Protective effect of mesenchymal stromal cells in diabetic nephropathy: the In vitro and In vivo role of the M-Sec-tunneling nanotubes

Diabetic nephropathy (DN), the major cause of end stage renal disease, is characterised by both podocyte damage and mesangial expansion, resulting in increased glomerular permeability to proteins and a progressive decline in renal function [1,2] The diminished expression of podocyte-associated proteins, including nephrin and podocin, along with podocyte apoptosis, are early features of DN and significant factors in the development of albuminuria [3,4].

Hyperglycemia and advanced glycation end products (AGEs) are important determinants of podocyte injury in DN [5,6]. Moreover, there is growing evidence indicating that mitochondrial dysfunction plays a key role in linking hyperglycemia to podocyte damage. Altered mitochondrial membrane potential (MMP) and bioenergetics, as well as increased mitochondrial oxidative stress, have been observed in renal cells exposed to a diabetic milieu and in kidneys from diabetic animals [7–12]. However, effective strategies for supplementing, repairing, and replacing damaged mitochondria are not yet available.

Recently, tunnelling nanotubes (TNTs) have emerged as an important mechanism for horizontal mitochondrial transfer. TNTs are straight open-ended membrane channels with diameters ranging from 50 to 200 nm and lengths spanning several cell diameters. They connect distant cells and, as characteristic properties, they do not make contact with the substrate and contain either a F-actin or a tubulin backbone [13–16]. Cells generate TNTs in response to stresses, including serum deprivation [17–19], inflammation [20–22], and oxidative stress [17,23], with the cytosolic protein M-Sec playing a crucial role in TNT formation in most cell types [24,25].

TNTs mediate the transfer of cellular components, including organelles from healthy donor cells to stressed target cells [26–31], thereby enhancing the resistance of recipient cells to insults. Heterotypic TNTs have also been observed between neighbouring cells as well as between progenitors/stem cells and adult cells. For instance, mesenchymal stromal cells (MSCs) [32] can form heterotypic TNTs with cardiomyocytes [33–35], endothelial cells [36–38], and cancer cells [38–40]. MSCs are particularly efficient as mitochondrial donors due to their high expression of Miro1, a mitochondrial Rho-GTPase responsible for fine-tuning actin- and tubulin-dependent mitochondrial motility and positioning [41,42].

We have previously demonstrated that podocytes can form TNTs with other podocytes in a M-Sec-dependent manner in response to stresses, including those related to diabetes [43,44]. Whether transfer of mitochondria from MSCs to podocytes via TNTs can occur in the context of DN remains unknown. Supporting this hypothesis, previous studies have shown the benefits of exogenous MSC treatment in experimental DN [45–48], though the role of horizontal M-Sec-TNT-mediated mitochondrial transfer was not investigated.

In this study, we aimed to explore whether a M-Sec-TNT-mediated exchange of mitochondria from MSCs to podocytes occurs in both in vitro and in vivo models of DN, and whether it can ameliorate podocyte damage and both functional and structural alterations of DN.

Conditionally immortalized human podocytes, generously provided by Prof. Saleem [49], were cultured and expanded at 33°C with 5% CO₂, in RPMI medium. The medium was supplemented with 10% foetal bovine serum (FBS), 2 mM L-glutamine, 100 U/mL penicillin/streptomycin, insulin, transferrin and sodium selenite. Experiments were conducted using podocytes that had been differentiated at 37°C for 15 days.

Primary murine podocytes were isolated from mice at 6 weeks of age. Briefly, mice were anaesthetized and perfused with a Hanks' Balanced Salt solution containing 8x10⁷ surface inactivated beads (Dynabeads, Thermo Fisher Scientific). Glomeruli were separated using a magnetic particle concentrator and seeded onto dishes coated with collagen type IV. Dishes were kept at 37°C in DMEM medium supplemented with 10% FBS, 100 mM HEPES, 100 U/ml penicillin/streptomycin, 1 mM sodium bicarbonate, and 1 mM sodium pyruvate. Subculture of primary podocytes was performed by detaching the glomerular cells using 0.25% trypsin-EDTA, then passing them through a 40-μm cell strainer (Falcon; BD Biosciences), and finally onto dishes coated with collagen type IV and used for experiments prior to the third passage.

Human MSCs (Lonza) were cultured in MSC basal medium with mesenchymal stromal cell growth supplement, 2 mM L-glutamine, 100 U/mL penicillin/streptomycin, and gentamicin sulfate-amphotericin. Cells were used for experiments until the sixth passage.

Bone marrow-derived murine MSCs (BM-MSCs) were obtained from 6-week old wild-type and M-Sec knockout (KO) mice as well as from 14-week old diabetic (DM) and non-diabetic (ND) mice. Briefly, BM cells were flushed from the tibial and femoral cavities in sterile conditions and seeded at 37°C for 7–10 days in MesenCult™ Expansion Medium with MesenPure™ (Stemcell Technologies), 2 mM L-glutamine, and 100 U/mL penicillin/streptomycin. Cells were subcultured until passage three and then characterised by flow cytometry using the following markers: FITC-CD44, FITC-SCA-1, FITC-CD49f, PE-CD45 (Thermo Fisher Scientific). To assess the ability of MSCs to differentiate into adipocytes, cells were cultured in MesenCult™ Adipogenic Differentiation Medium (Thermo Fisher Scientific) and then stained with Oil Red O.

M-Sec knock-down. To silence M-Sec, human MSCs were transfected (Lipofectamine 3000 Transfection Reagent, Thermo Fisher Scientific) with a plasmid construct encoding M-Sec-specific sh-RNA (5′-GATCCGACTGCTGGAGGCCACATTCCTGT-3′) or a mock plasmid (5′-GCACTACCAGAGCTAACTCAGATAGTACT-3′) cloned in a pRS vector (ExactHuSH, OriGene Technologies Inc).

Miro1 overexpression. BM-MSCs were transfected with a human adenovirus (Vector Biolabs, Malvern, PA) encoding Miro1 (98% homologous to the murine gene) at a standardised multiplicity of infection (MOI) of 100. An empty vector was used as a control for transduction. Experiments were performed 6 h after adenoviral transduction at 37C°.

Mitochondria-tracking. For mitochondrial tracking in vitro and in vivo, donor MSCs were transfected with a modified insect virus (Baculovirus), expressing a fusion construct of a red/green fluorescent protein with mitochondrial E1a pyruvate dehydrogenase (Cell Light™ Mitochondria-RFP/GFP BacMam 2.0, Thermo Fisher Scientific).

Podocytes were studied either in monocultures or in co-cultures with MSCs at a ratio of 1:1 in 35-mm μ-dishes (Ibidi, 80,136; final concentration 60,000 cells per dish) for 24 hours.

Cell tracking. CellTracker Blue and CellTracker Green CMFDA (Thermo Fisher Scientific) were used for cell tracking in co-cultures.

Fluorescence-Activated Cell Sorting (FACS). Podocytes (labelled with CellTracker Green) and MSCs were co-cultured at a ratio of 1:1 in 90-mm dishes (final concentration 2x10⁶ cells per dish) for 24 h. Cells were then detached using a non-enzymatic cell dissociation solution and washed in PBS. Green cells were sorted using a 100-µm microfluidic sorting chip on a SH800 Cell Sorter (Sony Corporation) equipped with a 525/50 optical filter. The viability of the cells was determined 24h post sort by microscopy.

To visualise of TNTs, cells were stained with Alexa Fluor®-488-wheat germ agglutinin (WGA; Thermo Fisher Scientific) and imaged using an Apotome 2 Zeiss Microscope, controlled by the Axiovision software and equipped with both an incubator for live cell imaging microscopy and Nomarski optics for differential interference contrast (DIC) microscopy. Sequences of optical planes (Z-stack) with a thickness of 0.20–0.25 µm were acquired for reconstruction in the x–z plane. To quantify TNTs, the number of podocytes connected to one or more MSCs by straight WGA-labelled structures, which were not adhered to the substrate, as determined by Z-stack analysis, and had a diameter thinner than 1 µm was blindly counted. Results were expressed as the percentage of total counted podocytes (minimum 150 podocytes). To visualised the F-actin cytoskeleton, cells were fixed with paraformaldehyde, which was added gently along the sides of the dishes to ensure that the culture medium remained undisturbed (final concentration 4% w/v) and then stained using Alexa Fluor® 488-phalloidin (Thermo Fisher Scientific).

Mitochondria Transfer-Flow cytometry. Transfer was assessed by both live fluorescent microscopy and flow cytometry. For transfer quantification, cells were fixed in 2% w:v PFA, passed through a cell strainer, and analysed by flow cytometry (Citoflex, Beckman Coulter). The instrument was calibrated using single-positive cells (either donor or recipient cells alone). Data analysis was carried out using the Cytexpert software (Version 2.3) and expressed as percentage of dually-labelled podocytes.

Mitochondrial Membrane Potential (MMP) was evaluated using the fluorescent probe JC-1 (Thermo Fisher Scientific) in live fluorescent microscopy in blind. Results were expressed as red to green JC-1 fluorescence intensity of podocytes (at least 150 podocytes).

Mitochondrial Superoxide Production was evaluated using the mitochondrial superoxide probe MitoSOX™ Red (Thermo Fisher Scientific) in live fluorescent microscopy in blind. Results were expressed as the percentage of red fluorescence per podocyte area (at least 150 podocytes).

Mitochondrial bioenergetics. Real-time measurement of oxygen consumption rate (OCR) was carried out using the Seahorse Bioscience XF Cell Mito Stress Test assay kit on the XFe24 Extracellular Flux Analyzer (Agilent Technologies). Cells were seeded at a density of 12,000 cells per well in 24-well assay plates and cultured in standard medium. After 24 h, the medium was replaced with XF Base Medium supplemented with 10 mM glucose, 2 mM L-glutamine, and 1 mM pyruvic acid. OCR was measured at baseline and after sequential addition of 1 μM oligomycin, 1 μM carbonyl cyanide-4- (trifluoromethoxy)-phenylhydrazone (FCCP), and 0.5 μM rotenone/antimycin A. Data were analysed using the WAVE software (Agilent Technologies) and normalized to total protein concentration. The metabolic parameters were calculated using the XFMito Stress Test Report Generator (Agilent Technologies).

Apoptosis was detected by transferase-mediated dUTP nick end-labelling (TUNEL) assay (ApopTag® Red In Situ Apoptosis Detection Kit) and results expressed as the percentage of apoptotic podocytes (at least 150 podocytes).

Animals. Animal studies was approved by the Ethical Committee and both housing and care of laboratory animals were in accordance with Italian law. Animals were housed at the Department of Drug Science and Technology, University of Turin, Italy. Male C57BL6 mice (Charles River Laboratories) and M-Sec-KO C57BL6 mice, generated by Prof. Ohno H, were used for the study. As previously reported [43], M-Sec KO animals exhibited normal vitality, fertility, and did not display any apparent phenotype. Diabetes induction. Diabetes (DM) was induced in eight-week old C57BL6 mice by intraperitoneal injection of streptozotocin (STZ) dissolved in citrate buffer (pH 4,5) at a dosage of 55 mg/kg body weight/day, delivered in 5 consecutive days. Non-diabetic (ND) animals received a citrate buffer alone. The onset of diabetes was confirmed 4 weeks after the first injection of STZ by blood glucose ≥250 mg/dl. MSC treatment. After eight weeks of diabetes, both diabetic mice with established albuminuria (DN) and ND animals (n=5 per group) were injected once a week for 6 consecutive weeks, via tail vein, with either vehicle or BM–MSCs (1.0×10⁴ cells/g body weight). BM–MSCs were obtained from both WT (MSC^WT) and M-Sec KO (MSC^KO) C57BL6 mice. Clinical parameters. Prior to killing, blood samples were collected from alert, 4-h–fasted animals via saphenous vein puncture. Glucose levels were measured using a glucometer (Accu-chek; Roche). Quantitative immunoturbidimetric latex (Sentinel Diagnostic) was used to determine glycated haemoglobin levels. Systolic blood pressure (SBP) was measured by tail-cuff plethysmography. Urine was collected over an 18-h period, while each mouse was individually housed in a metabolic cage with free access to food and water. Enzyme-linked immunosorbent assay (ELISA; Bethyl Laboratories) was used to measure urinary albumin concentration. Creatinine clearance was estimated based on the concentration of creatinine in both serum and urine as assessed by high-performance liquid chromatography (HPLC), in accordance with the Animal Models of Diabetic Complications Consortium (AMDCC) guidelines. Albuminuria was expressed as albumin excretion rate (AER, μg/18 h) or albumin-to-creatinine ratio (ACR, μg/mg). Tissue processing. After 14 weeks of diabetes, the animals were anaesthetized by using pentobarbital and euthanized through decapitation. The kidneys were quickly dissected, weighed, and processed for subsequent analyses. One kidney was frozen in N₂ and stored at −80°C for mRNA and protein analysis. The second kidney was divided into two halves: one half was fixed in formalin and embedded in paraffin for light microscopy, while the other half was embedded in optimal cutting temperature (OCT) compound and snap-frozen in liquid nitrogen. PAS staining. Paraffin-embedded renal sections were stained using the periodic acid–Schiff. Mesangial area was analysed (percentage of glomerular area) from digital pictures of 15–20 glomeruli per kidney per animal using the Axiovision 4.7 software.

Renal cortex pieces (1 mm³) were fixed in 2% glutaraldehyde, 4% paraformaldehyde in phosphate buffer 0.12 mol/l for 4 h at room temperature, postfixed in 1% osmium tetroxide for 2 h, dehydrated in graded ethanol, and embedded in Epon 812. Ultrathin sections were counterstained with uranyl acetate and lead citrate, and examined with an Energy Filter Transmission Electron Microscope (EFTEM, ZEISS LIBRA® 120) equipped with an yttrium aluminium garnet (YAG) scintillator slow-scan charge-coupled device (CCD) camera (Sharp eye, TRS, Moorenweis, Germany).

Total RNA was extracted using the TRIzol reagent (Thermo Fisher Scientific). After quantification, 1 µg of total RNA was reverse transcribed in cDNA (High-Capacity Reverse Transcription kit, Thermo Fisher Scientific). mRNA expression was analysed by real-time PCR using pre-developed TaqMan reagents (Supplementary Table S1).

Immunohistochemistry (IHC) was performed in 4 µm paraffin kidney sections of formalin-fixed tissue. Antigen retrieval was carried out using a citrate buffer (0.01 M, pH 6.0). To neutralise endogenous peroxidase activity, sections were exposed to 3% H₂O₂. Endogenous avidin-binding sites were blocked by incubation with 0.1% avidin and 0.01% biotin. Subsequently, the slides were incubated with 3% BSA in phosphate buffer saline (PBS) for 30 min. Overnight incubation at 4°C was performed with a primary rat antibody anti-MAC2 (Cedarlane; CL8942AP). After washing with PBS, sections were exposed to a secondary biotinylated goat anti-rat antibody (Jackson ImmunoResearch Laboratories; 112–065-003) for 1 h, followed by incubation with horseradish peroxidase (HRP)-conjugated streptavidin (DAKO, P0397) for 1 h. Finally, sections were mounted and 3,3′- diaminobenzidine (DAB) used as the chromogen substrate for HRP. A negative control was included in which the primary antibody was preincubated with a control peptide. Section visualisation was performed using an Olympus-BX4I microscope and images were captured and digitised with a high-resolution camera (Carl Zeiss). On average 20 randomly selected glomeruli were assessed per mouse, and the results were expressed as the number of MAC-2 positive cells per glomerular area.

Immunofluorescence (IF). Snap-frozen kidney sections (3 µm) were fixed in cold acetone for 5 min and blocked in 3% BSA in PBS. Slides were then incubated overnight at 4°C with guinea pig anti-nephrin (Progen Biotechnik, GP-N2), rabbit anti-podocin (Merck Life Science, P0372), rabbit anti-fibronectin (Merck Life Science, F3648) primary antibodies. After washing with PBS, fluorescein isothiocyanate-conjugated (FITC) swine anti-rabbit (DAKO, F0205), Alexa Fluor 568 goat anti guinea pig (Thermo Fisher Scientific, A11075) secondary antibodies were added for 1 h. After the incubation, sections were mounted and then examined using a APOTOME 2 microscope (Carl Zeiss). On average 20 randomly selected glomeruli were assessed per mouse. Results were expressed as the percentage area of positive staining per glomerulus. For double IF, renal cortex sections obtained from ND^WT, DM^WT, and DM^KO mice were incubated overnight at 4°C with a rabbit anti-GFP (Thermo Fisher Scientific; A11122). After washing with PBS, sections were incubated with a fluorescein isothiocyanate-conjugated (FITC) swine anti-rabbit (DAKO, F0205) for 1 h. After further blocking in 3% BSA, sections were incubated with guinea pig anti-nephrin (Progen Biotechnik, GP-N2) overnight at 4°C, followed by incubation with an Alexa Fluor 568 goat-anti guinea pig secondary antibody for 1 h. Digitised images were colour-combined and assembled into photomontages by using Adobe Photoshop (Universal Imaging Corporation, West Chester, PA).

Immunoblotting. Cells were homogenised in a RIPA buffer containing 0.5% NP40 (vol./vol.), 0.5% sodium deoxycholate (wt/vol.), 0.1% SDS (wt/vol.), 10 mmol/L EDTA, and protease inhibitors. Protein extracts were obtained by centrifugation at 14,000×g for 20 min at 4°C, following a 45-min incubation on ice. The total protein concentration was determined using the DC Protein Assay Kit (Bio-Rad, Milan, Italy). Equal amounts of protein samples were separated on SDS polyacrylamide gel electrophoresis and electro-transferred onto nitrocellulose membranes. After blocking in 5% non-fat milk in TBS (pH 7.6), the membranes were incubated overnight at 4°C with a primary antibody targeting either M-Sec (Abcam, ab196659), Miro1 (Abcam, ab188029), active caspase-3 (Novus Biological, NB110-55658). Subsequently, the membranes were washed, and secondary HRP-linked antibodies (GE Healthcare, RPN4301; Santa Cruz Biotechnology, sc-2005) added for 1 h. Detection was performed using SuperSignal PICO (Euroclone, Milan, Italy) and visualised on a Gel-Doc system (Bio-Rad, Milan, Italy). Band intensities were quantified by densitometry, with tubulin serving as the internal control.

The results were presented as means ± SEM, geometric mean (25^th–75^th percentile), or fold change over control. Non-normally distributed variables were log-transformed prior to the analyses. The Student's t-test or ANOVA was used for data analysis, as appropriate. Post-hoc comparisons were performed using the Least Significant Difference test. Statistical significance was defined as P<0.05. The number of independent experiments is reported in the figures to legends.

To assess whether MSCs can form heterotypic TNTs with podocytes, cultured human podocytes were exposed to serum deprivation (0.1% FCS), a well-known TNT inducer, for 24 h and then incubated with human MSCs. Live cell fluorescence microscopy analysis showed bridge-like structures interconnecting MSCs and podocytes, displaying the specific features of TNTs. Specifically, they were straight, with a width of less than 1 µm and a length spanning several cell diameters (Figure 1A). Serial Z-stack images proved that these structures did not adhere to the substrate, a characteristic specific to TNTs that distinguishes them from filopodia (Figure 1A z-stack and z-stack depth-colour coding). Fluorescent staining for F-actin showed the presence of an actin backbone within TNTs (Figure 1B). After fixation, co-cultures were immunostained for nephrin to confirm podocyte identity (Figure 1C).

To clarify whether diabetes-related insults can induce formation of heterotypic TNTs, human podocytes were exposed to high glucose (HG, 25 mM, 48 h), AGE-Bovine Serum Albumin (AGEs, 100 µg/ml, 72 h), or Monocyte Chemoattractant Protein-1 (MCP-1, 10 ng/ml 24 h) and then co-cultured with human MSCs for 24 h. Normal glucose concentrations (NG, 10 mM made iso-osmolar with mannitol) and vehicles were used as controls. As shown in the images in Figure 2A–C, membrane channels between podocytes and MSCs, which fulfilled the criteria for defining TNTs described above, were observed. To clarify the importance of M-Sec in TNT formation, we assessed whether MSCs expressed M-Sec by immunoblotting. As shown in Figure 2D, MSCs constitutively expressed M-Sec. Then we generated MSCs lacking M-Sec by transfecting them with an M-Sec-specific sh-RNA (MSC^shMSec) or a mock plasmid (MSC^mock) and confirmed knockout efficiency by immunoblotting (Figure 2E). As shown in the graph (Figure 2F), pre-exposure of podocytes to HG, AGEs, and MCP-1 significantly increased the number of podocytes connected to MSCs through TNTs compared with controls, but this effect was not observed when podocytes were co-cultured with MSCs lacking M-Sec (MSC^shMSec), proving that diabetes-induced heterotypic TNT formation requires M-Sec expression in MSCs.

Human podocytes were exposed to HG, AGEs, and MCP-1, as described above. Rotenone (10 nM), which induces mitochondrial dysfunction was used as a positive control. Subsequently, Cell Tracker Blue-labelled podocytes were co-cultured with human MSCs, carrying Cell-Light Mito-RFP-labelled mitochondria. After 24 h, we observed red fluorescent mitochondria along TNTs and within the cytosol of recipient podocytes (Figure 3A–D). To quantify mitochondrial transfer, the percentage of blue podocytes containing MSC-derived red mitochondria (dually labelled) was assessed by flow cytometry. As shown in the graph (Figure 3E), mitochondrial transfer from MSCs to control podocytes was modest, but it was significantly increased by pre-exposure of podocytes to HG, AGEs, and MCP-1. Importantly, the transfer was almost abolished when podocytes were incubated with MSCs lacking M-Sec (MSC^shMSec) (Figure 3E) or when experiments were performed in the presence of latrunculin-B (200 nM), which inhibits TNT formation (Figure 3F). Taken together, these data show that exposure of podocytes to diabetes-related insults enhances the transfer of mitochondria from donor MSCs to damaged podocytes in a M-Sec-TNT-dependent manner.

To explore the functional relevance of mitochondrial transfer, human podocytes were exposed to AGEs or vehicle for 72 h, and then incubated for a further 24 h with MSCs or vehicle (see experimental design in Figure 4A). Exposure of podocytes to AGEs induced the transition of the fluorescent probe JC-1 from red to green fluorescence, indicating a decrease in Mitochondrial Membrane Potential (MMP) and hence mitochondrial damage (Figure 4B). AGEs also increased mitochondrial oxidative stress, as assessed using the mitochondrial superoxide probe Mitosox Red (Figure 4C). Importantly, incubation of AGE-treated podocytes with MSCs expressing M-Sec (MSC^Mock) abolished both MMP loss and mitochondrial oxidative stress, while incubation with MSCs either lacking M-Sec (MSC^shMSec) or pre-exposed to rotenone (MSC^Rot) to damage donor mitochondria was ineffective (Figure 4B,C). To obtain quantitative data, podocytes were separated from the mono/cocultured system by FACS sorting and podocyte bioenergetics was studied using the Cell Mito Stress Test on a Seahorse Analyser.

Exposure to AGEs reduced basal and maximal oxygen consumption rate (OCR), ATP-linked respiration, and reserve capacity, indicating alterations in mitochondria-dependent cellular bioenergetics (Figure 4D). Moreover, podocyte expression of the mitochondrial genes ND4L and COX-1 was modestly but significantly down-regulated (Figure 4E,F) as assessed by real-time PCR. These AGE-induced effects were ameliorated by incubation with MSC^Mock, but left unchanged by incubation with either MSC^shMSec or MSC^Rot (Figure 4D–F). Therefore, donor MSCs can improve mitochondrial parameters in recipient podocytes only if they express M-Sec and carry healthy mitochondria.

We then investigated the impact of mitochondrial transfer on the phenotype of diabetes-injured podocytes. As expected, exposing human podocytes to AGEs induced nephrin down-regulation (Figure 4G) and apoptosis, as assessed by both immunoblotting for active caspase 3 on FACS-sorted podocytes (Figure 4H) and in situ TUNEL assay (Figure 4I). Coculturing of AGE-treated podocytes with MSCs reduced apoptosis and increased nephrin gene expression even above control levels. However, MSCs lacking M-Sec (MSC^shMSec) or treated with rotenone (MSC^Rot) failed to improve nephrin down-regulation and apoptosis (Figure 4G-I). Taken together, these results show that the transfer of healthy mitochondria from MSCs to podocytes protects diabetes-injured podocytes via an M-Sec-dependent mechanism.

To explore whether diabetes can affect the efficiency of MSCs as mitochondrial donors, MSCs were isolated from the bone marrow of diabetic (MSC-DM) and non-diabetic (MSC-ND) mice and characterized. MSCs were then incubated with AGEs/vehicle-treated murine podocytes (Figure 5 A–C). Exposure of podocytes to AGEs increased the percentage of podocytes connected via TNTs to MSCs without differences between MSC-ND and MSC-DM in their ability to form TNTs (Figure 5D). However, differences were seen in mitochondrial transfer (Figure 5E). The percentage of blue-labelled podocytes containing red mitochondria derived from non-diabetic MSCs was significantly increased in AGE-treated podocytes compared with vehicle-treated podocytes, indicating enhanced mitochondrial transfer. By contrast, this increase was not observed when AGE-treated odocytes were incubated with MSCs from diabetic mice. Collectively, these findings demonstrate that donor MSCs from diabetic mice retain the capability to form TNTs with injured podocytes, but they fail to effectively transfer mitochondria to recipient podocytes.

To elucidate the underlying mechanism, we focussed on Miro1, which localizes to the mitochondrial outer membrane and anchors mitochondria to the actin/tubulin cytoskeleton. Miro1 is of relevance in this context because previous studies showed that Miro1 is essential for the movement of mitochondria along TNTs rather than for their formation [41,42]. Both Miro1 mRNA and protein levels were significantly reduced in MSC-DM compared with MSC-ND (Figure 5F,G). Furthermore, exposure of MSC-ND to either AGEs or MCP-1 significantly decreased Miro1 mRNA expression (Figure 5H), providing a potential mechanism for diabetes-induced Miro1 down-regulation.

To functionally link Miro1 down-regulation to impaired mitochondrial transfer, MSC-DM were transfected with either an adenovirus expressing Miro1 (DM-Miro) or a mock vector (DM-Mock). As shown in Figure 5I, MSC-DM-Miro1 expressed levels of Miro1 mRNA comparable to those of MSC-ND, proving the efficacy of transfection. Re-expression of Miro1 ameliorated mitochondrial transfer from MSC-DM to AGEs-treated podocytes (Figure 5J), confirming that down-regulation of Miro1 can at least partially explain the lower efficiency of diabetic MSCs in transferring mitochondria to podocytes.

Consistent with this, coincubation of MSCs from diabetic mice with AGE-treated podocytes failed to improve MMP (Figure 6A), mitochondrial oxidative stress (Figure 6B), mitochondrial bioenergetics (Figure 6C), expression of ND4L, COX-1, Nephrin (Figure 6D–F), and apoptosis (Figure 6G–I). However, re-expression of Miro1 in MSC-DM partially restored their ability to rescue these podocyte abnormalities (Figure 6A–I).

The in vivo relevance of these findings was explored in the animal model of STZ-induced diabetes with established albuminuria. Eight weeks after diabetes induction, levels of albuminuria were significantly higher in DM mice compared with ND mice, indicating the development of glomerular injury [ND: 17.24 (15.4–18.7) vs. DM: 35.73 (28.40–45.51); P<0.01]. Both DN and control animals were then treated for six weeks with either vehicle or exogenous MSCs, which were isolated from WT (MSC^WT) and M-Sec KO (MSC^KO) mice. Mice with similar levels of albuminuria were assigned to the different treatment groups. Table 1 displays the metabolic and physiological parameters evaluated at the conclusion of treatment. DN mice exhibited significantly higher blood glucose and glycated haemoglobin levels, as well as a lower body weight. Treatment with either MSC^WT or MSC^KO did not alter these parameters. DN mice showed a six-fold increase in ACR compared with ND animals. Treatment with MSC^WT significantly reduced albuminuria, while treatment with MSC^KO was ineffective. The decline in renal function induced by diabetes was abolished by MSC^WT treatment, but it was left unchanged by MSC^KO. Therefore, expression of M-Sec is required for exogenous MSCs to improve functional alterations of DN.

Diabetes induced the down-regulation of both nephrin and podocin mRNA and protein expression. This effect was mitigated by treatment with MSC^WT, while treatment with MSC^KO had no effect (Figure 7A–E). Similarly, PAS staining revealed that diabetes-induced mesangial expansion was ameliorated by treatment with MSC^WT but not with MSC^KO (Figure 7F,G). Consistently, the overexpression of both fibronectin and collagen was reduced only in DN mice receiving MSC^WT treatment (Figure 7H–K).

The number of macrophages (MAC-2⁺ cells) within the glomeruli was significantly higher in DN mice compared with ND mice, and this effect was alleviated, though not abolished, by both MSC^WT and MSC^KO treatments (Figure 8A,B). Moreover, diabetes induced overexpression of both the chemokine MCP-1 and its cognate receptor CCR2, and these effects were abolished by both MSC^WT and MSC^KO therapy (Figure 8C,D). Therefore, in contrast with other functional and structural alterations caused by diabetes, the protective effect of exogenous MSCs on glomerular inflammation was independent of M-Sec expression in MSCs.

Given the crucial role of the M-Sec-TNT system in mitochondrial transfer, we next investigated whether treatment with exogenous MSC^WT and MSC^KO affected diabetes-induced mitochondrial abnormalities in vivo. Diabetes decreased the expression of TFAM, a mitochondrial transcription factor, COX1, and ND4L, while mitochondria ultrastructure, as well as the expression of genes regulating mitochondrial biogenesis, fusion, fission, and mitophagy, remained unaltered (Figure 9A–C, Supplementary Figure S1 and S2). Importantly, treatment with MSC^WT either improved or reversed the down-regulation of TFAM, COX1, and ND4L, while injection of MSC^KO was ineffective (Figure 9A–C).

Finally, we tested whether exogenous MSCs, carrying CellLight Mitochondria-GFP, could transfer mitochondria to podocytes in vivo. Renal cortex sections were immunostained for nephrin and GFP. As expected, GFP-mitochondria were not detected within the glomeruli of ND mice treated with MSC^WT. By contrast, fluorescent-tagged mitochondria were seen in the glomeruli of DN-MSC^WT mice, and some puncta localised to podocytes as shown by costaining for nephrin in the merged image. However, this was rarely observed in DN mice treated with DM-MSC^KO, suggesting that M-Sec expression in donor MSCs favours mitochondrial transfer to podocytes (Figure 9D).

Our study demonstrates that M-Sec-TNT-mediated horizontal transfer of mitochondria from healthy MSCs to diabetes-injured podocytes ameliorates podocyte damage. Additionally, the expression of M-Sec in exogenous MSCs is essential for providing renoprotection in experimental DN.

TNTs are membrane channels that bridge cells and allow the exchange of various cargos, including organelles, between cells. Homotypic TNTs have been described in many cell types, including podocytes [43,44]. However, cells can also form heterotypic TNTs with other cell types, such as stromal cells, macrophages, and progenitors/stem cells [50–53]. Our study provides the first evidence that podocytes, under diabetes-related conditions, can establish heterotypic connections with MSCs that fulfil the current criteria for TNTs. Similarly, MSCs have been found to form heterotypic TNTs with corneal epithelial cells, airway epithelial cells, brain epithelial cells, pulmonary alveolar cells, cardiomyocytes, macrophages, and endothelial cells [40]. The difference between homotypic and heterotypic TNTs is unclear. However, homotypic TNTs create a supercellular system that probably plays a role in diluting stresses, reducing the burden of damaged organelles, and maintaining homeostasis. On the other hand, heterotypic TNTs are primarily important in cell-to-cell cooperation and cell repair/differentiation, particularly when precursor/stem cells are one component of the pair.

The mechanism underlying the formation of TNTs is not fully understood. However, it is known that both podocytes and MSCs express M-Sec, and in our study M-Sec silencing in MSCs resulted in the abolishment of TNT formation. This indicates that the presence of M-Sec on both MSCs and podocytes is essential for the development of TNTs. Exposure to high glucose is known to induce M-Sec overexpression in podocytes [43] and this probably favoured TNT development in our coculture system. M-Sec cooperates with the small GTPase RalA and the exocyst complex to initiate F-actin polymerization and TNT development [24]. M-Sec is required for the formation of TNTs among podocytes and enables MSCs to form TNTs with cardiomyocytes and lung epithelial cells in a manner dependent on M-Sec [42]. However, the reliance on M-Sec for TNT formation appears to vary across cell types. Unlike podocytes and MSCs, other cell types such as renal tubular epithelial cells [43] and neurons [54] form TNTs independently of M-Sec.

TNTs can transfer various cellular components, and in our study, we specifically focused on mitochondria due to their emerging role in the pathogenesis of DN. In keeping with this, we observed mitochondrial abnormalities in podocytes subjected to diabetes-related insults, as well as in mice with DN, such as TFAM deficiency, impaired mitochondrial respiration, down-regulation of mtDNA genes, decreased MMP, and increased mitochondrial oxidative stress.

Our results show that TNTs interconnecting MSCs and podocytes are functionally active and can transfer mitochondria. Although mitochondria are larger than TNTs, mitochondria can enlarge TNTs, creating a bulge, and this makes progression of mitochondria along TNTs possible. To track the mitochondria, donor MSCs were transfected with a modified insect virus (Baculovirus), expressing a fusion construct of a red fluorescent protein with mitochondrial E1a pyruvate dehydrogenase, instead of being stained with a fluorescent mitochondria dye. This eliminates the possibility of false-positive staining caused by fluorescent dye leakage from MSCs into the medium and subsequent uptake by podocytes.

The efficiency of mitochondrial transfer was enhanced when podocytes were pre-exposed to either diabetes-related conditions or rotenone, suggesting that the transfer was intended to rescue podocytes that had dysfunctional mitochondria by delivering healthy organelles. Accordingly, the transfer of MSC-derived mitochondria to AGE-treated podocytes improved MMP and bioenergetics, abolished mitochondrial oxidative stress, and restored or even increased expression of mitochondrial genes. Previous studies reported that exposure of podocytes to AGEs also induces markers of senescence [55,56] and it would be valuable for future research to explore whether mitochondrial transfer from MSCs could also provide protection against podocyte senescence. Mitochondrial transfer via TNTs has been proven in several cell types [14,16,29,57–59] and even shown to restore aerobic respiration in cells lacking mitochondria [60–62]. Although TNTs are generally considered as the prevalent mechanism of intercellular mitochondrial transfer, other mechanisms have also been described, such as extracellular vesicles, fusion, and extrusion/internalisation [63]. However, abrogation of mitochondrial transfer by both latruculin and M-Sec deletion confirmed that the TNT-M-Sec system was the predominant mechanism of mitochondrial exchange between MSCs and podocytes. Mitochondrial replacement is of particular relevance in cells, as podocytes, that are terminally differentiated as mitochondrial damage tends to accumulate in cells that cannot replicate.

Following TNT-mediated transfer, the mitochondria in recipient podocytes were derived from both the host (podocytes) and the donor (MSCs), thus they originated from different human subjects. Mixing donor and host mitochondria probably increased heteroplasmy (variants present only in a proportion of mtDNA copies rather than in all copies), while reducing the variant load. This might have contributed to the observed benefit. However, the final effect of mixing mtDNA of different origins is unpredictable. For potential future clinical applications, sequencing the mtDNA of MSCs prior to injection to detect and quantify somatic mtDNA variants would be a safer approach.

Both loss of slit diaphragm proteins and podocyte apoptosis are markers of podocyte injury and well-established features of DN. Notably, M-Sec-dependent mitochondrial transfer from MSCs to AGE-treated podocytes reduced podocyte apoptosis and even enhanced nephrin expression, providing evidence of cytoprotection. Besides mitochondria, TNTs can also transfer other compounds and organelles, such as mRNA, proteins, lipid droplets, lysosomes, and endosomes [64–67]. Therefore, it is possible that the transfer of other cargo molecules also played a role in rescuing podocytes. However, donor MSCs carrying mitochondria that were damaged with rotenone failed to improve mitochondrial function, nephrin expression, and apoptosis, indicating that MSCs protected podocytes predominantly through the transfer of healthy mitochondria.

In DN, all cells, including MSCs, are subjected to the harmful effects of diabetes, which can impair the cytoprotective functions of MSCs. Consistent with this, MSCs isolated from diabetic mice formed TNTs with podocytes, but their ability to transfer mitochondria was significantly diminished, reducing their effectiveness in rescuing mitochondrial function, nephrin expression, and preventing apoptosis. This inability to transfer mitochondria was due to the down-regulation of Miro1. Indeed, restoring Miro1 expression re-established the efficiency of MSCs as mitochondrial donors and their protective effect on podocytes. Miro-1 is a transmembrane protein located in the outer mitochondrial membrane that interacts with molecular motors, such as kinesin, dynein, and Myo19, which drive the movement of mitochondria along microtubules and/or actin filaments. TNTs contain an actin and/or microtubules cytoskeleton, which serves as “track” for mitochondrial movement and a large body of evidence proves that Miro1 is essential in mitochondrial movement along TNTs [42,68–74]. Furthermore, higher expression of Miro1 was found responsible for the superior efficacy of iPSC-MSCs compared with MSCs in mitochondria transfer [75]. Extracellular vesicles and mitochondria-derived vesicles (MDVs) can also transfer mitochondrial components or whole mitochondria [76,77] to distant recipient cells and Miro proteins have been involved in MDV formation [78]. Therefore, it is possible that the down-regulation of Miro1 in diabetic MSCs may also affect other mechanisms of horizontal mitochondrial exchange. Regardless of the mechanism, our results provide the first evidence that diabetes-related insults down-regulate Miro1 in MSCs, thereby limiting their ability to transfer mitochondria. This finding may have clinical relevance, as autologous non-immunogenic MSCs from patients with diabetes could be engineered to overexpress Miro1 for cell-based therapy.

In the in vivo part of our study, the administration of exogenous MSCs to DN mice ameliorated markers of mitochondrial dysfunction, preserved both nephrin and podocin expression, reduced mesangial expansion, improved albuminuria, and maintained renal function. Importantly, these beneficial effects were observed without any changes in blood glucose levels or glycated haemoglobin. Several studies have reported that treatment with MSCs can improve hyperglycemia in animal models of STZ-induced diabetes [79–82] by promoting the repair of β-cells or mitigating their destruction. However, data on this topic are conflicting [83], and efficacy of MSCs depends on factors such as the age of the animals and the timing of MSC treatment after diabetes induction [79]. The relative advanced age of our animals and the delayed initiation of MSC treatment after DM induction may explain why we did not observe amelioration of hyperglycemia.

Several studies have previously reported that exogenous MSC treatment has a protective effect on both functional and structural alterations of DN in experimental models [83–88]. The release of extracellular vesicles and/or soluble factors by MSCs was proposed as the underlying mechanism. Our aim was to demonstrate that the benefit of MSC treatment was at least partially mediated by the M-Sec-TNT system. Treatment with exogenous MSCs lacking M-Sec did not ameliorate slit diaphragm protein down-regulation, mesangial expansion, albuminuria, or renal function and this supports the hypothesis of an important role of M-Sec in the protective action of exogenous MSCs. Additionally, MSC-derived mitochondria were observed in the glomerular podocytes of diabetic mice, suggesting in vivo mitochondrial transfer. Importantly, this phenomenon was drastically reduced when the inoculated MSCs did not express M-Sec. A previous study reported mitochondria transfer from exogenous MSCs to renal tubular cells in the context of DN [89], but the mechanism was not investigated. Furthermore, tubular epithelial cells do not express M-Sec and likely other mechanisms were involved. Further studies are, however, required to demonstrate the formation of TNTs in vivo between exogenous MSCs and podocytes and to confirm in vivo TNT-mediated mitochondrial exchange.

Clinical benefit was observed despite most intravenously injected MSCs becoming trapped in the lungs, with only a fraction reaching the kidneys [90]. The higher injection frequency in our study, compared with previous studies [86], may have led to a greater overall number of MSCs reaching the podocytes. Furthermore, even if only a limited number of MSCs reach the kidneys, their ability to deliver mitochondria to damaged podocytes via TNTs may exert a small yet powerful effect with substantial therapeutic impact. Additionally, using freshly isolated MSCs may have further enhanced their mitochondrial donation efficacy, thus increasing therapeutic potential. Finally, although MSC engraftment in the kidney is generally limited, mitochondrial transfer via TNTs requires only transient contact with target podocytes rather than permanent engraftment.

Inflammation was the only diabetes-induced abnormality improved by MSCs therapy independently of M-Sec. MSCs are known to release factors, such as interleukin-1 receptor antagonist (IL1-RA), Prostaglandin E2 (PGE2), Tumour necrosis factor- (TNF) stimulated gene-6 (TSG6), which inhibit the recruitment of inflammatory monocytes and promote the polarisation of macrophages toward the M2 phenotype [91]. This may explain the anti-inflammatory effect of MSCs. However, we cannot exclude the possibility that MSCs can also transfer mitochondria to inflammatory cells through a TNT-independent mechanism [92]. Supporting this, a recent study reported that mitochondrial transfer from MSCs to macrophages reduces inflammation and improves kidney injury in an animal model of DN, although the underlying mechanism of mitochondrial transfer was not explored [93].

No obvious side effects were observed in mice undergoing MSC therapy. This might be attributed to the use of syngeneic BM-MSCs, which were injected at low concentrations and after only three passages in culture. However, our histological and functional analyses were limited to the kidney, and the duration of treatment was relatively short. Therefore, further studies are needed to investigate both the beneficial and detrimental effects of long-term BM-MSC treatment. Additionally, caution is necessary when translating data obtained from experimental animals to humans, as animal models might not perfectly reflect human DN progression and renal damage is relatively mild in mice made diabetic with STZ.

In conclusion, our study provides evidence that MSCs have an important protective effect in both in vitro and in vivo models of DN via the M-Sec-TNT system. Given that podocyte injury is a key factor in various proteinuric glomerulopathies, our findings may have broader implications for other kidney diseases. The discovery that podocytes can be rescued through TNT-mediated horizontal transfer of mitochondria from MSCs may open possibilities for future research in the field of regenerative cell-based therapy.

• Mitochondrial dysfunction plays an important role in the development of podocyte injury in diabetic nephropathy and effective strategies for supplementing, repairing, and replacing damaged mitochondria are not yet available.

• Our present study demonstrates that MSCs have an important protective effect in DN by allowing replacement of dysfunctional mitochondria through the M-Sec-TNT system.

• Demonstrating that MSCs are able to replace dysfunctional podocyte mitochondria through horizontal TNT-mediated transfer may pave the way for regenerative cell-based therapy strategies.

Upon submission, authors agree to make any materials, data, and associated protocols available upon request.

The Authors declares that there are no competing interests associated with the manuscript.

The work was supported by the Ferrero Foundation (Alba), Turin University (ex. 60% grant), the European Foundation for the Study of Diabetes (EFSD), 1/2021 Novo Nordisk and the Italian Ministry for Education, University and Research (Ministero dell'Istruzione, dell'Università e della Ricerca - MIUR) under the program “Dipartimenti di Eccellenza 2018–2022” project D15D18000410001.

Federica Barutta: Conceptualization, Investigation, Methodology, Writing—original draft. Beatrice Corbetta: Conceptualization, Investigation, Methodology, Writing—original draft. Stefania Bellini: Data curation, Investigation. Roberto Gambino: Investigation. Stefania Bruno: Investigation. Shunsuke Kimura: Resources. Koji Hase: Resources. Hiroshi Ohno: Resources. Gabriella Gruden: Conceptualization, Supervision, Funding acquisition, Writing—review & editing.

ACR

albumin-to-creatinine ratio

AGEs

advanced glycation end products

CCR2

C-C chemokine receptor type 2

COX1

Cytochrome c oxidase subunit I

DAPI

4′, 6-diamidino-2-phenylindole

DIC

differential interference contrast

DM

diabetes

DN

diabetic nephropathy

FBS

foetal bovine serum

FITC

fluorescein isothiocyanate-conjugated

HG

high glucose

HRP

horseradish peroxidase

IF

immunofluorescence

MCP-1

monocyte chemoattractant protein-1

MMP

mitochondrial membrane potential

MSCs

mesenchymal stromal cells

ND

non-diabetic

ND4L

NADH dehydrogenase subunit 4L

NG

normal glucose

OCR

oxygen consumption rate

PBS

phosphate buffer saline

RFP

red Fluorescent Protein

SBP

systolic blood pressure

STZ

streptozotocin

TFA

transcription factor A mitochondrial

TNTs

tunnelling nanotubes

TUNEL

transferase-mediated dUTP nick end-labelling

WGA

wheat germ agglutinin