Extracellular vesicle-mediated crosstalk between NPCE cells and TM cells result in modulation of Wnt signalling pathway and ECM remodelling.

Abstract Primary open‐angle glaucoma is a leading cause of irreversible blindness, often associated with increased intraocular pressure. Extracellular vesicles (EVs) carry a specific composition of proteins, lipids and nucleotides have been considered as essential mediators of cell‐cell communication. Their potential impact for crosstalk between tissues responsible for aqueous humour production and out‐flow is largely unknown. The study objective was to investigate the effects of EVs derived from non‐pigmented ciliary epithelium (NPCE) primary cells on the expression of Wnt proteins in a human primary trabecular meshwork (TM) cells and define the mechanism underlying exosome‐mediated regulation that signalling pathway. Consistent with the results in TM cell line, EVs released by both primary NPCE cells and NPCE cell line showed diminished pGSK3β phosphorylation and decreased cytosolic levels of β‐catenin in primary TM cells. At the molecular level, we showed that NPCE exosome treatment downregulated the expression of positive GSKβ regulator‐AKT protein but increased the levels of GSKβ negative regulator‐PP2A protein in TM cells. NPCE exosome protein analysis revealed 584 miRNAs and 182 proteins involved in the regulation of TM cellular processes, including WNT/β‐catenin signalling pathway, cell adhesion and extracellular matrix deposition. We found that negative modulator of Wnt signalling miR‐29b was abundant in NPCE exosomal samples and treatment of TM cells with NPCE EVs significantly decreased COL3A1 expression. Suggesting that miR‐29b can be responsible for decreased levels of WNT/β‐catenin pathway. Overall, this study highlights a potential role of EVs derived from NPCE cells in modulating ECM proteins and TM canonical Wnt signalling.


| INTRODUC TI ON
An estimated 8.4 million people globally are blind due to glaucoma, making it the second leading cause of irreversible blindness. 1 This is projected to increase with rising life expectancy. 2 The disease is characterized by progressive damage in the optic nerve fibres and characteristic visual field loss. 3 Primary open-angle glaucoma (POAG) is the predominant form of glaucoma in Western countries.
Intraocular pressure (IOP) is the major modifiable risk factor for POAG, but other IOP-independent risk factors may be involved in the pathogenesis of the disease. 4 The control of IOP is dependent on the dynamic balance between aqueous humour (AH) production mainly by non-pigmented ciliary epithelium (NPCE) and its out-flow through the conventional and alternative pathways. 5 In humans, IOP is primarily controlled by the out-flow of the AH through the trabecular meshwork (TM), also known as the conventional out-flow pathway. The TM is a porous structure composed of endothelial cells embedded in a dense extracellular matrix (ECM), composed of collagen, elastic fibres, proteoglycans and other macromolecules. 6 Fibrotic changes or deposits of ECM into the TM can slow or block the out-flow which leads to an elevation of the IOP. 7,8 Glaucoma is thus linked to TM structure and function where structural changes in the TM likely alter tissue architecture and biomechanical properties which, in turn, influence its resistance to AH out-flow. 6,8 It has previously been proposed that crosstalk communication may exist between inflow tissue and out-flow tissues through potential modulators in AH. 9 Increasing numbers of papers describe the presence of many biologically active factors, signalling molecules, immune mediators and steroid hormones in the AH. [10][11][12] Extracellular vesicles (EVs) were identified in AH and most biologically active compounds in the AH may be EV-associated. 13 Among EVs, exosomes are of particular interest because of their roles in cell-to-cell communication. Exosomes are 30-150 nm nanoparticles that are released from most cell types under normal and pathological conditions. 14 Nucleic acids and proteins are embedded in the phospholipid exosome membrane. 15 These rich exosomal cargos have the potential to regulate gene expression, cell death, oxidative stress, cellular metabolism and signal transduction pathways. 16 Once released, exosomes can be internalized by neighbouring or distal cells, and equip cells with the ability to influence cellular functions. The term extracellular vesicles will be used with agreement with the international society for extracellular vesicles recommendation. 17,18 Research on exosome and small EV function in the eye has been limited. Although the origin of AH EVs is presently unknown, we assume that their source may be from the NPCE cells, where AH is actively secreted. NPCE cells contain a wide range of secretory vesicles in their cytoplasm, ranging from 10 nm to up to 1.5 μm in diameter. 9 Smaller-sized vesicles (10-100 nm) include translucent and dense-core secretory vesicles and may possibly represent EVlike nanoparticles. Previously, we provided evidence that NPCE cells secrete EVs that were internalized specifically by TM cells in culture. In addition, we demonstrated that incubation of TM cells with NPCE-derived EVs induced a significant decrease in the expression of the Wnt signalling pathways proteins and genes in human NPCE and TM cell lines. 19 The Wnt signalling pathway, one of the best-studied signalling cascades, has been implicated in POAG disease. While the underlying mechanism is still unclear, addition of exogenous sFRP1, a specific Wnt/β-catenin signalling inhibitor, to ex vivo perfusion cultured eyes, decreased out-flow facility and increased the IOP. 20 This active Wnt signalling is found to be involved in TM cell-mediated ECM expression and TM cell stiffening. 21 Overall, Wnt signalling appears to be a key player in TM regulation, with strong evidence suggesting that abnormal Wnt signalling may promote IOP.
Wnt signalling is known to crosstalk with other signalling mechanisms including Notch, NF-κB, mTOR, transforming growth factor β, protein kinase C and PI3K/AKT. 22 Among them, PI3K/AKT has been shown in several investigations to act upstream of Wnt signalling, resulting in β-catenin stabilization via phosphorylation on Ser 9 of GSK-3β . 23 Previous studies have supported a molecular intersection between components of the Wnt and PI3K/Akt signalling pathways. 24 Induced IOP elevation in rats has been shown to activate the PI3K/Akt pathway. 25 Along the cascade of the PI3K/AKT signalling pathway, an additional potential element involved in the regulation of the Wnt pathway is protein phosphatase 2A (PP2A). PP2A has been shown to negatively regulate Wnt pathway. 26,27 PP2A was found to be significantly higher in AH samples from the POAG group as compared to a cataract group. 28 After treatment with DEX, TGFβ2 alkaline phosphatases activity was reported to be increased in TM cells. Alkaline phosphatase activity was also found to be significantly higher in intact TM tissues from glaucomatous patients. 29 In the present study, we turned to a primary cell model and in-

| EV isolation
Dulbecco's modified Eagle's medium with EV-depleted FBS was prepared by ultracentrifugation of culture medium for 16 hours at 110 000 g overnight at 4°C using a SW28 rotor followed by filtration through a 0.22 µm filter (Millipore Express PLUS (PES) membrane).
EVs were isolated from conditioned media (CM) of cultured NPCE cell line or from CM of Primary NPCE cells using a polyethylene glycol (PEG) based approach as previously described. 30,31 Supernatant was collected from cells that were grown in medium containing EVdepleted FBS for 48 hours and was subjected to centrifugation steps at 300 g for 10 minutes and 2000 g for 10 minutes to remove cell debris and larger vesicles. The resulting supernatant was filtered using 0.2 µm filter and added to the filtered precipitation solution (50% PEG8000 (Sigma), 0.5 mol/L NaCl in PBS), to achieve final PEG concentration (8.3% w/v), and samples were then mixed by repeated inversion and incubated overnight at 4°C. EVs were precipitated by centrifugation at 1500 g for 30 minutes, and the supernatant was removed. Residual fluid centrifugation was eliminated by centrifugation at 1500 g for 5 minutes and pelleted EVs were resuspended in PBS for further analysis.

| Transmission electron microscopy at cryogenic temperature (cryo-TEM)
For cryo-TEM, 4 µL of the vesicle suspension was applied to a copper grid coated with a perforated lacy carbon 300 mesh (Ted Pella Inc) and blotted with filter paper to form a thin liquid film of solution.
The blotted sample was immediately plunged into liquid ethane at its freezing point (−183°C) in an automatic plunger (Lieca EM GP). The vitrified specimens were transferred into liquid nitrogen for storage.
Sample analysis was carried out under a FEI Tecnai 12 G2 TEM, at 120 kV with a Gatan cryo-holder maintained at −180°C. Images were recorded with the Digital Micrograph software package, at low-dose conditions, to minimize electron beam radiation damage. The measurements were performed at the Ilse Katz Institute for Nanoscale Science and Technology Ben-Gurion University of the Negev, Beer Sheva, Israel. Izon Science) were used to calibrate size and concentration, following the manufacturer's instructions. Samples were diluted 1000-fold with PBS buffer and measured over 10 minutes. The movement of the particle through the membrane was identified as change in the ionic stream causing voltage changes. The power of the signal is proportional to the particle size. According to the amount of particles and their velocity at a specific time, the qNano determines the EVs' concentration.   The refractive index of each fraction was measured at 22°C using an Abbe refractometer (Atago 1211 NAR-1T liquid) and converted to density using the conversion table published in the Beckman Coulter ultracentrifugation manual. Next, each fraction was washed in 20 mmol/L HEPES, ultra-centrifuged at 100 000 g for 1 hour and mixed with lysis buffer prior to analysis by Western blotting.

| Image stream analysis
Cellular uptake of the NPCE RNA and proteins via EVs was analysed using image stream. TM cells (5 × 10 5 ) were seeded in a 6-well plate in high glucose DMEM, containing 10% EV-depleted FBS. After 24 hours, the TM cells were incubated for 2 hours with Exo-Red (EXOC300A-1, SBI, CA, USA) or Exo-Green (EXOC300A-1, SBI) stained EVs isolated from NPCE cells. Labelling of exosomal RNA by Exo-Red and proteins by Exo-Green was performed according to the manufacturer's protocol. 10 μL of Exo-Red or Exo-Green was added to 20 μg of EVs resuspended in 500 μL 1×PBS. The EV suspension was mixed with the stain solution and incubated for 10 minutes at 37°C. The labelling reaction was stopped by adding ExoQuick-TC reagent. Labelled EVs were placed on ice for 30 minutes following centrifugation at 17136 g for 3 minutes. 100 μL of labelled NPCE EVs were co-cultured with TM cells for two hours washed and analysed on the ImageStreamX Mark II imaging cytometer equipped with 60× objective. Exo-Green and Exo-Red fluorescence were recorded using excitation with 488 nm and 642 nm lasers, respectively at 3 mW intensity.

| Microarrays analysis
Isolation and purification of RNA from NPCE cell line EVs were carried out using a miRCURY RNA isolation kit (Exiqon, Woburn, MA) according to the recommended procedure. RNA concentration and purity were determined at 260 nm by a NanoDrop2000 spectrophotometer (Thermo Scientific). RNA quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). miRNA expression analyses were performed at the Life Sciences Core Facilities, Israel Weizmann Institute of Science using SurePrint G3 Human miRNA 8 × 60k microarrays (Agilent Technologies) according to the manufacture's protocol.
Hybridization signals were detected with a DNA Microarray Scanner (Agilent Technologies), and the intensity of each hybridization signal was evaluated using Agilent Feature Extraction software.

| Statistical analysis
Data are presented as the means ± SD. Data were analysed using two-way ANOVA followed by Bonferroni's post-test. Statistical analysis was performed using GraphPad Prism version 5 software (GraphPad Software, Inc). Differences were considered statistically significant at P < .05.

| NPCE EV traffic RNA and proteins between NPCE and TM cells
To study whether encapsulated RNAs and proteins in NPCE EVs are transferred to TM cells, we stained exosomes with Exo-Red (RNA dye) or Exo-Green (protein dye). These were then added to the TM cells for 2 hours. Cell images were then collected by Image Stream Flow Cytometry. After 2 hours, both the Exo-Red ( Figure 1A) and Exo-Green ( Figure 1B) signals were detected in the cytoplasm of TM cells exposed to these labelled EVs. These results suggest that NPCE EVs were internalized by TM cells and effectively delivered their cargo into the TM cytosol.

| Characterization of exosomal proteins by LC-MS/MS
We further performed protein analysis of NPCE EVs derived from culture media to identify proteins that can potentially function as mediators of cell-to-cell signalling. Using the LC-MS technique, the combination of three independent exosome preparations resulted in the identification of 182 proteins. The 91 proteins detected in all three samples are shown in Table 1. This list included proteins known to be involved in cell adhesion, cytoskeleton regulation, oxidative stress response and others. Several of the identified proteins are associated with membrane trafficking and fusion processes. An assortment of cytosolic enzymes and various membrane transporters were also present. The list of proteins identified in this study is broadly consistent with that of other exosome proteome studies, further supporting the enrichment of EVs in samples prepared from NPCE supernatants. Along with our hypothesis that NPCE EVs participate in signalling within the drainage system, we also identified cellular signal transduction related proteins that were described previously as regulators of the Wnt signalling pathway (14-3-3 protein and histidine triad nucleotide-binding protein 1). 32,33 In addition, our proteomic analysis revealed the presence of CD9 in EVs released by NPCE cells. CD9, a member of the tetraspanin family, has been implicated in inhibition of Wnt/β-catenin signalling through the reduction of the cellular pool of β-catenin. 34 These results suggest that EVs carry proteins involved in Wnt signalling regulation, and the presence of these proteins further supports the idea that EVs may facilitating crosstalk between NPCE and TM cells by transferring their contents to target cells.

| NPCE exosome miRNA characterization
We demonstrated from the proteomics experiments the presence of RNA related proteins in EVs isolated from NPCE cells. To identify potential miRNAs that regulate genes involved in the regulation of the Wnt pathway, the miRNAs found in the microarray analysis were matched to previously published data of miRNAs that were reported in the AH of cataract patients. 33 When comparing our data vs. AH data, we found that over two-thirds of the most overlapped AH miRNAs detected in five studies were also presented in the NPCE EVs. By generating a Venn diagram, we found 77 miRNAs which were common between NPCE EVs and AH miRNAs (Figure 2A,B). As shown in Figure 2C, the most abundant of these 77 miRNAs identified in the NPCE EV samples was miR-21, followed by miR-638, let-7a, miR-100 and miR-16. Many of the abundant miRNAs in EVs isolated from NPCE cells have been reported to regulate Wnt signalling (Table 2), and other miRNAs such as miR-29, miR-17 and miR-21 have been related to the modulation of collagen synthesis. These results suggest that EVs carry miRNAs that have regulatory functions and can perhaps give rise to proteins or perform regulation on the recipient cells.

| Characterization of EVs isolated from primary NPCE cells
To analyse EVs isolated from primary NPCE cells, cryo-TEM and TRPS were performed. Using the cryo-TEM approach, we observed that primary NPCE cells generated typical round shape heterogeneous membrane encapsulated vesicles ( Figure 3A) Figure 3D).

| Effect of NPCE EVs on PDK1/Akt/mTOR signalling pathway in a TM cell line
Inhibitory serine-phosphorylation is the most frequently examined mechanism that regulates the activity of GSK3. It is well established that AKT inhibits GSK3 kinase activity via phosphorylation of Ser-9 in GSK3β. 35 A possible mechanism by which the NPCE EVs can mediate the inhibition of Wnt signalling pathway in TM cells is by decreasing the protein level of phosphorylated Akt protein. Akt has been shown to phosphorylate and inactivate GSK3β, which is required for the deactivation of β-catenin. Therefore, AKT protein level was measured in protein extracts derived from the TM samples using Western blot analysis with anti-AKT monoclonal antibody and anti-β-actin as control. As the full deactivation of Akt requires de-phosphorylation at both Thr308 and Ser473, 36 we examined two-site phosphorylation in TM cells following NPCE EV treatment at different time-points (1,2,4 hours).
As controls, we used untreated TM cells and TM cells that were exposed to RPE EVs, which we had previously shown to be unable to reduce β-catenin levels in TM cells. When the TM cells treated with NPCE EVs for 2h, we observed greater than a threefold decrease (P < .05) of Akt phosphorylated at Ser473 compared with untreated cells (Figure 5A,B). A trend of lower Akt phosphorylation levels at Thr308 was observed upon NPCE exposure, without reaching significance ( Figure 5C,D). No differences were detected in the expression levels of Akt phosphorylation at both the Ser473 and Thr308 positions upon RPE EV exposure, as compared to the untreated control. In the attempt to better define the molecular mechanism underlying the ability of EVs to modulate Akt protein we further investigated the phosphorylation state of proteins that lies upstream (PDK1) and downstream (mTORC1) of Akt in TM cells following EV treatment. In response to NPCE EVs at 2 hours, TM cells exhibited 2.5-fold (P < .05) decreased phosphorylation of PDK1 but pre-treatment of TM cells with RPE EVs had no effect on the amount of PDK1 at the respective time-point ( Figure 5E,F).
Next, we examined how the presence or absence of NPCE-derived EVs affects mTORC1 phosphorylation. Results revealed that no significant differences were detected following NPCE and RPE EV treatment regarding their effects on phosphorylation of the mTORC1 at Ser2448 (Figure 5G,H). Collectively, these results suggest that NPCE EVs seem likely to play some role in phosphorylation of Akt and Pdk1 proteins. However, as EVs were unable to inhibit mTORC1 phosphorylation by Akt protein, and Akt was partially activated by phosphorylation of T308, Akt protein in our system cannot account for the reduced phosphorylation of p-GSK3β in TM cells. protein levels as compared to control cells ( Figure 6A,B). Next, we investigated whether the NPCE EVs are positive for PP2A protein.

| Effect of NPCE EVs on PP2A protein levels in a TM cell line
Western blot analysis of density gradient fractions confirmed the presence of PP2A in NPCE EVs ( Figure 6C). We observed, using a continuous sucrose gradient, that most of the PP2A was recovered in fractions 4 to 6, ranging from densities of 1.18-1.188 g/ mL, corresponding to the density range classically assigned to EVs. 39 These results suggest that NPCE cells secrete EVs containing PP2A protein and thus may be responsible for the elevation levels of PP2A protein in TM cells and downregulation of Wnt signalling by PP2A protein delivery and activation of GSK3β in recipient cells.

| NPCE EV effect on collagens protein levels in TM cells
Trabecular meshwork tissue consists of various extracellular matrix (ECM) proteins, including collagen (types I, III, IV and V), proteoglycans and laminin. 40 Changes in ECM are thought to have a role in the increased out-flow resistance of the TM in POAG. 41 Activation of the Wnt signalling pathway in humans has been shown to promote fibrosis of various organs. 42 We have been suggested that down-regula-   including collagens and genes involved in ECM deposition. 43 We found that members of 29-miRNA family were abundant in NPCE exosomal samples and that treatment of TM cells with NPCE EVs significantly decreased the expression of the COL3A1. In addition, fifteen miRNAs were found to be either potent activators or inhibitors of Wnt signalling.

| D ISCUSS I ON
Taken together, these data indicate that EVs released by NPCE cells may regulate Wnt signalling in TM tissue. Regarding the Wnt signalling pathway attenuation in TM cells by NPCE-derived EVs, the cell lines used are relevant as they match primary cells. The exact mechanism that underlies the inhibition of Wnt signalling following NPCE exosome treatment in TM cells remains incompletely understood. We propose that one potential mechanism for the observed F I G U R E 5 Effect of NPCE cell line EVs on activity of AKT, PDK1 and mTOR proteins. Experimental conditions included untreated TM cells and TM cells treated with either control (RPE EVs) or NPCE EVs for the indicated times. Total cell lysates were prepared from TM cells and subjected to the western blot analyses with p-AKT, t-AKT, p-PDK1, p-mTORC1 and t-mTORC1 antibodies. Actin was used to verify equal loading. Representative western blots demonstrate (A) pAKT 473, (c) pAKT 308, (E) p-PDK1 (Ser 241) and (G) p-mTORC1 (Ser 2448) protein levels. Densitometric evaluation of (B) p-AKT 473, (D) p-AKT 308, (F) p-PDK1 and (H) t-mTORC1 protein expression. For all graphs, data are means ± SEM of at least three independent experiments. (*P < .05, **P < .01 and ***P < .001 in two-way ANOVA with Bonferroni's test) of 89% identified AH miRNAs in NPCE EVs suggests that NPCE tissue may be the pivotal source of miRNAs found in AH. Our understanding is that exosome miRNAs may target genes in TM cells and contribute to the ocular drainage system homeostasis by modulating levels of β-catenin and levels of several ECM proteins such as collagen. We believe that exosomal cargo may vary in response to different physiological or pathological conditions, and AH out-flow system may utilize multiple miRNAs, creating a sophisticated balance between activation and repression of Wnt signalling.
Accumulating evidence suggests that the Wnt pathway is a major signalling pathway contributing to the TM AH drainage resistance. 44,45 It was shown that Wnt antagonist increases out-flow facility in perfused human anterior segments. 20 Wnt signalling components were found in different ocular drainage tissues, among them sFRP-1, an antagonist of the Wnt signalling pathway. Inhibition of the Wnt pathway using sFRP1 increases IOP in vivo in mice animal models. 46 Wnt signalling pathway has a complex relation with the ECM. It intersects with TGF-beta pathway to regulate ECM gene expression in multiple cell types, and plays a role in ECM assembly. 47 Evidence from open-angle glaucoma studies suggests a link between Wnt inhibition and increased TM cell stiffness. 48  EVs contain significant amounts of miR-638, which was linked to the down-regulation of the Wnt/β-catenin signalling pathway. 14 In addition, miR-21 was detected in NPCE cells derived EVs at high concentration compared with other miRNAs detected. miR-21 is indeed involved in the Wnt/β-catenin signalling pathway through its upstream target genes and positive regulation of Wnt, thereby activating the Wnt/β-catenin signalling pathway and causing β-catenin expression to increase. 50 miR-21 could negatively modulate the activity of Wnt/β-catenin signalling via targeting Wnt-1, which likely accounts for the Wnt/β-catenin cascade activation. 51 Furthermore, miR-21 was shown to target TIMP3, a tissue inhibitor of MMPs, promoting MMP activity, leading to degradation of ECM. 52 Another F I G U R E 6 Expression of PP2A in TM cells and NPCE EVs. Western blot showing increased PP2A protein levels in the TM cells after NPCE EV exposure. A, Western blot representative images. B, Quantification of PP2A in TM cells exposed to NPCE or RPE EVs or untreated cells. Results are representative of three independent experiments (*P < .05; ***P < .001 in two-way ANOVA with Bonferroni's test). C, NPCE EVs were floated into a linear sucrose gradient (0.25-2.5 mol/L) and then subjected to overnight centrifugation. Gradient fractions were collected and analysed by Western blot analysis using PP2A antibody recent study on human tissues suggests that miR-21 promotes ECM degradation through inhibiting autophagy via the PTEN/Akt/mTOR signalling pathway. 53 Along the cascade of PI3K/AKT signalling pathway, potential elements involved in the regulation of WNT pathway is PP2A. PP2A was found to be significantly higher in AH samples from the POAG group as compared to a cataract group. 28 After treatment with DEX, TGFβ2 alkaline phosphatases activity was reported to be increased in TM cells. Alkaline phosphatases activity was also found to be significantly higher in the intact TM tissues from glaucomatous patients. 29 PP2A has been shown to negatively regulate signalling networks such as the Wnt pathway. Yokoyama  In conclusion, our findings propose exosome-mediated crosstalk between NPCE cells and TM cells that affect the behaviour of target TM cells through modulation of Wnt signalling pathway and ECM remodelling, probably, by the transfer of proteins and RNA cargo.
Exploration of the precise physiological roles of NPCE EVs may allow the development of novel therapeutic approaches, with minimally invasive procedures.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.