Rictor/mTORC2 signalling contributes to renal vascular endothelial‐to‐mesenchymal transition and renal allograft interstitial fibrosis by regulating BNIP3‐mediated mitophagy

2.1. Patients and tissue samples

We enrolled 20 allogeneic kidney transplanted patients diagnosed with CAD at the First Affiliated Hospital of Nanjing Medical University, all of whom had received a pathological confirmation of renal allograft interstitial fibrosis via nephrectomy or biopsy of the transplanted kidneys. We procured normal kidney tissues from specimens obtained during surgical nephrectomy, ensuring they were located at least 5 cm from the margin of the tumour. Our study protocols were approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University and followed the guidelines of the Declaration of Helsinki and Istanbul. All participants signed informed consent. Table 1 presents the demographic characteristics of participants in the Normal and CAD groups.

2.2. Animals

Rictor^flox/flox mice and Tie2‐Cre mice (C57BL/6 background) were generously provided by Prof. Chunsun Dai (The Second Affiliated Hospital of Nanjing Medical University) and Prof. Aihua Zhang (Nanjing Medical University), respectively. Hybridization of Rictor^flox/flox and Tie2‐Cre mice created offspring by specifically deleting the Rictor gene in endothelial cells (Genotype: Rictor ^fl/fl, Cre^+/− ). The other littermates with the genotype Rictor ^fl/fl, Cre^−/− served as controls. The primer for genotyping was as follows:

Rictor genotyping: sense:5′‐GACACTGGATTACAGTGGCTTG‐3′

anti‐sense: 5′‐GCTGGGCCATCTGAATAACTTC‐3′

Tie2‐Cre genotyping, sense: 5′‐ATTTGCCTGCATTACCGGTC‐3′

anti‐sense: 5′‐ATCAACGTTTTCTTTTCGG‐3′.

Additionally, C57BL/6 and BALB/C mice were sourced from the Animal Center of Nanjing Medical University. All animals were housed in the specific pathogen‐free Laboratory Animal Center at Nanjing Medical University.

2.3. Renal allogeneic transplantion mice models

All animal experiments were approved by the Animal Research Ethics Committee of Nanjing Medical University. Renal allogeneic transplantation mice models were constructed as described below. Mice were anesthetised with isoflurane inhalation (RWD Life Science), and the right kidney of the recipient mouse was excised in situ. Subsequently, the left kidney from donor mouse was subsequently transplanted into the recipient mouse. The average total ischemia time during the surgery ranged from 40 to 60 min. BALB/C mice were used as recipients and syngeneic donors (Syn group), whereas C57BL/6 mice were applied as allogeneic donors to create chronic renal allograft interstitial fibrosis mice models (Allo group). Contralateral nephrectomies on the recipient mice were performed on day 7 after kidney transplantation. Tacrolimus (Astellas Pharmaceutical) was administered intraperitoneally at 1.5 mg/kg q.d. for the first 7 days after transplantation and 1.5 mg/kg q.w. for the next 4 weeks. At 16 weeks post‐transplantation, mice were sacrificed humanely, and the harvested transplanted kidneys were immediately fixed in paraffin or stored in liquid nitrogen.

Likewise, BALB/C mice receiving donor kidneys from Rictor^fl/fl Cre^+/− or Rictor^fl/fl Cre^−/− C57BL/6 mice were designated as the Rictor^−/− Allo group or the WT Allo group, respectively. Two weeks after the resection of the left native kidneys in recipient mice, tamoxifen (cat: T5648, Sigma‐Aldrich) was intraperitoneally injected for 5 days at 60 mg/kg to induce Rictor gene ablation.

2.4. Adeno‐associated virus infection in vivo

For in vivo reduction of BNIP3 expression, AAV9 vectors expressing a shRNA targeting BNIP3 (AAV‐shBNIP3) or a control shRNA (AAV‐shNC) were generated by Jiangsu KeyGEN BioTECH to achieve BNIP3 depletion. Each mouse received a tail intravenous injection of the adenoassociated virus (AAV) vector with a final titer of 1 × 10¹² viral genome, in the third week after kidney transplantation.

2.5. Single‐cell transcriptomic analysis

The profiles of GSE195718 and GSE131882 were obtained from the GEO database (www.ncbi.nlm.nih.gov/geoprofiles/). Uniform manifold approximation and projection (UMAP) was used for dimension reduction. The identification of hub genes was performed using Cytoscape software v3.10.1. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) enrichment analysis were conducted using DAVID (https://david.ncifcrf.gov/tools.jsp). ²⁹ , ³⁰

2.6. Cell cultures and treatment

Human Umbilical Vein Endothelial Cells (HUVECs), obtained from Jiangsu KeyGEN BioTECH (KeyGEN, KG419), were cultured in Ham's F‐12K medium (Gibco, 21127030) with 1% Endothelial Cell Growth Supplement (ECGS) (KeyGEN, KGY1052), 1% penicillin‐streptomycin, 10% fetal bovine serum (Gibco, 10099141), and .1 mg/mL heparin (KeyGEN, KGY3149H) at 37°C in a humidified atmosphere containing 5% CO₂. After being seeded on appropriate culture plates overnight, HUVECs were exposed to 100 ng/mL tumour necrosis factor‐α (TNFα) (Novoprotein, C008) for different time periods.

Rictor gene knockout of HUVECs and corresponding control cells were generated using the CRISPR/Cas9 system obtained from Hanbio biotechnology. Rictor gene knockout of HUVECs were confirmed by sequencing and validated by western blotting. For Rictor overexpression, HUVECs were transfected with either Empty vector lentivirus or Flag‐Rictor lentivirus (Genechem) following the manufacturer's instruction.

2.7. Plasmids, small interfering RNA and reagents

The BNIP3‐HA (P29482), HA‐UB‐K48 (P31802), Myc‐UB‐K48 (P44473), HA‐UB‐K63 (P31800), and Rictor (P41727) plasmids were obtained from MIAOLING BIOLOGY. HA‐UB‐WT, HA‐UB‐K63R, and HA‐UB‐K48R plasmids were sourced from Addgene. HA‐tagged site‐directed mutations of BNIP3 (K130R, K135R) were generated by Jiangsu KeyGEN BioTECH. His‐tagged MARCH5 plasmid was obtained from Beijing Tsingke Biotech Company. All the constructed plasmids were confirmed by DNA sequencing. The plasmids transfection was performed using Lipofectamine 3000 (Invitrogen, L3000008).

The small interfering RNA (siRNAs) were synthesized by Jiangsu KeyGEN BioTECH and transfected using RFect siRNA Transfection Reagent (Baidai biotechnology, 11013). The targeted siRNA sequences were shown as follows:

siRictor: 5′‐GCAGACUCUAUGCAACAAATT‐3′

siBNIP3: 5′‐GAACUGCACUUCAGCAAUAAUTT‐3′

siMARCH5: 5′‐GGUUGUAGGUCAUAAAGAATT‐3′

CCCP (S6494), Mdivi‐1 (S7162), MG132 (S2619), CHX (S7418), and CQ (S6999) were purchased from Selleck.

2.8. Western blotting and IP assays

Cells or tissues were lysed using lysis buffer (Fdbio science, FD009) supplemented with 1 mM phenylmethylsulfonyl fluoride (Fdbio science, FD0100) as well as a protease and phosphatase inhibitors cocktail (ThermoFisher Scientific, 78442). The mitochondrial protein of HUVECs was isolated and extracted using a mitochondrial protein extraction kit (BestBio). Protein concentration was quantified with a BCA protein quantitative kit (Beyotime, P0010).

For western blotting assays, proteins were separated by SDS‐polyacrylamide gel electrophoresis and transferred to PVDF membranes (Millipore, ISEQ00010). After blocking in 5% nonfat milk solution at room temperature for 2 h, the membranes were incubated with the primary antibodies at 4°C overnight, then with the HRP‐conjugated secondary antibodies at room temperature for 2 h and detected by ECL reagent (Biosharp, BL520A).

For immunoprecipitation (IP) assays, cells were lysated using IP lysis buffer (ThermoFisher Scientific, 87787) containing 1 mM PMSF. Protein samples were incubated with the primary antibodies at 4°C overnight and then mixed with Protein A/G magnetic beads (Bimake, B23201) at room temperature for 30 min. For proteins tagged with HA or Flag, protein lysates were directly incubated with anti‐HA (Bimake, B26201) or anti‐Flag (Bimake, B26101) magnetic beads at 4°C overnight. After magnetic separation, the protein‐beads complexes were washed with lysis buffer three times, then boiled in 1X SDS loading buffer and subjected to western blotting analysis.

The primary antibodies and secondary antibodies used in our research are listed in Table S1.

2.9. Ubiquitination assays

To evaluate the ubiquitination of endogenous BNIP3, we immunoprecipitated BNIP3 using anti‐BNIP3 antibody and analyzed it by western blot assays with anti‐Ub antibody. For the analysis of BNIP3 ubiquitination, HUVECs were transfected with plasmids expressing the target genes. The whole cell lysates were immunoprecipitated with indicated antibodies or magnetic beads and subjected to western blotting assays.

2.10. Cycloheximide chase assays

Cycloheximide (CHX) chase assays were performed to determine the half‐lives of proteins. Cells were treated with 200 μg/mL CHX at different time points, and then cell proteins were extracted to analyze BNIP3 levels by western blotting assays.

2.11. Transmission electron microscopy analysis

Preparation of HUVECs for transmission electron microscopy (TEM) analysis was as described. ³¹ The prepared electron microscope samples were processed and scanned using TEM (FEI Tecnai G2) at the Analysis and Testing Center, Nanjing Medical University.

2.12. RNA extraction and reverse‐transcriptase PCR assay

Total RNA was isolated from cell lines using the RNA‐Quick Purification Kit (Shanghai YISHAN biotechnology, RN001) and reverse transcribed with the HiScript II Q RT SuperMix (Vazyme, R222‐01). The qPCR was carried out using the ChamQ SYBR qPCR Master Mix (Vazyme, Q341‐02) on the StepOnePlus Real‐Time PCR system (Applied Biosystems). The specific primer sequences used were as follows:

BNIP3: forward: 5′‐CAGGGCTCCTGGGTAGAACT‐3′, reverse: 5′‐CTACTCCGTCCAGACTCATGC‐3′

GAPDH: forward: 5′‐GGAGCGAGATCCCTCCAAAAT‐3′, reverse: 5′‐ GGCTGTTGTCATACTTCTCATGG‐3′

2.13. Renal allograft histological pathology staining analyses

Renal allograft histological pathology analyses were evaluated by Masson trichrome staining and hematoxylin and eosin (H&E) as previously described. ³¹ HE staining was utilized to assess allograft structure changes, while Masson staining was employed to assess renal interstitial fibrosis degree.

2.14. Indirect immunofluorescence and immunohistochemical staining assays

After renal tissues were fixed and embedded, renal sections underwent indirect immunofluorescence (IF) and immunohistochemical (IHC) staining. IHC staining of tissues was performed with the HRP Kit (CWbio, CW2069S). In the case of IF staining, renal sections were initially baked at 65°C for 2 h. Subsequently, they were deparaffinized with xylene, rehydrated through graded ethanol solutions (100%, 100%, 95%, 85%, and 75%, respectively), and rinsed with distilled water. Following this, the renal sections were undergone antigen retrieval using an enhanced citrate solution (Beyotime, P0083). Subsequently, they were incubated with 3% hydrogen peroxide and then blocked with a Blocking Buffer (Beyotime, P0260). Subsequent to these steps, the renal sections were incubated with the primary antibodies at 4°C overnight, and then with the secondary antibodies at room temperature for 1 h. Visualization of nuclei was achieved using DAPI Fluoromount‐G (SouthernBiotech, 0100−20).

In preparation for cells’ IF staining, cells were seeded onto glass coverslips at 30% confluence. Following fixation of the cells using 4% formalin, permeabilization, and blocking with Blocking Buffer (Beyotime, P0260), the cell samples were incubated with target protein antibodies at 4°C overnight. Following that, the coverslips were incubated with fluorescence‐conjugated secondary antibodies and stained with DAPI Fluoromount‐G (SouthernBiotech, 0100−20). To label the mitochondria, the cells were incubated with MitoTracker Red CMXRos (Thermo Fisher, M7512), followed by fixation and immunostaining using the specified antibody. Images were captured either through an Olympus optical microscope or a Zeiss LSM 880 NLO confocal microscope.

The primary antibodies and secondary antibodies used for this study are listed in Table S1.

2.15. ELISA assay

The levels of TNFα in serum of the Syn and Allo groups were assessed with Mouse TNFα ELISA kit (Elabscience, E‐EL‐M3063) according to the manufacturer's instructions.

2.16. Statistical analysis

The data were presented as mean ± SEM and each experiment was repeated independently at least three times. Student's t‐test was applied to determine statistical differences between the two groups, while one‐way ANOVA followed by Dunnett or Tukey post hoc tests were used for statistical comparisons among multiple groups. The GraphPad Prism Software 8.0 was used for statistical analyses. A P‐value of <0.05 was considered statistically significant.

3.1. The upregulation of Rictor in allogeneic renal vascular endothelial cells of CAD patients is unveiled through single‐cell transcriptomic analysis

To delineate potential key genes involved in the progression of renal vascular EndMT and CAD progression, we systematically analyzed publicly available GSE195718 and GSE131882 datasets. A comprehensive summary of the included datasets is presented in Figure 1A. Patients were categorized into three groups: the native group (comprising healthy adult kidney samples from three individuals), the normal allograft group (consisting of normal renal allograft tissues from three stable kidney transplanted patients), and the fibrotic allograft group (encompassing fibrotic renal allograft tissues from six CAD patients). Subsequently, we acquired 44716 cells from 12 patients for single‐cell RNA sequencing (scRNA‐seq) analysis. UMAP analysis identified all major cell types within the kidney tissues based on the expression of lineage‐specific markers (Figure 1B). EndMT has been established as an important contributor to the development of renal allograft fibrosis. ¹¹ Therefore, our focus was on the differential expressed genes (DEGs) in endothelial cells, identified using PECAM1 (CD31) and CD34 markers (Figure 1C). We identified 2779 genes comprising 1771 upregulated genes and 1008 downregulated genes between the native and fibrotic allograft groups (Figure 1D). Hub genes in endothelial cells, including Rictor, were illustrated in Figure 1E. GO analysis for the DEGs in endothelial cells revealed enrichment in focal adhesion, actin cytoskeleton organization, actin filament organization, cell‐cell adhesion, and actin filament organization (Figure 1F). KEGG analysis depicted an upregulated pathway including focal adhesion, regulation of adherents junction, actin cytoskeleton, and tight junction (Figure 1G). Subsequently, we analyzed DEGs across the three groups and found that Rictor was highly expressed in PECAM1‐positive endothelial cells in fibrotic allograft group compared to the native group, with its expression in the fibrotic allograft group exhibiting an increasing trend compared to the normal allograft group (Figure 1H). Pseudotime analysis further validated a significant elevation in Rictor expression in endothelial cells along a trajectory from normal allograft tissues to fibrotic allograft tissues (Figure 1I). These findings robustly supported the important role of Rictor in the pathogenesis of EndMT and renal allograft fibrosis.

3.2. Rictor/mTORC2 signalling is upregulated in renal vascular endothelial cells of CAD patients and mice

To explore the potential involvement of the Rictor/mTORC2 pathway in renal allograft interstitial fibrosis, we initially analyzed Rictor expression in renal allograft tissues from normal individuals (normal group) and CAD patients (CAD group). The results of histological pathology staining indicated disordered renal structure and significant interstitial fibrosis in the CAD group (Figure 2A). Furthermore, IHC assays revealed heightened expression of FN and Rictor, accompanied by downregulation of CD31 expression in the CAD group (Figure 2B, C).

In addition, we generated an allogeneic kidney transplanted mice model to confirm the involvement of Rictor/mTORC2 signalling in the development of renal allograft fibrosis. BALB/C mice received kidneys from either BALB/C or C57BL/6 mice were designated as the Syn (syngeneic) or Allo (allogeneic) group, respectively. The renal allografts of the Allo group displayed interstitial inflammatory cell infiltration, glomerulosclerosis, and significant interstitial fibrosis, as revealed by HE and Masson staining (Figure 2D). Moreover, IHC staining showed increased expressions of FN and αSMA, along with decreased CD31 expression in the Allo group (Figure 2E, F). These findings were confirmed by western blotting assays of Collagen I, αSMA, and CD31 (Figure 2G, H), suggesting the presence of reanl vascular EndMT and interstitial fibrosis in the Allo group. Furthermore, increased Rictor expressions in renal vascular endothelial cells from the Allo group were detected via IF assay (Figure 2I, J). Consistently, remarkable inductions of Rictor, p‐Akt^Ser473, p‐Akt^Thr308, p‐PKCα, and p‐SGK1 were detected in kidney tissues from the Allo group (Figure 2K, L), suggesting the activation of the mTORC2 signalling pathway.

In our previous study, TNFα was conclusively identified as a contributor to EndMT. ¹² IHC assays revealed an augmented expression of TNFα in both the CAD group and Allo group, compared to the control and Syn group (Figure S1A–D). ELISA assay also demonstrated a significant elevation in TNFα protein levels in the serum of the Allo group in comparison to the Syn group (Figure S1E). Correspondingly, a decreased CD31 expression and elevated expressions of FSP‐1 and FN were observed in HUVECs following TNF‐α treatment (Figure S1F, G). Besides, the media of the Rictor/mTORC2 signalling pathway were activated in HUVECs treated with TNFα at different time points (Figure S2A, B). Collectively, these findings indicate that Rictor/mTORC2 signalling is upregulated in vascular endothelial cells from the fibrotic kidneys with allograft transplantation, and may be associated with development of renal vascular EndMT.

3.3. Activation of Rictor/mTORC2 signalling mediates EndMT formation

To elucidate the influence of the Rictor/mTORC2 pathway on EndMT, Rictor siRNAs were applied to reduce Rictor expression in HUVECs, followed by stimulation with 100 ng/mL TNFα for 48 h. Subsequent analyses revealed a significant reduction in Rictor expression and notable downregulation of mTORC2 signalling media in HUVECs transfected with Rictor siRNA compared with control siRNA under TNFα treatment (Figure 3A, B). As depicted in Figure 3C, D, TNFα treatment resulted in upregulation of FN, Collagen I, and αSMA expression, alongside a decrease in CD31 expression, all of which were effectively reversed by Rictor siRNA. These changes at the protein level were also corroborated by IF staining assays (Figure 3E, F). In a subsequent experiment, we transfected HUVECs with either control or Rictor overexpression plasmids for 24 h, followed by TNFα treatment (100 ng/mL) for 48 h. In line with our previous results, Rictor overexpression in HUVECs activated the mTORC2 pathway and exacerbated TNFα‐induced EndMT, as confirmed by western blotting and IF staining assays (Figure 3G–L). Furthermore, in the human renal glomerular endothelial cells, Rictor knockdown with Rictor siRNA could also attenuate TNFα‐induced EndMT (Figure S3A, B). Collectively, these findings robustly demonstrate that Rictor/mTORC2 signalling has a significant impact on mediating the development of EndMT.

3.4. Rictor gene knockout could enhance BNIP3‐mediated mitophagy in HUVECs

To assess the potential mechanism of Rictor on EndMT formation, we established a Rictor gene knockout (KO) HUVEC line utilizing CRISPR/Cas9‐mediated gene editing (Figure S4A, B) and subsequently confirmed its validity via a western blotting assay (Figure 4A). Additionally, we observed a substantial inactivation of the Rictor/mTORC2 signalling pathway after Rictor knockout under TNFα treatment (Figure 4A). To further explore how Rictor/mTORC2 signalling regulates EndMT progression in endothelial cells, we performed a transcriptomic analysis of Rictor KO HUVECs and control cells. The KEGG pathway analysis of the RNA‐seq data showed significant enrichment of mitophagy‐related genes in Rictor KO HUVECs compared to the controls (Figure 4C). IF staining revealed that colocalization of MitoTracker with LC3, indicating mitophagosome formation, was more prominent in HUVECs after Rictor KO (Figure 4D, E). Transmission electron microscopy (TEM) scan analysis also provided the final piece of evidence, revealing an increased number of mitophagosomes in HUVECs with Rictor KO (Figure 4F, G). Western blotting assay further confirmed that Rictor KO induced a significant increase in LC3 expression, accompanied by a noticeable downregulation of mitochondrial membrane protein VDAC1 and TIMM23, particularly after TNFα co‐stimulation (Figure 4H, I), indicating that Rictor KO could induce mitophagy activation.

Mitophagy, a selective process for eliminating dysfunctional or damaged mitochondria via autophagy, is regulated by two major mitophagy pathways: the PINK1–Parkin pathway and the mitophagy receptors pathway. ³² In our endeavour to elucidate the mechanisms underlying Rictor's modulation of mitophagy, we assessed several pivotal molecules, including PINK1, NIX, FUNDC1, and BNIP3, all of which were reported to play crucial roles in mitophagy mediation. Western blotting analyses revealed a marked upregulation of BNIP3 expression following Rictor gene knockout in HUVECs (Figure 4J, K). Conversely, the expression of PINK1, NIX, and FUNDC1 remained statistically unchanged between the Rictor KO and control groups (Figure 4J, K). Mitochondrial protein extraction results further verified the augmented BNIP3 expression in HUVECs after Rictor gene knockout (Figure 4L, M). Additionally, Rictor gene knockout in HUVECs resulted in increased colocalization of BNIP3 with LC3‐II and MitoTracker (Figure 4N–Q). Taken together, these findings strongly suggest that Rictor expression depletion could enhance BNIP3‐mediated mitophagy in renal vascular endothelial cells, subsequently attenuating EndMT and renal allograft interstitial fibrosis.

3.5. Changes in mitophagy activity and BINP3 expression can influence TNFα‐induced EndMT in HUVECs

To assess the impact of mitophagy activity on TNFα‐induced EndMT, mitophagy agonist (CCCP) and inhibitor (Mdivi‐1) were employed in HUVECs. HUVECs were pretreated with 20 μM CCCP for 2 h, followed by 100 ng/mL TNFα treatment for 24 h. Western blotting analysis demonstrated that CCCP reversed TNFα‐induced upregulation of Collagen I and αSMA expression, and the downregulation of CD31 expression (Figure 5A, B). Additionally, HUVECs were pretreated with 10 μM Mdivi‐1 for 24 h, followed by 100 ng/mL TNFα treatment for 24 h. Figure 5C, D illustrated that Mdivi‐1 treatment exacerbated TNFα‐induced EndMT, as confirmed by western blotting assay.

Subsequently, to further validate the importance of BNIP3‐mediated mitophagy in the mitigation of TNFα‐induced EndMT resulting from Rictor downregulation, BNIP3 siRNA was transfected into the control and Rictor KO groups. In HUVECs, following transfection with BNIP3 siRNA, the Rictor knockout‐induced upregulation of mitophagy was essentially nullified, manifesting as comparable levels of BNIP3, TOMM20, and LC3 expression (Figure 5E, F). Meanwhile, BNIP3 knockdown abrogated the alleviation of EndMT induced by downregulated Rictor expression (Figure 5G, H). In summary, these results strongly suggest that BNIP3‐mediated mitophagy activation could ameliorate TNFα‐induced EndMT in HUVECs.

3.6. Rictor/mTORC2 signalling could promote the proteasome depended degradation of BNIP3 at K130 through K48‐linked ubiquitination

To explore how the regulatory mechanism of Rictor on BNIP3, we initially assessed BNIP3 expression at the transcriptional level in HUVECs in the Rictor KO and the control groups, observing comparable BNIP3 mRNA levels in both groups (Figure 6A). Therefore, we hypothesized that Rictor may influence BNIP3 protein degradation rather than its mRNA transcription. To confirm this, we utilized CHX, a well‐established protein synthesis inhibitor, to measure protein half‐lives. Our observation indicated an extension in the half‐life of BNIP3 in HUVECs following Rictor knockout (Figure 6B, C). Furthermore, we discovered that the proteasome inhibitor MG132, but not the autophagy inhibitor CQ, impeded the BNIP3 upregulation induced by Rictor knockout in HUVECs (Figure 6D–G), further supporting the involvement of Rictor in mediating the proteasome depended on degradation of BNIP3. Consistently, Rictor knockout in HUVECs resulted in a reduction in the ubiquitination of endogenous BNIP3 (Figure 6H), while Rictor overexpression led to an increase in the ubiquitination of endogenous BNIP3 (Figure 6I). Mammalian cells harbour a diverse array of ubiquitin chains, each serving specific biological functions. Predominantly, K48‐linked polyubiquitination facilitates proteasome‐dependent degradation of substrate proteins, while K63‐linked polyubiquitination specifically modifies certain proteins, thereby initiating corresponding signal pathways. To ascertain the particular polyubiquitination modification of BNIP3 affected by Rictor, we transfected either K48‐linked or K63‐linked polyubiquitination plasmids into HUVECs. Our results indicated that Rictor knockout led to a decrease in K48‐linked polyubiquitination of BNIP3, leaving K63‐linked polyubiquitination unaffected (Figure 6J). To verify this, we expressed K48‐resistant (K48R) or K63‐resistant (K63R) polyubiquitin chains in Rictor‐overexpressed HUVECs. Consistently, K48R polyubiquitin abolished the augmented ubiquitination of BNIP3 (Figure 6K). These results confirmed that Rictor regulated the proteasome‐dependent degradation of BNIP3 through K48‐linked polyubiquitination. Our subsequent objective was to identify the specific lysine residues on BNIP3 responsible for its ubiquitination and subsequent degradation. Among the 12 lysine residues in human BNIP3 protein, two were predicted as potential ubiquitination sites by CKSAAP_UbSite (http://systbio.cau.edu.cn/cksaap_ubsite/) (Figure 6L). To test these predictions, we created two HA‐tagged BNIP3 mutant plasmids with individual lysine residues substituted by arginine. The following immunoprecipitation assays demonstrated that the K130R mutant showed reduced ubiquitination, suggesting that the lysine residue K130 is crucial for the BNIP3 ubiquitination mediated by Rictor (Figure 6M). Furthermore, the K130 mutant displayed a substantial reduction in BNIP3 degradation mediated by Rictor overexpression (Figure 6N, O). Therefore, these results elucidate that Rictor regulates mitophagy by promoting the K48‐linked ubiquitination of BNIP3 at the K130 site.

3.7. MARCH5 is the E3 ubiquitin ligase that specifically regulates the proteasome‐dependent degradation of BNIP3 mediated by Rictor/mTORC2 signalling pathway

To investigate the specific E3 ubiquitin ligase regulating Rictor‐medicated degradation of BNIP3, we used the UbiBrowser platform (http://ubibrowser.ncpsb.org) to predict the ligases interacting with BNIP3 (Figure 7A). According to our predicted results, we screened the MARCH5 protein that was downregulated by Rictor knockout (Figure 7B, C). The following immunoprecipitation assays revealed endogenous interactions between BNIP3 and MARCH5 in HUVECs (Figure 7D), suggesting the specific E3 ubiquitin ligase mediating BNIP3 degradation is MARCH5. The CHX chase assay revealed that MARCH5 knockdown significantly prolongs the half‐life of BNIP3 (Figure 7E, F). We found that it was MG132 and not CQ that could reverse the upregulated BNIP3 caused by MARCH5 knockdown in HUVECs (Figure 7G, H). The ubiquitination assays showed that MARCH5 knockdown decreased BNIP3 ubiquitination (Figure 7I), while MARCH5 overexpression promoted K48‐linked, but not K63‐linked ubiquitination of BNIP3, consistent with the effect of MARCH5 on the degradation of BNIP3 (Figure 7J). To identify the ubiquitinated sites of BNIP3 regulated by MARCH5, the His‐MARCH5 plasmid and HA‐BNIP3 plasmid or other mutants were co‐transfected into HUVECs. The ubiquitination assays indicated that the ubiquitination of BNIP3 in the K130R mutant of BNIP3 was significantly reduced compared with that of wild‐type or K135R mutant (Figure 7K), suggesting the Lys130 site was the ubiquitinated site of BNIP3 mediated by MARCH5. To explore the influence of MARCH5 on the proteasomal degradation of BNIP3 mediated by Rictor, we transfected MARCH5 siRNA into the control or Rictor overexpressed HUVECs. The ubiquitination assays indicated that Rictor overexpression promoted the ubiquitination of BNIP3, which could be reversed by silencing MARCH5 (Figure 7L). All these results demonstrate that MARCH5 acts as the specific E3 ubiquitin ligase regulating proteasome‐dependent degradation of BNIP3 mediated by the Rictor/mTORC2 signalling pathway.

3.8. Ablation of Rictor gene in endothelial cells can alleviate renal vascular EndMT and allograft interstitial fibrosis

To further investigate the influence of Rictor expression in endothelial cells during the progression of renal allograft interstitial fibrosis, we created the animal model with endothelial cell‐specific Rictor gene deletion using the Cre‐LoxP system. The crossbreeding strategy employed to generate the knockout mice is shown in Figure 8A. Subsequently, we established chronic renal allograft interstitial fibrosis mice model, the procedure is shown in Figure 8C. BALB/C mice received donor kidneys from Rictor ^fl/fl Cre^+/− (Figure 8B, Lane 4) or Rictor ^fl/fl Cre^−/− (Figure 8B, Lane 2) mice were divided into the Rictor^−/− Allo group or the WT Allo group, respectively. We employed western blotting analysis and IF staining assays to evaluate Rictor expression and mTORC2 signalling in the transplanted kidneys (Figure 8D–F). The results demonstrated a significant reduction in Rictor protein levels and a substantial suppression of mTORC2 signalling in the Rictor^−/− Allo group than the WT Allo group (Figure 8E, F).

We performed HE and Masson staining assays on transplanted kidney tissues from various groups. The Rictor^−/− Allo group demonstrated substantial reductions in renal glomerular sclerosis and interstitial ECM deposition (Figure 8G). The fibrotic area in the WT Allo group was significantly larger than that in the Rictor^−/− Allo group (Figure 8G). Additionally, we assessed the extent of renal fibrosis and vascular EndMT. The western blotting assay demonstrated that the targeted deletion of Rictor expression in endothelial cells resulted in a reduction in Collagen I and αSMA expression, while concurrently enhancing CD31 expression (Figure 8H, I). IHC staining for FN, αSMA, and CD31 expression provided further confirmation of the western blotting results (Figure 8J, K). In summary, those results underscored the ameliorative impact of Rictor deletion in endothelial cells on renal vascular EndMT and interstitial fibrosis in allogeneic kidney transplanted mice.

3.9. Rictor gene ablation could enhance BNIP3‐mediated mitophagy in chronic renal allograft interstitial fibrosis mice model

In chronic renal allograft interstitial fibrosis mice model, we explored the effects of Rictor ablation on mitophagy activity. Elevated LC3 expression in transplanted kidney tissues was observed in the Rictor^−/− Allo group, while TOMM20 expression was reduced in the Rictor^−/− Allo group compared with the WT Allo group (Figure 9A, B). These findings suggested an augmentation of mitophagy in the Rictor^−/− Allo group. Furthermore, Rictor ablation resulted in a significant upregulated BNIP3 expression and downregulated MARCH5 expression (Figure 9C, D). Moreover, the Rictor^−/− Allo group exhibited increased colocalization of CD31 with LC3 and BNIP3 (Figure 9E–H), further confirming the upregulated BNIP3‐mediated mitophagy in renal vascular endothelial cells. These findings collectively demonstrate that BNIP3 expression and mitophagy activity are augmented in the renal vascular endothelial cells following Rictor knockout.

3.10. Knockdown of BNIP3 reverses mitophagy activation, the improved renal vascular EndMT and interstitial fibrosis caused by Rictor gene knockout

To delve further into the role of the mitophagy receptor BNIP3 in the enhanced mitophagy and amelioration of EndMT and renal allograft interstitial fibrosis associated with Rictor ablation in endothelial cells, we injected the kidney‐transplanted mice from the WT Allo group or Rictor^−/− Allo group with AAV‐shBNIP3 or AAV‐shNC. We found that BNIP3 knockdown substantially negated the effect of Rictor ablation on mitophagy, as evidenced by the diminished expressions of LC3 and BNIP3 in transplanted kidney tissues from both the WT Allo and Rictor^−/− Allo groups (Figure 10A, B). In concordance, the knockdown of BNIP3 led to reduced colocalization of CD31 with BNIP3 and LC3, denoting inhibited BNIP3‐mediated mitophagy in endothelial cells within the transplanted kidney tissues (Figure 10C, D).

Moreover, BNIP3 knockdown effectively mitigated the disparities in the expression of EndMT markers, including FN, αSMA, and CD31, observed between transplanted kidney tissues from the WT Allo and Rictor^−/− Allo groups (Figure 10E, F). Additional validations were provided through HE, MASSON, and IHC staining assays, which confirmed that BNIP3 knockdown counteracted the ameliorative effects on renal allograft damage and fibrosis attributed to Rictor ablation in endothelial cells (Figure 10G–J). Hence, these results strongly support the hypothesis that inhibition of Rictor/mTORC2 signalling could enhance BNIP3‐mediated mitophagy in renal vascular endothelial cells, which, in turn, alleviates the development of EndMT and renal allograft interstitial fibrosis.