Potential neuroprotective effect of stem cells from apical papilla derived extracellular vesicles enriched by lab-on-chip approach during retinal degeneration

Retinal degeneration (RD) is recognized as a frequent cause of visual impairments, including inherited (Retinitis pigmentosa) and degenerative (age-related macular) eye diseases. Dental stem cells (DSCs) have recently demonstrated a promising neuroprotection potential for ocular diseases through a paracrine manner carried out by extracellular vesicles (EVs). However, effective isolation of EVs is still challenging, and isolation methods determine the composition of enriched EVs and the subsequent biological and functional effects. In the present study, we assessed two enrichment methods (micro-electromechanical systems and ultrafiltration) to isolate the EVs from stem cells from apical papilla (SCAP). The size distribution of the corresponding isolates exhibited the capability of each method to enrich different subsets of EVs, which significantly impacts their biological and functional effects. We confirmed the neuroprotection and anti-inflammatory capacity of the SCAP-EVs in vitro. Further experiments revealed the possible therapeutic effects of subretinal injection of SCAP-EVs in the Royal College of Surgeons (RCS) rat model. We found that EVs enriched by the micro-electromechanical-based device (MEMS-EVs) preserved visual function, reduced retinal cell apoptosis, and prevented thinning of the outer nuclear layer (ONL). Interestingly, the effect of MEMS-EVs was extended to the retinal ganglion cell/retinal nerve fiber layer (GCL/RNFL). This study supports the use of the microfluidics approach to enrich valuable subsets of EVs, together with the choice of SCAP as a source to derive EVs for cell-free therapy of RD.


Background
A large body of evidence suggests the beneficial effects of mesenchymal stem cell (MSC) mediated therapies on retinal degeneration (RD) [1,2]. Dental stem cells (DSCs), originally derived from neural crest cells, appear to involve in nervous system development and mediate neuroprotection [3] that may be attributed to the neurotrophic factors (such as nerve growth and brain-derived neurotrophic factors, NGF and BDNF, respectively) released by these types of cells [4,5]. Interestingly, DSCs promote paracrine-mediated neuroprotection significantly greater than human bone marrow and adipose-derived mesenchymal stem cells [6], which reportedly preserves retinal ganglion cells [7]. Stem cells from apical papilla (SCAP) originating from the cranial neural crest and found in the root apex of immature permanent teeth [8] actively contribute to their environment by producing immunosuppressive, anti-inflammatory, and trophic factors [9]. As a traditional "biowaste" cell source, SCAP is a novel source of MSCs representing a stromal cell-derived inducing activity (SDIA) to retinal cells differentiation [10] and can be considered a promising candidate for retinal regeneration [11].
Despite the beneficial effects of cell-based therapy in regenerative medicine, undesired differentiation or malignant transformation remain concerns for clinical application [12]. Profound trophic effects of MSCs on the retina are widely accepted, and this effect is primarily mediated through paracrine signaling of the extracellular vesicles (EVs) carrying soluble retina-regulatory molecules [13,14]. EVs are small membrane-enclosed delivery vehicles released by most live cells and play a vital role in mediating intracellular communication through the exchange of their cargo [15]. EVs are considered a heterogeneous group of particles typically categorized as exosomes (40-100 nm in diameter) and microvesicles (MVs: 0.1-1 µm in diameter) [16]. According to the MISEV2018 guideline, the authors were urged to refer EV subtypes based on their size ("small EVs" (sEVs) [< 200 nm] and "medium/large EVs" (m/lEVs) [> 200 nm]) instead of terms such as exosomes and MVs [17]. Emerging studies have demonstrated the therapeutic effects of stem cell-derived sEVs (mostly exosomes) on ocular diseases. In particular, subretinally injected human neural progenitor cell-derived sEVs promote photoreceptor regeneration in the Royal College of Surgeons (RCS) rats through inactivating microglia [18]. Photoreceptor protection has also been achieved after intravitreal delivery of MSC-sEVs, indicating exosomal miR-21 as a therapeutic for N-methyl-N-nitrosourea-induced retinal injury [14]. Further studies have uncovered the survival of RGCs using human embryonic [19] and bone marrow [20] stem cell-derived EVs in the optic nerve crush model.
The content, quality, and quantity of the EVs influence their potential bioactivity. For example, a recent study has shown that the isolation method can determine EVs' biological and functional effects at single and sub-population levels, which correlate with their composition [21]. Conventional methods of EV "enrichment," which mostly rely on concentrating, not isolating, face significant limitations, as they are often time-consuming (4-5 h), low throughput (5-25% exosome recovery), and expensive [22][23][24].
Micro-electromechanical (MEMS) microfluidics, a miniaturized machine that combines mechanical and electromechanical elements onto a single device, offer a rapid, efficient, and sensitive method for bio-particle manipulation [25,26]. Alternating current (AC) electrokinetic driven chips have been investigated for EVs capturing, where electrokinetic and hydrodynamic forces work together to relocate the nanoparticles around the electrodes [27][28][29]. This approach offers label-free isolation and detection method in which small EV enrichment occurs based on their intrinsic properties such as size, shape, and dielectric properties [30,31]. The isolation of a pure population of EVs is crucial for reliably studying physiological function and biological activity. The inconsistency between studies might be one of the reasons behind the fact that very few EV-based therapeutics have progressed to clinical use. This is especially important for ocular therapeutic strategies in which a small and pure EV population is needed. Specifically, subretinal injection offers a relatively precise and direct effect on targeting cells in the subretinal space that requires a lower dose (0.1-3 μl in animal studies) of therapeutic compared to intravitreal delivery (2-20 μl in rats) [32][33][34]. Microfluidics can provide an excellent opportunity to prepare the pure EV solution for subretinal injection.
This study employed a MEMS device to enrich SCAPderived small EVs injected subretinally against retinal degeneration in RCS rats. Both in vivo and in vitro confirmed potential neuroprotective effects of the MEMS enriched SCAP-EVs. It revealed that MEMS enriched SCAP-EVs inhibited apoptotic and inflammatory signal pathways in the retina by targeting Bax, Bcl-2, Il-10, and Il-6. Collectively, elucidating the potential effects of SCAP-EVs would provide promise for establishing novel therapeutics of visual loss in RD disorders.

Animals
RCS rats, regardless of sex, were provided by the Royan institute and were raised in the specific pathogen-free room of the Animal Care Center of Royan Institute, at temperatures between 20 and 25 °C with regular daylight exposure and free access to food and water. Animal care and experimental procedures were performed in accordance with the Helsinki Declaration and compiled with the ARRIVE guideline (https:// arriv eguid elines. org/) and were approved by the institutional research ethics committee at Royan institute (IR.ACECR.ROYAN.REC.1400.043). At least ten animals were used for each group.
Cell culture and conditioned medium preparation PC12, a rat pheochromocytoma cell line, was cultured with Dulbecco's modified Eagle's medium (DMEM), 10% foetal bovine serum (FBS), 5% Horse serum, 1% L-Glutamine, 1% Nonessential amino acids (NEAA), and 1% penicillin/ streptomycin (all from Gibco, Paisley, UK). PC12 cells were seeded in a 6-well plate at a density of 80 × 10 4 cells/ cm 2 and cultured for 24 h. Four experimental groups were defined. The H2O2 exposure group, to induce oxidative stress, was administrated with 250 μM H2O2 alone for 12 h. Two protective groups were treated with 60% SCAP-EVs (by value), enriched by ultrafiltration method (UF-EVs) and MEMS device (MEMS-EVs), 24 h before the administration of 250 μM H2O2. SCAP-EVs control group was only treated with 60% serum-reduced basal medium. To evaluate cell numbers, we washed PC12 cells with PBS, fixed in 4% PFA for 15 min at ambient temperature, stained with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma, Munich, Germany), and scanned by the fluorescence microscope. For each group, cell counts were obtained from data acquisition of the 5-10 random low magnification fields of 2-3 wells using Image J software (Version 1.42q, NIH, USA).
The SCAP was acquired by Royan institute [10] and grown in standard culture conditions supplemented with DMEM, 10% FBS, 1% L-Glutamine, 1% NEAA, and 1% penicillin/streptomycin until they reached a confluency of 70-80%. The cells were then washed three times with PBS − , dissociated using trypsin-EDTA, and 5 × 10 4 cell/ cm 2 were plated in T150 flasks. One day later, the culture medium was replaced with a serum-free medium for 24 h. The collected supernatant was filtered with a 0.45-μm filter (Millipore) to discard cell debris.

EVs enrichment
As previously described, the collected supernatant was loaded into the MEMS device and subsequently enriched [35]. Briefly, a 400 μL of a freshly filtered conditioned medium was placed on the chip and isolated for 20 min at 10 kHz, and the peak-peak voltage 5 V (Vpp). The MEMS device is composed of a MEMS layer and a polydimethylsiloxane (PDMS) chamber. The MEMS layer, which comprises two sets of interdigitated electrode arrays, connects to a function generator and produces AC electrokinetic force fields. Once the EV capture was completed, the chip was washed gently with a low-conductive buffer (85 g/l sucrose in DI water) three times to remove bulk proteins and nucleic acids. The isolated EVs were subsequently released and collected from the chip surface by elution with the basal medium.
Accordingly, a freshly filtered supernatant was added to the Amicon ultra-0.5 (Millipore, 3000 MWCO) filter and centrifuged at 14,000×g for 20 min. Concentrated samples were retrieved by upside-down centrifugation at 2000×g for 2 min. All centrifugation steps were performed at 4 °C. The concentration factor was kept consistent for both enrichment methods.

Protein content measurements
The conditioned medium was lysed by adding an equal amount of Radioimmunoprecipitation (RIPA) buffer and then incubated at 4 °C for 30 min, followed by 30 s of sonication in an ice-cold bath. Protease and phosphate inhibitors (Sigma Aldrich, Missouri, USA) were added to the lysis buffer at 1:10 concentration. Following the lysis with RIPA, total protein contents were measured using the BCA protein assay kit (KIAZIST, Kermanshah, Iran). Briefly, BSA standard or samples (5 μl) were transferred to PCR tubes to which 5 μl working reagent was added (working reagent 50:1 ratio of assay reagents A and B). The tubes were incubated for 30 min at 37 °C before being analyzed with Nanodrop (NANODROP 2000c, Thermo Scientific, USA).

Transmission electron microscopy (TEM)
Six microliters of conditioned medium were pipetted onto a carbon-coated grid, counterstained using 1% uranyl acetate, briefly washed with distilled water, and allowed to settle and dry for one hour at ambient temperature. To visualize the morphological characteristic of the cells releasing EVs, SCAP was fixed with 2.5% glutaraldehyde at 37 °C for 20 min. Fixed cells were then embedded in resin, sectioned using an ultra-microtome, and stained with 1% toluidine blue for visualization using light microscopy. The ultrathin sections were double-stained with 1% uranyl acetate and lead citrate. The grids were analyzed with a Philips EM208S electron microscope (FEI, The Netherlands) operating at 100 kV.

Dynamic light scattering (DLS)
The DLS experiment evaluated EVs size distribution using a Nanoparticle sizer SZ-100 (HORIBA Scientific) performing at a 90° angle.

MTS assay
The viability and metabolic activity of PC12 cells following SCAP-EVs treatment were carried out using the MTS assay by applying cell titer 96 aqueous one solution (Promega, WI, USA) according to the manufacturer's instruction. Briefly, PC12 wells from each group were incubated with MTS/PMS solution for 3.5 h. Therefore, the absorbance of the resulting colored solution was measured at 490 nm by exploiting a micro-plate reader (AWARNESS Technology, Inc., USA) and normalized to the cell-free solution.

Subretinal transplantation
All animals were divided into three groups with ten individuals at least. The animals were anesthetized by intraperitoneal injection of ketamine (50 mg/kg) and xylazine (5 mg/kg). The right eyes were enrolled in this study. Subretinal transplantations were performed in RCS rats (postnatal 21 days) with UF-EVs, MEMS-EVs, or unconditioned basal medium (vehicle). The retinas that did not receive any injections were the untreated control group. The right eye of RCS rats was injected with a 2 μL concentrated SCAP-EVs resuspended solution that contained 4 μg protein, which is approximately equivalent to 1.76 ± 0.2 e 09 sEVs [36]. The subretinal transplantation method was adapted from Ben M'Barek et al. [37]. Briefly, a local subretinal detachment was created through air injection via an insulin syringe under an operating microscope (OMS-90) following a small sclera incision. Generating a hole into the cornea reduced the intraocular pressure. After the subretinal injection of the solution, the syringe was held in place for 30 s to prevent any leakage. The animals exhibiting any complications such as massive intraocular, retinal hemorrhage, or endophthalmitis after surgery were excluded from the study. Although EVs require a milder retinal detachment due to their small size compared to cell transplantation therapies, postoperative H&E examinations showed that the retinal detachment was completely closed in the evaluation period used in this study. All RCS rats, including untreated groups, received oral cyclosporine (Novartis Pharma AG, Basle, Switzerland) dissolved in their drinking water 24 h before transplantation.

Tunel staining
Apoptotic cells of the retina were detected using in situ cell death detection kit (DeadEnd ™ Fluorometric Tunel System kit, Promega, Madison, WI) and PI counterstain (Sigma-Aldrich, USA) according to the manufacturer's protocol. All slides were visualized under a fluorescent microscope (Olympus, Center Valley, PA, USA) equipped with an Olympus DP70 camera.

Histological assessment
RCS rats were euthanized on day 28 after transplantation by carbon dioxide inhalation, and the right eyes were enucleated for further analysis. These eyecups were fixed in 4% PFA overnight at 4 °C. Sections (4 μm thickness) were cut from tissue embedded in paraffin wax, attached to glass slides perpendicular to the optic disc, and stained with hematoxylin and eosin (H&E) for light microscopy. For all eyes, the sections were obtained at the same distance (1600 μm) from the iris. Z-projected thickness and nucleus count of retinal layers were analyzed by Image J. The retinal neovascularization was evaluated by H&E staining the RCS rat eyes three months after transplantation. The average number of intravitreal vessels per section was determined by counting the vascular lumens on the vitreous side of the inner limiting membrane. Three independent experiments were considered for all corresponding groups (UF-EVs, MEMS-EVs, vehicle, and control). The untreated left eyes were assigned as the control group.

Immunofluorescence staining (IF)
Heat-induced antigen retrieval was first carried out by autoclaving sections in citrate buffer (sodium citrate-citric acid) at pH 6. All slides were then air-dried, and washed in PBS, and the cell membranes were permeabilized by 0.4% Triton X-100 for 20 min. Then, the sections were incubated with primary antibodies (diluted into 10% normal goat serum) at 4 °C overnight. Following two wash steps with PBS, the sections were incubated with secondary antibodies (diluted into 5% normal goat serum) at 37 °C for one hour, counterstained with DAPI or PI, and scanned and photographed via fluorescence microscopy. The antibodies that have been used in our study are as follows: Anti-CD81 (

Flow cytometry
For identification, enriched EVs were incubated with CD81 antibody (diluted into 0.2% (wt/vol) BSA solution) for 1 h at 37 °C in the dark and mixed every 10 min. A parallel sample with an isotype control antibody was incubated under the same condition to control the nonspecific binding. The gates were positioned and calibrated using 300-nm beads. To preclude the noise derived from the buffer, optics, and electronics interfering in light scattering results of nanosized vesicles, we adapted a fluorescence threshold triggering protocol from Vlist et al. [38].

Reverse-transcription quantitative polymerase chain reaction (RT-qPCR) analysis
The retinal layer of the enucleated eyes was isolated mechanically under dissecting stereomicroscope. Total RNA of the PC12 cells and isolated retina was extracted using TRIzol reagent per the manufacturer's instructions. Biotechrabbit ™ Kit (Cat. No. BR0400403, Germany) was used to transcribe RNA to cDNA per the manufacturer's instructions. Expression levels of the apoptotic and inflammatory mediated genes were measured with RT-qPCR. Normalizing to GAPDH levels, all expression values were calculated using the comparative threshold cycle method (ΔΔCT). The mean and standard deviations were computed from four independent experiments. The sequences of primer pairs that are used in this study are as follow: Bcl-2:

Western blot
Whole lysates of EVs were prepared using RIPA lysis buffer. Proteins were separated and loaded on 12% SDS-PAGE gel. After being transferred to a PVDF membrane (Millipore, USA), the samples were blocked with 5% skimmed milk in PBS with 0.1% Tween for 2 h at room temperature and incubated with primary antibody overnight at 4 °C. The membranes were then incubated with HRP-conjugated secondary antibody for 1 h at ambient temperature.

Visual cliff avoidance test
Visual function of the RCS rats on day 28 post-operation was evaluated using a depth perception visual cliff, as previously described [39]. In brief, rats were placed on a stage at the center of a Perspex platform, creating a cliff appearance. Animal's activity was recorded for 2 min with a phone camera, and the latency to dismount the stage (in seconds) was analyzed. Ten individuals per group (30 total) were tested once (for each rat) to avoid the memory effect.

Statistical analysis
All data were presented as the mean ± SE (standard error of the mean) from at least three biological samples for each test and ten for each group. Statistical differences were calculated using one-way analysis of variance (ANOVA) for comparison between two groups and Dunnett or Sidak tests to compare multiple groups. P values < 0.05 were considered to be significant.

SCAP releases EVs into the extracellular environment
Cells constitutively release EVs during cell-cell communication [16]. In this study, the SCAP, with a homogenous spindle-shaped morphology (Fig. 1a), was used as the parent cells to drive EVs. Figure 1b-d reveal the secretion and ultrastructural morphology of SCAP-derived EVs using TEM. Electron micrographs show EVs biogenesis that involves fusing multivesicular bodies (MVB) (Fig. 1b) with the plasma membrane and subsequent release of a heterogeneous population of membrane-bilayered vesicles of different sizes (Fig. 1c, d). MVB can either fuse with lysosomes to degrade their cargo or with the plasma membrane to release their cargo extracellularly (Fig. S1). To enrich EVs, we first collected SCAP conditioned medium according to a standard workflow (Fig. 1e). We further examined EVs' presence in the collected conditioned medium using TEM, flow cytometry, and Western blot. TEM showed that SCAP-derived EVs exhibited typical cup-shaped vesicles with diameters ranging from 50 to 300 nm (Fig. 1f, insert box). Phenotype characterization by flow cytometry and Western blot analysis confirmed the expression of the CD81, a feature biomarker of EVs (Fig. 1g, h).

Enrichment method determines the functional effect of EVs
To investigate the functionality of EVs, we employed two enrichment methods, including ultrafiltration (UF-EVs) and MEMS microfluidics (MEMS-EVs). The isolation principles are based on size exclusion forced by AC electrokinetic relocation for the MEMS method and sedimentation pressure for the UF method, as demonstrated in Fig. 2a and b, respectively. Our previous work found that the proposed MEMS device captured sEVs at 10 kHz frequency and 5Vpp [35]. We analyzed the enrichment capacity of the MEMS device at two frequencies (10 and 100 kHz) and three isolation periods (10, 20, and 40 min) by BCA protein assay. For each column, the protein recovery efficiency (%) was measured as follows: To recover the concentrate once the EV capture was completed, the MEMS device was washed with 300 μL of elution buffer (85 g/l sucrose in DI water) to remove bulk proteins and nucleic acids, and then the isolated EVs were repelled and collected from the chip surface by elution with the basal medium. The highest EVs' yield was obtained after 20 min isolation at 10 kHz frequency (Fig. 2c). To assess the size distribution of the enriched population from MEMS and UF methods, we compared the number-weighted size distribution and z-average (intensity-weighted) mean diameter of enriched EVs from both ways (Fig. 2d, e). The size analysis by DLS showed that isolates produced by ultrafiltration Proteinre covery efficiency(%) = (protein concentration after enrichment∕ protein concentration before enrichment) × 100. Fig. 2 The principles of EVs enrichment approaches. a MEMS-based EVs collection: briefly, particles that have similar size, shape, and intrinsic dielectric properties trap around the edge of the electrodes and along its surface under the influence of AC electrokinetic forces. Entrapment occurred at a specified 10 kHz AC frequency and 5 Vpp. b Ultrafiltration-based EVs collection: sedimentation pressure forced particles to pass through a filter with a particular pore size. In this study, a regenerated cellulose filter with a 3KD cut-off was employed. c Protein recovery efficiency (%) [(Protein concentration after enrich-ment/Protein concentration before enrichment) × 100] of the MEMS method, d Number-weighted size distribution of isolates from each isolation method, e Pooled data of Z-average showing mean particle diameter of isolates from each extraction method. Collected supernatant is considered as the control. Compared to the UF-EVs and control, the Mean diameters of MEMS-EVs indicate a significant statistical difference (*P < 0.05, **P < 0.01). Error bars represent the mean ± SE. MEMS micro-electromechanical systems, Vpp voltage peak to peak, AC alternative current, UF ultrafiltration contained larger vesicles (an average of 267 ± 21.2 nm) with somewhat similar protein recovery (72% for UF method compared to 74% for MEMS method) evaluated by BCA protein assay. In comparison, isolates produced by the MEMS method contained smaller vesicles (an average of 113.2 ± 18.4 nm). Collected supernatant considered as the control contained a distribution of nano-sized vesicles with an average of 310 ± 9.69 nm. The number-weighted size distribution of UF-EVs shows a peak at a higher diameter than that in the MEMS-EVs. Since the number-weighted size measurement calculates the most frequent particle size in the sample [40], this shows the presence of smaller vesicles in the enriched population by the MEMS method. This is due to the strong electrokinetic forces acting on the smallest EVs (< 200 nm) in the micro-device [35]. To further demonstrate the size-based separation of sEVs by the MEMS method, size distribution of enriched EVs was analysed by TEM and Image J (Fig. S2). The average size of EVs observed in 24 captured images by TEM was 93.4 ± 4.6 nm; most vesicles were between 50 and 90 nm in diameter, with a peak around 70-90 nm. The difference between mean diameter results from DLS and TEM methods is ascribed to the presence of dispersant that induces error in the DLS particle sizing measurements and shifts its results to the higher values [41].
Flow cytometry of isolated EVs showed a higher level of CD81 expression in MEMS-EVs compared to UF-EVs, which may indicate a higher EVs recovery, by MEMS method compared to UF method (Fig. S3).

Functional inhibition of apoptosis and inflammation signal pathways by SCAP-EVs protect PC12 cells exposed to H2O2 in vitro
To investigate the neuroprotection effect of SCAP-EVs, we employed H2O2-induced oxidative stress in cultured PC12 cells as an in vitro model and evaluated whether pretreatment of PC12 cells with SCAP-EVs enriched by both methods could inhibit the cytotoxic effects of H2O2 (Fig. 3a). Serum content was reduced to replace it with SCAP-EVs. The viability changes were normalized as percentages of the control, unexposed cells and plotted against various concentrations. When serum starvation was performed in the presence of SCAP-EVs, the apoptosis nuclei were clear, and viability was negatively affected, suggesting that SCAP-EVs did not protect PC12 during 24-h serum starvation as discussed earlier by Heino et al. [42]. While, there was no significant difference in the viability of PC12 cells between the control group (100% serum content by value) and 40, 50, and 60% serum reduced groups (Fig. S4a). Therefore, 60% of the serum content was replaced with enriched SCAP-EVs groups. The dose-response curve of H2O2 exposure after 12 h was calculated with an IC50 (the half-maximal inhibitory concentration) of ~ 250 μM (Fig. S4b).
Metabolic activity of the PC12 cells cultured in 60% serum reduced medium was not changed significantly compared to the control group, implying that cell viability was not affected by 60% depletion of serum content for 24 h (Fig. 3b). While, the viability of PC12 cells in the 250 μM H2O2 exposure group was significantly lower than that in the serum-reduced control group. Furthermore, the reduced viability of PC12 cells exposed to H2O2 was recovered to about 13%, and 66% in pretreated groups with UF and MEMS enriched SCAP-EVs, respectively. To confirm the protective effect, we assessed the cell numbers using DAPI staining and data analysis. Cell count quantification revealed a more significant number of the cells in the MEMS-EVs group than in the controls and UF-EVs (Fig. 3c, d).
SCAP possesses a strong immunomodulatory capacity accompanied by the modulation of the inflammatory and apoptosis responses [43][44][45]. The Bcl2 family proteins are known as key regulators of apoptosis cell death [46]. Prosurvival (anti-apoptosis) Bcl2 protein suppresses cell death, and pro-apoptotic Bax protein promotes cell death [47]. On the other way, Il-10 appears to have a vital role in suppressing the inflammatory while Il-6 shows proinflammatory and proatherogenic properties [48]. We further examined the expression of pro/anti-inflammatory (Il-6, Il-10) and pro/anti-apoptosis (Bax, Bcl-2) cytokines under H2O2 and SCAP-EVs treatments at mRNA level with RT-qPCR. H2O2 stimulation markedly decreased Il-10 and Bcl-2 mRNA levels and increased Il-6 and Bax mRNA levels compared to the control group, while MEMS-EVs treatment reversed the effect of H2O2 stimulation (Fig. 3e, f). These results suggest that SCAP-EVs suppress the H2O2-induced cytotoxic effect in vitro. However, in vitro modulatory potential of UF-EVs against the H2O2 stimulation is statistically significant just for Il-6 and Bax cytokines.

SCAP derived EVs modulate inflammation and apoptosis to preserve visual function in RCS rats
To evaluate whether SCAP-EVs exerted therapeutic effects during retinal degeneration in RCS rats and further define the underlying mechanism, we performed in vivo experiments (Fig. 4a). To meet this goal, we injected 2 μL SCAP-EVs enriched by both UF and MEMS methods into the subretinal space of the RCS rats. The 2 μL concentrates approximately contained 4 μg protein according to the BCA protein assay. The left eye served as an untreated group for comparison. We traced the injected EVs labelled with CD81 in the subretinal space of RCS rats. In the retina sections of EVs-treated RCS rats, the labelled EVs were observed as green dots in the subretinal space (Fig. 4a). Visual function assessment using the visual cliff test showed that MEMS-EVs treatment rescued the visual acuity of RCS rats 28 days post-injection, compared with the vehicle and UF-EVs (Fig. 4b). The expression of pro/ anti-inflammatory (Il-6, Il-10) and pro/anti-apoptosis (Bax, Bcl-2) cytokines under enriched SCAP-EVs treatment at mRNA level with RT-qPCR reveal that MEMS-EVs treatment markedly up-regulated the levels of Il-10 and Bcl-2 mRNA expressions and down-regulated the levels of Il-6 and Bax mRNA expressions, compared to the vehicle (Fig. 4c, d).
It has been reported that genetic mutation results in apoptosis of photoreceptors and reduction in thickness of ONL in RCS rats [49]. Tunel staining showed that the MEMS-EVs group significantly reduced the number of apoptotic cells in the ONL of RCS rats, compared with the untreated control, vehicle group, and the UF-EVs treated eyes at days 28 post-injection (Fig. 4e, f), which is consistent with the anti-apoptotic results in Fig. 4c.
In the RCS rat, the retinal pigment epithelium (RPE) cells are not phagocytizing the rod outer segment (OS) debris [49]. Moreover, the OS became disorganized, and the ONL was reduced to a single layer of cells associated with an autofluorescence debris zone [50]. Therefore, photoreceptor regeneration could be one of the mechanisms that mediate the observed protective effects, as previously described [37,51]. To confirm the potential of SCAP-EVs for photoreceptor regeneration, we examined rhodopsin protein expression (Fig. 5). Rhodopsin staining exhibited OS disappearance of rod photoreceptors in untreated control and vehicle, whereas rhodopsin-positive surface was markedly increased in the Metabolic activity (viability) of the PC12 cells that received H2O2 and were under serum reduced condition alone or with SCAP-EVs pretreatment (n = 3 per group). EVs were enriched by MEMS, and UF approaches. Data are presented as mean ± SE. c Cell count of the PC12 cells received H2O2 and was under serum reduction condition alone or with enriched SCAP-EVs by MEMS and UF approaches (n = 3 per group). Data are presented as mean ± SE. d PC12 cells morphology of untreated control and H2O2-induced group alone or with SCAP-EVs pretreatments enriched by MEMS or UF methods. *P < 0.05, **P < 0.01. RT-qPCR analysis showing relative mRNA expression levels of e Apoptosis markers Bax and Bcl-2 and f Inflammatory markers Il-10 and Il-6 compared to the negative control (n = 3 per group). Data are presented as mean ± SE. SCAP stem cells from apical papilla, MEMS micro-electromechanical systems, UF ultrafiltration MEMS-EVs group. Although the UF-EVs group showed an improved effect, the difference is not statistically significant. These results demonstrate that SCAP-EVs that are particularly enriched by the MEMS method preserve the photoreceptors from apoptosis and modulate the inflammatory response to protect the visual function in an RD model.

SCAP derived EVs did not promote proliferation in RCS rats
To visualize the proliferating nuclei post-injection, we performed proliferating cell nuclear antigen (PCNA) staining (white arrowhead). No statistically significant difference was e Apoptosis detection by Tunel (green) and PI staining (red) in RCS retinas (inner nuclear layer, INL, and outer nuclear layer, ONL) of untreated control, vehicle, UF-EVs, and MEMS-EVs enriched groups at day 28 post-injection. Scale bar, 100 μm. f Numbers of apoptosis cells (white arrowhead: Tunel positive cells) in ONL were statistical analysed in untreated control, vehicle, UF-EVs, and MEMS-EVs enriched groups at day 28 post-injection (n = 3 per group) *P < 0.05, **P < 0.01, ***P < 0.001. Data are shown as mean ± SE. SCAP stem cells from apical papilla, MEMS micro-electromechanical systems, UF ultrafiltration observed in the average count of the PCNA-positive nuclei between controls and protective groups. Moreover, we did not find uncontrolled cell proliferation expressed PCNA in the retinal cross-sections of the RCS that had received SCAP-EVs (Fig. S5). PCNA-positive nuclei observed through the retina cell layers might belong to microglia cells [18,52].

SCAP derived EVs preserve retinal morphology in RCS rats
In RCS rats, progressive thinning of retinal layers was detected beginning from week 3 [53]. Retinal nerve fiber layer (RNFL), GCL, and inner plexiform layer (IPL) were thin and difficult to measure separately after week 6. Therefore, the combination of RNFL, GCL, and IPL was measured as ganglion cell complex (GCC). H&E staining revealed that the ONL thickness was preserved significantly in the MEMS-EVs group, compared with that in untreated control, vehicle, and UF-EVs with relatively similar retinal thicknesses (Fig. 6a, b). Interestingly, the GCC thickness, a measure of RGC axonal density, increased in animals receiving MEMS-EVs (Fig. 6c). The numbers of cells in ONL and GCL corresponding layers suggest that cell bodies are also affected during treatment with SCAP-EVs (Fig. 6 d,  e). Previous studies have demonstrated that sEVs play an essential role in retaining photoreceptor degeneration [54] and preventing degenerative thinning of RNFL [55]. The protective effects observed in the MEMS-EVs compared to the UF-EVs are potentially due to the size differences between EV populations enriched by these two methods as MEMS-EVs mostly contained sEVs while UF-EVs are considered as m/lEVs. In addition to the H&E staining for counting the ganglion cell, we examined specific immunostaining for surviving RGCs using both the Brn3a and neuronal-specific nuclear protein (NeuN) markers (Fig. 6f). Evaluation of the Brn3a + RGCs and neurons expressing NeuN in the RCS retina of the untreated control, vehicle, and UF-EVs groups exhibited a loss compared to MEMS-EVs group, which is inline with the H&E quantification results.

SCAP derived EVs did not promote neovascularization in RCS rats
We further assessed the angiogenesis effect of SCAP-EVs three months after transplantation in RCS rats. There is no statistically significant difference among the average number of vascular lumens counted on the vitreous side of the boundary between the retina and the vitreous body in the MEMS-EVs and UF-EVs compared to vehicle and untreated groups (Fig. S6).

Discussion
Transplantation of DSCs is a promising therapeutic approach for neurodegenerative disorders, including retinal degeneration (RD) [56,57]. Of particular importance to this purpose, DSC-EVs could be an effective therapy and potentially avoid the drawbacks of stem cell therapy, including probable immune response and ethical issues. Recent studies have reported diverse biomedical function activity for dental pulp stem cells-derived EVs in animal models, including antiapoptotic, hematopoietic recovery, and bone regeneration capacity [58,59]. Treatment of the RCS rats by EVs derived from human embryonic [60] and neural progenitor cells [18] have proved to be therapeutically efficacious via microRNA exchange. These previous studies possess similarities to our findings and help us better understand the underlying mechanism in RD treatment. The present study aimed to introduce SCAP as a novel source of EVs with substantial neuroprotective potential that is particularly advantageous at the early stages of RD treatment. Another aspect of the novelty of this work lies in the micro-electromechanical microfluidics (MEMS) device that we exploit to enrich the precise population of small EVs. Our previous work characterized and designed the MEMS device capable of enriching sEVs [35]. Briefly, this device offers a label-free and relatively fast method to isolate and enrich particles using AC electrokinetic phenomena. When placed in an AC electric field, a particle will polarize (i.e., develop a dipole moment) dependent on the intrinsic properties of the particle, such as dielectric properties, size, and shape. Therefore MEMS microfluidics approach was utilized to enrich EVs using the AC electrokinetic net force to relocate the EVs laterally within a chamber where the target population is pulled away from the rest of the population. This method could serve as a novel, precise enrichment method, particularly for ocular treatment, as it mainly requires a small and effective dose of therapeutics.
As a heterogeneous group of cell-derived nanoparticles, EVs include several subtypes of similar size and morphology, making it difficult to differentiate. Research is still ongoing to decipher whether and how each subset mediates the diverse biological effects. Here, we discovered that SCAP-EVs, which are enriched by the MEMS method, demonstrate a neuroprotective effect as well as the capacity to alleviate the RD in the RCS model. We further observed that UF-EVs revealed a partial therapeutic capacity in vitro, though the difference between protective effects of MEMS and UF enrichment methods was more prominent in vivo, most likely due to the diverse biological capabilities of EVs subsets. We also found that MEMS-EVs mostly contained sEVs subsets, although both UF and MEMS methods seemed to have a relatively similar protein recovery. Moreover, compressible and deformable particles such as EVs can be pushed through the membrane pores or damaged due to the applied shearing force even if larger than the pore size [61][62][63]. Therefore, the UF or repeated centrifugation methods are associated with partial particle loss, presumably the sEVs subsets. Furthermore, several studies have evaluated the effect of electrokinetic force on isolated particles at different working ranges of frequency and Vpp. Particularly, Zhu et al. [27] have identified and compared nucleic acids and proteins in EVs isolated by electrokinetic force to those enriched by ultracentrifuge method. They revealed no noticeable differences in RNA species between the EVs isolated by electrokinetic force and those isolated by UC. Another contribution has compared EVs from electrophoretic-based isolation and UC. They found higher (14 times) EVs recovery than U/C, so Western blot and RT-PCR tests demonstrated that EV contents (surface protein and RNA) were preserved intact [64].
Due to the multiple mechanisms, RD and ultimate vision loss are aligned with retinal cells death, especially photoreceptors [65]. Moreover, retinal neuroinflammation begins at the early stages of RD and is amplified at advanced stages. Therefore, the inflammatory modulation response seems to be a critical factor in the success of retinal regeneration therapies [66]. Importantly, SCAP can secrete a large variety of neuroprotective and immunomodulatory factors with a higher capacity than conventional MSC origins, making them a desirable candidate for neural regeneration [4]. Given that PC12 cells are neural progenitor cells and considered a significant model system to study many aspects of neuronal activity [67], we employed H2O2-induced oxidative stress in cultured PC12 cells as an in vitro model and evaluated neuroprotection and anti-inflammatory response in cells pretreated with SCAP-EVs. We found that the MEMS-EVs group, notably, revealed a higher anti-apoptosis and antiinflammatory response that confirmed the therapeutic effect of MEMS-EVs against neuroinflammation.
Previous studies have thoroughly documented that the RCS rats have a mutation in the merTK gene that ultimately leads to photoreceptor cell death to such an extent that at seven weeks of age, almost half of the nuclei of the photoreceptors have degenerated [49]. Moreover, the genetic mutation prevents RPE phagocytosis of photoreceptors shed outer segments, leading to a disorganized OS and a thinner ONL [49]. It has been reported that dental pulp stem cells have the capability to respond to cues from the retina and express mature photoreceptor marker rhodopsin ex vivo [68]. In the present study, we observed that the ONL of the MEMS-EVs group had fewer apoptotic cells than the untreated control, vehicle, and UF-EVs treated groups. Furthermore, immunohistochemical staining for the photoreceptor cell markers rhodopsin confirmed more rhodopsin-positive OS in rat eyes treated with MEMS-EVs compared to untreated control, vehicle, and UF-EVs treated groups. These results support the notion that conservation of the visual acuity in the MEMS-EVs treated eyes was correlated with the preservation of photoreceptors in the ONL, identified by rhodopsin expression in the OS of photoreceptors, and observed antiapoptotic response. Besides the preserved photoreceptors, we examined the surviving RGCs using two reliable and efficient neuron-specific proteins (neuronal nuclei [NeuN]) [69] and Brn3a [70] in retina treated with MEMS-EVs compared to untreated control, vehicle, and UF-EVs treated groups. NeuN-positive neurons, as well as Brn3a-positive RGCs, were increased in retina treated with MEMS-EVs compared to untreated control, vehicle, and UF-EVs treated groups, that is in line with the RGCs nuclei quantification measured by H&E staining. Since NeuN labels displaced amacrine cells as well as RGCs [71], the relative increase in NeuNpositive neurons and Brn3a-positive RGCs exhibit that neural retina was preserved in retina treated with MEMS-EVs supporting the neuroprotection potential of MEMS-EVs.
Morphologically, MEMS-EVs increased the cell nuclei density and delayed the thinning of ONL and GCC. The proposed MEMS isolation method that mostly enriches a higher number of sEVs subset relies solely on physical interactions, so possible damage caused by chemical and biological reactions may be prevented. The use of 10 kHz (frequency), 5Vpp (voltage), and 0.03 mS/m (medium conductivity) working conditions reduces the likelihood to EVs damage and altering the particles' nature during experiments, as described by several studies [28,64,72,73]. Similarly immunocytometry test on MEMS-EVs demonstrated that EVs contents (surface protein) were preserved intact.
The numbers of retinal cells labelled with the proliferating cell nuclear antigen (PCNA) were not significantly different among the control and protective groups, reflecting that majority of retinal cells did not proliferate, not even in the presence of SCAP-EVs, but when SCAP-EVs were present, the survival or anti-apoptotic pathways were promoted. These data align with previous neuroprotection studies that aim to preserve retinal cells' viability in the early stage of vision loss and postpone the development of retinal degeneration [74]. Angiogenic factors such as vascular endothelial growth factor (VEGF), which stimulate neovascularization in the pathological processes, are closely involved in RD pathogenesis [75,76]. However, profiling SCAP secretion capacity exhibited that SCAP conditioned medium and EVs could potentially promote angiogenesis [9,77]. It has been shown that microenvironmental cues present in damaged tissue may induce endothelial transdifferentiation of SCAP, enhancing their regeneration potential [78]. Therefore the effect of SCAP-EVs on neovascularization has to be studied before taking any further steps toward the clinic. Our quantitative assessment of the retinal neovascularization three months after transplantation revealed no statistically significant difference among the average count of vascular lumens in protective groups compared to the control groups giving insight into the neovascularization property of SCAP-EVs.
To further evaluate the underlying protective mechanisms of SCAP-EVs, we sought to evaluate the inflammatory and apoptotic-related marker expressions (Il-10, Il-6, Bax, and Bcl-2). In agreement with the previous immunohistochemical results, we found suitable therapeutic functionality via regulation of apoptotic-related factors (Bax and Bcl-2) in the MEMS-EVs treated group compared to the vehicle. Moreover, we detected the up-regulation of Il-10 and downregulation of Il-6, indicating the anti-inflammatory response of the MEMS-EVs treated group compared to the vehicle, which is in line with the reported neuroprotective potential of these cells [11].
Compared to the UF isolation method, we found that the MEMS device isolates a relatively smaller subset of EVs that effectively suppressed RD when administered subretinally. However, smaller EVs diffuse rapidly across the injected area, which results in their short retention time in a single dose [18]. Even though we observed that the visual function of the RCS rats was partially rescued following subretinal delivery of SCAP-EVs, further visual acuity measurements are needed to reflect the function of the localized part of the retina as well as the long-term visual function of the entire retina. Accordingly, additional studies are required to completely recognize the pharmokinetics of the EVs subsets for prolonged therapeutic effects in RD therapy.

Conclusion
Our data suggest that SCAP-EVs mediate neuroprotection in the RD model through mitigation of apoptosis and inflammation responses. MEMS enrichment method is capable of isolating a smaller and effective EVs population that protects the visual acuity, reinforcing the notion that EVs' isolation methods determine the compositions and therapeutic capacity of the isolates. The proposed MEMS device offers flexible manufacturing options, with the coupled advantage of relatively low cost. These endeavors are crucial for a wide range of biomedical and clinical applications as a remarkable practical and vital candidate since a MEMS device can be employed to generate patient-specific therapeutics. Our findings suggest that SCAP-EVs is potentially a promising candidate for RD treatment.