Introduction

Photoreceptor apoptosis is recognized as one of the major form of photoreceptor death and one important contributor to visual loss in retinal degenerative disorders [1, 2], such as inherited retinal dystrophies encompassing the retinitis pigmentosa (RP) [3]. Recently, mesenchymal stem cell transplantation (MSCT) has demonstrated immense potential to ameliorate a variety of ocular dysfunctions [4, 5]. Particularly, MSCT retards retinal degeneration with the prevention of loss of retinal ganglion cells (RGCs) and retinal pigment epithelium (RPE) [1, 2]. Extensively efficient to ameliorate retinal traumatic, ischemic, and oxidative injuries [7, 18] as well as retarding inflammatory and diabetic retinopathies [8, 52], MSCT represents promising cell-based therapeutics which also rescues photoreceptor deficiency in mice with genetic defects [9,10,11]. Notably, there are studies reporting retinal damages and ocular complications induced by intravitreal or subretinal MSCT, indicating that stem cell delivery should still be applied with caution [53, 54]. In this study, we for the first time revealed that MSCT protected photoreceptors against specific apoptotic stimulus in vivo, and further identified EXOT as cell-free alternative strategy to overcome potential drawbacks of cell transplantation with the beneficial effects preserved. Moreover, we discovered that a single intravitreal injection of MSCs or exosomes has a long-lasting protection of photoreceptors in both the MNU pharmacological model and the Pde6bmut genetic model with ameliorated loss of either scotopic (mixed rod and cone responses) or photopic (cone-mediated) ERG responses, which therefore provides promise for establishing novel therapeutics of retinal degeneration.

It is recognized that MSCT maintains tissue homeostasis through either paracrine effects to establish beneficial microenvironments or through inhabitation in recipient tissues to replenish deficient cells [14, 55]. After intravitreal transplantation, exogenous MSCs have been traced to remain in the vitreous body, in which no retinal incorporation has been reported [56,57,58]. Nevertheless, it has been claimed in MSCT treating a mouse model of RP that transplanted MSCs morphologically integrated into the RPE, while not being detected in other retinal layers [10]. Additionally, in vitro experiments indicated regulation of MSCs by the retinal microenvironment, in that RPE cell-conditioned medium and photoreceptor outer segments stimulate differentiation of MSCs toward the RPE cell phenotype, but the in vivo functional evidence of MSC replacing retinal cells is lacking [59]. Here, we showed that the biodistribution of injected MSCs was primarily at the ONL, indicating close functional correlationships with recipient photoreceptors and potential needs of damaging photoreceptors for recovery. However, MSCs were able to release exosomes for therapeutic effects at the ONL, and effects but not biodistribution of MSCs were diminished upon blockade of exosomal generation. Therefore, it can be expected that MSCT improves retinal integrity and function via paracrine effects, as also confirmed by the identification of various neuroprotective and anti-inflammatory factors secreted, such as nerve growth factor, basic fibroblast growth factor, and tumor necrosis factor alpha-stimulated gene-6 [8, 16]. It has additionally been reported that MSC-derived EVs enhance functional recovery of retina after ischemia-reperfusion damages and optic nerve crush injuries, and that transplanted EVs can be uptaken by RGCs and microglia [17, 18]. In this study, we further revealed that EXOT efficiently and continuously counteracted phototoxin-induced and Pde6bmut retinal degeneration by targeting photoreceptors, adding another dimensional to the current knowledge of paracrine-based MSC therapy of retinal disorders. Whether the injected MSCs can also proliferate and differentiate toward photoreceptors in vivo remain to be investigated in future studies.

Shuttle of microRNAs by EVs have been widely proved as significant contributors to tissue homeostasis and feasible therapeutics for diseases [23,24,25]. In the ocular system, it has been reported that EV-mediated microRNA communication may influence posttranscriptional regulation of retinal development, and that microRNAs carried by circulating EVs can be used as prognostic biomarkers for retinopathy [20, 21]. microRNA-dependent mechanisms have also been implicated in MSC-derived EVs to treat corneal fibrosis and optic nerve injury [17, 60]. In this study, we first identified an individual microRNA, miR-21, as a contributor to photoreceptor viability in vivo and a mediator of therapeutic effects of EXOT on retinal dysfunction. Notably, miR-21 has been previously documented mainly as a negative regulator of retinal health, such as inhibition of neovascularization in the ischemic retina by targeting tissue inhibitor of metalloproteinase 3 [61], induction of Müller cell gliosis after optic nerve crush by regulating glial fibrillary acidic protein [39], promotion of autoimmune uveoretinitis by targeting interleukin-10 [62], and facilitation of the progression of retinoblastoma by targeting phosphatase and tensin homolog [63]. Besides, previous studies have shown that miR-21 can induce cell apoptosis by targeting S-phase kinase-associated protein 2 and B-cell lymphoma-2 [64, 65]. Our results showing miR-21 protection against retinal disorders only join with another study demonstrating that the miR-21/Pdcd4 axis regulates MSC-induced neuroprotection in glaucoma, indicating target-specific regulatory effects based on Pdcd4 inhibition [48]. Actually, miR-21 inhibition of Pdcd4 expression for antiapoptotic effects has additionally been reported in other systems [66, 67]. The detailed function and mechanisms of Pdcd4 in regulating photoreceptors and retinal degeneration in vivo remains to be elucidated in future studies.

Materials and methods

Animals

Animal experiments were performed following the Guidelines of Intramural Animal Use and Care Committees of Fourth Military Medical University, **’an Jiaotong University and the ARRIVE guidelines. Twelve-week WT C57BL/6 mice and miR-21−/− mice (C57BL/6 background) (weight, 20–22 g; three male or female mice per cage) were purchased from the Jackson Laboratory, as we have used in previous studies [42, 43]. Two-week-old male or female Pde6bmut mice with a nonsense mutation of Pde6b gene on a C57BL/6J background that result in the failure of protein production [38] were obtained from TC at Department of Clinical Medicine, Fourth Military Medical University. Mice were maintained with good ventilation and a 12-h light/dark cycle, and were kept feeding and drinking ad libitum.

Cell culture

Isolation and culture of MSCs from mouse bone marrow were as previously described [68]. Briefly, whole bone marrow cells were seeded, incubated overnight, and rinsed with phosphate buffer saline (PBS) to remove the non-adherent cells. The adherent cells were cultured with alpha-Minimum Essential Medium supplemented with 20% fetal bovine serum (FBS), 2-mM l-glutamine, 100-U/ml penicillin, and 100-g/ml streptomycin (all from Invitrogen, USA) at 37 °C in a humidified atmosphere of 5% CO2. Primary MSCs were digested with 0.25% trypsin (MP Biomedicals, USA) and passaged for the following experiments after seeding at appropriate densities.

MSCs were verified according to the current standard [32, 33]. For flow cytometric analysis of the surface markers, MSCs at the first passage were collected and suspended in PBS supplemented with 3% FBS at 1 × 106 cells/ml. MSCs were added with FITC-conjugated anti-mouse CD11b antibody (11-0112-82; eBioscience, USA), PE-conjugated anti-mouse CD29 antibody (12-0299-42; eBioscience, USA), FITC-conjugated anti-mouse CD34 antibody (11-0341-82; eBioscience, USA), PE-conjugated anti-mouse CD45 antibody (12-0451-82; eBioscience, USA), PE-conjugated anti-mouse CD31 antibody (12-0311-82; eBioscience, USA), and FITC-conjugated anti-mouse Stem cell antigen 1 (Sca1) antibody (11-5981-82; eBioscience, USA) at concentrations of 1:100. Nonimmune immunoglobulin of the same isotype was used as the NC. MSCs were incubated in 4 °C for 30 min in dark, and then washed twice with PBS supplemented with 3% FBS. The percentages of positively stained cells were determined with a flow cytometer (FACSAria; BD Biosciences, USA) equipped with the FACSDiva Version 6.1.3 software. For cell viability examinations, the seeded MSCs at the first passage were incubated with 20-μl 5-mg/ml methyl thiazolyl tetrazolium (MP Biomedicals, USA) for 4 h [55, 69]. The precipitates were extracted with 180-μl dimethyl sulfoxide (DMSO) and the absorbance was measured at the optical density of 490 nm. Cell viability fold changes were calculated accordingly. For osteogenic differentiation, the seeded MSCs at the first passage were induced in media containing 100-μg/ml ascorbic acid (MP Biomedicals, USA), 2-mM β-glycerophosphate (Sigma-Aldrich, USA) and 10-nM dexamethasone (Sigma-Aldrich, USA) for 14 d, and Alizarin red (Sigma-Aldrich, USA) staining was performed to determine the mineralization. For adipogenic differentiation, the seeded MSCs at the first passage were induced in media containing 0.5-mM isobutylmethylxanthine (MP Biomedicals, USA), 0.5-mM dexamethasone and 60-mM indomethacin (MP Biomedicals, USA) for 14 d, and Oil red O (Sigma-Aldrich, USA) staining was performed to determine the lipid droplet formation. Photographs were taken using an inverted optical microscope (CKX41; Olympus, Japan).

Culture of the 661W cell line was according to our previous study and verified for identity and non-contamination [70]. The 661W cell line was derived from mouse retinal tumors and has been characterized previously to be of cone photoreceptor cell lineage [71]. The 661W cell line was cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% FBS, 2-mM l-glutamine, 100-U/ml penicillin, and 100-g/ml streptomycin (all from Invitrogen, USA) at 37 °C in a humidified atmosphere of 5% CO2. For co-culture of MSCs with 661W cells, 661W cells were seeded in 12-well plates as the bottom of a Transwell system (0.4-μm pore size; Corning, USA), while MSCs at the first passage were added into the upper chamber of the Transwell system for 48 h.

Chemical treatments

MNU-provoked retinal degeneration model was established accordingly [70, 72] by intraperitoneal injection of MNU (Sigma-Aldrich, USA) at 40 mg/kg. MNU was freshly dissolved in sterile saline. Mice were sacrificed at indicated time points according to study design. In vitro treatment of 661W cells with MNU was performed at a concentration of 200 μg/ml for 6 h. For blockade of MSC generation of exosomes [36, 37], 10-μM GW4869 (Sigma-Aldrich, USA) was applied in MSC culture for 24 h, which was initially dissolved in DMSO into a stock solution of 5 mM and was diluted in culture media. The effects of GW4869 on MSC viability and osteogenesis were determined after wash-out procedures.

Collection and identification of exosomes

Collection of MSC-derived exosomes was performed as stated before [73]. Briefly, MSCs at nearly confluence of the first passage were cultured in exosome-depleted media (complete media supernatant after overnight centrifugation at 100,000 g) for 48 h. Exosomes from supernatants were then isolated by the differential centrifugation protocol at 4 °C: 300 g for 10 min, 3000 g for 10 min, 10,000 g for 20 min, 100,000 g for 70 min, followed by washing with PBS for another centrifugation at 100,000 g for 70 min. Quantification of exosomes were performed by determining the concentration of total proteins using the Pierce BCA Protein Assay (Thermo Fisher Scientific, USA).

Exosomes were identified according to the criteria of EVs [74]. The number and size distribution of exosomes was quantitated using dynamic light scattering with a NanosizerTM instrument (Malvern Instruments, UK) [75]. Transmission electron microscopy (TEM) [76] was performed on whole mounts of isolated exosomes using 2.5% glutaraldehyde, postfixed with 1% osmium tetroxide, embedded and stained with uranyl acetate and lead citrate, and observed using a TEM microscope (JEM-1230; JEOL, Japan). Purified exosomes were further characterized by western blot using anti-CD63 and anti-CD81 antibodies [73], as stated below.

Intravitreal injection

MSCs at the first passage were collected and suspended in PBS at 1 × 107 cells/ml and were injected at 1 μl into each vitreous chamber with a 33-G Hamilton syringe (Hamilton Company, USA) at 6 h after MNU treatment or at 2-week old of Pde6bmut mice. The collected exosomes were adjusted to a protein concentration of 1 μg/μl and were also injected at 1 μl into each vitreous chamber at 6 h after MNU treatment or at 2-week old of Pde6bmut mice. For the intravitreal delivery, the injection site was selected just posterior to the limbus and the needle was injected under anesthesia (100 mg/kg ketamine plus 20 mg/kg xylazine intraperitoneally), which was retracted after 2 min to minimize backflow [17].

Retinal histology and morphology

At sacrifice, retinal tissues were rapidly isolated, fixed overnight with 4% paraformaldehyde, and embedded in paraffin. 5-μm serial sections were prepared (RM2125; Leica, Germany) and underwent H&E staining for tissue histology and morphology, according to previous reports [70]. Quantification of ONL thickness was determined using the ImageJ 1.47 software along the vertical meridian of the eyeball from nasal to temporal side and through the optic nerve head (ONH) [77], and statistically analyzed as area under the curve (AUC).

TUNEL and IF staining

At sacrifice, retinal tissues were rapidly isolated, fixed in 4% paraformaldehyde, cryoprotected with 30% sucrose, and embedded in the optimal cutting temperature (OCT) compound. The specimens were snap-frozen and sectioned into 15-μm sagittal sections (CM1950; Leica, Germany). Sections were then underwent TUNEL assay using DeadEndTM Fluorometric TUNEL System according to the manufacturers’ instructions (Promega, USA), counterstained by Hoechst 33342 (Sigma-Aldrich, USA) [32]. The images were further analyzed using the ImageJ 1.47 software from at least five consecutive microscopic fields for TUNEL+ cells over total ONL cells. For IF analyses on cone photoreceptors, sections were blocked with 5% bovine serum albumin (BSA) (Sigma-Aldrich, USA) dissolved in PBS for 1 h at room temperature, and stained with a rabbit anti-mouse S-opsin primary antibody (ABN1660; Millipore, USA) overnight at 4 °C at a concentration of 1:1000. After washing with PBS, sections were stained with an AF647-conjugated goat-anti-rabbit secondary antibody for 1 h at room temperature, and were counterstained with Hoechst 33342 (Sigma-Aldrich, USA). Quantification of the percentages of positively stained area over total retinal fields was performed using the ImageJ 1.47 software.

ERG analysis

The UTAS Visual Diagnostic System with a Big Shot Ganzfeld (LKC Technologies, USA) was employed. Mice were dark-adapted overnight before ERG analysis. Under dim red light conditions, anesthesia was induced by an intraperitoneal injection of 80 mg/kg ketamine and 10 mg/kg chlorpromazine, and mice were then lightly secured to a stage to ensure a stable position for recording. Cornea was anesthetized with a drop of 0.5% proxymetacaine and was kept moist with physiological saline. Platinum circellus record electrodes were placed on each cornea, while reference and ground electrodes were respectively located in the mouth and inserted in the tail. White flashes with the intensity of 3.0 cd.s/m2 were applied for stimulating the responses. The band-pass (1–300 Hz) was used to amplify the recorded signals. The line noise was wiped off by a 50-Hz notch filter. Totally 10 scotopic (dark-adapted, mixed rod- and cone-mediated) and 60 photopic (light-adapted, cone-mediated) responses were recorded for analyzing the amplitudes of a-wave and/or b-wave by ERG View v4.380 R software [78].

Transfection of microRNA mimics, siRNA, and plasmids

The mimics and NC of miR-21, and the siRNA and NC of Pdcd4 were designed by RiboBio (Guangzhou, China) and were transfected into cells at final concentrations of 50 nM, according to previous studies [42, 43]. The Lipofectamine 2000 reagent (Invitrogen, USA) was used for transfection according to the manufacturer’s recommendation. CD63-EGFP plasmids were purchased from Hanheng (Shanghai, China) and were also transfected into MSCs using the Lipofectamine 2000 reagent (Invitrogen, USA). The transfected MSCs or 661W cells were further used for downstream experiments after 48 h incubation.

In vitro apoptosis assay

661W cells were harvested and evaluated by FITC-conjugated Annexin V and propidium iodide (PI) double staining according to the manufacturer’s instruction of Annexin V Apoptosis Detection Kit I (BD Biosciences, USA). Cell apoptosis was detected with a flow cytometer (CytoFLEX; Beckman Coulter, USA) equipped with the CXP 2.1 software. The percentages of early apoptotic (FITC+PI−) plus late apoptotic (FITC+PI+) cells were expressed as apoptotic percentages [55, 79].

Tracing of MSCs, exosomes, and miR-21

For tracing of MSCs and exosomes, PKH26 (Sigma-Aldrich, USA) was used according to manufacturer’s instructions [73]. For biodistribution of MSCs in the eye, freshly dissected eyes were scanned using the Xenogen IVIS imaging system (Xenogen IVIS Lumina II; PerkinElmer, USA) at 6 h after injection of PKH26-labeled MSCs. For in vivo tracing in the retina, retinal tissues were harvested at 24 h after injection of PKH26-labeled MSCs or exosomes. CD63-EGFP plasmid-transfected MSCs were also injected for in vivo examination of exosome release. For tracing of MSC-derived miR-21 in vivo, Cy5-labeled miR-21 mimics (RiboBio, China) was transfected into MSCs using the Lipofectamine 2000 reagent (Invitrogen, USA) before injection. The retina were then fixed in 4% paraformaldehyde, cryoprotected with 30% sucrose, and embedded in the OCT compound. The specimens were snap-frozen and sectioned into 15-μm sagittal sections (CM1950; Leica, Germany) and counterstained by Hoechst 33342 (Sigma-Aldrich, USA). For in vitro tracing, 661W cells were treated by PKH26-labeled exosomes for 48 h, fixed in 4% paraformaldehyde, blocked by 5% BSA, and immunostained with a rabbit anti-mouse primary antibody at a concentration of 1:100 for β-tubulin (ab18207; Abcam, UK) overnight at 4 °C. After washing with PBS, the sections were stained with a FITC-conjugated goat-anti-rabbit secondary antibody and counterstained by Hoechst 33342 (Sigma-Aldrich, USA). For tracing of MSC-derived miR-21 into 661W cells in vitro, Cy3-labeled miR-21 mimics (RiboBio, China) was transfected into MSCs using the Lipofectamine 2000 reagent (Invitrogen, USA) before MSCs being added in the upper chamber of the Transwell system. MSCs and 661W cells were then co-cultured for 48 h, and 661W cells were fixed in 4% paraformaldehyde and counterstained by Hoechst 33342 (Sigma-Aldrich, USA).

qRT-PCR

Total RNA was extracted from retinal tissues or cells by direct adding of Trizol Reagent (Takara, Japan) and purified by phenol–chloroform extraction. For mRNA, cDNA synthesis was performed using the PrimeScriptTM RT Reagent Kit (Takara, Japan), and the primer sequences of the genes detected in this study were listed in the Supplementary Information (Table S1). For microRNA, reverse transcription primers and qRT-PCR primers were designed by RiboBio (Guangzhou, China) as the Bulge-loopTM miRNA qRT-PCR Primer Sets. qRT-PCR detection was performed using the SYBR Premix Ex Taq II Kit (Takara, Japan) by a Real-Time System (CFX96; Bio-Rad, USA). The relative expression level of each gene was obtained by the cycle number after normalizing against Glyceraldehyde-3-phosphate dehydrogenase (Gapdh, for mRNA) and RNU6 (for microRNA) abundances using the 2−ΔΔCT method [42, 43].

Western blot

Traditional western blot was performed as previously described [79]. Whole lysates of exosomes were prepared using the Lysis Buffer (Beyotime, China). Proteins were extracted, loaded on sodium dodecyl sulfate-polyacrylamide gels, transferred to polyvinylidene fluoride membranes (Millipore, USA), and blocked with 5% BSA (Sigma-Aldrich, USA) in PBS with 0.1% Tween for 2 h in room temperature. The membranes were incubated overnight at 4 °C with the following primary antibodies: a rabbit anti-mouse primary antibody at a concentration of 1:1000 for CD63 (sc-5275; Santa Cruz Biotechnology, USA); and a rabbit anti-mouse primary antibody at a concentration of 1:1000 for CD81 (sc-70803; Santa Cruz Biotechnology, USA). The membranes were then incubated with peroxidase-conjugated goat anti-rabbit secondary antibodies at a concentration of 1:40000 for 1 h in room temperature. The blotted bands were visualized using an enhanced chemiluminescence Kit (Amersham Biosciences, USA) and a gel imaging system (5500; Tanon, China).

Capillary-based immunoassay was performed for retinal and 661W cell lysates using the Wes-Simple Western method with the anti-rabbit detection module (ProteinSimple, USA) [80]. Protein expression was measured by chemiluminescence and quantified as AUC using the Compass for Simple Western program (ProteinSimple, USA). Proteins were detected with the following primary antibodies: a rabbit anti-mouse primary antibody for Pdcd4 (9535S; Cell Signaling Technology, USA) and a rabbit anti-mouse primary antibody for Gapdh (5174S; Cell Signaling Technology, USA) at concentrations of 1:1000 (both from Cell Signaling Technology, USA).

Sample processing

Each experiment was performed in at least three independent times and data represented biological replicates. Sample size of animals was determined based on preliminary tests for variation to ensure adequate power for analysis. Samples were blindly allocated to each of the experimental group based on randomization and were also blind for data analysis. No samples were excluded from the results.

Statistical analysis

The results are represented as median ± range or as box (25th, 50th, and 75th percentiles) and whisker (range) plots for indicated experiments. The data were analyzed using non-parametric methods, i.e. the Mann–Whitney U tests for two group comparisons or the Kruskal–Wallis tests for multiple group comparisons, in the GraphPad Prism 5.01 software for not meeting normal distribution. Values of P < 0.05 were considered to be statistically significant.