Our University’s Institutional Ethics Committee for the use of animals in scientific research approved this study (approval protocol no. 0096/2022), and its standards comply with those of the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Bethesda, MD, USA). Eighty male Wistar rat pups (45 g ± 5 g; 25 days old) from the university vivarium were used. All animals were housed in an experimental room at 22 ± 2 °C and maintained on a reversed light–dark cycle (dark period from 8 a.m. to 8 p.m.). The animals had free access to water and food. All efforts were made to minimize animal suffering and to use the fewest possible animals to obtain valid results.

The rats were weaned on postnatal day 21 (P21) and housed in group cages. The animals were randomized into eight experimental groups, with 10 animals per group. From P21 to P24, the animals underwent dietary adaptation with a normal protein AIN-93 diet and physical exercise adaptation. From P25 to P55, the animals underwent MLT administration and physical exercise protocols. At the end of these protocols, the rats underwent behavioral testing from postnatal day 56 (P56) to P61, followed by electrophysiological recordings of CSD from P62 to P70. Nutritional manipulation with an AIN-93 M diet was maintained from P25 until the day of electrophysiological recording (Figure 1).

The diets were formulated according to the guidelines of the American Institute of Nutrition (AIN) (Reeves et al., 1993). Wellnourished rats received the AIN-93 M diet containing 12% protein, while undernourished rats received a modified version of the same diet with 7% protein. Table 1 presents the composition of the experimental diets. Micronutrients (vitamins and minerals) were provided through a standardized mixture (AIN-93-MX and AIN-93-VX), previously obtained from Rhoster, Limited, in accordance with the nutritional guidelines established by the AIN protocol. For detailed information on these mixtures, refer to the Supplementary Tables 1, 2.

MLT (purchased from Sigma-Aldrich, St. Louis, United States) was administered from P25 to P55 on alternate days at a dose of 10 mg/kg body weight by subcutaneous injection at the beginning of the dark cycle (8 h), for a total of 15 days of application. This dose was determined a priori based on evidence from our laboratory, obtained in male Wistar rats during development, indicating that 10 mg/kg, compared to a higher dose, is associated with reduced CSD propagation and increased cortical markers of antioxidant capacity (Araújo et al., 2023). MLT was dissolved in 0.1 mL saline solution (0.9% NaCl) containing 5% ethanol (Araújo et al., 2023). The subcutaneous route was chosen because it is a safe method for administering drugs to animals, especially when gradual and sustained absorption is required. This route allows for relatively slow absorption, helping maintain constant plasma drug levels for a longer period. Additionally, it is less invasive than intramuscular, intraperitoneal, or intravenous routes, is easier to perform, causes less pain and stress to the animal, and carries a lower risk of complications compared to the other routes mentioned (Kagan et al., 2007; McDonald et al., 2010).

The animals were divided into two groups: sedentary and exercised. The experimental groups performed forced physical exercise on a motorized treadmill (INSIGHT, model EP-131) for rodents. This modality was chosen because it allows standardization of exercise intensity and frequency (Segabinazi et al., 2019). Initially, the animals underwent familiarization sessions, during which, for 3 days (P22-24), the rats were kept on the treadmill for 10 min with it turned off for visual and olfactory adaptation. Then, at a speed of 8 m/min, the rats remained active for 5 min, to adapt to sound and movement, according to González-Chávez et al. (2019). The physical exercise protocol was applied in moderate-intensity sessions from P25 to P55, lasting 40 min per session, following a protocol adapted from Braz et al. (2023). Each session included a 5-min warm-up at 12 ± 2 m/min, followed by 30 min at 20 ± 2 m/min, and a 5-min cool-down at 12 ± 2 m/min. The sessions always started at 10 a.m., with the treadmill kept at a flat incline (0°), three times a week. Forced treadmill sessions were conducted without electrical shocks; animals that refused to run were gently encouraged only with a wooden stick. If they continued to refuse, they were excluded from the study and accounted for in the final n = 10 per group. Animals in the sedentary group were placed on the treadmill for the same period, but the device remained off (Téglás et al., 2019). It is also important to state that forced treadmill exercise is considered a low-stress paradigm (Svensson et al., 2016). Since we used this method in previous studies and did not observe conspicuous signs of stress (e.g., changes in body weight) (Vitor-de-Lima et al., 2019), no physiological stress markers were collected in the present study. Further observations on this topic are discussed in the study limitations section.

Each rat was placed in the center of the apparatus, a circular arena 90 cm in diameter surrounded by a circular wall 52 cm high. The OF device was in a room with dim red lighting and sound attenuation. Rat movements were recorded for 5 min using a digital camera. After each test, the OF apparatus was cleaned with 70% ethanol. Video-recorded activity was stored on a computer and later analyzed with ANYmaze™ software (version 4.99 m).

Two object recognition tasks (ORTs) were conducted in the OF arena on two consecutive days, starting 24 h after the open-field test. Each ORT consisted of two sessions. In the first session (training phase), two identical objects were placed equidistant from the walls and symmetrically positioned around the arena’s center. Rats were allowed to explore the objects for 5 min. After a 40-min intersession interval, they returned for the second (test) session. During the test session on day 1, one of the objects was moved to a new spatial location (spatial recognition test). On day 2, one of the objects was replaced with a novel object of a different shape (novel object-shape recognition test). Based on the exploration times of the novel (N) and familiar (F) locations or objects, a discrimination index (DI) was calculated using the formula: DI = (TN – TF) / (TN + TF), where TN is the time spent exploring the novel object, and TF is the time spent exploring the familiar one.

Twenty-four hours after the last object recognition task, the animals underwent the elevated plus maze (EPM) test. The EPM apparatus had a cross-shaped design, with four arms, each measuring 49 cm in length and 10 cm in width: two open arms and two closed arms (with side walls), arranged perpendicular to each other. All arms extended from a central square platform measuring 10 × 10 cm. At the start of the test, each animal was individually placed in the center of the maze, facing one of the open arms. Rats were allowed to explore the maze freely for 5 min under dim red lighting in a sound-attenuated room. Behavioral activity was recorded by a video camera, stored on a computer, and later analyzed using ANYmaze™ software (version 4.99 m). After each trial, the apparatus was cleaned with 70% ethanol.

Electrophysiological recording of CSD was performed 1 to 5 days after the behavioral tests. Animals were anesthetized intraperitoneally with a combination of urethane (1 g/kg) and alpha-chloralose (40 mg/kg), and three burr holes were made in the right side of the skull to expose portions of the cortical surface (bregma coordinates are given in millimeters along the anteroposterior (AP) and mediolateral (ML) axes, as shown in Figure 2). The first hole was drilled in the frontal bone (AP + 2/ML + 2), and the second and third holes in the parietal bone (AP –2/ML + 3 and AP –6/ML + 3, respectively). The frontal hole served as the site for applying the stimulus that triggered the CSD, which then propagated and was recorded through the parietal openings. CSD was induced every 20 min by placing a small cotton ball (1–2 mm in diameter) soaked in 2% potassium chloride (KCl) solution (approximately 270 mM) on the frontal opening for 1 min. For 4 hours, reductions in electrocorticogram (ECoG) amplitude and shifts in direct current (DC) potential associated with CSD were recorded using a digital acquisition system (MP-150, Biopac, USA) and the AcqKnowledge software (version 4.1) with 16-bit digitization and a sampling rate of 1.0 kHz per channel. To obtain DC, the signal was filtered at 0.0–300 Hz and amplified 50 times; while to obtain the ECoG, the signal was filtered at 0.5–35 Hz and amplified 5,000 times. This system allowed real-time visualization and storage on a computer. Throughout the recording session, rectal temperature was maintained at 37 ± 1 °C using an electric heating pad. The following CSD parameters were calculated: (1) mean propagation velocity, (2) amplitude, and (3) duration of the negative component of the slow potential shift.

Samples from the left (CSD-free) portion of the cerebral cortex were used to determine levels of lipid peroxidation and reduced glutathione (GSH). Lipid peroxidation was assessed by measuring thiobarbituric acid-reactive substances (TBARS) in tissue homogenates, following the method described by Ohkawa et al. (1979). Tissue samples were homogenized (5 g:1 mL) in phosphate-buffered saline (PBS) containing 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 1.8 mM KH2PO4, supplemented with 0.15 mg/mL trypsin inhibitor (type II-S) and 1 mM PMSF, on ice. The homogenate was then centrifuged at 12,000 g for 12 min at 4 °C, and the supernatant (100 mg/mL) was added to a reaction medium containing 0.3% thiobarbituric acid, 0.4% SDS, and 7.5% acetic acid (pH 3.5). The mixture was heated to 95 °C for one hour. Samples were centrifuged (1,000 g for 10 min), and the absorbance of the supernatant was measured at 535 nm. TBARS concentration was calculated using 1,1,3,3-tetraethoxypropane as the standard. Data were corrected for the protein concentration of the homogenate.

The same supernatant obtained from the tissue homogenate was used to determine GSH levels by measuring non-protein sulfhydryl groups (Sedlak and Lindsay, 1968). The homogenate supernatant was treated with 1 volume of 10% trichloroacetic acid to precipitate proteins, then centrifuged at 5,000 g for 10 min. The resulting supernatant, containing non-protein sulfhydryl groups, was reacted with 4 mM 5,5′-dithiobis(2-nitrobenzoic acid) in a solution of 250 mM TRIS, 2.5 mM EDTA, and 10% methanol. GSH concentration was calculated using L-cysteine as the standard. Data were corrected for the protein concentration of the homogenate.

SOD activity was determined by assessing the tissue homogenate’s ability to decrease the formation of the pink chromophore (adrenochrome) produced by the oxidation of epinephrine (Misra and Fridovich, 1972). The sample (10 mg protein/mL) was added to a 50 mM sodium carbonate buffer, pH 10.2, and then supplemented with 3.0 mM epinephrine. The rate of adrenochrome formation was measured in triplicate by recording absorbance at 480 nm over 2 min, with readings every 15 s. SOD activity was expressed in arbitrary units, based on the capacity to reduce the spontaneous rate of adrenochrome formation, and was normalized to the protein content of the sample.

Catalase activity was determined by measuring the decrease in absorbance due to the reduction of H2O2 to water, as described by Aebi (1984). The sample (10 mg protein/mL) was added to 10 mM hydrogen peroxide in 50 mM potassium phosphate buffer (pH 7.0). After homogenization, the rate of H2O2 decomposition was measured at 240 nm every 15 s over 2 min. Samples were analyzed in triplicate, and values are expressed as mmol H2O2 consumed per minute, calculated using the H2O2 extinction coefficient and normalized to the sample’s protein content.

NADPH oxidase activity was evaluated by quantifying superoxide produced in the presence of NADPH (100 μM), as described by Lima et al. (2021). Superoxide production was measured using lucigenin-derived chemiluminescence. Tissue homogenate samples (100 μg/mL) were added to a reaction medium containing 10 μM lucigenin and 100 μM NADPH diluted in PBS, and luminescence was measured with a luminometer. Results were expressed as relative light units (RLU) per minute and normalized to the protein content in the sample. Superoxide release was also measured in the absence of NADPH to represent basal superoxide anion production.

To ensure the samples were sufficient to detect the expected effect sizes, we performed sample size calculations using an alpha value of 5%, a power of 80%, and standard deviation estimates from previous studies in our laboratory (Guedes, 2011; Guedes and Abadie-Guedes, 2019; Figueira-de-Oliveira et al., 2025). All samples in this study met the calculated minimum requirement.

Results were expressed as mean ± standard deviation (SD). The statistical software used was Sigmastat® version 3.5. Intergroup differences were analyzed using a three-way ANOVA, with nutritional status (well-nourished versus malnourished), exercise condition (sedentary versus exercised), and systemic treatment (vehicle versus MLT) as factors. When appropriate, ANOVA was followed by the Holm-Sidak test. Differences with p < 0.05 were considered significant.

In the open field test (OF), well-nourished animals (Figure 3) showed that physical exercise led to more entries and longer stays in the central area of the apparatus compared to sedentary animals. This pattern was also seen in exercised rats treated with melatonin compared to those given only a vehicle solution. In the malnourished groups (Figure 4), exercised animals also had more entries and longer stays in the central region, indicating an anxiolytic effect of exercise. Furthermore, MLT administration further enhanced this effect, as MLT-treated rats performed better than vehicle-treated rats.

The elevated plus maze (EPM) task was also used to assess anxiety behavior in well-nourished (Figure 5) and malnourished (Figure 6) animals. In both groups, exercised animals made more entries into and spent more time in the maze’s open arms than sedentary animals. This effect was more pronounced in the malnourished groups, suggesting that these animals were more sensitive to the anxiolytic effects of exercise.

Short-term memory was assessed using the object recognition test in two modalities: shape recognition and position recognition. Performance was measured by the discrimination index, defined as the proportion of time spent exploring the novel object relative to the total time spent exploring all objects. Positive values of this index indicate a preference for the novel object, reflecting better recognition memory performance. The results showed that animals subjected to physical exercise and/or treated with melatonin had higher discrimination indices than the sedentary group treated with a vehicle. In both the shape and position recognition tests, the exercised and/or melatonin-treated groups performed better than the sedentary control group, suggesting that both treatments facilitated recognition of the novel stimulus (Figure 7).

After topical application of 2% KCl to the cortical region, CSD was confirmed by slow depolarization and reduced ECoG activity (Figure 2). Malnutrition and sedentary lifestyles enhanced CSD propagation compared with well-nourished, physically active groups. In contrast, both physical exercise and melatonin treatment showed neuroprotective effects, acting as antagonists of the phenomenon’s propagation. Animals subjected to physical activity and/or treated with melatonin exhibited a significant reduction in the speed and amplitude of CSD propagation, along with an increase in its duration (see Figure 8).

A low-protein diet disrupted cerebral redox balance. The malnourished-sedentary group treated with vehicle showed higher TBARS levels than the corresponding well-nourished group (well-nourished sedentary vehicle), indicating greater lipid peroxidation and oxidative damage (Figure 9A). In contrast, all malnourished groups subjected to physical exercise, melatonin (MLT) treatment, or both, had lower cerebral cortex TBARS levels than the malnourished sedentary vehicle group. The low-protein diet also significantly reduced glutathione (GSH) levels in the cerebral cortex of sedentary, malnourished rats compared to well-nourished controls (Figure 9B). Melatonin increased these GSH levels, and exercise produced a similar improvement. The combination of melatonin and exercise brought the values even closer to those observed in well-nourished animals, indicating a more robust restoration of the antioxidant system.

Basal production of superoxide anions in the cerebral cortex was significantly increased in the malnourished sedentary vehicle group compared to the well-nourished groups, confirming a basal pro-oxidant state (Figure 10A). In the tissue, NADPH oxidase activity, a major source of reactive oxygen species (ROS), was also elevated due to malnutrition. Both melatonin administration and physical exercise alone effectively reduced basal superoxide anion production and mitigated the hyperactivity of NADPH oxidase, bringing the levels closer to those of the well-nourished groups (Figure 10B). The combination of exercise and melatonin did not further inhibit basal superoxide anion production or NADPH oxidase activity in malnourished rats (Figure 10).

Superoxide dismutase (SOD) activity was reduced in the cerebral cortex of the malnourished sedentary vehicle group compared to the well-nourished group (Figure 11A). In malnourished rats, physical exercise increased SOD activity than in the corresponding sedentary group. Additionally, both sedentary and exercised malnourished groups treated with melatonin showed greater SOD activity than their respective vehicle-treated groups. In the well-nourished groups, melatonin administration in the exercised group also increased SOD activity compared to the other well-nourished groups (Figure 11A).

Catalase activity was also reduced in the cerebral cortex of the malnourished sedentary vehicle group relative to the wellnourished sedentary vehicle group (Figure 11B). In malnourished rats, exercise and/or melatonin treatment led to greater catalase activity than in the malnourished sedentary vehicle group. However, the malnourished group that received both exercise and melatonin exhibited the highest catalase activity compared to the malnourished groups subjected exclusively to exercise or MLT treatment (Figure 11B).

A summary of all statistical differences is shown in Table 2.

Pharmacological evidence from our laboratory supports the plausibility that the cortical oxidative environment can modulate CSD dynamics in a dose-dependent manner. Araújo et al. (2023) observed divergent effects of melatonin depending on the dose: 10 mg/kg was associated with a profile more compatible with an antioxidant effect, slowing CSD and improving behavioral performance, whereas a higher dose showed the opposite effect on CSD propagation. Similarly, Mendes-da-Silva et al. (2014) demonstrated that ascorbic acid can exert antioxidant or pro-oxidant actions depending on the dose, with divergent effects on oxidative stress and CSD. In the present study, the melatonin dose selected for its antioxidant profile was associated with behavioral improvement and increased cortical resistance to CSD, reinforcing the notion that strengthening endogenous antioxidant capacity helps reduce deleterious effects associated with a pro- oxidant context, such as malnutrition. However, because this is a comparative design, our data do not establish causality. Future studies should employ specific mechanistic manipulations (e.g., pro-oxidant challenges in well-nourished animals, blocking antioxidant pathways in the exercise group, and independent modulation of cortical excitability) to test whether changes in redox state and/or CSD are necessary and sufficient for behavioral changes.

Although we used shock-free treadmill running with an adaptation phase, and previous studies from our laboratory have already observed a reducing effect of physical exercise on CSD propagation and anxiety-like behavior regardless of the paradigm adopted (forced or voluntary) (Braz et al., 2023; Silva-Gondim et al., 2019; Vitor-De-Lima et al., 2024), we recognize that imposed exercise can, under certain conditions, act as a stressor and influence behavioral outcomes (Svensson et al., 2016). However, in the present study, we did not observe anxiogenic effects of this intervention compared to sedentary groups. Future studies should include direct measures of stress (e.g., corticosterone or other physiological markers), as well as comparisons between imposed and voluntary exercise, to more accurately dissociate the effects of exercise from the stress component associated with the method.

Finally, in this study, only male rats were used as a common strategy to avoid modulation of the CSD phenomenon by natural fluctuations in estrogens and progestogens (Accioly and Guedes, 2020; Kudo et al., 2023). However, we recognize sex-related differences in redox balance (Kamper et al., 2009) and melatonin pharmacodynamics (Adamah-Biassi et al., 2014). Therefore, we acknowledge this limitation and recommend that future studies include both sexes.