The study was performed in strict accordance with the guidelines specified in the Guide for the Care and Use of Laboratory Animals, with approval from the Ethics Committee of the Medical Faculty at Universitas Padjadjaran. The methodologies applied in this research align with the ethical standards of the relevant national and institutional bodies that oversee animal experimentation. The methodologies have received approval from the Research Ethics Committee at Universitas Padjadjaran (Number 1344/UN6.KEP/EC/2023).

In this study, 25 12-week-old male Wistar rats with an average weight of 200 g were housed in either experimental or control cages. The rats were procured from Biofarma Company, located in Bandung, Indonesia, and housed at the Animal Research Innovation center, Universitas Padjadjaran, Bandung, Indonesia. ^{56, 57} Acclimatization after transfer was 2 weeks. All animals were maintained under a 12-h light-dark cycle (lights activated at 19:00) under controlled environmental conditions (22–24°C). Food and water were provided ad libitum throughout the two-week acclimatization period. ⁵⁸

Experimental Procedure

Social Isolation Procedure and Unpredictable Chronic Mild Stress. ^{56, 57, 59– 64}

All the Wistar rats that passed the acclimatization phase were randomly divided into groups. The Wistar rats were divided into 5 groups: a normal housing group (NCG), a positive control group (social isolation and chronic unpredictable mild stress (PCG), a positive control group with antidepressant fluoxetine treatment (PCG + FLX), a depression model group with moderate intensity continuous training (PCG + MICT), and a positive control group with combination antidepressant and moderate intensity continuous training (PCG + FLX + MICT).

The impact of antidepressants, continuous moderate-intensity training, and the combination of antidepressants with continuous moderate-intensity training on depressive-like behavior in Wistar rats was evaluated through weekly weight measurements, serum corticosterone levels, and the gene expression of Srm, V-GluT1, m-GluR1, and FGF-2 within the brain throughout the duration of the 8-week study. The body weight of the rats was recorded weekly using a digital scale (Jiuyue) to monitor changes throughout the intervention period. Figure 1 presents overview of experimental design and allocated treatment to each group. At the conclusion of the experiment, all groups of rats were euthanized using 5% isoflurane (Mersi) inhalation methods. Subsequently, the rats were dissected from the cranial region to facilitate the extraction of the brain tissue. The organs were placed in tubes according to their respective groups and stored at −80°C. (Thermo Scientific Forma 900 Series −86°C Upright Ultra-Low Temperature Freezers).

Rats were subjected to a protocol to establish a depressed state with chronic unpredictable mild stress and social isolation. The unpredictable stress was induced by multiple stressors, including food and drink restrictions. Another stress comes from a 12-hour light/dark cycle (three days a week (Wednesday – Friday – Saturday): a 12-hour light/dark cycle is changed and continuous lighting is applied on Tuesdays and Sundays (24 hours) – change every week. During splash testing, a 10% sucrose solution is randomly sprayed on the back of the mouse. Every Wednesday, Friday, and Sunday, 200 ml of water is added to the bedding (100 g) and left for 24 h. Every Tuesday for 30 minutes, the Wistar rat was positioned within a cylindrical apparatus with a diameter of 80 mm and a height of 315 mm (Taffhome). Every 10 minutes, the cage was sprayed with a solution of 10% peppermint and 10% lemongrass (RealEagle). ⁶³

Intervention and behavioral test

Continuous training with moderate intensity and fluoxetine medication were the intervention for this study. Continuous training consisted of daily moderate-intensity treadmill running (Ideas). The treadmill speed was set at 15 m/min for 30 min. The training was held at 09.00 am. ⁶⁵ The pharmacotherapy used in this research was Fluoxetine (Nopres ^, DexaMedica), given daily by oral gavage at a dosage of 10 mg/kg for 2 months, at 08.00–09.00 am. ⁶⁶ The behavioral assessment employed in this study was the FST in weeks 4 and 8 at 08.00–10 am. The FST commenced with a 20-minute acclimatization period in the room containing the cylinders, during which both animals underwent simultaneous testing. The Wistar rats were evaluated within a transparent glass cylinder, with a diameter of 20 cm, filled with water to a depth of 39 cm at a temperature of 24°C. A Swanbee B-650/DC cylinder was used, measuring 65 cm in height and 24 cm in width. This test was conducted to assess the rat’s escape attempts from the cylinder during weeks 4 and 8, with each session lasting 6 minutes. The assessment focused on the frequency and duration of behaviors such as swimming, climbing, and immobility. Fresh water was replenished for each rat. During the FST, the Wistar rats were visually isolated from one another by placing dark gray dividers between the beakers. Following the completion of the experiment, all rats were dried using dry towels and subsequently placed in a warm cage for a duration of 15 minutes. ^{56, 64, 67, 68} Documentation of treadmill running and swimming test is presented on Figure 2 and Figure 3, respectively.

Real-time polymerase chain reaction

Approximately 10 mg of prefrontal cortex tissue was homogenized in a Sodium Dodecyl Sulfate (SDS) Lysis Buffer Solution (Promega Wizard). The homogenate was centrifuged at 10,000 g for 10 minutes at 4°C, after which the supernatant was analyzed by RT-PCR.(Bioline SensiFAST SYBR No-ROX One Step Kit, cat. no. BIO-72005) to know the gene expression of Fibroblast Growth Factor-2 (FGF-2), Spermidine Synthase (Srm), Glutamate Metabotropic Receptor1 (m-GluR1), Vesicular Glutamate Transporter1 (VGLUT1) in the brain (The primer manufactured by Integrated DNA Technologies). The level of FGF-2 (Gene ID 54250), Spermidine Synthase (Srm), m-GluR1, and V-GLUT 1 was evaluated relative to that of the housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The design of the primers was using the NCBI Primer-BLAST tool ( https://www.ncbi.nlm.nih.gov/tools/primer-blast/) based on NCBI Reference Sequences (RefSeq). Primer specificity was verified against the Rattus norvegicus RefSeq mRNA database. ( https://www.ncbi.nlm.nih.gov/genbank/, accessed on 4 May 2023. The design of the primers was using the NCBI Primer-BLAST tool based on NCBI Reference Sequences (RefSeq). Primer specificity was verified against the Rattus norvegicus RefSeq mRNA database. The primer sequences used in this study are shown in Table 1.

Blood was collected for corticosterone serum test from the tail of a rat between 9 and 10 am after. ^{69, 70} Using a serum separator tube, the samples were clotted for 30 min at room temperature before centrifugation for 15 min at 1000 × g. The blood serum was tested using the ELISA procedure for serum corticosterone level (ABClonal Corticosterone ELISA Kit 96 wells, cat. no. RK09054) in weeks 1, 4, and 8. The assays were performed in duplicate according to the manufacturer’s instructions.

Data analysis was conducted using the Statistical Package for the Social Sciences, version 24 (Chicago, IL, USA), with results expressed as mean ± SEM. The Kruskal-Wallis test was employed to determine statistical differences among the four groups, contingent on the normality test. The significance of the findings was further evaluated using the Mann–Whitney test and ANOVA, depending on the normality of the data. A p-value of 0.05 was considered statistically significant.

At baseline, body weight did not differ significantly among the five groups (p > 0.05), thereby confirming comparable initial conditions before the start of experimental procedures. The change in body weight (ΔBW) was determined by calculating the difference between BW at each subsequent time point and the baseline (H-0) for weeks 1, 4, and 8. Inter-group comparisons were conducted using the Mann–Whitney U test, with the PCG serving as the reference group.

By week 4, the divergent trajectory became apparent. The PCG displayed the highest cumulative weight gain among all groups (ΔBW: 57.80 ± 8.54 g), which was significantly greater than that of the NCG (ΔBW: 27.20 ± 5.51 g; p = 0).016). The treatment groups — PCG + FLX (ΔBW: 32.00 ± 14.94 g; p = 0.249), PCG + MICT (ΔBW: 47.20 ± 6.98 g; p = 0.421), and PCG + FLX + MICT (ΔBW: 39.80 ± 5.54 g; p = 0.310)—showed numerically lower weight gain compared to the PCG, but these differences did not reach statistical significance at this time.point.

At week 8, the inter-group divergence was most prominentnced. The PCG exhibited the greatest overall weight gain (ΔBW: 95.20 ± 10.37 g), significantly exceeding that of the NCG (ΔBW: 46.80 ± 5.46 g; p = 0).012). Notably, the combination therapy group (PCG + FLX + MCT) demonstrated significantly attenuated weight gain relative to the PCG (ΔBW: 70.80 ± 4.73 g; p = 0.036), whereas neither fluoxetine monotherapy (PCG + FLX; ΔBW: 64.60 ± 10.67 g; p = 0.059) nor exercise monotherapy (PCG + MICT; ΔBW: 82.40 ± 5.91 g; p = 0.143) reached significance individually. The near-significant trend observed in the PCG + FLX group (p = 0.059) suggests a potential weight-attenuating effect of fluoxetine that may be amplified when combined with exercise.

The results indicate that the combination of social isolation and the CUMS model resulted in significant weight gain in the PCG, which is consistent with stress-induced metabolic imbalances and possible overeating as a behavioral desperation symptoms. ⁷¹ Changes in body weight across treatment groups over time are presented in Figure 4. The notable reduction in weight gain observed exclusively in the combination group at week 8—despite the lack of individual significance for either monotherapy—indicates a synergistic interaction between fluoxetine and CICT in normalizing stress-induced changes.

Baseline corticosterone concentrations were consistent across all five groups (one-way analysis of variance, F(4,20) = 0.462, p = 0.763), indicating a uniform pre-intervention HPA-axis status. By week 8, the PCG exhibited the highest corticosterone level (36,091.87 ± 278.83 ng/mL), representing a 58.62% increase from baseline, with notably low inter-individual variability. This finding aligns with a consistent and pronounced elevation in corticosterone levels induced by chronic stress, driven by the combined social isolation and CUMS paradigm (paired t-test, t = −35.54, p < 0.0001). ^{72, 73} The negative control group maintained a relatively stable corticosterone trajectory over the 8-week period (23,602.12 ± 10,515.81 ng/mL; Δ baseline = +14.5%, paired t-test p = 0.466), confirming the absence of pathological HPA-axis activation in non-stressed animals. Among the treatment groups, the combination therapy group (PCG + FLX + MCT) demonstrated the most effective reduction of stress-induced corticosterone elevation, with a week-8 mean of 23,797.63 ± 3,116.32 ng/mL, a value virtually indistinguishable from the non-stressed NCG (t = −0.040, p = 0.969) and significantly lower than the PCG (t = 8.786, p < 0.0001). The MICT-alone group (PCG + MCT) showed a moderate but significant reduction compared with PCG (29,649.05 ± 2,489.75 ng/mL; t = 5.750, p = 0.0004), while fluoxetine monotherapy (PCG + FLX) yielded a numerically lower but highly variable week-8 corticosterone level (30,828.76 ± 11,724.35 ng/mL; p = 0.345 vs. PCG), likely reflecting individual differences in SSRI responsiveness within the depression model. Figure 5 illustrates the level of corticosterone serum across treatment groups over time.

By the eighth week, both social isolation and CUMS caused a notable decrease in four indicators of brain health. The positive control group (PCG) showed significantly lower levels compared to the normal control group (NCG) for all indicators: FGF-2 (0.186 ± 0.061 vs 1.931 ± 0.327, p = 0.008), spermidine synthase/SRM (0.161 ± 0.040 vs 1.213 ± 0.108, p = 0.012), V-GluT1 (0.571 ± 0.029 vs 3.478 ± 0.501, p = 0.012), and mGluR1 (0.501 ± 0.063 vs 1.521 ± 0.494, p = 0.008; Mann-Whitney U test). The Kruskal-Wallis test indicated significant differences among groups for all markers (FGF-2, p = 0.001; SRM, p < 0.001; V-GLUT1, p < 0.001; mGluR1, p = 0.011). All three treatments led to increased levels of FGF-2, SRM, and V-GLUT1 compared with PCG (all, p < 0.05). However, a difference was noted with mGluR1 levels: fluoxetine alone did not bring mGluR1 back to normal levels (1.072 ± 0.506 vs PCG 0.501 ± 0.063, p = 0.841), whereas moderate-intensity continuous training (MICT) (1.726 ± 0.330, p = 0.008) and combination therapy (2.527 ± 0.239, p = 0.008) did. This selective resistance of metabotropic glutamate receptor expression to SSRI monotherapy, despite successful restoration of neurotrophic (FGF-2) and presynaptic glutamatergic (V-GLUT1) markers in the same treatment group, indicates that fluoxetine-mediated serotonergic enhancement does not uniformly engage all neuroplasticity domains and that exercise-dependent activation is required for postsynaptic Group I mGluR signaling.

Expression of neuroplasticity biomarkers across treatment groups at week 8 is presented by Figure 6. The combination therapy (PCG + FLX + MICT) consistently exhibited the highest fold-change relative to PCG across all four biomarkers: FGF-2 (20.14-fold), SRM (16.78-fold), V-GLUT1 (5.59-fold), and mGluR1 (5.01-fold). The combination group was the only intervention that fully normalized V-GLUT1 expression to levels statistically indistinguishable from normal controls (3.198 ± 0.270 vs NCG 3.478 ± 0.501, p = 1.000), whereas fluoxetine alone (1.434 ± 0.241, p = 0.008 vs NCG) and MICT alone (1.811 ± 0.483, p = 0.056 vs NCG) remained significantly below physiological levels. This indicates that the complete restoration of vesicular glutamate packaging capacity necessitates the combined effect of both interventions. Furthermore, SRM expression in the combination group (2.752 ± 0.151) significantly exceeded normal control levels (1.213 ± 0.108, p = 0.012), indicating a supraphysiological upregulation not observed in any of the monotherapy groups. The overshoot pattern in SRM expression among all biomarkers analyzed suggests that the combination of fluoxetine and MICT converges on the polyamine biosynthesis pathway through distinct upstream mechanisms. This convergence results in an amplification effect that elevates the expression of spermidine synthase beyond the homeostatic set points. Collectively, these findings demonstrate a clear dose-response hierarchy across depression intervention modalities. (PCG < PCG + FLX < PCG + MICT < PCG + FLX + MICT) and the results suggest that each biomarker exhibits a distinct therapeutic response profile. Specifically, mGluR1 is uniquely reliant on exercise, V-GLUT1 requires combination therapy for full normalization, and SRM and FGF-2 are particularly responsive to synergistic induction.

Several limitations of this study should be acknowledged. First, the research exclusively used male Wistar rats; given the well-documented sex-dependent differences in HPA axis regulation and neuroplasticity, this may limit the generalizability of the findings to female subjects. ^{96, 97} SecondMoreover, although gene expression was quantified at the mRNA level, protein-level validation was not conducted, and post-translational modifications could influence the activity of functional transporters or growth factors. Third, the mechanisms underlying the synergistic effects of combined therapy remain speculative and require further investigation through pathway-specific pharmacological blockade or genetic manipulation. Future research should incorporate both sexes, protein-level analyses, option from other antidepressant medication, comprehensive behavioral testing, and dose-response exercise paradigms to further elucidate the optimal therapeutic window for combined exercise–pharmacotherapy interventions in depression.

During the intermediate stage of the editorial review, various phrasing and wording alternatives were crafted using the generative AI tool Paperpal Pro, while the figure was developed with the assistance of Claude Pro. All original revisions were selected, modified, and reviewed by the authors to verify their accuracy and precision. The authors assume full responsibility for the final paper and affirm that the incorporation of AI did not affect the study’s outcomes or scientific integrity.