The MSCs were isolated from adult male mice inguinal adipose tissue, washed in sterile PBS (2×), minced in small pieces, and submitted to enzymatic digestion by 37°C with collagenase IA 0.1% (Sigma, USA). In sequence, these cells were filtered in cell strainer (100 µm, BD, Beckton Dickson, USA), rewashed (2×) using complete medium, and cultivated at 37°C in a humidified atmosphere with 5% CO₂ in D-MEM low glucose culture medium (45 mM NaHCO₃, 10% FBS, 100 U/ml penicillin, 100 U/ml streptomycin) (n = 5). The differentiation into mesenchymal lineages in vitro (i.e., adipocytes, osteoblasts, and chondrocytes) was performed using six-well plates (TPP, USA) during 15–20 days with adipogenesis, chondrogenesis, and osteogenesis Mesenchymal Stem Cell Kit (Millipore, USA), according to the manufacturer’s specifications (n = 5). For MV isolation, the 24h MSC serum-free supernatant was submitted to pre-centrifugation at 2,000 g for 5 min and after to ultracentrifugation at 100,000 g for 2 h. Then, the pellet was suspended in sterile PBS with 0.1% of BSA and kept at −80°C. Furthermore, MVs were evaluated according to their size properties using a NanoSight instrument (Malvern, UK). The content of small RNAs or miRNAs inside of MVs was verified with a capillary electrophoresis method using the Agilent small RNA kit, strictly following the manufacturer’s instructions (Agilent Technologies, Santa Clara, CA, USA) (n = 5). Two additional groups consisting of MVs treated with RNAse A (100 µg/mL, Thermo Scientific, USA) and ultracentrifuged medium without cells were carried out concomitantly, and all samples were read with an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) (n = 5). The immunophenotyping of MSCs and MVs was performed using specific sets of antibodies (i.e., CD44, CD90, CD105, CD73, CD45, CD34, CD11c, CD106, CD31, CD9, and CD69, BD Bioscience, USA), according to the manufacturer’s recommendations (n = 5). Briefly, cells were washed (PBS), centrifugated (2,000 g at 5 min), and incubated (45 min in dark room) with specific antibody at 1:100 dilution and rewashed afterward in FACs buffer (PBS 2% FBS). A FACSCanto II flow cytometer (BD, Beckton Dickson) was used for cell acquisition, and the FlowJo software was used for data analysis.

For in vitro assays, approximately 2 × 10⁵ of renal epithelial tubular cells (MM55.K, ATCC^® CRL-6436TM) were seeded in six-well plates (TPP, USA) and treated with nephrotoxic drug cisplatin (8 µg/mL) for 48 h, and two additional treated groups with cisplatin were co-cultured in contact with MSCs (v/v 1:1, 1 × 10⁵) or MVs (50 µg/ml, sequentially each 6 h) for 48 h (Figure S4C in Supplementary Material). Subsequently, cells were trypsinized and subjected to analysis for apoptosis, cell proliferation, and oxidative stress analysis using the respective kits: Alexa Fluor^® 488 annexin V/Dead Cell Apoptosis kit, CellTrace™ Violet Cell Proliferation kit, and MitoSOX™ Red Mitochondrial Superoxide Indicator kit (Life Technologies, USA), following the manufacturer’s recommendations (n = 6). For apoptosis analysis, a range of cisplatin was utilized as death curve (1–10 µg/mL) and newly a dose of 8 µg/mL was established for additional assays (Figure S4D in Supplementary Material). Two additional groups without co-culture were processed with tubular cells treated and untreated with cisplatin. During the flow cytometry analysis to distinguish the tubular cells population (MM55.K cells) from MSCs, the MSCs population were previously labeled with the fluorescent dye CellTrace™ Violet (Cell Trace Violet Proliferation Kit, Life Technologies, USA) and excluded from analysis using negative gate strategy in flow cytometry workflow (Figure S4E in Supplementary Material). For visualizing the MVs incorporation process, the MVs were prelabeled with PKH26 red-fluorescent dye (Sigma, USA) and co-cultured in chamber slide system with tubular cells (1 × 10⁴/well) treated with cisplatin (8 µg/mL) in a time-course assay per 24 h. Images were recorded at each time point to observe MVs incorporation into tubular cells (NIS elements microscope Nikon).

For all animal experiments, we used adult C57BL/6 mice, aged 8–12 weeks with 20–30 g weight (n = 8 per group). The animals were housed in polypropylene boxes at controlled room temperature and light conditions (22°C and 12-h light/12-h dark). They received a solid diet and filtered water ad libitum. All animal handling protocols were approved by the local Animal Care committee CEP/UNIFESP No. 1058/10. Animals were separated into three study groups according to the experimental procedures: mice treated with saline solution, mice treated with cisplatin, and mice treated with cisplatin but also transplanted with MSCs or MVs.

To develop the in vivo experimental model of toxic AKI, an intraperitoneal infusion with 15 mg/kg of cisplatin (Pfizer, USA) was performed (n = 8 per group). The control group was administered with 300 µL phosphate buffer (PBS). After 24 h of cisplatin treatment (day + 1), 1 × 10⁶ of MSCs or 300 µL of saline solution (Sham group) were infused intraperitoneally in the animals. Conforming pre-established by our lab (data not showed) and in classical works in the literature (6, 10), around 100 µg of MVs suspension was infused intravenously on day +1 and day +3 after cisplatin infusion (Figure S4B in Supplementary Material). To evaluate the action mechanism of MVs during in vivo infusion, three additional groups were set in parallel: (i) MVs pretreated with Proteinase K 20 µg/mL (Sigma, USA) and DNase I (10 U/µL; Stratagene, USA); (ii) MVs pretreated with a cocktail of proteases [Proteinase K 20 µg/mL (Sigma, USA), trypsin (Life Technologies, USA) and collagenase 1 A (Sigma, USA)]; and (iii) MVs pretreated with Proteinase K 20 µg/mL (Sigma, USA) and RNase A (100 mg/mL; Thermo Scientific, USA), according to the manufacturer’s instructions. At day 4 after cisplatin administration, with or without MSC or MV infusion, all groups were euthanized using an overdose of anesthetic [xylazine (30 mg/kg) + ketamine (200 mg/kg)], and urine, blood, and renal tissues were collected and maintained at −80°C until analyses.

The creatinine, urea, and lactate analysis were performed to measure the status of renal function (n = 8). Creatinine was measured using the Jaffe modified method (Labtest Diagnostic, Brazil). Urea concentration was analyzed with a urea kit (Labtest Diagnostic, Brazil), according to the manufacturer’s specifications. The percentage of lactate was measured using Advia 1650 equipment (Bayer, Germany), following the datasheet instructions. All animals were periodically (each 24 h during 4 days) monitored for gain or loss of body mass (n = 8). Finally, an injury score evaluation was determined according to the absolute mean of clinical parameters (Table S3 in Supplementary Material).

Kidney fragments were fixed in buffered formaldehyde 4% (pH = 7.0) and embedded in paraffin block. For each group, histological sections of 4–5 µm were performed and the slides were stained with hematoxylin/eosin (H&E). For a quantitative evaluation of renal injury, clinical pathologic score of tissue damage was performed. Twenty-five fields from five slides were selected (in magnification 400×), and for all groups, the percentage of necrosis, expansion of glomerular cavity, tubular dilation, detachment of epithelial cells, formation of renal casts, and inflammatory infiltrates were evaluated. All these parameters were determined here as tissue injury index. A pathological score was assigned for each parameter, according to injury severity (0 = absent, 1 = mild, 2 = moderate, and 3 = severe). Finally, an overall percentage score (e.g., arithmetic mean of all absolute parameters) was generated following a direct comparison between the experimental groups.

Tissue sections of 3–4 µm were processed on silanized slides (n = 5). The immunohistochemistry was performed using protein block reagent, an EnVision+/HRP-Dual Link kit, DAB (3-3) tetrahydrochloride and diaminobenxidine solution (Dako, Denmark), and caspase 3 antibodies (1:1,000, Cell Signaling, USA), following the manufacturer’s specifications. Slides were hydrated and dehydrated through alcohol–xylene gradient (30–100%), washed in buffer, and stained and mounted in Permount solution (Sigma, USA). To check the percentage of extracellular matrix deposition, picrosirius staining (Sirius red in picric acid 0.1%) was performed. For immunofluorescence evaluation, kidney tissue fragments were frozen in Tissue Tek (Sakura, Japan) and 4-µm cryosections were processed. Cells were fixed, permeabilized, and incubated with primary antibodies (1:200), i.e., anti-Ki-67 (Abcam, USA), followed by secondary antibodies (1:400) conjugated to FITC (Abcam, USA). Cells were also counter-stained with DAPI for nuclei visualization (Vector Labs, USA). The fluorescent labeling was observed using fluorescence microscopy (Nikon, Japan). For the detection of in situ necrosis/apoptosis, the TUNEL assay was used with the in situ Cell Death Detection Kit (Roche, Germany), following the general information contained in the manufacturer’s protocol. The following in situ assays such as (i) renal fibrosis by picrosirius method, (ii) cell proliferation (immunohistochemistry to Ki-67), and (iii) cell death (immunohistochemistry to Caspase-3) were performed using the quantification of at least 20–30 fields (magnification 200×) with software developed by Nikon (NIS elements microscope Nikon).

The in vivo tissue oxidative stress was evaluated indirectly by nitric oxide (NO) measurement using the Griess modified method (n = 6). In addition, the renal content of reduced (GSH) and oxidized (GSSG) glutathione levels were assessed using a glutathione fluorescent detection kit (BioVision, USA), following the manufacturer’s instructions (n = 6). The animal serum cytokine profile was performed using a flow cytometry beads array systems with Cytometric Bead Array Kit for inflammation (BD Bioscience, USA), according to the manufacturer’s recommendations (n = 6).

Total RNA, including small RNAs (miRNAs), was extracted from all kidney tissues with a miRNeasy Mini Kit (50) (Qiagen, South Korea), according to the manufacturer’s instructions (n = 6 per group). The concentration, quality, and integrity of RNA samples were, respectively, checked using a Nanodrop spectrophotometer (Thermo Scientific, USA) and Bioanalyzer equipment (Agilent Technologies, USA) (n = 6 per group). Furthermore, the reverse transcription of total RNAs and miRNAs was performed using, respectively, the High Capacity kit (Life Technologies, USA) and the RT² miRNA First Strand Kit (12) (Qiagen, South Korea) (n = 6 per group). Gene expression levels between the groups of samples were measured by real-time PCR (RT-qPCR) using the TaqMan system (Life Technologies, USA) (n = 6). To translate miRNAs into cDNAs was used the miScript II RT Kit, following all recommendations in datasheet (Qiagen, South Korea). For quantitative detection of pre- and mature miRNAs forms, the Syber Green PCR kit was utilized (Qiagen, South Korea). At gene expression assays normalization, the endogenous HPRT gene was used as a housekeeping gene, whereas that for miRNA reactions, the endogenous control SNOR73A was used. Finally, the reaction was carried out using the ABI Prism 7300 sequence detection system (Life Technologies, USA). The results were analyzed through relative quantification method according to formula: 10,000/2^ΔCt, using SDS software (Life Technologies, USA). The global miRNA PCR arrays were performed in 96-well plates per each set (3 set per group), following the recommendations specified in the products catalogs: RT² SYBR Green ROX qPCR Master Mix (24) and Mouse miRNome RT²miRNA PCR array (528 miRNAs; Qiagen, South Korea). Data analysis and normalization were performed using the web-based application provided on the manufacturer’s website (Qiagen, South Korea).

Differentially expressed miRNAs were selected between the groups using the QIAGEN software website analysis. Only the miRNAs that had at least fivefold changes in expression were included in further analysis. Furthermore, the MIR@NT@N resource was used to identify mRNAs that were targeted by the selected differentially expressed miRNAs (18). Then, we searched for overlaps in the miRNA sets using GeneVenn, a web application for comparing set lists using Venn diagrams (19). The known and predicted relationships between miRNAs and target genes were obtained from the DIANA Tools and MIR@NT@N resource, according to the criteria in the default parameters (18–21). We only selected those genes targeted by at least two miRNAs to restrict our analysis to the most relevant pairs of connected miRNAs and mRNAs. Pathways significantly enriched with target genes were identified with the DIANA Tools. We also generated miRNA–mRNA networks to visualize the relations between miRNA and their predicted target genes. The gene ontology (GO) analysis were generated using a web-based enrichment software analysis, FunNet, and the networks were built using the Cytoscape software, which allows network editing and visualization of the molecular interactions (22).

To validate our findings, cultures of renal tubular cells (2 × 10⁵ cells per well) were seeded in six-well plates to further addition of RNA oligos, in order to inhibit or to overexpress specific miRNAs (i.e., miR-21, miR-377, miR-880, and miR-141). The cells were submitted to a transfection process using the N-TER kit (Sigma, USA), conforming to the manufacturer’s instructions (Qiagen, South Korea) (n = 6). According at its respective oligos, the cells were transfected with miRNAs mimics, inhibitors or scramble oligos. Additionally, a control group was run in parallel with transfection reagent alone. Subsequently, all groups were submitted to cisplatin for 48 h, and MV treatment was performed in an additional group conforming previously described.

For each sample, 50 µg of renal tissue protein extract was used for electrophoresis on 10% polyacrylamide gels (SDS-PAGE). The immunostaining was carried out with primary antibodies (β-actin/1:1,000, Sigma, USA; IKK-α/1:1,000, Cell Signaling, USA; Argonaute 2/1:1,000, Cell Signaling, USA; Drosha/1:1,000, Cell Signaling, USA; Dicer/1:1,000, Imgenex, India), followed by conjugated secondary antibodies (anti-mouse or anti-rabbit peroxidase/1:5,000, Sigma, USA). Then, the membrane was revealed by chemiluminescence methods using the ECL kit (Millipore, USA), and the images were acquired on GEN-BOX equipment (Syngene, UK) (n = 5). The GeneSnap software and GeneTools (Syngene, UK) were used to identify, analyze, and quantify the gel bands.

Data were collected and presented according to classical descriptive statistics. The data sets were tested for a Gaussian distribution by a Kolmogorov–Smirnov test with Dallal–Wilkison–Lillie for a p value, and further statistical inference parametric methods were performed for all analyses. Student’s t and one-way ANOVA tests (followed by Tukey post-test) were used to determine significant differences, respectively, between two or more independent sample groups. All experimental data obtained in this study are represented by the mean and SD of the mean (±SD), and statistical analyses were conducted considering a minimal statistical significance (α) at 5% (p < 0.05).

First, we performed an extensive characterization of MSCs and their MVs. MSCs in vitro showed the classical features of mesenchymal cells such as fibroblast colony-forming unit formation, fibroblast-like morphology, differentiation into mesodermal lineages (i.e., adipocytes, chondrocytes, and osteoblasts), high proliferation index, and a distinctive immunophenotype with classical MSCs features (Figure S1 in Supplementary Material). On the other hand, MVs secreted by MSCs displayed spheroid morphology, had small size (≈125 nm), exhibited positivity to MSC (i.e., CD105, CD90, and CD44), exosomes (i.e., CD9 and CD63), and surface markers, and internally had the presence of miRNAs (<22 nt) and their biogenesis machinery (Dicer, Drosha, and Argonaut 2) (Figure S2 in Supplementary Material). Subsequently, we assessed the repair potential of MSCs and MSC-MV treatment in a toxic model of AKI. Interestingly, we demonstrated in vivo after cisplatin administration that injections of both MSCs and MVs ameliorated physiologic injury parameters such as weight loss, injury scores, and renal damage markers (serum creatinine and urea levels; Figures 1A–F; Movies S1 and S2 in Supplementary Material). Moreover, to understand which molecular pattern inside MVs was promoting tissue repair, we performed a pretreatment of MVs with RNAses and observed the abrogation of the therapeutic effect. By contrast, DNAses and proteases pretreatments did not show any effect (Figures 1E,F). Furthermore, we identified in situ in histological renal sections treated with MSCs and MVs an improvement in tissue regeneration status as observed by increase in the proliferation rate (Ki-67 staining) and reduction of the injury score, detected mainly by reduction of apoptosis index (active caspase 3 and TUNEL assays) and interstitial extracellular matrix deposition (Sirius red staining) (Figure S3 in Supplementary Material).

In an attempt to investigate the beneficial therapeutic effect of MSCs, we analyzed in vivo the tissue expression of molecules that are involved in the modulation of renal toxic injury. With this analysis, we detected in kidney tissues after MSC and MV therapy a reduction of the injury markers (e.g., lactate index and Kim-1 levels) and apoptosis index (i.e., Bcl-2/Bax ratio) when compared to damage group alone (cisplatin) (Figures 2A–C). Moreover, the MSC and MV infusion decreased the renal oxidative stress by enhancing anti-oxidant enzymes levels (e.g., GSH/GSSG ratio and HO-1) and reducing free radicals presence (i.e., NO; Figures 2D–F). Concomitantly, we observed in kidney tissues treated with both MSCs and MVs an increase in the expression of classical cytoprotective genes such as hepatocyte growth factor, vascular endothelial growth factor, and IGF (Figures 2G–I). In addition, mice transplanted with MSCs or MVs showed a decrease in the serum inflammatory cytokine levels (i.e., interleukin 2, interleukin 17, TNF-α, and INF-γ) and an increase in anti-inflammatory molecules such as interleukin 10 and interleukin 4 (Figures 2J–O). Finally, the NF-κB signaling pathway, which is associated with acute inflammation, was also downregulated in renal tissues after MSCs and MVs intervention, as demonstrated by the reduction of IKK-α protein expression (Figures 2P,Q).

To support our in vivo findings, we evaluated the cross-talking between MSCs and/or MVs with tubular epithelial cells. Using in vitro co-culture assays, we observed after cisplatin stimulus that both treatments (MSC and MVs) reduced cell death frequency (i.e., apoptosis/necrosis ratio) and oxidative stress index (i.e., anion superoxide level) while promoting an increase in the cellular proliferative rates, thereby indicating that MSCs and MVs can restore the tubular epithelial cell physiology after toxic injury (Figure 3).

In search for precise molecular mechanisms involved in the MSC-mediated cytoprotection, we analyzed the global miRNA profile changes in the renal tissue submitted to toxic injury alone or treated with MSCs (Figure 4A). We detected a total of 528 miRNAs expressed in common between all the tested conditions (Figure 4A; Figure S1 in Supplementary Material), in which 39 were upregulated and 37 were downregulated in the cisplatin condition when compared with saline at a fold change ≥3 (Figures 4B,D). After MSC therapy, 50 miRNAs were upregulated and 11 miRNAs were downregulated in the cisplatin + MSCs condition when compared with cisplatin alone, considering a fold change ≥3 (Figure 4B). Further, we observed that the proteins related to miRNA biogenesis (e.g., Dicer, Drosha, and Argonaut 2) were all modulated during toxic injury, and after MSC administration, they had their physiologic levels re-established (Figure 4C). These results were complemented by miRNA amount analysis, which showed higher miRNA levels during injury process and lower levels after MSC treatment (Figure S4A in Supplementary Material). Additionally, we observed that the differential changes in the miRNA signatures were associated with a broad modulation of specific miRNAs associated with cisplatin or MSC intervention (Figures 4C,D). Therefore, we decided to narrow our downstream analysis to the highly modulated miRNAs (fold change ≥5) at each comparison. In the cisplatin condition compared to saline, we found that miR-377, miR-741, miR-675-3p, miR-21, miR-463, and miR-293* were the most upregulated miRNAs, whereas miR-141, miR208a, miR-92a*, miR-292-3p, miR-673-5p, and miR-1190 were the most downregulated miRNAs (Figure 4D). When we evaluated the cisplatin + MSCs condition in comparison with cisplatin alone, we detected the miR-141, miR-880, miR-433, miR-92a, miR-463, and miR-295 as most upregulated miRNAs, while the miR-30e, miR-377, miR543, miR-693-5p, miR-201, and miR148a* were the most downregulated ones (Figure 4D). Subsequently, we identified the protein-coding genes that were predicted to be targets of the top differentially expressed miRNAs at each comparison and looked for biological pathways that were significantly enriched with those genes. Thus, with this approach, we found signaling pathways related to tissue repair and homeostasis recovery, such as regulation of actin, axon guidance, ErbB, P13K-Akt, Wnt, MAPK, insulin, protein processing, and ubiquitin/proteolysis (Figure 4E). A brief description of all pathways associated with the differentially expressed miRNAs at each comparison and their predicted target genes is illustrated in Table S2 in Supplementary Material. In addition, the GO analysis using the mRNAs targeted by miRNAs showed that the majority of the highly regulated miRNAs in renal tissues are linked to positive regulation of transcription DNA-dependent biological processes, thereby highlighting a presence of intrinsic regulatory mechanism (Figure 4F). In attempt to verify if the miRNAs differentially expressed in renal tissues were directly transferred by MVs, we carried out a miRNAs profile of MVs and MSCs and detected a higher number of miRNAs upregulated in MVs than MSCs (Figure S5A in Supplementary Material). Moreover, we identified that the top modulated miRNAs were not directly associated to kidney miRNA profile and the miR-377 and miR-141 (mostly regulated in renal tissues), although present inside of MVs, they were not found to be enriched into MVs (Figures S5B,C in Supplementary Material). These findings suggest that MVs and MSCs did not share the same miRNA profile and its effect in the renal tissue can be conducted by an indirect regulatory process.

In order to identify some biological functions attributed to MSC renoprotection, we carried out an in silico analysis with the most upregulated and downregulated miRNAs in the cisplatin condition compared to saline and connected these miRNAs with previously described differentially expressed genes which were associated to cisplatin damage in renal tissues (Figure S6A in Supplementary Material) (23, 24). After cisplatin intervention, we observed that the upregulated and downregulated miRNAs (red/green triangles) in renal tissues were strongly integrated in the network, considering that the top miRNAs in terms of fold change (yellow border triangles) showed interactions with key molecules altered in kidney injury (blue and orange rectangles) (Figure S6A in Supplementary Material). Moreover, using a functional and enrichment approach, we verified in this miRNA–mRNA network that the top miRNAs in fold change (yellow border triangles) also interacted with specific genes involved in TGF-β/Wnt (purple border rectangles) and fibrosis/epithelial–mesenchymal transition (EMT) (orange rectangles) signaling pathways (Figure S6A in Supplementary Material). Furthermore, to heighten our inference analysis, we compared in a Venn diagram all the miRNA profiles and selected the 10 miRNAs (overlapping*) that were exclusively related to MSC cytoprotection (Figure S6B in Supplementary Material). Interestingly, we also identified in this overlap two miRNAs (i.e., miR-141 and miR-377) that were inversely regulated between the “injury profile” (Cis vs. Saline) vs. the “treatment profile” (Cis + MSC vs. Cis) (Figure 4D; Figure S6B in Supplementary Material). After, using these 10 pre-selected miRNAs, we constructed a second miRNA–mRNA network (“regulatory network”) and identified their corresponding predicted target genes (Figure 5A). Then, we selected only the genes that were targeted by at least two miRNAs in order to visualize the most relevant miRNA–mRNA interactions (Figure 5A). In this second regulatory network, the most connected miRNAs were over again the miR-141 and miR-377 (yellow border triangles) (Figure 5A). Subsequently, after enrichment analysis, we extracted semantic relationships between these miRNAs and again verified connection with genes related to the TGF-β/Wnt (purple border rectangles) and fibrosis/EMT (orange rectangles) signaling pathways (Figure 5A). Furthermore, in search of specific functions associated to these most connected miRNAs (e.g., miR-141 and miR-377), we built a second Venn diagram with additional putative target genes predicted by three distinct databases, i.e., Targetscan, miRanda, and miRDB (Figure 5B). Finally, the results from this second Venn diagram revealed three potential target genes for miR-141 (i.e., Yaf2, Ulk2, and Ccne2) and eight potential target molecules for miR-377 (i.e., Ncoa6, Bend6, Nts, Cul1, Pitx2, Zfp36l1, Ssfa2, and Rsbn1) (Figure 5B). Thus, based on its biologic importance and considering the renal context of injury and repair, only the most representative target genes (i.e., Ulk2 and Cul1) were selected to validation step analysis.

To validate the participation of miR-377 and miR-141 in tissue injury and repair, we evaluated the dynamic expression of the precursor and mature forms of these miRNAs. Cisplatin treatment in vivo increased both mature and pre-miR-377 levels, proportionally to the damage severity (Figure 6A). Conversely, the renal tissue expression of both mature and pre-miR-141 species was higher when the injury was absent and lower when the damage process was already established (Figure 6B). Conforming observed at MSCs treatment in our global miRNA profile (Cis + MSCs vs. Cis), the treatment with MVs also reduced abruptly the miR-377 expression and increased the levels of miR-141 in renal tissues submitted to toxic injury (Figures 6A,B). Interestingly, the detection of the pre-miR-377 in early stages of the renal tissue injury suggests that its expression could be used as a potential marker of AKI (Figure 6A). Concordantly, we observed similar results using tubular cells in vitro, which showed elevated expression of mature miR-377 during toxic injury progression, considering that the levels of mature miR-141 decreased at the same time points (Figures 6C,D). Surprisingly, we also observed in vitro that treatment with MVs was able to decrease miR-377 expression and increase the miR-141 levels when compared to cisplatin-treated cells alone (Figures 6C,D). According to the damage severity, we also observed in vitro a premature expression of pre-miR-377 than its mature form (miR-377), whereas the pre-miR-141 expression remained unaltered (Figures 6C,D). For elucidating the incorporation process of MVs by tubular epithelial cells, we added pre-stained MVs suspensions (MVs labeled with fluorescent red dye) to injured tubular cells cultures (cultured with cisplatin) and analyzed its progressive incorporation per 24 h in a time-lapse assay. We observed that MVs, although in aggregated status, are internalized by epithelial tubular cells (dim red dots and black arrows) with predominant localization in the cytoplasm (Figure 6E; Movie S3 in Supplementary Material).

To investigate the role of miRNA–mRNA interactions on kidney injury progression and tissue recover after MSC treatment, we conducted an in vitro transfection assay in tubular epithelial cells using specific oligos that work as mimics and inhibitors of some of the most relevant miRNAs found in our miRNA profile analysis (i.e., miR-880, miR-141 miR-377, and miR-21). First, we validated our transfection assays by demonstrating that all inhibitors and mimic oligos efficiently overexpressed or silenced their correspondent target molecules (Figure S7 in Supplementary Material). Then, we observed in tubular cells submitted to cisplatin treatment a protection against cell death (reduction in apoptosis/active caspase-3 and necrosis/7-AAD expression levels) only in cells incubated with inhibitors of miR-21 and miR-377 (Figure 7A). Complementary, the intracellular level of anion superoxide (O2−, a stress oxidative radical) decreased only in cells treated with inhibitors of miR-21 and miR-377 (Figure 7B). By contrast, the use of miR-880 and miR-141 mimics, which were elevated in tissues treated with MSCs, did not show any direct protective effect on tubular cells submitted to toxic injury (Figures 7A,B). In addition, the in vitro inhibition of miR-377 in tubular epithelial cells promoted an increase in the expression of cytoprotective genes (i.e., BCL-2 and HO-1; Figures S8A,B in Supplementary Material) and a reduction in damage-related molecules (i.e., Kim-1 and iNOS; Figures S8C,D in Supplementary Material). The absence of any significant effects in the control (cells without cisplatin stimulus), the negative control (cells only exposed to transfection reagents), and the scrambled oligos groups (miRNA control, cells treated with not associated miRNAs) corroborated the relevance of our results (Figures 7A,B; Figures S7A–D and S8A–D in Supplementary Material). In sequence, we selected potential target genes for miR-141 (i.e., Ulk2) and miR-377 (i.e., Cul1) and confirmed in vitro and in vivo their inverse relationships. Specifically, we identified in vitro that Ulk2 expression increased under injury conditions but after MVs treatment or miR-141 overexpression had its levels decreased when compared to cisplatin-treated cells alone (Figure 7C). These results were further confirmed in vivo by demonstrating that the Ulk2 levels were elevated in tissues damaged by cisplatin and decreased after MSC or MV treatment (Figure 7C). In parallel, we also showed in vitro that Cul1 expression was higher in normal conditions and decreased after cell injury (Figure 7D). Conversely, the in vitro Cul1 levels were enhanced after MV therapy or when miR-377 was silenced in comparison with cisplatin group alone (Figure 7D). Finally, in vivo measurements showed that Cul1 levels were lower during AKI but increased in renal tissues after MSC or MV administration (Figure 7D).