjustments were made weekly, with increments of 15% BW, culminating in a maximum of 130% BW by the final week of training. To demonstrate strength gain resulting from our protocol, animals were subjected to a progressive overload test on the final day of training. The test consisted of five sets of five climbs in the black chamber, with progressively increasing overloads of 0%, 50%, 75%, 100%, and 150% of the animal's body weight, and 1‐min rest intervals between sets. Total work performed was calculated and expressed in Joules (J).

2.4 Surgical Procedure and OUA Administration

Animals were anesthetized with ketamine (75 mg/kg, i.p.), xylazine (10 mg/kg, i.p.), and acepromazine (2.5 mg/kg, i.p.), and then positioned in a stereotaxic apparatus for rodents. The scalp was shaved, and the surgical field was disinfected with chlorhexidine. A 2 cm longitudinal incision was made along the midline of the scalp, and a 0.8 mm diameter hole was drilled into the right parietal bone to accommodate the implantation of a 9 mm guide cannula. The cannula was positioned 1 mm above the right lateral ventricle using the following stereotaxic coordinates relative to bregma (in mm): anteroposterior (AP), 0.9; laterolateral (LL), 1.5; and ventral (V), 2.6, by Paxinos et al. (2006). The cannula was secured in place with dental acrylic cement, and the incision was sutured, leaving the upper end of the cannula externally accessible. Immediately after cannula implantation surgery, animals received an injection of dipyrone (200 mg/kg), diluted in 1 mL of a 50% dextrose solution (i.p.). This was administered to provide postoperative analgesia and to prevent hypoglycemia resulting from fasting and surgical stress. Following surgery, animals were monitored twice daily for 48 h by a veterinarian. Assessments included general condition (activity level, posture, and coat appearance), food and water intake, surgical site integrity, and specific signs of pain or distress. The body weight was also monitored daily. No additional doses of analgesics were required, as animals did not show overtly signs of pain or distress during the observation period. After a 4‐day recovery period, animals received a single i.c.v. injection of OUA (cat. no. O3125) at a concentration of 10−3 M (5 μL) diluted in artificial cerebrospinal fluid (aCSF), or 5 μL of aCSF alone. A 10 mm infusion needle was inserted into the guide cannula and connected via polyethylene tubing to a Hamilton microsyringe. The infusion needle extended 1.0 mm beyond the tip of the guide cannula to reach the right lateral ventricle (El‐Mallakh et al. 1995). The infusion was administered over 4 min to minimize solution reflux.

2.5 Behavior Tests

2.5.1 Open Field

Exploratory behavior was assessed using the open field test on the 7th and 14th days following OUA administration, as previously described (El‐Mallakh et al. 1995; Valvassori et al. 2019, 201). Rats were individually placed in a wooden open field arena (50 cm × 50 cm × 50 cm), with the floor divided into 16 equal squares. Each animal was gently positioned at the center of the arena and allowed to explore for 5 min on each testing day. All sessions were video‐recorded using a fixed overhead camera, and locomotor behavior was quantified by manually counting the total number of square crossings by three evaluators blinded to the experimental groups; the quantifications were then averaged, and this mean value was used as the representative data point for statistical analysis.

2.5.2 Elevated Plus Maze

To assess anxiety‐like or risk‐taking behavior (Kukula et al. 2022), the elevated plus‐maze test was employed. The apparatus consisted of two open arms and two closed arms arranged at right angles to each other, forming a plus‐shaped configuration. All arms measured 50 cm × 10 cm, with the closed arms enclosed by 40 cm high walls. The arms converged at a central platform measuring 10 cm × 10 cm. The entire structure was elevated 50 cm above the floor. On the 7th and 14th days following OUA administration, each animal was gently placed in the central zone of the maze and allowed to freely explore for 5 min, and the total time spent in open arms was quantified by three evaluators blinded to the experimental groups; the quantifications were then averaged, and this mean value was used as the representative data point for statistical analysis.

2.5.3 Novel Object Recognition

To evaluate potential cognitive impairments in recognition memory associated with OUA‐induced BD (Valvassori et al. 2022), the object recognition test was conducted in the same arena used for the open field assessment. The objects employed for discrimination were figures of similar size and texture (8–10 cm in height), affixed to the floor. On the 16th day following OUA administration, animals were placed in the arena containing two identical objects (A1 and A2) and allowed to explore freely for 10 min. Short‐term memory was assessed 1 h after this initial exposure: animals were returned to the same arena for 5 min, where one of the familiar objects was replaced with a novel one (A and B). Long‐term memory was evaluated 24 h after the first trial, in a similar 5‐min session where one of the original objects was substituted with a different novel object (A and C). Exploration behavior was tracked and quantified using ToxTrac software, which recorded the cumulative time each animal spent within the defined zones surrounding the objects. Recognition memory was inferred from the proportion of time spent exploring the novel object, expressed as the Discrimination Index (DI), calculated using the following formula: (DI) = [time spent exploring novel object/(time spent exploring both objects)] × 100 (Lueptow 2017).

2.6 Biochemical Assays

2.6.1 Sample Processing

After the experimental protocol, animals were anesthetized intraperitoneally with ketamine (100 mg/kg) and xylazine (10 mg/kg), followed by transcardial perfusion with 0.9% NaCl solution (600 mL) containing heparin (5 IU/mL). The left side of the brain was immediately rinsed with cold aCSF, extracted from the skull, and placed in a Petri dish containing cold aCSF, positioned over ice, with filter paper lining the dish to facilitate tissue stabilization and dissection. Samples were immediately flash‐frozen in liquid nitrogen and stored at −80°C until further processing. The frozen samples were weighed and homogenized in Tris–HCl buffer (50 mM, pH 7.4). The homogenate was centrifuged at 10.000×g for 10 min at 4°C, and the resulting supernatant (S1 fraction) was used for the analysis of oxidative stress parameters. Protein content was determined using the colorimetric Bradford method, with bovine serum albumin (BSA; 1 mg/mL) as the standard (Bradford 1976). The right side was processed as described below for immunofluorescence assay.

2.6.2 Reactive Species Quantification

To estimate reactive species production, the S1 fraction was diluted 1:10 in 50 mM Tris–HCl buffer (pH 7.4) and incubated with 10 μL of 2′,7′‐dichlorodihydrofluorescein diacetate (DCFH‐DA; 1 mM) for 60 min, at 37°C in the absence of light. Fluorescence resulting from DCFH‐DA oxidation was measured as an indicator of oxidative stress following the methodology (Loetchutinat et al. 2005), using an Agilent Cary Eclipse Fluorescence Spectrometer (Santa Clara, CA, USA) with excitation at 488 nm, emission at 525 nm, and beam setting 2.5. The dichlorofluorescein diacetate (DCF) fluorescence intensity was expressed as a percentage relative to the control group.

2.6.3 Non‐Protein Thiol Content

The evaluation of non‐protein thiol (NPSH) content was performed as previously described in the literature (Ellman 1959). The supernatant was mixed with 10% trichloroacetic acid and centrifuged at 10,000×g for 5 min at 4°C. The protein pellet was discarded, and the free thiol groups in the clear supernatant were analyzed. An aliquot of the supernatant was added to 1 M potassium phosphate buffer (pH 7.4) and 5,5′‐dithiobis‐(2‐nitrobenzoic acid) (DTNB). The colorimetric reaction was measured at 412 nm using a spectrophotometer. Results were expressed as a percentage relative to the control group.

2.6.4 Superoxide Dismutase Activity

Superoxide dismutase (SOD) activity relies on the enzyme's ability to inhibit the autoxidation of epinephrine to adrenochrome mediated by the superoxide anion (O₂−·), with concomitant formation of hydrogen peroxide (H₂O₂) (Misra and Fridovich 1972). The reaction was initiated by adding 6–18 μL of sample (S1) to 30 μL of 6 mM epinephrine in a reaction mixture containing 242–260 μL of 57.7 mM sodium carbonate buffer, maintained at 30°C. Absorbance was recorded at 480 nm every 10 s over 2 min using a spectrophotometer. One unit of SOD activity was defined as the amount of enzyme required to inhibit epinephrine autoxidation by 50% under the assay conditions. Enzymatic activity was normalized to total protein content and expressed as units per milligram of protein (U/mg protein).

2.6.5 Catalase Activity

Catalase (CAT) activity was determined by quantifying the enzymatic decomposition of hydrogen peroxide (H₂O₂) by monitoring the decline in absorbance at 240 nm (Aebi 1984). The reaction was initiated by adding 20 μL of the sample (S1) to 105 μL of 0.3 M H₂O₂ in a reaction mixture containing 100 μL of 50 mM phosphate buffer (pH 7.0) at 25°C. Enzyme activity was expressed in units (U), where one unit is defined as the amount of enzyme that decomposes 1 μmol of H₂O₂ per minute under the assay conditions.

2.6.6 Glutathione Peroxidase Activity

Glutathione peroxidase (GPx) activity method quantifies the reduction of hydrogen peroxide (H₂O₂) via oxidation of reduced glutathione (GSH) to oxidized glutathione (GSSG) (Wendel 1981). The activity was measured indirectly by monitoring the consumption of reduced nicotinamide adenine dinucleotide phosphate (NADPH) at 340 nm, as NADPH is oxidized to NADP+ during the regeneration of GSH from GSSG by glutathione reductase (GR). The reaction mixture consisted of 800 μL of reductase system—containing 0.15 mM NADPH, 1 mM GSH, 10 μL of GR (500 U), and 250 μL of 100 mM sodium azide—dissolved in 20 mL of 100 mM TFK buffer (pH 7.0). To this, 50 μL of sample (S1) and 100 μL of 4 mM H₂O₂ were added to initiate the reaction. Absorbance was recorded every 30 s over a 2‐min period at 340 nm. Enzymatic activity was expressed as nanomoles of NADPH consumed per minute per milligram of protein (nmol NADPH/min/mg protein).

2.6.7 Glutathione S‐Transferase Activity

Glutathione S‐transferase (GST) activity was measured based on the conjugation of reduced glutathione (GSH) with 1‐chloro‐2.4‐dinitrobenzene (CDNB), a model electrophilic substrate (Habig et al. 1976). GST plays a key role in phase II detoxification by catalyzing the conjugation of xenobiotic compounds with GSH. For the assay, 50 μL of sample (S1) was added to a reaction mixture containing 0.1 M potassium phosphate buffer (pH 7.5), 50 mM CDNB, and 100 mM GSH. The reaction was monitored by measuring the increase in absorbance at 340 nm every 30 s for 2 min using a spectrophotometer. Enzyme activity was expressed as micromoles of CDNB conjugated per minute per milligram of protein (μmol/min/mg protein).

2.6.8 Western Blotting

Western blot analysis was performed as described by (Funck et al. 2014), with minor modifications. S1 fraction of cerebral cortex and hippocampus were diluted on radioimmunoprecipitation assay (RIPA) buffer and centrifuged at 12.700 rpm for 20 min at 4°C. Protein concentrations were determined using the bicinchoninic acid (BCA) protein assay (Thermo Fisher Scientific; cat. no. 1859078), and all samples were adjusted to a final concentration of 2 μg/μL. Equal amounts of protein (30 μg per sample) were separated on a 9% SDS‐polyacrylamide gel (SDS‐PAGE) and transferred to nitrocellulose membranes using the Trans‐Blot Turbo Transfer System (Bio‐Rad; RRID: SCR_023156). Equal protein loading was confirmed with Ponceau S staining (Sigma‐Aldrich, P7170). Following transfer, membranes were blocked with 4% BSA in TBS‐T (Tris‐buffered saline with 0.1% Tween‐20) and washed twice for 10 min at room temperature. Membranes were then incubated overnight at 4°C with mouse antibody 4‐HNE (SAB5202472; Sigma Aldrich; 1:1000; RRID:AB_2801464), 3‐NT (Sc‐55 256; Santa Cruz Biotechnology; 1:1000; RRID:AB_1126663), NF‐κB (SC‐8008; Santa Cruz Biotechnology; 1:1000; RRID:AB_628017), p‐NF‐κB (SC‐136548; Santa Cruz Biotechnology; 1:1000; RRID:AB_10610391), Caspase 3 (total and cleaved; 8G10; Cell Signaling Technology: 1:1000; RRID:AB_2069872), NeuN (MAB377; Merck Millipore; 1:1000; RRID:AB_2298772) and GFAP (G9269; Sigma‐Aldrich; 1:1000; RRID:AB_477035). After primary antibody incubation, membranes were washed twice for 15 min in TBS‐T and incubated for 2 h at room temperature with anti‐mouse (sc‐2005; Santa Cruz Biotechnology; 1:10000; RRID:AB_631736) or anti‐rabbit (A6154; Sigma‐Aldrich; 1:10000; RRID:AB_258284) horseradish peroxidase‐conjugated secondary antibodies. Protein bands were visualized using enhanced chemiluminescence (ECL) substrate (Pierce ECL, Bio‐Rad; cat. no. 1705061), and signals were captured with the ChemiDoc XRS+ Imaging System (Bio‐Rad; cat. no. 12003153). Band intensities were quantified using Image Lab software (Bio‐Rad) and normalized to total protein using Ponceau S staining (Romero‐Calvo et al. 2010). Data are expressed as a percentage of the control group.

2.6.9 Immunofluorescence and Fluoro Jade C (FJC) Staining

The right side of the brain was extracted and post‐fixed in 4% paraformaldehyde (PFA) for 24 to 48 h. After immersion in 30% sucrose for 3 days, the tissues were rapidly frozen in liquid nitrogen and stored at −80°C. Coronal sections of 16 μm were obtained using a cryostat (CM2850, Lupetec), mounted on gelatin‐coated slides, and left to dry at room temperature for 2 h.

For FJC staining, slides were sequentially immersed in 0.05% sodium hydroxide in 80% ethanol for 5 min, 70% ethanol for 2 min, distilled water for 2 min, and 0.06% potassium permanganate for 10 min. Sections were then incubated for 1 h in a solution containing 0.01% FJC (MilliporeSigma; cat. no. AG325) and 0.01% DAPI (Santa Cruz Biotechnology; cat. no. sc‐3598) in 0.1% acetic acid. After staining, slides were rinsed with distilled water, cleared in xylene for 1 min, mounted with DPX, coverslipped, and left to dry at room temperature.

In the Glial Fibrillary Acidic Protein (GFAP) analysis, sections were delimited with a hydrophobic barrier and incubated in blocking solution (5% fetal bovine serum in 0.1 M PBS with 0.2% Triton X‐100) for 1 h. They were then incubated overnight at 4°C with the primary antibody rabbit anti‐GFAP (1:500; ab4648, Abcam; RRID:AB_449329). After PBS washes, the secondary antibody goat anti‐rabbit IgG‐HRP (1:250; sc‐2004, Santa Cruz Biotechnology; RRID:AB_631746) was applied for 2 h at room temperature in PBS‐Tx (0.2%). After final washes, sections were counterstained with DAPI, mounted with coverslips, and stored in the dark at 4°C until analysis.

Images from the retrosplenial granular cortex, c region (RSGc), retrosplenial dysgranular cortex (RSD), medial parietal association cortex (MPtA) and lateral parietal association cortex (LPtA) in cerebral cortex and cornu ammonis 1 (CA1), cornu ammonis 3 (CA3) and dentate gyrus (DG) in the hippocampus were acquired using a 20× objective lens on a Zeiss AX10 ScopemA1 microscope equipped with a Zeiss Axiocam 105 color camera. Acquisition was performed with Zen software (version 2.3, Zeiss), and image analysis was conducted using Fiji software (version 1.51J). The “Color Threshold” function was used to isolate and enhance the specific labeling. Quantification of GFAP staining was based on the percentage of astrocyte‐labeled areas per image and manual counting of GFAP‐positive cells, while FJC‐positive cells were counted manually.

2.6.10 Sholl Analysis

Sholl analysis was performed to quantitatively evaluate the morphological complexity of activated astrocytes (Zhang et al. 2024). GFAP‐positive astrocytes were manually reconstructed in ImageJ/Fiji by tracing each individual process to obtain a complete representation of their morphology. The Sholl Analysis plugin was then applied to these reconstructed cells.

The center of the cell body was manually defined as the reference point, and a series of concentric circles was generated starting at 3.5 μm from the center, with 1.5 μm increments, up to 50 μm. The software automatically quantified the number of intersections between astrocytic processes and each radius. From these outputs, we obtained: (i) the number of intersections at each radius; (ii) the total number of intersections; and (iii) the length of the longest astrocytic process. All parameters were kept constant across analyses to ensure methodological consistency.

In each region, five astrocytes were selected according to two criteria: (1) their morphology was not obscured by neighboring cells or blood vessels; and (2) as many of their processes as possible were fully contained within the tissue section. The measurements from the five cells within each region were then averaged, and this mean value was used as the representative data point for statistical analysis.

2.7 Statistical Analysis

Data were presented as box‐and‐whisker plots, in which the lower boundary of the box indicates the 25th percentile, the line within the box represents the median, and the upper boundary corresponds to the 75th percentile; whiskers depict the range of observed values. For the total work test, data were analyzed by repeated measures ANOVA and are represented as mean ± SEM. Normality of the data distribution was assessed using the Shapiro–Wilk test. Depending on the distribution characteristics, data were analyzed using a two‐way analysis of variance (ANOVA) for parametric data or the Scheirer–Ray–Hare test for non‐parametric data. For sholl analysis, data were analyzed using three‐way repeated measures ANOVA. When applicable, post hoc comparisons were conducted using Tukey's multiple comparisons test. Statistical significance was defined as p < 0.05. All analyses were performed using GraphPad Prism version 8.0 (GraphPad Software, San Diego, CA, USA).

3 Results

To evaluate the effect of strength training, both sedentary and exercised animals were subjected to an overload test at the end of the training period. Statistical analysis revealed that trained animals showed higher total work compared to sedentary animals from 75% to 150% of body weight overload (F(4, 88) = 181.3; p < 0.0001; Figure 1C).

3.1 Behavior Analysis

To evaluate the effects of ouabain administration and prior physical exercise on locomotor activity, anxiety‐like behavior, and recognition memory, animals were subjected to the open field and elevated plus maze tests at 7 and 14 days post‐infusion, and to the novel object recognition test on the 17th day after infusion (Figure 2). Two‐way ANOVA revealed that previous strength training prevents OUA‐induced increase in the number of crossings in the open field (F(1, 28) = 12.20; p = 0.0016; Figure 2B) and time spent in the open arms in the elevated plus maze test (F(1, 28) = 13.07; p = 0.0012; Figure 2E) on the seventh day after OUA injection. On the 14th day after OUA, physical exercise prevents the OUA‐induced decrease in time spent in the open arms of the elevated plus maze test (F(1, 28) = 5.471; p = 0.0267; Figure 2F). Besides, strength training protects against the OUA‐induced decrease in the discrimination index in both short‐ (F(1, 28) = 5.322; p = 0.0287; Figure 2H) and long‐term memory (F(1, 28) = 5.626; p = 0.0248; Figure 2I).

3.2 Oxidative Stress and Antioxidant Defenses

To investigate the effects of ouabain and the potential neuroprotective role of strength training on oxidative stress, redox status, and inflammation, we quantified key enzymatic and molecular markers in both the cerebral cortex and hippocampus (Figure 3). Statistical analysis showed that prior physical training prevents OUA‐induced increases in the activity of superoxide dismutase and glutathione‐S‐transferase in both the cerebral cortex (SOD: [F(1, 20) = 23.30; p = 0.0001; Figure 3A]; GST: [F(1, 20) = 17.34; p = 0.0005; Figure 3D]) and hippocampus (SOD: [F(1, 20) = 7.362; p = 0.0134; Figure 4A]; GST: [F(1, 20) = 3.280; p = 0.0852; Figure 4D]), as well as the OUA‐induced decrease in catalase activity in these same regions (cerebral cortex: [F(1, 20) = 8.357; p = 0.0090; Figure 3B]; hippocampus: [F(1, 20) = 9.336; p = 0.0062; Figure 4B]). In addition, previous strength training protects against the OUA‐induced elevation of DCFH and non‐protein thiol levels in the cerebral cortex (DCFH: [F(1, 20) = 7.871; p = 0.0109; Figure 3E]; NP‐SH: [F(1, 20) = 3.972; p = 0.0601; Figure 3F]) and hippocampus (DCFH: [F(1, 20) = 3.692; p = 0.0690; Figure 4E]; NP‐SH: [F(1, 20) = 5.669; p = 0.0273; Figure 4F]), and also prevents the increase in the p‐NF‐κB/NF‐κB total ratio in both structures (cerebral cortex: [F(1, 20) = 4.470; p = 0.0473; Figure 3I]; hippocampus: [F(1, 20) = 16.43; p = 0.0006; Figure 4I]). Strength training further prevents the OUA‐induced decrease in glutathione peroxidase activity in the cerebral cortex (F(1, 20) = 4.665; p = 0.0431; Figure 3C) and the increase in immunoreactivity of 3‐nitrotyrosine (F(1, 20) = 13.46; p = 0.0015; Figure 4G) and 4‐hydroxynonenal (F(1, 20) = 6.030; p = 0.0233; Figure 4H) in the hippocampus. Two‐way ANOVA revealed a main effect of OUA on glutathione peroxidase activity in the hippocampus (F(1, 18) = 9.666; p = 0.0061; Figure 4C).

3.3 Cell Viability Assays

To further investigate the neuroprotective effects of strength training, we evaluated astrocytic reactivity and neuronal degeneration in the cerebral cortex and hippocampus. Immunofluorescence for GFAP and Fluoro‐Jade C staining and western blotting analysis for Caspase‐3, NeuN and GFAP were used to assess gliosis (GFAP) and neurodegeneration (Fluoro‐Jade C, Caspase‐3 and NeuN). Two‐way ANOVA showed that previous strength training protects against OUA‐induced GFAP reactivity in the RSD (F(1, 12) = 20.36; p = 0.0007; Figure 5E), MPrA (F(1, 12) = 10.75; p = 0.0066; Figure 5H), and LPtA (F(1, 12) = 16.76; p = 0.0015; Figure 5K) regions of the cerebral cortex, as well as in the CA3 (F(1, 12) = 13.62; p = 0.0031; Figure 6E) and DG (F(1, 12) = 12.61; p = 0.0040; Figure 6H) regions of the hippocampus. In GFAP‐positive cells, statistical analysis shows a OUA‐induced increase in RSD (F(1, 12) = 33.63; p < 0.0001; Figure 5F) and CA3 (F(1, 12) = 43.76; p < 0.0001; Figure 6F) and strength training prevents OUA‐induced increase in RSGc (F(1, 12) = 9830; p = 0.0086; Figure 5C) and DG (F(1, 12) = 12.66; p = 0.0039; Figure 6I). Also, a main effect of OUA in CA1 (F(1, 12) = 9571; p = 0.0093; Figure 6C), and main effect of exercise in LPtA (F(1, 12) = 17.24; p = 0.0013; Figure 5L).

Astrocytic complexity in the cerebral cortex and hippocampus was assessed by Sholl analysis across increasing distances from the soma. Three‐way repeated‐measures ANOVA showed that previous strength training protects against OUA‐induced increase in the number of intersections at several distances from the soma in MPtA (F(31, 372) = 7570; p < 0.0001; Figure 7C), LPtA (F(31, 372) = 2388; p < 0.0001; Figure 7D), CA3 (F(31, 372) = 4431; p < 0.0001; Figure 8B) and DG (F(31, 372) = 5765; p < 0.0001; Figure 8C), total number of intersections (a) and branches length (b) in MPtA (a. F(1, 12) = 8579; p = 0.0126; Figure 7C1 ; b. F(1, 12) = 31.78; p = 0.0001; Figure 7C2 ), CA3 (a. F(1, 12) = 11.53; p = 0.0053; Figure 8B1 ; b. F(1, 12) = 9573; p = 0.0093; Figure 8B2 ) and DG (a. F(1, 12) = 6185; p = 0.0286; Figure 8C1 ; b. F(1, 12) = 14.01; p = 0.0028; Figure 8C2 ). Besides, statistic analysis shows a main effect of OUA in the number of intersections per radii in RSGc (F(1, 12) = 26.63; p = 0.0002; Figure 7A) and RSD (F(1, 12) = 9964; p = 0.0083; Figure 7B), total number of intersections in CA1 (F(1, 12) = 10.03; p = 0.0081; Figure 8A1 ) and branches length in LPtA (F(1, 12) = 5847; p = 0.0324; Figure 7D2 ).

Physical exercise also prevents the OUA‐induced increase in Fluoro‐Jade C‐positive cells in the MPtA (F(1, 12) = 7.799; p = 0.0163; Figure 9F) region of the cerebral cortex, and in the CA1 (F(1, 12) = 8.587; p = 0.0126; Figure 10B), CA3 (F(1, 12) = 17.07; p = 0.0014; Figure 10D), and DG (F(1, 12) = 27.93; p = 0.0002; Figure 10F) regions of the hippocampus. Statistical analysis revealed a main effect of OUA on Fluoro‐Jade C‐positive cells in the RSGc (F(1, 12) = 7.459; p = 0.0182; Figure 9B), RSD (F(1, 12) = 46.90; p < 0.0001; Figure 9D), and LPtA (F(1, 12) = 4.886; p = 0.0472; Figure 9H) regions of the cerebral cortex.

Finaly, on western blotting analysis, two‐way ANOVA shows that prior strength training prevents OUA‐induced increase in Cleaved Caspase‐3/Caspase‐3 total ratio (F(1, 19) = 15.09; p = 0.0010; Figure 11A) and GFAP immunoreactivity (F(1, 19) = 8396; p = 0.0092; Figure 11C), but no alters OUA‐induced decrease in NeuN (F(1, 19) = 5570; p = 0.0291; Figure 11B) immunoreactivity in cerebral cortex. In hippocampus, a single administration of OUA increased Caspase‐3/Caspase‐3 total ratio (F(1, 19) = 9312; p = 0.0066; Figure 11D) and GFAP immunoreactivity (F(1, 19) = 20.05; p = 0.0003; Figure 11F) and decreased NeuN immunoreactivity (F(1, 19) = 9332; p = 0.0065; Figure 11E).

3.4 Correlations

To explore potential relationships between biochemical, histological, and behavioral outcomes, Pearson's correlation analyses were performed (Figure 12). A positive correlation was observed between neurodegeneration (FJC) in the MPtA region of the cerebral cortex and oxidative stress levels (DCFH) in the cerebral cortex (r = 0.6232; p = 0.0099; Figure 12A), as well as between FJC in the DG region of the hippocampus and DCFH levels in the hippocampus (r = 0.8038; p = 0.0002; Figure 12B). Inflammatory signaling, as assessed by the p‐NFκB/NFκB ratio, was positively correlated with GFAP immunoreactivity in both the cerebral cortex (MPtA; r = 0.7004; p = 0.0025; Figure 12C) and hippocampus (DG; r = 0.5446; p = 0.0292; Figure 12D), supporting an association between astrogliosis and neuroinflammation. Moreover, p‐NFκB/NFκB ratio positively correlated with DCFH fluorescence in both the cerebral cortex (r = 0.7285; p = 0.0014; Figure 12E) and hippocampus (r = 0.8002; p = 0.0002; Figure 12F), reinforcing the link between inflammation and oxidative stress. Furthermore, higher levels of neuronal degeneration in the DG region were significantly associated with poorer cognitive performance. Specifically, FJC signal negatively correlated with short‐term memory (r = −0.7935; p = 0.0002; Figure 12G) and long‐term memory discrimination indices (R = −0.6335; p = 0.0084; Figure 12H) in novel object recognition test.

4 Discussion

The present study demonstrates, for the first time, that structured strength training prevents behavioral, neurochemical, and cellular alterations in a well‐validated rat model of BD. Our findings show that strength training not only attenuates manic‐ and depressive‐like behaviors and memory impairments but also exerts significant antioxidant and neuroprotective effects in brain regions critically involved in cognition and prominently affected in BD, namely the hippocampus and cerebral cortex.

In line with previous behavioral studies (Maidana et al. 2025), our results confirm that prior physical exercise mitigates the hyperlocomotion and risk‐taking behaviors typically observed in the early manic‐like phase following ouabain infusion (El‐Mallakh et al. 1995; Valvassori et al. 2019). Notably, we also observed a reversal of recognition memory deficits in both short‐ and long‐term phases, further supporting the hypothesis that strength training improves cognitive resilience in neuropsychiatric contexts. It is important to note that the discrimination index observed for the control group in the present study was lower than that reported in our previous work (Maidana et al. 2025). Such variability is not unexpected, as the Novel Object Recognition test is highly sensitive to subtle differences between experimental cohorts (Cohen and Stackman Jr. 2015). Factors such as the specific set of objects used, minor variations in illumination or contextual cues within the testing arena, and differences in handling or baseline exploratory drive can influence overall discrimination performance. Despite these variations, control animals in the current experiment still exhibited positive discrimination values, consistent with intact recognition memory, while OUA‐induced animals showed the expected impairment. Therefore, the pattern of group differences remained robust across studies. These behavioral benefits are particularly relevant given the persistent cognitive impairments observed in BD patients, even during euthymic periods (Bora 2018).

Oxidative stress is a key feature in the pathophysiology of BD and has been consistently associated with mitochondrial dysfunction and neuronal damage (Scaini et al. 2021). In the present study, ouabain administration induced marked oxidative stress, as evidenced by increased DCFH oxidation, elevated levels of 3‐NT and 4‐HNE, and altered activities of key antioxidant enzymes, including SOD, CAT, GPx, and GST. These findings are consistent with previous reports showing that ouabain disrupts redox homeostasis and compromises the brain antioxidant defense systems (Valvassori et al. 2022). Notably, previous strength training was effective in restoring redox balance, normalizing enzymatic activity, reducing the accumulation of reactive species, and ultimately attenuating oxidative damage. These results suggest that strength exercise improves endogenous antioxidant defenses and protects vulnerable brain regions from redox‐related injury.

Beyond its direct effects on cellular integrity, oxidative stress is also a well‐established trigger of neuroinflammatory responses (Scarian et al. 2024). Excessive production of ROS activates redox‐sensitive transcription factors, such as NF‐κB (Nakajima and Kitamura 2013), which in turn promote the expression of pro‐inflammatory cytokines, chemokines, and enzymes including COX‐2 and iNOS (Anilkumar and Wright‐Jin 2024). In support of this mechanism, we found a significant positive correlation between DCFH oxidation and NF‐κB activation, indicating that ROS accumulation likely contributed to the pro‐inflammatory profile observed in the ouabain group. This inflammatory cascade may lead to the activation of glial cells—particularly astrocytes and microglia—contributing to a neurotoxic environment (Kwon and Koh 2020). Astrocyte activation, often reflected by increased GFAP expression, may initially represent an adaptive response; however, sustained or dysregulated activation is associated with neuroinflammation and neuronal dysfunction (Ben Haim et al. 2015). In the present study, ouabain‐induced GFAP upregulation was accompanied by marked increases in astrocytic branching complexity, as revealed by Sholl analysis, consistent with a reactive astroglial phenotype. Importantly, prior strength training attenuated both GFAP immunoreactivity and the associated morphological hypertrophy, as well as reduced GFAP protein levels assessed by Western blot, indicating that exercise limits the transition from adaptive astrocytic remodeling to pathological astrogliosis. Furthermore, oxidative and inflammatory insults impair mitochondrial function and activate apoptotic signaling cascades, ultimately contributing to neuronal loss (Dong et al. 2024). Consistent with this mechanism, ouabain increased the cleaved caspase‐3/total caspase‐3 ratio and reduced NeuN immunoreactivity, indicating activation of apoptosis and compromised neuronal integrity. Prior strength training prevented caspase‐3 activation, supporting its anti‐apoptotic effect, although NeuN reduction was not fully reversed, suggesting that exercise primarily limits apoptotic progression rather than completely restoring neuronal protein levels.

Importantly, strength training appears to counteract these pathological processes. Physical exercise is known to boost antioxidant capacity and modulate redox‐sensitive signaling pathways, thereby reducing ROS accumulation and restoring redox equilibrium (Radák et al. 2001). Additionally, physical training has been shown to downregulate pro‐inflammatory signaling through inhibition of NF‐κB activation and to decrease the production of inflammatory mediators in both peripheral and central systems (Ayari et al. 2023). These anti‐inflammatory effects are believed to contribute to reduced glial reactivity and enhanced neuronal survival. Consistently, prior strength training attenuated OUA‐induced increases in GFAP immunoreactivity, astrocytic morphological complexity, and the number of FJC‐positive cells in both the cortex and hippocampus. These findings were further supported by biochemical evidence showing reduced GFAP protein levels and attenuation of caspase‐3 activation, reinforcing the conclusion that strength training limits both astroglial reactivity and neuronal injury through convergent cellular and molecular mechanisms. Altogether, these findings suggest that the neuroprotective effects of strength training in BD are mediated, at least in part, by its capacity to disrupt the feed‐forward loop between oxidative stress, neuroinflammation, and cell death.

In addition to its antioxidant and anti‐inflammatory effects, strength training may promote mitochondrial resilience and activate intracellular pathways associated with neuroplasticity and survival (Tuon et al. 2015). A previous work from our group demonstrated that strength training activates the PI3K/AKT/mTOR/P70S6K and ERK/CREB pathways in the hippocampus (Maidana et al. 2025), which are involved in regulating cell growth, synaptic plasticity, and antioxidant defense. Furthermore, modulation of Na+/K+‐ATPase activity and NMDA receptor signaling—both of which are disrupted by ouabain—may also contribute to the behavioral and neurochemical improvements observed (Arnaiz and Bersier 2014).

Taking together, our findings highlight the potential of structured strength training as an adjunctive therapeutic strategy in bipolar disorder. By attenuating oxidative stress, modulating neuroinflammation, preserving neuronal integrity, and improving behavioral outcomes, strength exercise emerges as a low‐cost, accessible, and biologically plausible intervention for BD. Moreover, our results reinforce the utility of the ouabain model for investigating the neurobiological underpinnings of BD and for evaluating novel treatment strategies targeting redox and inflammatory pathways.

Nonetheless, some limitations should be considered. Although the ouabain model reproduces core features of BD, it does not fully capture the complexity of the disorder in humans. A limitation of this study is the absence of classical depressive‐like behavioral tests, such as the sucrose preference test and the forced swim test. These assays were not included to avoid additional stress or prolonged individual housing, which could have interfered with oxidative, inflammatory, and glial endpoints assessed in the same cohort. Although our behavioral battery effectively captured locomotor, anxiety‐like, and cognitive domains, this restricts the depth of interpretation regarding the depressive phase. Another limitation concerns the evaluation of neurodegeneration using Fluoro‐Jade C (FJC), whose visualization under brightfield microscopy may reduce contrast and sensitivity compared to fluorescence‐based imaging, restricting the analysis to a primarily qualitative interpretation. To overcome this constraint, complementary biochemical approaches were employed, including Western blot quantification of NeuN as a marker of neuronal integrity and analysis of the cleaved caspase‐3/total caspase‐3 ratio as an index of apoptotic signaling. Together, these molecular measures provide resolution‐independent support for neuronal injury and strengthen the overall interpretation of the findings. Nonetheless, while our results provide evidence of neuroprotective effects at multiple levels, further research is needed to unravel the precise molecular mechanisms linking strength training to mitochondrial function, oxidative regulation, and long‐term neuroplastic adaptations.

Author Contributions

Luan Machado Maidana: conceptualization, investigation, writing – original draft, methodology, formal analysis. Jozyê Milena da Silva Guerra: writing – review and editing, investigation, methodology. Adson Souza‐Pereira: methodology, investigation, writing – review and editing. Mustafa Munir Mustafa Dahleh: investigation, writing – review and editing. Douglas Buchmann Godinho: investigation, writing – review and editing. Marina Prigol: writing – review and editing, resources. Luiz Fernando Freire Royes: resources, writing – review and editing. Leonardo Magno Rambo: resources, writing – original draft, conceptualization, methodology, formal analysis.

Funding

This work was supported in part by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES) and Fundação de Amparo à pesquisa do Estado do RS (FAPERGS). CNPq—L.F.F. Royes (grant no. 307382/2017‐6); M. Prigol (grant no. 307068/2023‐4). CAPES—L.M. Maidana, J.M.S. Guerra, A.S. Pereira, M.M.M. Dahleh, D.B. Godinho. FAPERGS—M. Prigol (23/2551‐0001900‐0).

Conflicts of Interest

The authors declare no conflicts of interest.