Exosome-shuttled miRNAs contribute to the modulation of the neuroinflammatory microglia phenotype by mesenchymal stem cells. Implication for amyotrophic lateral sclerosis

Background Mesenchymal stromal/stem cells (MSCs) are characterized by neuroprotective, immunomodulatory, and neuroregenerative properties, which support their therapeutic potential for neurodegenerative diseases driven by microglia-associated inflammation, such as amyotrophic lateral sclerosis (ALS). One mode of action through which MSCs exert their immunomodulatory effects is the release of extracellular vesicles, including exosomes, that carry proteins, mRNAs, and microRNAs (miRNAs), which, once transferred, modify the function of target cells. We have investigated the role of miRNAs present in exosomes derived from IFN-γ-primed mouse MSCs in the modulation of microglia activation, and analysed their effect on target genes and signaling pathways. Methods We compared miRNA expression in IFN-γ-primed vs unprimed mouse MSCs by microarray and measured the levels of relevant miRNAs in their respective exosomes through RT-PCR. To assess the effect of dysregulated MSC-derived miRNAs, we transfected lipopolysaccharide-activated N9 microglial line cells and primary microglia from late-symptomatic SOD1G93A ALS mice with their specific mimics and analysed the mRNA expression of pro/anti-inflammatory genes in the cells. We used mirWalk and Panther and KEGG Pathway databases to predict target genes of specific miRNAs and possible pathways they regulate. Data were compared using Student’s t-test. Results We identified nine miRNAs that were significantly dysregulated in IFN-γ-primed MSCs, but present at different levels in their derived exosomes. Transfection with three of the four miRNAs significantly upregulated in IFN-γ-primed MSC-derived exosomes, namely miR-467f, miR-466q and miR-466m-5p, could modulate the pro-4 expression, and/or upregulating Cx3cr1 expression. We obtained similar results in primary microglia from SOD1G93A mice transfected with miR-467f and miR-466q. Further analysis of the mode of action of miR-467f and miR-466q indicated that they dampen the pro-inflammatory phenotype of microglia by modulating the p38 MAPK signaling pathway via inhibition of the expression of their target genes, Map3k8 and Mk2. Conclusion These results suggest that exosome-mediated transfer of functional miRNAs could be one mode of action through which MSCs exert their therapeutic effect in ALS by downregulating neuroinflammatory microglia, and identify miR-467f and miR-466q as immunomodulatory miRNAs involved in this process.

4 inflammatory phenotype of N9 microglia by downregulating Tnf and/or Il1b expression, and/or upregulating Cx3cr1 expression. We obtained similar results in primary microglia from SOD1G93A mice transfected with miR-467f and miR-466q.
Further analysis of the mode of action of miR-467f and miR-466q indicated that they dampen the pro-inflammatory phenotype of microglia by modulating the p38 MAPK signaling pathway via inhibition of the expression of their target genes, Map3k8 and Mk2.
Conclusion These results suggest that exosome-mediated transfer of functional miRNAs could be one mode of action through which MSCs exert their therapeutic effect in ALS by downregulating neuroinflammatory microglia, and identify miR-467f and miR-466q as immunomodulatory miRNAs involved in this process. protective M2-like, by inhibiting the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway, thereby reducing pro-inflammatory cytokine expression [5]; similarly, MSCs inhibited the activation of NF-κB and mitogenactivated protein kinase (MAPK) pathways in LPS-stimulated BV2 microglial cells and, therefore, their polarization to the pro-inflammatory phenotype through the secretion of tumor necrosis factor-α-induced gene/protein 6 (TSG-6) [6]. In other studies, colony-stimulating factor-1 was found to be one MSC secretome molecule responsible for the anti-inflammatory effect of MSCs on LPS-activated microglia [7], and MSCs were shown to exert a "calming" effect on pro-inflammatory microglia through the release of CX3CL1 that upregulates the CX3CR1/CX3CL1 axis involved in the control of microglia activation [8]. It should be noted that MSC-derived microvesicles were also recently demonstrated to be modulators of LPS-induced microglia activation [9].
Such studies have led to administration of MSCs being considered as a possible alternative therapeutic approach for controlling neurological diseases associated with neuroinflammation [10], including multiple sclerosis [11,12] and amyotrophic lateral sclerosis (ALS) [13][14][15]. In vivo studies in our laboratory [13] have shown that intravenous administration of MSCs in mice that express the human Cu, Zn superoxide dismutase-1 carrying the G93A point mutation (SOD1 G93A ), a widely used experimental model for ALS, during the symptomatic stage of disease, significantly improve the clinical outcome and pathological scores [13]. The beneficial effect is related to the decrease of oxidative stress and the inhibition of glutamate-mediated excitotoxicity, but also to a reduction in astrocyte and microglia proliferation and related neuroinflammation [13]. Similarly, ALS mice treated with human MSCs through transplantation into the spinal cord [16] or 6 multiple systemic administration [17] showed decreased microglia and astrocyte activation and improved motor performance.
In addition to soluble factors, extracellular vesicles are a key instrument in cell-cell communication [18]. Among the many subtypes of extracellular vesicles, exosomes have emerged as physiologically relevant and powerful components of the MSC secretome [19,20]. Exosomes are small vesicles with a diameter of 40-120 nm, with a specific molecular composition that depends on the cell of origin and the cellular context [21]. Together with specific proteins, lipids, and mRNAs, the exosome cargo is rich in various microRNAs (miRNAs), which are small sequences of RNA that, when transferred to the cytoplasm of target cells, govern various processes, preventing protein translation. Indeed, they modulate gene expression at post-transcriptional level via mRNA degradation, translational repression, or both, in target cells [22].
The aim of this study was to assess, in vitro, if the immunomodulatory effect of MSCs on neuroinflammation could be attributed, at least in part, to their release of exosomes that shuttle specific miRNAs able to downregulate the pro-inflammatory phenotype of activated microglia, and to define the mode of action of these "immunomodulatory" miRNA(s) through identification and validation of their target genes involved in the inflammatory pathway.

Microarray analysis to compare miRNA expression in MSCs primed or not with IFN-γ
Bone marrow-derived MSCs were isolated from 6-to 8-week-old C57BL/6J mice (Harlan Laboratories), expanded, and characterized as described previously [11].
The experiments were approved by the Animal Ethics Committee of Ospedale

Isolation and characterization of MSC-derived exosomes
In order to increase their production of exosomes, expanded IFN-γ-primed and unprimed MSCs were stimulated for 20 minutes with 1 mM ATP (Sigma-Aldrich) at 37° C [25]. The resulting supernatant was centrifuged at 2,000 x g at 4° C for 20 minutes to eliminate cells and debris, and incubated overnight at 4° C with 0.5 volume of Total Exosome Isolation Kit (Invitrogen). After the incubation, the sample was centrifuged at 10,000 x g at 4° C for 1 hour and the pellet containing the exosomes was resuspended accordingly to the experimental needs.
For characterization by western blot analysis, 15 μg of exosome proteins were 8 loaded on a precast polyacrylamide gel (from 4% to 12% gradient, Life Technologies), using the Bolt® Mini Gel Tank (Life Technologies) system. Protein were then transferred on a nitrocellulose membrane (BioRad) using XCell II™ Blot Module (Life Technologies). After blocking in 5% BSA in PBS/Tween 20 for 1 hour, the membrane was incubated overnight at 4° C with primary rabbit anti-ALIX For characterization by electron microscopy, exosomes collected from 7x10 6 MSCs were fixed in a volume of 50-100 μl of 2% paraformaldehyde, according to a published protocol [26]. 5 μl of resuspended pellet was allowed to adhere to electron microscopy grids (Formvar-Carbon) for 20 minutes at 42° C. Subsequently, the grids were washed 2 times with 100 μl PBS for 3 minutes, once with 1% glutaraldehyde for 5 minutes, and finally seven times with 100 μl of distilled water for 2 minutes each. For contrast phase microscopy, the samples were transferred to 50 μl 2% uranyl acetate (UA) solution for 5 minutes and then to 50 μl of methylcellulose (MC) and UA (9 ml MC + 1 ml UA 4%) for 10 minutes in ice. The sections were dried on a filter paper and then in the air, visualized using a FEI CM10 microscope, and acquired via a Leo912ab camera.

Culture and activation of N9 microglia line cells
The murine microglial cell line N9 (Neuro-Zone srl, Italy) was plated in 75 cm 2 cell culture flasks at a concentration of 5-6x10 5 cells in 15 ml RPMI (Sigma-Aldrich) containing 10% fetal bovine serum (FBS) (Lonza), 100 U/ml penicillin, and 100 μg/ml streptomycin, and maintained at 37° C and 5% CO 2 in incubator. The cells were activated by exposure to 1 μg/ml LPS (Sigma-Aldrich) for 30 minutes for the immunofluorescence experiments and for 24 hours for the RT-PCR experiments, as indicated in the legends to the relevant figures.

Mice
B6SJL-TgN SOD1/G93A1Gur mice expressing a high copy number of mutant human SOD1 with a Gly to Ala substitution at position 93 (referred to thereafter as SOD1 G93A mice) and B6SJL-TgN (SOD1)2Gur mice expressing wild-type human SOD1 [27] (referred to thereafter as wtSOD1 mice) were originally obtained from Jackson Laboratories (Bar Harbor, ME, USA) and bred at the animal facility of the Pharmacology and Toxicology Unit, Department of Pharmacy at the University of Genoa, Genoa, where they were kept until experiments were carried out.
All mice were housed in pathogen-free conditions with food and water ad libitum. All applicable international, national, and/or institutional guidelines for the care and The onset of overt clinical symptoms in our SOD1 G93A mouse colony occurs at approximately day 90 [28]. Animals were sacrificed at the end stage of disease, established according to an homogeneous motor impairment severity score (extension reflex and gait impairment score: 4.5 units, at around 135 days of age) as previously described [13] and characterized by an overactivation of microglia [29].

Isolation of adult primary microglia
Primary microglia were isolated from the brain of late stage SOD1 G93A and age- at 450 x g at 4° C for 5 minutes. Supernatants were aspirated and the pellets were resuspended in 7.2 ml of wash solution (RPMI and 1M HEPES); 1.2 ml of FBS was mixed with the cell suspension, followed by 3.6 ml of 100% Percoll (Sigma-Aldrich).
Finally, 1 ml of 10% FBS in RPMI was layered over the cell suspension and samples were centrifuged at 800 x g at 4° C for 15 minutes without brake. Pellets were resuspended in 1 ml RPMI with 10% of FBS and cells were counted. An average of 5-11 6 x 10 6 cells was obtained per single brain and the primary microglia were further purified on CD11b (Microglia) MicroBeads (Miltenyi Biotec) according to the manufacturer's instructions. It is notoriously difficult to obtain pure mouse microglia from adult brain and we reached an average yield of 3-5x10 5 microglia per single brain with a final purity of 85-90%.

Microglia exposure to IFN-γ primed MSC-derived exosomes
1x10 5 LPS-activated N9 cells or 2-3x10 5 primary microglia (higher concentration of cells was used because of the low survival of primary microglia in culture) resuspended in 1 ml RPMI were plated per well in a 24-well plate in presence or absence of IFN-γ primed MSC-derived exosomes (exos IFN-γ-MSC ). The quantity of exosomes added to the cultures was equivalent to that produced by MSCs at a microglia:MSC ratio of 1:3. After 24 hours at 37° C and 5% CO 2 , cells were processed for RNA extraction.

RNA isolation and Real Time quantification
Total RNA was isolated from N9 cells and primary microglia using QIAzol Lysis Medium (Sigma-Aldrich). Fluorescence image acquisition was performed by a Leica TCS SP5 laser-scanning confocal microscope, through a plan-apochromatic oil immersion objective 63X/1.4 NA. The quantitative estimation of co-localized proteins was performed by calculating the 'co-localization coefficients' [31].
According to Costes et al. [32], the correlation between the green and red channels was evaluated with a significance level > 95%. Costes' approach was carried out by macro routines (WCIF Colocalization Plugins, Wright Cell Imaging Facility, Toronto Western Research Institute, Canada) integrated as plugins in the ImageJ 1.52q software (Wayne Rasband, NIH, USA).

Statistical analysis
The results are presented as mean ± standard error (SEM). Statistical analysis was performed on independent experiments using Student's t-test through the Prism 5 program (GraphPad Software, La Jolla, CA). In all analyses, P<0.05 is considered statistically significant.

IFN-γ-primed MSCs
As already known from the literature, the immunomodulatory effects of MSCs are promoted through environmental factors, in particular inflammatory cytokines that "prime" MSCs to become immunosuppressive through the concerted action of nitric oxide [33] and many other factors [34].

Exosomes derived from IFNg-primed MSCs affect genes related to the inflammatory and neuroprotective phenotype of microglia
Exosomes were isolated from MSCs activated with IFN-γ (exos IFN-γ-MSC ) and characterized through electron microscopy and Western blot analyses. These analyses revealed a preparation composed of purified nanovesicles, with a diameter ranging from 30 to 100 nm, which expressed ALIX and CD9 (Additional file 3) as exosomal markers [37].
To understand if exos IFN-γ-MSC could modulate the molecular phenotype of activated microglia, we exposed LPS-activated N9 cells to exos IFN-γ-MSC for 24 hours and assessed the mRNA expression of pro-and anti-inflammatory markers. The results showed that exos IFN-γ-MSC were able to significantly downregulate the expression of pro-inflammatory genes such as Tnf, Il1b and Il18, whilst upregulating the expression of markers associated with an anti-inflammatory/neuroprotective phenotype including Cx3cr1 [38], Cd206 [39], Nr4a2 [40], on activated microglial cells (Fig. 1).
We further assessed if exos IFN-γ-MSC could affect microglia isolated from the brain of late symptomatic SOD1 G93A mice. At this stage, these cells display an overactivated pro-inflammatory phenotype, with upregulation of Tnf expression and downregulation of alternative activation phenotype markers (Cx3cr1 and Nr4a2), as compared to microglia isolated from wtSOD1 mice (Fig. 2a). While pure microglia can be isolated from adult mouse brain, albeit with a low yield, these cells do not grow well in culture. Nevertheless, a 24-hour exposure of the primary SOD1 G93A microglia culture to exos IFN-γ-MSC yielded data which, albeit not reaching significance, suggest that exos IFN-γ-MSC could modulate the pro-inflammatory phenotype of SOD1 G93A microglia by inducing a decrease in Tnf expression and an increase in expression of Cx3cr1 and Nr4a2 (Fig. 2b).

miRNAs dysregulated in IFNg-primed MSCs are differentially expressed in their derived exosomes
To first ascertain if exosomes derived from IFNg-primed MSCs contain the same miRNAs dysregulated in the cells themselves (Additional File 2), we measured miRNA expression in exosomes derived from both unprimed and IFN-γ-primed MSCs (Fig. 3).
The results show that all nine miRNAs are present in unprimed MSC-derived exosomes (exos MSC ). Only four of these miRNAs, namely miR-467f, miR-466q, miR-466m-5p and miR-466i-3p, are significantly upregulated in exos IFN-γ-MSC (Fig. 3), supporting the findings of Squadrito et al. that sorting in exosomes is apparently influenced by the cellular environment [41], an observation which could be relevant in the context of their potential effect on their mRNA targets and suggesting a possible active role played by these specific miRNAs in the immunomodulatory capacity of MSCs.

In-vitro transfection with specific mimics reduces the pro-inflammatory phenotype of activated microglia
To understand if the four miRNAs upregulated in exos IFN-γ-MSC could affect the proinflammatory phenotype of activated microglia, we transfected LPS-activated N9 microglia with their respective mimics (synthetically generated oligonucleotide of sequence identical with that of endogenous miRNAs), and measured the mRNA expression of the pro-inflammatory phenotype markers, Tnf and Il1b, and of Cx3cr1, as a marker of anti-inflammatory microglia (Fig. 4a). As expected, activation with LPS induced microglia to overexpress pro-inflammatory cytokines and to downregulate Cx3cr1 expression. Transfection with mimics demonstrated that some of the miRNAs could modulate microglia phenotype; in particular, miR-467f and miR-466q significantly reduced the expression of Tnf and Il1b, whereas miR-466m-5p induced an upregulation of Cx3cr1 expression; miR-466i-3p did not have any effect.
To understand if the modulatory effect of miR-467 and miR-466q on the pro-inflammatory phenotype of LPS-activated N9 microglia-like cells translated to the SOD1 G93A microglia, we transfected primary microglia from late symptomatic SOD1 G93A mice with miR-467f and miR-466q and analysed the expression of the representative pro-inflammatory genes, Tnf and Il1b. In line with what we observed with LPS-activated N9 cells, transfection with miR-467f and miR-466q reduced Tnf and Il1b mRNA expression in primary SOD1 G93A microglia (Fig. 4b).
These results suggest a selective role for specific miRNAs in microglia phenotype modulation, with miR-467f and miR-466q in particular showing an anti-inflammatory potential upon transfection in activated microglia.
To assess whether miR-466i-5p, miR-467g, miR-3082-5p, miR-5126 and miR-669c-3p that were dysregulated in IFN-γ-primed MSCs but not upregulated in their derived exosomes, could also have some effect on the phenotype of microglia, we transfected LPS-activated N9 microglia-like cells with mimics of these miRNAs and analysed the expression of Tnf, Il1b and Cx3cr1 in the cells by RT-PCR. The results show that miR-3082-5p and miR-5126 significantly increased the expression of Cx3cr1, while none of the miRNAs was able to affect the expression of the proinflammatory genes (Additional file 4).

miR-467f and miR-466q act on their target genes to reduce the activation of the p38 MAPK pathway and, thereby, the inflammatory phenotype of activated microglia
Based on their anti-inflammatory effect, we further investigated the mechanism of action of miR-467f and miR-466q. To predict possible miRNA targets for these miRNAs, we used miRWalk online database, which predicts the possible targets of miRNAs through algorithms applying several different criteria, such as perfect base pairing, conservation criteria, AU content, and free energy of miRNA-mRNA heteroduplex. Hence, we identified 1718 possible target genes for miR-467f and 1157 for miR-466q (Additional files 5 and 6, respectively). To define pathways that could involve components encoded by these target genes, we used two different databases, KEGG, which predicts pathways based on the involvement of the miRNA itself in regulating the pathway, and Panther Classification System, which predicts pathways on the basis of the predicted target genes we identified through miRWalk database. By combining data from these two databases, we predicted a number of pathways, listed in Table 1, which could be affected by the specific miRNAs, and we decided to focus on MAPK signaling, since this pathway can be modulated by both miRNAs (Table 2). Most importantly, since MAPKs are a family of serine/threonine kinases whose activation is correlated with the synthesis of inflammation mediators, the inhibition of this pathway by both miRNAs could explain their anti-inflammatory effect in activated microglia.
Of the several components of MAPK pathway which could be regulated by miR-467f and miR-466q (Table 2), we focused on Map3k8, target of miR-467f, and Mk2, target of miR-466q, important steps of activation of the p38 MAPK signaling pathway (Fig.   5a), which plays a key role in neuroinflammation [42]. To assess if the expression of Map3k8 and Mk2 in microglia is affected by exposure of the cells to exos IFN-γ-MSC , we cultured LPS-activated N9 cells in the presence of exos IFN-γ-MSC for 24 hours and evaluated the expression of the two genes through RT-PCR analysis. The results show that exos IFN-γ-MSC significantly decreased the mRNA expression of Map3k8 and Mk2 in pro-inflammatory microglia (Fig. 5b). To ascertain that Map3k8 and Mk2 are indeed specific targets of miR-467f and miR-466q in microglia, we performed RT-PCR for Map3k8 and Mk2 mRNAs on LPS-activated N9 cells transfected with the relevant mimics. As shown in Fig. 5c, transfection with miR-467f induced a downregulation of its direct target, Map3k8, in activated N9 cells, whereas transfection with miR-466q reduced not only the expression of its target Mk2, but also that of Map3k8 presumably due to an indirect mechanism still unknown.
We obtained similar results in ex-vivo experiments in SOD1 G93A primary microglia, in which transfection with miR-467f modulated the expression of Map3k8, which should affect the whole pathway, and of Mk2, probably due to the upstream effect on its direct target (Fig. 5a); as seen with activated N9 cells, miR-466q reduced the expression of both its direct target, Mk2, and Map3k8 (Fig. 5d). To confirm that miR-467f and miR-466q affected the activation of p38 MAPK pathway, through their inhibition of their target gene expression, we investigated phosphorylated p38 (p-p38) through immunofluorescence analyses in LPS-activated N9 cells transfected with each miRNA separately or with a mixture of both ( Fig. 6a and 6b; N9+Mix+LPS). The co-localized fluorescence of p-p38 with the house-keeping protein, GAPDH, in N9 cells was significantly increased upon stimulation with LPS, but when treatment with LPS was preceded by transfection with miR-467f and miR-466q, individually or together, N9 cells showed reduced levels of p-p38. This was specific for the action of miR-467f and miR-466q in the cells, as transfection with Cneg did not have any effect, while Cpos inhibited only the expression of GAPDH, as expected. discussion Administration of MSCs, which have shown neuroprotective and immunomodulatory potential [2], is being increasingly considered as a potential treatment for neurodegenerative diseases, particularly those where microglia activation plays a key role.
There is increasing evidence that the therapeutic features of MSCs may rely on paracrine mechanisms with positive effects mediated in part by exosomes released from the administered cells [43,44]. For example, in the rat model of stroke, intravenous treatment with MSC-derived exosomes enhanced neurite remodeling, neurogenesis, and angiogenesis, as demonstrated through augmented numbers of neuroblasts and endothelial cells [45]. Similarly, MSC-derived exosomes induced functional recovery of rats after traumatic brain injury [46], by promoting an increase in mature neuronal and endothelial cells and a reduction in neuroinflammatory astrocytes in the lesion [47].
In this study, we show that the therapeutic effect of IFNg-primed MSCs in the SOD1 G93A mouse model of ALS could be mediated by their released exosomes acting on activated microglia. This can be particularly relevant in view of the welldocumented non-cell autonomous nature of this disease [48,49]. In fact, although motor neurons (MNs) are the principal target in ALS, non-neuronal cells are also involved the etiopathogenesis and it is now clear that astrocytes [48,50] and microglia [51,52] play a crucial role in MN degeneration during disease progression, mainly by regulating the pro-inflammatory milieu surrounding MNs in motor cortex and spinal cord. In particular, microglia play a primary role during the late fastprogressing phase of the disease [53,54]. Therefore, the attenuation of the proinflammatory microglial phenotype at this stage, as in our experiments, may in turn impact positively on MNs viability and disease progression.
There is a growing interest in exosomes as potential functional vehicles that deliver 23 their cargo, and in particular miRNAs and proteins, to target cells [55]. For example, exosomes from mouse primary cortical neurons are enriched in miR-124a which they deliver to astrocytes leading to upregulation of glutamate transporter 1 (GLT1), a protein involved in glutamate uptake [56]. Such an observation could be of therapeutic relevance in ALS where astrocytes display a pathological downregulation of GLT1 expression. On the other hand, exosomes can also be deleterious, depending on the parent cell milieu; thus, in-vitro studies show that astrocytes derived from SOD1 G93A mice secrete exosomes that transfer SOD1 G93A to non-transgenic spinal neurons inducing motor neuron death, which could contribute to progression of ALS pathology [57]. This suggests that microenvironmental stimuli could affect the cargo of the exosomes released by cells, and thereby the response of the target cells. Molecular sorting into exosomes is affected by the cell milieu [58] and sorting of miRNAs in particular is regulated by the cell status, as elegantly demonstrated by Squadrito et al. [41]. In this context, we observed that priming of MSCs with IFN-γ, which licenses their immunosuppressive capacity [36], induced an overexpression of specific miRNAs, suggesting their possible role in the immunomodulatory action of MSCs. Of interest, we observed differences in the levels of the studied miRNAs in the exos IFN-γ-MSC compared to exos MSC , which did not necessarily mirror their expression in the parental cells, supporting the demonstration that exosomal content does not reflect the cytoplasm of the cell from which they originate [59] and is affected by environmental conditions, such as oxidative stress [60].
We showed that exos IFN-γ-MSC are enriched in specific miRNAs able to modulate the pro-inflammatory phenotype of target microglial cells, a demonstration on par with other studies demonstrating functional transfer of miRNAs through exosomeshuttling, whereby miRNA-containing exosomes can affect the response of recipient cells to the environment [58,59]. This is of particular importance in the context of inflammation where such intercellular communication through exosome-transferred miRNAs has been shown to influence inflammatory responses [58,61]. This is exemplified by the study of Alexander et al. [62] in mice, whereby in-vivo administration of exosomes containing miRNAs with contrasting functions, miR-155 and miR-146a, altered the capacity of the recipient cells to respond to inflammatory cues in ways reflecting their respective regulatory functions. Thus, while miR-155containing exosomes induced an increase of the cellular response to LPS with overexpression of TNF and IL-6 in serum in mice treated with LPS, administration of miR-146a-containing exosomes led to a reduction of the inflammatory response to LPS, with decreases in TNF and IL6 serum concentrations [62]. These data confirm that depending on the milieu of the parent cell and differential miRNA enrichment of the exosomes released, these could induce target cells to react to the same inflammatory stimulus in different ways.
Interestingly, in our study we noted a common "beneficial" anti-inflammatory effect on pro-inflammatory microglia exerted by transfected miR-467f and miR-466q and by exos IFN-γ-MSC in which they are enriched, suggesting that the anti-inflammatory effect of exos IFN-γ-MSC could be related to the miRNA content. Indeed, transfection of selected miRNAs in pro-inflammatory microglia showed that miR-467f and miR-466q are able to affect microglia activation, inhibiting the expression of pro-inflammatory cytokines. In this context, exosomes have been studied as modulators of neuroinflammation through the shuttling of their cargo, which includes miRNAs, to 25 target cells [61]. Thus, it has been shown that several miRNAs play an important role in the control of neuroinflammatory mechanisms. In the animal model of traumatic brain injury, miR-200b is downregulated in pro-inflammatory microglia, but when transfected in these cells it is able to modulate the inflammatory response, decreasing c-Jun N-terminal kinase activity, inducible nitric oxide synthase expression, and nitric oxide production [63]. However, the antiinflammatory mode of action of transfected miRNAs does not depend necessarily on restoring their expression in the cells; indeed, activation of microglia was not associated with a decrease in the expression of miR-467f and miR-466q (additional file 7); rather, the anti-inflammatory effect of the transfected miRNAs suggests that a quantitative threshold might be necessary for their effect on the gene targets to result in a qualitative response [64].
Based on our data, we propose that exos IFN-γ-MSC can attenuate pro-inflammatory microglia phenotype through the activity of specific miRNAs able to modulate the p38 MAPK pathway which is involved in the neuroinflammatory process. An aberrant activation of p38 MAPK has been demonstrated in ALS [65], suggesting a crucial role for this pathway in the disease. In this context, the demonstration that miR-467f and miR-466q play a role in inhibiting the p38 MAPK signaling pathway could be exploited to further support the use of MSCs or MSC-derived exosomes to treat neurodegenerative diseases characterized by microglia activation, such as ALS.

conclusions
In conclusion, we demonstrated that MSC-derived exosomes can regulate neuroinflammation through specific miRNAs that modulate the pro-inflammatory phenotype of activated microglia. In particular, we showed that miR-467f and miR-466q exert their anti-inflammatory role affecting target genes involved in the p38 MAPK signaling pathway, which participates in the increased microglial production of pro-inflammatory cytokines. Therefore, this study suggests another mode of action through which MSCs can control microglia activation, and identifies possible relevant immunomodulatory miRNAs that could lead to novel therapies that dampen neuroinflammation. Our data with SOD1 G93A primary microglia support the hypothesis that MSC-derived exosomes and their delivered miRNAs can have a positive impact in experimental ALS by reducing the pro-inflammatory phenotype of microglia, possibly ameliorating the disease course.
The potential value of exosomes as therapeutic tool is increasingly promising and would provide several advantages compared to classical cell therapy, mainly linked to the ability to mitigate risks associated with cell transplantation, to cross the blood-brain barrier, which is highly impenetrable to most drugs, and to impact on the behavior of adjacent or distant cells.
In addition, exosomes are naturally carriers of miRNAs and this intrinsic property could be exploited as a gene delivery system. Moreover, one of the main advantages of using exosomes is the possibility to engineer them "made-to-order" to improve their functions, allowing the delivery of appropriate cargo, including miRNAs, directly to the brain.    Figure 1 Exposure to exosIFN-γ-MSC affects the molecular phenotype of activated microglia. RT-PCR q 41 Figure 2 Exposure to exosIFN-γ-MSC affects the overactivated phenotype of SOD1G93A primary micro 42 Figure 3 Exosomes derived from MSCs primed or not with IFN-γ are differentially enriched in miRNAs.