Different pro-angiogenic potential of γ-irradiated PBMC-derived secretome and its subfractions

Secretomes from various cell sources exert strong regenerative activities on numerous organs, including the skin. Although secretomes consist of many diverse components, a growing body of evidence suggests that small extracellular vesicles (EVs) account for their regenerative capacity. We previously demonstrated that the secretome of γ-irradiated peripheral blood mononuclear cells (PBMCs) exhibits wound healing capacity. Therefore, we sought to dissect the molecular composition of EVs present in the secretome and compared wound healing-related activities of these EVs to other subfractions of the secretome and the fully supplemented secretome (MNCaposec). Compared to EVs derived from non-irradiated PBMCs, γ-irradiation significantly increased the size and number and changed the composition of released EVs. Detailed characterization of the molecular components of EVs, i.e. miRNA, proteins, and lipids, derived from irradiated PBMCs revealed a strong association with regenerative processes. Reporter gene assays and aortic ring sprouting assays revealed diminished activity of the subfractions compared to MNCaposec. In addition, we showed that MNCaposec accelerated wound closure in a diabetic mouse model. Taken together, our results suggest that secretome-based wound healing represents a promising new therapeutic avenue, and strongly recommend using the complete secretome instead of purified subfractions, such as EVs, to exploit its full regenerative capacity.

regeneration has not entered the routine clinical practice 7 . More recently, cell transplantation has been shown to elicit tumor formation, graft versus host disease, unwanted immune responses, thrombosis, and infections 8,9 . The main reasons for what so far can be considered a failure are poor cell engraftment and low, if any, capacity to truly transdifferentiate into mature cells. Regulatory restrictions and high costs further complicate the whole scenario. Despite all that, we have learned a lot from these first experiences. The key finding was that adult stem cells exert beneficial paracrine actions, mediated by soluble factors produced and released by the transplanted cells, rather than through transdifferentiation [10][11][12][13][14] . The first clear demonstration of the paracrine hypothesis came from two studies in which the administration of conditioned medium from mesenchymal stem cells was able to repair cardiac damage 10,15 .
Few years later, in 2008, Timmers et al. demonstrated that fractionated conditioned medium products of mesenchymal stem cells and containing only soluble factors smaller that 100-200 nm in size provides cardio protection in an ischemia reperfusion (I/R) injury model 16 . The same group later demonstrated that extracellular vesicles (EVs) are responsible for a reduction in I/R injury in a Langendorff Model 17 . Recently, small EVs derived from stem cells or platelet-rich plasma were shown to significantly contribute to wound healing processes in different rodent models 18,19 .
The literature in regenerative medicine has recently undergone an oblique twist. The "stem cell-centric vision" was silently exchanged for a "stem cell-derived EV-centric vision". As the availability and accessibility of stem cells of different origins are limited and it is time-consuming and extremely expensive to produce them under Good Manufacturing Practice (GMP) conditions 20 , we focused on peripheral blood mononuclear cells (PBMCs) as a source for secretome production. These cells can be obtained easily and are a current waste product in all transfusion units worldwide. In the past, we have shown that the secretome of γ-irradiated PBMCs (MNC aposec ) has regenerative capacity in several preclinical settings. MNC aposec abrogates hypoxia-induced cell damage in acute myocardial infarction and chronic ischemia, attenuates stroke and spinal injury, resolves myocarditis, and augments wound healing [21][22][23][24][25][26][27] . Multiple modes of action have been described, such as the induction of cytoprotection, inhibition of platelet aggregation, induction of vasodilation, endothelial cell proliferation, migration of epithelial and mesenchymal cells, and immune modulation and inhibition of bacterial growth 28 . More recently, we demonstrated the safety and tolerability of autologous MNC aposec produced under GMP conditions in human full-thickness skin wounds 29 .
Diabetic wounds and impaired wound healing represent clinical challenges with an unmet need 30,31 . The process of wound healing involves well-orchestrated biological events, including macrophage invasion, reepithelization, fibroblast migration and proliferation, extracellular matrix deposition, and neo-angiogenesis [32][33][34] . However, in diabetic wounds, the healing process is impaired, which can lead to chronic non-healing wounds, such as diabetic foot ulcers, and, in the worst case, requires limb amputation 32 . Similar to myocardial infarction, the regenerative medicine literature on diabetic wound repair has undergone a change in paradigm. The "mesenchymal stem cell -based therapies" are currently superseded by "EV-based therapies" in experimental wound models 19,[35][36][37][38][39][40] .
Two of our previous publications 27, 41 set the stage for the present study. We have conclusively shown that MNC aposec is superior to the secretome of non-irradiated PBMCs in regards to the healing capacity in a porcine model of acute burn wounds, and that PBMCs release a plethora of proteins, lipids, and EVs into the culture medium, all with different biological activities 41 .
Therefore, the aims of the current study were: 1) to investigate whether γ-irradiation changes the miRNA, protein, and lipid content of EVs released by PBMCs; 2) to test if subfractions of MNC aposec have better wound healing properties in vitro than the whole secretome; 3) to verify whether MNC aposec exerts its regenerative activity in an animal diabetic wound model.

Results
PBMCs exhibit an altered EV secretion profile in response to ionizing radiation. We previously showed that the secretome of γ-irradiated PBMCs contains EVs, which had wound healing properties in in vitro cell culture experiments on epithelial and mesenchymal cells 41 . Therefore, we aimed to investigate in detail the amount, size, and composition of MNC aposec -derived EVs. Figure 1 shows the production of the secretome (Fig. 1A), the analytical procedures performed (Fig. 1B), and the functional tests carried out (Fig. 1C). Analysis of the size and number of EVs released from irradiated and non-irradiated EVs by NanoSight technology revealed a 4-fold higher concentration of EVs derived from irradiated PBMCs compared to non-irradiated cells ( Fig. 2A). The size of EVs from both irradiated and non-irradiated PBMCs ranged from 80 to 280 nm, but the mean diameter of vesicles derived from irradiated PBMCs was significantly larger than the diameter of vesicles released by non-irradiated PBMCs (170.5 ± 4.8 nm vs. 143.8 ± 4.3 nm; Fig. 2B and C), suggesting that different vesicle species are secreted by cells cultivated under stressed and non-stressed conditions. γ-irradiation induces the release of EV-packed proteins involved in intracellular trafficking, immunomodulation, and wound healing. To investigate the protein content of EVs, we cultured irradiated PBMCs from 10 donors in EV-free (pre-centrifuged) medium. After 24 hours, the EVs were harvested, pooled, and the protein content analyzed by SDS-PAGE and HPLC-MS. Silver staining revealed a single band at approximately 67 kDa, corresponding to human serum albumin, in the EV-free medium (Fig. 3A). Uncropped micrographs are available at Supplementary Fig. S1. In contrast, EVs released by irradiated and non-irradiated PBMCs presented numerous proteins of various molecular weights. Because the quantitation of EVs derived from irradiated and non-irradiated cells by silver staining was comparable, we investigated whether they differ qualitatively using a whole proteomics approach. A total of 129 proteins were detected in EVs derived from non-irradiated PBMCs, while 490 secreted proteins were found in EVs after γ-irradiation (Fig. 3B). All proteins identified in the different EVs are listed in Tables S1-S3. Interestingly, almost all proteins released by non-irradiated cells were also present in irradiated PBMCs. Only 16 proteins were exclusively detected in EVs from non-irradiated PBMCs. In contrast, 377 proteins were specifically present in EVs of irradiated PBMCs, suggesting that γ-irradiation significantly changes vesicle formation and/or packaging.
To investigate the possible biological functions of the proteins identified in EVs from irradiated PBMCs, Gene Ontology functional classification was performed. The most significant processes were related to vesicle formation, trafficking, immunological processes, and wound healing (Table 1).

EVs from irradiated PBMCs contain a variety of bioactive phospholipids.
In order to compare the prevalence of different lipid species in EVs from irradiated and non-irradiated PBMCs, we performed thin layer chromatography (TLC). EVs released by PBMCs predominantly contained cholesterol, followed by phospholipids, according to their co-migration with corresponding standards (Fig. 4A). Full silica gel is available at Supplementary Fig. S2. Irradiation of PBMCs strongly increased EV release and the amount of detectable lipids. Detailed analysis of the phospholipids by liquid chromatography-mass spectrometry (LC-MS/MS) indicated that the EVs contained a variety of native and oxidized bioactive phospholipids (Fig. 4B-E). We observed that the amount of phospholipids was increased in the EV preparation from irradiated PBMCs compared to non-irradiated PBMCs (Fig. 4B-E), corroborating our TLC findings (Fig. 4A). The major phosphocholine species detected in the EV lipid preparations by our screening method 42 were 1,2-dipalmitoyl-sn-glycero-3-p hosphocholine (DPPC), 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (PAPC), 1-stearoyl-2 -arachidonoyl-sn-glycero-3-phosphocholine (SAPC), and 1-stearoyl-2-linoleoyl-sn-glycero-3-phosphocholine (SLPC) ( Fig. 4B and C). The relative amount of 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphocholine (PLPC) in EVs could not be determined because its signal in the growth medium lipid extract control introduced high background.
Next, we analyzed whether the EVs contain oxidation products of the unsaturated phospholipids. Oxidized phospholipids have emerged as versatile signaling compounds 43 and can be secreted from cells similar to protein signaling compounds in the regulation of inflammation 44 . The EVs contained the hydroperoxides and hydroxides   γ-irradiated PBMCs exhibit a distinct miRNA secretion signature in EVs. EVs reportedly function as cargo for miRNA from donor cells to recipient cells, mediating communication between different cell types 46,47 . Therefore, we investigated the miRNA composition of EVs and whether γ-irradiation affected the miRNA content of EVs. RNA length profile analysis revealed that EVs carried short RNA fragments up to 80 nucleotides (Fig. S3), mainly representing miRNAs (20-25 nucleotides), whereas PBMCs had RNAs from the entire spectrum of length. In order to investigate the miRNA expression profile of EVs and PBMCs, RNA was isolated and analyzed using an Affymetrix miRNA 4.0 Array. γ-Irradiation significantly changed the miRNA composition in EVs ( Fig. 5A and B). Of the 229 miRNAs detected in EVs from γ-irradiated PBMCs, only 35 (Tables S4 and  S5) were also present in EVs derived from non-irradiated PBMCs, suggesting a strong influence of γ-irradiation on miRNA packaging into EVs. When we compared the miRNA content of γ-irradiated PBMCs to the content of EVs derived from the same cells, we found that a high number of all detected EV-derived miRNAs (188 out of 229; 82%) (Table S6) were exclusively present in EVs, suggesting enrichment of certain miRNAs in EVs. In contrast, a rather small number of detectable miRNAs in PBMCs (41 out of 455; 9%) (Table S7) were transferred to EVs (Fig. 5C, Table S8). The heat map in Fig. 5D shows a quantitative comparison of the 229 miRNA detected in EVs from γ-irradiated PBMCs and the donor cells. These data indicate that γ-irradiation strongly affects the miRNA composition of EVs, which could significantly contribute to its regenerative activity.

MNC aposec exhibits angiogenesis-promoting features.
We previously demonstrated that MNC aposec exhibits strong angiogenic potential in vitro 24 and in mouse and porcine wound models in vivo, and was associated with accelerated wound healing 26,27 . In addition, we demonstrated that purified small EVs were more potent in inducing keratinocyte and fibroblast migration in in vitro scratch assays 41 , and recent publications have suggested that small EVs are key players in neo-angiogenic processes [35][36][37][38][39] . Therefore, we investigated whether purified EVs derived from irradiated PBMCs also demonstrate angiogenic potential in an ex vivo aortic ring assay and compared them to the protein and lipid fraction, as well as the whole secretome. To address the question of whether the pro-angiogenic effects can also be achieved in a diabetic setting, we additionally analyzed and compared vessel sprouting by aortas obtained from wildtype and diabetic (LepR db/db ) mice. The effective dose of MNC aposec was titrated to a concentration of 4 × 10 6 PBMCs/ml (Fig. S4). Though MNC aposec strongly induced sprouting in all of the aortic rings tested, the pro-angiogenic effect was even more pronounced in aortas obtained from diabetic mice (Fig. 6). Next, we investigated which component(s) of the secretome (EVs, proteins, lipids) conferred angiogenesis-promoting properties. The whole secretome had the highest efficiency in inducing vessel sprouting when compared to subfractions (Fig. 6). Though proteins increased sprouting by 45% and 20% in aortic rings from wildtype and diabetic mice, respectively, an induction of vessel sprouting by EVs was only observed in diabetic mice (30% increased sprouting compared to medium alone). Lipids failed to induce vessel sprouting in any setting tested. These data indicate that the full angiogenic potential of MNC aposec is only achieved by applying the whole secretome. In line with the pro-angiogenic effect of MNC aposec and the EV fraction in the aortic ring assay, Vegfa and Ena-78 mRNA expression was induced in aortas from diabetic animals, suggesting a distinct pro-angiogenic response in a hyperglycemic system (Fig. 7) We also investigated whether gene activation is involved in the pro-angiogenic activities of MNC aposec and the purified fractions. As the involvement of activator protein 1 (AP-1) in angiogenic processes is well documented 48 , we chose an AP-1 reporter gene assay to address this question. As expected, we observed a strong induction of AP-1 promotor activity after the addition of MNC aposec (Fig. 8A). The protein and lipid fraction significantly induced the AP-1 promoter by 35% and 55%, respectively. Interestingly, EVs had no effect on the activity of the AP-1 promotor, suggesting that they act most likely via repression of gene expression by miRNAs. As the mitogen-activated protein (MAP) kinase pathways involving heat shock protein 27 (HSP27) have also been shown to be crucial for wound healing 49,50 , we established an HSP27 potency assay and investigated the effects of EVs, proteins, and lipids on this wound healing-associated signaling pathway. Corroborating the findings of the AP-1 promotor assay, HSP27 phosphorylation was strongly induced by the whole secretome. The lipid fraction exhibited weak activity (~20%) compared to MNC aposec , and EVs and proteins exhibited little to no effect (Fig. 8B). Pooling subfractions to generate MNCaposec failed to display an additive effect of EVs, proteins, and lipids, suggesting the presence of further bioactive molecules secreted by γ-irradiated PBMC.

GO Biological Process
MNC aposec accelerates wound healing in LepR db/db mice. As all subfractions of MNC aposec demonstrated only weak activities in our ex vivo and in vitro potency assays, we evaluated the effects of MNC aposec on diabetic wound healing. We excised full-thickness skin wounds on the mid-dorsum of LepR db/db mice, followed by topical administration of MNC aposec or drug vehicle. Compared to control mice, wound closure was accelerated in MNC aposec -treated LepR db/db mice, as indicated by smaller wound circumferences on days 11, 14, 18, and 25 post-wounding (Fig. S5A,B). On day 25 post-wounding, mice were sacrificed and wound morphology was investigated. H&E staining revealed a significantly smaller wound area in MNC aposec -treated mice compared to drug vehicle-treated control group (Fig. S5C), indicating that MNC aposec ameliorates diabetic wound healing. In contrast to our observations in other wound models (i.e., full-thickness skin wounds in wildtype mice and burn wounds in pigs 26,27 ), where blood vessel formation was increased after MNC aposec treatment, CD31 staining of blood vessels remained unaffected by MNC aposec treatment in the present study (Fig. S5D,E).

Discussion
For a long time, cell-based approaches have been considered the promising therapeutic future for promoting tissue regeneration of almost every damaged organ 51,52 . Although this idea was supported by a large number of in vitro cell culture and in vivo animal experiments, there was considerable disappointment when clinical trials in humans revealed only minor effects or failed 7 . Gnecchi et al. were the first group to report that concentrated paracrine factors derived from mesenchymal stem cells are responsible for the tissue regenerative effect in an experimental model of acute myocardial infarction 10 . Since then, numerous studies have further corroborated this finding, partly clarified the biology of paracrine mechanisms and identified the putative underlying mechanisms of secretome-based tissue regeneration 40,53,54 . We reported for the first time that regenerative secretomes consist primarily of proteins, lipids, and EVs, and much effort has been made to identify the factor(s) or subfraction(s) responsible for in vitro activities 41 . In particular, the contribution of small EVs (containing exosomes) to regenerative processes, including wound healing, has been investigated extensively [35][36][37][38][39] . In line with these investigations, we recently demonstrated in in vitro scratch wound assays of human keratinocyte and fibroblast cultures that MNC aposec -derived EVs accelerated wound healing processes 41 . Using a multi-omics approach, we analyzed the constituents of PBMC-derived EVs and their altered composition after γ-irradiation. Recent studies have reported that certain stimuli affect vesicle formation processes, increasing the release of EVs 55 . In line with these data, we demonstrated that γ-irradiation augments the release of  EVs from PBMCs. Moreover, we provide evidence that PBMCs dynamically fine-tune lipid, protein, and miRNA species packaged and released in EVs in response to radiation. As EVs shuttle biomolecules from one cell to the other, γ-irradiation is able to influence functional properties of non-irradiated recipient cells. Our multi-omics analyses showed that the protein, lipid, and miRNA composition of secreted EVs shifts towards a regenerative phenotype upon irradiation. Interestingly, comparative proteome analysis of EVs derived from irradiated and non-irradiated cells revealed 377 proteins that were exclusively found in EVs from irradiated PBMCs compared to non-irradiated counterparts. E.g. galectin-1 56 , serpin B9 56 , lysozym C 57 , HLA class I antigen, leukocyte elastase inhibitor 58 , leukotriene A-4 hydrolase 59 , proteins reportedly involved in immunological processes, were secreted in EVs by irradiated PBMCs. We furthermore detected proteins predominantly associated with wound healing, such as calreticulin 60 , fibrinogen α/β/γ chains 32 , platelet factor 4 61 , and platelet glycoproteins 62 . We therefore hypothesize that the tissue regenerative effect attributed to irradiated PBMC-derived EVs is, at least in part, due to selected proteins packed and secreted in EVs. Delineating their specific functions merits future investigations. The miRNA composition of secreted vesicular EVs changed remarkably after γ-irradiation. Intriguingly, we found that a large number of miRNAs were exclusively present in EVs derived from irradiated PBMCs. The majority of detected miRNAs have not yet been well characterized, making data interpretation delicate. Nevertheless, our results suggest that at least some of these miRNAs are associated with regenerative processes; thus, further studies are needed to clarify their contribution to wound healing. For example, we detected several miRNAs reportedly associated with major wound healing events. One of the best characterized miRNAs present in EVs from γ-irradiated PBMCs was miRNA-21, which has been shown to be upregulated in the skin after injury and to contribute to wound healing by enhancing keratinocyte, fibroblast, and endothelial cell migration, wound contraction, collagen deposition, and angiogenesis 35,63 . In addition, we detected miRNA-126 64-66 , miRNA-30 67 , miRNA-31 68,69 , and miRNA-335, all involved in angiogenic and skin wound healing events, even in a diabetic setting 70 , suggesting that γ-irradiation promotes the release of vesicular miRNAs with tissue regenerative capacity. Intriguingly, purified EVs, containing miRNAs, showed no or only minor effects on vessel sprouting in wildtypeand diabetic mice-derived aortic rings, respectively, suggesting that the main regenerative function of EVs is not dedicated to the induction of endothelial cell proliferation. This discrepancy could be explained by the fact that the majority of studies investigating pro-angiogenic potential of miRNA apply high doses of purified miRNAs without EV packaging.
As bioactive oxidized phospholipids play a crucial role in two critical phases of wound healing, the sensing of cell damage patterns 71 by the innate immune system, and the resolution phase of inflammation 72 , we analyzed the phospholipid content of EVs released by non-irradiated and γ-irradiated PBMCs. Compared to EVs from non-irradiated PBMCs, EVs from γ-irradiated PBMCs contained high amounts of bioactive phospholipids. The altered lipid composition in EVs after irradiation was comparable to our previous study on soluble lipids in MNC aposec41 . PAPC-OOH was enriched in EVs from irradiated PBMCs and has recently been implicated in dendritic cell maturation and differentiation 73 , suggesting that oxidized lipids play a role in shaping the immune response rather than inducing angiogenesis or proliferation and the migration of skin cells as shown previously 41 . Furthermore, lysophosphatidylcholines, a phospholipid species highly abundant in EVs derived from stressed PBMCs, are important precursors of lysophosphatidic acid, a major factor in tissue remodeling and immune regulation during wound healing 74 . Notably, the wound fluid from diabetic rat wounds differs in lysophospholipid composition from controls 75 , suggesting a potential application of PBMC secretome. Finally, POVPC were detected in the EVs from PBMCs after γ-irradiation. This carbonyl-containing, fragmented, oxidized lipid can induce inflammasome activation 76 , which has been shown to accelerate the healing of skin wounds 77 . Tentatively, our results indicate that irradiated PBMCs contribute to wound healing, at least partially, by releasing distinct bioactive phospholipid species in EVs. In a previous study, we found that the EV fraction exerted proliferative and stimulatory effects in vitro 41 , and a growing number of studies have focused on EV-based therapies in wound healing. Therefore, we addressed whether MNC aposec -derived EVs may solely be responsible for skin wound regeneration. In wound healing settings, numerous studies have attributed pro-angiogenic capacities to stem cell-derived EVs, and various underlying mechanisms have been identified thus far (30)(31)(32)(33)(34). In contrast to the existing literature on MSC-derived EVs, we observed diminished sprout-inducing activity in ex vivo aortic rings and reduced activation of AP-1 and HSP27 with the EV-enriched fraction compared to MNC aposec . These data suggest that EVs released by stem cells, PBMCs, or other cells have distinct properties with regard to angiogenesis. Pro-angiogenic activity, activation of the transcription factor AP-1, and HSP27 phosphorylation were remarkably compromised in all subfractions tested. Though secretome-derived proteins exhibited a moderate pro-angiogenic effect, the lipid subfraction failed to induce endothelial cell sprouting in aortic rings obtained from wildtype and diabetic mice. However, the pro-angiogenic effect of EVs derived from MNC aposec was only observed in the diabetic setting. The pro-angiogenic factor VEGFA and wound healing-promoting factor Ena-78 were strongly induced by these EVs in diabetic aortas, but not in wildtype aortas. These findings suggest that different mechanisms account for the pro-angiogenic activity in the two settings. More sophisticated experiments are needed to fully elucidate the signaling pathways involved in diabetic and physiological wound healing events.
In addition, the combination of these three subfractions did not lead to a synergistic effect on AP-1 promoter activity and phosphorylation of HSP27, suggesting that other pro-angiogenic components are present in MNC aposec . Currently we are investigating the contribution of free nucleic acids, present in the secretome (data not shown), to the pro-angiogenic effect of MNC aposec . Together, our data suggest that secretome subfractions concertedly confer tissue-repairing characteristics. Based on the data presented here, we recommend the use of complete secretomes rather than apportioned constituents for clinical wound management. However, since we only investigated neo-angiogenesis in the present study, we cannot rule out that other important regenerative properties of secretome-promoted wound healing are attributed solely to EVs. Further in vivo wound healing studies with secretome subfractions have to be conducted to gain additional functional insights into their mechanism and contribution to tissue regeneration.
Having demonstrated that the whole cell secretome is necessary for full regenerative capacity, we applied MNC aposec to full-thickness skin wounds in a hyperglycemic mouse model (LepR db/db ) to investigate whether MNC aposec represents a promising therapeutic option for non-healing diabetic foot ulcers. In accordance with previous observations from wildtype mouse models 26 and a porcine burn model 27 , MNC aposec also accelerated wound closure in LepR db/db mice. In contrast to these models 26,27 , where pro-angiogenic effects were observed during the early phase of wound healing, we detected no difference in blood vessel formation in MNC aposec -treated LepR db/ db compared to wildtype 25 days after wounding. Assessment of angiogenesis-related events at earlier time points has to be conducted to fully elucidate the pro-angiogenic potential of MNC aposec in diabetic mice.
Although we could demonstrate less angiogenic potential of single purified subfractions of cellular secretomes compared to the whole secretome, our study has some limitations. Though we have shown that the use of human secretomes in rodents is harmless and efficient in several previous studies 28 , potential xenogenic influences cannot be completely ruled out in our diabetic mouse model. Further experiments with allogenic MNC aposec will deepen our understanding of the regenerative action of MNC aposec and its subfractions in diabetic wound healing. For consistency with the medicinal product used in clinical studies in humans, we pooled secretomes and subfractions from 10 donors. We were therefore not able to determine possible donor-to-donor variations in the present study.
In conclusion, our study provides encouraging data that have paved the way for a clinical trial testing the efficacy of allogenic MNC aposec on non-healing diabetic foot ulcers. As MNC aposec has already passed phase I safety trials, patient recruitment for phase II is already in progress.

Materials and Methods
Ethics statements. Human leukocyte concentrates were obtained from healthy volunteers as a byproduct of the thrombocyte donation procedure. This study was approved by the local ethics committee at the Medical University of Vienna (vote: EK No: 1326/2013). All donors provided informed written consent and demographic data of individual donors are listed in Table S9. Animal experiments were approved by the Animal Protocol Review Board of the City Government of Vienna, Austria (vote: 190097/2015/9) and were performed in accordance with the Austrian guidelines for the use and care of laboratory animals. All methods were performed in accordance with relevant guidelines and applicable regulations.
Generation of the secretome of human PBMCs. PBMCs were isolated as described previously 27,41 .
After isolation, cells were resuspended in serum-free CellGenix culture medium (Freiburg, Germany) at a density of 25 × 10 6 cells/ml and γ-irradiated with 60 Gy. After 20 hours of incubation, cell suspensions were centrifuged to eliminate cells and cellular debris and passed through a 0.2-µm pore filter (Whatman, GE Healthcare, Little Chalfont, UK). In anticipation of potential clinical use, the secretome of irradiated PBMCs was also produced according to GMP guidelines as described previously 24 . To reduce high donor-to-donor variability, pools of 10 donors were used (Table S10). Concentrations of interleukin 8 (Il-8) and transforming growth factor beta 1 (TGF-β1) were assessed in MNC aposec of individual donors and pools of MNC aposec and revealed smaller variability between pooled samples compared to individual samples (Table S11). To reduce donor-to-donor variability, pooled samples were used for experiments. Isolation of MNC aposec components. For functional in vitro assays, different components of the MNC aposec were isolated (Fig. S6). Lyophilized GMP MNC aposec was reconstituted in 0.9% NaCl to obtain the initial concentration of 25 × 10 6 cells/ml. Proteins were precipitated by the addition of polyethylene glycol 4000 (30% w/v; Fluka BioChemika, Sigma-Aldrich) on ice for 15 minutes and centrifuged at 20,000 g for 15 minutes at 4 °C. Total lipids were extracted using the chloroform/methanol-assisted Folch method as described previously 78 . EVs were isolated as described below. All fractions were reconstituted in serum-free CellGenix culture medium, corresponding to a concentration of 25 × 10 6 cells/ml, and stored at −80 °C until further use. To ensure comparability, proteins and lipids were used in the same concentrations as present in MNC aposec .
Isolation of EVs and NanoSight analysis. EVs were isolated by ultracentrifugation of the secretome of irradiated and non-irradiated human PBMCs. To obtain EVs, secretomes were centrifuged at 20,000 g for 20 minutes and at 110,000 g for 120 minutes at 4 °C in an SW 41 swinging bucket rotor (Beckman Coulter). For lipid, protein, and RNA analysis, irradiated and non-irradiated PBMCs were cultured in pre-centrifuged (220,000 g for 24 hours at 4 °C) serum-free CellGenix to deplete EVs from the culture medium. The absolute number and size of EVs were assessed using a NanoSight, NS500 instrument (Malvern Instruments, Malvern, UK). The nanoparticles were visualized by laser light scattering and the motion of each nanoparticle tracked from frame to frame by NTA3.1 software. Recorded videos were analyzed by the NTA3.1 software. For comparability, isolated EVs were employed in the same concentrations as EVs present in MNC aposec . Resolution was set to 70.000 for full MS scans, ions with single charge were excluded from MS/MS analysis and fragmented ions were excluded for 60 seconds from further fragmentation. Generated raw files were converted into mgf by applying MS Convert (http://proteowizard.sourceforge.net/tools.shtml). Database search (SwissProt_ human, version from November 2017)) was performed by submitting the mgf files to MASCOT v. 2.6.0 (Matrix Science, London, UK) through ProteinScape (Bruker, Bremen, Germany) using following parameters: Taxonomy: Homo sapiens; Modifications: carbamidomethyl on C as fixed, deamidation on N and Q, carboxymethylation on M and phosphorylation S and T as variable; Peptide tolerance was set to 20 ppm and the MS/MS tolerance to 0.05 Da; Trypsin was selected as the enzyme used; False discovery rate (FDR) was set to 1% and the decoy database search was used for estimating the FDR; The Mascot search results were further refined by applying an additional data analysis step using Scaffold v4.8.4 (www.proteomesoftware.com).

RNA isolation and transcriptome analysis.
Functional analysis of EV-derived proteins. The WEB-based Gene Stet Analysis Toolkit (WebGestalt) database was used to functionally analyze the 490 proteins found in EVs released by irradiated PBMCs. WebGestalt enables the detection of enrichment of gene ontology (GO) terms in a set of proteins of interest. The whole human genome was used as a reference set for enrichment analysis and the Benjamini-Hochberg method for multiple testing was applied with a significance level of p ≤ 0.05 and FDR < 5%. Biological processes were manually assigned to clusters associated to vesicle formation and transportation, wound healing, and immunological processes, respectively.
Thin layer chromatography. Total lipids were extracted from EVs from irradiated and non-irradiated PBMCs (pool of 5 donors) cultured in pre-centrifuged serum-free CellGenix medium. Purified EVs were resuspended in 150 µl deionized water and 1350 µl chloroform/methanol (2:1 v/v) supplemented with 0.01% butylated hydroxytoluene (BHT) and aqueous formic acid (1/4 the volume of chloroform/methanol). After centrifugation, the lower organic phase was collected and organic solvents evaporated under an argon stream to avoid oxidation. TLC was performed as described previously 79 .
Liquid chromatography-mass spectrometry. For phospholipid analysis by mass spectrometry, lipids were isolated from irradiated and non-irradiated PBMCs (pool of 10 donors) as described previously 42 . LC-MS/ MS analysis of oxidized and non-oxidized phospholipids was performed at the FTC-Forensisch-Toxikologisches Labor BetriebsgmbH (Vienna) as described previously 42 . High technical reproducibility was verified 42 . Sample peak areas were normalized to the peak areas of the spiked DNPC (internal standard). Animal model. C57BL/KsJm/Leptdb (db/db) mice were obtained from Charles River. Twelve-week-old symptomatic diabetic db/db mice were used for the experiments. To determine blood glucose levels, blood was drawn from the tail vein and analyzed using an Akku-Chek Go glucose meter (Roche Diagnostics, Germany). All animals exhibited severely increased blood glucose levels ≥ 300 mg/dl. For skin wounding, the animals' dorsal hair was shaved and completely removed using a depilatory cream. Aortic ring assay. Aortic ring assay was performed as described previously 1 with minor modifications.
Murine aortas were excised, cut into rings of approximately 1 mm length and sandwiched between two layers of 2 mg/ml human fibrinogen (EMD Millipore, Darmstadt, Germany), 43.3 µg/ml aprotinin (Sigma-Aldrich), and 0.6 U/ml thrombin (Sigma-Aldrich). After polymerization, the fibrin-aorta-fibrin gel was equilibrated with M199 medium (Gibco by Life Technologies) supplemented with 4 mM L-glutamine (Gibco by Life Technologies), antibiotic solution (100 U penicillin, 100 µg/ml streptomycin; Gibco by Life Technologies), 250 ng/ml amphotericin B (Fisher Bioreagents, Fisher Scientific), and 10% fetal calf serum (PAA Laboratories, Pasching, Austria) for 45 minutes. The medium was then changed to complete M199 medium supplemented with MNC aposec corresponding to 1 × 10 6 PBMCs, 2 × 10 6 PBMCs, 4 × 10 6 PBMCs, and 8 × 10 6 PBMCs and EVs, proteins, or lipids added at a concentration corresponding to 4 × 10 6 PBMCs. Aortic rings were incubated for up to 5 days and medium was changed every other day. On day 5, aortas were incubated with calcein (Thermo Fisher Scientific) for 1 hour to label viable sprouting endothelial cells. The outgrowth was assessed using an Olympus IX83 scanning microscope (Tokyo, Japan) with cellSens Imaging Software (Olympus, Tokyo, Japan). The outgrowth area was quantified using ImageJ software.

HSP27 phosphorylation.
To detect HSP27 phosphorylation at residue Ser82, A549 cells were treated for 30 minutes, fixed, and permeabilized. Phosphorylation was assessed by sequential addition of an antibody detecting phosphorylated HSP27, followed by a peroxidase-conjugated anti-rabbit IgG antibody and chemiluminescent peroxidase substrate. The plates were read in the luminescence reader (EnVision, Perkin Elmer or Centro LB960, Berthold). The relative potency was calculated for the analyzed samples compared to the standard using the PLA software (Stegmann Systems PLA).

Reverse transcription(RT)-quantitative polymerase chain reaction (qPCR). Cultured aortic rings
were manually excised together with outgrown cells from the surrounding fibrin matrix. RNA isolation, RT and qPCR were performed as described previously 80 . qPCR probes. Murine primers and probes were obtained from Thermo Fisher Scientific. Probes were conjugated to 5′ 6-carboxyfluorescein-major groove binder (FAM-MGB) and 3′ non-fluorescent quencher (NFQ matched subjects (time course data) was performed. Bonferroni post-tests were carried out to compare test groups versus controls. Data obtained by aortic ring assays were evaluated by two-way ANOVA and Sidak and Dunnett multiple comparison post hoc tests.