A retrospective study was conducted at the Department of Endocrinology and Metabolism, Nanfang Hospital of Southern Medical University, involving 350 DKD patients ≥ 18 years of age with an eGFR ≥ 15 ml/min/1.73 m² and urine albumin-to-creatinine ratio (ACR > 3.4 mg/mmol). Exclusion criteria were as follows: patients suffered from type 1 diabetes mellitus (T1DM), acute complications of diabetes, advanced renal failure, heart disease, urinary infection, cerebrovascular accident, kidney transplantation, and other confirmed renal disease (e.g., polycystic kidney, solitary kidney), and pregnant and lactating women. This study was approved by the Medical Ethics Committee of Nanfang Hospital of Southern Medical University, Guangzhou, China (Certificate Number: NFEC-2021-144).

Forty-two 7-week-old C57BL/6J male mice were obtained from the Guangdong Medical Laboratory Animal Center (Guangdong, China). Thirty mice (DM group) were administrated a single injection of streptozotocin (STZ, 120 mg/kg, i.p., pH = 4.5, Sigma-Aldrich, United States) dissolved in citrate buffer after a high-fat diet (HFD, D12492, HFK Bioscience, Beijing, China) for 4 weeks (Jia et al., 2019; Yang et al., 2020). A sustained blood glucose level, measured once a week, of > 16.7 mM indicated hyperglycemia. The other mice (n = 12) received injection of equal volume of sodium citrate.

Two weeks after STZ injection, DM group mice were randomly divided into two groups: 1) DM + HNa group mice (n = 12) were given HFD and HSD (60% fat and 4.0% salt, Guangdong Medical Animal Experimental Center, Guangdong, China); 2) DM group mice (n = 18) were fed with HFD and normal salt diet (NSD, 0.5% salt). Control mice were also randomly divided into two groups: 1) Control + HNa group mice (n = 6) were given HSD (Guangdong Medical Animal Experimental Center, Guangdong, China); 2) Control group mice (n = 6) were fed with NSD.

After 8 weeks HSD, mice were divided into the following groups: DM + Saline group (n = 6), DM + Dapa group (n = 6), DM + HNa + Dapa group (n = 6). The above mice were administrated with saline or dapagliflozin (1 mg/kg/day, MedChemExpress, China) by gavage for 12 weeks and continued their previous diet. All in vivo studies were performed from 2 to 4 pm, and approved by the Animal Ethics Committee of Southern Medical University, Guangzhou, China (Certificate Number: SYXK 2016-0167).

After 4 weeks HFD and intraperitoneal injection of STZ, IPGTT was conducted to evaluate the glucose responsiveness of mice. In short, mice (n = 6 per group) were fasted for 16 h with free access to drink water. Mice were injected intraperitoneally with glucose (2 mg/g body weight). Blood glucose levels were measured at 0, 15, 30, 60, 90 and 120 min.

Before the mice were sacrificed, IPITT was conducted to evaluate the insulin tolerance after mice were fasted for 6 h with free access to drink water. Mice (n = 6 per group) were injected intraperitoneally with insulin (0.75 U/kg body weight). Blood glucose levels were measured at 0, 15, 30, 60, 90 and 120 min.

For collecting urine, all experimental mice were housed individually in metabolic cages for 24 h with drinking water, without food. The mice were allowed to recover for 24 h before performing other operations. The urine output and water consumption were also measured. The urinary albumin (ml025061-j2; MIBio, Shanghai, China) and creatinine (ml026283-j2; MIBio, Shanghai, China) were detected by ELISA according to the manufacturer’s instructions. After fasting for 12 h, serum was collected. The Na levels in the urine and serum were measured by sodium assay kit (Nanjing Jiancheng Bioengineering Institute, Jiangsu, China) using a SpectraMax M5 Multifunctional Microplate Reader (Molecular Devices, United States).

Human renal tubular epithelial cell line HK-2 (China Center For Type Culture Collection) was cultured in 1.0 g/L Dulbecco’s modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS). After starvation with 2% FBS for 12 h, cells were cultured in 4.5 g/L DMEM medium, whether containing 15 mM NaCl (provided in the form of NaCl) or not, to simulate high glucose and salt conditions in vivo. Simultaneously, 15 mM sodium gluconate (Aladdin, Shanghai, China) or 25, 30 and 55 mM mannitol (Fuchen, Tianjing, China) were used to intervene cells to eliminate the influence of osmotic pressure and chloride ions. After culture, cells were treated with 5 nM Digoxin (APExBIO, United States), 40 μM Etomoxir (MedChemExpress, China) or 5 μM dapagliflozin and collected for corresponding examination.

The small interfering RNA targeting SLC5A2 (siSLC5A2) and the appropriate negative control (siNC) (Ribobio, Guangzhou, China) were transfected into HK-2 cells at a final concentration of 50 nM for 72 h. The sequences of si-SLC5A2 are shown in Supplementary Figure S1C. All transfections were conducted with Lipofectamine® 3000 (Invitrogen, Carlsbad, CA, United States) as instructed by the manufacturer’s protocols.

The renal cortex was fixed with 10% neutral formalin for 20–24 h, dehydrated and embedded in paraffin (Leica, German). The paraffin was cut into sections and tissue sections were subjected to Periodic Acid-Schiff (PAS) and Masson’s trichrome (Maiwei, Xiamen, China) staining as previously described (Wang et al., 2015). For immunohistochemical analysis, the sections were incubated with primary antibodies at 4°C overnight, including anti-FN (1:300; F3648; Sigma, United States), anti-CPT1A (1:800; 15184-1-AP; proteintech, United States), anti-ACOX1 (1:900; 10957-1-AP; proteintech), anti-FABP4 (1:50; 12802-1-AP; proteintech), anti-FASN (1:350; 10624-2-AP; proteintech), anti-SGLT2 (1:2,000; A03748-1; BOSTER, China), anti-ATP1A1 (1:200; 14418-1-AP; proteintech) and anti-ATP1B1 (1:50; 15192-1-AP; proteintech). They then were probed with secondary antibodies (pv-6001; ZSGB-BIO). The sections were photographed under a BX51 upright microscope (×200 or ×400) with four random fields and analyzed by the ImageJ software.

Total proteins were extracted using a RIPA lysis buffer (KeyGen BioTECH, Nanjing, China) and the protein concentration was measured by bicinchoninic acid (BCA) assay kit (TAKARA, Dalian, China). Proteins (20 μg) were separated using 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA). The membranes were blocked with 10% skimmed milk powder (Sigma, United States) for 1 h and incubated with following primary antibodies overnight at 4°C: anti-FN (1:2000, 66042-1-Ig, Proteintech Group, Inc. Rosement, United States), anti-COL-1 (1:1,000, A16891, ABclonal, Wuhan, China), anti-αSMA (1:500, A7248, ABclonal), anti-TGF-β1 (1:500, A18692, Santa Cruz, California), anti-ATP1A1 (1:5,000, 14418-1-AP, Proteintech), anti-ATP1B1 (1:1,000, 15192-1-AP, Proteintech), anti-CPT1A (1:1,000, 15184-1-AP, Proteintech), anti-ACC (1:10,000, 67373-1-Ig, Proteintech), anti-FASN (1:500, 10624-2-AP, Proteintech), anti-PGC-1α (1:1,000, A12348, ABclonal), anti-SGLT2 (1:600, 24654-1-AP, Proteintech), and anti-β-actin (1:2,000, 66009-1-Ig, Proteintech). The membranes then were probed with goat anti-mouse and anti-rabbit secondary antibodies for 2 h at room temperature. The bands were developed with enhanced chemiluminescent liquid (Millipore) and photographed. The images were analyzed with ImageJ software. We also validated SGLT2 rabbit polyclonal antibody after silencing SLC5A2 in HK-2 cells in Supplementary Figures S1A–C.

Total RNA was extracted using Trizol (TAKARA) and RNA quality was assessed by measuring the OD260/OD280 ratio using a NanoDrop micro spectrophotometer (Thermo Fisher). 1 ug RNA was reversely transcribed into cDNA with a PrimeScriptTMRT reagent Kit (TAKARA) to reverse transcribe RNA to cDNA. RT-qPCR was performed on Roche 480 PCR machine with 10 ul system of RT-qPCR reaction solution, namely 5 ul SYBR® Select Master Mix (2X) (TAKARA), 0.2 ul Forward Primers, 0.2 ul Reverse Primers, 1 ul cDNA and 3.6 ul dH2O. Then operate according to the following RT-qPCR reaction conditions: pre-denaturation at 95°C for 30 s, cycle at 95°C for 5 s and 60°C for 35 s for 50 times. The above experiment was repeated three times. The gene expressions of mouse and human were respectively normalized to β-actin and 18S mRNA and expressed as fold change. All primer sequences are presented in Supplementary Table S1.

Data were expressed as mean ± SEM. The pairwise comparison of normally distributed data used two independent sample t-tests, and the difference among multiple group was analyzed by one-way analysis of variance (One-Way ANOVA). Perform Levene’s homogeneity of variance test and analysis of variance first. If the variances are uniform (p > 0.05) and there is a significant difference in the means between groups (p < 0.05), the LSD method is further used for multiple comparisons; if the variances are not uniform (p < 0.05), the approximate F test Welch method is selected for correction; if there is a significant difference in the means between groups (p < 0.05), Dunnett’s T3 method is used for multiple comparisons. p < 0.05 was considered statistically significant.

To assess the potential relationship between 24 h-UNa excretion and renal function in DKD, we collected clinical data from 350 DKD patients, and evaluated the association between 24 h-UNa levels and renal function. Patients were grouped according to the degrees of albuminuria. Compared to the 24 h-UNa excretion of microalbuminuria group, patients of macroalbuminuria group had a higher urinary sodium excretion (Table 1, Figure 1A). Men were more likely to develop macroalbuminuria rather than microalbuminuria. Compared to the patients with microalbuminuria, patients with macroalbuminuria had a longer diabetic duration and higher systolic blood pressure, as we ll as a lower level of glycated hemoglobin (HbA1c), fasting blood glucose and low-density lipoprotein (LDL) with no significant difference on diastolic blood pressure between two groups. Besides, the levels of indexes of renal function including eGFR, serum creatinine and ACR were consistently correlated to the condition of albuminuria (Table 1). Binary logistic regression analysis indicated that LDL-cholesterol, duration of diabetes, systolic blood pressure, serum uric acid and 24 h-UNa excretion were all significantly associated with the degree of albuminuria. Interestingly, our data suggested that age was not the risk factor for DKD (Table 2). We equally divided the patients into four groups according to the excretion of 24-h urinary sodium, and we found that the excretion rate of urinary microalbumin was higher in the other three groups than in the lowest urinary sodium excretion group, but there was no significant increase in urinary protein level with the increase of urinary sodium level (Figure 1B). These data suggest that the excretion of urinary protein may be correlated with the excretion level of urinary sodium, but its correlation needs further validation.

To assess the association between high salt intake and the development of DKD, we first established a mouse model of T2DM through HFD and streptozotocin (Supplementary Figure S2A). IPGTT at 2 weeks after streptozotocin injection indicated impaired glucose tolerance in diabetic mice (Supplementary Figure S2B,C). Simultaneously, 24 h urine volume and drinking water volume of T2DM mice were higher than control mice (Supplementary Figure S2D,E). Then the diabetic mice were fed a high-salt diet or a normal-salt diet for 20 weeks, followed by evaluation of the renal fibrosis. Schematics illustrating the experimental protocols are shown in (Figure 2A). Urinary sodium and serum sodium concentrations were measured at 4 weeks and 20 weeks after HSD to determine whether mice received an effective salt loading. We found that 24 h-UNa excretion of the mice in the HSD-fed DM + HNa and Control + HNa group was significantly increased (Figure 2B), but their concentrations of serum sodium rarely changed (Figure 2C). The expression levels of Fn, αSma, Col-1 and Col-3a1 genes in the renal tubules of DM group and Control + HNa group mice were significantly higher than those in the Control group (Figure 2D). Further, their expressions were increased in the DM + HNa group (Figure 2D). In addition, the changes in renal fibrosis markers COL-1, TGF-β1 and αSMA detected by western blot were consistent with the above results (Figures 2E,F). The PAS staining results indicated that HSD thickened the glomerular basement membrane and increased vacuolar degeneration of renal tubular epithelial cells in DM mice (Figures 2G,H). Furthermore, there were more widespread collagen deposition by Masson staining (Figures 2I,J) and more expression of FN by immunohistochemistry analysis (Figures 2K,L) in the kidneys of the DM + HNa group. These findings suggested that high-salt intake triggers renal fibrosis in T2DM.

Fatty acid metabolism is essential to tubular epithelial cells and it is also the main source of energy in the kidney. To further investigate the mechanism underlying salt-induced exacerbation of renal interstitial fibrosis in diabetic mice, we tested the indicators of fatty acid metabolism in the renal cortex. It was indicated that HSD further restrained the fatty acid metabolism process in T2DM mice involving oxidation synthesis, transport and lipolysis as demonstrated by altered expression levels of related genes, including carnitine palmitoyltransferase 1A (Cpt1a), hormone sensitive lipase (Hsl), fatty triglyceride lipase (Atgl), fatty acid binding protein 4 (Fabp4), acetyl-CoA carboxylase α (Accα), and fatty acid synthase (Fasn) (Figures 3A,B). Even the expressions of Pgc-1α and Pparγ were down-regulated which were the important factors of mitochondrial biosynthesis (Figure 3C). Meanwhile, HSD induced increased expressions of mitochondrial biogenesis-related factors such as cytochrome C1 (CytoC1) and mitochondrial dynamin-related protein 1 (Drp1) (Figure 3C). Consistently, immunohistochemistry (Figures 3D,E) and western blot (Figures 3F,G) indicated the same changes in the protein expressions of these genes in T2DM mice fed with HSD. Additionally, the fasting blood glucose levels of HSD-fed diabetic mice were significantly reduced (Supplementary Figure S3A) and insulin sensitivity was restored (Supplementary Figures S3B,C), consistent with previous reports (Zhao et al., 2016). These data indicate that HSD induces changes in factors related to fatty acid metabolism in the renal tubules of diabetic mice.

To detect the in vitro impact of salt on the renal tubules in DKD, human renal tubular epithelial cells (HK-2) were cultured for 72 h in the medium containing high glucose (HG group), high glucose and NaCl (HG + HNa group) or normal glucose and NaCl (control group), followed by examination of fibrotic indexes. The expression levels of cell fibrosis indicators FN, α-SMA, and TGF-β were gradually increased with the increase in NaCl concentration (Figure 4A). When the concentration of NaCl climbed to 15 mM, TGF-β1 protein expression increased significantly by western blot (Figures 4B,C). Addition of high salt also induced cellular injury in HK-2 cultured in normal glucose (Figure 4D). The effect of high salt on HK-2 cells may be due to increased sodium, chloride, or elevated osmotic pressure. To determine the specific mechanism, we treated HK-2 with another 15 mM sodium gluconate or 25, 30 and 55 mM mannitol. The results showed that the expressions of fibrosis indicators were induced by the treatment with sodium rather than chloride (Figure 4E) and mannitol (Figure 4F). This indicates that high-salt-induced HK-2 cell damage hinges on the increase in sodium concentration, and increased osmotic pressure alone cannot change fibrotic gene expressions.

To identify the potential upstream mechanism of PTEC damage caused by high glucose and salt, we conducted an in vitro study in HK-2 cells, and cultured cells in a medium with 25 mM glucose and 15 mM NaCl for 72 h. Treatment with high salt further decreased the expressions of CPT1A, acyl coenzyme A oxidase 2 (ACOX2), and FABP4 genes (Figure 4G) in high-glucose cultured cells, which was consistent with the results of in vivo experiments. The high glucose and salt also induced decreased protein expressions of CPT1A, FASN and ACC (Figures 4H,I). Then the cells were treated with Etomoxir, an inhibitor of FAO for 72 h, and we found that the expressions of CPT1A decreased and the cell fibrosis was significantly alleviated (Figure 4J). These data indicate that high salt may induce dysregulation of fatty acid metabolism in PTEC to impair renal function.

To clarify the potential upstream mechanism whereby high salt causes disorder of renal tubular fatty acid metabolism, we first detected increased expressions of SGLT2 (Figures 5A–C) and Na+/K+-ATPase-related indicators ATP1A1, and ATP1B1 (Figures 5B,C) in the renal cortex of HSD-fed diabetic mice. In vitro, the levels of intracellular glucose, SGLT2, Na+/K+-ATPase and ATP in HK-2 cells were detected. The HG + HNa group exhibited elevated glucose levels (Figure 6A) and SGLT2 protein expression in cells (Figures 6B,C). With an increase in the concentration of NaCl added to HG-treated HK-2 cells, the expression levels of the above-mentioned Na+/K+-ATPase-related indicators gradually increased (Figure 6D), which was consistent with the changes of cell fibrotic indicators (Figure 4A). ATP1A1 expression was up-regualted in the HG + NaCl group (Figures 6E,F). When the cells in HG + HNa group were administrated 5 nM Na+/K+-ATPase inhibitor Digoxin for 72 h, the expressions of Na+/K+-ATPase decreased and the expressions of Fatty acid metabolism-related genes were restored, accompanied with improvement in the cell fibrotic phenotype (Figure 6G). Collectively, these findings demonstrate suppression of Na+/K+-ATPase might alleviate the fatty acid metabolism disorder, thereby relieving renal tubular damage.

So far, several studies have noted that administration of dapagliflozin could protect kidney through improving the energy metabolism of diabetic renal tubules. There are also clinical studies suggesting that HSD affects drug’s effectiveness. ACR in diabetic mice treated with dapagliflozin was significantly reduced, which was not obvious in diabetic mice on a high-salt diet (Figure 7A). Also, the recovery of αSma, Col-1 and Slc5a2 genes in diabetic mice with HSD was weaker than that of diabetic mice (Figure 7B), as were the protein levels of FN and SGLT2 (Figures 7C,D) in diabetic mice were restored by dapagliflozin treatment; the presence of HSD decreased the effect of dapagliflozin with lower expressions. Besides, high salt decelerated the recovery of the oxidation, transport, synthesis, and release of fatty acid in the renal tubules (Figures 7E–G) and weakened the dapagliflozin’s promoting effect on restoring the levels of mitochondrial-related factors Pparγ and Pgc-1α mRNA. And the inhibitory effect of dapagliflozin on CytoC1 (Figure 7H) and Na+/K+-ATPase-related indicators was also alleviated by HSD (Figures 7I,J).

For investigating the direct effect of high salt on HG- and dapagliflozin-treated PTEC, HG-treated HK-2 cells were treated with 1, 2, 5, 8, 10, 15 μM dapagliflozin for 72 h. It was found that mRNA expression levels of FN, COL1, CTGF and ATP1A1, ATP1B1 and ATP1B3 showed inverted-U-shape dose responses to increasing dapagliflozin concentration, with 5 uM dapagliflozin exhibiting the best intervention effect (Figures 8A,B). Western blot also proved that the expressions of SGLT2, FN and ATP1B1 proteins decreased most significantly in HG- and 5 μM dapagliflozin-intervened cells (Figures 8C,D). As the expression levels of Na+/K+-ATPase decreased in dapagliflozin-treated PTEC, the expression levels of fatty acid metabolism-related genes CPT1A and ACOX1 increased significantly (Figure 8E). These findings indicate that dapagliflozin reduces Na+/K+-ATPase expression in HK-2 cells and promotes fatty acid metabolism, thereby protecting the kidneys. However, the expression of ATP1A1 protein of HG + HNa group cells treated with dapagliflozin relative to HG + DMSO group was insignificantly down-regulated and there was no significant difference on CPT1A protein expression between both groups (Figures 8F,G). The above results indicate that high salt interferes with dapagliflozin to improve the energy metabolism of PTEC in DKD.