Small Extracellular Vesicles from Human Fetal Dermal Cells and Their MicroRNA Cargo: KEGG Signaling Pathways Associated with Angiogenesis and Wound Healing

The use of cell secreted factors in clinical settings could be an alternative to conventional cell therapy, with the advantage of limiting concerns generally associated with traditional cell transplantation, such as tumorigenicity, immunoreactivity, and carrying of infections. Based on our published data, we predict a potential role for extracellular vesicles (EVs) in contributing to the proangiogenic activity of human fetal dermal cell secretome. Depletion of nanosized EVs from secretome significantly impaired its ability to induce formation of mesh-like structures in vitro. The isolated EVs were characterized for size and concentration by nanoparticle tracking analysis, and for protein markers (Rab5+, Alix+, CD63+, and calnexin−). The microRNA profile of EVs revealed 87 microRNAs significantly upregulated (≥15-fold increase) in fetal compared to adult dermal cell-derived EVs. Interestingly, these upregulated microRNAs included microRNAs with a validated role in angiogenesis according to literature. Moreover, the DIANA-TarBase v7.0 analysis confirmed enrichment in the KEGG signaling pathways associated with angiogenesis and wound healing, with the identification of putative target genes including thrombospondin 1. To validate the in silico data, EVs were also characterized for total protein contents. When tested in in vitro angiogenesis, fetal dermal cell-derived EVs were more effective than their adult counterpart in inducing formation of complete mesh-like structures. Furthermore, treatment of fibroblasts with fetal dermal-derived EVs determined a 4-fold increase of thrombospondin 1 protein amounts compared with the untreated fibroblasts. Finally, visualization of CSFE-labeled EVs in the cytosol of target cells suggested a successful uptake of these particles at 4-8 hours of incubation. We conclude that EVs are important contributors of the proangiogenic effect of fetal dermal cell secretome. Hence, EVs could also serve as vehicle for a successful delivery of microRNAs or other molecules of therapeutic interest to target cells.


Introduction
The list of clinical conditions related to insufficient angiogenesis is wide, ranging from cardiovascular diseases to impaired wound healing [1]. Therefore, there is a great interest in developing clinical strategies ensuring vasculature formation, such as delivery of different cell populations, or administra-tion of proangiogenic growth factors. Among the cell-based therapies, the use of mesenchymal stromal cells (MSCs), which are closely related to pericytes and produce diverse proangiogenic factors, is a promising approach with the potential to stimulate vasculature tissue development [2].
MSC-based therapy applied to regenerative medicine counts hundreds of registered clinical trials with excellent records of safety and efficacy (http://www.clinicaltrials.gov and http://www.clinicaltrialsregister.eu). Despite earlier works ascribing the therapeutic effects of MSCs to their ability to engraft and differentiate to form new permanent tissues, the current consensus view is that MSCs are shortlived after delivery and exert therapeutic benefits through secretion of bioactive factors [3]. In support of MSC paracrine activity, animal studies have shown that administration of MSC-derived soluble factors recapitulates the effects of cell-based therapy [4]. Consequently, the attention has been brought to the vast array of molecules produced by MSCs [5,6]. The mixture of growth factors, cytokines, chemokines, and extracellular vesicles (EVs) released by cells is known as secretome and can be collected as cell culture conditioned medium (CM).
Human fetal skin cell therapy has been used to replace older skin cell therapy to treat patients with skin ulcers [7,8], as well as burns [9,10], thus resulting in a safe and more efficacious procedure. At that time, the authors suggested the differential gene expression profiling observed in fetal versus adult skin cells as responsible for the efficacy of fetal skin cell therapy [8]. Moreover, since no trace of fetal skin cells was found in recipient biopsies, a paracrine mechanism of healing was suggested [10]. Although an extensive characterization of cell phenotype and secreted factors was missing, those studies open up new research perspectives on the use of cell secretome for regenerative medicine applications.
In a previous study [11], we isolated human fetal dermal cells, which we named multipotent fetal dermal cells based on in vitro characteristics, such as the MSC-like immunophenotype, the multilineage differentiation potential, and the low immunogenicity. We also confirmed the practical advantages of culturing fetal skin cells in comparison to adult skin cells described by others [7,9,10], including faster isolation technique and higher proliferative capacity of fetal cells. More recently [12], we used liquid chromatography and tandem mass spectrometry (LC-MS/MS) to show a set of proteins related to angiogenesis and wound healing, which were significantly upregulated in fetal dermal cell secretome compared to adult dermal cell secretome. Proteome finding was corroborated by the remarkable in vitro proangiogenic capacity of fetal dermal cell secretome compared to its adult counterpart.
In the present study, we investigated whether the presence of EVs could contribute to the biological functions of fetal dermal cell secretome. EVs are released by several cell types and are essential for cell-to-cell communication [13]. These particles are internalized by target cells [14] and once in the cytosol, discharge their material such as proteins, mRNAs, and microRNAs (miRNAs). Administration of MSC-derived EVs has been shown to have beneficial effects in various animal models of organ injury by regulating angiogenesis, cell proliferation, cell migration, and collagen synthesis [15,16]. Nevertheless, EV-based therapy for skin repair consists only of one registered clinical trial aims at studying the effects of plasma-derived exosomes on cutaneous wound healing (http://www.clinicaltrials.gov, NCT02565264).
Herein, EVs were isolated from secretome of fetal and adult dermal cells and characterized for size and concentration by nanoparticle tracking analysis (NTA). Then, we tested the capacity of EV-depleted secretome in inducing in vitro angiogenesis and migration of target cells such as human umbilical vein endothelial cells (HUVECs) and fibroblasts. The miRNA expression profile of fetal dermal cell-vs. adult dermal cell-derived EVs was also analyzed, and a bioinformatics approach was used to identify the Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathways most likely affected by these miRNAs. The predicted angiogenic/wound healing-related effects of fetal dermal cellderived EVs were further validated in in vitro cell-based assays of angiogenesis and cell migration by preconditioning of target cells with different concentrations of EV preparations. Finally, we investigated whether carboxyfluorescein succinimidyl ester-(CSFE-) labeled EVs from fetal dermal cells might be taken up by target cells.

Materials and Methods
2.1. Cell Procurement. Fetal skin biopsies were taken from 20-to 22-gestational-week human fetuses obtained from therapeutic abortions, according to a protocol approved by ISMETT's Institutional Research Review Board (IRRB/00/2015) and Ethics Committee. Signed informed consent form was obtained from each donor. Fetal dermal cells were isolated and characterized as previously described [11]. Adult skin biopsies (45-55-year-old donors) were provided by Istituto Humanitas (Rozzano, Milan). Fetal and adult dermal cells were isolated under the same conditions. Normal dermal fibroblasts and HUVECs were purchased from Thermo Fisher Scientific (Waltham, MA, USA) and used as target cells in in vitro cell-based assays.

Cell
Cultures, CM Collection, Isolation, and Storage of EVs. Fetal dermal cells and adult dermal cells, previously cryopreserved in Dulbecco's modified Eagle's medium (DMEM) (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% dimethyl sulfoxide (DMSO) (CryoSure, WAK-Chemie Medical GmbH, Steinbach, Germany) and 30% fetal bovine serum (FBS), were grown in 75 cm 2 tissue flasks (SARSTEDT, Numbrecht, Germany) as previously described [11]. For CM collection, serum-free alphaminimum essential medium (MEM) (Gibco, Thermo Fisher Scientific) was added to 80% confluent cells and collected 24 hours later. EVs were isolated by differential ultracentrifugation of CM according to a protocol [17] and as previously described [18]. In brief, CM was centrifuged at 1800 × g for 10 minutes to remove cell debris, centrifuged again at 17000 × g for 15 minutes, and then at 160000 × g for 1 hour in the ultracentrifuge (Optima MAX-XP, Beckman Coulter Inc., Irving, TX, USA). All centrifugation steps were done at 4°C. The pellet was resuspended in phosphate-buffered saline (PBS) without Ca 2+/ Mg 2+ (Sigma-Aldrich) or subjected to total protein extraction or total RNA extraction.  [20]. We selected DIANA-TarBase v7.0 for analysis and set the p value ≤ 0.05 to analyze miRNAs and their target genes. KEGG enrichment analysis was used to identify signaling pathways most enriched by our miRNAs. In order to identify single genes targeted by multiple miRNAs, KEGG analysis was performed with the "genes intersection" option. . As a negative marker of EVs, a rabbit monoclonal antibody against calnexin (C5C9, 2679, Cell Signaling; 1 : 1000 dilution) was used. For this latter, total protein extracts from cells were also included in the analysis. For validation of the in silico analysis, fibroblasts were treated with 10 μg/ml of fetal-derived EVs for 24, 48, and 72 hours. 30 μg/lane of fibroblast total protein extracts was separated by SDS polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Bio-Rad Laboratories). The membranes were incubated 1 hour at room temperature with a rabbit monoclonal antibody against recombinant human thrombospondin 1 (THBS1) (ab267388, Abcam, Cambridge, UK; 1 : 1000 dilution). Betaactin was used as an internal loading control (sc-81178, Santa Cruz Biotechnology; 1 : 1000 dilution). After three washings with T-TBS, the membranes were incubated for 1 hour at room temperature with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology; 1 : 10000 dilution). After washing, the signal was detected with an enhanced chemiluminescence reagent (ECL; Amersham, Arlington Heights, IL, USA). Densitometric analysis of Western blot analysis was done with the Image Lab software, version 6.0.1 (Bio-Rad Laboratories).

EV/CM-Induced
In Vitro Angiogenesis. Serum-starved HUVECs were resuspended in serum-free culture medium supplemented with different concentrations of EVs (100, 50, 10, or 5 μg/ml) from fetal and adult dermal cells. Approximately 10000 cells/well were plated in triplicate onto Matrigel from the in vitro angiogenesis assay kit (Millipore, Billerica, MA, USA) in flat bottom 96-well plastic plates (Costar Corning Inc., Costar, NY, USA). HUVECs plated in fetal dermal cell CM or in serum-free culture medium were used as positive and negative control, respectively. Plates were incubated at 37°C in a humidified atmosphere with 5% CO 2 . Formation of mesh-like structures was monitored for 24 hours with an inverted microscope (Olympus CKX41, Tokyo, Japan) coupled with a camera (Olympus U-TV0.5XC-3) for image acquisition. Numerical values (score from 0 to 5) were assigned to each pattern according to manufacturer's specifications (Millipore) and as previously described [12]. Formation of mesh-like structures was quantified by calculating the number of junctions, nodes, total mesh area, and total segments length, with ImageJ software of the Angiogenesis Analyzer plugin (https://imagej.nih .gov/ij/). Analyzer (RTCA) dual purpose (DP) instrument (Acea Biosciences Inc., San Diego, CA, USA), which detects the realtime migration of cells [21]. Briefly, serum-starved fibroblasts were resuspended in culture medium supplemented with 100, 50, 10, or 5 μg/ml EVs from fetal and adult dermal cells or in culture medium without EVs, and added to the upper chamber (30000 cells/chamber) of the CIM plate. Culture medium 0.5% FBS was used as chemoattractant and loaded to the lower chamber. The CIM plates were assembled into the RTCA-DP instrument and placed in the incubator at 37°C in a humidified atmosphere with 5% CO 2 . Cell migration was recorded every 15 minutes for 7 hours. Each time point was calculated from duplicate values, and cell migration was expressed as cell index (CI) at a 7-hour time point. Analysis was performed with the RTCA Software 1.2 of the xCELLigence system.   Stem Cells International   7 Stem Cells International different samples corresponding to the remaining treatments (adult-derived EVs, fetal-derived CM, and the corresponding EV-depleted CM) were analyzed. For cell migration assay, three different samples of each condition were analyzed (fetal-derived EVs, adult-derived EVs, fetal CM, and the corresponding EV-depleted CM). For quantitative analysis by Luminex of whole CM vs. EV-depleted CM, ten fetal CM samples were analyzed. Data were analyzed with R [23] and expressed as the mean ± SD. Data from two different groups were compared with Student's t-test. Differences between the groups were considered significant at a p value of ≤ 0.05.
Angiogenesis data were analyzed with GraphPad Prism 8.4.2 and expressed as the mean ± SD. Data from each condition were compared to treatment with fetal-derived EVs with the one-way ANOVA.

Characterization of EVs by Nanoparticle Tracking
Analysis and Protein Marker Expression. The NTA of pellet preparations obtained by differential ultracentrifugation of CM from fetal and adult dermal cells indicated a "mode" diameter size (representing the size of the most abundant particles in a sample preparation) consistent with that of "small EVs" (<200 nm [24]). The mode sizes of a representative fetal sample (77:5 ± 0:8 nm) and that of a representative adult sample (87:2 ± 2:8 nm) are shown in Figure 1(a) and Figure 1(b), respectively. The concentration of EVs from both fetal and adult cells was in a range of 10 11 -10 12 particles/ml. Moreover, characterization by Western blot analysis revealed that pellet particles expressed Rab5, Alix, and CD63 proteins, and while were negative for calnexin protein, this latter found in total protein extracts from human fetal dermal cells (Figure 1(c)).

Differential Expression of EV miRNAs and Their
Association with Signaling Pathways Related to Angiogenesis and Wound Healing. We found 87 highly expressed miRNAs (Ct values ≤ 26) in fetal dermal cell-derived EVs. These highly expressed miRNAs were also considered significantly upregulated in fetal compared to adult dermal cell-derived EVs (Table 1; Supplemental Material S1). In addition, 21 miRNAs had a validated role in angiogenesis according to literature (Table 1; Supplemental Material S1).

Validation of the In Silico Analysis by Western Blot.
The amount of THBS1 protein in fibroblasts treated for 72 hours with 10 μg/ml of fetal-derived EVs was higher than the amount of THBS1 protein in both untreated fibroblasts (Figure 4(a)) and the earlier time points of treatment (24 and 48 hours, data not shown). The amount of THBS1 protein in fibroblasts following a 72-hour treatment with fetalderived EVs was approximately 4-fold higher than the amount of THBS1 protein in untreated fibroblasts at the same time point (Figure 4(b)).  (Table 3). Lower doses of fetal dermal cellderived EVs (≤5 μg/ml) were ineffective in inducing formation of mesh-like structures (data not shown). The results were confirmed by angiogenesis parameters quantified on images (n = 4 for treatment with fetal-derived EVs; n = 3 for the other treatments). In the case of HUVECs cultured in the presence of 10 μg/ml of adult dermal cell-derived EVs, histograms were lower to those of HUVECs cultured in the presence of fetal cell-derived EVs, but differences were not statistically significant, probably due to the high SD values. Data were expressed as total mesh area ( Figure 6(a)), number of junctions ( Figure 6(b)), number of nodes (Figure 6(c)), number of segments ( Figure 6(d)), and total length of segments ( Figure 6(e)) ( * p value ≤ 0.05; * * p value ≤ 0.001).  9 Stem Cells International respectively) migrated toward the lower chamber (containing culture medium 0.5% FBS as chemoattractant) with similar CI values. A lower concentration of EVs (5 μg/ml) (light blue curve) resulted in the absence of migration, and CI values similar to those of negative control fibroblasts seeded in the absence of EVs (pink curve) (Figure 7). The CI values of cell migration induced by 10 μg/ml of adult-derived EVs were similar to those induced by EVs of fetal origin (dark blue and orange curves, respectively) (data not shown).

Amount of Growth Factors and Chemokines in EV-
Depleted CM. The levels of growth factors and chemokines in EV-depleted CM were similar to the levels in whole fetal dermal cell CM. The growth factor detected with the highest amount was VEGF-A, while the chemokine detected with the highest amount was SDF-1-alpha (Table 4).

Cellular Uptake of EVs.
Labeling of fetal dermal cellderived EVs with CSFE was successfully verified by flow cytometry. We clearly identified a discrete population of fluorescent particles in the range of small EVs compared to control, unlabeled EVs (Figure 8(a)). The fluorescent signal of CSFE-labeled EVs was visualized into the cytosol starting from 4 hours of incubation (Figure 8(b)), while at later time  10 Stem Cells International points (8 hours of incubation; Figure 8(c)), the signal was mainly detected in the perinuclear region of target cells. No differences were found when using targeted HUVECs or fibroblasts (data not shown).

Discussion
As a continuation of the previous study [12] and in search for the molecules contributing to functional activity of secretome, we herein focused on isolation and characterization of EVs from secretome of human fetal dermal cells. We used equal numbers of cultured cells and collected equal volumes of secretome at the same time, in the attempt to standardize the EV source. We followed the updated guidelines of the International Society for Extracellular Vesicles (ISEV), which  11 Stem Cells International recommend to use the generic term "EVs" for particles naturally secreted by cells, whose characterization is mainly based on physical parameters, such as size and concentrations [24]. The NTA of pellet particles obtained by ultracentrifugation of secretome of both fetal and adult dermal cells revealed a size diameter of approximately 70 nm, thus suggesting that we may have isolated "small EVs." In fact, the general characterization suggested by the ISEV discriminates EV subtypes

12
Stem Cells International in small EVs (<200 nm size) and larger EVs [24]. Furthermore, the isolated EVs were positive for Rab5 and Alix, and CD63 while were negative for calnexin [24]. Since there is no a perfect quantification method, we followed the most common, which is based on total protein amounts in our dose-response studies [24]. Depletion of EVs by differential ultracentrifugation almost abrogated the in vitro proangiogenic effect of fetal dermal cell secretome. Since depletion of EVs only slightly reduced the efficiency of cell migration, we suggest that perhaps the amount of chemokines such as SDF-1 alpha and MCP-1 in EV-depleted secretome may be sufficient to ensure a migratory response. On the contrary, a VEGF-A concentration of approximately 5000 pg/ml in EV-depleted secretome could not be sufficient to ensure formation of mesh-like structures normally requiring higher doses of VEGF-A (e.g., 20000 pg/ml or more) [25]. Nevertheless, the multitude of factors contained in secretome is extremely difficult to establish which factor might be responsible for one activity or another.
According to the current version of the database Exocarta [26,27], 9769 proteins, 1116 lipids, 3408 mRNAs, and 2838 miRNAs have been identified in EV/exosomes from several cell types and organisms. Research studies have often hypothesized that the transfer of miRNAs in particular will account for the understood EV-mediated effects [14]. miRNA-based therapy entered in clinical studies mainly for cancer treatments, while it is still in early stages for  Plotted values (mean ± SD) represent EV-depleted CM (n = 10) compared to whole CM (n = 10). Differences not denoted with an asterisk are not significant. EV: extracellular vesicle; CM: conditioned medium.
13 Stem Cells International applications of regenerative medicine [28,29]. However, emerging reports are available indicating the vast potential of miRNAs for the repair of several tissues including bone/cartilage muscle, cardiovascular tissue, neurological tissue, skin, and even in angiogenesis [29,30]. With respect to organ repair and angiogenesis, the involvement of miRNAs in different phases of wound healing has been documented [31,32], and several proangiogenic miRNAs have been identified and validated [33,34] .
Both freshly isolated and frozen EVs were tested in cellbased assays of angiogenesis and cell migration, thus showing no differences in their performances. This observation was in agreement with a previous report showing that the storage in the absence of cryoprotectant at -20°C did not affect the biochemical activity of EVs [72]. While EVs from both cell types stimulated cell migration with a similar efficacy, fetal dermal cell-derived EVs were far more effective than adult dermal cell-derived EVs in inducing formation of mesh-like structures. Even if this result was not statistically significant, the slide in Supplemental Material S4 clearly shows this difference. Overall, we observed a delay in EV-induced cellular responses compared to secretome/CM-induced cellular responses. Since the delay was independent from the used EV concentrations (e.g., either 100 or 10 μg/ml), we suggest that the delay could be due to the dynamic of EV internalization by target cells.
In view of the promising prospective of miRNA-based therapeutics, the research is dedicated in solving some challenges in order to make it more translationally valuable. These challenges include the off-target effects of miRNA or their low internalization by target cells making their delivery difficult [73]. Typically, miRNAs for therapeutics have been delivered by direct injection, viral vectors, or coupled to scaffolds [74]. A number of studies indicate that EVs may exert their effect via horizontal transfer of their cargo [14]. Therefore, EVs could serve as a vehicle to successfully deliver miRNAs of interest to several therapeutic applications [75]. Confocal microscope observations of a punctuated green fluorescent pattern inside target cells suggest successful internalization of CSFE-labeled EVs. The signal was visualized in the cytosol of both HUVECs and fibroblasts at not earlier than 4 hours of incubation, and in the perinuclear region at later time points (from 8 hours of incubation). To conclude, the proangiogenic features of secretome of human fetal dermal cells appear largely related to the presence of small EVs. Although there is much to be learned in the field of EV research, the unique properties of these particles clearly represent a new therapeutic opportunity for tissue regeneration [15,76], since it could offer a number of advantages over traditional cell transplantation as a cell-free product. As with conventional drugs, EVs can be standardized and tested in terms of dose and biological activity. Furthermore, EVs can be produced in clinical grade, freeze, and easily delivered. Finally, EV-based therapy could overcome the challenge of a successful delivery of miRNA molecules in vivo [77].

Data Availability
The data concerning the NTA of EVs, the heat map graph, the figures and graphs of in vitro angiogenesis, the curves of cell migration, the flow cytometry graphs of CSFElabeled EVs, and the uptake of EVs used to support the findings of this study are included within the article. The miRNA data used to support the findings of this study are included as a table within the articles, but also in supplementary information file S1 The KEGG data used to support the findings of this study are included within the article (as graphs), and also in supplementary information files S2 and S3.

Conflicts of Interest
The authors declare no conflicts of interest.
Supplementary Materials S1 (excel file): list of significantly upregulated miRNAs in fetal vs. adult EVs; a list of all analyzed miRNAs is also included. S2 (excel file): complete list of the KEGG signaling pathways; list of the KEGG signaling pathways associated with angiogenesis and wound healing. S3 (excel file): complete list of the KEGG signaling pathways obtained with genes intersection 29, including putative target genes. S4 (power point slide): pictures of HUVECs treated with fetalderived EVs (n = 4) and adult-derived EVs (n = 3). Each picture represents an independent experiment. (Supplementary  Materials)