Mesenchymal stem cells (MSCs) are a heterogeneous population of adherent cells with self-renewable capacity and with a wide distribution in adult organisms; they can be isolated from adult tissues including bone marrow, adipose tissue, kidney and liver (Crisan et al., 2008). As multipotent progenitor cells, MSCs are able to differentiate into various cell types, especially of the mesodermal lineage, thus representing an important opportunity for regenerative medicine (Caplan, 1991). Besides this application, MSCs are able to control cell survival, organ function and inflammation. Remarkably, conditioned medium collected from MSCs can exert many of these paracrine effects, suggesting that the major mechanism of action of MSCs relies on soluble factors rather than cell–cell contact (Madrigal et al. 2014, Zanotti et al. 2013). A well-established feature of MSCs is their ability to inhibit inflammation and immunity, both in vitro and in vivo. Several studies have demonstrated that MSCs may be a useful therapeutic option for the treatment of immune-mediated disorders such as type 1 diabetes (T1D), rheumatoid arthritis (RA) and graft-versus-host disease (GVHD), but also transplantation rejection and neurodegenerative diseases (Ram Sharma et al. 2014, Ying Wang et al. 2014, Klinker et al. 2015) In order to better characterize the paracrine mechanism responsible for MSCs immunomodulatory properties, we performed a shotgun proteomic analysis of the mouse MSCs-conditioned medium, in collaboration with Prof. Gabriella Tedeschi from the University of Milan. Since the MSCs anti-inflammatory phenotype is induced by a pro-inflammatory environment (Bernardo ME et al. 2013, Groh ME et a. 2005), we stimulated MSCs with a cocktail of cytokines (IL1beta, IL6 and TNFalpha) in order to compare the medium conditioned by activated MSCs (st MSC-CM), with that derived from unstimulated MSCs. Interestingly, we found that the stimulation with pro-inflammatory cytokines (IL1beta, IL6 and TNFalpha) induces a remarkable change in the whole secretome; notably, most of proteins secreted by stimulated MSCs are involved in the regulation of angiogenesis. Among the factors up-regulated by conditioned MSCs, we identified the tissue inhibitor of metalloproteinases 1 (TIMP-1), a specific glycoprotein implicated in the endogenous inhibition of metalloproteinases (Lambert et al. 2004). We demonstrated that MSCs affect local inflammation trough the release of TIMP-1, both in in vitro and in vivo experimental settings. Specifically, by the secretion of TIMP-1, MSCs block the formation of new vessels at the draining lymph node of immunized mice. This event results in a decreased recruitment of leukocytes from the blood flow into the inflamed tissue, leading to a local suppression of the immune response (Zanotti and Angioni, Leukemia 2016). In order to design successful pre-clinical experiments as well as clinical trials, the analysis of human MSCs secretome was required. Thus, in collaboration with Prof. Gabriella Tedeschi, we performed a mass spectrometry based proteomic approach to analyse how pro-inflammatory cytokines modulate the composition of the human MSCs (hMSCs) secretome. Comparative analysis of hMSC-CM and mMSC-CM confirmed that the exposure to pro-inflammatory cytokines results in increased secretion of a number of immunomodulatory and angiogenesis-related proteins by MSCs from both species. Notably, 62% of the proteins identified in st hMSC-CM were also identified in st mMSC-CM, clearly highlighting the existence of a common signature in the secretome of human and mouse MSCs. However, despite the similar proteomic signature in response to stimulation by pro-inflammatory cytokines of human and mouse MSCs, our data indicate that they may induce different biological responses. For example, although the growth factor M-CSF/CSF1 is up-regulated in both human and mouse MSC-CM upon stimulation, only hMSC-CM efficiently induce macrophage differentiation; this is probably due to the different concentration of secreted M-CSF, which is higher in human-derived supernatants. Concerning angiogenesis, our data fully corroborate the anti-angiogenic role of stimulated MSCs for both mouse and human samples. In particular, we confirmed the key role of TIMP-1 as anti-angiogenic factor, both in human and mouse cells (Maffioli E et. al 2017). Complete mass spectrometry data are available via ProteomeXchange with identifier PXD005746. Although worldwide there are about 600 registered clinical trials evaluating the potential of MSCs-based cell therapy (see www.clinicaltrials.gov), this approach still remains far from a fully developed and safe clinical technology. Importantly, a standardized and unifying protocol addressing which source of MSCs should be used and which is the best route of administration is still missing. Moreover, even the parameters of quality and safety of MSCs are not universally established. To overcome all these issues, we focused our attention on an alternative approach, exploiting products derived from MSCs rather than MSCs themselves, which may represent a cost-effective and safer approach. One of the best-characterized MSCs product are MSCs-derived extracellular vesicles (EVs) (Biancone L et al., 2012). EVs arise from the plasma membrane and are released by cells as particles. In accordance with the recommendations of the International Society for Extracellular Vesicles (ISEV), EVs can be classified according on their size, origin, and isolation methods, into three main classes: (i) Microvesicles or shedding vesicles (size between 50 and 1000 nm, budding from the plasma membrane, and enriched in CD40); (ii) Apoptotic bodies (size between 800 and 5000 nm, derived from fragments of dying cells, and enriched in histones and DNA); and (iii) Exosomes, which are small (~30–120 nm) membrane vesicles of endocytic origin (enriched in late endosomal membrane markers, including Tsg101, CD63, CD9, and CD81) (Yáñez-Mó, 2015). Several studies have reported that MSCs-derived EVs display therapeutic potential in a similar fashion to their parent cells (Merin-Gonzàles et al. 2016). For example, it has been shown that MSCs-derived EVs reduce infarct size and enhance tissue repair in cardiovascular disease (Lai et al. 2010). More in general, it seems that EVs derived from unstimulated MSCs may reduce fibrosis and apoptosis, but sustain stem cell differentiation. Thus, MSCs-derived EVs are able to reinforce the regeneration process associated to not only cardiovascular disease but also liver, lung and acute kidney injury (Sweta Rani et al. 2015). To investigate MSCs-derived EV effect on angiogenesis, we first isolated and concentrated vesicles from the conditioned medium of MSCs stimulated with pro-inflammatory cytokines. We analysed the effect of EVs in vitro by endothelial cell tube formation assay, using EVs released by unstimulated MSCs as control. Our data confirmed that vesicles derived from stimulated MSCs affect the tubulogenesis process, perfectly mimicking the anti-angiogenic effect of the whole conditioned medium. Remarkably, the neutralization of TIMP-1 by a blocking antibody was able to revert this effect. Accordingly, through western blot analysis, we found that TIMP-1 is present in EVs derived from stimulated MSCs only. In the tubulogenesis assay, the degradation of the matrigel, with which plates are coated, is an essential step for the formation of an elongated and well-organized tube network. TIMP-1, secreted by st MSCs trough the release of EVs, plays a crucial role in the inhibition of this process, likely by inhibiting the matrix metalloproteinase activity required for the efficient matrigel digestion. However, endothelial cell migration represents another fundamental step during angiogenesis. In order to investigate this aspect of the angiogenic process, we set up a different assay based on the ability of cells to move in a free space: the wound healing scratch assay (Liang CC et al. 2007). Wound was made by scratching a line across the bottom of the dish on a confluent endothelial monolayer. After that, cells were treated with EVs from MSC-CM plus VEGF, with or without TIMP-1 blocking antibody. As expected, the pro angiogenic signal VEGF increases the migration of endothelial cells and st- MSCs-derived EVs block this process. However, in this condition, the TIMP-1 blocking treatment didn’t rescue the process. This suggests that st- MSCs-derived EVs are able to block angiogenesis through a second mechanism. Endothelial cell motility depends on the rapid reorganization of the actin network that, in turn, modulates not only the cell shape but also cell migration. In this context, reactive oxygen species (ROS) have emerged as crucial regulators. However, their effect, supporting or inhibiting angiogenesis, remains still controversial, and seems to depend on their localization and, in particular, on their concentration (Marcelo L. Lamers et al. 2011, Carlos Wilson et al. 2015). Indeed, low oxidant concentration regulates VEGF receptor cross-phosphorylation and redox-dependent downstream signalling (Lamalice et al. 2007). In contrast, high levels of oxidative species cause endothelial cell dysfunction due to an alteration in migration, increased apoptosis and an induction of senescence (Lum et al. 2001, Rong Liu et al. 2014). To investigate the role of ROS on the inhibition of endothelial cell migration induced by st MSCs-derived EVs, we analysed the production of oxidative species in migrating endothelial cells treated with unst- or st- MSCs-derived EVs. Interestingly, st MSCs-derived EVs were able to increase ROS production in the endothelial cells at the front of migration, thus suggesting that an oxidative unbalance could participate to the anti-angiogenic effect induced by st- MSCs-derived EVs. Nox2 represents the main key regulator of ROS production in endothelial cells, and adenosine has been implicated in its control (Sapna Thakur et al. 2010). Adenosine is catabolized from ATP (adenosine-5'-triphosphate) by the enzymes CD39 (nucleoside triphosphate dephosphorylase) and CD73 (ecto-5'-nucleotidase). Previous studies have demonstrated that exosomes may express this couple of adenosine-generating enzymes (Clayton A et al. 2011, Schuler PJ et al. 2014, Amarnath S et al. 2015). By western blot analysis, we demonstrated CD39 and CD73 expression in st- MSCs-derived EVs. Moreover, when performing the scratch assay in presence of an ecto-ATPase inhibitor ARL 67156 and AMP-CP, which inhibit respectively CD39 and CD73, we were able to rescue the migrating block caused by st MSC-derived EV treatment, and preliminary data suggest that the CD39 inhibition leads to a decrease in the ROS accumulation. This supports our hypothesis that st- MSCs-derived EVs, carrying CD39 and CD73, are able to increase the concentration of the extracellular adenosine, which binds adenosine receptors expressed on the surface of endothelial cells, leading to the activation of Nox2 and, in turn, to the production of ROS. Furthermore, the inhibition of ROS accumulation obtained using a generic antioxidant (NAC), fully rescued endothelial cell migration. We are planning to perform further experiments to identify the specific adenosine receptor involved in this process and its connection with Nox2. To validate our results in vivo, we analysed the anti-angiogenic effect of st- MSCs-derived EVs in the retinal vascularization mouse model, a very useful tool in the study of physiologic vessel sprouting (Andreas Stahl et al. 2010). Indeed, in contrast to humans, mouse pups have an immature retinal vasculature at the birth; the development completes in some weeks. The retinal vascularization proceeds in a tightly regulated and organized manner, reliable for detection of any defect (Stahl et al. 2010). Thus, C57BL/6J pups were intra-peritoneally injected with unstimulated or stimulated MSCs-derived EVs, and sacrificed 5 days later to collect retinas. Samples were dissected and stained to measure the vasculature formation of developing retina by confocal microscopy. Remarkably, we observed a decrease in the retina vascular arborisation of pups treated with st- MSCs-derived EVs, but no effect with the unstimulated counterpart. We also took advantage of another mouse model, the matrigel plug assay, which consists in the analysis of the vascularization of a matrigel plug that was previously supplemented with EVs and then implanted in the dorsal back of C57BL/6-N mouse. Both models confirm that st MSC derived-EVs show anti-angiogenic proprieties. Further experiments will be performed to measure the ROS concentration in retinas treated with st- MSCs derived-EVs. Moreover, we will use in vivo st- MSCs-derived EVs knock-down for TIMP-1 (that we already obtained by siRNA approach) with or without CD39 inhibitor to rescue the phenotype. To conclude, our data indicate that MSCs specifically target endothelial cells to control angiogenesis through the release of extracellular vesicles. In particular, when exposed to pro-inflammatory cytokines, MSCs release EVs that strongly inhibit the angiogenic process. This inhibition is mediated by at least two co-operating mechanisms targeting two aspects of the angiogenic process. Thus, on the one hand, EVs-delivered TIMP-1, by inhibiting MMPs, affects the ability of endothelial cells to degrade the surrounding extracellular matrix. On the other, the local production of adenosine, due to the presence of CD39/CD73 enzymes on EVs released by st- MSCs, induces Nox2 activation and ROS production in endothelial cells, thus affecting endothelial cell motility and proliferation. We believe that these findings are of utmost importance to understand the biological role of MSCs and to exploit them in therapy. Indeed, our data confirm the emerging concept of MScs as sensors of the microenvironment to control physiological and pathological events. Thus, while in hypoxic condition MSCs release vesicles enhancing the angiogenic process (Consuelo Merino-González et al. 2016, Gangadaran P. et al. 2017, Jiejie Liuet al. 2015,Suyan Bian et al. 2014), within the inflammatory milieu MScs acquire an anti-inflammatory phenotype by inhibiting both the immune cells’ activity (Soraia C. Abreu et al. 2016, Claudia Lo Sicco et al. 2017) and the immune-associated angiogenesis (Angioni et al., manuscript in preparation; Maffioli E. et al.2017; Zanotti, Angioni et al. Leukemia 2016). This evidence provides new perspectives to exploit MSCs in therapy. EVs derived from st-MSCs may represent a very effective approach for the treatment of pathological angiogenesis for several reasons: they seem to target specifically the endothelium, inhibit the angiogenesis acting on multiple pathways, and are more standardisable than MSCs.