Extracellular vesicles derived from patients with antibody‐mediated rejection induce tubular senescence and endothelial to mesenchymal transition in renal cells

2.1. Patient recruitment

Patients were recruited from Department of Translational Medicine, University of Eastern Piedmont, Novara and Department of Emergency and Organ Transplantation, University of Bari, Italy. All patients presented written informed consent for the use of this material for research purposes.

We enrolled 39 kidney transplant recipients who were transplanted between January 2009 and August 2016 in the two centers. Among these patients, 14 developed biopsy‐proven acute AMR (AAMR), 14 presented biopsy‐proven chronic AMR (CAMR) and 11 patients presented a stable graft function (TX Ctrl). In addition, as additional control, healthy volunteers collected‐EV were also used (n = 5).

AMR was diagnosed based on the Banff 2013 classification. One‐third of AMR patients showed positive glomerular and peritubular C4d staining. Median age of recipient was 48.8 years (range 26–60) for AAMR, 49 years (24–61) for CAMR and 61 years (36–75) for TX Ctrl group. AMR developed at median 12.5 months posttransplant, whereas CAMR developed at a median of 7 years after transplantation. Maintenance immunosuppression regimen encompassed calcineurin inhibitors (tacrolimus 95%, cyclosporine A 5%), mycophenolate mofetil as antiproliferative agent and corticosteroids (Table 1).

2.2. Characterization of EV

Plasma purified EV were characterized as previously described, ¹⁴ , ¹⁵ according to the criteria suggested by ISEV 2018 guideline. ⁴ For further details, see Supporting Information.

2.3. EV stimulation of renal tubular epithelial and endothelial cells

RPTEC (human renal proximal tubular epithelial cells) between 3 and 5 passages and human umbilical vein endothelial cells lines (HUVEC) (ATCC‐LGC Standards, Italy) were grown in RGEM medium (Lonza) and EndGro (Merck Millipore). Cells were exposed to AMR‐derived EV and TX Ctrl, healthy control EV at concentration of 5e⁺⁴ EV/ target cells for 24 and 48 h.

2.4. RNA extraction and qPCR analysis

Total RNA was isolated with RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions and quantified by NanoDrop ND‐1000 Spectrophotometer (Thermo Fisher Scientific). Retrotranscription was performed using the iScript cDNA Synthesis Kit (Bio‐Rad). qPCR was performed with SsoAdvanced™ Universal SYBR® Green Supermix (Biorad) and the Light Cycler@96 (Roche). The relative amounts of mRNA were normalized to β‐actin mRNA as the housekeeping gene. Data were analyzed using the ΔΔCt method. ¹⁶ Primer sequences for these genes are given in Table S1.

2.5. FACS analysis

The immunophenotype of endothelial cells was analyzed by FACS analysis as already described. ¹⁷ See Supporting Information.

2.6. Immunohistochemistry and confocal laser scanning microscopy

Renal formalin fixed‐paraffin embedded biopsies of the same patients enrolled for EV isolation were processed as previously described ¹⁷ , ¹⁸ , ¹⁹ and are indicated in Supporting Information.

2.7. RNA extraction from EV and miRNA analysis

RNA extraction was performed from EV by miRNeasy mini kit (Qiagen, cat no. 217004). The EV′ total RNA quantification was assessed using NanoDrop 1000 and the Qubit 2.0. Fluorometer.

RNA samples were processed using the Agilent’s miRNA complete labeling and hybridization kit and hybridized on SurePrint human miRNA microarrays (protocol v.3.1.1 [G4170‐90011]).

The Agilent microarray data are Minimum Information About a Microarray Experiment (MIAME) compliant, and raw data are accessible through Experiment Array Express accession number E‐MTAB‐10447.

2.8. Microarray data analysis

Microarray data analysis was performed using Agilent Feature Extraction Software and Genespring software. The raw expression signals were log‐transformed, normalized, and filtered for microarray analysis according to the quantile corrected signal of all miRNAs with an intensity of 100, resulting in the selection of 327 miRNAs from the original collection of 1205 miRNAs. Using a moderate t‐test, miRNAs showing differential expression between various comparison groups were detected. Probe sets were sorted according to a significant p‐value and adjusted to account for multiple tests using the Benjamini‐Hochberg FDR method. We applied a filter with FDR of 0.05 and a fold change <2 to determine the features that were differentially expressed. PCA was carried out using tools from Genespring (Agilent Technologies).

2.9. miRNA target analysis

We used computational algorithms to identify the validated and predicted interactions between miRNA dataset and mRNA targets. The algorithms used were: TargetScan 5.2 (http://www.targetscan.org/), miRTarBase47 and DIANA TarBase v7.04 and miRBase 17.0 (http://microrna.sanger.ac.uk).

2.10. Statistical analysis

All results were analyzed using GraphPad Prism 5.0 (GraphPad software, Inc.) with the integrated tests: nonparametric Kruskal–Wallis test (or Mann‐Whitney U test) and the Dunn’s multiple comparison post hoc test. Significance was considered at p < .05. All in vitro results represented the means of at least three independent experiments.

3.1. Platelet and endothelial cells are the main source of plasma EV in patients with AMR

We first investigated the size and the concentration of EV with nanoparticle tracking analysis (NTA) by Nanosight. By Kruskal–Wallis test, we found a significant difference in the number of EV among the four groups analyzed. In particular, we found an increased number of EV in AAMR group compared with CAMR by multiple comparison test (p = .02, Dunn's post hoc test) (Figure 1A). Plasma purified EV showed a homogenous pattern of spheroid particles with a size ranging from 60 to 170 nm as seen by Nanosight analysis and then confirmed by transmission electron microscopy (Figure 1B,C).

Next, we evaluated EV phenotype by MACS multiplex bead‐based flow cytometry assay described recently. ²⁰ , ²¹ Data analysis revealed that EV were positive for tetraspanins CD9, CD81, and CD63. Moreover, platelet markers (CD41b, CD42) and endothelial markers such as CD31 were increased in CAMR and AAMR compared with TX Ctrl. Extracellular vesicles adhesion molecules (CD29) was significantly augmented in CAMR. AMR‐EV were negative for epithelial makers as CD326 (epithelial cell adhesion molecule, EpCAM), mesenchymal markers (CD105). Moreover, EV exhibited low detectable expression of leukocyte (CD45) and antigen presenting cells markers (CD40, Figure 1D).

3.2. EV from AMR patients induced cellular senescence in RPTEC culture

To elucidate the role of EV on renal cells, we cultured tubular epithelial cells in the presence of AMR‐derived EV. Compared with untreated cells and to EV‐derived from healthy control, RPTEC exposed to AMR‐derived EV showed reduced proliferation (Figure 2A) without induction of early/late apoptosis or necrosis (Figure S1A). Next, we performed SA‐β‐gal staining in RPTEC to assess the occurrence of cellular senescence, defined as a state of stable growth arrest in the absence of apoptosis. Compared with untreated cells and to healthy control EV (Figure 2B), we found that acute and chronic AMR‐derived EV significantly induced cellular senescence as measured by the increased number of SA‐β‐gal‐positive cells (Figure 2B). Furthermore, we decided to evaluate the mRNA levels of cell cycle arrest markers as p21 together with well‐known anti‐aging factors such as Klotho. ¹⁹ , ²² , ²³ Interestingly, in RPTEC exposed to AMR‐derived EV, we found a statistically significant augment of p21 ^waf1/cip1 (CDKN1A) and IL‐6 transcripts and a reduction in KL mRNA levels (Figure 2C–E). Taken together, these data suggest that AMR‐derived EV can induce the establishment of a senescence state in RPTEC, as observed by a stable cell cycle arrest without apoptosis and the positivity to the common senescent and Senescence‐Associated Secretory Phenotype (SASP) markers. ²⁴ Based on the data obtained in cell culture, we then decided to evaluate whether the tubular senescence and parenchymal SASP occurred also in renal biopsies of the same patients with diagnosis of acute and chronic AMR (Figure 3). For that purpose, we stained for P16ink4a, a cell cycle inhibitor associated with senescence whose increased expression has been linked to the biological renal aging ⁹ and for IL‐6, one the most important SASP marker.

We found that patients with AAMR and CAMR presented increased tissue expression of tubular nuclear senescence marker p16INK4a and IL‐6 compared with TX Ctrl biopsies. Importantly, also glomerular cells appeared positive to p16 staining particularly in CAMR group.

3.3. EV from AMR patients induced endothelial to mesenchymal transition (EndMT)

We then decide to evaluate the effect of AMR‐derived EV on endothelial cells. Firstly, MTT test revealed that the endothelial cell treated with AMR‐derived EV did not show reduced proliferation compared with untreated cells (Figure 4A). By FACS analysis, we found that AMR‐derived EV induced EndMT by decreasing significantly constitutive endothelial markers, such as CD31 and VE‐cadherin. Furthermore, we observed an increased level of fibroblast markers, such as Vimentin expression with the production of Collagen I (Figure 4B). Representative plots are shown in Figure S2. Taken together, these data indicated that AMR‐derived EV promoted EndMT.

We then decided to investigate the possible occurrence of EndMT in renal biopsies of AMR patients (Figure 4C–H). Renal biopsies from TX Ctrl (Figure 4C) showed αSMA expression predominantly in smooth muscle cells (wall of renal arteries) and CD31 staining at level of interstitial peritubular capillaries. In the AAMR and CAMR renal tissues (Figure 4D,E, white arrows), interstitial cells upregulated αSMA together with an intense reduction in CD31 expression, indicating EndMT. The increase in αSMA associated with the decrease of CD31 occurred predominately in peritubular capillaries and arterioles and was more evident in CAMR compared with AAMR. (Figure 4I). Finally, we identified the presence of EndMT also at glomerular level where we found upregulated αSMA and co‐localization with CD31 (Figure 4H, white arrow).

3.4. EV from AMR patients induced complement activation in tubular and endothelial culture

Complement system is a well‐recognized player in AMR ²⁵ , ²⁶ ; for instance, CAMR biopsies exhibited higher C3 gene transcripts. ²⁷ To elucidate the role of circulating EV in modulating complement, we assessed gene transcripts C3 and CFH in RPTEC. After stimulation for 24 h with AMR‐derived EV, we found a significant increase of C3 gene level in RPTEC (Figure 5A,B) and a decrease of CFH gene expression. A similar trend was observed in primary endothelial cell culture (data not showed). We then decided to study the role of EV in functional complement activation. For this purpose, we cultured primary endothelial cells in serum‐free media in the presence of AMR‐derived EV. Cell culture supernatants were collected after 24 h and loaded in a complement enzyme immunoassay for the functional determination of classical, lectin, and alternative pathways activation (Figure 5C). Compared with untreated cells and TX Ctrl, culture media derived from endothelial cells exposed to AAMR and CAMR‐derived EV activated classical and lectin pathways (Figure 5C). Next, we evaluated the deposition of C4d in endothelial cells. Interestingly, immunofluorescence analysis revealed strong C4d deposits in AAMR and CAMR conditions (Figure 5D,E).

3.5. miRNA characterization from AMR‐derived EV

We used microarray analysis to compare the miRNA profiles of EV isolated from AAMR and CAMR comparing with TX Control patients (Figure 6). By microarray miRNA profiling, we detected 42 miRNA downregulated and 34 miRNA upregulated (Table 2) with FDR <0.05 and Fold change >2 (Figure 6A). The principal component analysis showed that these 76 miRNAs were able to distinguish ABMR‐derived EV from TX control EV (Figure 6B). We then searched whether differences in EV‐miRNAs content were present between AAMR and CAMR patients. We found nine miRNAs differentially expressed in the two AMR groups whose eight downregulated and only one upregulated in AAMR compared with CAMR (Table 3 and Figure 6C). Interestingly, these miRNAs were able to distinguish AAMR‐derived EV from CAMR EV by PCA (Figure 6D).

We investigated the possible target genes of identified miRNAs (Tables 2 and 3). miRNA candidates were picked according to their expression, their fold change, the p value, and prediction score. Several targets were associated with CDKN1A and CDKN2A genes regulation as miR‐24‐3p, miR‐let‐7d‐5p, miR‐101‐3p, mir‐345‐5p, miR‐20‐5p, and miR‐132‐3p. Since our group has previously demonstrated that premature senescence and EndMT represent central mechanisms involved in the progression of chronic graft deterioration after renal transplantation, ¹⁸ , ¹⁹ , ²⁸ , ²⁹ we selected miRNA that specifically target senescence‐associated genes as CDKN1A, CDKN2A, together with early fibrosis genes as Collagen I, TGFB, and complement‐related genes. Using different databases (TargetScan, miRDB, miRTarBase, and DIANA TarBase) and filtering for the prediction target score, we identified and validated two miRNA downregulated in AMR‐derived EV versus TX Ctrl: mir‐24‐3p that targets CDKN1A, CDKN1B, TGFB, CDK6, and CFP (Properdin: Position 18–24 of CFP 3' UTR, 7mer‐8m with 98 context score percentile rank); mir‐29a‐3p which modulates COL1A1, COL1A2, COL3A1, and COL4A1 pro‐fibrotic genes (Figure 7A,B). Furthermore, even if not yet validated, we also found the downregulation of miR‐101‐3p which was associated with Klotho (KL), CDKN1A, and C3 gene regulation (Table 2, Fold change 5.34, p = .053).

Four miRNAs were upregulated: miR‐604, miR‐590‐3p, miR‐let‐7d‐5p, and miR‐515‐3p. In particular, miR‐604, miR‐590‐3p were found to interact with canonical conserved 8mer and 7mer sites that fit the seed region of each miRNA at 3'‐UTR of CD59, CFD, CFHR4, and CFH complement components. Whereas miR‐let‐7d‐5p and miR‐515‐3p target CDKN1A, H2AFX, (associated to senescence and DNA damage response) and COL1A1, COL1A2, and TGFBR1/2 genes (associated to fibrosis). We validate their expression by qPCR, confirming the microarray data (Figure 7C–F). Regarding miR‐590‐3p, by microarray data, it was upregulated in AAMR‐EV compared with CAMR EV and displayed the same trend in qPCR analysis. However, no statistical significance was reached (Figure 7F).