Tr1 Cells Emerge and Suppress Effector Th17 Cells in Glomerulonephritis

This article is featured in Top Reads, p. 1607

Immune-mediated inflammatory diseases are characterized by an expansion of proinflammatory CD4⁺ T-cell subsets, such as Th1 and Th17 cells. Accordingly, inflammatory diseases in the kidney, such as antineutrophil cytoplasmic Ab (ANCA)-associated glomerulonephritis (GN), lupus nephritis, and anti–glomerular basement membrane GN, are characterized by an increase of effector CD4⁺ T cells, especially of Th1 and Th17 cells (1–5). Thus, one key aim is to understand how these cells can be therapeutically controlled. The immune system has several mechanisms by which effector CD4⁺ T cells can be controlled to maintain and reestablish homeostasis. A first mechanism to control effector CD4⁺ T cells is exerted via regulatory T (Treg) cells. Foxp3⁺ Treg cells are an anti-inflammatory T cell subset that has been studied extensively in the context of GN (6–9). It has been shown that Foxp3⁺ Treg cells via the production of IL-10 control Th17 cells in the kidney and the intestine (10, 11). There is another anti-inflammatory T-cell subset that is also known to produce high amounts of IL-10, namely T regulatory type 1 (Tr1) cells. Tr1 cells are CD4⁺ T cells that are defined by the absence of Foxp3, expression of IL-10, and regulatory function (12, 13). Interestingly, the suppressive activity of Tr1 cells also depends on functional IL-10 receptor signaling (11, 14, 15). In fact, when Tr1 cells cannot respond to IL-10, they lose the capacity to maintain their own IL-10 production; consequently, their suppressive activity declines (14). Mouse Tr1 cell differentiation can be induced in vitro by the addition of the cytokine IL-27 (16). Tr1 cells are identified via a combination of criteria: high IL-10 expression, lack of Foxp3, coexpression of the integrin α2 subunit (CD49b), and lymphocyte activation gene 3 (LAG-3) (13, 17). Additionally, there are more coinhibitory receptors (CIRs) that are known to be expressed by Tr1 cells, such as T cell Ig and ITIM domain (TIGIT), the transmembrane protein TIM-3, programmed cell death protein 1, and CCR5.

Recently, the presence of Tr1 cells has been identified in the kidneys of patients with IgA vasculitis (18). IL-10⁺ CD4⁺ T cells have also been shown to expand and mediate regulatory function in patients with lupus nephritis after methylprednisolone therapy (19). Similarly, Foxp3⁺ and Tr1-like cells expand after nasal myeloperoxidase peptide tolerization therapy in a murine myeloperoxidase-ANCA GN model and reduce disease severity (20). Thus, Tr1 cells seem to be present in different forms of renal autoimmune disease and might mediate tolerogenic mechanisms. However, whether these cells are bona fide Tr1 cells is not clear because the methods for identifying them relied mainly on the expression of IL-10 and lack of Foxp3 coexpression via flow cytometry or CD49b and LAG-3 costaining by immunofluorescence. Furthermore, whether these cells are able to suppress renal inflammation is unclear. Finally, the origin of Tr1 cells in the kidney is unknown. Indeed, Th17 cells can convert into Treg cells, referred to as Tr1^exTh17 cells in the intestine (21).

Therefore, we aimed to study the emergence and function of Tr1 cells in the context of GN. By doing so, we found that Tr1 cells emerge in the kidney, but only a small fraction of these cells are derived from Th17 cells. Moreover, functional in vitro as well as in vivo analysis confirmed that Tr1 cells have regulatory activity in the kidney in a mouse GN model. Finally, we showed the existence of Tr1 cells in human kidney biopsies. Thus, this study forms the basis to target Tr1 cells in patients with GN.

Mice were kept under specific pathogen-free conditions in the animal research facility of the University Medical Center Hamburg-Eppendorf. Food and water were provided ad libitum. Rag1^−/− mice were obtained from The Jackson Laboratory. Foxp3^mRFP, IL17a^eGFP, IL17A^Katushka, IL10^eGFP reporter mice and IL17A^Cre, Rosa26^YFP mice are described elsewhere (13, 21–24). Age- and sex-matched littermates between 8 and 12 wk old were used. All animals were cared for in accordance with the institutional review board Behörde für Soziales, Familie, Gesundheit und Verbraucherschutz (Hamburg, Germany), listed under animal protocol number 17/2012.

For the induction of the experimental crescentic GN (cGN; nephrotoxic nephritis), 8–12-wk-old male mice were injected i.p. with an antiserum raised against the glomerular basement membrane.

Mice were injected i.p. with 15 µg anti-CD3 specific Ab (self-made, clone 145-2C11) dissolved in 100 µl PBS. Injections were performed on days 8 and 10 after induction of GN. Mice were sacrificed 4 h after the second injection was applied.

First, isolated spleens and lymph nodes were smashed through a 100-µm cell strainer. Total cells were pelleted. For the transfer of CD4⁺ T cells, first CD4⁺ T cells were isolated via positive selection using MACS (Miltenyi Biotec, Bergisch Gladbach, Germany). APCs were isolated with the same system via CD4 and CD3 negative selection. To avoid proliferation during in vitro culture, APCs were irradiated with 30 Gy.

For Th17 cell proliferation, plates were coated with PBS containing 2 µg/ml anti-CD3 specific Ab. Plates were incubated overnight at 4°C or for 3 h at 37°C. A total of 1 × 10⁶ naive CD4⁺ T cells/ml isolated from reporter mice were cultured in full Click’s medium that was supplemented with 10% FBS, 1% l-glutamine, 1% penicillin/streptomycin, and 1:1000 2-ME. A total of 4 × 10⁶ irradiated APCs/ml, anti-CD3 specific Ab (3 µg/ml), anti-CD28 (2 µg/ml), TGF-β (0.5 ng/ml), IL-6 (10 ng/ml), IL-23 (20 ng/ml), anti-IFN-γ (10 µg/ml), and anti-IL-4 (10 µg/ml) were added to the full Click’s medium. Cells were cultured for 5 d at 37°C and 5% CO₂.

For Tr1 cell proliferation, 24-well plates were coated with PBS containing 2 µg/ml anti-CD3 specific Ab. Plates were incubated overnight at 4°C or for 3 h at 37°C. Prior to culture, the coating solution was removed completely. After depletion of CD25⁺ cells, 1 × 10⁶ CD4⁺ T cells/ml from reporter mice were resuspended in full Click’s medium that contained anti-CD28 (2 µg/ml) and IL-27 (30 ng/ml). Cells were cultured for 5 d at 37°C and 5% CO₂. The expression of Foxp3^mRFP and IL-10^eGFP was determined using flow cytometry.

FACS was performed on a BD FACSAria Illu or AriaFusion. Th17 cells were FACS sorted by negative expression of Foxp3^mRFP and positive expression for IL-17A^eGFP. Foxp3⁻ Tr1 cells were sorted for high expression of IL-10^eGFP. Foxp3⁺ Tregs cells were isolated from spleens of reporter mice and sorted according to Foxp3^mRFP expression. For the in vivo transfer, 1 × 10⁵ Th17 cells were either transferred alone or in combination with 2.5 × 10³ Foxp3⁺ Treg cells or 5 × 10⁵ Tr1 cells into Rag1^−/− mice. This transfer was performed 24 h prior to the induction of cGN.

Light microscopy was performed on paraffin-embedded, 1-µm-thick, 4% paraformaldehyde fixed kidney cross-sections stained with periodic acid-Schiff. The formation of crescents was assessed in 30 glomeruli per mouse in a blinded manner.

Spleens and lymph nodes were smashed through a 100-µm cell strainer and washed with PBS/1% FBS followed by centrifugation. Spleens were further processed for erythrocyte lysis using ammonium-chloride-potassium buffer.

Kidneys were crushed and incubated at 37°C for 45 min in RPMI medium containing 10% FBS, 200 µg/ml DNase I, and 2 mg/ml collagenase D. After incubation, the tissue was mechanically reduced to a pulp, and cells were pelleted. A 37% Percoll gradient centrifugation was performed. The remaining erythrocytes were lysed via ammonium-chloride-potassium buffer.

For the isolation of intraepithelial lymphocytes, 0.5-cm pieces of the small intestines were incubated in 10 ml RPMI medium containing 10% FBS and 1.5 mg DTT for 20 min at 37°C. For the isolation of lamina propria lymphocytes, digested tissue was cut to a pulp and transferred in 6 ml collagenase solution (RPMI medium containing 10% FBS, collagenase 100 U/ml, and DNase I 1000 U/ml) for 45 min at 37°C. Lamina propria lymphocytes were pelleted and pooled with intraepithelial lymphocytes. For further cell separation, a 40%/67% Percoll gradient was performed (GE Healthcare). The leukocyte-enriched interphase was isolated, washed, and processed for further staining steps.

In the case of reporter mice, cells were analyzed after surface staining. Fate⁺ mice, besides being YFP fate reporters for IL-17A, are also reporters for Foxp3^RFP, IL-10^eGFP, and IL-17A^Katushka (21). Both RFP (excitation 555 nm, emission 584 nm) and Katushka (excitation 588 nm, emission 635 nm) are excited by the 532-nm laser line. To properly analyze the expression of Foxp3 and IL17A, high compensation levels are required between both fluorophores. The spillover of RFP into Katushka is possible to correct. However, the spillover of Katushka into RFP would require high levels of compensation, which could affect other fluorochromes in the matrix. Thus, to correctly gate Foxp3⁺ and Foxp^neg cells, it is necessary to plot and gate the populations against each other. The diagonal in the plots corresponds to IL-17A⁺ cells, which has been corroborated by mRNA expression when generating these mice (21).

To enable intracellular staining, in the case of nonreporter mice, cells were stimulated for 3 h at 37°C with PMA (50 ng/ml; Merck Darmstadt) and ionomycin (1 mM; Sigma-Aldrich).

To discriminate dead from living cells, isolated cells were pelleted and stained with a fluorochrome-labeled violet DNA dye. Pacific Orange succinimidyl ester (Life Technologies, lot 2179293) was diluted (1:1000) in PBS and incubated for 30 min at 4°C. Cells were washed and pelleted.

Surface staining panel for flow cytometric analysis: CD11b (PE/cyanine 7 [Cy7], dilution 1:400; BioLegend, clone M1/70, lot B249268); CD11c (PE-Cy7, dilution 1:400; BioLegend, clone N418, lot B222652); CD195 (CCR5) (PE/Cy7, dilution 1:400; BioLegend, clone HM-CCR5, lot B224462); CD25 (BV650, dilution 1:200; dilution 1:100; BioLegend, clone PC61, lot B288551); CD3 (BUV379, BD, clone 17A2, lot 9080908); CD4 (PacBlue, dilution 1:600; BioLegend, clone RM4-5, lot B336509; BV650, dilution 1:400; BioLegend, clone RM4-5, lot B297638); CD45 (BV785, dilution 1:800; BioLegend, clone 30-F11, lot B336128); CD45.1 (allophycocyanin, dilution 1:400; BioLegend, clone A20, lot B209251); CD45.2 (PE/Cy7, dilution 1:400; BioLegend, clone 104, lot B307583); CD8α (PE/Cy7, dilution 1:400; BioLegend, clone 53-6,7, lot B295389); NK1.1 (PE/Cy7, dilution 1:400; BioLegend, clone PK136, lot B284829); CD279 PD-1 (BV 605, dilution 1:400; BioLegend, clone 29F.1A12, lot B227579); TCR-γδ (PE/Cy7, dilution 1:400; BioLegend, clone GL3, lot B222125); TIGIT (PerCP/Cy5.5, Dilution 1:400; eBioscience, clone GIGD7, lot 1988570); CD366 (TIM-3) (BV 421, dilution 1:400; BioLegend, clone RMT3-23, lot B259713). Surface staining was performed for 20 min at 4°C.

In the case of CD49b (PE, dilution 1:100; BioLegend, clone HMa2, lot B230820) and CD223 (LAG3) (allophycocyanin, dilution 1:100; BioLegend, clone C9B7W, lot B243438), the staining was performed for 30 min at 37°C.

In order to stain for intracellular markers, cells were fixed for 20 min with 4% formaldehyde solution, followed by permeabilization of the cell membranes by adding 0.1% Nonidet P-40 solution (Sigma-Aldrich) for 4 min. Last, a staining mixture with fluorochrome-labeled Abs for intracellular staining was added to the cells.

Intracellular staining panel for flow cytometric analysis: Foxp3 (allophycocyanin, dilution 1:80; eBioscience, clone FJK-16s, lot 2297415; PE, dilution 1:80; eBioscience, clone NRRF-30, lot 1927456); IFN-γ (BV785, dilution 1:100, BioLegend, clone XMG1.2, lot B343101; allophycocyanin, dilution 1:100; BioLegend, clone XMG1.2, lot B288855); IL-10 (PE-Dazzle, dilution 1:100; BioLegend, clone JES5-16E3, lot B265048); IL-17A (BV 421, dilution 1:100; BioLegend, clone TC11-18H10.1, lot B318238). Intracellular staining was performed for 1 h at room temperature.

Fluorochrome detection was performed on an LSR II flow cytometer using FACSDiva software. For analysis, data were exported from FACSDiva to FlowJo vX software for Macintosh or Windows.

Using MACS beads, responder cells (CD4⁺ CD25⁻) were isolated from the spleen and lymph nodes and labeled with 5 µM CellTrace Violet dye. In each well, 1.5 × 10⁴ responder cells together with 7.5 × 10⁴ APCs were plated in a 96-well round-bottomed plate. Next, 1 × 10⁴ Treg or Tr1 cells/well were added to the responder-APC mix isolated from the kidneys of nephritic mice. Additional soluble anti-CD3 specific Ab (1.5 µg/ml) led to further cell stimulation. After 96 h at 37°C in a 5% CO₂ incubator, the proliferation of responder cells was determined via flow cytometry by detecting the intensity of the violet dye per cell.

We induced cGN in 13 mice. Six of them were assigned to the anti-CD3 group, and the remaining seven mice were included in the control group. Foxp3^neg IL-10–producing CD4⁺ T cells from nephritic kidneys were sorted, and the cells of each group were pooled, so that we had one replicate per group for sequencing. The samples were processed with the Chromium Single Cell 3′ v2 kit according to the corresponding protocol of 10× Genomics. The libraries were sequenced on an Illumina NovaSeq 6000 system (S4 flow cell) with 150 bp and paired-end configurations.

Murine single-cell RNA-sequencing data were processed using CellRanger version 4.0.0. The further processing and downstream analysis of the single-cell data were done using R software version 4.1.1 (2021-08-10). The global seed was set to 0. Unless mentioned otherwise, methods were run with default parameters. The R-Package Seurat (version 4.1.0) was used for preprocessing, dimensional reduction, and cluster identification.

As part of the quality check, cells with the number of genes between >450 and <4000 (control) or >700 and <4000 (anti-CD3) were kept and with the percentage of mitochondrial genes <35%. Values were determined after visually inspecting the distribution. The two mouse datasets (control and anti-CD3) were integrated and scaled (with regression of library size and percentage of mitochondrial genes), and principal components (PCs) were computed according to Seurat’s pipeline. After Louvain clustering (PC 1 to 30, resolution 0.8) and Uniform Manifold Approximation and Projection (UMAP) visualization (PC 1 to 30), clusters with less than 150 cells, as well as one cluster expressing primarily ribosomal genes and another cluster expressing cell cycle genes, were removed. Finally, 10,802 cells (control 3,142, anti-CD3 7,660) were further analyzed. Next, reclustering (resolution 0.4 and PC 1 to 25) was performed, and a new UMAP was computed.

For the functional enrichment analysis, the fgsea method by the R-package fgsea (version 1.26.0) was used. The genes were ranked according to the log₂ fold change. GO Biological Process ontology annotation was received from https://www.gsea-msigdb.org/ as a gmt file (v2023.1). Moreover, we performed a gene set coregulation analysis to check the regulation of selected BPs on the single-cell level. Here, we applied the method plotCoregulationProfileReduction with default parameters.

The preprocessed and clustered human data were received from (25). The dataset was subset using only CD4⁺ clusters, annotated with Th1, Th17, CD4_Tcm, and Treg. Subsequently, the patients were reintegrated using methods FindIntegrationAnchords (k.filter = 50), IntegrateData (k.filter = 50), and cells were reclustered (resolution 0.5, PC 1 to 30).

Tr1 or CIR scores were calculated using Seurat’s method AddModulScore and the following mouse (or human equivalent) genes: Il10, Lag3, Havcr2, Pdcd1, Ctla4, Itga2, and Tigit. Significance between datasets was determined using the Wilcoxon rank-sum test.

Statistical evaluations were performed using GraphPad Prism software. Unpaired nonparametric Wilcoxon–Mann–Whitney U test, one-way ANOVA using Tukey multiple comparisons, or Kruskal–Wallis using Dunn multiple comparisons test was used for the evaluation of statistical significance. A p value < 0.05 was used to define significance.

As a first step, we wanted to analyze the infiltration of IL-10–producing CD4⁺ T cells during the time course of cGN. To that end, we used Fate⁺ reporter mice, which allowed us to analyze Foxp3, IL-10, and IL-17A expression simultaneously. Furthermore, because IL-17A⁺ cells are permanently marked with YFP, we can also identify IL-10–producing cells that have previously expressed IL-17A (21). Experimental cGN was induced in Fate⁺ mice, and we analyzed the kidneys throughout the course of the disease (Fig. 1A–1C). As a control, we analyzed the frequency of these cells under steady-state conditions (day 0). IL-10 expression was detected in Foxp3⁺ (Fig. 1D, 1E) as well as in Foxp3^neg CD4⁺ T cells (Fig. 1D, 1F). Both IL-10–producing T-cell subsets were below 1% of the total CD4⁺ T-cell population at steady state, but frequencies increased with the onset of the disease at day 3, reaching the peak at day 7 (Fig. 1E, 1F). However, within CD4⁺ T cells, the frequencies of Foxp3^neg IL-10–producing CD4⁺ T cells increased almost 10-fold (day 0, 0.5 ± 0.1 versus day 7, 3.8 ± 0.5), whereas Foxp3⁺ IL-10–producing CD4⁺ T cells increased only 5-fold (day 0, 0.5 ± 0.1 versus day 7, 2.4 ± 0.5) (Figure 1E, 1F). We furthermore found that a variable fraction (10–30%) of Foxp3^neg IL-10–producing CD4⁺ T cells were YFP⁺, indicating that they were Th17 cells or had expressed IL-17A before (Fig. 1G). However, the percentage of YFP⁺ cells within Foxp3^neg IL-10⁺ was independent of the induction of cGN.

Taken together, Foxp3^neg IL-10–producing CD4⁺ T cells emerge and expand 10 times during the course of cGN. Likewise, percentages of Foxp3⁺ IL-10–producing CD4⁺ T cells increase but only at ∼5-fold compared with steady-state conditions. Finally, ∼10%, on average, of Foxp3^neg IL-10–producing CD4⁺ T cells at some point produced IL-17A.

Next, we aimed to study whether we can further expand Foxp3^neg IL-10–producing CD4⁺ T cells in the kidney. To this end, we used anti-CD3-specific Ab treatment, which we have used before to suppress cGN (1) and which is known to induce IL-10 production in CD4⁺ T cells in the intestine (11, 17, 21, 24). Again, we induced experimental cGN in Fate⁺ mice that were treated with an Ab against CD3 on days 8 and 10 after disease induction (Fig. 2A). Kidney CD4⁺ T cells were separated into IL-10–producing Foxp3⁺ (Fig. 2B, 2C) and Foxp3^neg cells. Foxp3^neg IL-10–producing CD4⁺ T cells were further separated into Foxp3^neg YFP^neg cells (IL-10–producing cells, which did not emerge from Th17 cells) (Fig. 2B, 2D), Foxp3^neg YFP⁺ IL-17A⁺ (IL-10–producing Th17 cells) (Fig. 2B, 2E) and Foxp3^neg YFP⁺ IL-17A⁺ (exTh17 cells; IL-10–producing cells, which have produced IL-17A before) (Fig. 2B, 2F). In line with previous data studying the intestine (Supplemental Fig. 1), anti-CD3 Ab treatment strongly increased the numbers of IL-10⁺ cells in all analyzed subsets in the kidney (Fig. 2C–2F). Furthermore, we confirmed that some exTh17 cells expressed IL-10. Also, here the production of IL-10 was further promoted by anti-CD3 Ab treatment (Fig. 2F).

Next, we aimed to further characterize Foxp3^neg IL-10–producing CD4⁺ T cells in the kidney in an unsupervised approach. To this end, we used single-cell sequencing. We induced experimental cGN in Foxp3^RFP IL10^eGFP reporter mice and treated half of them with an Ab against CD3. The control group received PBS instead. Foxp3^neg IL-10–producing CD4⁺ T cells were subjected to FACS and then to single-cell sequencing using the 10× Genomics platform (Fig. 3A and Supplemental Fig. 2A). A total of 10,802 cells (control, 3,142 cells; anti-CD3, 7,660 cells) were analyzed. This analysis revealed a heterogeneous cell population that separated into seven distinct clusters (C1–C7) (Fig. 3B). Although most of the clusters showed a gene signature toward effector-like cells (Supplemental Fig. 2B, 2C), clusters 3 and 4 were enriched with CIRs (Fig. 3B, 3C, Supplemental Fig. 2D, 2E). The expression of differentially expressed and specific genes used for annotation within the clusters can be viewed in Supplemental Fig. 2. Moreover, we specifically focused on the identification of Tr1 cells and screened the genes to identify them. We employed a gene set composed of Il10 and CIR (Lag3, Havcr2, Pdcd1, Ctla4, Itga2, and Tigit) to compute a Tr1 score for CD4⁺ Foxp3^neg IL10⁺ T cells. This score was used to identify Tr1 cells based on our previous work (17). This analysis also revealed clusters 3 and 4 to express the highest Tr1 score (Fig. 3D). Additionally, the treatment with anti-CD3 specific Ab increased the frequency of CIR-expressing Foxp3^neg IL-10–producing CD4⁺ T cells. On this basis, we next looked at the clusters from the control and anti-CD3–treated groups separately and checked for the relative proportion of each cluster (Fig. 3C). Here, the additional treatment with an Ab against CD3 led to a relative shrink of clusters 1, 2, 6, and 7 but a relative increase of clusters 3, 4, and 5 (Fig. 3C). Moreover, also the Tr1 score was increased in cells derived from mice treated with anti-CD3 Ab (Fig. 3E). As expected, when differentially expressed genes were plotted comparing CIR-rich (C3,C4) versus CIR-poor clusters, many genes, such as Lag3, Ctla4, Pdcd1, and Havcr2, responsible for the bona fide Tr1 signature emerged (Fig. 3F, Supplemental Fig. 2G), and enrichment analysis of these genes revealed biological processes associated with regulation of inflammation (Supplemental Fig. 2F).

Taken together, we identified Foxp3^neg IL-10–producing CD4⁺ T cells in murine nephritic kidneys, which express the core transcriptional program of Tr1 cells (e.g., cluster 3 and 4). This fraction was further expanded upon anti-CD3 specific Ab treatment.

Having identified CD4⁺ T cells expressing the characteristic genes of Tr1 cells in the kidneys of nephritic mice, we then focused on identifying these cells via flow cytometry. Therefore, we specifically analyzed the expression of CIRs that in the single-cell sequencing analysis were shown to be expressed mainly by Tr1 cells, as shown by us before in the intestine (17). For this, we induced cGN in Foxp3^RFP IL10^eGFP reporter mice. On days 8 and 10, half of the mice received an injection of an Ab against CD3. Control mice received PBS injections instead (Fig. 4A). As shown before, treatment with anti-CD3 Ab increased the percentage of Foxp3^neg cells that expressed IL-10 (Fig. 4B). To identify CIR-rich cells within Foxp3^neg IL-10–producing CD4⁺ T cells, we analyzed the coexpression of IL-10 and CD49b, LAG-3, TIGIT, and TIM-3 via flow cytometry (Supplemental Fig. 3). We identified Foxp3^neg IL-10–producing CD4⁺ T cells that indeed additionally expressed CIRs either alone or in combination with other markers (Fig. 4C–4E). The most common combination at steady state corresponded to CD49b, LAG-3, TIM-3, and TIGIT. In correlation with the results from Fig. 3, anti-CD3 Ab treatment also strongly induced the frequency of CIRs (Fig. 4D, 4E). The four-marker combination was again the most expanded, followed by the combination of CD49b, LAG-3, and TIM-3 and then the double combination of CD49b and LAG-3. Other combinations also increased their proportion under the anti-CD3 treatment (Fig. 4E).

Thus, anti-CD3 specific Ab treatment induced Foxp3^neg IL-10–producing CD4⁺ T cells. These cells represent a heterogeneous population, some of which express CIRs, the extracellular markers characteristic of Tr1 cells.

On the basis of the above-mentioned data, we next aimed to evaluate the suppressive capacity of kidney-derived Foxp3^neg IL-10–producing CD4⁺ T cells. To this end, we first used an in vitro assay. CD4⁺ CD25-depleted cells (responder cells) were isolated from the spleens of mice at steady state and stained with a proliferation dye. During cell division, the intensity of the dye is halved in both daughter cells. To obtain IL-10–producing T cells, experimental cGN was induced in Foxp3^RFP IL-10^eGFP reporter mice, and Foxp3⁺ CD4⁺ T cells, as well as Foxp3^neg IL-10–producing CD4⁺ T cells, were sorted out from the kidneys 10 d after disease induction, and their suppressive function was assessed in vitro (Fig. 5A). As a negative control, Treg cells were exchanged with the same numbers of nonsuppressive CD4⁺ responder cells. The dye was measured after 5 d using flow cytometry (Fig. 5B). When the violet dye intensity was analyzed in the responder cells, the negative control did not show any suppression but rather enhanced proliferation of responder cells (Fig. 5B, 5C). The addition of kidney-derived Foxp3^neg IL-10–producing CD4⁺ T cells inhibited proliferation of almost 85% of the responder cells, which was comparable to the one observed by Foxp3⁺ Tregs (Fig. 5B, 5C).

Taken together, ex vivo isolated Tr1 cells from the kidney could suppress the proliferation of CD4⁺ responder T cells in vitro to a degree similar to that of classical Foxp3⁺ Tregs.

Tr1 cells are known to be efficient regulators of CD4⁺ T-cell–driven inflammation in the intestine (11, 17). Thus, having identified the presence of bona fide Tr1 cells within the IL-10⁺ Foxp3^neg population in the kidneys of nephritic mice, we next aimed to evaluate their suppressive capacity in the context of cGN. Of note, Th17 cells have been shown to be pathogenic on their own and to promote GN when transferred into these mice (1, 5). Thus, we designed an in vivo suppression assay by transferring Th17 cells plus Tr1 cells into Rag1^−/− mice (Fig. 6A). Because a high cell number is required for these experiments, both Th17 and Tr1 cells were differentiated in vitro. Foxp3⁺ Treg cells isolated from the spleens of untreated mice at steady state were used as a positive control (Fig. 6B). Twenty-four hours after the T cell transfer, GN was induced in the mice, and the survival rate and disease severity were evaluated (Fig. 6C). Although not significant, mice that received only Th17 cells showed a relative increase in mortality compared with mice that received additional regulatory cells (Fig. 6D). Also, the glomerular damage analyzed in kidney cross-sections revealed the highest score in mice that received Th17 cells alone (Fig. 6E, 6F). Both groups that received either Tr1 cells or Foxp3⁺ Tregs cells displayed not only increased survival but also less crescent formation (Fig. 6D–6F).

In conclusion, in vitro differentiated Tr1 cells were sufficient to suppress Th17 cell–mediated GN in vivo.

Finally, we aimed to translate these findings into human biology. To this end, we took advantage of already available single-cell data from kidney biopsies isolated from patients with ANCA (Fig. 7A) (25). UMAP analysis identified eight different clusters within human-derived CD4⁺ T cells from the kidney biopsies (Fig. 7B). We then looked at the expression of Foxp3 and CIRs within the clusters. We specifically aimed for the identification of Foxp3^neg CD4⁺ T cells that express at least two surface markers identifying Tr1 cells (Fig. 7C). Through this approach, we identified a large number of cells that expressed at least two or more CIRs at the same time and could correspond to bona fide Tr1 cells. Next, we identified cells that showed an increased CIR score that was previously defined by simultaneous expression of the genes for LAG3, ITGA2, HAVCR2, PDCD1, and TIGIT. This score was high in clusters 2 and 5 and relatively increased in cluster 7 (Fig. 7D). By further analyzing the expression levels for genes encoding for IL10, CTLA4, FOXP3, and IL2RA, we identified cluster 5 to be enriched in Foxp3⁺ Tregs (Fig. 7E). Thus, we were able to identify clusters 2 and 7, which showed low gene expression of FOXP3 and IL2RA, as clusters corresponding to Tr1 cells. These results are in correlation with our previous results in mice, where we could observe a high heterogeneity in Foxp3^neg IL-10–producing T cells.

These data indicate that bona fide Tr1 cells are also present in the kidney in patients with ANCA nephritis.

This study aimed to address the potential of Tr1 cells to be used as T-cell–based therapy for cGN. To this end, we analyzed the emergence of Tr1 cells, identified their molecular profile, and tested their functionality in the context of GN. We found that Tr1 cells emerge, albeit at low numbers, during experimental GN, and we proved their high level of suppressive activity in vitro. Importantly, Tr1 cell transfer was protective in a mouse cGN model. Thus, our study sets the premise to therapeutically target Tr1 cells in cGN.

Foxp3⁺ Tregs have been described as being able to ameliorate kidney injury by suppressing Th17 cell–mediated pathology via IL-10 (10). Their power to stabilize immune homeostasis was found to be common for different organs and diseases, such as kidneys and the intestine (11). It has been reported before that IL-10–producing Foxp3⁺ Tregs are important to suppress kidney injury in cGN (10). In the present study, we showed that Foxp3^neg Tregs, namely Tr1 cells, are able to produce high amounts of IL-10 and to suppress nephritis. This is in line with previous data showing that besides Foxp3⁺ Tregs, Tr1 cells can also suppress Th17 cell–driven colitis (11). Tr1 cells have already been described in other autoimmune kidney diseases (18, 19). However, in these studies, the identification of Tr1 cells was based solely on the expression of IL-10 and the lack of Foxp3. Thus, the role of bona fide Tr1 cells remained to be elucidated. To answer this question, we first studied IL-10–producing CD4⁺ T cells, including Tr1 cells, in the kidney. Even under steady-state conditions, we could detect Tr1 cells in the kidney, confirming that these cells may not be present only in the intestine (26). Indeed, as the disease progressed, frequencies of IL-10 strongly increased in both Foxp3⁺ and Foxp3^neg CD4⁺ T cells. As mentioned before, in the disease setting, the number of Foxp3^neg IL-10–producing CD4⁺ T cells in the kidney was expanded ∼10 times, whereas Foxp3⁺ IL-10–producing Tregs increased 5-fold.

We next aimed to determine the origin of the Foxp3^neg IL-10–producing CD4⁺ T cells identified in the inflamed kidney. Of note, it had been shown before that Th17 cells can coproduce IL-10. Furthermore, intestinal Th17 cells were shown to be able to completely transdifferentiate into Tr1 cells (referred to as Tr1^exTh17), a process called “T cell plasticity” (21). Although Th17 cells in the brain of experimental autoimmune encephalomyelitis (EAE) mice can convert to Th1 cells, Th17 cells in the kidney have been described to be more stable in terms of their cytokine expression and not to convert to Th1 cells to a large extent (1). To investigate whether Th17 cells found in the kidneys of nephritic mice were able to convert to Tr1 cells, we used Fate⁺ mice. We found that a fraction of Foxp3^neg IL-10–producing CD4⁺ T cells originated from Th17 cells. Thus, we demonstrated Th17 cell plasticity toward Tr1 cells. By tracking Th17 cells, we also identified cells coproducing IL-17 and IL-10. However, the number of Tr1^exTh17 cells was low.

To further characterize Foxp3^neg IL-10–producing CD4⁺ T cells in an unsupervised way, we performed single-cell RNA sequencing of cells isolated from the kidneys of mice with GN and nephritic mice that were additionally treated with an anti-CD3 Ab. This approach confirmed that a fraction of Foxp3^neg IL-10–producing CD4⁺ T cells from the kidney showed the transcriptional signature of a Tr1 cell in the inflamed kidney. For both samples, the clustering of the total cells displayed a heterogeneous cell population. Although some clusters revealed gene expression levels associated with effector cells, others displayed a regulatory-like gene expression pattern. This is in line with previous data showing that in the small intestine and the spleen, there was a strong heterogeneity of IL-10–producing CD4⁺ T cells (17). Recently, it was shown that IL-10 arising from CD4⁺ T cells can have a pathogenic role in EAE by promoting the survival of effector T cells (27). Although one could speculate about a similar role in the kidney, especially considering the presence of proinflammatory-like IL-10⁺ CD4⁺ T cell clusters, it is unknown whether these cells are pathogenic in nephritis. Furthermore, in both EAE and experimental cGN, a complete lack of IL-10 results in increased disease severity (10, 27). Thus, further studies will be essential to decipher the role of Tr1 cells, which suppress nephritis, from other cellular sources of IL-10, which may promote disease.

To build on these descriptive data, we aimed to gain confirmation regarding whether kidney-derived Foxp3^neg IL-10–producing CD4⁺ T cells would fulfill the ultimate criterion defining Tr1 cells (i.e., regulatory activity). We performed in vitro suppression assays with Foxp3^neg IL-10–producing CD4⁺ T cells from the kidneys of nephritic mice showing that these cells indeed have a regulatory function in vitro. We have shown before that the in vitro suppressive capacity of Tr1 cells isolated from the intestine is mediated via IL-10, TGF-β1, and CTLA-4 (11). Unfortunately, due to the limited number of Tr1 cells that could be isolated from the kidney, we were unable to test the relevance of these factors for the suppressive capacity of Tr1 cells isolated from the kidney. Nonetheless, one could envision using these cells as a means of T-cell–based therapy.

To form the basis for this, one needs to further show their regulatory function in vivo. Indeed, during ongoing kidney inflammation, Foxp3⁺ Tregs are present and increase in numbers during disease progression (9). These cells are able to suppress CD4⁺ T-cell–driven inflammation in the kidneys (9, 10, 28). Thus, we wanted to go further and determine the suppressive capacity of Tr1 cells in an in vivo suppression assay combined with GN. To avoid confounding factors and to be able to study the interaction between pathogenic and regulatory CD4⁺ T cells only, we performed a transfer experiment of Th17 cells alone or together with Tr1 cells and Foxp3⁺ Tregs, respectively, into Rag1^−/− prior to induction of cGN. To this end, Treg cells first had to be generated in vitro because it was not possible to obtain a sufficient number of Tr1 cells from the kidneys. Moreover, because the relative contribution of each IL-10⁺ cluster was not defined, we performed these experiments by transferring the whole IL-10⁺ Foxp3^neg population. Strikingly, our data indicate that in these conditions, Tr1 cells are capable of suppressing Th17 cell–mediated GN. Thus, our data form the basis to establish a Tr1-based T-cell therapy for patients with GN, similar to what is already done in patients with inflammatory bowel disease (29, 30).

Finally, to further support this, we aimed to prove the existence of bona fide Tr1 cells in human kidneys. For this, we took advantage of available single-cell transcriptome data from patients with ANCA to establish an identification method for human Tr1 cells (25). We focused on CD4⁺ T cells that expressed CIRs and other surface markers. We shed light on the population of IL-10–producing CD4⁺ T cells and finally showed the gene expression pattern of Tr1 cells. The results of this study prove that Tr1 cells are present in the human kidney. However, and in agreement with the mouse data, there is a high degree of heterogeneity in IL-10⁺ CD4⁺ T cells. The contribution of each cluster to the regulatory function or whether they represent a source of pathogenic IL-10 remains to be studied.

In conclusion, we identified Tr1 cells in mouse and human diseased kidneys. These cells have strong suppressive activity in a mouse model of cGN. Thus, our data form the basis not only to identify Tr1 cells in patients with cGN but also to potentially develop personalized T-cell therapies for patients with cGN.

The authors have no financial conflicts of interest.

We thank the FACS Sorting Core Facility and the Single Cell Core Facility of the Universitätsklinikum Hamburg-Eppendorf for their support. We thank Cathleen Haueis, Sandra Wende, and Anett Peters for excellent technical assistance.