Ten-week-old male C57BL/6 mice (wildtype, WT) or CX3CR1^−/− mice purchased from Jackson Laboratory were injected intraperitoneally with a single dose of STZ (Sigma Chemicals, United States) at 150 mg/kg dissolved in 10 mmol/L sodium citrate buffer (pH 4.5), while control mice were injected with the buffer alone. Three days later, mice with a random blood glucose level >20 mmol/L were assigned to the hyperglycemic groups, whereas citrate buffer-treated mice were assigned to the normoglycemic groups. The following experimental groups were formed: (1) WT mice (WT&Vehicle); (2) CX3CR1^−/− mice (CX3CR1^−/−&Vehicle); (3) WT mice with STZ-induced DM (WT&STZ); (4) CX3CR1^−/− mice with STZ-induced DM (CX3CR1^−/−&STZ); (5) WT mice with DM and insulin treatment for the final 2 weeks of the study (WT&STZ&Insulin), and (6) WT mice with DM and CX3CL1 neutralizing antibody treatment for the final 2 weeks of the study (WT&STZ&anti-CX3CL1). For insulin treatment, diabetic WT mice were injected subcutaneously with long-acting insulin at 40 IU/kg/day (Lantus; Sanofi-Aventis Deutschland GmbH, Germany). For treatment intervention, diabetic WT mice were injected intraperitoneally with anti-CX3CL1 antibody at 40 mg/kg/day (TP233; Torrey Pines Biolabs, Houston, TX, United States) as described previously (Xuan et al., 2011) or oral administration of canagliflozin at 10 mg/kg/day (Selleck Chemicals, Germany). Mice were euthanized by an overdose of pentobarbital (150 mg/kg, i.p) at 6 weeks after STZ injection to induce DM, and cardiac and renal function, apoptosis, and fibrosis were analyzed.

Female NOD mice, 10–30 weeks of age, were monitored for diabetes onset twice time per week by measurement of blood glucose. The mice with twice random blood glucose level >20 mmol/L were assigned to the hyperglycemic groups (day 1 of study). At 30^th week, the NOD mice with normal glucose were used as controls (NG). The treatment with CX3CL1 neutralizing antibody and insulin treatment in NOD mice was the same as in STZ mice mentioned above. Under anaesthetized by intraperitoneal injection of a mixture of xylazine (5 mg/kg) and ketamine (100 mg/kg), some NOD mice were injected with adeno-associated virus serotype 9 (AAV9) carrying the short hairpin-CX3CR1 plasmid (sh-CX3CR1) or a scramble plasmid (sh-scr) into the myocardium or kidneys. NOD mice were euthanized by an overdose of pentobarbital (150 mg/kg, i.p) and tissues harvest were performed for further analysis.

The neonatal rats were sacrificed by 2% isoflurane inhalation and cervical dislocation. Isolation and culture of ventricular cardiomyocytes (NRCM) and cardiac fibroblasts (NRCF) were performed as described previously with modifications (Korhonen et al., 2009; Tiede et al., 2010). After digesting the ventricular tissue, the cells were pelleted and the cells were re-suspended in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. The resulting cell mixture was pre-plated for 1 h to plate out cardiac fibroblasts. The isolated cardiomyocytes were plated into 6-well plates and cultured for 4–5 days. Pre-plated cardiac fibroblasts were cultured for 2 days and then 2-4 cells were used to passage.

A rat proximal tubular cell line, NRK-52E (ATCC, United States), and a rat glomerular mesangial cell line, HBZY-1 (ATCC, United States) were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) containing 10% FBS at 37°C.

Cultured NRCM and HBZY-1 were starved of serum for 12 h and then were exposed to recombinant soluble CX3CL1 (chemokine domain; 537-FT-025; R&D) at 200 ng/ml for 24 h in the presence or absence of 10 μM Y27632 (a ROCK inhibitor; Selleck) or 5 μg/ml of an anti-CX3CR1 antibody (GTX27200; Genetex). Cultured NRCF and NRK-52E were also exposed to recombinant soluble CX3CL1 (chemokine domain) at 200 ng/ml for 24 h in the presence or absence of 500 nM LY2157299 (a TGF-β/Smad signaling inhibitor; Selleck) or the anti-CX3CR1 antibody (5 μg/ml). Cultured H9C2, HBZY-1, NRCF, and NRK-52E were respectively exposed for 5 days either to normal glucose (5 mM, NG) or high glucose concentration (33.3 mM, HG) as well as to high glucose for 3 days followed by normal glucose for the remaining 2 days (HN) with/without different dose of canagliflozin (1, 5, and 10 μM).

RhoA GTPase activity was measured in cell lysates prepared from cardiomyocytes or glomerular mesangial cells in the presence or absence of 200 ng/ml sCX3CL1 by using a pull-down assay that employed the Rho-binding domain of the Rho effector protein rhotekin (Cytoskeleton, Inc., Denver, CO, United States, BK036). Total cell extracts and precipitates were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting with a monoclonal antibody targeting RhoA (diluted 1:500).

Confocal microscopy was used to delineate the distribution of cytochrome C. Cells were incubated with 25 nM of Mitotracker dye (Mitotracker red; Molecular Probes, Invitrogen) at 37°C for 10 min. Slides were washed with PBS and fixed in 4% formaldehyde. They were then incubated with primary anti-cytochrome C or AIF antibody followed by incubation with Alex488-labeled secondary antibody. The cells were also subjected to DAPI staining to visualize the nuclei. Images were taken using different excitation filters and merged.

For immunofluorescence staining, cultured cells were fixed in 3.7% paraformaldehyde for 10 min and then permeabilized with 0.1% Triton X-100 for 5 min. Next, the cells were incubated for 30 min at room temperature with blocking buffer (PBS containing 20% goat serum), before being incubated with the primary antibodies overnight at 4°C. The negative control was incubated with PBS instead of the primary antibody. For staining of heart and kidney sections, tissues were first embedded in Tissue-Tech OCT compound in cryomolds and then were snap-frozen in 2-methlybutane pre-cooled in liquid nitrogen and stored at −80°C. Heart and kidney sections were cut at a thickness of 5 μm and fixed for 10 min with chilled acetone. After blocking with 10% normal donkey serum in PBS for 30 min, sections were incubated with the primary antibodies overnight at 4°C. Next, the corresponding secondary antibodies conjugated to Alex Fluor 488 dyes (Invitrogen, United States) or Cy3 dyes were added and the sections were incubated for 60 min at room temperature. Nuclei were counterstained with DAPI. Images were acquired with a confocal laser microscope (Olympus, Japan).

Heart and kidney tissues from the different groups mice were fixed in 4% paraformaldehyde and embedded in paraffin. Then 4 μm sections were prepared for immunostaining. After antigen retrieved by heating in citrate buffer (pH 6.0), the sections were incubated with goat anti-CX3CL1 and anti-Bax antibodies, followed by incubation with an HRP-labeled secondary antibody. The Dako EVision + System-HRP (DAB) was used to visualize CX3CL1 and Bax staining. Images were captured using an upright microscope (Olympus, Japan).

After sCX3CL1 treatment for 24 h, cardiomyocytes or glomerular mesangial cells were loaded with 2 μM calcein-AM (Molecular Probes, United States) and 5 mM cobalt chloride for 15 min at 37°C to quench cytosolic calcein fluorescence (Dai et al., 2011). Then the cells were rinsed to remove residual dye and opening of the mPTP was visualized as redistribution of calcein from the mitochondrial matrix to the cytosol. For assessment of the mitochondrial membrane potential, cells were stimulated with sCX3CL1 for 24 h, loaded with 50 nmol/L tetramethylrhodamine ethyl (TMRE) for 30 min at 37°C, and then stained with DAPI for 10 min at room temperature. Viable cells were observed through a confocal laser microscope (Olympus, Japan). The fluorescence intensity of calcein-AM was monitored at 515 nm, while the fluorescence intensity of TMRE was monitored at 582 nm. A total of 200 cells from two replicate dishes were assessed for fluorescence intensity in three independent experiments, with the relative fluorescence intensity being determined by Image-Pro Plus 6.0 software (United States).

To localize nuclear DNA fragmentation in cultured cells or heart and kidney tissues, in situ terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) was performed using commercial apoptosis detection kit (Merck, United States or Roche, United States). All procedures were done according to the directions of the manufacturers. Cells were counterstained with DAPI to visualize nuclei. The number of TUNEL-positive cells was determined by randomly counting 10 fields of a section and was expressed as a percentage of the cells with normal nuclei.

Heart and kidney tissues were processed for electron microscopy as described previously to assess mitochondrial changes in the myocardium and renal tubules (Bernardi et al., 1999).

Tissues were cut into pieces smaller than 1 mm3 and fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for 4 h. Ten the tissues were post-fixed in 1% osmium tetroxide in 1% K4Fe(CN)6 buffer with 0.1 M sodium cacodylate, dehydrated through a graded series of ethanol and propylene oxide, and embedded in Epon 812. Ultrathin sections were cut, mounted on copper grids, and stained with lead citrate and uranyl acetate. A Hitachi H-7600 transmission electron microscope (Pleasanton, CA, United States) equipped with a digital camera was used to capture images of the sections. Damaged mitochondria were defined as those showing vacuolation, disruption of cristae, or deformity.

Nuclear DNA fragmentation was assessed by gel electrophoresis using the apoptotic DNA-ladder kit (Roche, United States) according to the manufacturer’s instructions. Briefly, after DNA extraction, sample DNA concentrations were quantified by reading the relative absorbance at 260/280, and 40 mg of DNA from each sample was loaded onto 1% agarose gel, separated by electrophoresis, stained by ethidium bromide, and visualized with an ultraviolet trans-illuminator.

Paraffin-embedded left ventricular sections were stained with Masson’s trichrome to measure interstitial fibrosis with the Image-Pro Plus 6.0 software. Myocardial interstitial fibrosis and perivascular fibrosis were quantified using a modification of the technique described (Liao et al., 2003; Ti et al., 2011). Briefly, ten 400x fields from 10 random stained sections from the mid left ventricle were digitized. For myocardial interstitial fibrosis analysis, an area stained blue was then selected for its color range, and the proportional area stained blue (matrix) was then determined. Perivascular fibrosis was excluded from the interstitial fibrosis measurement. For perivascular fibrosis measurement, perivascular fibrosis area/luminal area (PVCA/LA) was analyzed. To normalize the PVCA around vessels with different sizes, the perivascular fibrosis was represented as the PVCA-to-LA ratio. Kidney sections were stained with Masson’s trichrome. For the semiquantitative evaluation of fibrosis by Masson-Trichrome staining in the kidney, renal fibrotic lesions were expressed as percentage of Masson’s trichrome-positive fibrotic area. 20 randomly selected glomerulus or tubulointerstitial areas per mouse were graded as reported previously (Kitada et al., 2011).

After 6 weeks of DM induction, individual mice were placed in metabolic cages for 24 h urine collection. The urine samples were stored at -80°C until analysis. Blood samples were collected from the left cardiac ventricle. Enzyme-Linked Immunosorbent Assay (ELISA) was used to measure the level of neutrophil gelatinase-associated lipocalin (NGAL) in serum or 10% supernatant fluid of kidney tissues, serum creatinine (Cr), urine Cr, and serum blood urea nitrogen (BUN) from mice by using the kits (cusabio, United States) according to the protocols of the manufacturers.

Cardiac function was evaluated in mice by echocardiography using a Sequoia 512 system with a 15L-8 probe (Siemens, Germany). Mice were anaesthetized with inhalational isoflurane at a concentration of 1.5%, and two-dimensional parasternal short-axis images of the left ventricle (LV) were obtained at the level of the papillary muscles. From M-mode tracings, the LV end-diastolic diameter (LVEDD), LV end-systolic diameter (LVESD), and LV fractional shortening (LVFS) were measured.

Total RNA was extracted from culture cells and mouse heart and kidney tissues (total RNA isolation system, Omega, United States). Reverse transcription was carried out in 20 μL of reaction mixture containing 1 μg of total RNA. Digital images of CX3CL1, CX3CR1, and β-actin bands separated on ethidium bromide-stained agarose gels were obtained. Primer sequence was listed in the Supplementary Table S1.

Samples containing equal amounts of protein were separated by 8–12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto polyvinyl difluoride membranes. The membranes were blocked with 5% skim milk at room temperature for 1 h, incubated overnight at 4°C with the primary antibodies (Supplementary Table S2), and then washed three times with TBST buffer before incubation for 1 h with HRP-conjugated secondary antibodies at room temperature. The blots were detected using with Dylight 800-labeled secondary antibodies in a Western blotting detection system (Infrared Imaging System, Licor, United States). Relative expression was quantified by densitometry using the ImageJ Analysis Software (National Institutes of Health, Bethesda, MD, United States).

Data are expressed as the mean ± SEM. Comparison of two groups was performed by Student’s t-test, while multiple comparisons were done by one-way ANOVA followed by Bonferroni’s multiple comparison exact probability test. All analyses were performed with GraphPad Prism 8.0 software (GraphPad Software Inc., CA, United States), and Significance was accepted at p < 0.05.

Three days after STZ injection in mice, blood glucose levels were increased by about six folds. And they were almost normalized by insulin treatment for the final 14 days of the study (Figure 1A). Similar results were obtained in NOD mice (Figure 1B). CX3CL1 protein expression in the heart and kidneys was significantly increased in a time-dependent manner throughout the 42-day study period (Figures 1C,D). Immunohistochemistry showed that CX3CL1 expression was dramatically enhanced in the heart and kidney tissues of mice with STZ injection (Figure 1E). The gene expression levels of CX3CL1 and its receptor CX3CR1 were also increased in hearts and kidneys of NOD mice (Figures 1F,G).

We then investigated whether enhanced CX3CL1 expression would persist after hyperglycemia was controlled. From 4 weeks after STZ injection, mice with DM were treated with insulin for 2 weeks. Western blotting revealed that myocardial and renal CX3CL1 protein expression was not significantly lower after the control of hyperglycemia than in the untreated group (Figures 1H,I), suggesting that upregulation of CX3CL1 may represent a molecular memory of hyperglycemia. Moreover, we measured CX3CL1 expression in the principal cell types of the heart or kidney, which were exposed to high glucose. Compared with exposure to 5.5 mM D-glucose for 72 h, exposure to high concentrations of D-glucose (33 and 50 mM) resulted in a concentration-dependent increase of CX3CL1 expression in NRCM, NRCF, HBZY-1, and NRK-52E cells (Figure 1J).

Treatment with soluble CX3CL1 (sCX3CL1) for 1 h directly activated RhoA GTPase in both NRCM and HBZY-1 cells (Figures 2A,C). The expression of intact and cleaved ROCK1 protein was significantly upregulated in NRCM and HBZY-1 cells after exposure to sCX3CL1 for 24 h (Figures 2B,C). There was a decrease of fluorescence intensity detected by the calcium plus CoCl₂ assay in NRCM and HBZY-1 cells following sCX3CL1 stimulation (Figures 2D,F). Confocal microscopy revealed loss of TMRE staining, which indicates a decrease of the mitochondrial membrane potential, in cardiac and renal cells after stimulation with sCX3CL1 (Figures 2E,G). These findings indicate that the mitochondrial permeability transition was enhanced by sCX3CL1. In NRCM and HBZY-1 cells, stimulation with sCX3CL1 significantly increased mitochondrial-dependent apoptosis pathway related marker, include Bax and cytochrome C (CytoC) (Figures 2H–J). And after identifying mitochondria and nuclei by staining with Mitotracker (red) and DAPI (blue), respectively, the subcellular distribution of AIF and cytochrome C was monitored with confocal microscopy by detecting green immunofluorescence. This showed that sCX3CL1 stimulation induced translocation of AIF into the nucleus (Figure 2K) and the release of CytoC from the mitochondria into the cytoplasm (Supplementary Figures S1A,B). Moreover, the TUNEL and DNA ladder assays showed that sCX3CL1 increased apoptosis and DNA fragmentation in cultured cells (Supplementary Figures S1C–G). We also found that sCX3CL1 stimulated caspase-9 activity (Supplementary Figures S2A,B). The above effects of sCX3CL1 were antagonized by co-treatment with either a CX3CR1 neutralizing antibody (5 μg/ml) or a Rho kinase inhibitor (Y-27632 at 10 μM) (Figures 2A–K; Supplementary Figures S1A–G; Supplementary Figures S2A,B). Taken together, our findings indicate that sCX3CL1 promotes the mitochondrial apoptotic pathway through activation of RhoA/ROCK1-Bax signaling.

In cultured NRCF and NRK-52E cells, sCX3CL1 treatment increased the expression of TGF-β protein (Figures 3A,B). Exposure to sCX3CL1 also induced phosphorylation of Smad2 and Smad3, as shown by Western blotting (Figures 3C–F). Immunofluorescence also confirmed that sCX3CL1 induced a marked increase of phosphorylated Smad2 and Smad3 in the nucleus of cells (Supplementary Figures S3A–D). Both Western blotting and immunofluorescence demonstrated that exposure to sCX3CL1 induced the fibroblasts into myofibroblasts transition (FMT), which were characterized by the presence of a microfilamentous contractile apparatus enriched with α-smooth muscle actin (α-SMA) (Figures 3G,H). Both Western blotting and immunofluorescence showed a decrease in expression of the epithelial marker E-cadherin by NRK-52E cells in response to sCX3CL1 stimulation, while expression of the mesenchymal markers like Fibronectin, Vimentin, and α-SMA was increased (Figures 3I–K). The above-mentioned effects of sCX3CL1 were blocked by co-treatment with either the CX3CR1 neutralizing antibody (5 μg/ml) or with a TGF-β/Smad inhibitor (LY2157299 at 500 nM) (Figures 3C–K; Supplementary Figures S3A–D), suggesting that sCX3CL1-induced phenotypic trans-differentiation of cardiac and renal cells is dependent on the TGF-β/Smad pathway.

At 6 weeks after induction of DM by injection of STZ, we noted that wild-type (WT) mice developed rough hair and became glassy-eyed with a staggering gait, while these changes were much less marked in CX3CR1^−/− STZ-induced diabetic mice (Supplementary Figure S4A; Supplementary video). At 6 weeks, compared to WT&Vehicle group mice, WT diabetic mice had decreased left ventricular fraction shortening (LVFS), left ventricular ejection fraction (LVEF), and increased left ventricular end-systolic diameter (LVEDs), while CX3CR1^−/− diabetic mice had improvement of above echo parameters (Figures 4A,B; Supplementary Figures S4D–F). CX3CR1^−/− diabetic mice also had lower 24 h urinary albumin excretion (microalbuminuria) (Figure 4C), lower serum levels of creatinine (Figure 4D), and lower serum levels of neutrophil gelatinase-associated lipocalin (NGAL) (Figure 4E). Interestingly, these beneficial effects of CXRCR1 deficiency in STZ-induced diabetic mice were independent of the blood glucose level (Supplementary Figure S4G). Moreover, knockdown of the CX3CR1 results in similar cardiorenal improvement in NOD mice (Supplementary Figures S5A–E, Figures 4F–J). The above results suggest that CX3CR1 deficiency preserves cardiorenal function in diabetic mice.

After hyperglycemia had persisted for 4 weeks, we treated STZ-induced diabetic WT mice and NOD mice with insulin or an CX3CL1 neutralizing antibody for 2 weeks. As a result, the insulin-treated STZ mice remained in a poor condition similar to untreated WT diabetic mice, while the appearance and behavior of mice treated with the CX3CL1 neutralizing antibody showed marked improvement (Supplementary Figure S4A, Supplementary Video). Additionally, in WT diabetic mice, treatment with the CX3CL1 neutralizing antibody, resulted in significant improvement of cardiorenal dysfunction (higher LVFS, lower renal tissue NGAL levels, less microalbuminuria, and lower urine and serum creatinine levels), while insulin treatment only resulted in an improving tendency of cardiorenal dysfunction (Figures 4A–E). In NOD mice, treatment with either CX3CL1 neutralizing antibody or insulin significantly improved cardiorenal dysfunction (Figures 4F–J; Supplementary Figures S5C–E).

TUNEL assay revealed an increase of apoptotic cells in both the heart and the kidneys of STZ-induced diabetic WT mice, while fewer apoptotic cells were noted in CX3CR1^−/− diabetic mice or WT mice treated with the CX3CL1 neutralizing antibody for 2 weeks (Figures 5A–D). In contrast, insulin treatment did not cause a significant reduction of apoptosis (Figures 5A–D). Immunohistochemical analysis revealed that expression of Bax was enhanced in the heart and kidneys of WT diabetic mice, but not in CX3CR1^−/− diabetic mice and WT mice treated with the CX3CL1 neutralizing antibody (Figures 5E,F). The level of Bax expression was similar in WT diabetic mice with or without insulin treatment (Figures 5E,F). Electron microscopy also revealed deformities of some of the myocardial and tubular mitochondria in WT diabetic mice, including attenuation and angulation along the longitudinal axis with focal enlargement and swelling of the cristae (Figures 5G,H). In contrast, the majority of the mitochondria were intact with few broken cristae in CX3CR1^−/− diabetic mice or WT diabetic mice treated with the CX3CL1 neutralizing antibody (Figures 5G,H).

Myocardial interstitial and perivascular fibrosis were less prominent in CX3CR1^−/− diabetic mice or WT diabetic mice treated with the CX3CL1 neutralizing antibody (Figures 6A,B, Supplementary Figure S6A). When renal fibrosis was assessed by Masson-trichrome staining, there was a significantly lower fibrotic score for renal glomerular and interstitial area in CX3CR1^−/− diabetic mice or WT diabetic mice receiving the neutralizing antibody than in untreated WT diabetic mice (Figures 6C,D, Supplementary Figure S6B). Similarly, cardiorenal benefits from knockdown CX3CR1 by AAV or neutralizing antibody improved cardiorenal fibrosis in NOD mice. As for insulin treatment, just in NOD mice, could reverse cardiorenal fibrosis while it did not work in STZ-induced diabetic mice (Figures 6A–H, Supplementary Figures S4A–D).

Immunofluorescence showed that the number of vimentin-positive cardiac fibroblasts with abundant α-SMA protein was markedly increased in STZ-induced diabetic WT mice, indicating that phenomenon of cardiac fibroblasts to myofibroblasts transition (FMT) (Figure 6I). Furthermore, in the kidneys of WT diabetic mice, E-cadherin expression was decreased along with increase of fibronectin, vimentin, and α-SMA expression, suggesting that the epithelial-mesenchymal transition (EMT) had been induced (Figure 6J). In CX3CR1^−/− and neutralizing antibody-treated diabetic mice, there was marked suppression of cellular trans-differentiation in the heart and kidneys, but this was not seen in insulin-treated diabetic mice (Figures 6I,J).

Finally, we examined if CX3CL1 is a therapeutic target of SGLT2i because it is a consensus that SGLT2i is protective for cardiorenal dysfunction (Rangaswami et al., 2019) and possibly are associated with SGLT2-independent mechanisms. Four weeks after STZ-induced hyperglycemia, we treated the diabetic mice with canagliflozin for 2 weeks. As a result, canagliflozin significantly improved cardiorenal dysfunction (Figures 7A–D), dramatically suppressed the cardiac and renal expression level of CX3CL1 (Figures 7E–G). Considering the glucose-lowering effect of canagliflozin, we tried to test whether canagliflozin had a glucose-lowering independent effect on CX3CL1 expression. Three days after exposure to high glucose (HG, 33 mM), the medium of cardiac and renal cells was replaced with normal glucose medium (HN, 5.5 mM) and cultured for 2 days in the presence/absence of canagliflozin. As shown in Figures 7H–M, there was still a high level of CX3CL1 expression in the HN groups, while co-treatment with canagliflozin 10 μM significantly downregulated the protein expression of CX3CL1 in the four types of cardiac and renal cells.