Mesenchymal stem cell-derived exosomes protect trabecular meshwork from oxidative stress

This study aims to investigate the beneficial effects of exosomes derived from bone marrow mesenchymal stem cells (BMSCs) on trabecular meshwork cells under oxidative stress and predict candidate genes associated with this process. Trabecular meshwork cells were pretreated with BMSC-derived exosomes for 24 h, and exposed to 0.1 mM H2O2 for 6 h. Survival rate of trabecular meshwork cells was measured with CCK-8 assay. Production of intracellular reactive oxygen species (iROS) was measured using a flow cytometer. RT-PCR and ELISA were used to detect mRNA and protein levels of inflammatory cytokines and matrix metalloproteinases (MMPs). Sequencing of RNA and miRNA for trabecular meshwork cells from Exo and control groups was performed on BGISEQ500 platform. Phenotypically, pretreatment of BMSC-derived exosomes improves survival rate of trabecular meshwork cells exposed to H2O2, reduces production of iROS, and inhibits expression of inflammatory cytokines, whereas increases expression of MMPs. There were 23 miRNAs, 307 lncRNAs, and 367 mRNAs differentially expressed between Exo and control groups. Exosomes derived from BMSCs may protect trabecular meshwork cells from oxidative stress. Candidate genes responsible for beneficial effects, such as DIO2 and HMOX1, were predicted.

www.nature.com/scientificreports/ HBMSC-derived exosomes improve viability of hTMCs exposed to H 2 O 2 . After 24 h of pretreatment with hBMSC-derived exosomes and washing 3 times with PBS, hTMCs were exposed to H 2 O 2 (0.1 mM) for certain periods i.e., 6, 12, and 24 h, respectively. Notably, H 2 O 2 (0.1 mM) significantly reduced survival rate of hTMC in a time-dependent manner. Meanwhile, pretreatment with hBMSC-derived exosomes improved viability of hTMCs exposed to 0.1 mM H 2 O 2 (Fig. 3), and 6 h was chosen as the time point for the subsequent study.
Effects of hBMSC-derived exosomes on production of intracellular reactive oxygen species (iROS) in H 2 O 2 -exposed hTMCs. After pretreatment with hBMSC-derived exosomes for 24 h and washing 3 times with PBS, hTMCs were exposed to H 2 O 2 (0.1 mM) for 6 h. Exposure to H 2 O 2 significantly induced a high amount of iROS production in hTMCs, whereas pretreatment with hBMSC-derived exosomes decreased production of iROS at 6 h, implying hBMSC-derived exosomes protected hTMCs against oxidative stress by reducing iROS (Fig. 4).
Effects of hBMSC-derived exosomes on expression of MMPs in H 2 O 2 -exposed hTMCs. HTMCs were treated as described above. Expression levels of MMP-2 and MMP-3 were analyzed by RT-PCR and ELISA. After exposure to H 2 O 2 , expression levels of MMP-2 and MMP-3 were unaffected. By contrast, pretreatment with hBMSC-derived exosome up-regulated MMP-2 and MMP-3 in hTMCs exposed to H 2 O 2 (Fig. 6).  (Fig. 7a). Moreover, 4500 target genes were identified through database queries, including 42 DE mRNAs by calculation. After further analysis of gene expression, 12 miRNA-mRNA target pairs were predicted (Fig. 7b). Using Pearson correlation coefficient models, co-expression of DE lncRNA and DE mRNA was predicted. There were 3634 relationship pairs, including 3375 pairs of positive correlation and 259 pairs of negative correlation. DElncRNA-DEmRNA interactional network (power-law distribution R 2 = 0.664) was constructed based on characteristics of a biological process. Notably, 9 subnets were identified using MCODE. Additionally, 21 lncRNA, 26 mRNA, and 263 pairs of interactional relations were identified in the highest score (11.435) subnet. Their levels were positively correlated and up-regulated in Exo group (Fig. 7c). Moreover, using DAVID, functional enrichment of DE mRNA was explored. GO and KEGG analysis displayed that the DE mRNAs were enriched in 138 pathways (Supplementary Table S8). HTMCs were pretreated with hBMSC-derived exosomes or PBS for 24 h, then exposed to 0.1 mM H 2 O 2 for different time points i.e., 6, 12, and 24 h, respectively. Cell viability was measured by CCK-8 assay. n = 3 for each condition. A P value was obtained by a two-tailed unpaired t-test. *P < 0.05 and **P < 0.01 compared to PBS + H 2 O 2 group. Data were presented in Supplementary

Discussion
Detailed mechanisms underlying dysfunctional trabecular meshwork in POAG are not completely understood. However, reports 27,28 have revealed that oxidative stress triggers trabecular meshwork dysfunction in vitro and in vivo. HTMCs under oxidative stress produce excessive iROS via mitochondria. Increased iROS production indicates oxidative stress and thus up-regulates inflammatory markers in hTMCs, including IL-1, IL-6, and IL-8. Additionally, a polymorphism in IL-1α correlated with increased IL-1 gene expression and elevated risk for POAG 29 . As a result, sustained high production of iROS and inflammatory factors are crucial features of hTMC dysfunction in POAG 30,31 . H 2 O 2 -exposed hTMCs are widely adopted in glaucoma model in vitro. In this study, survival rate of trabecular meshwork cells exposed to H 2 O 2 was reduced, iROS was continuously produced, and IL-1α, IL-1β, IL-6, and IL-8 were up-regulated compared to control group. Besides, we successfully constructed classical glaucoma model in vitro. Previous studies 30,31 were confined to antioxidant effect on trabecular meshwork cells exposed to H 2 O 2 . Either sufficiency or deficiency in antioxidant supplement is not always associated with eye pathology 32 . Whereas, exosomes, as a subcellular structure containing RNAs, might affect gene expression and protect trabecular meshwork cells from oxidative stress. MSC-derived exosomes have exhibited remarkable therapeutic effects for degenerative eye diseases. For instance, MSC-derived exosomes accelerated recovery of corneal epithelium 33 . In addition, protective effects of MSC-derived exosomes on retinal ganglion cells were identified in glaucoma models 34 . However, one study 13 had explored potential effects of MSC-derived exosomes on trabecular meshwork, a key component of aqueous drainage channels. Herein, we focused on hTMCs, which were pretreated with hBMSC-derived exosomes, then exposed to hydrogen peroxide. Interestingly, hTMCs pretreated with hBMSC-derived exosome demonstrated enhanced survival rate, lower iROS production, and lower expression of IL-1α, IL-1β, IL-6, and IL-8 compared to those only exposed to hydrogen peroxide. These findings imply that BMSC-derived exosomes potentially alleviate dysfunction of trabecular meshwork induced by oxidative stress.
Besides up-regulated proinflammatory cytokines, iROS induced by oxidative stress potentially promotes pathophysiological changes in the outflow tract by increasing intracellular oxidative damage. Matrix metalloproteinases (MMPs), a class of zinc-containing neutral proteases implicated in regulating extracellular matrix degradation as well as interacting with microenvironment, have been proposed as a vital group of enzymes to maintain outflow tract homeostasis [35][36][37][38][39][40] . In human outflow models, MMP-2 and MMP-3 could significantly increase the outflow capacity 41 , and regulate extracellular matrix remodeling 42 . This study confirmed that cultured hTMCs secrete MMPs 43 , and detected changes in MMP-2 and MMP-3 levels upon stimulation of BMSC-derived exosome in hTMCs. Surprisingly, although oxidative stress might not influence MMPs levels in hTMCs, MMPs were up-regulated in hTMCs pretreated with hBMSC-derived exosomes. This indicates that BMSC-derived exosomes may regulate MMPs to enhance trabecular meshwork function. . Pretreatment of hBMSC-derived exosomes reduced production of iROS in hTMCs when exposed to H 2 O 2 . HTMCs were pretreated with hBMSC-derived exosomes or PBS for 24 h, then exposed to H 2 O 2 (0.1 mM) for 6 h, and a blank control group cultured with medium containing exosome-free serum. (a) Intracellular ROS was identified with DCFDA staining and measured by flow cytometry. (b) Quantification of iROS fluorescence. n = 3 for each condition. A P value was obtained by a two-tailed unpaired t-test. *P < 0.05 compared to PBS + H 2 O 2 group. Data were presented in Supplementary Table S3 www.nature.com/scientificreports/ To identify candidate genes responsible for alleviating oxidative stress damage in trabecular meshwork cells, gene expression was profiled in Exo and control groups. Differentially expressed genes, including 23 DE miR-NAs, 307 DE lncRNAs, and 367 DE mRNAs, were identified. As shown in Fig. 7a, we analyzed the differentially expressed miRNAs in Exo and control group. Many of DE miRNAs were reported [44][45][46][47] to be involved in the development of glaucoma, oxidative stress response, and the regulation of MMPs. For example, the expression of miR-126 in micro-vesicles was up-regulated in tears of patients diagnosed with open-angle glaucoma 46 . In our study, miR-126-5p was down-regulated in Exo group, indicating that hBMSC-derived exosomes may reduce risk of glaucoma through it. MiR-451a and miR-125b were reported 45,47 to inhibit the expression of MMP-2. In our study, both of them were down-regulated in Exo group, indicating that hBMSC-derived exosomes up-regulated expression of MMP-2 through them.
After further analysis of gene expression, we predicted the regulatory networks of miRNA-mRNA (Fig. 7b) and lncRNA-mRNA (Fig. 7c). Many of these DE genes have been reported to be related to trabecular meshwork dysfunction and oxidative stress response. For example, in DEmiRNA-DEmRNA regulation network, miR-3529-3p was up-regulated in Exo group, while its target gene, CXCL5, was down-regulated. CXCL5, an inflammatory chemokine, was significantly elevated in aqueous humor in patients with glaucoma 48 . HBMSC-derived exosomes may reduce the inflammatory response of hTMCs under oxidative stress through the action of miR-3529-3p on CXCL5. In lncRNA-mRNA interactional subnet, DIO2 and HMOX1 act as hub nodes, co-expressing with 10 lncRNAs, respectively. DIO2 was down-regulated in Exo group. It was reported 49 to regulate phagocytosis in trabecular network and extracellular matrix remodeling, thereby maintaining homeostasis of outflow tract. Hmox1 was up-regulated in Exo group. It is one of crucial factors in Nrf2 pathway, which plays a pivotal role in inflammation and oxidative stress response 50,51 . Several studies 50,51 have demonstrated that up-regulated expression of Hmox1 and its metabolites have significant anti-inflammatory and antioxidant defense effects mediated by Nrf2. The expression of Hmox1 was reported 44 to be down-regulated with increased production of iROS in trabecular meshwork. In addition, C5AR1 was down-regulated in Exo group, which was reported to be up-regulated in retinal pigment epithelium cells under oxidative stress 52 . All of the genes above may play roles in the protective effect of hBMSC-derived exosomes on trabecular meshwork under oxidative stress.
Subsequently, the results of GO and KEGG analysis demonstrated that the DE mRNAs were enriched in 138 pathways, including cell division, extracellular matrix organization, regulation of ERK1 and ERK2 cascade, etc. ERK1/2 pathway, for example, was involved in the ROS-induced cell injury 53 and affect MMP-2 secretion in hTMCs 54 . It could be predicted that many of the 138 pathways play roles in the protective effect of hBMSCderived exosomes on trabecular meshwork under oxidative stress.
Our study was limited to in vitro experiments on one strain of hTMCs. The conditions of hTMCs exposed to H 2 O 2 might not entirely mirror those in POAG in vivo, however, pathological conditions may present similarity to a certain extent. Besides, hTMCs are constantly exposed to aqueous humor in vivo, while H 2 O 2 level in human aqueous humor can reach as high as 300 μM 28 . However, this finding needs to be subjected to more strains of hTMCs or in vivo experiments. In addition, the findings from transcriptome sequencing need to be validated by a serial of experiments. Nonetheless, this study has demonstrated that hBMSC-derived exosomes could be absorbed by hTMCs, which have exerted protective effect against oxidative stress damage, and thus enabling functional preservation of hTMCs. The transcriptomic sequencing and network analysis indicate meaningful regulatory RNAs and nodes, thus providing a basis for future research on glaucoma therapy by integraing exosomes with stem cells.
Isolation and characterization of exosomes. Fresh exosome-free medium was changed at 80% confluence of hBMSCs. After 48 h, conditioned medium was collected and centrifuged at 300g for 10 min, 2000g for 10 min and 10,000g for 30 min, to discard the pellets and collect the supernatant each time. Thereafter, the supernatant was subjected to centrifugation in a SW 32 Ti rotor (Beckman Coulter, USA) at 100,000g for 70 min. After washing twice, the pellets were re-suspended in PBS. The exosome preparation was filtered through a 0.22 μm filter and stored at − 80 °C until use. At passage 2 to 5, exosomes were isolated from hBMSCs. Figure 5. Pretreatment of hBMSC-derived exosome down-regulated the expression of inflammatory factors in hTMCs when exposed to H 2 O 2 . HTMCs were pretreated with hBMSC-derived exosomes or PBS for 24 h, then exposed to H 2 O 2 (0.1 mM) for 6 h and a blank control group cultured with medium containing exosomefree serum. Then, hTMCs were assayed by RT-PCR to assess mRNA expression of IL-1α (a), IL-1β (c), IL-6 (e), and IL-8 (g). Gene expression levels were normalized to GAPDH and presented as relative ratios. Total hTMC supernatants were analyzed for IL-1α (b), IL-1β (d), IL-6 (f), and IL-8 (h). n = 3 for each condition. A P value was obtained by a two-tailed unpaired t-test. *P < 0.05; **P < 0.01 compared with PBS + H 2 O 2 group. Data were presented in Supplementary Tables S4 and S5 www.nature.com/scientificreports/ Exosomal morphology was characterized by transmission electron microscopy (HT7800; Hitachi, Japan). The size distribution of exosomes was measured using a ZetaView analysis system (PMX 110; Particle Metrix, Germany). Classic exosomal surface markers were examined by Western blotting analysis. Total protein from exosomes was extracted in lysis buffer, and a protein BCA assay kit was used to measure concentration. Lysates were separated on 10% SDS-polyacrylamide gel and proteins were transferred to polyvinylidene difluoride membranes. Then, membranes were incubated with 5% milk for 1 h and probed overnight at 4 °C with antibodies targeting HSP70 (ab181606, 1:1000 dilution, Abcam, USA) and CD9 (ab92726, 1:2000 dilution, Abcam, USA). After washing, blots were incubated with peroxidase-conjugated anti-rabbit IgG horseradish and peroxidase secondary antibodies (ab205718, 1:2000 dilution, Abcam, USA) at room temperature for 1 h. The Alpha FluorChem E system (ProteinSimple, USA) was used to visualize the protein bands.

Human trabecular meshwork cells culture and characterization.
A human normal trabecular meshwork cell line was obtained from ScienCell Research Labs (Carlsbad, CA, USA). HTMCs were cultured in DMEM/Ham's F12 (M&C gene technology, Beijing, China) supplemented with penicillin/streptomycin and 10% fetal bovine serum (FBS). Cell cultures were maintained at 37 °C in 5% CO 2 with medium being changed at intervals of 3 days and cells passaged with 0.05% trypsin/EDTA at 80% confluence. HTMCs were characterized through Dexamethasone treatment. HTMCs were exposed to dexamethasone (Dex, 100 nmol; Sigma-Aldrich, Figure 6. Pretreatment of hBMSC-derived exosome up-regulated the expression of matrix metalloproteinases in hTMCs when exposed to H 2 O 2 . HTMCs were pretreated with hBMSC-derived exosomes or PBS for 24 h, then exposed to H 2 O 2 (0.1 mM) for 6 h, and a blank control group cultured with medium containing exosome-free serum. Then, hTMCs were assayed by RT-PCR to assess the mRNA expression of MMP-2 (a) and MMP-3 (c). Gene expression levels were normalized to GAPDH and presented as relative ratios. Total hTMC supernatants were analyzed for MMP-2 (b) and MMP-3 (d). n = 3 for each condition. A P value was obtained by a two-tailed unpaired t-test. *P < 0.05; **P < 0.01 compared to PBS + H 2 O 2 group. # P < 0.05; ## P < 0.01 compared to control. Data were presented in Supplementary Tables S4 and S5 Fig. S1a). And protein expression of myocilin in cell lysate was detected by ELISA ( Supplementary Fig. S1b). Data were presented in Supplementary Table S1. For all experiments, hTMCs were used at passage 2 to 5.
Exosome tracking. HBMSC-derived exosomes were stained with a green fluorescent dye (PKH67, Sigma-Aldrich, USA). Briefly, 200 μl exosome suspension with 1.0 × 10 8 exosomes were mixed with equal volume dilution containing 2 μl dye and incubated at 37 °C for 5 min in dark. An equal volume of 10% BSA-PBA solution was added to stop dyeing. Then, they were subjected to ultracentrifugation at 100,000g for 70 min. The pellets were re-suspended in 100 μL PBS. HTMCs were cultured as described above. At 60% confluence, the cells were exposed to the stained exosomes or dye for 0, 6, 12, and 24 h respectively. Then, the hTMCs were stained with phalloidin (Sigma-Aldrich, USA) and 4, 6-diamidino-2-phenylindole (DAPI, Solarbio Science & Technology, Beijing, China) sequentially, and mounted on a slide and observed under a fluorescence microscope.
Oxidative stress and human BMSC-derived exosomes pretreatment. HTMCs were seeded at a density of 1 × 10 6 cells per well then cultured in normal growth medium for 24 h. After washing 3 times with PBS, the cells were cultured in a complete media (containing exosome-free serum) with hBMSC-derived exosomes (1 × 10 8 particles/ml) or PBS for 24 h. After another 3-time-washing, hTMCs were treated with 0.1 mM H 2 O 2 for 6, 12, and 24 h respectively. Then, the supernatant and cell lysis solution were collected for further analysis.
Cell counting kit-8. After treatments, cell viability was quantified at 6, 12, and 24 h using a cell counting kit-8 (Sigma-Aldrich, USA). A total of 100 ml aliquots of medium was transferred to 96 well plates and absorbance at 450 nm was measured using a spectrophotometer (Bio-rad; USA) and normalized to control.  To obtain clean reads, small RNA sequencing data were filtered by SOAPnuke 55 to removes reads based on the following criteria: low-quality, less than 18nt, with 5′ adapter contamination or poly A, or without 3′ adapter sequence or insert fragment. Bowtie2 56 was used to compare the clean reads with reference sequence. The expression level of miRNAs was measured by transcripts per million (TPM). The differentially expressed miRNAs were detected using DEGseq 57 with the standard of false discovery rate (FDR) ≤ 0.001, and fold change ≥ 2 or fold change ≤ 0.5.

Sequencing of lncRNA and mRNA and identification of DE lncRNA and DE mRNA.
Total RNA was extracted as described above, then qualified and quantified using a Nano Drop and Agilent 2100 bioanalyzer. TruSeq Stranded Total RNA Kit (Illumina, USA) was used to construct the cDNA libraries. After removal of RNA (rRNA), the RNA was fragmented into small pieces. The RNA fragments were used to synthesize first strand cDNA with random primers. In the process of second strand cDNA synthesis, dUTP replaced dTTP. After the addition of a single 'A' base, ligation of the adapter and UDG (uracil-DNA glycosylase) treatment, the products are enriched with PCR and final cDNA library was created. The Agilent 2100 bioanalyzer was used to check the fragments size distribution. The final ligation PCR products were sequenced using the BGISEQ-500 platform (BGI-Shenzhen, China).
RNA sequencing data were filtered by SOAPnuke to eliminate reads containing adapter, unknown base N ratio greater than 5%, or ratio of bases with mass value less than 10 greater than 20%. After obtaining clean reads, HISAT 58 was used to compare the clean reads with reference genome sequence. RSEM 59 was used to calculate gene expression, while FPKM was used to standardize gene expression. The differentially expressed mRNAs and lncRNAs were detected using DEGseq with with the standard of false discovery rate (FDR) ≤ 0.001, and fold change ≥ 2 or fold change ≤ 0.5.

Bioinformatics analysis.
According to the results of differentially expressed genes, hierarchical clustering analysis was performed using PheatMap function in R software. The co-expression relation pair was selected by Pearson's correlation coefficient with the standard of P value < 0.01 and |cor| > 0.9. The target genes of miRNA were quested through the miRTarBase 60 database. The network graph was drawn using Cytoscape software, while the subnets were mined by MCODE in Cytoscape. David (the Database for Annotation, Visualization and Integration Discovery, https:// david. ncifc rf. gov/ summa ry. jsp) was used for enrichment analysis of KEGG 61-63 pathway and GO biological process.

Statistical analysis.
All experiments were conducted at least 3 biological replicates. Results were presented as mean ± standard deviation (SD) of three independent experiments. A P value for difference was determined by t-test using GraphPad Prism software 9 (URL: https:// www. graph pad. com/) and Microsoft Excel. A differences was considered significant when a P value were less than 0.05 (P < 0.05). Figure 7. HBMSC-derived exosomes modulate gene expression of hTMCs exposed to H 2 O 2 . (a) Heat map of 23 DE miRNA between Exo group and control group. The horizontal axis is log2 (expression value + 1) and the vertical axis is a gene. Line indicates DE miRNA, red indicates high expression, and green indicates low expression. (a) was drawn using R 3.6.3 for Windows (URL: https:// cran.r-proje ct. org/ bin/ windo ws/ base/ old/3. 6.3/ R-3. 6.3-win. exe). (b) DEmiRNA-DEmRNA regulation network. Circle indicates DE mRNA, inverted triangle indicates DE miRNA, green indicates down-regulation and red indicates up-regulation. 12 miRNA-mRNA targeting pairs were predicted in the network. (c) Delncrna-DEmRNA interactional subnet. Square indicates lncNRA, Circle indicates mRNA. The data of microRNA, LncRNA and mRNA were presented in Supplementary Tables S6 and S7. (b,c) were drawn using Cytoscape 3.7.2 (URL: https:// github. com/ cytos cape/ cytos cape/ relea ses/ downl oad/3. 7.2/ cytos cape-3. 7.2. zip).

Data availability
Supplementary information accompanies this paper.