Micro-RNAs Shuttled by Extracellular Vesicles Secreted from Mesenchymal Stem Cells Dampen Astrocyte Pathological Activation and Support Neuroprotection in In-Vitro Models of ALS

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease with no effective cure. Astrocytes display a toxic phenotype in ALS and contribute to motoneuron (MN) degeneration. Modulating astrocytes’ neurotoxicity can reduce MN death. Our previous studies showed the beneficial effect of mesenchymal stem cell (MSC) administration in SOD1G93A ALS mice, but the mechanisms are still unclear. We postulated that the effects could be mediated by extracellular vesicles (EVs) secreted by MSCs. We investigated, by immunohistochemical, molecular, and in vitro functional analyses, the activity of MSC-derived EVs on the pathological phenotype and neurotoxicity of astrocytes isolated from the spinal cord of symptomatic SOD1G93A mice and human astrocytes (iAstrocytes) differentiated from inducible neural progenitor cells (iNPCs) of ALS patients. In vitro EV exposure rescued mouse and human ALS astrocytes’ neurotoxicity towards MNs. EVs significantly dampened the pathological phenotype and neuroinflammation in SOD1G93A astrocytes. In iAstrocytes, exposure to EVs increased the antioxidant factor Nrf2 and reduced reactive oxygen species. We previously found nine miRNAs upregulated in MSC-derived EVs. Here, the transfection of SOD1G93A astrocytes with single miRNA mimics reduced astrocytes’ activation and the expression of neuroinflammatory factors. Moreover, miR-466q and miR-467f mimics downregulate Mapk11, while miR-466m-5p and miR-466i-3p mimics promote the nuclear translocation of Nrf2. In iAstrocytes, transfection with miR-29b-3p mimic upregulated NQO1 antioxidant activity and reduced neurotoxicity towards MNs. MSC-derived EVs modulate astrocytes’ reactive phenotype and neurotoxicity through anti-inflammatory and antioxidant-shuttled miRNAs, thus representing a therapeutic strategy in ALS.


In Vitro Treatment with Extracellular Vesicles
After EV isolation, MSCs were detached with Trypsin-EDTA 1X and counted by a Neubayer Chamber to set up a standard protocol for the number of EVs to be added to the astrocyte media during in vitro treatment. In each experiment, the ratio of 1:6 between cultured astrocytes and EV-generating MSCs was maintained for in vitro treatments. Accordingly, we used a number of EVs generated by 6 × 10 5 or 2 × 10 4 MSCs for the treatment of astrocytes plated on 35 mm Petri dishes or 24-well plates, respectively.
Mouse SOD1 G93A astrocytes and iAstrocytes were treated with MSC-derived EVs in DMEM without FBS or knockout serum, respectively. After 24 h treatment, astrocytes were collected, pelleted, and stored for WB and RT-qPCR experiments, or fixed with paraformaldehyde (PFA) 4% in PBS for immunofluorescence. For MN viability experiments, the medium of mouse or human astrocytes was removed, and MNs were seeded on confluent astrocytes previously treated (24 h) or not with EVs.

Mouse Spinal Cord Primary Astrocyte Cell Culture Preparation
Spinal cord astrocyte primary cell cultures were prepared from late symptomatic (120-day-old) adult SOD1 G93A and age-matched WT mice as previously described [53,54]. In detail, late symptomatic (120-day-old) SOD1 G93A mice and age-matched WT mice were euthanized by cervical dislocation by well-trained personnel and spinal cords were rapidly removed. The dissected tissue was mechanically chopped with a scalpel and the chopped spinal cord was dispersed and further fragmented in Dulbecco's Modified Eagle Medium (DMEM; Euroclone, S.p.A, Milan, Italy, Cat# ECM0728L) containing 10% Fetal Bovine Serum (Euroclone, Cat# ECS0180L), 1% glutamine (Euroclone, S.p.A, Milan, Italy Cat# ECB3004D) and 1% Penicillin/Streptomycin (Euroclone, Cat# ECB3001D); then, the preparation was seeded at the optimal density in two 35 mm Petri dishes, pre-coated with poly-L-ornithine hydrochloride (1.5 µg/mL; Merck, Milan, Italy, Cat# P2533) and laminin (3 µg/mL; Merck, Milan, Italy, Cat# L2020). The preparation was placed at 37 • C in humidified 5% CO 2 incubator; then, after 5 days, the medium containing tissue fragments was replaced with fresh complete DMEM. After 7DIV, adherent astrocytes were detached by Trypsin-EDTA 1× (Euroclone, S.p.A, Milan, Italy, Cat# ECB3052B) and re-plated on pre-coated 35 mm Petri dishes to obtain the homogeneous confluence at the optimal density of 10 5 cells. The purity of spinal cord adult astrocyte cell cultures was checked with flow cytometry and immunofluorescence as previously described [54]. For immunofluorescence (IF) analyses, astrocytes were also re-plated on pre-coated glass coverslips placed at the bottom of 24-multiwell plates, at a density of 3 × 10 4 cells per well. After 20DIV, astrocytes were treated for 24 h with Evs or transfected for 48 h with synthetic mimics; after treatment, astrocytes were finally detached and collected for Western blot and RT-qPCR experiments, or fixed with 4% PFA (Merck, Milan, Italy" Cat# 47608) for IF studies.

Mouse Spinal Cord Motor Neuron Preparation and MN/Astrocyte Co-Cultures
MNs were isolated from the spinal cord of SOD1 G93A E13.5 mouse embryos as previously described with some modifications [55]. Briefly, the spinal cord was isolated from Cells 2022, 11, 3923 5 of 29 embryos by microscopy dissection, then meninges and the dorsal root ganglia were removed. The tissue was digested with 0.5% trypsin (Merck, Milan, Italy, Cat# T4799) in Hank's Balance Salt Solution (HBSS) for 20 min at 37 • C. Then, trypsin solution was replaced with a mix of 0.4% BSA (Merck, Milan, Italy, Cat# A3311) in Leibovitz-15 medium (Merck, Milan, Italy, Cat# L5520) and 0.02 mg/mL Deoxyribonuclease I (DNAse; Merck, Milan, Italy, Cat#DN25) and gently triturated. The tissue homogenate was stratified on 6.2% OptiPrep (Merck, Milan, Italy" Cat# D1556) cushion and centrifuged 500× g for 15 min at room temperature. After centrifuging, the MN-enriched cell population was localized at the interface between the OptiPrep solution and the medium. The MN band was collected and re-suspended in MN medium, composed of neurobasal medium (Thermo Fisher Scientific, Monza, Italy, Cat# 21103-049), 2% B27 supplement (Thermo Fisher Scientific, Monza, Italy, Cat# 17504044), 2% horse serum (Thermo Fisher Scientific, Monza, Italy, Cat# 16050130), 0.5 mM stable L-Glutamine (Thermo Fisher Scientific, Monza, Italy, Cat#35050038), 25 µM Mercapto ethanol (Sigma-Aldrich, Cat# M6250), 10 ng/mL ciliary neurotrophic factor (CNTF; Merck, Milan, Italy, Cat# C3835), 100 pg/mL glial-derived neurotrophic factor (GDNF; Merck, Milan, Italy, Cat# G1401), and 5 µg/mL Penicillin/Streptomycin. MN suspensions were centrifuged at 75× g for 20 min and the pellet was suspended in 1 mL of MN medium plus 50 µL of Chick Embryo Extract (US Biological, Salem, Massachusetts, United States, Cat# C3999). MNs were seeded at a density of 5 × 10 4 MNs in a 35 mm Petri dish on confluent adult astrocytes prepared from 120-day-old SOD1 G93A mice, previously treated or not with EVs. Confocal microscopy representative images showing MN/astrocyte co-cultures labelled with specific markers for astrocytes and MNs are reported in Supplementary Figure S1 (panels a-d). After 3 days, and then three times a week, the medium was replaced with fresh MN medium. To assess MN viability, MN were counted in an area equal to 1 cm 2 (exploiting a 10 × 10 mm grid pre-designed at the bottom of the Petri dish) from day 4 after seeding. The average number of MNs at the count, starting from day 4, in a 1 cm 2 square area, was around 700 ± 200 MNs. The number of MNs was recorded three times a week for 14 days. The percentage of surviving MNs at each time-point was calculated as % vs. the total number of MNs, counted in the same 1 cm 2 square area of the respective dish, at the day 4 of co-culture (starting day of the cell count experiment, reported as 100% of total MNs).

iAstrocytes Preparation and iAstrocyte/MN Co-Cultures
Induced neural progenitor cells (iNPCs) were reprogrammed from the fibroblast of ALS patients (C9orf72 patients reported as C9, or patients carrying the SOD1 A4V mutation, reported as SOD1) or control donors (reported as CTR), as previously reported [56], and banked. The details of the donors are listed in Supplementary Table S1. iNPCs were grown on fibronectin-coated plates and were fully differentiated into iAstrocytes on day 7 after plating. Immunofluorescence representative images showing iAstrocytes differentiated from iNPCs, of healthy donors and ALS patients, stained for cell identity markers, are reported in Supplementary Figure S2 (panels a,b). iAstrocytes were maintained in 37 • C and 5% CO 2 in DMEM (Fisher Scientific, Milan, Italy) supplemented with 10% fetal bovine serum (FBS) (Life science production), and 0.2% N-2 (Gibco).
Hb9-GFP-positive motor neurons were prepared from mouse embryonic stem cells (mESC) containing a GFP gene under the Hb9 motor neuron promoter, a kind gift from Prof. Thomas Jessell (Columbia University, New York, NY, USA). Hb9-GFP mESCs were maintained on a mouse embryonic fibroblast (Merck, Burlington, MA, USA) feeder layer in mESC media (KnockOut-DMEM, 15% (v/v) embryonic stem cell FBS, 2mML-glutamine, 1% (v/v) nonessential amino acids (from Thermo Fisher Scientific, Waltham, United States) and 0.00072% (v/v) β-mercaptoethanol (Merck, Milan, Italy). mESCs were then differentiated into MN-enriched cultures via embryoid bodies (EBs). Briefly, mESCs were lifted with trypsin, resuspended in EB medium (DMEM/F12), 10% (v/v) knockout serum replacement, 1% N 2 , 1 mM L-glutamine (from Thermo Fisher Scientific, Waltham, United States), 0.5% (w/v) glucose and 0.0016% (v/v) β-mercaptoethanol and seeded into non-adherent Petri For co-culture, day 5 iAstrocytes were re-seeded in Greiner 96-wellplates at a density of 10,000 cells/well, followed by addition of treatment (EVs or miRNA) on day 6 in serum-free media. On day 7, EBs were dissociated with 200 U/mL papain (Merck, Milan, Italy) and GFP+ neurons were seeded into 96-wellplates containing iAstrocytes. Immunofluorescence representative images showing iAstrocytes/MNs co-cultures, stained for cell identity markers, are reported in Supplementary Figure S2 (panel c). For experiments assessing motor neuron survival, co-culture plates were imaged on days 1 and 3 using the InCell high content microscope and the number of Hb9-GFP+ neurons with axons was quantified in the whole well. Images obtained with InCell were used for semi-automated high-throughput analysis with Columbus microscopy software [57]. The percentage of Hb9-GFP+ neurons with axons was calculated as % of the total number of MNs at day 1 over day 3. Typically, MN survival at day 3 on healthy control iAstrocytes was between 65-75%, while on control astrocytes we typically observed a decrease between 50-70% depending on the toxicity of the individual ALS donor. To reduce inter-week variation, the data were normalised to the average of the healthy control co-culture in the same 96-well plate = 100%. Each iAstrocyte line was assessed in 3 independent experiments (i.e., from 3 independent iNPC differentiations at different passages) and 3 technical replicates were run in each experiment.

Immunofluorescence Experiments
Adult astrocytes cultured from SOD1 G93A (120-day-old) and age-matched WT mice, seeded on 12 mm diameter glass coverslips, at the bottom of 24-multiwell plates, were fixed with 4% PFA (Merck, Milan, Italy, Cat# 47608). After fixing, cells were permeabilized with methanol for 5 min at −20 • C. Bovine serum albumin (BSA) 0.5% in PBS, for 15 min at room temperature, was used to arrest the process. All primary antibodies were diluted in 3% PBS-BSA blocking solution. Primary antibodies were incubated overnight at 4 • C (Supplementary Table S3). The day after, cells were washed in 0.5% PBS-BSA before the incubation of 1 h at room temperature with the secondary antibody (Supplementary  Table S3). Secondary antibodies were diluted 1:3000 in PBS containing 3% BSA. Then, cells were washed in PBS and the coverslip was assembled on a microscopy glass slide through Fluoroshield TM with DAPI (Merck, Milan, Italy, Cat# F6057). Fluorescence image (512 × 512 × 8 bit) acquisition was performed by a three-channel Leica TCS SP5 laserscanning confocal microscope equipped with 458, 476, 488, 514, 543 and 633 nm excitation lines through a plan-apochromatic oil immersion objective 63×/1.4. Light collection was optimized according to the combination of the chosen fluorochromes, and sequential channel acquisition was performed to avoid crosstalk. The Leica "LAS AF" software package was used for image acquisition. The quantitative analyses related to immunofluorescence confocal microscopy studies were performed by calculating the co-localization coefficients of the proteins of interest, according to Manders' and Costes' theories [58,59], thus allowing a direct quantitative correlation between the intensity of co-localization of the protein of interest with respect to a stable housekeeping protein. Quantitative results were expressed as the relative co-localization intensity of the protein of interest, respect to the stable reference housekeeping protein 3-phosphate dehydrogenase glyceraldehyde (GAPDH). All the immunofluorescence studies were assessed by performing at least 3 independent experiments, run in triplicate (i.e., 3 wells per experiment). Each acquired image included an average of at least 6 to 12 cells homogenously distributed, thus resulting in 50 to 100 assessed cells for each sample, in each experiment.
To check MN/astrocyte co-culture purity (see Supplementary Figure S1), the cells were seeded on 12 mm diameter glass coverslips at the bottom of 24-multiwell plates; after 8 days of co-culture, cells were fixed with 4% PFA (Merck, Milan, Italy, Cat# 47608). After fixing, cells were permeabilized with methanol for 5 min at −20 • C. Bovine serum albumin (BSA) 0.5% in PBS, for 15 min at room temperature, was used to arrest the process and saturate the non-specific binding sites. MN/astrocyte co-cultures were stained with cell-specific primary antibodies for astrocytes (rabbit anti-GFAP plyclonal antibody, Merck, Milan, Italy, # HPA056030, 1:1000) and motor neurons (chicken anti-beta tubulin III polyclonal antibody, Abcam #ab41489, 1:1000; mouse anti-ChAT monoclonal antibody, Merck, Milan, Italy #AMAB91130, 1:500; rabbit anti-HB9 polyclonal antibody, Thermo Fisher #PA5-23407, 1:500; rabbit anti-Islet1 polyclonal antibody Abcam #ab20670, 1:500). All primary antibodies were diluted in 3% PBS-BSA blocking solution. Primary antibodies were incubated overnight at 4 • C. The day after, cells were washed in 0.5% PBS-BSA before incubation of 1 h at room temperature with the secondary antibody conjugated with specific fluorophores. Secondary antibodies were diluted 1:3000 in PBS containing 3% BSA. Then, cells were washed in PBS and the coverslip was assembled on a microscopy glass slide through Fluoroshield TM (Merck, Milan, Italy, Cat# F6057). Confocal microscopy fluorescence images were acquired as described above for spinal cord astrocyte primary cultures.

Western Blot
Human. On day 5, iAstrocytes were re-seeded into 6-well plates at 250,000 cells/well, treated with EVs in serum free media on day 6, and full media was replaced 24 h later to allow cells to recover for another 24 h. Cells were scraped on day 7, pelleted and lysed on ice with immunoprecipitation lysis buffer (150 mM NaCl, 50 mM HEPES, 1 mM EDTA, 1 mM DTT, 0.5% (v/v) Triton X-100, pH 8.0) with protease inhibitor cocktail for 15 min. Lysates were centrifuged at 17,000× g for 5 min at 4 • C, and supernatants quantified for protein content with Bradford assay measured with a WPA S1200 Diode Array Spectrophotometer (Biochrom, Nottingham, UK). Protein was diluted in 4X Laemmli buffer (228 mM Tris-HCl, 38% (v/v) glycerol, 277 mM SDS, 0.038% (w/v) bromophenol blue, 5% (v/v) βmercaptoethanol, pH 6.8) and immunoprecipitation buffer, and samples were boiled at 93 • C for 5 min. A total of 15 µg of protein or 2 µL of Blu-Eye pre-stained ladder was loaded in each well of 12% SDS-acrylamide gels, and gels were run at 120V using a mini-PROTEAN Tetra Handcast systems (Bio-Rad, Hercules, California, USA). Semi-dry transfer was completed onto a nitrocellulose membrane, in transfer buffer (47.9 mM Tris, 38.6 mM glycine, 1.38 mM SDS, 20% (v/v) methanol) in a semi-dry transfer apparatus applying 0.15 A per membrane for 1 h at room temperature. Membranes were blocked in 5% milk in Tris-buffered saline/0.1% Tween-20 (TBST) for 30 min, before incubating in primary antibodies NLRP3 (1:100, Abcam, Cambridge, UK, Cat# ab210491) and p-p65 (1:200, Cell Signalling, Leiden, The Netherlands, Cat#3033S) in 5% milk/TBSTat 4 • C overnight on rollers. Anti-rabbit or anti-mouse-HRP secondary antibodies (Promega, Milan, Italy, W4011 and W4021, respectively) were diluted 1:5000 in 5% milk/TBST and incubated with the membrane in room temperature for 60 min under agitation. Membranes were incubated in enhanced chemiluminescence (ECL) developer (Pierce) for 30 s before imaging with the Gbox imaging system (Syngene, Karnataka, India). Protein expression was quantified by band densitometry using GeneTools software (Syngene, Karnataka, India) and normalised to GAPDH as a loading control.

ELISAs and Cytokine Multiplexing
Human. iAstrocytes in monoculture were treated with EVs in serum-free media on day 6, followed by washing and replacing with full media on day 7 and collecting media and pelleting cells on day 8. Media was centrifuged at 2000× g for 5 min to pellet debris, and the supernatant was aliquoted and snap-frozen in liquid N 2 and stored at −80 • C until use. Human TNF-α and IL-1β Quantikine ® ELISA kits were purchased (R&D systems, Minneapolis, Canada, United States) and used according to manufacturer's instructions. Plates were read using a Pherastar spectrophotometer. For multiplex measurement of IL-6 and CCL-2, BD™ Cytometric Bead Array (CBA) kits were used according to manufacturer's instructions.
Mouse. Mouse astrocytes, cultured in 35 mm Petri dishes, were treated with MSCderived EVs for 24 h in serum-free DMEM. DMEM containing EVs was replaced with complete fresh DMEM, and astrocyte-conditioned medium was collected 24 h after EVs withdraw. Astrocyte-conditioned medium was centrifuged 2000× g for 5 min. Then, IL-1β, TNF-α, IL-6 and CCL2 concentrations were measured with a specific enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Minneapolis, Canada, United States Cat# DY401, DY410, DY406 and DY479, respectively) according to the manufacturer's protocol.

Astrocyte Transfection with miRNA Mimics and Quantitative RT-PCR
Human and mouse synthetic miRNA mimics were used to transfect (for 48 h) iAstrocytes or SOD1 G93A mouse spinal cord astrocytes using the HiPerFect ® Transfection Reagent (Qiagen, Manchester, UK), according to the manufacturer's instructions. The list and sequence of the synthetic mimics are reported in Supplementary Table S2. Human. Synthetic miRNA mimic hsa-miR-29b3P was obtained from Ambion ® Pre-miR™ (ThermoFisher, Waltham, United States,). The miRNA mimic was resuspended in DEPC-treated water and stored at −20 • C until further use. On day 6, iAstrocytes were treated with synthetic miRNA by adding the miRNA mimic to serum-free DMEM to a final concentration of 50 nM, along with transfection reagent HiPerfect (Qiagen, Waltham, United States,) used according to manufacturer's instructions. The mixture of media, miRNA and HiPerfect was briefly vortexed, allowed to incubate for 5-10 min at room temperature, and added onto the cells drop-wise. Full media was added onto the cells 6 h later, and cells were kept at 37 • C until the endpoint 24 h after transfection. For miRNA transfection validation, RNA was extracted with a Direct-zol RNA isolation kit (Zymo Research, Cambridge, UK) and miRNA content quantified with the small RNA kit in Bioanalyser (Agilent, Technologies LDA UK Limited, Cheshire, UK). For quantification of downstream genes with RT-qPCR, RNA was extracted from n = 3 independent experiments with the RNeasy kit (Qiagen) and concentration were quantified with Nanodrop. A total of 400 ng of RNA was converted to cDNA with the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, United States) in a Stratagene thermocycler. cDNA was stored at −20 • C until use. SYBR Green qPCR Master Mix (Low ROX-Bimake, Munich, Germany, Cat# B21702) was pipetted in low profile 96-wellplates, loading 20 ng of cDNA and 5 µM of forward and reverse primers (Merck, Life Science UK Limited, Gillingham, UK) in technical replicates. Plates were loaded onto a CFX 96TMReal-Time System (Bio-Rad) and fluorescence for SYBR was measured. Quantitative RT-PCR data were analysed using CFX Manager 3.1 (Bio-Rad, Hercules, California, USA) GraphPad Prism and Sigma Plot/Sigma Stat software.
Mouse. In total, 7 × 10 4 or 2 × 10 5 astrocytes were plated in 24-well plates or 35 mm-Petri dishes, respectively, in DMEM w/o FBS and transfected using the HiPerFect ® Transfection Reagent (Qiagen, Milan, Italy), according to the manufacturer's instructions, with mimics specific for miRNA (miRNA Mimic miRNA, Qiagen, Milan, Italy) and with MIS-SION miRNA Mimic Negative Control (Merck, Milan, Italy,), a synthetic miRNA which does not recognize any mRNA target in cells (Cneg), and with iBONi siRNA positive control-P4M (Riboxx), which inhibits the translation of GAPDH in cells, as an indicator of efficient transfection. After 48 h incubation at 37 • C in humidified 5% CO 2 , astrocytes were fixed with 4% PFA for confocal microscopy or were detached with Accutase ® (Euroclone S.p.A, Milan, Italy, Cat# ECB3056D) and the cell pellet re-suspended in 500 µL of QIAzol Lysis reagent (Qiagen, Milan, Italy, Cat# 79306) for qPCR experiments. The sequence of mimics used are summarised in Supplementary Table S2. Total RNA was isolated from astrocytes using QIAzol Lysis Reagent according to the manufacturer's instructions. First-strand cDNA was synthesized with 500 ng of total RNA from primary astrocytes using QuantiTect Reverse Transcription kit (Qiagen, Milan, Italy), in a final volume of 20 µL.

CellROX Reactive Oxygen Species Probe
Human iAstrocytes were re-seeded in black clear-bottom 96-well plates (Greiner) on day 5. On day 6, cells were incubated with Evs in serum-free DMEM for 24 h at 37 • C, followed by washing out EVs and replacing with full-serum DMEM for 24 h. On day 8, CellROX ® Orange (Thermo Fisher Scientific, Waltham, United States, Em 565 nm) was added to a final concentration of 5 µM at 37 • C for 30 min; the cells were washed with PBS and the medium replaced with phenol-red-free DMEM media. Live cells were imaged with an Opera Phenix high-content imager (PerkinElmer) and analysed with the Columbus Image Data Storage and Analysis System™ version 2.8 (PerkinElmer, Buckinghamshire, UK). The software allows for the generation of a cell mask that separates individual cells and calculates fluorescence intensity against the background signal. Fluorescence intensity per pixel in each well was calculated and these values were normalised to the average of the 2 wells containing untreated healthy control fluorescence intensity, set as 100 in each experiment. All samples were assessed in 3 independent experiments in duplicate (i.e., 2 wells per experiment). Whole-well image analysis was performed in each experiment, thus resulting in >1000 cells assessed for each sample in each experiment per cell donor.

Immunofluorescence Quantification of Nrf2 and NQO1
Mouse. Mouse-derived spinal cord astrocytes cultured from adult SOD1 G93A (120day-old) and age-matched WT mice, were seeded on 12 mm diameter glass coverslips, at the bottom of 24-multiwell plates, treated or not with MSC-derived EVs or miRNA mimics, were fixed and stained with specific primary and secondary antibodies as detailed above to study Nrf2 cellular-specific localization (nuclear/cytoplasm ratio). The nuclear Nrf2 signal was quantified using DAPI as a regional mask to separate nucleus from cytoplasm regions, then the Nrf2 signals, subtracting the background signal outside the cells, were selectively quantified in these two regions, and the ratio between Nrf2 (nucleus) over Nrf2 (cytosol) fluorescence intensity was calculated and reported as quantitative analyses.
Human. On day 6, iAstrocytes in 96-well plates were treated with EVs for 24 h followed by a 24 h wash-out step, as described in previous sections. Cells were fixed at day 8 with 4% paraformaldehyde for 10 min, followed by PBS washing and blocking with 5% donkey serum in 0.05% Triton-X/PBS for 30 min. Anti-nuclear factor erythroid 2-related factor 2 (Nrf2) (rabbit polyclonal, Abcam ab31163) and goat anti-NAD(P)H Quinone Dehydrogenase 1 (NQO1) (Abcam ab2346) antibodies were added at a dilution of 1:200 in 1% donkey serum/0.05% Triton-X/PBS and the cells were incubated overnight at 4 • C, washed with PBS, and incubated with goat anti-rabbit-488 (Abcam, Cambridge, UK, ab150077) and donkey anti-goat-555 (Abcam, Cambridge, UK, ab175704) antibodies, diluted at 1:500 at room temperature for 60 min. Cells were washed and stained with Hoechst 33342 (MedChem Express LCC, Monmouth Junction, United States) and imaged with an Opera Phenix high content imager. Images were analysed with Columbus software, in which cell masks were created and fluorescence intensity was quantified in the nuclei and cytoplasm. Immunofluorescence data were normalised per cell and reported as % of the untreated healthy control. Nuclear Nrf2 signal was quantified using DAPI as a regional mask, while NQO1 was quantified in the cytoplasm excluding the nuclear area. As described above, >1000 cells for each sample per experiment, per cell donor, were assessed.

Statistics
The Sigma Stat Software (Version 3. The threshold for statistical significance (P) was set at p < 0.05. Data are always presented as mean ± standard error of mean (SEM).

MSC-Derived EVs Reduce the Toxicity of Mouse-and Human-Derived ALS Astrocytes towards MNs
Mouse-or human-bone-marrow-derived MSCs were treated with interferon-γ (IFNγ; 24 h) to activate their immunosuppressive and neuroprotective functions [60]. We have previously detected the morphology and purity of mouse-MSC-derived EVs [45]. The human-MSC (hMSCs)-derived EVs, in resting conditions and after IFN-γ priming, displayed typical vesicle biogenesis markers that increased, albeit not significantly, after exposure to IFN-γ ( Figure S1a). The size distribution analysis produced prominent peaks between 100 and 175 nm, confirming the achievement of high-concentration EV preparations ( Figure S1b).
We set up mouse primary astrocyte-MN co-cultures to determine whether EVs have a neuroprotective effect on ALS MNs. We pre-treated adult SOD1 G93A astrocytes with MSC-derived EVs for 24 h; after wash-out, we seeded WT or SOD1 G93A mouse embryonic spinal cord MNs on WT untreated, SOD1 G93A EV-treated or SOD1 G93A untreated astrocytes (Figure 1a-c) and assessed the viability of MNs over 4-14 days of co-culture (Figure 1d). On day 4, the MN number between WT, SOD1 G93A treated and untreated conditions was comparable; however, from day 6 onwards, MNs co-cultured with untreated SOD1 G93A astrocytes displayed a constant viability decline [day 4: 100% viability; day 14: 4.9% viability]. Conversely, the number of MNs co-cultured with EV-treated SOD1 G93A astrocytes pretreated with EVs was higher at each time point than MNs seeded on untreated astrocytes (Figure 1d).  We performed similar experiments using iAstrocytes in a phenotypic screen standardised with MN survival as the primary read-out for iAstrocyte toxicity [57]. We differentiated patient-derived iAstrocytes from induced neural progenitor cells (iNPCs) reprogrammed from skin fibroblasts of two healthy controls (CTR155, CTR3050), two C9orf72 patients (pat78, pat183), and two SOD1 patients (pat100, pat102) over our standardised 7-day differentiation protocol [56]. As to mouse astrocytes, iAstrocytes were pre-treated for 24h with EVs isolated from hMSCs stimulated with IFN-γ and co-cultured with murine Hb9-GFP+ motor neurons seeded on top (Figure 1e,f). Confirming our previous findings [30,56], iAstrocytes from C9orf72 and SOD1 patients were toxic to motor neurons (Figure 1g). Pretreating iAstrocytes with EVs resulted in a significant rescue of MN survival after 3 days of co-culture in all patient donor cell lines. Of note, in the most toxic C9orf72 astrocyte line (pat183), astrocyte survival was 50% greater upon EV treatment.
Thus, exposure of mouse or patients' ALS astrocytes to EVs significantly reduced their toxicity towards MNs.

MSC-Derived EVs Reduce the Pathological Activation and the Inflammatory Phenotype in Mouse SOD1 G93A Astrocytes
The involvement of astrocytes in propagating inflammation in ALS has been extensively described [31,61,62], and we have previously demonstrated that in vivo administration of MSCs reduced astrocyte pathological activation and IL-β and TNF-α expression in the spinal cord of SOD1 G93A mice [39]. Western blotting experiments, aimed at evaluating the effects of the exposure to MSC-derived EVs on astrocytes' reactive phenotype markers, revealed that the expression of vimentin (Figure 2a,b), GFAP (Figure 2a,c), and S100β (Figure 2a,d) were significantly upregulated in SOD1 G93A astrocytes compared to WT astrocytes. Treating SOD1 G93A astrocytes with MSC-derived EVs reverted the upregulation of vimentin, GFAP, and S100β to control levels. We confirmed the reduction of vimentin (Figure 2e,f), GFAP (Figure 2g and Figure S2a), and S100b (Figure 2h and Figure S2b) overexpression after exposure to MSC-derived EVs using semi-quantitative immunocytochemistry.
These findings demonstrate that EVs derived from IFNγ-stimulated MSCs efficiently reduce the expression of astrogliosis markers, thus reflecting a lower astrocyte pathological reactivity.
We investigated in parallel the pro-inflammatory phenotype of patient-derived iAstrocytes. As opposed to mouse SOD1 G93A astrocytes, NLRP3 did not consistently increase in iAstrocytes from SOD1 and C9orf72 patients and, consequently, did not show significant changes after EV treatment (Figure 4a,b). We then assessed the levels of phospho-p65 (p-p65), the phosphorylated and active form of the nuclear factor kappa B (NFkB). We did not find any significant changes between p-p65 expression in iAstrocytes from healthy subjects vs. ALS patients, nor after EVs exposure (Figure 4a,c), Then, we assessed the levels of IL-1 β, IL-6, and CCL-2 secreted from human astrocytes. Samples from the healthy controls and C9orf72 patients showed levels of IL-1β below the detection limit. In contrast, and consistent with the murine astrocyte results, SOD1 iAstrocytes displayed higher IL-1β levels than controls, significantly reduced after EV treatment (Figure 4d). Secreted baseline levels of IL-6 ( Figure 4e) were similar in patients and controls, while EVs exposure resulted in IL-6 increased expression. The baseline for CCL2 was higher in SOD1 and C9orf72 patients, as observed in the SOD1 G93A mouse astrocytes (Figure 4f), however, the exposure to MSC-derived EVs had no effect in patients or controls.
limit. In contrast, and consistent with the murine astrocyte results, SOD1 iAstrocytes displayed higher IL-1β levels than controls, significantly reduced after EV treatment ( Figure  4d). Secreted baseline levels of IL-6 ( Figure 4e) were similar in patients and controls, while EVs exposure resulted in IL-6 increased expression. The baseline for CCL2 was higher in SOD1 and C9orf72 patients, as observed in the SOD1 G93A mouse astrocytes (Figure 4f), however, the exposure to MSC-derived EVs had no effect in patients or controls.
These data indicate that the inflammatory cytokine profiles of iAstrocytes and SOD1 G93A mouse astrocytes are clearly different and treatment with EVs resulted in a more marked decrease in neuro-inflammatory markers in mouse SOD1 G93A astrocytes than human SOD1 or C9orf72 iAstrocytes.

MSC-Derived EVs Modulate Nrf2 Nuclear Translocation and Promote Antioxidant Response in ALS Astrocytes
Activation of the Nrf2 antioxidant pathway ameliorates astrocyte toxicity [64]. We thus investigated whether Nrf2 and downstream NQO1 activation is a mechanism of EV-mediated reduction of astrocyte toxicity towards MNs. Western blot experiments showed that the total expression of Nrf2 was markedly decreased in SOD1 G93A compared to age-matched WT astrocytes, confirming a deficit of the antioxidant response, and that exposure of SOD1 G93A astrocytes to mouse MSC-derived EVs did not affect Nrf2 expression (Figure 5a,b). NQO1 expression was slightly reduced in SOD1 G93A astrocytes, and EV treatment did not affect the expression of NQO1 (Figure 5a,c). Since Nrf2 exerts its activity in the nucleus, where it promotes the transcription of cytoprotective enzymes, we assessed the translocation of Nrf2 into the nucleus by confocal microscopy, analysing the nucleus/cytoplasm cellular localization ratio of the Nrf2 factor ( Figure 5d). The nuclear localization of Nrf2 in SOD1 G93A astrocytes was significantly lower than in WT astrocytes. Treatment with MSC-derived EVs led to almost complete restoration of Nrf2 levels in the nuclear compartment (Figure 5e). Thus, even if total Nrf2 did not change following EV exposure, there was an augmentation of Nrf2 nuclear translocation, which we predict boosts the antioxidant response in SOD1 G93A astrocytes.  To assess whether MSC-derived Evs also improved the antioxidant response in patientderived iAstrocytes, we quantified the level of Nrf2 and NQO1 (via immunofluorescence) and reactive oxygen species (ROS) using the CellROX cell dye (see representative Figure 5f,g) in control, C9orf72, and SOD1 iAstrocytes treated or not with EVs. Both nuclear Nrf2 and total NQO1 levels were significantly lower in untreated iAstrocytes than in controls (Figure 5h,i). Treating iAstrocytes with EVs, however, led to an increase in nuclear Nrf2 and total NQO1 levels in SOD1 iAstrocytes. To explore whether the increase in the antioxidant response led to a consequent reduction of reactive oxygen species, we performed a CellROX ® live cell assay. iAstrocytes displayed higher levels of intracellular reactive oxygen species (ROS) in C9orf72 and SOD1 patients compared to controls; interestingly, ROS levels were significantly reduced upon EV treatment in both groups (Figure 5j).
The combination of increased activation of Nrf2/NQO1 signalling with a reduction in ROS production indicates that promoting the antioxidant response is one of the mechanisms involved in MSC-derived EV-mediated neuroprotection in human astrocytes.

EV-Shuttled miRNAs Are Responsible for the EV-Mediated Neuroprotection
We recently found that miRNA expression significantly increased in IFNγ-primed MSCs [45]. The same overexpressed miRNAs were present in EVs isolated from MSCs, which proved to affect the neuroinflammatory phenotype of activated mouse microglia [45]. As in the previous study, we transfected SOD1 G93A astrocytes with the EV-shuttled miRNA synthetic mimics. After 48 h of transfection with single mimics, we observed that seven out of nine miRNA mimics induced a significant reduction of GFAP, TNF-α, and IL-1β expression, except for miR-5126 and miR-467g (Figure 6a-c; Supplementary Table S4). are significantly increased in untreated or scramble-miRNA-treated SOD1 G93A astrocytes vs. WT astrocytes. The exposure to seven out of nine specific synthetic mimics of miRNAs shuttled by EVs significantly reduces the expression of the GFAP, TNF-α, and IL-1β, indicating a decreased astrocyte reactivity. Data are presented as means ± SEM of n = 3 independent experiments, run in triplicate; statistical significance for p < 0.05 at least (* p < 0.001 vs. WT astrocytes; # p < 0.001 vs. SOD1 G93A astrocytes; § p < 0.001 vs. SOD1 G93A astrocytes + scrambled miRNA; F (11,24) = 17.967, F (11,24) = 49.372, F (11,24) = 38.634, for GFAP, TNF-α, and IL-1β, respectively; one-way ANOVA, followed by Bonferroni post hoc test).
Additionally, we investigated the ability of miR-466m-5p and miR-466i-3p to modulate the antioxidant pathway linked to Nrf2, considering that the transcription factors BTB and CNC homology 1 (Bach1) and Kelch-like ECH associated protein 1 (Keap1) are two endogenous inhibitors of Nrf2 and predicted targets of miR-466i-3p and miR-466m-5p. Bach1 competes for the same binding site as Nrf2 on the enhancer antioxidant response element (ARE), preventing its activation and the synthesis of antioxidant enzymes [68]. Keap1 binds Nrf2 in the cytoplasm and prevents its translocation into the nucleus [69] ( Figure 7f). Confocal microscopy experiments showed that the total expression of Nrf2, which decreased in SOD1 G93A astrocytes compared to WT astrocytes, was not significantly affected by transfection with either mimic (Figure 7g,h). As also shown above, the nuclear expression of Nrf2 decreased in SOD1 G93A astrocytes vs. WT astrocytes. Interestingly, the treatment with miR-466m-5p or miR-466i-3p reversed the defective translocation of Nrf2 into the nuclear compartment (Figure 7g,i).
These results suggest a possible mechanism for the MSC-derived EV-induced nuclear translocation of Nrf2 observed in SOD1 G93A astrocytes.
Similarly to mouse MSCs [45], we performed microarray analysis of a miRNAs expressed by hMSCs (Supplementary Table S5) and found that miR-29b3P was one of the upregulated miRNAs in IFN-γ-primed hMSCs. Interestingly, this miRNA acts as a direct target of Nrf2 [70] and regulates the Nrf2 axis by silencing the downstream inhibitor Bach2 (www.mirdb.org; http://mirdb.org/cgi-bin/target_detail.cgi?target; accessed on 14 March 2022; ID = 2387339; [71,72]). Indeed, the expression of miR-29b-3p reduced ROS production in cancer cells [73]. We assessed whether this miRNA could be responsible for the antioxidant response triggered in iAstrocytes by MSC-derived EVs and, consequently, for MN rescue. We transfected iAstrocytes with miR-29b-3p mimic before seeding motor neurons. Because miR-29b-3p has been linked to activation of the antioxidant pathway via Nrf2 [70], we performed RT-qPCR to investigate the expression of the genes encoding for Nrf2 and NQO1 in control iAstrocytes transfected with the miR-29b-3p mimic (Figure 7j). Consistent with the Nrf2 downstream action of miR-29b-3p, miR-29b-3p transfection had no effect on Nrf2, but increased NQO1 expression. Similarly to EVs, miR-29b-3p mimic restored MN survival in co-culture in three out of four lines, leading to a 40% increase viability in the most aggressive ALS C9orf72 line (Figure 7k).
These data suggest that miR-29b-3p may mediate MN rescue through antioxidant effects.

Discussion
We show that in vitro exposure of ALS astrocytes to MSC-derived EVs in human and mouse models reduced their neurotoxicity towards MNs, possibly by EV-shuttled miRNAs. Our results support previous findings showing that MSC-conditioned medium exerts a comparable ameliorative effect in different in vitro models of ALS. To our knowledge, we describe here the first demonstration that MSC-derived EVs are directly responsible for reducing astrocyte neurotoxicity.
We studied the effects of human MSC-derived EVs on i-Astrocytes from two SOD1 and two C9orf 72 ALS patients, and two healthy controls, unveiling the amelioration of the i-Astrocyte phenotype and the reduction of neuronal toxicity in specific ALS genetic subtypes. Consistently, pre-treating adult SOD1 G93A mouse astrocytes with mouse MSCderived EVs also reduced neurotoxic effects. Thus, our evidence gathered on iAstrocytes and SOD1 G93A mouse astrocytes demonstrated that both in vitro models are useful to shed light on the mechanisms of MSC-derived EV effects.
In the mouse cell model, exposure to the MSC-derived EVs reduced the expression of the astrogliosis markers vimentin, GFAP, and S100β, NLRP3 inflammasome, and the expression and secretion of IL-1β, TNF-α, IL-6, and CCL2 pro-inflammatory cytokines. In contrast, they did not increase antioxidants via Nrf2 or NQO1. These results suggest that in SOD1 G93A astrocytes, MN rescue is primarily based on reduced inflammatory mechanisms, while in the i-Astrocyte the modulation of inflammation might play a minor role, with EVs being more effective in modulating the oxidative stress. However, consistent with the mouse counterpart, iAstrocytes displayed high levels of IL-1β and CCL2, and EV treatment significantly decreased IL-1β, but not CCL2.
The differences between the anti-inflammatory or antioxidant effects of EVs in adult mouse astrocytes and iAstrocytes can rely on their different origins. Adult mouse astrocytes from the spinal cord of symptomatic SOD1 G93A mice mature in vivo in a neuroinflammatory pathological environment until isolation [53]. In human iAstrocytes, we found a mild increase of neuroinflammatory markers, but instead a higher ROS accumulation and a lower total Nrf2, the master regulator of the antioxidant cascade, and EV treatment successfully overturned these pathological features. Moreover, SOD1 and C9orf72 astrocytes differ in their pro-inflammatory profile, with the expression of IL-1β being higher in iAstrocytes from SOD1 patients, and this aspect is worth further investigation.
We have profiled the miRNAs in mouse MSC-derived EVs and used their synthetic miRNA mimics to assess the effect on the reactive astrocyte properties. We found that seven out of nine mimics of miRNAs upregulated in IFN-γ-primed MSCs reduced the pathological reactive astrocyte phenotype. Four of them, miR-466q, miR-467f, miR-466m-5p, and miR-466i-3p, overexpressed in both MSCs and EVs, were tested for their ability to modulate the inflammatory response. After in silico analysis aimed at identifying specific inflammatory pathways, we focused on the NFҡB and MAPK pathways. Transfection with each single miRNA mimic had a modest effect on the p38 MAPK pathway; in contrast, MapK11, another predicted target of the four miRNAs, which promotes TNF-α and IL-1β synthesis, was significantly upregulated in SOD1 G93A astrocytes and reduced by miR-466q or miR-467f. The identified miRNAs also acted on oxidative stress. Indeed, in silico analysis indicated that miR-466i-3p and miR-466m-5p are potential modulators of Keap1 and Bach1, two endogenous inhibitors of Nrf2 that prevent its nuclear translocation and activity [68,69]. Interestingly, miR-466m-5p promoted Nrf2 translocation, possibly boosting the antioxidant response. Thus, MSC-derived miRNAs could be responsible for the beneficial effect of the EVs in mouse astrocytes by modulating both the inflammatory and the antioxidant pathways.
In human iAstrocytes, the addition of the synthetic miR-29b-3p mimic to C9orf 72 and SOD1 iAstrocytes before co-culture with MNs significantly upregulated NQO1 and rescued MN survival. Further work is required to elucidate the pathways regulated by the EV-shuttled miRNAs.
We have shown for the first time that the application of MSC-derived EVs to SOD1 and C9orf 72 patients' astrocytes and adult mouse SOD1 G93A astrocytes significantly rescued MN death, suggesting the possibility of replacing the MSC cell grafts in ALS with a less invasive and immunogenic administration of EVs [42,74]. Accordingly, the literature reports that EVs from adipose stem cells ameliorate the disease course in SOD1 G93A mice [75]. EVs have been demonstrated to play a key role in pathological processes since their cargo can contribute to the spread or modulation of a disease, but indeed they can be also exploited as a therapeutic strategy to overcome neurodegeneration associated with CNS pathologies, and this has been demonstrated with particular emphasis on Alzheimer's disease and Parkinson's disease, other than amyotrophic lateral sclerosis [76,77]. In particular, EVs secreted by MSCs have aroused considerable interest as a potential cell-free therapy [78,79]. In the last decade, it has become evident to the scientific community that EV cargo can indeed contain molecules such as miRNAs, as we confirmed in the former and present paper, that can be themselves new targets or can target key cellular pathways in the receiving cells, suggesting that they could be considered as a tool for new potential gene therapies. Moreover, EVs can easily cross the blood-brain barrier and can be exploited to deliver drugs [80,81].
In this framework, our findings surely support a new potential approach further contributing to the landscape of innovative ALS therapies. Very recently, this scenario has been strongly enriched with the ground-breaking discovery and application of the antisense oligonucleotide (ASOs) Tofersen, which efficiently mediates the degradation of superoxide dismutase 1 (SOD1) messenger RNA to reduce SOD1 protein synthesis in human patients affected by inherited SOD1 mutations [82,83]. At the same time, Mueller et al. showed the efficacy of a single infusion of adeno-associated virus encoding a microRNA targeting the SOD1 [84]. These results are also supported by the possibility of exploiting the CRISPR base editors in SOD1 G93A mice, reinforcing the potentiality of gene therapies [85]. Of note, all the above experimental treatments are delivered by an invasive route, such as intrathecal infusion, and are limited to SOD1 or C9ORF72 mutations. Conversely, EVs can be administered by a non-invasive route, such as intranasally, entering the brain quickly and efficiently, thus representing a very translational opportunity for the treatment of neurological diseases [86][87][88].
Based on the above evidence, as the subsequent translational application of the data reported in this paper, our research groups are currently involved in carrying out studies aimed at testing the efficacy of MSC-derived EVs intranasally administered in SOD1 G93A ALS mice by performing a comprehensive panel of in vivo functional studies, as well as ex vivo histological and molecular analyses, as previously reported [89][90][91][92].
Our results pave the way also for using miRNA-loaded EV-mimicking synthetic particles, bypassing the culture and amplifying MSCs. Moreover, they are significant because they extend beyond SOD1 as a model for ALS and demonstrate a broader therapeutic potential for EVs in different subtypes of ALS.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cells11233923/s1. Figure S1: Confocal microscopy representative images showing MN/astrocyte co-cultures labelled with specific markers for astrocytes and motor neurons. Figure S2: Immunofluorescence representative images showing iAstrocytes and MN/iAstrocyte co-cultures labelled with specific markers for astrocytes and motor neurons. Figure S3: Characterization of exocytosis markers and size of the extracellular vesicle populations in IFNγ-stimulated MSCs. Figure S4: MSC-derived EVs reduce astrogliosis in astrocytes from the spinal cord of adult SOD1 G93A mice. Figure S5: MSC-derived EVs reduce the NLRP3 inflammasome in astrocytes from the spinal cord of adult SOD1 G93A mice. Figure S6: MSC-derived EVs reduce the expression of pro-inflammatory cytokines in astrocytes of adult SOD1 G93A mice. Table S1: Details of fibroblast donors. Table S2: List and sequence of the synthetic mimics used for mouse and human astrocyte transfections. Table S3: Details of antibodies. Table S4: Percentage of variation of GFAP, TNF-α and IL-1β expression in SOD1 G93A mouse-derived astrocytes after 48h transfection with single miRNA synthetic mimics vs. untreated SOD1 G93A astrocytes. Table S5: List of the up-regulated miRNAs detected in human MSCs (hMSCs) primed with IFN-γ.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Conceptualization