We selected 2-week-old male SD rats, isolated their femurs and tibias, washed the bone marrow cavities with a complete medium, pipetted the marrow into a cell suspension, and separated mononuclear cells using density gradient centrifugation with Ficoll-Paque PREMIUM 1.084 (Cytiva, Boston, MA, USA). Cells from the tibia and femur of one rat were inoculated into two T25 cell culture flasks. The isolated and purified cells were grown in DMEM/F-12 medium (Gibco, Grand Island, CA, USA) containing 20% fetal bovine serum (Gibco, CA, USA) and 1% penicillin-streptomycin solution (Gibco, CA, USA) in a humidified cell culture incubator at 37 °C with 5% CO${}_2$. After 24 hours, the non-adherent cells were aspirated, and the culture medium was replaced with a complete medium. Thereafter, the complete medium was replaced every 48 h. The primary cells were subcultured when they grew above 80%–90% confluency. Third-generation BMSCs were used for subsequent experiments.

In addition to the morphological observations under the light microscope, the P3 generation cells were induced to differentiate into adipogenic, osteogenic, and chondrogenic lines, and were subsequently characterized by oil red O staining, alizarin red staining, and allicin blue staining, respectively. The expression of cell surface markers CD29 (positive), CD44 (positive), and CD34 (negative) was detected by flow cytometry (Supplementary Fig. 1).

After the BMSCs grew to 50%–60% confluence, the original medium was removed, washed three times with PBS (Gibco, Grand Island, CA, USA), and replaced with DMEM/F-12 containing 5% exosome-free FBS (OriCell, Wuhan, China). The cell culture supernatant was collected after 48–72 h, and centrifuged at 500 $\times$g for 10 min, 2000 $\times$g for 15 min, and 10,000 $\times$g for 30 min at 4 °C. The obtained cell supernatant was transferred to an ultracentrifuge tube, centrifuged twice at 100,000 $\times$g for 70 min at 4 °C, and then resuspended in 50 uL PBS and stored at –80 °C. The above stock solution was diluted with 0.9% saline to a final concentration and sterilized with a 0.22 µm filter before injection.

To identify the extracted BMSC-Exo, transmission electron microscopy (TEM, Hitachi, Tokyo, Japan) was used to capture images of exosome morphology. Nanoparticle tracking technology (NTA, ParticleMetrix, Ludwigshafen, Germany) was applied to analyze the particle size distribution of exosomes, and western blotting was used to detect expressions of CD9 and CD81 on the exosomal surface.

5-week-old male SD rats (150–200 g) were purchased from HFK Bioscience Co, Ltd. (Beijing, China) and kept in the SPF animal laboratory while being fed a high-fat diet (HFD). After two weeks of acclimation, streptozotocin (STZ, 35 mg/kg/day, dissolved in 0.01 M citrate buffer, PH4.5) (Sigma, Saint. Louis, MO, USA) was intraperitoneally injected into the rats in the experimental group for three consecutive days, while the rats in the control group were injected with an equal volume of vehicle. To measure the postprandial blood glucose levels by a blood glucose meter (Roche, Basel, Switzerland), blood samples were collected each morning (8:00 AM) by acupuncturing the tail vein with a 28G needle. Rats with fasting glucose continuously above 11.1mM were considered to be successful diabetic models [17]. A total of 32 rats were divided into 4 groups and given different treatments up to 18 weeks. Excluding modeling failure and death, the remaining 20 rats were finally included in the follow-up experiments: the control group (NC group, n = 5), a diabetes group (DM group, n = 5), a metformin treatment group (Met group, n = 5), and an exosome treatment group (Exo group, n = 5). From the fourth week, the Met group was treated with 300 mg/kg/day of metformin by gavage, while the other three groups were given the same volume of pure water by gavage. The Exo group was given 100 µg (dissolved in 200 µL 0.9% sodium chloride solution) of exosomes through the tail vein every week, and the other three groups were given the same amount of normal saline at the same time points [17]. After 14 weeks of continuous administration, the rats were anesthetized with sodium pentobarbital (45 mg/kg body weight). Carotid blood was collected to assess blood glucose (GLU), urea nitrogen (BUN), and creatinine (Cr) levels. Each rat’s left kidney was fixed in a 4% paraformaldehyde solution overnight. The right kidney was kept in liquid nitrogen for qRT-PCR and Western blot detection. The rat grouping treatment strategy is shown in Fig. 1.

Fig. 1.

Preparation of rat diabetes model and group administration. HFD, high-fat diet; STZ, streptozotocin.

Blood samples obtained from rat carotid artery blood collection were allowed to stand at 37 °C to promote coagulation. After the blood was coagulated, it was centrifuged at 2500 rpm for 10 min, and the supernatant was aspirated. We used an Indiko automatic biochemical analyzer (Thermo Scientific, CA, USA) to measure GLU, BUN, and Cr concentration.

The kidneys fixed in 4% paraformaldehyde were clarified, dehydrated, embedded in paraffin, and cut into micron-thick sections. Sections were stained with hematoxylin-eosin (H&E) and Masson and Periodic Acid-Schiff (PAS) staining to assess the severity of diabetic kidney disease. Type IV collagen immunohistochemical staining was performed on the remaining sections. The sections were first incubated with Collagen IV polyclonal antibodies (1:200, Bioss, Beijing, China) at 4 °C overnight, washed with PBS for 5 min 3 times, and then incubated with biotin-labeled secondary antibodies for 1 hour at room temperature. The stained sections were scanned with an Aperio Scan Scope AT Turbo scanner (Leica, Wetzlar, Germany) to obtain images, and Image Scope software was used to analyze glomeruli morphology, renal tubules, and collagen deposition. For each kidney tissue, we selected three different slices and five fields of view in each slice. Twenty glomeruli were randomly selected from each kidney to assess the lesion.

Paraffin sections of left rat kidneys were stained with a TUNEL staining kit (Roche, Basel, Switzerland) to identify apoptotic cells in kidney tissue via fluorescein-dUTP labeling of DNA fragmented strands. Images were acquired using an Eclipse NI microscope (Nikon, Tokyo, Japan).

Total RNA was extracted from right kidney tissue frozen in liquid nitrogen using TRIzol (Thermo Fisher Scientific, MA, USA). The Revertaid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, MA, USA) was used for reverse transcription, and cDNA amplification was carried out using an SYBR Green Master Mix Kit (Takara, Otsu, Japan). Relative mRNA levels were calculated using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal standard. All primers were synthesized by Sangon Biotech and are listed in Table 1.

The right kidney tissue was homogenized with RIPA and centrifuged to obtain total tissue protein. The protein molecules were separated with 12.5% SDS PAGE gel electrophoresis, and the protein was transferred to a 0.22 µm PVDF membrane by a wet transfer method. A rapid blocking solution (Servicebio, Wuhan, China) was used for blocking at room temperature for 1 hour. PVDF membrane and rabbit-derived primary antibodies against against Bax (1:1000, Cell Signaling Technology, Boston, MA, USA), Caspase 9 (1:1000, CST, MA, USA), Caspase 3 (1:1000, CST, MA, USA), NF-$\kappa$B (p65) (1:1000, Abcam, Cambriage, UK), GAPDH (1:5000, Proteintech, Wuhan, China) were placed in horizontal shaking incubators overnight at 4 °C. After rinsing with TBST (5 min, 3 times), they were incubated with fluorescein or Horseradish peroxidase-labeled secondary antibodies for 1 hour at room temperature. We used X-ray film to develop bands. X-ray films were digitally scanned and analyzed quantitatively using the Image J software (V1.8.0, National Institutes of Health, Bethesda, MD, USA). Three rat kidneys in each group were randomly selected for protein extraction; each sample was repeated three times, and a representative band was selected for analysis.

Data are presented as means $\pm$ standard deviations (SD). Statistical analyses were performed using Prism 9.0.0/121 software (San Diego, CA, USA). Significant differences between the two groups were determined using one-way ANOVA (Dunnett’s t tests). For all comparisons, the other groups were compared with the data presented in the model group. p values less than 0.05 were considered to be statistically significant.

TEM images revealed that BMSC-Exo had a typical saucer-like structure (Fig. 2a). NTA showed that the average particle size of the particles in the sample was 136.5 $\pm$ 68 nm, that the main particle size peak was located at 102.9 nm, and that the main peak accounted for 97.4%, which was in line with the exosome particle size distribution characteristics (Fig. 2b). Western blot detection of exosome surface marker molecules showed that the obtained samples were positive for CD63 and CD81 (Fig. 2c).

Fig. 2.

Identification of BMSCs-Exo. (a) Transmission electron microscopy (TEM) shows that BMSCs-Exo have a “sauce-like” structure. (b) Analysis of BMSCs-Exo Particle Size Distribution by Nanoparticle Tracking Technology (NTA). (c) Exosome markers (D63, and CD81) are detected by Western Blot (S1-S5, 5 batches of samples).

GLU, BUN, and Cr were notably increased in the DM group compared with the NC, Met, or Exo groups (p 0.01). Treatment with Met significantly decreased these parameters (p 0.05). Exo appeared to show a stronger effect than Met, but there was no statistically significant difference between the two groups (p $>$ 0.05; Fig. 3, Supplementary Table 1).

Fig. 3.

BMSCs-Exo reduces blood glucose and improves renal function in diabetic rats. (a) Rats grouping and dosing schedule. (b,c) Comparison of GLU, BUN, and Cr in the NC group, DM group, Met group, and Exo group. *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001.

To evaluate the pathological changes of kidney damage in diabetic rats, we performed PAS staining and Masson staining in addition to conventional H&E staining and detected the synthesis and secretion of collagen IV (the most important matrix collagen in diabetic kidney disease) with immunohistochemical techniques. PAS staining showed that, compared with the NC group, the glomerular basement membrane of the DM group was homogeneously thickened, the mesangial matrix was increased, and there were no Kimmelstiel-Wilson (K-W) nodules with obvious arteriolar hyalinization. The basement membrane in the renal tubular region was markedly thickened (Fig. 4a). Masson staining showed that the glomerular basement membrane, mesangial matrix, surrounding tubules, and surrounding blood vessels were stained blue in the DM group, indicating that there was obvious collagen deposition in the kidney (a manifestation of renal fibrosis; Fig. 4b). Immunohistochemistry for collagen IV showed that collagen IV in the DM group was slightly distributed in the glomerular mesangial matrix, but was mainly in the tubular cells and interstitium and that the staining was more obvious than it was in the NC group (Fig. 4c). The above-mentioned manifestations were significantly alleviated after the administration of BMSCs-Exo and metformin. Compared with the Met group, the renal pathological manifestations of the BMSCs-Exo group were lighter.

Fig. 4.

BMSC-Exo alleviated diabetic kidney disease in the Exo group. (a) PAS staining of rat renal tissues (200$\times$). (b) Masson staining of rat renal tissues (200$\times$). (c) Immunohistochemical techniques used to detect collagen IV in kidney tissue (200$\times$).

To evaluate the renal cell apoptosis in diabetic rats, we performed TUNEL staining on the pathological sections of rat kidneys. The results showed that the DM group had more severe apoptosis than the NC group. Administration of Met significantly reduced apoptosis, and fewer apoptotic cells were detected with Exo than with Met (Fig. 5a). The pro-apoptosis protein Bax was detected by western blot, and the results corresponded with TUNEL staining. The relative expression of Bax in the rat kidneys in the DM group was significantly elevated compared with those in the NC group (p 0.01). Additionally, the gray value was significantly decreased in the Exo group (Exo vs. DM, p 0.05) compared with the DM group. There was no statistical difference between the Met goup and the DM goup (p 0.05). We also detected the expression levels of apoptosis-executing proteins Cleaved caspase 9 and Cleaved caspase 3, and showed that the expression level of Cleaved caspase 9 and Cleaved caspase 3 representing the activated form of Caspase, was significantly increased in the DM group compared with the NC group (p 0.05). Additionally, the expressions of both the Exo group and Met group were significantly lower than those in the DM group (Exo vs. DM p 0.01, Met vs. DM p 0.01). There was also no statistical difference between the Met group and the Exo group (p $>$ 0.05) (Fig. 5b).

Fig. 5.

BMSCs-Exo reduced apoptosis in renal tissue. (a) TUNEL staining of rat renal tissues (200$\times$). (b) Protein expression of Bax, Cleaved Caspase 9, Cleaved Caspase 3 detected by western blot. The relative expression of Bax, Cleaved Caspase 9 and Cleaved Caspase 3 were calculated. *p 0.05, **p 0.01.

To observe the renal inflammatory state of diabetic rats and the interventional effects of Exo, we observed the distribution of inflammatory cells in kidney tissue using H&E staining, detected the transcription level of pro-inflammatory factor TNF-$\alpha$ in kidney tissue using qRT-PCR, and detected the NF-$\kappa$B (p65) expression levels in kidney tissue with western blots. H&E results showed that a large number of inflammatory cells infiltrated the kidney tissue of rats in the DM group, and the distribution of inflammatory cells in the Exo and Met groups was significantly reduced (Fig. 6a). The TNF-$\alpha$ transcription level was significantly increased in the DM group compared with the NC group (p 0.01). Compared with the DM group, the Exo group showed significant decreases (p 0.05), but the Met group had no significant differences compared with the DM group (p $>$ 0.05) (Fig. 6b). Finally, the expression level of NF-$\kappa$B was significantly increased in the DM group compared with the NC group (p 0.005), the Met group (p 0.05), and the Exo group (p 0.05) compared with the DM group. There was no statistically significant difference between the two groups (Fig. 6c,d).

Fig. 6.

BMSCs-Exo attenuates kidney inflammation in diabetic rats. (a) H&E staining showed the rat kidney tissue structure and inflammatory cell infiltration (400$\times$). (b) Detection of TNF-$\alpha$ transcription levels in rat kidney tissue using qRT-PCR. (c) Western blot detection of the expression of inflammatory signaling molecule NF-$\kappa$B (p65) in rat kidney tissue. (d) Quantitative analysis of the gray value of western blot bands. *p 0.05, **p 0.01.