We downloaded the following four gene expression datasets from the Gene Expression Omnibus (GEO) database¹ : (1) GSE12682 (Si et al., 2009) comparing human kidney tissue between 29 healthy controls and 23 CKD samples, measured by the Affymetrix Human Genome U133A 2.0 Array (Affymetrix; Thermo Fisher Scientific, Waltham, MA, United States); (2) GSE12683 (Si et al., 2009) comparing Balb/c mouse kidney tissue between 10 healthy controls and 10 samples with AKI, measured by Affymetrix Mouse Genome 430A 2.0 Array; (3) GSE102513 (Vandenbussche et al., 2018) comparing CD-1 mouse kidney tissue between four healthy controls and four samples CKD induced by Tacrolimus (1 mg/kg/day for 28 days), measured by the Agilent-028005 SurePrint G3 Mouse GE 8x60K Microarray; and (4) GSE85957 (Pavkovic et al., 2014) comparing Rattus norvegicus (brown rat) kidney tissue samples between 19 healthy controls and 38 rats with kidney disease induced with 1 or 3 mg/kg cisplatin. In this last dataset, tissue was collected on days 3, 5, 8, and 26, and gene expression was measured using the Rat Genome 230 2.0 Array.

If one gene matched multiple probes, the average value of all probes was calculated as the expression of the corresponding gene. All sample collection and experimentation procedures were approved by the Guangxi Medical University Ethics Committee.

Differentially expressed genes (DEGs) between healthy control and diseased human kidney tissue were identified using the limma package (Ritchie et al., 2015) in R from dataset GSE12682. Differential threshold was calculated based on two criteria: false discovery rate (FDR) P-value < 0.05 and | log₂(fold change)| > 1. The ClusterProfiler package (Yu et al., 2012) in R was used to assess enrichment in biological processes (BPs) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Gene set enrichment was analyzed using the JAVA program² based on MSigDB immunologic signatures (c7.all.v6.2.symbols.gmt) (Godec et al., 2016). Gene sets with a NOM P-value < 0.05 after performing 1000 permutations were considered to be significantly enriched (Subramanian et al., 2005).

We used xCell web tool³ (Aran et al., 2017) with Charoentong signature (Charoentong et al., 2017) to generate cell enrichment scores based on gene signature profiles. Then these data were used to quantify immune cell proportions in all four datasets. Enrichment scores were compared using Student’s t-test, and P < 0.05 was considered to indicate a significant difference.

Animal experiments were approved by the Guangxi Medical University Ethics Committee. C57BL/six mice (8 weeks old, 20 ± 2 g) were purchased from the Experimental Animal Center of Guangxi Medical University. Mice were randomly divided into a control group (n = 27) and a treatment group (n = 27). The treatment group received 10.5 mg/kg Adriamycin (Zhejiang Hisun Pharmaceutical, Taizhou, China) three times per week for 2 weeks via the tail vein (Lu et al., 2013). The control group received injections of physiological saline via the tail vein. Urine was collected for 16 h overnight on day 27 for measurement of urinary protein. Blood was obtained on day 28 for measurement of serum creatinine and creatinine clearance. All urine and blood specimens were analyzed using automated analyzers at the Experimental Animal Center of Guangxi Medical University. Animals were sacrificed using cervical dislocation on day 28, and kidneys and spleens were removed for further study.

Kidneys were perfused with saline before removal, cut into 1–2 mm³ pieces and digested in Dulbecco’s modified Eagle medium (DMEM) containing 1 mg/ml collagenase IV (Sigma–Aldrich) and 100 mg/ml DNase I (Roche) for 40 min at 37°C with intermittent agitation. The digested cell suspension was then passed through a 40-μm cell strainer. Mononuclear cells were separated using 1.077 g/ml NycoPrep gradient (Axis-Shield, Oslo, Norway). Spleens from C57BL/6 mice were also isolated, minced, and digested for 30 min at 37°C in RPMI 1640 containing 1 mg/ml collagenase D and 100 mg/ml DNase I (both from Roche). The digested cell suspension was passed through a 40-μm cell strainer.

Single-cell suspensions from kidney were subjected to magnetic bead separation, and CD3+ cells were purified using the CD3+ T cell kit (Miltenyi Biotec, Bergisch Gladbach, Germany). The surface of CD3+ cells was labeled with antibodies against CD25 and FOXP3 (BD Accuri C6, United States). Double-positive cells were considered Tregs and were sorted using a flow cytometer (BD Accuri C6, United States) according to the manufacturer’s instructions.

Tregs were extracted from spleens of healthy C57BL/six mice, and single-cell suspensions were prepared as in Section “Preparation of Cell Suspensions.” Cells were incubated in vitro in RPMI 1640 medium and allowed to proliferate for 7 days. Approximately 1.2 × 10⁶ Tregs were injected via the tail vein into mice on day 5 after Adriamycin administration.

All mice were sacrificed on day 28. Coronal sections of renal tissue were stained with rat anti-mouse CD103 antibody (M290, 1:200) and hamster anti-mouse CD11c antibody (N418, 1:100), then incubated with AF488-conjugated goat anti-rat IgG or AF546-conjugated goat anti-hamster IgG antibody (both 1:500). Tissue sections were analyzed on an inverted fluorescence microscope (BX50, Olympus). Consecutive sections from kidney cortex to kidney medulla were photographed.

Bone marrow from the femurs of C57BL/6 mice was harvested and triturated with sterile syringes, and the resulting cell suspension was filtered through 40-μm nylon mesh. Approximately 2 × 10⁶ cells were incubated in RPMI 1640 containing 50 ng/ml granulocyte-macrophage colony stimulating factor (GM-CSF) at 37°C for 10 days. The adherent marrow-derived macrophages (M0) were rinsed three times in RPMI 1640 medium and then incubated in RPMI 1640 medium containing both IL-10 and TGF-β (10 ng/ml each) for 4 days to become M2c macrophages (Lu et al., 2013).

For co-culture experiments, M2c macrophages (1 × 10⁶ cells/ml) were plated in six-well plates with 1.5 ml of Tregs (1 × 10⁶ cells/ml). Co-cultures were incubated for 24 h at 37°C and 5% CO₂. Control cultures contained only Tregs. After 24 h, the floating Tregs were collected, washed in RPMI 1640, and used in subsequent experiments.

Statistical tests included the unpaired, two-tailed t-test using Welch’s correction for unequal variances and one-way ANOVA with Tukey’s multiple comparisons test. Statistical analyses were performed using Prism software (version 5, GraphPad) and R software (version 3.5.3). Results were expressed as mean ± SEM. Differences associated with P < 0.05 were considered statistically significant.

In human kidney tissue (GSE12682), we found a total of 345 DEGs by comparing healthy control tissue with CKD tissue (Figure 2A), of which 123 genes were up-regulated and 222 down-regulated. The down-regulated genes were enriched in various renal system development-related BPs and KEGG pathways (Supplementary Tables S2, S3), whereas up-regulated genes were immune-related (Figures 2B,C). Although no DEGs were identified in Balb/c mice (GSE12683) using the initial threshold (FDR P < 0.05 and | log₂(fold change)| > 1), using a less stringent threshold (P < 0.05) identified 660. These included 330 up-regulated genes enriched in various immune-related BPs (Figure 2E). These results suggest that genes of the immune system are activated in injured kidney, and that identification of immune cells associated with these up-regulated genes may provide candidate therapeutic targets against kidney injury.

Gene set enrichment analysis confirmed that among the up-regulated immune-associated gene sets, those associated with Tregs were significantly enriched in both human CKD tissues (Figure 2D) and mouse AKI tissues (Figure 2F). Extraction of genetic profiles associated with different immune cells from all datasets and scoring them for enrichment (Figure 3A and Supplementary Table S4) further confirmed Tregs to be more abundant in diseased kidney tissue than in healthy control tissue from humans, mice, or rats, regardless of the cause of tissue injury (Figures 3B–F).

Histology confirmed Adriamycin-induced structural kidney injury in mice (Figure 4A), including glomerular sclerosis, tubular atrophy, and interstitial expansion. Severe functional injury was indicated by increased serum creatinine (Figure 4B) and proteinuria (Figure 4C), as well as reduced creatinine clearance (Figure 4D). Numbers of CD4+ CD25+ Tregs were significantly higher in Adriamycin-injured kidneys than in control kidneys (5.1 versus 0.7%) (Figure 4E).

To examine whether the influx of Tregs mitigates or aggravates pre-existing kidney damage, we used a mouse model of adoptive cell transfer. CD4+ CD25+ Tregs were isolated from spleens of healthy mice and expanded in vitro (Figures 5A,B). Treg cultures of 90% purity were injected via the tail vein into mice with Adriamycin-induced kidney injury (Figure 5C). Kidneys were isolated on day 28 after Adriamycin administration and histology revealed that large numbers of Tregs had been recruited to the injured kidney, such that Tregs were more abundant in injured kidney than in healthy kidney (Figure 5D). Fluorescence imaging (Figure 5E) and histological imaging (Figure 5F) showed that adoptive transfer of Tregs aided recovery of structural defects in the kidney injury model.

M2c macrophages may induce Tregs to prevent Adriamycin-induced injury in mice (Kinsey et al., 2012). Indeed, we found that Tregs co-cultured with M2c macrophages showed significant up-regulation of chemokines CCR2, CCR5, CCR7, CD62L, CX3CR1 at the mRNA level (Figures 6A–C) and protein level (Figures 6B,C). M2c macrophages may promote the protective effects of Tregs in kidney by increasing expression of these chemokines.