miR-137-5p-Loaded Milk-Derived Small Extracellular Vesicles Modulate Oxidative Stress, Mitochondrial Dysfunction, and Neuroinflammatory Responses in an In Vitro Alzheimer’s Disease Model

TL;DR

Milk-derived extracellular vesicles loaded with miR-137-5p can reduce Alzheimer’s disease-related damage in lab-grown brain cells.

Contribution

Milk-derived sEVs loaded with miR-137-5p show multi-target therapeutic potential in an Alzheimer’s disease model.

Findings

Aβ exposure increased oxidative stress, inflammation, and mitochondrial dysfunction in SH-SY5Y cells.

miR-137-5p-loaded sEVs normalized multiple AD-related pathological markers toward control levels.

Unloaded sEVs partially modulated but did not fully reverse AD-like cellular alterations.

Abstract

Background/Objectives: Alzheimer’s disease (AD) is characterized by progressive neurodegeneration driven by interconnected mechanisms, including oxidative stress, mitochondrial dysfunction, neuroinflammation, synaptic impairment, and abnormal protein aggregation. MicroRNAs (miRNAs) have emerged as post-transcriptional regulators of these complex pathways; however, efficient delivery remains a major limitation. Small extracellular vesicles (sEVs) have been proposed as biologically compatible carriers for miRNA delivery. Methods: In this study, milk-derived sEVs were isolated, characterized, and loaded with microRNA-137-5p (miR-137-5p). Their effects were evaluated in an amyloid-β (Aβ)-induced in vitro AD model using SH-SY5Y human neuroblastoma cells. Oxidative stress markers, including reactive oxygen species (ROS), malondialdehyde (MDA), superoxide dismutase (SOD), lactate dehydrogenase (LDH), and glutathione peroxidase 1 (GPX1), were assessed. Inflammation- and neuroprotection-related gene expression analyses included intercellular adhesion molecule 1 (ICAM1), tumor necrosis factor alpha (TNF-α), and brain-derived neurotrophic factor (BDNF). Cytoskeletal injury was evaluated using neurofilament light chain (NfL). Mitochondrial stress markers included cytochrome c (Cyt-c), 8-hydroxy-2′-deoxyguanosine (8-OHdG), PTEN-induced kinase 1 (PINK1), dynamin-1-like protein (DNM1L), and mitochondrial transcription factor A (TFAM). Synaptic and extracellular matrix-associated proteins, including complexin-2 (CPLX2), SPARC-related modular calcium-binding protein 1 (SMOC1), and receptor tyrosine kinase-like orphan receptor 1 (ROR1), as well as AD-related biomarkers, including total tau, phosphorylated tau at threonine 181 (pTau-181), phosphorylated tau at threonine 217 (pTau-217), and amyloid-β 1–40 (Aβ1–40), were evaluated using molecular and biochemical approaches. Results: Aβ exposure was associated with increased oxidative stress, inflammatory activation, mitochondrial and cytoskeletal alterations, synaptic-related disturbances, and elevations in tau- and amyloid-associated proteins. Treatment with unloaded sEVs was associated with partial modulation of several parameters, whereas miR-137-5p-loaded sEVs were consistently associated with normalization of multiple pathological markers toward control levels. Conclusions: These findings indicate that miR-137-5p-enriched sEVs may represent a useful experimental platform for multi-target modulation of AD-related cellular alterations. Further mechanistic and in vivo studies are required to clarify translational relevance.

Figures

Full text

1. Introduction

Alzheimer’s disease (AD) is a complex and progressive neurodegenerative disorder characterized by synaptic dysfunction, neuronal loss, and cognitive decline, with extracellular β-amyloid (Aβ) deposition representing a central pathological hallmark [1]. Despite substantial advances in elucidating disease mechanisms, current therapeutic strategies remain largely symptomatic and fail to effectively interfere with the underlying neurodegenerative process [2]. This therapeutic limitation underscores the urgent need for disease-modifying approaches that target early and interconnected molecular drivers of AD progression.

Accumulating evidence indicates that oxidative stress, mitochondrial dysfunction, and chronic neuroinflammation constitute a self-amplifying pathogenic network that critically contributes to neuronal vulnerability and disease advancement [3]. These interrelated processes not only exacerbate Aβ-mediated neurotoxicity but also compromise cellular resilience, synaptic integrity, and neuronal survival. Consequently, therapeutic strategies capable of simultaneously modulating these converging pathways are increasingly recognized as promising candidates for effective intervention in AD [4].

In this context, transcriptional and post-transcriptional regulators of gene networks governing neuroinflammation, redox homeostasis, mitochondrial function, and neuronal survival have emerged as attractive targets for understanding and modulating AD-related pathology [5]. Among these regulators, non-coding RNAs, particularly microRNAs (miRNAs), have gained considerable attention due to their ability to fine-tune gene expression at the post-transcriptional level and their involvement in multiple neurodegenerative disease processes [6]. These properties position miRNAs as compelling candidates for multi-target therapeutic modulation in AD.

Previous studies have investigated the therapeutic potential of miRNA mimics in various neurodegenerative disorders, demonstrating their capacity to regulate disease-associated molecular pathways [7,8]. However, effective clinical translation of miRNA-based therapies remains challenging due to issues related to instability, off-target effects, and inefficient delivery [9]. To overcome these limitations, recent research has focused on nanocarrier-based delivery systems designed to enhance miRNA stability, bioavailability, and target specificity [10]. Among these platforms, small extracellular vesicles (sEVs) have emerged as particularly promising due to their lipid bilayer structure, which protects RNA cargo from enzymatic degradation and facilitates efficient cellular uptake [11,12].

In the present study, milk-derived sEVs were employed as a natural nanocarrier system owing to their low immunogenicity, inherent bioactivity, and feasibility for large-scale isolation [13]. Beyond their carrier function, milk-derived sEVs have been reported to exert intrinsic antioxidant and anti-inflammatory effects, further supporting their therapeutic relevance in neurodegenerative conditions [14]. Studies have identified reduced miR-137 levels in AD patient samples [15], and experimental restoration of miR-137-5p has been shown to ameliorate AD-like pathology and cognitive impairment in vivo, supporting its potential as a therapeutic candidate [16,17].

MiRNA-137-5p was actively loaded into milk-derived sEVs, and the effects of miRNA-137-5p-loaded sEVs were evaluated in an in vitro amyloid-β-induced AD model using human SH-SY5Y neuroblastoma cells. By integrating miRNA-based multi-target regulation with a biologically compatible vesicular delivery system, this study explores a novel and mechanistically rational therapeutic strategy aimed at mitigating key molecular processes underlying AD-associated cellular pathology.

2. Materials and Methods

2.1. Isolation and Characterization of Milk-Derived EVs

Raw bovine milk samples were obtained from the Atatürk University Farm and subjected to microbiological analysis to confirm sterility and suitability for subsequent experimental procedures. Samples meeting sterility criteria were processed for sEV isolation.

sEVs were isolated from milk using a combination of sequential centrifugation and ultracentrifugation steps, following the protocol described by Del Pozo-Acebo et al., with minor modifications [18]. Briefly, milk samples were subjected to differential centrifugation to remove fat globules, cellular debris, and casein aggregates prior to vesicle enrichment. To eliminate cells and milk fat globules, 1 mL of milk was centrifuged at 13,000× g for 30 min at 4 °C using a JLA-16.250 rotor (Beckman Coulter, Brea, CA, USA) in an Avanti J-26XPI (Beckman Coulter, Brea, CA, USA) centrifuge. The resulting supernatant was subsequently centrifuged at 35,000× g for 60 min at 4 °C to remove residual debris and large protein aggregates. For sEV pelleting, the supernatant was ultracentrifuged at 100,000× g for 105 min at 4 °C using a Optima L-90K ultracentrifuge equipped with a 50.2 Ti rotor (Beckman Coulter, Brea, CA, USA). After discarding the supernatant, the EV-containing pellet was resuspended in 25 mL of phosphate-buffered saline (PBS).

An additional ultracentrifugation step was performed at 100,000× g for 105 min at 4 °C to further wash and concentrate the vesicles. The final pellet was resuspended in 700 µL of PBS and subsequently passed through a 0.22 µm syringe filter (Millex^®^-GP, Merck Millipore, Burlington, MA, USA) to remove larger particles and improve sample purity.

For further purification, size-exclusion chromatography (SEC) was performed using an Izon qEV column (Izon Science, Christchurch, New Zealand), with minor modifications based on the work of Aliakbari et al. [19]. The column was first equilibrated with 10 mL of cold PBS. Subsequently, 700 µL of the filtered sEV suspension was loaded onto the column, followed by elution with PBS. The initial 3 mL of eluate (fractions 1–6) was discarded, and the subsequent fractions enriched in sEVs (fractions 7–12; total volume 2 mL) were collected into sterile Eppendorf tubes. The collected sEV-containing fractions were concentrated using a Vivaspin centrifugal concentrator (Sartorius, Göttingen, Germany) at 4000× g for 40 min at 20 °C, yielding a final volume of approximately 2 mL [20].

Following isolation, sEVs were characterized in accordance with the MISEV2018 guidelines. Vesicle morphology was assessed by transmission electron microscopy (TEM), while size distribution and particle concentration were determined using dynamic light scattering (DLS) and nanoparticle tracking analysis (NTA).

2.2. Loading of miRNA-137-5p into sEVs

For miRNA loading, 200 µg of sEV protein and 1000 µg of miRNA-137-5p mimic (Thermo Fisher Scientific, Waltham, MA, USA) were incubated with 0.2% (w/v) saponin (Merck, Darmstadt, Germany) at a final volume of 1 mL of 1× PBS. Saponin was used as a membrane-permeabilizing agent to facilitate the encapsulation of miRNA into sEVs. Based on nanoparticle tracking analysis (NTA), this preparation contained approximately 1 × 10^8^ sEV particles per mL, indicating that the applied dose corresponds to this particle range, which is consistent with doses commonly used in extracellular vesicle-loading studies.

The identical process was used to create control EV samples, but the miRNA-137-5p mimic was not added. To guarantee homogeneity, the resultant mixtures were vortexed for ten seconds before being incubated under the same circumstances. After incubation, residual saponin and unencapsulated miRNA were eliminated by purification using PD-10 size-exclusion chromatography columns (Cytiva, Marlborough, MA, USA). According to the protocol described by Albaladejo-García et al. [21], the first 1.5 mL of the eluate was collected and used for subsequent physicochemical characterization and downstream experiments.

2.3. Dynamic Light Scattering (DLS)

A Zetasizer Nano ZS (Malvern Instruments, Malvern, UK) was used to measure the hydrodynamic diameter and size distribution of sEV samples. Large aggregates were eliminated by passing vesicle suspensions through a 0.45 µm syringe filter after being diluted 1:250 in 1 mL of 1× PBS. Measurements were carried out at 25 °C in compliance with the manufacturer’s instructions and [22].

2.4. Nanoparticle Tracking Analysis (NTA)

NTA used a NanoSight NS300 system (Malvern Technologies, Malvern, UK) with a 488 nm laser and a high-sensitivity scientific CMOS camera to examine the particle concentration and size distribution of sEVs. According to the manufacturer’s instructions, sEV samples were diluted in particle-free PBS (Gibco, Waltham, MA, USA) prior to analysis in order to reach an ideal particle concentration within the suggested detection range (serum-derived samples: 1:500 dilution). Measurements were made under continuous flow circumstances (flow rate: 50) at regulated room temperature (23–25 °C). Fifteen 60 s video recordings were made at various camera locations for each sample. NTA software version 3.1.54 was used for data collection and processing, with a bin size of 2 and a detection threshold of 5 [21].

Vesicle size was evaluated using both DLS and NTA, as these techniques provide complementary information. DLS Z-average values represent intensity-weighted measurements and may be influenced by larger particles, whereas NTA provides number-based size distributions of individual vesicles. Accordingly, NTA was primarily used to assess the size distribution of milk-derived sEVs.

2.5. Imaging Using Transmission Electron Microscopy (TEM)

A total of 10 µL of isolated sEV suspension was applied to formvar/carbon-coated copper grids for TEM examination, and the grids were left to adsorb for 20 min at room temperature. After three gentle PBS rinses, the grids were fixed with 2.5% glutaraldehyde for ten minutes. The grids were negatively stained with 2% (w/v) uranyl acetate for 15 min after being fixed and cleaned with ultrapure water. After that, the grids were allowed to air dry at ambient temperature for 20 min. As previously mentioned, TEM imaging was carried out using a Tecnai 120 transmission electron microscope (FEI, Hillsboro, OR, USA) running at an accelerating voltage of 120 kV [21].

2.6. RT-qPCR-Based Validation of miR-137-5p Encapsulation in sEVs

Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was employed to verify the successful loading and intravesicular localization of miRNA-137-5p within sEVs. To distinguish encapsulated miRNA from surface-associated or unbound miRNA, RNase protection assays were performed. Accordingly, pre-loading (input), miRNA-loaded sEVs, mock-loaded sEVs, and free miRNA controls were analyzed under RNase-only and RNase plus membrane-disruptor conditions.

To control RNA extraction efficiency and technical variability, a constant amount of an exogenous non-mammalian miRNA spike-in was added to each sample prior to RNA isolation. Following total RNA extraction, miRNA-specific reverse transcription was performed, and quantitative PCR was carried out using validated primer/probe sets. Absolute quantification of miRNA-137-5p was achieved by generating standard curves from serial dilutions of synthetic miRNA-137-5p, enabling precise determination of miRNA copy numbers. Data were expressed as both absolute copies per vesicle and relative fold changes compared with control groups.

miRNA-loaded sEVs exhibited significantly higher levels of miRNA-137-5p compared with mock-loaded sEVs and free miRNA controls, indicating efficient loading. Importantly, miRNA-137-5p levels were preserved following RNase-only treatment but were markedly reduced after combined RNase and membrane-disruptor treatment, confirming that the miRNA cargo was protected within intact vesicles rather than adsorbed on the vesicle surface.

Ct values and calculated copy numbers were subjected to appropriate statistical analyses, including predefined sample size determination and corrections for multiple comparisons.

2.7. Cell Culture

The American Type Culture Collection (ATCC, Manassas, VA, USA) provided the human neuroblastoma SH-SY5Y cell line. The cells were grown in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Waltham, MA, USA) supplemented with 1% penicillin-streptomycin and 10% fetal bovine serum (FBS; Gibco). Cultures were kept at 37 °C in a humidified incubator with 5% CO_2_. SH-SY5Y cells were treated with amyloid-β (Aβ; C008-4, Meilunbio, Dalian, China) for 48 h to create an in vitro AD model. Subsequent experimental evaluations were conducted using Aβ-treated cells (SH-SY5Y-Aβ).

To establish an in vitro model of AD, SH-SY5Y cells were treated with Aβ [23]. Oligomeric amyloid-β_1–42_ (Aβ_1–42_) was prepared according to established protocols. Briefly, the synthetic human Aβ_1–42_ peptide (C008-4, Meilunbio, Dalian, China) was initially dissolved in hexafluoroisopropanol (HFIP) to obtain monomeric species. The solvent was then evaporated under sterile conditions, and the resulting peptide film was re-dissolved in dimethyl sulfoxide (DMSO). This solution was subsequently diluted in cell culture medium to induce oligomer formation and incubated at 4 °C for 24 h before use. The resulting oligomeric Aβ_1–42_ preparation was freshly used for cell treatments.

2.8. Cellular Viability

The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay kit (Merck, Germany) was used to measure cell viability to determine the cytotoxic effects of miRNA-137-5p-loaded sEVs (sEV_miR-137-5p) on SH-SY5Y cells. In order to enable cell attachment, SH-SY5Y cells were seeded into 96-well plates at a density of 1 × 10^4^ cells per well and incubated under standard culture conditions (37 °C, 5% CO_2_, humidified environment). After incubation, cells were exposed to two distinct quantities (0.5 µg/mL and 5 µg/mL) of either miRNA-137-5p-loaded sEVs (sEV_miR-137-5p) or unloaded control sEVs (sEV). Following treatment, cell viability was assessed in accordance with the manufacturer’s guidelines.

2.9. Assessment of Intracellular Reactive Oxygen Species (ROS)

As directed by the manufacturer, intracellular ROS production was measured using a one-step fluorometric test kit. In order to facilitate cell attachment, SH-SY5Y cells were seeded into black, clear-bottom 96-well plates at a volume of 100 µL per well and incubated for 24 h at 37 °C in a humidified environment with 5% CO_2_. After incubation, the produced master reaction mixture was applied directly to each well in place of the culture medium. After that, the cells were incubated for the suggested reaction time at 37 °C with 5% CO_2_. A microplate reader was used to measure the fluorescence intensity at excitation wavelengths (λ_ex) of 520 nm and emission wavelengths (λ_em) of 605 nm. ROS production in the treatment groups was expressed as a percentage in relation to the control group, and intracellular ROS levels were normalized to the control group, which was defined as 100%.

2.10. Measurement of Total Tau, pTau-181, pTau-217, Nfl, and Aβ1–40

Using commercially available sandwich ELISA kits (MyBioSource, San Diego, CA, USA) and following the manufacturer’s instructions, the concentrations of total Tau, phosphorylated Tau-181 (pTau-181), phosphorylated Tau-217 (pTau-217), neurofilament light chain (NfL), and amyloid beta 1–40 (Aβ_1–40_) were measured.

Cells were separated using trypsin and cleaned with ice-cold PBS after the experimental treatments. Cell suspensions were gathered and centrifuged for five minutes at 4 °C at 1000× g. The resultant cell pellets were resuspended in 800 µL of PBS after the supernatants were disposed of. Samples were repeatedly frozen and thawed to guarantee total cell lysis. After centrifuging cell lysates at 1500× g for 10 min at 2–8 °C, the cleared supernatants were collected and kept at −80 °C until ELISA analysis.

The ELISA kits used for the quantification of human proteins were as follows: Total Tau (MBS812766), pTau-181 (MBS744973), pTau-217 (MBS1608795), NfL (MBS9399603), and Aβ_1–40_ (MBS2506221).

2.11. Assessment of Oxidative Stress Parameters

The concentrations and enzymatic activities of malondialdehyde (MDA), superoxide dismutase (SOD), lactate dehydrogenase (LDH), and glutathione peroxidase 1 (GPX1) in SH-SY5Y cells were determined in accordance with the manufacturers’ instructions. Cell lysates were prepared as described above.

Human-specific ELISA kits obtained from MyBioSource (USA) were used for the analyses as follows: MDA (MBS7606463), SOD (MBS2800400), LDH (MBS009535), and GPX1 (MBS026180).

2.12. Real-Time Quantitative Polymerase Chain Reaction (RT-qPCR)

Following the manufacturer’s instructions, total RNA was extracted from SH-SY5Y cells using QIAzol reagent (Qiagen, Hilden, Germany). RNA integrity was confirmed by agarose gel electrophoresis, and RNA concentration and purity were evaluated spectrophotometrically using a NanoDrop device (Thermo Fisher Scientific, Waltham, MA, USA) by calculating the A260/A280 ratio. Using the Omniscript RT Kit (Cat. No. 205113, Qiagen, Hilden, Germany), 1 µg of total RNA was reverse-transcribed into complementary DNA (cDNA) in accordance with the manufacturer’s instructions for mRNA expression analysis. Using the same Omniscript RT Kit (Cat. No. 205113, QIAGEN, Hilden, Germany) and a stem-loop RT primer specific for miRNA-137-5p, reverse transcription was carried out for miRNA quantification.

QuantiNova SYBR Green PCR Kit (Cat. No. 208052, Qiagen, Hilden, Germany) was used for RT-qPCR on a Rotor-Gene Q MDx 5plex HRM system (Qiagen, Hilden, Germany). The thermal cycling settings included a 5 min initial denaturation step at 95 °C, 40 cycles of denaturation at 95 °C for 10 s, and annealing/extension at 60 °C for 30 s. Following amplification, a melt-curve analysis was carried out to verify the PCR products’ specificity.

Table 1 lists primer sequences for BDNF, ICAM1, TNF-α, and miRNA-137-5p along with the internal reference genes U6 small nuclear RNA (for miRNA normalization) and GAPDH (for mRNA normalization). The presence of a single peak in the melting curve analysis and the amplified products’ agarose gel electrophoresis were used to confirm primer specificity. To rule out contamination and genomic DNA amplification, each run contained no-template controls (NTCs) and minus reverse transcriptase (−RT) controls.

Relative gene expression levels were calculated using the 2^−ΔΔ^Ct method [25]. mRNA expression levels were normalized to GAPDH, while miRNA-137-5p expression levels were normalized to U6 snRNA. All RT-qPCR reactions were performed in triplicate as technical replicates, and a minimum of three independent biological replicates were analyzed.

2.13. Measurement of Inflammation-Related and Mitochondrial Injury Markers

Using commercially available human-specific ELISA kits, the levels of cytochrome c (Cyt-c), 8-hydroxy-2′-deoxyguanosine (8-OHdG), mitochondrial transcription factor A (TFAM), PTEN-induced kinase 1 (PINK1), dynamin-related protein 1 (DNM1L), complexin 2 (CPLX2), receptor tyrosine kinase–like orphan receptor 1 (ROR1), and SPARC-related modular calcium-binding protein 1 (SMOC1) were quantified.

The ELISA kits used were as follows: Human Cyt-c ELISA Kit (MBS162423, MyBioSource), Human 8-OHdG ELISA Kit (MBS267161, MyBioSource), Human TFAM ELISA Kit (MBS2020891, MyBioSource), Human PINK1 ELISA Kit (MBS2124222, MyBioSource), Human DNM1L ELISA Kit (MBS2124218, MyBioSource), Complexin 2 (CPLX2) ELISA Kit (MBS450651, MyBioSource), Receptor Tyrosine Kinase–Like Orphan Receptor 1 (ROR1) ELISA Kit (MBS2000330, MyBioSource), and SPARC-Related Modular Calcium-Binding Protein 1 (SMOC1) ELISA Kit (MBS2707129, MyBioSource).

Every test was carried out in compliance with the manufacturer’s guidelines. In short, samples were produced at the suggested dilutions, added to microtiter plates that had already been coated, and then subjected to a series of incubation, washing, and substrate development procedures. After stopping the enzymatic reaction, a microplate reader was used to measure absorbance at 450 nm. By interpolating from standard curves created for each kit, biomarker concentrations were determined from the optical density (OD) measurements.

2.14. Statistical Analysis

R software (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism (version 9.5.0; GraphPad Software, San Diego, CA, USA) were used for statistical analyses. The Shapiro-Wilk test was used to evaluate the data normality for each experimental group, confirming that all datasets had a normal distribution. Parametric statistical tests were therefore used. One-way analysis of variance (ANOVA) was used to examine differences between the four experimental groups. Tukey’s post hoc test was then used for multiple pairwise comparisons. The mean ± standard error of the mean (SEM) is used to display all data. Statistical significance was defined as a p-value of less than 0.05. GraphPad Prism was used to create graphical representations. Detailed numerical data, including ANOVA statistics and post-hoc pairwise comparisons, are provided in Supplementary Table S1.

3. Results

3.1. miR-137-5p Expression

When comparing miRNA-loaded sEVs to blank sEV controls, RT-qPCR analysis showed a significant increase in miR-137-5p expression levels (p < 0.001; Figure 1). These findings unequivocally demonstrate that miR-137-5p was successfully and effectively loaded into sEVs. Furthermore, the significant enrichment of miR-137-5p in sEVs confirms that sEV-mediated miRNA transport is a viable method for delivering functional miRNAs to target cells.

3.2. miR-137-5p Loading Does Not Significantly Influence the Morphology, Size Distribution, or Particle Concentration of Milk-Derived sEVs

Both blank and miR-137-5p-loaded sEVs exhibited the characteristic round, cup-shaped morphology of sEVs as observed by TEM (Figure 2a). The absence of vesicle aggregation or deformation suggests that vesicle structure was unaffected by the miRNA-loading process. Native sEVs had a mean particle size of roughly 190–200 nm, according to DLS analysis, and the miR-137-5p-loaded group had a similar size distribution (Figure 2b). Both samples showed hydrodynamic diameters of about 160–170 nm, and Z-average measurements showed no discernible changes between groups (Figure 2c). By displaying overlapping particle-size distribution curves with peak vesicle diameters primarily between 80 and 250 nm in both groups, NTA further corroborated these findings (Figure 2d). These results confirm that miR-137-5p loading does not significantly alter the morphology, size distribution, colloidal stability, or particle concentration of milk-derived sEVs.

3.3. Cytotoxicity Assessment of sEV_miR-137-5p in SH-SY5Y Cells

As shown in Figure 3, treatment of SH-SY5Y neuroblastoma cells with blank sEVs or miR-137-5p-loaded sEVs across increasing concentrations (0.5, 1, 2.5, 5, and 10 µg/mL) revealed a dose-dependent pattern in cell viability. At low concentrations (0.5 and 1 µg/mL), cell viability remained comparable to the untreated control group, with mean values clustered around ~100%, indicating no detectable cytotoxic effect (p > 0.05).

At 2.5 and 5 µg/mL, a mild reduction in viability was observed; however, cell survival remained above ~85–90%, suggesting limited biological impact at these intermediate concentrations. In contrast, treatment with 10 µg/mL resulted in a marked and statistically significant decrease in cell viability (approximately 50–60%), indicating pronounced cytotoxicity at this highest dose (p < 0.001–0.0001, as indicated in the figure).

Overall, these findings demonstrate that sEV-mediated miR-137-5p delivery does not adversely affect SH-SY5Y cell viability at concentrations up to 5 µg/mL, whereas 10 µg/mL exceeds the biocompatibility threshold under the present experimental conditions. Based on these results, 0.5 and 5 µg/mL were considered biologically safe and appropriate for subsequent functional analyses.

3.4. sEV_miR-137-5p Treatment Modulates Oxidative Stress Levels in Aβ-Treated SH-SY5Y Cells

To evaluate oxidative stress status, four experimental groups were compared: untreated SH-SY5Y cells (Control), Aβ-treated SH-SY5Y cells (SH-SY5Y-Aβ), SH-SY5Y-Aβ cells treated with blank sEVs (sEV), and SH-SY5Y-Aβ cells treated with miR-137-5p-loaded sEVs (sEV_miR-137-5p). Intracellular ROS levels and LDH, GPX1, MDA, and SOD activity were assessed. Statistical significance was determined relative to the untreated control group to define the oxidative stress status of each group under pathological conditions.

Aβ exposure resulted in a marked increase in oxidative stress, as evidenced by significant elevations in ROS, LDH, GPX1, and MDA levels compared with the control group (Figure 4; p < 0.001 for all), accompanied by a significant reduction in SOD activity (p < 0.001). These findings indicate pronounced oxidative damage and impaired antioxidant defense mechanisms in SH-SY5Y cells following Aβ treatment.

Treatment with blank sEVs partially modulated oxidative stress parameters; however, ROS, LDH, GPX, and MDA levels remained significantly elevated compared with the control group (p < 0.05 for %ROS, LDH, GPX; p < 0.01 for MDA), despite a modest but significant increase in SOD activity (p < 0.01 vs. control). In contrast, treatment with sEV_miR-137-5p markedly attenuated Aβ-induced oxidative stress. In this group, ROS, LDH, GPX1, and MDA levels were significantly reduced, while SOD activity was restored toward control levels. Notably, no significant differences were observed between the control group and the sEV_miR-137-5p-treated group (p > 0.05), indicating effective mitigation of Aβ-induced oxidative stress by miR-137-5p-loaded sEVs.

3.5. sEV_miR-137-5p Treatment Regulation of Inflammation and Neuroprotection-Related Gene Expression

To evaluate inflammation-related markers and neuroprotective status among the experimental groups, mRNA expression levels of BDNF, ICAM1, and TNF-α were analyzed by RT-qPCR in SH-SY5Y neuroblastoma cells. RT-qPCR analysis demonstrated that Aβ exposure significantly decreased BDNF mRNA expression levels compared with the untreated control group (Figure 5a; p < 0.001). Conversely, Aβ treatment significantly increased ICAM1 (Figure 5b; p < 0.001) and TNF-α (Figure 5c; p < 0.01) mRNA expression levels relative to control cells, indicating an imbalance between inflammatory and neuroprotective gene expression in the neuroblastoma cell model.

Treatment with unloaded sEVs partially modulated this inflammatory response. Although ICAM1 and TNF-α mRNA levels were significantly reduced compared with the Aβ-treated group, their expression remained significantly different from control values (p < 0.01 for both). In parallel, BDNF expression was increased following sEV treatment; however, it did not fully return to control levels (p < 0.05 vs. control).

Notably, treatment with sEV_miR-137-5p resulted in a more pronounced regulatory effect. In this group, ICAM1, TNF-α, and BDNF mRNA expression levels showed no statistically significant differences compared with the untreated control group (p > 0.05), indicating effective normalization of both inflammation-related and neuroprotective gene expression profiles in SH-SY5Y cells.

Collectively, these findings demonstrate that sEVs exert regulatory effects on inflammation-associated gene expression in an Aβ-induced neuroblastoma cell model. Moreover, miR-137-5p-loaded sEVs exhibit superior anti-inflammatory and neuroprotective gene-modulating effects compared with unloaded sEVs, highlighting their enhanced therapeutic potential in vitro.

3.6. Regulation of NfL Levels in Aβ-Exposed SH-SY5Y Cells by sEV-Mediated miR-137-5p Delivery

Intracellular levels of NfL, a cytoskeletal protein expressed in neuronal lineage cells, were quantified in SH-SY5Y cells. Exposure to amyloid-β (SH-SY5Y–Aβ group) resulted in a marked and statistically significant increase in intracellular NfL levels compared with untreated SH-SY5Y control cells (Figure 6; p < 0.001), indicating cytoskeletal stress and neuronal injury-like alterations at the cellular level.

Treatment of Aβ-exposed SH-SY5Y cells with milk-derived sEVs did not significantly attenuate NfL expression, and intracellular NfL levels remained significantly elevated compared with control cells (p < 0.001). In contrast, treatment with miR-137-5p-loaded sEVs (sEV_miR-137-5p) markedly reduced intracellular NfL levels, restoring them to values comparable to those observed in untreated control cells (p > 0.05).

3.7. sEV_miR-137-5p Regulating Oxidative Stress and Mitochondrial-Related Markers in SH-SY5Y Cells Induced by Aβ

Levels of oxidative stress and mitochondria-related markers were markedly altered in SH-SY5Y neuroblastoma cells following amyloid-β exposure (Figure 7). Compared with the control group, amyloid-β-treated cells exhibited a significant increase in Cyt-c (Figure 7a), 8-OHdG (Figure 7b), PINK1 (Figure 7d), and DNM1L (Figure 7e) levels (p < 0.001 for all markers), accompanied by a significant decrease in TFAM levels (Figure 7c), indicating enhanced mitochondrial stress, oxidative DNA damage, and dysregulated mitochondrial dynamics.

When compared with the control group, treatment with blank sEVs significantly modulated these parameters, partially reducing Cyt-c, 8-OHdG, PINK1, and DNM1L levels and partially restoring TFAM expression (p < 0.01 vs. control), suggesting a limited but detectable protective effect in this in vitro neuroblastoma model.

In contrast, treatment with miR-137-5p-loaded sEVs (sEV_miR-137-5p) resulted in a more pronounced normalization of mitochondrial and oxidative stress-related markers. Compared with the control group, no significant differences were observed in Cyt-c, 8-OHdG, PINK1, DNM1L, or TFAM levels in the sEV_miR-137-5p-treated cells (p > 0.05), indicating that miR-137-5p-enriched sEVs effectively restored these parameters toward baseline control levels. Collectively, these findings demonstrate that miR-137-5p loading substantially enhances the capacity of milk-derived sEVs to counteract Aβ-induced mitochondrial and oxidative stress alterations in SH-SY5Y neuroblastoma cells.

3.8. sEV_miR-137-5p Regulating Levels of Synaptic and Extracellular Matrix-Associated Proteins in SH-SY5Y Cells Induced by Amyloid-β

Amyloid-β exposure significantly altered the expression of synaptic and extracellular matrix-associated proteins in SH-SY5Y neuroblastoma cells (Figure 8). Compared with the untreated control group, SH-SY5Y-Aβ cells exhibited a marked decrease in CPLX2 (Figure 8a) and SMOC1 (Figure 8c) levels, accompanied by a significant increase in ROR1 (Figure 8b) concentrations (all p < 0.001), indicating pronounced synaptic and extracellular matrix-related disturbances.

Treatment with sEVs partially modulated these alterations when compared with the control group; however, significant differences in CPLX2, SMOC1, and ROR1 levels persisted (p < 0.01 vs. control), suggesting a limited restorative effect in this in vitro neuroblastoma model.

In contrast, treatment with miR-137-5p-enriched sEVs (sEV_miR-137-5p) resulted in a substantial normalization of CPLX2, SMOC1, and ROR1 levels. Notably, no statistically significant differences were observed between the sEV_miR-137-5p group and the control group for any of these markers (p > 0.05), indicating near-complete restoration toward baseline levels. Collectively, these findings demonstrate that miR-137-5p loading markedly enhances the functional efficacy of sEVs in counteracting amyloid-β-induced alterations in synaptic and extracellular matrix-associated protein expression in SH-SY5Y cells.

3.9. sEV_miR-137-5p Modulates Tau Pathology and Amyloid-β Levels in Aβ-Treated SH-SY5Y Cells

Amyloid-β exposure significantly elevated key AD-related protein biomarkers in SH-SY5Y neuroblastoma cells (Figure 9). Compared with untreated control cells, SH-SY5Y-Aβ cells exhibited marked increases in total tau (Figure 9a), pTau-181 (Figure 9b), pTau-217 (Figure 9c), and amyloid-β_1–40_ (Figure 9d) levels (all p < 0.001).

Treatment with sEVs resulted in a modest but statistically significant attenuation of these biomarkers relative to the amyloid-β-treated group; however, biomarker levels in the sEV-treated cells remained significantly different from those observed in the control group (p < 0.01), indicating a limited capacity of sEVs to fully normalize amyloid-β-induced molecular alterations in this neuroblastoma cell model.

In contrast, cells treated with miR-137-5p-loaded sEVs exhibited levels of total tau, pTau-181, pTau-217, and Aβ_1–40_ that were not significantly different from those observed in control cells (p > 0.05). These findings indicate that miR-137-5p enrichment enhances the capacity of sEVs to normalize amyloid-β-associated tau and amyloid-related protein alterations at the cellular level.

4. Discussion

In the present study, the effects of milk-derived sEVs, either unloaded or loaded with miR-137-5p, were evaluated in an amyloid-β (Aβ)-induced SH-SY5Y neuroblastoma cell model. The results indicate that Aβ exposure was associated with the development of a neurodegeneration-like cellular profile, including oxidative stress, inflammatory activation, cytoskeletal alterations, mitochondrial dysfunction, synaptic-related changes, and tau-associated protein abnormalities. While treatment with unloaded sEVs was associated with partial modulation of several parameters, miR-137-5p-loaded sEVs were consistently associated with a closer normalization of multiple molecular endpoints toward control levels.

Aβ exposure was accompanied by a marked imbalance between inflammation-related and neuroprotective gene expression, as reflected by decreased BDNF mRNA levels and increased ICAM1 and TNF-α expression. This expression pattern is consistent with previous reports indicating that Aβ exposure is associated with suppression of neurotrophic support and enhancement of inflammatory signaling in neuronal cell models [26,27,28]. BDNF has been widely recognized as a key mediator of neuronal survival and synaptic plasticity [29], whereas ICAM1 and TNF-α are commonly used indicators of inflammatory activation in neurodegenerative contexts [30,31].

Treatment with unloaded sEVs was associated with a partial reduction in ICAM1 and TNF-α expression and a modest increase in BDNF levels; however, these changes did not fully restore gene expression to control values. In contrast, treatment with miR-137-5p-loaded sEVs was associated with expression levels of BDNF, ICAM1, and TNF-α that were not statistically different from those of untreated control cells. These findings suggest that miR-137-5p loading may enhance the regulatory capacity of sEVs on inflammation-related and neuroprotective gene expression, although direct molecular targets were not examined in the present study.

Intracellular NfL levels were significantly increased following Aβ exposure, indicating cytoskeletal stress and neuronal injury-like alterations at the cellular level. Such changes are in line with previous studies in which increased NfL concentration has been associated with axonal damage and neurodegenerative processes [32,33]. Unloaded sEV treatment was not associated with a significant reduction in intracellular NfL levels, whereas miR-137-5p-loaded sEV treatment was associated with NfL levels comparable to those observed in control cells. This observation suggests that miR-137-5p delivery via sEVs may be linked to improved maintenance of cytoskeletal homeostasis under Aβ-induced stress conditions.

Consistent with the central role of mitochondrial dysfunction and oxidative stress in Aβ-related neurotoxicity [34], Aβ exposure was associated with increased oxidative damage markers and altered expression of mitochondrial stress-related proteins, including Cyt-c, 8-OHdG, PINK1, DNM1L, and TFAM. Treatment with unloaded sEVs resulted in partial modulation of these parameters, whereas miR-137-5p-loaded sEVs were associated with values that did not differ significantly from control levels. These findings indicate that miR-137-5p-enriched sEVs may be associated with improved mitochondrial homeostasis and reduced oxidative burden, although the underlying regulatory mechanisms remain to be elucidated.

Alterations in synaptic and extracellular matrix-related markers, including CPLX2, SMOC1, and ROR1, were also observed following Aβ exposure, supporting the presence of synaptic dysfunction and cytoskeletal instability. While unloaded sEVs were associated with limited normalization of these markers, miR-137-5p-loaded sEVs were associated with expression patterns comparable to those of control cells. Given the importance of synaptic integrity in neurodegenerative disease progression, these findings suggest a potential association between miR-137-5p delivery and the preservation of synaptic-related molecular features.

In addition, Aβ exposure resulted in increased levels of total tau, phosphorylated tau (pTau-181 and pTau-217), and Aβ1–40, confirming the induction of AD-like protein alterations in SH-SY5Y cells. Unloaded sEV treatment was associated with partial reductions in these protein levels, whereas miR-137-5p-loaded sEVs were associated with levels comparable to those observed in control cells. Although these results suggest a regulatory association between miR-137-5p delivery and tau- and Aβ-related pathways, direct mechanistic interactions were not investigated and therefore cannot be conclusively established.

The partial modulation of several molecular parameters observed following treatment with unloaded milk-derived sEVs is consistent with previous reports indicating that these vesicles possess intrinsic bioactive properties [35]. Moreover, milk-derived sEVs carry endogenous proteins, lipids, and regulatory RNAs that may contribute to neuroprotective effects by attenuating oxidative stress, including ROS accumulation, and suppressing pro-inflammatory cytokine signaling [36,37]. Such inherent vesicular components may therefore contribute to baseline antioxidant and anti-inflammatory effects independent of the exogenously loaded miRNA cargo. In this context, the more pronounced normalization observed with miR-137-5p-loaded sEVs likely reflects a synergistic effect between the intrinsic bioactivity of milk-derived sEVs and the additional regulatory capacity conferred by miR-137-5p.

Although the present study demonstrates broad normalization of oxidative, inflammatory, mitochondrial, cytoskeletal, synaptic, and tau-related markers following sEV-mediated miR-137-5p delivery, the precise intracellular targets responsible for these effects were not directly interrogated. Previous studies have identified several experimentally validated targets of miR-137-5p that are mechanistically relevant to the observed molecular outcomes. In particular, miR-137-5p has been reported to suppress TNFAIP1, leading to attenuation of NF-κB-dependent inflammatory signaling and reduced amyloid-β-induced neurotoxicity, and regulate USP30, a mitochondrial deubiquitinase involved in mitochondrial quality control and mitophagy regulation [17]. Modulation of these pathways has been associated with improved mitochondrial integrity, reduced oxidative stress, and enhanced neuronal resilience in AD models [38]. Consistent with these reports, the present findings suggest that miR-137-5p delivery via milk-derived sEVs may exert coordinated multi-pathway regulatory effects; however, the observed molecular changes should be interpreted as associative rather than direct evidence of specific target engagement.

A key limitation of the present study is the absence of direct target validation experiments for miR-137-5p. Luciferase reporter assays, transcriptomic or proteomic profiling, and gain- or loss-of-function approaches were not performed to confirm specific miRNA–mRNA interactions in the current experimental setting. Accordingly, while prior literature supports several biologically plausible intracellular targets of miR-137-5p, definitive conclusions regarding direct target regulation cannot be drawn from the present data.

In addition, cellular uptake dynamics and subcellular trafficking of sEVs and their miRNA cargo were not directly visualized. Nevertheless, the consistent modulation of multiple AD-related molecular endpoints supports effective intracellular delivery, and future imaging-based studies may further clarify the underlying mechanisms.

Moreover, experiments were conducted using undifferentiated SH-SY5Y neuroblastoma cells, a widely used in vitro neuronal model for investigating amyloid-β-induced molecular stress responses. While this system enables controlled mechanistic assessment of neuronal signaling pathways, it does not capture the cellular heterogeneity of the central nervous system. In particular, astrocytic and microglial contributions to neuroinflammatory and neuroimmune processes, EV-mediated intercellular communication, and amyloid-β clearance could not be addressed. Therefore, the observed protective effects of sEV-mediated miR-137-5p delivery should be interpreted primarily as neuron-centered responses within a simplified experimental framework.

Finally, the present study focused on molecular and biochemical alterations in an acute amyloid-β-induced in vitro model, which provides mechanistic insight into early disease-related cellular responses. However, extrapolation of these findings to more physiologically relevant systems remains speculative. Therefore, validation in differentiated neuronal models, neuron–glia co-culture systems, and in vivo studies, including assessments of functional outcomes and prolonged exposure paradigms, will be essential for translational interpretation. Collectively, while the findings support sEV-mediated delivery of miR-137-5p as a promising multi-target experimental approach, further confirmation in multicellular and in vivo models is required before clinical implications can be considered.

5. Conclusions

In conclusion, the present study demonstrates that milk-derived sEVs loaded with miR-137-5p are associated with broad normalization of oxidative stress, inflammatory responses, mitochondrial dysfunction, cytoskeletal injury, synaptic alterations, and tau-related abnormalities in an amyloid-β-induced SH-SY5Y cell model. While unloaded sEVs exerted partial protective effects, miR-137-5p loading consistently enhanced these responses, supporting a potential multi-pathway regulatory role. Although direct target validation was beyond the scope of this study, the findings are consistent with previous reports implicating miR-137-5p in mitochondrial and inflammatory signaling pathways relevant to Alzheimer’s disease. Taken together, these results provide proof-of-concept evidence supporting sEV-mediated miR-137-5p delivery as an experimental strategy warranting further validation in physiologically relevant and in vivo models.