ST2⁺ T-Regulatory Cells in Renal Inflammation and Fibrosis after Ischemic Kidney Injury

Visual Abstract

Abstract

Key Points

• IL-33/ST2 alarmin pathway regulates inflammation, fibrosis, and resolution of ischemia-reperfusion injury of kidneys.

• ST2 regulates the transcriptome of T-regulatory cells related to suppressive and reparative functions.

• The secretome of ST2⁺ T-regulatory cells regulates hypoxic injury in an amphiregulin-dependent manner.

Background

Inflammation is a major cause of kidney injury. IL-1 family cytokine IL-33 is released from damaged cells and modulates the immune response through its receptor ST2 expressed on many cell types, including regulatory T cells (Tregs). Although a proinflammatory role of IL-33 has been proposed, exogenous IL-33 expanded Tregs and suppressed renal inflammation. However, the contribution of endogenous IL-33/ST2 for the role of Tregs in the resolution of kidney injury has not been investigated.

Methods

We used murine renal ischemia-reperfusion injury and kidney organoids (KDOs) to delineate the role of the ST2 and amphiregulin (AREG) specifically in Tregs using targeted deletion. Bulk and single-cell RNA sequencing were performed on flow-sorted Tregs from spleen and CD4 T cells from postischemic kidneys, respectively. The protective role of ST2-sufficient Tregs was analyzed using a novel coculture system of syngeneic KDOs and Tregs under hypoxic conditions.

Results

Bulk RNA sequencing of splenic and single-cell RNA sequencing of kidney CD4 T cells showed that ST2⁺ Tregs are enriched for genes related to Treg proliferation and function. Genes for reparative factors, such as Areg, were also enriched in ST2⁺ Tregs. Treg-specific deletion of ST2 or AREG exacerbated kidney injury and fibrosis in the unilateral ischemia-reperfusion injury model. In coculture studies, wild-type but not ST2-deficient Tregs preserved hypoxia-induced loss of kidney organoid viability, which was restored by AREG supplementation.

Conclusions

Our study identified the role of the IL-33/ST2 pathway in Tregs for resolution of kidney injury. The transcriptome of ST2⁺ Tregs was enriched for reparative factors including Areg. Lack of ST2 or AREG in Tregs worsened kidney injury. Tregs protected KDOs from hypoxia in a ST2- and AREG-dependent manner.

Introduction

AKI is a major contributor to morbidity and mortality in hospitalized patients, including those with ischemia-reperfusion injury (IRI).¹ The role of IL-33 has been assigned contradictory functions.^2,3 The binding of IL-33 to its receptor ST2 (Il1rl1) activates both proinflammatory and anti-inflammatory immune cells.⁴ ST2 is also expressed on the regulatory T cells (Tregs), which offer a dominant anti-inflammatory mechanism. While exogenous IL-33 or IL-33–containing fusion protein (IL233) promoted Tregs to protect from AKI, endogenous IL-33 was deemed proinflammatory.^5,6 The ST2⁺ Tregs are enriched in tissues in response to IL-33 treatment during homeostasis and injury⁷ and to endogenous IL-33 in solid tumors to suppress antitumor immunity.⁸ ST2⁺ Tregs regulate adipogenesis and inflammation in adipose tissue⁹ and also play a reparative role in multiple organs.^10,11 However, a definitive contribution of ST2 expression on Tregs, especially because it pertains to the resolution of kidney injury, has remained understudied. Tregs can produce amphiregulin (AREG), an epidermal growth factor (EGF) family member, which has been shown to promote repair in many tissues.^12–15

We hypothesized that ST2⁺ Tregs play an important role in the resolution of kidney injury. We investigated whether the IL-33/ST2 pathway, in addition to regulating anti-inflammatory functions of Tregs, also promotes production of proreparative factors for resolution of kidney injury. Here, we used unbiased transcriptomics to identify ST2-regulated factors in Tregs isolated from the spleen and postischemic kidneys, and IRI model in mice carrying deletion of ST2 or AREG in Tregs. We also used a novel in vitro coculture assay system using syngeneic kidney organoids (KDOs) to understand the role of ST2⁺ Tregs and AREG in preserving the viability of kidney progenitor cells during hypoxic injury.

Methods

Mice and IRI

Studies were performed in accordance with protocols approved by the University of Virginia Animal Care and Use Committee. The Il1rl1^tm1 (ST2KO), Il1rl1^fl/fl, and Areg^fl/fl mice were derived from cryopreserved sperms obtained from University of California, Davis, The Knockout Mouse Project Repository. ST2KO mice were crossed with B6.Foxp3^GFP mice to generate ST2KO-Foxp3-GFP mice for flow-sorting of splenic Tregs for bulk RNA sequencing (Supplemental Figure 2). B6.Foxp3^YFP-Cre mice were purchased from the Jackson Laboratory. The Il1rl1^fl/flFoxp3^YFP-Cre and Areg^fl/flFoxp3^YFP-Cre mice were generated by crossing Il1rl1^fl/fl and Areg^fl/fl with Foxp3^YFP-Cre mice (Supplemental Figures 1 and 3) and genotyped by PCR (primers shown in Supplemental Table 4). IRI was induced as reported,^10,16 with details presented in Supplemental Methods. Plasma creatinine and BUN were analyzed using enzymatic (Diazyme Laboratories), and colorimetric assays (Arbor Assay), respectively.¹¹ Gene expression analysis for kidney injury and fibrosis was performed as described in Supplemental Table 1. Kidney sections were analyzed by hematoxylin and eosin and Masson trichrome staining as before,^10,11 and fibrosis was quantified using ImageJ.¹⁷

Next-Generation Bulk RNA Sequencing

RNA, prepared from flow-sorted cells, was assessed for quality as described (Supplemental Figure 14 and Supplemental Table 3). Library preparation and sequencing were performed with High Throughput Nextseq kit (Illumina) using the Nextseq 500 Sequencing System (Illumina) at University of Virginia Genome Analysis and Technology Core, research resource identifier: SSR_018883. After quality control with FastQC and Cutadapt,¹⁸ reads were aligned using HISAT2¹⁹ to mouse reference genome mm10/GRCm38 and counted using FeatureCounts.²⁰ Differential gene expression was analyzed through principal component analysis (PCA), distance matrix, and heatmap using DESeq2.²¹ Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed using GOseq.²² Datasets are available in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) database²³ (GSE224295).

Flow Cytometry, Treg Sorting, Suppression Assay, and Kidney Organoid Coculture

Flow cytometry was performed as before^10,24 using labeled antibodies (Supplemental Table 2). Tregs were sorted from spleens using Influx Cell Sorter (BD Biosciences), and Treg-suppression assay was performed as before²⁵ (Supplemental Figure 4). For coculture, splenic Tregs were isolated using a kit (STEMCELL Technologies). Tregs (1×10⁵) were cocultured in 0.4μ trans-well system (Costar), with approximately 50 KDOs (generated as described in Supplemental Figures 9 and 11), in alpha modification MEM supplemented with ribonucleosides, deoxyribonucleosides, L-glutamine (2 mM), and 10% FBS (Gibco). For in vitro hypoxia reoxygenation,²⁶ the culture was placed at 37°C in a humidified sealed hypoxic chamber (Billupus-Rothenberg) purged with 1% oxygen, 5% carbon dioxide, and balance nitrogen (Praxair), maintaining 6-hour hypoxia, followed by 1-hour normoxia (5% carbon dioxide, 20% oxygen, balance nitrogen). Recombinant AREG (Biolegend) was used at 100 ng/ml. Viability of organoids was assayed using cell counting kit-8 (Dojindo Molecular Technologies) following manufacturer's instructions.

Single-Cell RNA Sequencing of Kidney T Cells after Ischemia

CD4⁺ T cells were flow-sorted from kidneys of mice with or without ST2 expression in Tregs and subjected to ischemia reperfusion before performing single-cell RNA sequencing as described (Supplemental Figure 15). Single-cell RNA sequencing datasets are available in National Center for Biotechnology Information GEO database (GSE272380).

Statistical Analyses

Samples with wild-type (WT) and mutant alleles were compared using one-way ANOVA, followed by the Tukey test for multiple comparisons using GraphPad Prism. The results are expressed as mean±SEM, with P < 0.05 considered significant.

Results

RNA Sequencing Identified Distinct ST2-Dependent Gene Expression Profiles of Tregs

To elucidate the role of ST2 in Tregs, we performed RNA sequencing of flow-sorted ST2-high, ST2-low, and ST2KO Tregs. Hierarchical clustering demonstrated unique gene expression signatures between the three Treg subsets (Figure 1A). The 2D PCA plot indicated that ST2-high, ST2-low, and ST2KO Tregs possessed distinct properties with no major differences within the replicates (Figure 1B). Euclidean clustering showed that ST2-high and ST2-low were more similar, as compared with ST2KO Tregs (Figure 1C). Volcano plot to depict the significantly upregulated (red) and downregulated genes (blue) between ST2KO and ST2-high Tregs showed a substantial upregulation of several genes in the ST2-high Tregs, as compared with the ST2KO group (Figure 1D). Expression of genes for reparative factors, AREG, and growth differentiation factor-15 (GDF15) was highly enriched in the ST2-high and ST2-low Tregs compared with ST2KO Tregs (Figure 1E). Areg was higher in the ST2-high Tregs than both ST2-low and ST2KO Tregs, but Gdf15 did not differ between the ST2-high and ST2-low groups. Furthermore, ST2-high and ST2-low Tregs presented a more activated phenotype with higher expression of Ctl1a4, Il10, Tgfb1, Cd73, Tnfr2, and Cd44 and a reduced expression of Cd39, Nrp1, Sell, and Ghrl compared with ST2-deficient Tregs.

Figure 1.

RNA sequencing of flow-sorted Tregs subtypes revealed distinct ST2-regulated signatures. (A) Heatmap representing unsupervised hierarchic clustering of gene expression profiles from splenic Tregs of ST2KO/ST2-low (ST2m)/ST-high (ST2p) (n=3 mice per group). (B) Characterization of samples based on a 2D principal component analysis (PCA). (C) Euclidean distances between the samples represented as a heatmap of the sample-to-sample distance matrix with clustering. (D) Volcano plot representing significantly upregulated and downregulated genes between ST2-high versus ST2KO Tregs. (E) Expression changes for selected Treg-associated genes across the three Treg subsets.

GO analysis was performed using the Goseq algorithm, using the top ten statistically significant upregulated and downregulated GO classes (Figure 2Ai). In ST2-high Tregs, positive regulation of cell cycle, cell proliferation, cytokine-mediated signaling, cellular responses to cytokine, cellular response to oxygen, and inflammatory response were significantly upregulated compared with ST2KO Tregs (Figure 2Aii), whereas ubiquitination, nucleocytoplasmic transport regulation, IFN-β production, chaperon-mediated protein complex assembly, and IL-8 production were downregulated (Figure 2Aiii). KEGG pathway analysis results for the top ten significantly upregulated and downregulated pathways were plotted as a bar graph between ST2KO and ST2-high Tregs, with the combined scores of the top features represented by a heatmap (Figure 2Bi). In ST2-high Tregs, EGF receptor, Jak-STAT, T-cell receptor, Oncostatin-M, IL-1, IL-3, NF-κB, thymic stromal lymphopoietin, and cell cycle signaling pathways were upregulated (Figure 2Bii), whereas Fas (CD95), stress induction, MyD88-independent toll-like receptor 3/4 cascade, inflammasome, stimulator of interferon response cGAMP interactor 1, toll-like receptor, TNF, DNA damage, and ataxia telangiectasia mutated signaling were downregulated in the ST2-high Tregs compared with ST2KO (Figure 2Biii). Similar differential expression of GO attributes and KEGG pathways was observed in ST2-high Tregs compared with ST2-low with cell proliferation, NF-kB signaling, and chemotaxis upregulated in ST2-high Tregs. Pathways for negative regulation of transcription and mitogen-activated protein kinase signaling and positive regulation of programmed cell death and fibrosis were downregulated in the ST2-high Tregs compared with ST2-low Tregs (Supplemental Figure 5). Thus, ST2-high Tregs possessed a unique set of attributes, which could play a vital role in the homeostasis and proliferation of Tregs, immunomodulation, and contribution to cellular repair.

Figure 2.

Categorization of expressed genes to GO classes and KEGG pathway analysis. (A) GO enrichment analysis for ST2-high relative to ST2KO Tregs to predict the gene association with biological processes, molecular functions, and cellular components, by comparing against experimentally validated datasets. (i) Heatmap presenting the GO score based on combined P value and Z score. (ii) Bar chart representing the top ten significantly upregulated GO attributes. (iii) Bar chart representing the top ten significantly downregulated GO attributes. (B) KEGG pathway enrichment analysis for ST2-high relative to ST2KO Tregs to map the molecular interaction, reaction, and relation networks. (i) Heatmap presenting the KEGG pathway score based on combined P value and Z score. (ii) Bar chart representing the top ten significantly upregulated KEGG pathways. (iii) Bar chart representing the top ten significantly downregulated KEGG pathways. ATM, ataxia telangiectasia mutated; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; STING, stimulator of interferon response cGAMP interactor 1; TCR, T-cell receptor; TSLP, thymic stromal lymphopoietin.

Loss of ST2 in Tregs Exacerbated Kidney Dysfunction

ST2 deletion in Tregs in Il1rl1^fl/flFoxp3^YFP-Cre, Il1rl1^fl/wtFoxp3^YFP-Cre, and Il1rl1^wt/wtFoxp3^YFP-Cre was confirmed by genotyping.²⁷ Although no significant difference was seen in the proportion of splenic-Foxp3⁺ Tregs between Il1rl1^fl/fl and Il1rl1^wt/wtFoxp3^YFP-Cre (Figure 3A), the expression of ST2⁺ on Tregs was significantly different, confirming ST2 deletion (Figure 3B). In vitro, Treg suppression assay (Figure 3C) indicated that ST2-sufficient Tregs were more efficient in suppressing T-cell proliferation compared with ST2-deficient Tregs. To test the role of ST2⁺ Tregs in AKI, we performed bilateral ischemic injury followed by reperfusion for 24 hours (Supplemental Figure 6A). This resulted in a similar loss of function in both Il1rl1^fl/fl and Il1rl1^fl/flFoxp3^YFP-Cre mice (Supplemental Figure 6, B and C), suggesting that ST2 expression on Tregs may be more critical for recovery from AKI rather than suppression of injury in the early phase. Therefore, we performed unilateral ischemia to clamp only the right renal pedicle, followed by reperfusion for 14 days with contralateral nephrectomy performed on day 13 to measure the function of the injured kidney (Figure 3D). Indeed, the Il1rl1^fl/flFoxp3^YFP-Cre mice had higher plasma creatinine and BUN levels than Il1rl1^fl/fl mice (Figure 3, E and F), suggesting the importance of ST2 Tregs in restoring homeostasis after injury.

Figure 3.

Loss of ST2 expression on Tregs worsened loss of kidney function on ischemic injury. (A) Flow cytometry–based splenic Treg analysis: (i) Il1rl1^fl/fl, (ii) Il1rl1^fl/fl, Foxp3^YFP-Cre, (iii) bar plot estimating the proportion of Tregs in the spleen of naïve mice (n=3). (B) Flow cytometry for expression of ST2⁺ on Tregs in spleens of naïve mice (n=3). (C) In vitro Treg suppression assay to measure dilution of CFSE for proliferation of labeled CD4 T cells. (D) Schematic of unilateral IRI experimental design. Mice were subjected to unilateral ischemia (right kidney); 24 hours before euthanasia, contralateral nephrectomy was performed, n≥6. (E) Plasma creatinine, day 14. (F) BUN, day 14. Symbols represent individual mice; mean±SEM is shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. CFSE, carboxy-fluorescein succinimidyl ester; IRI, ischemia-reperfusion injury.

Greater Kidney Injury, Fibrosis, and Inflammation Were Observed in Mice with ST2-Deficient Tregs

Compared with the sham controls (Figure 4A), Il1rl1^fl/fl IRI mice showed tubular necrosis, cast formation, tubular dilatation, and interstitial leukocyte infiltration (Figure 4B), which appeared further exacerbated in Il1rl1^fl/flFoxp3^YFP-Cre IRI mice (Figure 4C), indicating greater injury in the absence of ST2-expressing Tregs (Figure 4D). The expression of Kim1 and Ngal (Figure 4E, i and ii) was also significantly upregulated in Il1rl1^fl/flFoxp3^YFP-Cre IRI mice compared with the Il1rl1^fl/fl mice. Collagen deposition, measured (Figure 4, F–H) using Masson trichrome staining (Figure 4I), and higher mRNA levels of the fibrosis markers, Col1a1, Col3a1, Sma, and Vim using semiquantitative real-time PCR (Figure 4J and Supplemental Figure 8), showed that Il1rl1^fl/fl Foxp3^YFP-Cre mice exhibited significantly more fibrosis than Il1rl1^fl/fl mice. Together, the data suggested that deletion of ST2 in Foxp3-expressing Tregs results in further aggravation of kidney injury, when analyzed 14 days after the ischemic injury.

Figure 4.

Increase in the severity of injury and worsening of kidney fibrosis on loss of ST2 expression in Tregs. Representative hematoxylin and eosin–stained sections from mice subjected to ischemic injury (IRI): (A) Il1rl1^fl/fl; Sham, (B) Il1rl1^fl/fl; IRI, and (C) Il1rl1^fl/fl Foxp3^YFP-Cre; IRI. (D) Kidney injury score; (E) real-time PCR analysis of kidney injury markers (i) Kim1 and (ii) Ngal. Representative images of Masson trichrome staining (F) Il1rl1^fl/fl; Sham, (G) Il1rl1^fl/fl; IRI, and (H) Il1rl1^fl/fl Foxp3^YFP-Cre; IRI. (I) Quantification of fibrosis using image J, expressed as a percentage of the total surface area of kidney sections. (J) Heatmap representing real-time quantification of transcript levels of fibrosis markers, collagen 1a (Col1a1), collagen 3a (Col3a1), smooth muscle actin (Acta2/Sma), and vimentin (Vim), relative to Gapdh (See Supplemental Figure 7 for individual markers). Symbols represent individual mice; n≥6, mean±SEM is shown. *P < 0.05; **P < 0.01; ****P < 0.0001.

The levels of CD4⁺Foxp3⁺ Tregs in the blood (Figure 5Ai) and kidneys (Figure 5Aiii) were significantly higher in Il1rl1^fl/fl and Il1rl1^fl/flFoxp3^YFP-Cre IRI mice compared with sham, whereas splenic Tregs (Figure 5Aii) were lower in all IRI mice than sham. There were more Tregs in the blood and spleen of Il1rl1^fl/flFoxp3^YFP-Cre mice compared with the Il1rl1^fl/fl mice, which did not reach significance (Figure 5A, i and ii). The neutrophil count was increased in Il1rl1^fl/flFoxp3^YFP-Cre than Il1rl1^fl/fl (Figure 5B). The CD4⁺ T cells, although showing different proportions among the CD45⁺ cells, their absolute number did not show any difference (Supplemental Figure 7, B and C). Immunofluorescence staining also showed greater neutrophilic infiltrates in the kidney tubulointerstitium of the Il1rl1^fl/flFoxp3^YFP-Cre than Il1rl1^fl/fl mice; however, the CD4 T-cell infiltration and Foxp3⁺ cells were similar in both groups (Supplemental Figure 7D). Spleens from the Il1rl1^fl/flFoxp3^YFP-Cre mice displayed higher percentages of CD4 T cells producing IFNγ and TNFα than the Il1rl1^fl/fl mice (Figure 5C, i and ii), with no difference in IL-4 production (Figure 5Ciii). Importantly, the level of anti-inflammatory cytokine, IL-10 was lower in Il1rl1^fl/flFoxp3^YFP-Cre IRI mice compared with the Il1rl1^fl/fl IRI mice (Figure 5Civ), which correlated with RNA-seq data (Figure 1E). The data indicated that the presence of ST2 on Tregs was critical for the suppression of inflammation.

Figure 5.

ST2 expressing Tregs attenuated inflammatory milieu. Quantification of immune cell subsets in blood, spleen, and injured kidney using flow cytometry at 14 days postischemic injury. (A) CD4⁺ Foxp3⁺ Tregs in (i) blood, (ii) spleen, and (iii) kidney; (B) kidney neutrophils (Ly6G⁺CD11b⁺Ly6C⁺); (C) single-cell suspension of splenocytes was exposed to PMA and ionomycin, in presence of monensin, for a duration of 5 hours. After this treatment, intracellular cytokine expression was assessed using flow cytometry. Proinflammatory cytokines (i) CD4⁺IFNγ⁺ and (ii) CD4⁺TNFα⁺; anti-inflammatory cytokines (iii) CD4⁺IL-4⁺ and (iv) CD4⁺IL-10⁺. Symbols represent individual mice; n≥6, mean±SEM is shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. PMA, phorbol 12-myristate 13-acetate.

ST2-Deficient Tregs Were Unable to Protect KDOs from Hypoxic Injury

We developed an in vitro cell culture system using KDOs to recapitulate renal IRI starting with syngeneic pluripotent C57BL/6J murine embryonic stem cells (Supplemental Figure 9), which could be differentiated into KDOs as characterized extensively by histology, gene expression analysis, and immunostaining techniques (Supplemental Figure 10) for the expression of markers related to glomerular (nephrin), proximal tubular cells, and some features of a developing kidney (Lgr4, Lgr5, Pax8, Sox9, etc.). We tested the hypothesis that ST2⁺ Tregs may be important to preserve the viability of these potential progenitor cells because loss of ST2 in Tregs worsened ischemic injury in mice. Since there were no established protocols for the coculture of KDOs with Tregs, we optimized conditions for coculture by experimenting with various culture conditions under hypoxic conditions. Alpha modification MEM supplemented with ribonucleosides and deoxyribonucleosides gave the best results, on the basis of the viability of both KDOs and Tregs (Figure 6A).

Figure 6.

Secretome of ST2-sufficient splenic Tregs improved cellular viability of KDO during in vitro hypoxic injury. (A) Optimization of coculture conditions as described in methods using RPMI 1640 media with 10% FBS and supplements (RPMI), STEMdiff™ hematopoietic - EB basal medium (EB Basal), and αMEM complete media for 24 hours: (i) KDO, (ii) Tregs (technical replicates, n=3; biological replicates, n=1). (B) (i) Trans-well culture of KDO with Tregs (CD4⁺CD127⁻) from Il1rl1^fl/fl (WT) or ST2-KO mice. (ii) Cell viability assay to estimate in vitro hypoxic injury of trans-well culture (technical replicates, n=3; biological replicates, n=2 sets). (C) (i) Illustration depicting strategy to test the effect of Treg secretome on KDO. (ii) ELISA-based estimation of AREG from FSM (technical replicates, n=3; biological replicates, n=2 sets). (iii) Cell viability assay to evaluate in vitro hypoxic injury on KDO under Treg secretome culture conditions (technical replicates, n=3; biological replicates, n=2 sets). Mean±SEM is shown. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. αMEM, alpha modification minimal essential media; AREG, amphiregulin; FSM, filtered spent media; KDO, kidney organoids; WT, wild-type.

To measure the viability of KDOs after hypoxia-reoxygenation, we cultured Tregs and organoids, separated by a 0.4 μm membrane using the Trans-well culture system, such that Tregs cultured in the top chamber could be removed to monitor the viability of the organoids (Figure 6Bi). Subjecting the organoids to hypoxia reoxygenation resulted in severe reduction in viability, which was prevented by coculture with ST2-sufficient Tregs, but not with ST2KO Tregs (Figure 6Bii). Because Tregs were separated from organoids by a membrane, we hypothesized that soluble factors secreted from Tregs must be responsible for the protection. We cultured isolated Tregs for 24 hours and transferred the filtered spent media (FSM) from the Treg culture to kidney organoid culture before subjecting them to hypoxia reoxygenation (Figure 6Ci). AREG (Areg) is one of the genes highly expressed in the ST2-high Tregs compared with the ST2KO Tregs (Figure 1D). Indeed, AREG levels in the FSM of WT Tregs were higher than the ST2-deficient Tregs (Figure 6Cii). Culturing KDOs with the FSM from ST2KO Treg supplemented with recombinant AREG to the levels observed in the WT Treg FSM (40 pg/ml) restored the loss of viability of organoids (Figure 6Ciii) and reduced the expression of Il-33 and Hif1a in the organoids subjected to hypoxia indicative of hypoxic injury and its suppression by AREG among the soluble factors from ST2⁺ Tregs (Supplemental Figure 12). Our data suggested that secretome from ST2⁺ Tregs possessed AREG to boost survivability of organoids that represent tubular progenitors.

Kidney Injury Was Exacerbated by Deletion of AREG in Tregs

To investigate the relevance of AREG in Tregs, we deleted AREG in Tregs using cre-loxp–mediated recombination. The mice were subjected to unilateral ischemic injury. Mice with AREG-deficient Tregs (Areg^fl/flFoxp3^YFP-Cre) had higher plasma creatinine, BUN, and exacerbated injury compared with AREG-sufficient Areg^fl/fl mice on ischemia (Figure 7, A and B). Hematoxylin and eosin staining indicated greater tubular atrophy and inflammation (Figure 7, C and D) in Areg^fl/flFoxp3^YFP-Cre mice subjected to ischemic injury. Levels of kidney injury markers Kim1 and Ngal and fibrosis markers Col1a1, Col3a1, Sma, and Vim (Figure 7E and Supplemental Figure 13) were also higher in Areg^fl/flFoxp3^YFP-Cre mice. In addition, Tregs showed a lower trend in blood (Figure 7Fi), and significantly lower Treg levels in the spleen (Figure 7Fii) and kidney (Figure 7Fiii) of Areg^fl/flFoxp3^YFP-Cre mice than WT mice. There was upregulation of inflammatory factor TNFα and decrease in anti-inflammatory cytokine IL-10 and T helper cell 2 cytokine IL-4 in the CD4 T-cells (Figure 7G). These findings collectively underscored the role of AREG in Treg homeostasis and the protective role of AREG-expressing Tregs in mitigating kidney injury.

Figure 7.

Loss of AREG expression in Tregs exacerbated kidney injury by impairing Treg-mediated immune regulation and promoting inflammation. The Areg^fl/fl and Areg^fl/fl Foxp3^YFP-Cre mice were subjected to unilateral ischemic injury (IRI) as shown in Figure 3D. (A) Plasma creatinine, day 14. (B) BUN, day 14. (C) Kidney injury score. (D) Representative hematoxylin and eosin–stained sections (i) Areg^fl/fl, Sham; (ii) Areg^fl/fl, IRI; and (iii) Areg^fl/fl Foxp3^YFP-Cre, IRI. (E) Heatmap representing real-time quantification of mRNA levels of kidney injury markers, Kim1, Ngal, and fibrosis markers, collagen 1a (Col1a1), collagen 3a (Col3a1), smooth muscle actin (Acta2/Sma), and vimentin (Vim), relative to Gapdh. (F) Flow cytometry for quantification of CD4⁺ Foxp3⁺ Tregs in (i) blood, (ii) spleen, and (iii) kidney. (G) Flow cytometry for quantification of proinflammatory cytokines (i) CD4⁺IFNγ⁺, (ii) CD4⁺TNFα⁺; anti-inflammatory cytokines (iii) CD4⁺IL-10⁺, (iv) CD4⁺IL-4⁺ after activation of splenic cells with PMA/ionomycin. Symbols represent individual mice; n≥4, mean±SEM is shown. *P < 0.05; **P < 0.01; ****P < 0.0001.

Single-Cell Transcriptomics Identified ST2-Regulated Genes in Tregs from Injured Kidneys

Owing to a rarity of CD4⁺ T cells and Tregs among total renal immune cells, CD4 T cells were flow-sorted after unilateral IRI from the kidneys of Il1rl1^fl/fl (IRIWT) and Il1rl1^fl/flFoxp3^YFP-Cre (IRIKO) mice and subjected to single-cell RNA sequencing. The samples from Il1rl1-sufficient and Il1rl1-deficient genotypes (Figure 8A) were merged as IRIWT and IRIKO, respectively, and cell clusters were identified (Figure 8B) on the basis of distinct gene expression profiles (Supplemental Figure 15, A and D). Genes for activated T cells and Th1 T cells were enriched in the IRIKO group (Stmn1, Pclaf, Mki67, Top2a, Arsb, Vav3) compared with IRIWT, which was enriched for genes for naïve and quiescent T cells (Igfbp4, Satb1, Bach2, Aff3, Actn1). Foxp3, Areg, and Il1rl1 were coexpressed and upregulated among the Tregs (Figure 8C). IRIWT Tregs, subsetted for Il1rl1 expression into IL1RL1-positive and IL1RL-negative populations (Figure 8D), showed that the IRIWT-IL1RL1–negative group clustered with IRIKO Tregs (Figure 8E). Foxp3, Areg, and Il1rl1 were also coexpressed at higher levels in the IRIWT-IL1RL1 than IRIWT-IL1RL1–negative and IRIKO groups (Figure 8F). Expression of Treg function–related genes (Il2ra, Ctla1, Entpd1, Nt5e, Nrp1, Tnfsf1b, Cd44) was higher in the IRIWT-IL1RL1–positive group (Supplemental Figure 15E). The transcripts for Il10 or Ghrelin (Ghrl) and Gdf15 were not reliably detected in single-cell RNA sequencing. The expression of some genes (Tgfb1, Entpd1, Nrp1, and Sell) in kidney Tregs was different than splenic Tregs (Figure 1E), suggesting differences in properties of Tregs between lymphoid and kidney tissues. The signaling pathways for cell cycle, mammalian target of rapamycin, 5' adenosine monophosphate-activated protein kinase, and TCR were upregulated in the IRIWT-IL1RL1–positive than IRIWT-IL1RL1–negative and ST2KO Tregs (Figure 8, G and H), thus, overall implying a higher activation status of the ST2-expressing Tregs in the injured kidneys.

Figure 8.

Single-cell RNA sequencing analysis of ST2-sufficient and ST2-deficient Tregs in injured kidneys. The Il1rl1^fl/fl (IRIWT, n=2) and Il1rl^fl/fl Foxp3^YFP-Cre (IRIKO, n=3) mice were subjected to unilateral ischemic injury (IRI), after which CD4⁺ T cells were sorted by flow cytometry and subjected to 10× single-cell RNA sequencing workflow. (A) Merged UMAP plot displaying clustering of single-cell RNA-seq data from IRIWT and IRIKO groups. (B) Integration of datasets using the Harmony algorithm to correct batch effects and improve clustering accuracy. (C) Dot plot of gene expression levels across different cell types. (D) UMAP plot focusing on the Treg-specific subset, providing a more granular view of the distribution and relationship within the subset. (E) Heatmap showing hierarchical clustering of groups based on gene expression profiles. (F) Dot plot detailing differentially expressed genes across condition groups, highlighting significant expression differences. KEGG pathway enrichment analysis for upregulated genes comparing (G) IRIWT-IL1RL1–positive (ST2-positive) versus IRIKO. (H) IRIWT-IL1RL1–positive versus IRIWT ST2–negative (ST2-negative). AMPK, 5' adenosine monophosphate-activated protein kinase; MAPK, mitogen-activated protein kinase; UMAP, uniform manifold approximation and projection.

Discussion

ST2 interacts with its only known ligand IL-33, which is released by injured cells.²⁸ In previous studies,^10,11 treatment with an IL-2 and IL-33 hybrid cytokine (IL233) induced accumulation of ST2⁺ Tregs in injured kidneys. However, the role of ST2 expression on Tregs in AKI remained a knowledge gap. RNA sequencing analysis of flow-sorted Tregs revealed major differences between ST2KO, ST2-high, and ST2-low Tregs. We hypothesized that ST2⁺ Tregs are pivotal in limiting inflammation, injury, and fibrosis and also promote tissue restoration (Figures 1 and 2). ST2-high Tregs showed upregulation of transcripts for reparative factors AREG and GDF15. Also secreted by Tregs,²⁹ AREG was shown to induce proliferation and cellular differentiation, thus indicating their role in repair of tissues,³⁰ including kidneys, as was shown by a role for AREG from ILC2 in protection from AKI.³¹

To identify a role for ST2 in AKI, we generated Treg-specific ST2 knockout mice (Il1rl1^fl/flFoxp3^YFP-Cre) and subjected them to IRI. Deletion of ST2 in Tregs resulted in worsening of kidney function and exacerbation of kidney injury, inflammation, and fibrosis (Figures 3–5). We showed earlier that ischemic injury caused a decrease in Tregs in spleen and increase in kidneys, especially in mice treated with IL233,¹⁰ suggesting that ST2⁺ Tregs were mobilized from lymphoid reservoirs to kidneys on injury. Other studies have also described the enrichment of ST2-expressing Tregs in kidneys during injury and their expansion using IL-33 in conjunction with IL-2/anti-IL2 complex.³² Upon injury, Tregs were shown to be recruited to the site of inflammation via chemokine receptor CCR6.³³ Indeed, a recent study demonstrated a core signature of ST2⁺ Tregs distinct from ST2⁻ Tregs across multiple organs, such that ST2⁺ Tregs represented a more activated and migratory phenotype.⁷ ST2⁺ Tregs behaved similarly in most tissues with higher expression of Klrg1, Ki67, Id2, Cd44, and Gata3 than ST2⁻ Tregs with subtle differences in tissue-specific chemokine receptors. We observed elevated levels of Tregs in the blood, spleen, and kidney of Il1rl1^fl/flFoxp3^YFP-Cre mice (Figure 4), yet there were significantly more neutrophils in the kidneys and higher production of TNFα and IFNγ in mice with ST2-deficient Tregs, suggesting that ST2 may dictate the suppressive function of Tregs in the kidneys. Indeed, ST2⁺ Tregs were more suppressive than non–ST2-expressing Tregs (Figure 3C) as previously reported in both mice and humans.³⁴ IL-33 enhanced the expression of Foxp3 and GATA3 expression in colonic Tregs.³⁵ In turn, phosphorylated GATA3 recruited RNA polymerase II to Foxp3 and ST2 promoters, thus enhancing ST2 on the surface of Tregs.^35,36 Our data suggested that ST2 expression on Tregs may be more important during the repair process post-AKI because there was no significant effect of Treg-ST2 deletion 24 hours after injury in the bilateral ischemia model (Supplemental Figure 5), but exacerbated fibrosis in the unilateral ischemia model. In the case of sustained inflammation in the kidneys, maladaptive repair, vascular rarefaction, and fibrosis could have occurred, which led to CKD and kidney failure.³⁷

KDOs have served as a potential platform to study cellular interactions and to screen compounds related to organ viability.^38,39 We generated KDOs from syngeneic embryonic stem cells and cocultured them with Tregs separated by a membrane. Our in vitro study supported the role of secretory factors from ST2⁺ Treg in protecting the viability of organoids under hypoxia. Our RNA sequencing data showed that ST2-high Tregs had high expression of Areg and Gdf15, which could enhance cell viability.^40,41 AREG level in ST2-deficient Tregs was reduced and correlated with the lack of protection in organoids cultured with FSM from ST2KO Tregs as well as its restoration by AREG supplementation (Figure 6). AREG signals through EGF receptor (EGFR), which also played a role in nephrogenesis.⁴² AREG from CD4⁺ T cells was shown to induce the proliferation of intestinal epithelial cells during worm infection, underlying the importance of immune cell and tissue cell cross-talk in regulating the AREG-EGFR pathway.⁴³ Because Tregs are drawn to the sites of inflammation and injury, we proposed that AREG secretion by Tregs played a role in enhancing repair mechanisms. To further delineate the role of AREG with respect to Tregs, we generated Areg^fl/flFoxp3^YFP-Cre mice to show (Figure 7) that loss of AREG expression in Tregs exacerbated kidney injury and fibrosis. Loss of AREG in Tregs also worsened lung damage after influenza infection, with AREG expression in Tregs enhanced by IL-33.¹²

Flow cytometry (Figure 7F) showed significantly more Tregs in the spleen and kidneys of the WT IRI mice compared with Areg^fl/flFoxp3^YFP-Cre mice, indicating a role of AREG in Treg-homeostasis, as was also suggested earlier.^34,41 Areg^fl/flFoxp3^YFP-Cre mice had upregulation of inflammatory cytokine TNFα and a decrease in anti-inflammatory cytokine IL-10 and IL-4 in the CD4 T cells. Our earlier studies²⁴ showed that adoptive transfer of IL-10–deficient Tregs was unable to offer protection in ischemic kidney injury. However, mycophenolate mofetil treatment promoted recovery from ischemia and was accompanied with reduced levels of IL-10,⁴⁴ suggesting that IL-10 may not play a role in the repair process in renal ischemic injury. Furthermore, treatment of mice with IL-2/anti-IL2 complex was shown to increase Tregs and protect mice from ischemic injury even in the IL-10 KO mice.⁴⁵ Treatment of KDO with AREG, but not IL-10, rescued the loss of organoid viability by hypoxia (not shown), suggesting that IL-10, although important for regulating inflammation, might be redundant for repair of renal ischemic injury.

GDF15, a member of the TGFβ superfamily,^46,47 was demonstrated to have renoprotective and immunomodulatory effects after ischemic⁴⁷ and other kidney diseases^48,49 in part via promoting Klotho expression.⁴⁰ Both soluble ST2 and GDF15 were proposed as biomarkers for AKI.^50,51 Contrastingly, GDF15 expression on tubular organoids was proposed as a marker of maladaptive repair.⁵² Although Gdf15 was expressed highly on ST2-high Tregs compared with ST2KO Tregs, its expression did not differ between ST2-high and ST2-low Tregs (Figure 1E). Although GDF15 treatment could rescue kidney organoid viability during hypoxia (not shown), future careful studies are needed for deletion of GDF15 specifically in the Treg compartment.

Single-cell transcriptome analysis of kidney Tregs highlighted the role of renal ST2⁺ Tregs in ischemic injury. Uniform manifold approximation and projection plots (UMAP) revealed distinct clusters of kidney CD4⁺ T cells with Foxp3, Areg, and Il1rl1 being upregulated and coexpressed in total and Il1rl1 expressing (IL1RL1-positive) Tregs, indicating their role in kidneys postischemic injury. Tregs not expressing Il1rl1 (IL1RL1-negative) were closer to ST2KO Tregs than IL1RL1-positive Tregs. Enrichment of signaling for cytokine receptors, chemokine, cell cycle, and mammalian target of rapamycin in ST2⁺ Tregs suggested a more activated phenotype. There was overlap in the genes and pathways enriched for ST2⁺ splenic and kidney Tregs, especially for Foxp3, Il1rl1, and Areg, implying that ST2⁺ Tregs from different tissues had functional overlaps.⁷

Several Treg activation–related genes were enriched in the Il1rl1 expressing WT Tregs compared with the IL1RL1-negative or ST2KO Tregs in kidneys (Il2ra, Ctla1, CD39 [Entpd1], Nt5e [CD73], Nrp1, Tnfsf1b [TNFR2], Cd44). Although CD44 showed differential transcript expression, cell surface CD44 expression on ST2-sufficient or deficient kidney Tregs was not different in flow cytometry analysis (Supplemental Figure 16i). The proportion of CD44⁺ Tregs was higher in kidneys from mice subjected to ischemia compared with sham, suggesting an accumulation of CD44⁺-activated Tregs in injured tissues, unrelated to ST2 genotype (Supplemental Figure 16). Surprisingly, no or minimal transcripts of Gdf15, Ghrl, and Il10 were detected in kidney Tregs postischemia (Figure 8 and Supplemental Figure 15). Although the role of IL-10 was appreciated in ischemic injury in early time points,⁵³ we may have missed the relevant time point. Nevertheless, mice sufficient for ST2 expression had higher IL-10 production than ST2-deficient Tregs (Figure 5). These results may highlight a limitation of transcriptomics-based approaches, necessitating protein-level studies for critical targets.

In summary, using bulk and single-cell RNA sequencing of splenic and renal Tregs, ischemic injury model, and KDO, we showed the critical role of ST2⁺ Tregs in maintaining kidney homeostasis in late stages of ischemic injury. This study pioneers the use of KDOs cocultured with Tregs as an experimental model to identify soluble factors in ST2 Tregs, such as AREG, which could be leveraged for potential treatments of AKI.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/E831.

Funding

R. Sharma: National Institute of Diabetes and Digestive and Kidney Diseases (R01DK105833, R01DK104963, and R21DK112105), Virginia Catalyst (Award 1303), Juvenile Diabetes Research Foundation United States of America (3-SRA-2021-1005-S-B), and UVA School of Medicine LaunchPad Diabetes Fund. V. Sabapathy: National Institute of Diabetes and Digestive and Kidney Diseases post-doctoral fellowship (5TL1DK132771) and Indiana University George M. O’Brien Center (Advanced Microscopy Fellowship).

Author Contributions

Conceptualization: Vikram Sabapathy, Rahul Sharma.

Data curation: Vikram Sabapathy.

Formal analysis: Vikram Sabapathy, Rahul Sharma, Rajkumar Venkatadri.

Funding acquisition: Rahul Sharma.

Investigation: Nardos Tesfaye Cheru, Gabrielle Costlow, Murat Dogan, Airi Price, Vikram Sabapathy, Rahul Sharma, Rajkumar Venkatadri.

Methodology: Nardos Tesfaye Cheru, Gabrielle Costlow, Murat Dogan, Saleh Mohammad, Airi Price, Vikram Sabapathy, Rajkumar Venkatadri.

Project administration: Rahul Sharma.

Resources: Rahul Sharma.

Supervision: Rahul Sharma.

Validation: Rajkumar Venkatadri.

Writing – original draft: Vikram Sabapathy, Rahul Sharma.

Writing – review & editing: Vikram Sabapathy, Rahul Sharma.

Data Sharing Statement

Data related to transcriptomics can be accessed using the following accession numbers at the NCBI GEO database. Analyzable Data; Raw Data/Source Data. GEO. 1. Accession number: GSE224295; Figure 1, Bulk RNA-Seq https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE224295. Accession number: GSE272380; Figure 8, Single-cell RNA-Seq https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE272380.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E830.

Supplemental Methods

Supplemental Figure 1. Breeding strategy for Il1rl1^fl/flFoxp3^YFP-Cre.

Supplemental Figure 2. Breeding strategy for IL1RL1 (ST2) KO Foxp3-GFP.

Supplemental Figure 3. Breeding strategy for Areg^fl/flFoxp3^YFP-Cre.

Supplemental Figure 4. Flow cytometry, cell sorting, and Treg suppression assay.

Supplemental Figure 5. GO and KEGG analysis ST2-high versus ST2-low Tregs.

Supplemental Figure 6. Acute renal bilateral ischemia-reperfusion injury.

Supplemental Figure 7. uIRI flow cytometry for renal ST2 Tregs and CD4 T cells.

Supplemental Figure 8. uIRI qPCR for renal fibrosis markers.

Supplemental Figure 9. Culture, expansion, and characterization of murine ESC.

Supplemental Figure 10. Kidney organoid differentiation and characterization.

Supplemental Figure 11. Schema of kidney organoids differentiation timeline.

Supplemental Figure 12. Taqman assay for IL-33 and Hif1a expression in organoids.

Supplemental Figure 13. uIRI, qPCR for injury and fibrosis in Areg^fl/flFoxp3^YFP-Cre.

Supplemental Figure 14. RNA QC of flow-sorted ST2-high/ST2-low/ST2KO Treg.

Supplemental Figure 15. Detailed results scRNA-seq—renal CD4⁺ T cells post-IRI.

Supplemental Figure 16. CD44 expression on renal Tregs using flow cytometry.

Supplemental Table 1. Real-time PCR analysis and primers.

Supplemental Table 2. Antibodies used for flow cytometry/immunostaining.

Supplemental Table 3. RNA quality control for next-generation sequencing.

Supplemental Table 4. Genotyping primers.