MiR‐146a engineered extracellular vesicles derived from mesenchymal stromal cells more potently attenuate ischaemia–reperfusion injury in lung transplantation

An in situ left LTx model was established. To determine whether nebulising hucMSC‐EVs in donor rats could protect the lungs from IRI during LTx, we adopted a treatment protocol as depicted in Figure 1A. The actual operation images of rat nebulisation are provided in Figure S1. MSCs were isolated from the Wharton's jelly of human umbilical cords using the tissue explant adherence method. We confirmed the successful acquisition of huc‐MSCs by examining cell characteristics, including their spindle‐shaped morphology, MSC‐specific surface markers and multipotent differentiation potential (Figure S2). The hucMSC‐EVs were subsequently isolated from huc‐MSCs through differential and ultra‐high speed centrifugation (Figure 1B, a–c). TEM revealed that hucMSC‐EVs displayed the characteristic cup‐shaped morphology typical of EVs (Figure 1B, d). Nanoparticle tracking analysis (NTA) indicated an average concentration of 4.14 × 10¹⁰ ± 1.60 × 10⁹ particles per millilitre (Figure 1C). Western blotting verified exosomal markers CD9, CD63 and TSG101, while the absence of the organelle‐specific marker calnexin further affirmed the purity of the isolated hucMSC‐EVs (Figure 1D).

Twenty‐four hours after nebulised inhalation of PKH67‐labelled hucMSC‐EVs, immunofluorescence staining revealed a widespread distribution of PKH67‐labelled EVs in lung tissues (Figure 1E). In addition, immunofluorescence staining revealed that PKH67‐labelled EVs were localised in CD68‐positive AMs within the BALF. The positive endocytosis rate within 24 h was 94.1% (Figure 1F). These results indicate that AMs in lung tissues can efficiently take up hucMSC‐EVs. Subsequently, orthotopic left LTx was carried out in rats. After 2 h of reperfusion, the explanted lungs were harvested. Gross inspection showed that the left lung grafts treated with PBS exhibited severe pulmonary congestion and oedema. In contrast, these pathological manifestations were significantly alleviated in the group treated with hucMSC‐EVs; H&E staining demonstrated diffuse alveolar damage, which was characterised by the disruption of alveolar walls, alveolar oedema, as well as the infiltration of monocytes/macrophages and red blood cells. Significantly, these alterations were remarkably lessened in the hucMSC‐EVs‐treated group (Figure 1G).

Quantitative analysis revealed that the hucMSC‐EVs treated group exhibited significantly lower lung injury scores and wet/dry ratios compared with the PBS‐treated group (Figure 1H,I). The photo of lung oedema in rats after LTx are provided in Figure S3. To assess the functionality of the explanted lung, the PaO₂/FiO₂ of blood in the left pulmonary vein was measured. The findings demonstrated that the hucMSC‐EVs treated group demonstrated a significantly increased PaO₂/FiO₂ than the control group (Figure 1J). Additionally, lung tissue inflammatory cytokine levels (IL‐1β, IL‐6 and TNF‐α) were markedly higher in the PBS group than in the hucMSC‐EVs group (Figures 1K–M).

To explore the possible mechanism of IRI in LTx, lung tissues from grafts and a sham group were collected for RNA‐Seq analysis. A volcano plot was used to display the differentially expressed genes (DEGs) between the two groups (Figure 2A). Compared with the sham group, the LT‐PBS group had 3520 up‐regulated and 2715 down‐regulated transcripts. KEGG analysis identified 10 signalling pathways, such as the NF‐κB and NOD‐like receptor signalling pathways (Figure 2B). Thirty key genes from these two pathways were selected to create an expression cluster heatmap (Figure 2C). It showed that many genes in the NOD‐like receptor and NF‐κB signalling pathways were differentially expressed in LTx rat model compared with the sham group. Gene enrichment analysis (GSEA) was performed on NOD‐like receptor and NF‐κB signalling pathway, respectively, which are involved in NLRP3 inflammasome formation and activation (Figure 2D,E).

To validate the findings from the rat experiment described above, we generated a human single‐cell RNA sequencing (scRNA‐seq) atlas of IRI in LTx19 (Figure 2F). Focusing on macrophage subtypes, we found that both the positive ratio and expression level of NLRP3 were markedly elevated in the reperfusion group (Figure 2G,H). Our findings indicate that macrophage‐expressed NLRP3 may be a key mediator in donor lung IRI. Thus, we found NLRP3 expression was mainly concentrated in AMs (CD68 positive, CD11b negative) within lung tissue using immunofluorescence staining (Figure S4). Meanwhile, the majority of NLRP3 was localised in CD68‐positive macrophages, with significantly increased levels of expression observed in the PBS‐treated group compared with the hucMSC‐EVs‐treated group (Figures 2I–K). Additionally, Western blot analysis confirmed a significant increase in NLRP3 protein levels in the PBS‐treated group relative to the hucMSC‐EVs‐treated group (Figure 2L,M).

To elucidate the impact of hucMSC‐EVs on IRI in LTx, we performed miRNA sequencing on hucMSC‐EVs from three independent biological replicates. A total of 222 miRNAs were identified across the three preparations (Figure 3A), with the ten most abundant miRNAs highlighted in Figure 3B. Among these, miR‐34a‐5p, miR‐146a‐5p and miR‐127‐3p rank as the top three miRNAs with high expression abundance. Previously, research on miR‐34a‐5p mainly focused on tumourigenesis and development. For example, miR‐34a‐5p acts as a tumour suppressor in cutaneous squamous cell carcinoma cells by targeting SIRT6.20 Regarding miR‐146a‐5p, it functions as a critical modulator of inflammatory and immune responses. For example, miR‐146a‐5p suppresses the TLR4/NF‐κB pathway, thereby mitigating intestinal inflammation. Moreover, miR‐146a‐5p can also regulate intestinal homeostasis.21 Our findings confirmed that following LTx, the LT‐PBS group exhibited a reduced expression of miR‐146a‐5p compared with the sham group. However, after pre‐treatment with nebulised hucMSC‐EVs, the expression level of miR‐146a‐5p increased in the LT‐EVs group (Figure 3C). The experimental findings indicate that miR‐146a‐5p plays a critical role in the hucMSC‐EVs‐mediated amelioration of donor lung IRI during LTx.

Bioinformatic analysis indicated that IRAK1 and TRAF6 were identified as putative targets of miR‐146a‐5p (Figures 3D–F). This was validated through luciferase reporter assays (Figure 3G,H), and NR8383 cells transfected with a miR‐146a mimic displayed a marked reduction in IRAK1 and TRAF6 mRNA levels, as determined by RT‐qPCR (Figure 3I,J).

However, exogenous miR‐146a‐5p can induce a robust immune response via TLR7 activation through its UU‐containing motif. Therefore, directly introducing miR‐146a‐5p mimics into hucMSC‐EVs may not be optimal. Instead, we selected double‐stranded miR‐146a as the target miRNA for constructing engineered EVs, as its precursor does not activate the immune response. Next, we constructed engineered EVs by transfecting hucMSCs with a miR‐146a mimic (Figure 3K). Twenty‐four hours after transfection, hucMSCs transfected with miR‐146a‐5p exhibited a significant up‐regulation in its expression compared with both non‐transfected cells and those transfected with a negative control mimic, as determined by RT‐qPCR (Figure 3L). After 48 h, the cell culture supernatant from exosome‐free foetal bovine serum (FBS) medium was harvested, and engineered hucMSC‐EVs (EVs‐miR‐146a) isolated via differential ultracentrifugation contained significantly higher levels of miR‐146a‐5p than standard hucMSC‐EVs (Figure 3M). The time‐dependent uptake of PKH67‐labelled hucMSC‐EVs by NR8383 cells was observed at 2, 6 and 24 h. Specifically, at 24 h, the endocytosis rate of hucMSC‐EVs by NR8383 cells reached 96.6% (Figure 3N). To track the uptake of miR‐146a directly, cy5‐labelled miR‐146a was utilised to transfect hucMSCs, revealing cy5‐miR‐146a (red) in NR8383 cells 24 h after exposure to EVs‐miR‐146a (Figure 3O).

In the rat LTx model, donor lungs underwent cold preservation and reperfusion, resulting in NLRP3 activation in AMs. To evaluate the impact of engineered hucMSC‐EVs on NLRP3 inflammasome activation in macrophages, we established a cold preservation and warm reperfusion cell culture model using NR8383 cells (Figure 4A). To simulate in vivo delivery, NR8383 cells were treated with hucMSC‐EVs, EVs‐miR‐146a or PBS for 24 h. After simulated cold preservation for 24 h and warm reperfusion for 2 h, treatment with both hucMSC‐EVs and EVs‐miR‐146a substantially diminished IRAK1 and TRAF6 expression levels. Notably, hucMSC‐EVs exhibited a more pronounced inhibitory effect compared with EVs‐miR‐146a. (Figure 4B–E,H–J). Furthermore, compared with the PBS‐pretreated group, the phosphorylation of p65 (p‐p65) was significantly reduced, and nuclear translocation was decreased in the hucMSC‐EVs and EVs‐miR‐146a groups. These findings were confirmed by immunofluorescence staining and Western blotting (Figure 4F,G,H,K). Likewise, the EVs‐miR‐146a group demonstrated a more significant effect than the hucMSC‐EVs group.

The activation of NF‐κB serves as an initial signal for NLRP3 activation, thus regulating subsequent NLRP3 activation, while IRAK1 and TRAF6 enhance NF‐κB signalling, thereby playing critical roles in activating innate immune cells and triggering the subsequent inflammatory response. Moreover, NLRP3 inflammasome activation was inhibited in the groups treated with hucMSC‐EVs and EVs‐miR‐146a. Notably, the inhibitory effect was more prominent in the EVs‐miR‐146a‐treated group, as demonstrated by immunofluorescence staining and Western blot analysis (Figure 5A–D).

During NLRP3 activation, the inflammasome is formed through the recruitment of apoptosis‐associated speck‐like protein with a caspase recruitment domain (ASC) and pro‐caspase‐1. Western blot analysis showed that after NR8383 cells were subjected to simulated IR treatment, ASC expression was up‐regulated and cleaved caspase‐1 was activated (Figure 5C,E,I). The activated caspase‐1 then processes pro‐IL‐1β and pro‐IL‐18 into their mature, active forms, IL‐1β and IL‐18. Subsequently, Gasdermin D (GSDMD) is cleaved into GSDMD‐N (Figure 5C, F–H), inducing pyroptosis and leading to the secretion of IL‐1β and IL‐18. These elevated protein expressions were observed in the cold preservation and reperfusion NR8383 cell model, and administration of hucMSC‐EVs and EVs‐miR‐146a markedly inhibited NLRP3 activation and reduced the secretion of IL‐1β and IL‐18, with a more pronounced decrease observed following EVs‐miR‐146a intervention (Figure 5J–R).

In the rat LTx model, we further examined how hucMSC‐EVs and engineered EVs‐miR‐146a modulate NLRP3 inflammasome activation and lung injury. Immunofluorescence staining was performed to evaluate the expression levels of IRAK1, TRAF6 and phosphorylated p65 (p‐p65) in lung tissues, with AMs identified by CD68 (red) (Figure 6A–F). We observed significantly increased expression of IRAK1, TRAF6 and p‐p65 in CD68‐positive AMs, which was reduced in both the hucMSC‐EVs and EVs‐miR‐146a treatment groups, with EVs‐miR‐146a showing superior efficacy. Similar results were confirmed by Western blot analysis (Figure 6G–J).

Following the activation of NF‐κB, we further examined NLRP3 inflammasome activation (Figure 7A–G). Western blotting demonstrated an up‐regulation of NLRP3, cleaved caspase‐1, GSDMD‐N/GSDMD and ASC in the LT‐PBS group. In contrast, treatment with hucMSC‐EVs and EVs‐miR‐146a markedly reduced the levels of these NLRP3‐related proteins in explanted lungs, with the engineered EVs‐miR‐146a exhibiting the most pronounced effect.

The expression levels of IL‐1β and IL‐18 were examined using Western blot analysis and immunofluorescence staining (Figure 7H–Q). The LT‐PBS group exhibited elevated levels of these cytokines, with co‐localisation indicating significantly higher expression in CD68‐positive AMs compared with other lung cells. Both hucMSC‐EVs and EVs‐miR‐146a interventions effectively suppressed IL‐1β and IL‐18 expression, with engineered EVs‐miR‐146a showing the best results. Additionally, we quantified the mRNA levels of inflammatory cytokines IL‐1β, IL‐6 and TNF‐α in lung tissue (Figure 7Q–S). Both hucMSC‐EVs and EVs‐miR‐146a interventions reduced the expression of these cytokines, though the effects were not uniform. Furthermore, both treatments effectively inhibited lung cell apoptosis. TUNEL staining of lung tissue showed that the number of apoptotic cells was markedly decreased after the application of hucMSC‐EVs and EVs‐miR‐146a (Figure 7T, U). Notably, the effect of EVs‐miR‐146a was more pronounced (Table 1).

Transmission electron microscopy (TEM) of ultrathin sections of rat lung tissue was conducted to assess lung injury (Figures 7V). The LT‐PBS group exhibited oedema of the capillary basement membrane and discontinuity of the capillary endothelium, consistent with the lung injury scores obtained through H&E staining. This confirmed that the donor lungs experienced IRI following prolonged cold preservation. In contrast, lung tissue from the hucMSC‐EVs and EVs‐miR‐146a treatment groups showed more intact capillary structures and reduced oedema. Together with previous wet/dry ratio experiments, these findings indicate that the interventions alleviated lung oedema, with engineered EVs‐miR‐146a demonstrating the most effective results.

Lewis rats were procured from Changzhou Kavins Experimental Animal Co. Ltd. The animals were maintained under specific pathogen‐free conditions at the Animal Experimental Center of Wuxi People's Hospital, with free access to food and water. Ambient temperature was maintained at 22°C with a 12‐h light/dark cycle. Adult male rats weighing 300–350 g were used in this study.

MSCs were derived from the Wharton's jelly of human umbilical cords using an explant culture technique. The cells were cultured in DMEM/F12 medium containing 10% FBS and penicillin/streptomycin (100 U/100 mg/mL; Thermo Fisher Scientific, Burlington, Canada) at 37°C in a 5% CO₂ environment. Their identity was confirmed by flow cytometry, which demonstrated the expression of positive markers (CD73, CD105 and CD90) and the absence of negative markers (CD34, CD45, CD79α, CD11b and HLA‐DR). The trilineage differentiation potential of MSCs was induced using appropriate differentiation media. Cells were cultured in adipogenic differentiation medium (PD‐019; Procell, Wuhan, China) and osteogenic differentiation medium (PD‐017; Procell) for 3 weeks, and in chondrogenic differentiation medium (PD‐018; Procell) for 4 weeks.

HucMSC‐derived EVs were isolated through differential ultracentrifugation as previously described.39 HucMSCs at passages 5–6 were cultured in medium containing exosome‐free FBS for 48 h. The conditioned medium was collected and sequentially centrifuged at 300×g for 10 min and 3000×g for 30 min to remove cells and debris. Following a 10 000×g centrifugation for 40 min to eliminate apoptotic bodies and larger vesicles, the supernatant was passed through a 0.22 µm filter. Subsequently, ultracentrifugation was conducted at 120 000×g for 90 min, after which the supernatant was discarded, and the pellet containing EVs was resuspended in DPBS. This ultracentrifugation step was repeated, yielding a gelatinous precipitate containing purified EVs, all conducted at 4°C. TEM was utilised to examine EV morphology, while NTA quantified particle concentration and size distribution. Specific exosomal markers (positive markers: TSG101, CD9 and CD63; negative marker: calnexin) were examined via Western blotting. The isolated hucMSC‐EVs were subsequently utilised for both in vivo and in vitro experiments.

Twenty‐four hours before harvesting the lungs from the donor rats, the donor rats were administered nebulised hucMSC‐EVs or PBS using a vibrating mesh nebuliser (Contec Ltd., Qinhuangdao, China). The hucMSC‐EVs were prepared at a relative concentration of 1 mg dissolved in 1 mL of PBS. The nebulisation duration was 5 min, and it was performed only once. For a picture of the nebulised donor rats, please refer to Figure. Donor rats were anaesthetised with 5% isoflurane and positioned supine. Following hair removal and disinfection, a median thoracoabdominal incision was performed. Heparin (500 IU/kg) was administered via the inferior vena cava, and the pulmonary artery was perfused with a cold low potassium dextran (LPD) solution (4°C, 20 mL). The heart‐lung blocks were then excised, inflated with a 50% oxygen and 50% nitrogen gas mixture (2.5 mL), and the main bronchus was ligated. The heart‐lung block was stored on ice in cold LPD solution at 4°C for 12 h. An orthotopic left LTx model was established using the cuff technique. After 2 h of reperfusion, arterial blood was collected from the left pulmonary vein for blood gas analysis (i‐stat300G; Abbott Pharmaceutical Co. Ltd., Lake Bluff, IL, USA). At the conclusion of each experiment, grafted lungs were harvested and perfused with 0.9% saline. For the assessment of lung wet/dry ratios, the wet weight of each lung tissue sample was normalised to 0.1 g, with dry weight measured after heating the tissue at 65°C for 48 h. Lung injury scores were calculated in accordance with American Thoracic Society guidelines. S1

The NR8383 rat AM cell line (Procell) was employed to develop an ischaemia/reperfusion cell culture model. NR8383 cells were seeded in six‐well plates and cultured in low‐serum Ham's F‐12K (Kaighn's) medium (5% FBS). Cells were treated with hucMSC‐EVs, EVs‐miR‐146a or PBS for 24 h to simulate their in vivo delivery for donor management. After treatment, NR8383 cells were incubated in a hypoxic chamber at 4°C with 50% O₂, 5% CO₂ and the balance of N₂ for 24 h. Following 2 h of reperfusion at 37°C, the cells were moved to a standard culture incubator at 37°C and treated with ATP (5 mM) for 30 min.

Twelve human lung biopsies were collected at the end of the cold ischemic period and after 2 h of reperfusion for single‐cell sequencing. Details regarding sample collection, data preprocessing and quality control are outlined in our previous study.19 The GSE code of sequence data was GSE220797↗. The ‘FeaturePlot’ and ‘DotPlot’ were performed based on R package ‘Seurat’.

MiR‐146a engineered hucMSC EVs were constructed using a parental cell engineering approach.40 HucMSCs at passages 5–6 were transfected with 60 nM of miR‐146a mimic (Ribobio, Guangzhou, China) following the manufacturer's protocol. After 24 h, the cells were washed and cultured in exosome‐free FBS medium. The culture supernatant was collected after an additional 48 h of incubation, and engineered EVs were isolated using differential ultracentrifugation as previously described.

HucMSC‐EVs were labelled with PKH67 (Sigma) following the manufacturer's instructions. The rats were nebulised with PKH67‐labelled EVs, as previously described. The inflated lungs were harvested and rapidly immersed in liquid nitrogen after 24 h. Frozen lung sections were prepared for immunofluorescent analysis. NR8383 cells were fixed using 4% PFA after incubated with PKH67‐labelled exosomes at separate time points (2, 6, 24 h). Immunofluorescence analysis of these cells was performed as previously described.

According to a previously reported method,34, 35 multicolour fluorescence staining was used to determine the co‐localisation of the target proteins and AM (CD68) in lung tissue. In the different intervention groups, we assessed IRAK1, TRAF6, p‐p65, NLRP3, IL‐1β and IL‐18, with DAPI staining for cell nuclei. In addition, cell immunofluorescence staining was conducted to evaluate the expression of IRAK1, TRAF6, NLRP3, IL‐1β and IL‐18, along with the nuclear translocation of p‐p65, after the cell model was established. Briefly, NR8383 cells were seeded in glass‐bottom dishes. After treatment, cells were stained according to the standard protocol. DAPI was used to stain the cell nuclei, and imaging was conducted using a confocal microscope (Leica TCS SP8).

This study was approved by the ethics committees of Zhejiang University, Wuxi People's Hospital(No. 2023‐01) and Wuxi Second People's Hospital (No. 2022‐Y‐119). All experiments were performed in compliance with the applicable ethical guidelines and regulations.

All experiments were conducted independently at least three times. Data analysis was carried out using GraphPad Prism 10.1.2 software (GraphPad Software Inc., La Jolla, CA, USA). Results are expressed as mean ± standard deviation (SD). Comparisons between two groups were performed using unpaired t‐tests, with statistical significance set at p < .05.

The analysed data sets generated during the present study are available from the corresponding author on reasonable request.