Approx. 35-50% of the patients receiving matched related or unrelated allogeneic stem cell transplantation, develop severe forms of graft-vs.-host disease (GvHD; Grade II-IV) that cannot be controlled with corticosteroids in up to 50% of the GvHD patients.Owing to the lack of confirmed treatment options Le Blanc et al. introduced mesenchymal stem cells (MSCs) as a strategy to treat severe therapy-refractory acute GvHD.Since then, a number of different studies have addressed the impact of MSC administration on GvHD with different outcomes.New data suggest that the beneficial effects of MSCs rather derive from secreted, immune response-modulating factors than from their tissue intercalation themselves.On the basis of a preclin. myocardial infarction model, evidence was provided that the immune-modulating factors of MSCs are also secreted ex vivo and reside in supernatant fractions that are highly enriched for small extracellular vesicles (EVs) such as exosomes4 and presumably small microvesicles (MVs) that bud from the plasma membrane.Exosomes are naturally occurring EVs (70-140 nm) that are released by exocytosis upon fusion of the so-called multivesicular bodies with the plasma membrane.Containing lipids, proteins and RNA, exosomes seem to participate in the intercellular communication processes.Together with the other EVs, exosomes are released from many cell types and can be enriched from virtually all body fluids including blood plasma, urine and saliva.Depending on their origin, some exosomes exert immune stimulatory or immune suppressive functions.Considering MSC exosomes as the therapeutically active component of MSCs, the application of exosomes provides a number of advantages compared with the MSC application itself.Exosomes are non-self-replicating and owing to their small size can be sterilized by filtration.Thus, the regulatory items to produce exosome fractions for clin. treatment strategies should be less complicated than for any cellular therapeutic of in vitro expanded cells.Furthermore, it might be possible to harvest exosomes from supernatants of immortalized MSC cell lines, which otherwise could not be used for cellular therapies.Given the history of a therapy-refractory GvHD patient (Supplementary Figure S1), we decided to treat the patient in an individual treatment attempt with an exosome-enriched fraction processed from the collected MSC supernatants instead of administering the MSCs themselves.According to reported difference in the outcome of MSC therapies in GvHD patients, we assumed that the exosome preparations from different MSCs might differ in their immune-modulating activities.Hence, MSCs of four different unrelated bone marrow donors were raised initially.The MSC nature of expanded cells was confirmed by flow cytometry and in differentiation assays according to the standards (data not shown).MSC-conditioned media were harvested every 48 h.To sterilize the supernatants and to remove larger EVs the conditioned media were passed through 0.22 μm filter membranes.As small EVs and viruses share a variety of features, and viruses can be purified by polyethylenglycol (PEG) precipitation, we set up a PEG-based protocol to enrich for exosomes and potentially other EVs (Ludwig et al., in preparation).We used PEG6000 (10% final concentration) to enrich for the EV fractions of the sterilized supernatants.To remove residual polyethylene glycol and soluble proteins that might have been coprecipitated, pellets were washed with 0.9% NaCl and reprecipitated at 100.000 g for 2 h.Obtained pellets were dissolved in 0.9% NaCl to a final concentration of 1 unit EVs per 1 mL and stored as 1 mL aliquots at -80 °C until usage.The yield of an EV fraction prepared from supernatants of 4 × 107 MSCs that had been conditioned for 48 h was defined as 1 unit.The EV content of the four processed MSC supernatants was determined by nano-particle tracking anal. on the Zetaview (Particle Metrix, Diessen, Germany) and the NanoSight (Amesbury, UK) platforms.The particle and protein content of the four independent MSC supernatant fractions were all in same ranges (1.3-3.5 × 1010 particles/unit; 0.5-1.6 mg/unit).The average particle sizes calculated varied for the different preparations between 99-123 nm (ZetaView) and 133-138 nm (NanoSight).Western blot analyses demonstrated the presence of the exosomal marker proteins Tsg101 and CD81 in all fractions.These findings together with the vesicular appearances of obtained particles within transmission electron microscopy (Supplementary Figure S2C) indicated that the obtained samples were highly enriched for exosomes.However, owing to the general lack of methods to phys. sep. small MVs and exosomes from each other, we cannot exclude the fact that the MSC exosome-enriched fractions also contain proportions of other small EVs such MVs.As we and others have associated human leukocyte antigen-G (HLA-G) and transforming growth factor beta with exosomes, the four MSC exosome-enriched fractions were analyzed for their content of anti- and pro-inflammatory as well as apoptosis-inducing mols.Apart from the pro-inflammatory cytokines interferon gamma (IFN-γ) and interleukin-8 (IL-8), the exosome preparations contained high quantities of the anti-inflammatory mols. IL-10, transforming growth factor beta and HLA-G.Neither the pro-inflammatory cytokines IL-1β, IL-2, IL-6, IL-17a, IL-21 and TNF-α, the anti-inflammatory cytokines IL-1βRa and IL13, nor the apoptosis-inducing mol., soluble Fas Ligand were detected in any of the exosome preparationsEspecially, the MSC1 exosome preparation (MSC1 Exo) contained elevated transforming growth factor beta levels.In addition, MSC1 Exo differed from the other preparations with regard to the anti-inflammatory cytokine IL-10 and the pro-inflammatory cytokine IFN-γ.The ratio of IL-10 to IFN-γ varied between the four exosome preparations for almost two orders of magnitude; the highest ratio was found for MSC1 Exo (1.02).To avoid the lack of any material required for the clin. application and owing to its high content of anti-inflammatory mols., we considered MSC1 Exo as the fraction of choice.To reduce the risk of any unexpected reaction of the patient's immune response on MSC1 Exo administration, the preparation's immune-modulatory impacts on peripheral blood mononucleated cells (PBMCs) and natural killer cells of the patient were tested in mixed lymphocyte reactions (owing to the limitations of the patient's material, the other MSC exosome-enriched fractions could not be included here).To reflect different and reproducible allogeneic transplantation settings, HLA class I-neg. K562 as well as stable HLA-E*01:03 or HLA-B27-transfected K562 variants were used as allogeneic target cells.In the presence of MSC1 Exo the number of stimulated patient-derived PBMCs releasing IL-1β, TNF-α and IFN-γ (P=0.026, Mann-Whitney test) or stimulated natural killer cells releasing TNF-α or IFN-γ were decreased (Figure 1a).After demonstrating the immunosuppressive activities of MSC1 Exo and according to the dismal situation of the patient, we decided to treat the patient with the MSC1 Exo.Following the application regime of MSCs in GvHD patients, according to which 0.4-9.0 × 106 MSCs per kg body weight are usually administered,15 the amount of the MSC1 Exo obtained from the supernatant of 4 × 107 MSCs was calculated as a corresponding dosage for the body weight of this patient and was defined as 1 unit.To reduce the risk of potential side effects, only a tenth of a MSC1 Exo unit was initially administered.As within 2 days no side effects were observed, unit amounts were gradually increased and administered every 2-3 days until a four-time dosage (4 units) was reached (Figure 1b).All applications were tolerated very well and no side effects were detected.To monitor potential impacts of MSC1 Exo on the immune response of the patient's PBMCs, blood samples were taken after each administration.The cytokine responses of the different PBMC fractions, that is their release of IL-1β, TNF-α and IFN-γ toward K562 cells and their HLA-E*01:03 and HLA-B27 expressing counterparts, were analyzed simultaneously.Remarkably, following the third MSC1 Exo application the patient's PBMCs reduced their cytokine response toward the different K562 variants (Figure 1b).Compared with the cytokine responses of the patient's PBMCs before the MSC-exosome therapy, the numbers of IL-1β (P=0.014, Kruskal-Wallis test), TNF-α (P=0.0012, Kruskal-Wallis test) and IFN-γ (n.s. owing to the low number of measurements) producing PBMCs were reduced more than 50% after the last application.As this reactivity profile corresponds to the results of the in vitro experiments, it is tempting to speculate that MSC1 Exo impaired the in vivo capability of the patient's PBMCs to release pro-inflammatory cytokines.In line with the reduced pro-inflammatory cytokine response of the patient's PBMCs during the course of MSC-exosome therapy, the clin. GvHD symptoms improved significantly shortly after the start of the MSC-exosome therapy.The diarrhea volume was objectively reduced after exosome therapy (Figure 2a).The cutaneous and mucosal GvHD showed a remarkable response within two weeks, which was stable even after 4 mo following the MSC-exosome therapy (Figure 2b).Owing to the clin. response, the dosage of the steroids could be reduced from 125 mg/d before to 30 mg/d after the MSC-exosome therapy.The patient was stable for several months.Even though the patient died of pneumonia 7 mo post exosome application, the obtained results let us conclude that MSC-derived exosomes may provide a potential new and safe tool to treat therapy-refractory GvHD and potentially other inflammation associated diseases.