Macrophage Niche Reconstitution Reveals Dynamic Transcriptional and Communication Networks Renal Macrophage-Epithelial Communication
Mohammad Islamuddin, Lixuan Ji, Yilin Chen, Kejing Song, Calder R Ellsworth, Jack D Rappaport, Chenxiao Wang, Shumei Liu, Jay Kolls, Xiaojiang Xu, Xuebin Qin
TL;DR
This study shows how kidney macrophages regenerate after depletion, revealing communication networks between macrophages and epithelial cells that drive tissue repair.
Contribution
The study identifies epithelial-derived chemokines and macrophage-specific signaling networks that orchestrate niche restoration in the kidney.
Findings
Proximal tubule epithelial cells are the main source of CX3CL1, which drives macrophage recruitment and maintenance.
Regenerating macrophages show a transient injury-adaptive program with metabolic activation and reduced inflammatory signaling.
Spp1/osteopontin is essential for efficient macrophage regeneration following depletion.
Abstract
Renal-resident macrophages (RMs) are essential regulators of kidney homeostasis and repair, yet the cellular and molecular mechanisms governing RM niche regeneration after acute depletion remain poorly defined. How epithelial-immune interactions coordinate RM repopulation is particularly unclear. We employed an inducible hCD59 intermedilysin (ILY) ablation system to achieve rapid and specific depletion, and subsequent replenishment of RMs, followed by longitudinal single-cell RNA sequencing (scRNA-seq) of kidneys at baseline and days 1, 3, and 7 post-ablation. Integrated transcriptomic, pathway, transcription factor, and cell-cell communication analyses were combined with functional validation using clodronate-mediated macrophage depletion in Spp1 (Opn)-deficient mice. Acute ILY-mediated ablation resulted in rapid and selective RM depletion, followed by robust regeneration reaching ~…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7- —NIH, X.Q., X.Q, Tulane start-up funds, XQ
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsRenal and related cancers · Chronic Kidney Disease and Diabetes · Single-cell and spatial transcriptomics
Background
Renal macrophages (RMs) play indispensable roles in maintaining kidney homeostasis, orchestrating immune surveillance, and mediating responses to injury and repair[1, 2]. These cells exist as a heterogeneous population derived from embryonic yolk sac progenitors, fetal liver monocytes, and bone marrow-derived monocytes, contributing to distinct resident and infiltrating macrophage pools[3-6]. Their maintenance and phenotype are tightly regulated by local cues from tubular epithelial cells, endothelial cells, and extracellular matrix (ECM) components[7, 8]. Following kidney injury, macrophages act as central effectors of inflammation, resolution, and regeneration, dynamically shifting from pro-inflammatory to reparative states[9-11]. This phenotypic versatility enables RMs to coordinate immune cell recruitment, matrix remodeling, and epithelial recovery[12]. Recent advances in single-cell RNA sequencing (scRNA-seq) have revolutionized our understanding of RM heterogeneity, revealing distinct subsets such as Ccr2^+^Ly6C^+^ inflammatory macrophages, Fn1^+^ remodeling macrophages, and Pglyrp1^+^Ace^+^ immunomodulatory cells[12-18]. These subsets engage in temporally structured communication with proximal tubular epithelial (PTE) cells and other immune populations through defined ligand-receptor pairs most notably Fn1-Cd44, Fn1-Sdc4 and Spp1-Cd44, to coordinate ECM remodeling, immune regulation, and epithelial regeneration[17-20]. At the transcriptional level, repair-associated regulators such as Klf4, Irf8, and E2f families integrate microenvironmental signals (hypoxia, cytokines, and metabolic stress) to modulate macrophage polarization, proliferation, and reparative signaling[21-24]. Macrophage-derived growth factors, including Csf-1 and Spp1 (osteopontin), further promote tubular cell proliferation and differentiation, underscoring macrophage epithelial reciprocity in renal repair[25-27]. Despite these insights, the temporal and mechanistic hierarchy linking macrophage depletion, niche repopulation, transcriptional activation, and intercellular communication during kidney injury remains undefined.
Addressing this knowledge gap requires a system that can acutely and specifically perturb the macrophage compartment while preserving the renal microenvironment for unbiased molecular interrogation. Historically, experimental macrophage depletion strategies, including clodronate liposomes, irradiation, and diphtheria toxin receptor (DTR) models have provided mechanistic insights into macrophage biology but are constrained by incomplete ablation, off-target toxicity, and compensatory hematopoietic activation[28-30]. To overcome these limitations, we have developed an inducible human CD59-intermedilysin (hCD59-ILY) ablation system, enabling rapid, specific, and reversible depletion of targeted macrophage populations[5, 31, 32]. The ILY toxin from Streptococcus intermedius binds exclusively to hCD59, forming pores that lyse expressing cells within seconds, allowing temporally controlled depletion[32, 33]. This unique system facilitates dynamic tracking of macrophage regeneration, offering a clean platform to probe the transcriptional and cellular events underpinning RM niche restoration. Here, we employed our hCD59-ILY ablation platform in combination with high-throughput scRNA-seq and cell-cell communication mapping to characterize the temporal kinetics, transcriptional architecture, and signaling hierarchy underlying RM depletion and repopulation. We demonstrated temporally distinct RM subsets and signaling pathways-including Spp1-Cd44 and Fn1-integrin axes that orchestrate immune-epithelial crosstalk during niche restoration and linked 18 upregulated transcription factors to interleukin-1 responses, myeloid differentiation, and tissue remodeling for a transcriptionally coordinating RM regeneration.
Materials and methods
Animal model strains and their maintenance
Cx3cr1CreER^+/+^ mice (JAX 021160) were obtained from The Jackson Laboratory (Bar Harbor, ME) and housed at the Tulane University School of Medicine. The ihCD59^+/+^ mice, previously generated and backcrossed onto a C57BL/6 genetic background for at least seven generations[32], were used for breeding. Homozygous Cx3cr1CreER+/+ mice are deficient in CX3CR1 expression (Cx3cr1 knockout), whereas heterozygous Cx3cr1CreER^+/−^ mice retain functional CX3CR1 expression. To generate experimental genotypes, Cx3cr1CreER^+/+^ mice were crossed with ihCD59^+/+^ mice to produce the following offspring: Cx3cr1CreER^+/−^/ihCD59^+/−^. All animal experiments were conducted in accordance with the ethical guidelines of the Institutional Animal Care and Use Committee (IACUC) at Tulane University (Protocol number 1482). Mice were housed in a specific pathogen-free (SPF) facility at the Tulane University School of Medicine, maintained under a 12-hour light/dark cycle with controlled environmental conditions. Detailed experimental protocols, including intermedilysin purification, tamoxifen treatment, RM depletion and regeneration, sample collection schedules, flow cytometry staining and acquisition. RNA isolation and qRT-PCR, scRNA-Seq by 10x genomics and data analysis, and statistical analysis are provided in the Supplementary Methods.
Intermedilysin purification.
Recombinant His-tagged intermedilysin (ILY) was purified using the HisBind Purification Kit (EMD 70239) following the manufacturer’s instructions, as previously described[31, 32]. Briefly, bacterial cultures expressing His-tagged ILY were lysed, and the protein was captured using a nickel-affinity chromatography column. After washing to remove non-specifically bound proteins, the His-tagged ILY was eluted under optimized conditions. The concentration and purity of the purified ILY were assessed by SDS-PAGE, ensuring adequate purity for downstream applications.
Tamoxifen Treatment, RM Depletion, and Regeneration.
Tamoxifen (Sigma-Aldrich) was dissolved in corn oil at a concentration of 20 mg/mL. To induce Cre-mediated hCD59 expression in Cx3cr1CreER^+/−^/ihCD59^+/−^ mice, 10-12-week-old mice were administered tamoxifen at a dose of 100 μg/g body weight via intraperitoneal (i.p.) injection for 3 consecutive days. After a 15-day waiting period post-tamoxifen treatment, Cx3cr1CreER^+/−^/ihCD59^+/−^ mice and Cx3cr1CreER^+/−^/ihCD59^−/−^ littermate controls were injected with a single dose of intermedilysin (ILY) at 120 ng/g body weight. RM (RM) ablation and subsequent regeneration were monitored and confirmed by flow cytometry at 1-, 3-, and 7-days post-ILY administration[5, 6].
Preparation of single cells from mouse tissue.
Mice were euthanized using carbon dioxide (CO_2_) asphyxiation followed by perfusion with phosphate-buffered saline (PBS) to remove blood from tissues. Kidneys were harvested post-perfusion and processed for single-cell suspension preparation. Tissue dissociation was initiated by mechanical disruption of the kidney, followed by enzymatic digestion. The dissected kidney was incubated in 10 mL of Hank's Balanced Salt Solution (HBSS), containing 0.75 mg/mL collagenase IV (Worthington Biochemical) and 20 μg/mL DNase I (Sigma-Aldrich), for 40 minutes at 37°C. After digestion, the cell suspension was passed through a 40 μm cell strainer to remove debris and larger tissue fragments, and the resulting single-cell suspension was maintained on ice. The cell suspension was then washed twice with PBS by centrifugation at 500 x g for 10 minutes. To enrich hematopoietic cells, the cells were subjected to density gradient centrifugation using Percoll. The pellet obtained from the previous wash was resuspended in 10 mL of 30% Percoll solution and carefully layered over 3 mL of 70% Percoll solution. The gradient was centrifuged at 500 x g for 30 minutes without the application of a brake. Following centrifugation, the interphase layer between the 30% and 70% Percoll solutions, containing enriched single cells, was carefully collected. The collected cells were washed twice with 10 mL of PBS (500 x g, 10 minutes). The final cell pellet was resuspended in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) for downstream single-cell RNA sequencing (scRNA-seq) analysis. Cell number and viability were assessed using an automated cell counter, yielding a viability rate of approximately 84%, suitable for scRNA-seq. Simultaneously, a portion of the cells post-density gradient centrifugation was prepared for flow cytometry. The cells were washed twice with 10 mL of PBS (500 x g, 10 minutes) and then resuspended in 100 μL of flow cytometry staining buffer (PBS with 2% FBS). The suspension was further processed for antibody staining following standard protocols for flow cytometric analysis [5].
Clodronate Liposome-Mediated Depletion and Repopulation Analysis of RMs in Wild-Type and Osteopontin Knockout Mice
Ten- to twelve-week-old C57BL/6 wild-type (B6 WT) and osteopontin knockout (Opn−/−) mice were used in all experiments. To deplete RMs, mice received an intraperitoneal (i.p.) injection of 200 μL clodronate liposomes (LIPOSOMA, Netherlands). Control animals received an equal volume of PBS. Mice were sacrificed at 2- and 7-days and 9 days post-injection (DPI) to assess depletion and repopulation, respectively. Kidneys were perfused with PBS, minced, and digested with collagenase D and DNase I for 30 min at 37°C.RMs were identified as CD45^+^ CD11b^+^ F4/80hi by flow cytometry.
Flow cytometry staining, acquisition.
The single-cell suspensions prepared from mouse kidney tissue were processed for flow cytometric analysis. To block non-specific binding to Fc receptors, cells were incubated with anti-CD16/32 antibody (FcγRIII/II, Clone 93, Cat# 48-0161-80, eBioscience) at a dilution of 1:200 for 15 minutes at room temperature. Aqua live/dead dye (Invitrogen, Cat# L34957A) was used to distinguish live from dead cells according to the manufacturer’s instructions. For phenotypic analysis of RM ablation, cells were stained with the following pre-conjugated antibodies at a 1:100 dilution: CD45-e450 (Clone 30-F11, eBioscience, Cat# 48-0451-82), CD11b-PE-Cy7 (Clone M1/70, Invitrogen, Cat# 25-0112-82), hCD59-PE (Clone OV9A2, Invitrogen, Cat# 12-0596-42), and F4/80-BV605 (Clone BM8, BioLegend, Cat# 123133). The antibody cocktails were added to the cell suspensions, and samples were incubated for 30 minutes at 4°C in the dark to protect fluorophores from photobleaching. Following staining, the cells were washed twice with flow cytometry buffer (PBS containing 2% fetal bovine serum). Cells were then fixed in 1% paraformaldehyde (PFA) for 30 minutes on ice, followed by two additional washes with flow cytometry buffer. The stained and fixed cells were acquired using a BD LSR Fortessa flow cytometer, and data were analyzed using FACSDiva software, version 6.1.3. (BD: https://www.bdbiosciences.com/en-ca/products/instruments/softwareinformatics/instrument-software/bd-facsdiva-software-v-6-1-3.643629.
RNA isolation and qRT-PCR
Total RNA was isolated from mouse kidney tissue using the RNeasy Plus Mini Kit (Cat#74134; Qiagen, Germany) following the manufacturer’s protocol. RNA integrity and quantity were assessed using spectrophotometric analysis. For reverse transcription, 1 μg of RNA was used as input for cDNA synthesis, performed with the High-Capacity cDNA Reverse Transcription Kit (Cat# 4368814; Applied Biosystems, USA)[11]. The reaction was carried out according to the manufacturer’s instructions. Quantitative PCR (qPCR) was conducted on the synthesized cDNA using Perfecta SYBR Green FastMix Low ROX (Cat# 95074-250; QuantaBio, USA) as the detection chemistry. Amplification was performed using the QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific, USA). The cycling conditions included an initial denaturation at 95°C for 30 second, followed by 40 cycles of denaturation at 95°C for 5 seconds and annealing/extension at 60°C for 5 seconds. Target gene expression levels were normalized to the housekeeping gene Ribosomal Protein L32 (Rpl32). The relative expression levels of target genes were calculated using the 2^-ΔΔCT method. The following qPCR primers were used, synthesized by Integrated DNA Technologies (IDT, USA): Cx3cl1 (forward: 5'-GGACAAGCCACATAGGAAAGA, reverse: 5'-CACATGCACAAGTCCCTACA) Gene expression data were analyzed and plotted to determine relative fold changes in mRNA levels.
scRNA sequencing by 10x genomics
scRNA-seq was performed using the 10× Genomics Chromium platform to analyze transcriptomic profiles of kidney cells. A target of 5000 viable cells per sample was set based on prior optimization using 10× Genomics Single Cell 3' RNAseq technology (10x Genomics). Live cells were quantified and sorted before processing, ensuring the cell viability exceeded 80% for downstream analysis. Following cell capture, full-length barcoded cDNA libraries were generated from each sample according to the manufacturer's protocol. Briefly, individual cells were partitioned into droplets containing reagents for reverse transcription, generating unique cDNA barcodes for each cell. The cDNAs were then amplified via polymerase chain reaction (PCR) to obtain sufficient material for library preparation. The cDNA libraries were pooled and diluted to a final concentration of 1.8 pM for sequencing. The pooled libraries were sequenced using an Illumina NextSeq 2000 platform, with paired-end single index reads. Sequencing output was analyzed to ensure adequate coverage and depth for each sample. Raw sequencing data were processed using Cell Ranger version 7.1.0 (10× Genomics) to perform alignment to the mouse reference genome, read counting, and the generation of cell-specific gene expression matrices. To identify differentially expressed genes across cell clusters, Loupe Cell Browser (10× Genomics) was used.
scRNA data cell identification and annotation of cell clusters
We performed a standard sequence of filtering, highly variable gene selection, dimensionality reduction, and clustering were performed using the scRNA-seq analysis R (version 4.3.0)-based package Seurat (version 5.0.0) for quality control and downstream analysis. After alignment and initial preprocessing, we began our workflow with a total of 32,285 genes across 31,100 cells from single-cell RNA sequencing (scRNA-seq) on the 10X Genomics Chromium Platform using kidneys from Cx3cr1CreER^+/−^/ihCD59^+/−^ mice treated with ILY. For the next QC process, we filtered out cells with gene counts (nFeature) of less than 50 and greater than 8,000 and with a UMI (nCount) of greater than 40,000 to remove low quality cells and possible multiple captures. We also eliminated low-quality cells with greater than 50% mitochondrial gene expression. A total of 21,181 genes across 27,396 single cells were included in the downstream analysis after applying the quality control criteria, which include 19,564 genes across 4,782 cells in Normal kidney group, 19,479 genes across 5,004 cells in D0 group, 19,161 genes across 5,216 cells in D1 group, 19,358 genes across 5,830 cells in D3 group and 19,341 genes across 6,564 cells in D7 group. Normalization was performed using the Seurat package to reduce biases introduced by technical variation, sequencing depth, and capture efficiency. A correlation analysis and principal component analysis were performed and uniform manifold approximation and projection (UMAP) was used to classify cells into different cell clusters at a proper resolution. Batch effect were corrected using “rPCA” from Seurat. Highly variable genes were identified by iterative selection based on the dispersion versus average expression of these genes. Principal component analysis was used for dimension reduction, and the top 30 principal components were selected by a permutation-based test implemented in Seurat and passed to UMAP to visualize cell clusters. Cell cluster annotation was conducted based on the known biomarkers of different cells in the kidney as published in previous studies[9, 34-36]. DEGs among different clusters were identified to validate the reasonableness of cell cluster annotation.
Identification of DEG and Hallmark pathways using Gene Set Enrichment Analysis
Differentially expressed genes (DEGs) were identified using the FindMarkers function in Seurat (test.use = Mast). P-adjustive-value < 0.01, min.pct = 0.01 and logfc.threshold = 0.01 were used as thresholds for significant differential expression. Gene Set Enrichment Analysis (GSEA) was then performed via GSEA function in package clusterProfiler (4.4.4) to clarify the of RM recruitment and dynamic biological change in different experiment time points based on DEG. Human MSigDB Collections (https://www.gsea-msigdb.org/gsea/msigdb/mouse/collection_details.jsp) is utilized as database. To identify genes with consistent transcriptional changes across experimental conditions, differential expression results from each cell population (Macro/PTC1/PTC2/Cx3cr1 + cells) were processed as follows. First, low-confidence or predicted genes were removed by excluding gene symbols beginning with Gm-, mitochondrial genes (mt-), and genes ending with Rik, as implemented in the filtering step of the analysis pipeline. For each gene that passed filtering, the log2 fold-change (log2FC) values from the three comparisons (D1_vs_D0, D3_vs_D0, and D7_vs_D0) were averaged to obtain a mean log2FC representing the overall transcriptional trend. Genes were then separated into upregulated (mean log2FC > 0) and downregulated (mean log2FC < 0) groups and ranked within each direction based on the magnitude of the averaged log2FC. For visualization, we selected the top 30 upregulated and top 30 downregulated genes from each cell population for visualization. These ranked gene sets were used for producing the comparative heatmaps displayed in the main and supplementary figures.
Cell-cell communication analysis
CellChat(v2.1.0)(https://github.com/jinworks/CellChat) is utilized with in R programming environment for quantitatively characterizing and comparing the inferred cell-cell communication networks using an integrated approach by combining social network analysis, pattern recognition, and manifold learning approaches. A database of known cytokine/chemokine receptor and ligand pairs was constructed using combined information from CellTalkDB v2 (http://tcm.zju.edu.cn/celltalkdb/), which contains ~ 3,300 validated molecular interactions, including ~ 40% of secrete autocrine/paracrine signaling interactions, ~ 17% of extracellular matrix (ECM)-receptor interactions, ~ 13% of cell-cell contact interactions and ~ 30% non-protein signaling, here we just included Secreted Signaling, ECM-Receptor, Cell-Cell Contact and Non-protein Signaling. To ensure the analysis’s relevance and accuracy, only ligand-receptor pairs exhibiting a p-value less than 0.05 were considered, allowing for a focused evaluation of the intricate relationships between diverse cell types.
Transcription factor regulatory network analysis
pySCENIC (version 0.11.2) ( https://github.com/aertslab/pySCENIC), a computational approach for predicting critical regulators and identifying cell states from scRNA-seq data, is conducted for accessing the transcription factor (TF) regulons among all sub-cell types of RM and PTC cells across five-time experimental points. We filtered out genes expressed in fewer than 1% of cells and only keep Top 100 TFs with p-value < 0.005.TFannotations were obtained from the cisTarget database(https://github.com/aertslab/create_cisTarget_databases). This approach provided insights into the regulatory mechanisms driving cellular states.
Statistical Analysis
Data were presented as the mean ± standard error of the mean (s.e.m.) to summarize central tendency and variability. For statistical comparisons between multiple experimental groups, a one-way analysis of variance (ANOVA) was employed. In instances where experimental groups were compared over time or across multiple conditions, a two-way ANOVA was performed. For pairwise comparisons between two groups, an unpaired Student’s t-test was used to evaluate the significance of differences in group means. The unpaired version of the t-test was chosen as data were assumed to be independent between the two groups. A threshold for statistical significance was set at p < 0.05. Values that did not reach this threshold were considered not statistically significant and marked as "n.s." to indicate p > 0.05. All statistical analyses were conducted using appropriate software, such as GraphPad Prism and or R, to ensure robust and reproducible results.
Results
Single-cell RNA-seq reveals rapid renal macrophage ablation, regeneration, and coordinated kidney-wide transcriptional remodeling.
To effectively and swiftly target the RM while minimizing unintended effects on other tissues, we followed our previously established method[5, 6] to administer ILY 15 days post-tamoxifen induction in Cx3cr1CreER+/−/ihCD59^+/−^ mice (Fig. 1a) [5, 6]. By flow cytometry analysis (Fig. 1b), we found that the depletion of RMs was effectively and specifically achieved one day following ILY-injected Cx3cr1CreER^+/−^/ihCD59^+/−^ mice but not vehicle-injected Cx3cr1CreER^+/−^/ihCD59^+/−^ mice (Fig. 1b-c). The population begins to replenish by day 3 post-injection, reaching approximately 25% of its initial level and by day 7 reached approximately 75% (Fig. 1b). This result is consistent with our previous finding that we can specifically manipulate RMs by rapidly ablating RMs and monitoring their regeneration in mice[5, 6]. Then, we conducted scRNA-seq on ILY-treated and control Cx3cr1CreER^+/−^/ihCD59^+/−^ mouse kidneys across baseline and days 1, 3, and 7 after RM ablation to define cellular and molecular responses (Fig. 1a). Analysis of scRNA-seq data identified distinct clusters corresponding to major kidney epithelial populations, including PTC1 and PTC2, as well as immune cells such as macrophages and neutrophils (Fig. 1d, Suppl. Fige 1,). These primary cell populations were consistently observed across four groups, facilitating a comparative analysis of cellular distributions and gene expression profiles. The scRNA-seq results showed significant ablation (3.6% to 0.9%) of RMs (Cx3cr1 + Adgre1 + Itgam + cells) at day 1 post-ILY treatment (Fig. 1d, Suppl. Figure 1b-1c). By day 3, the scRNA-seq data confirmed the regeneration of these cells (0.9% to 3.6%) (Fig. 1d, Suppl. Figure 1c). This result is comparable with our finding observed by the flow cytometry (Fig. 1b), further demonstrating the feasibility of utilization of scRNA-seq for monitoring the RMs changes after emptying niche of RMs.
Previously, we demonstrated that Cx3cl1 levels in serum and kidney significantly increased at day 1 to day 7 and gradually declined close to the baseline at day 7 post-rapid RM depletion[5]. This increase most likely results from RMs ablated kidney[5]. CX3CL1 is mainly produced by glomerular endothelial cells and the tubular epithelium and can be detected in many other cells, such as podocytes, mesangial cells, and stromal cells[5, 37]. However, the exact cell population in kidney responsible for producing Cx3cl1 remains unclear. Our current RT-qPCR analysis revealed marked upregulation of Cx3cl1 mRNA at days 1 and 3 following ablation (Fig. 2a). To interrogate this process at single-cell resolution, we performed global scRNA-seq mapping, which not only corroborated these findings but also uncovered a sustained and progressive induction of Cx3cl1 across days 1, 3, and 7, indicating a prolonged chemotactic program activated in response to macrophage loss (Fig. 2a). The Cx3cl1-positive cell population is primarily composed of multiple renal epithelial cell types, including PTC1, PTC2, DTC, and LOH in our analysis (Fig. 2b). Together, these data confirm the previous finding at transcription level that Cx3cl1 expression is increased after emptying RM niche and further demonstrated that renal epithelial cells, such as the PTC-1 and PTC-2, mainly produced Cx3cl1, a critical niche signaling in RM maintenance and regeneration.
To further evaluate the rigor of our scRNA-seq dataset, we performed global and single-cell transcriptomic analyses at baseline (D0) and from day 1 to day 7 post-ablation (D1-D7). We focused on the Macro, PTC1, PTC2, and Cx3cr1^+^ cell compartments and selected the top 30 upregulated (red) and top 30 downregulated (blue) differentially expressed genes (DEGs). Genes were ranked by the mean log_2_ fold-change (log_2_FC) across the three comparisons (D1 vs D0, D3 vs D0, and D7 vs D0) (Fig. 2c; Supplementary Fig. 2). Macrophages robustly upregulated inflammatory/stress-associated and proliferative programs (e.g., S100a8, Lcn2, Cxcl10, Saa3, Rrm2, Cdc6, Pclaf) while downregulating regulatory and angiogenic-associated signals (e.g., Il10, Nr4a3, Cryab, Kdr), consistent with the emergence of an injury-associated, repopulating macrophage state (Fig. 2c, left panel). Collectively, these patterns suggest that replenishing macrophages in the post-ablation kidney niche adopt a more tissue-repairing, metabolically specialized, and microenvironment-modulating phenotype rather than a purely cytotoxic immune program. Hallmark pathway analysis of macrophage consistently upregulated genes further supported this interpretation (Fig. 2c, right panel). Across D1, D3, and D7, macrophage signatures showed reproducible enrichment of metabolic and proliferative pathways, including Glycolysis, Oxidative Phosphorylation, E2F Targets, MYC Targets (V1/V2), G2M Checkpoint, and DNA Repair, together with stress-adaptation programs such as Unfolded Protein Response, Reactive Oxygen Species, and Hypoxia (Fig. 2c, right panel). In contrast, several canonical inflammatory/immune activation modules were relatively reduced including TNF-α-signaling via NF-κB, IL6-JAK-STAT3 signaling and Inflammatory Response, aligning with a shift toward reparative remodeling rather than overt inflammatory activation. The alignment of DEG-level changes with pathway enrichment across times highlights the internal consistency and biological relevance of macrophage responses captured by our scRNA-seq experiments, while the agreement between bulk and single-cell signatures across macrophages, tubular cells, and Cx3cr1^+^ populations underscore the rigor and accuracy of our approach.
Major cell-cell communication dynamics following RM ablation
To characterize intercellular signaling changes after RM ablation, we applied CellChat across eight major kidney cell types[17, 38]. Global communication peaked at D1, decreased at D3, and partially recovered by D7, whereas D0 showed minimal signaling (Fig. 3a). Circle plots of the top 10% differential interactions revealed strengthened crosstalk among RMs, epithelial cells (PTC2, LOH, DTC), and immune cells (neutrophils, T cells) after ablation (Fig. 3b). RMs acted as dominant signal senders-particularly toward PTC2 and maintained autocrine signaling; incoming/outgoing analyses confirmed macrophages as major exporters throughout recovery (Fig. 3c). The most dynamic pathways were Spp1, Fn1, and Ccl, with App emerging at D7. Fn1 signaling (notably Fn1-Cd44 and Fn1-Sdc4, plus integrin interactions) mediated RM-RM, RM-immune, and RM-epithelial communication (Fig. 3d-f), while Spp1-Cd44 showed a transient dip at D3. Late-stage signaling featured App-Cd74 and App-Sorl1, and early chemokine recruitment was driven by Ccl6-Ccr2 and Ccl9-Ccr1. Overall, residual RMs rapidly reestablished themselves as communication hubs, coordinating epithelial and immune responses through temporally regulated Fn1, Spp1, and Ccl mediated networks.
Spp1 Signaling Promotes RM Regeneration Following Clodronate Induced Ablation
To define mechanisms underlying macrophage-mediated communication during renal repair, we focused on the Spp1 (osteopontin, OPN) signaling axis and tested its role in RM regeneration using Opn^−/−^ (Spp1^−/−^) mice. RMs were depleted by a single injection of clodronate liposomes (CL). In C57BL/6 (B6) mice, flow cytometry at 2 days post-injection (2 DPI) showed an ~ 65% reduction in CD45^+^CD11b^+^F4/80hi macrophages compared with PBS controls, confirming efficient RM ablation (Fig. 4c). To assess Spp1-Cd44 signaling in RM repopulation, B6-WT and Opn^−/−^ mice underwent the same CL depletion, and recovery was quantified at 7 DPI and 9 DPI. WT mice rapidly regenerated RMs, returning to near-baseline levels by 7 DPI and remaining stable thereafter. In contrast, Opn^−/−^ mice showed significantly impaired recovery, with macrophage frequencies remaining low at 7 DPI and only partially restored by 9 DPI; RM density and distribution never reached WT levels (Fig. 4d). Opn^−/−^ mice also exhibited a modest baseline reduction in RM abundance under steady-state conditions (Fig. 4d). Together, these data identify Spp1/OPN signaling as a critical driver of RM regeneration and renal immune niche reconstitution following depletion.
Temporal Dynamics of Macrophage transcription factor activity following RM ablation
To investigate the transcriptional programs underlying RM dynamics after ablation, we performed Transcription Factor Regulatory Network Analysis-“SCENIC” on RM scRNA-seq data across five time points (Normal, D0, D1, D3, D7) (Fig. 5a). SCENIC identified 18 TF regulons with increased activity after ablation; however, focusing on TFs most relevant to myeloid cell differentiation, we highlighted a core subset that was consistently upregulated at D1, D3, and D7: Tfe3, Mitf, Hif1a, Myc, Gabpa, and Rcor1 (Fig. 5a). This pattern suggests that post-ablation RM restoration is driven by a differentiation-linked regulatory module rather than baseline homeostatic programs.
To connect these TFs with macrophage-centered communication programs, we used STRING to integrate the upregulated TFs with the ligands identified by CellChat (Spp1, Fn1, App, Ccl6, Ccl9) (Fig. 5b). STRING revealed significant network connectivity (PPI enrichment p = 7.66 × 10^−12^) and highlighted biological processes enriched in this integrated network, including interleukin-1–associated responses and myeloid differentiation-related programs (Fig. 5c). Especially, Myc, Hif1a, Mitf and Tfe3 these three TFs are enriched in regulation of myeloid cell differentiation. Together, these data support a model in which a Tfe3/Mitf/Hif1a/Myc/Gabpa/Rcor1-centered transcriptional network coordinates macrophage regeneration and niche reconstitution by coupling differentiation programs with key signaling pathways active during renal repair.
Identification of sub-RM and Sub-PTC cell population features after RM ablation
We further dissected RM and PTC dynamics after RM ablation. Because RMs were ablated at Day 1, most macrophages detected afterward likely represent newly recruited populations and thus greater heterogeneity than at Day 0. Major cell-level analysis also revealed two distinct PTC populations, a feature rarely reported previously[9, 36]. Targeted sub-clustering across four time points identified five transcriptionally distinct RM subsets (Fig. 6a-c) and ten PTC subtypes (Fig. 6d-f). Among RMs, Fn1^+^Cd11b^+^ cells expressed Fn1, Itgam, Ccl6, Spp1, and Arg1, consistent with profibrotic tissue-remodeling “builder” macrophages [39-41]. Pglyrp1^+^Ace^+^ cells expressed Ace, Pglyrp1, and Nr4a1, suggesting antimicrobial and regulatory functions [42-44]. Ccr2^+^Ly6c2^+^ cells were enriched for recruited monocyte-derived markers (Ccr2, Ly6c2, Ccr9) and pro-inflammatory activity [45, 46]. Mmp13^+^Ccl12^+^ cells expressed Mmp13/Mmp12, Ccl12, and Mrc1, indicating ECM remodeling and chemotaxis, while Ly6c1^+^Adgrf5^+^ cells (Ly6c1, Adgrf5, Egfl7) resembled niche-surveillance “sensor/guardian” macrophages. Temporally, Fn1^+^Cd11b^+^ and Mmp13^+^Ccl12^+^ peaked at Day 1 and declined by Days 3–7, whereas Pglyrp1^+^Ace^+^ and Ccr2^+^Ly6c2^+^ were reduced at Day 1 but expanded later, approaching baseline by Day 7 (Fig. 6b), consistent with monocyte-driven replenishment and repair-associated transitions [39]. For PTCs, combining unsupervised clustering with S1-S3 anatomical segmentation and SLC gene expression resolved ten subtypes (Fig. 6d). Three PTC1 subclusters (PTC1-S2_1/_2/_3) showed pronounced temporal shifts: all were depleted at Day 1 but rebounded by Days 3 and 7 (Fig. 6e), consistent with compensatory epithelial repair potentially promoted by macrophage-derived CSF-1[47]. PTC1-S2_1 was enriched for Phgdh, Car3, Mt1, S100g, and S100a13 (stress-adaptive), PTC1-S2_2 shared this program but additionally expressed Apoe, Spp1, and Mdk (injury-associated metabolic reprogramming), whereas PTC1-S2_3 retained high S100g/S100a13/S100a10 with low Apoe/Spp1/Mdk, suggesting roles in calcium signaling and cytoskeletal remodeling.
These results show that RM ablation induces dynamic remodeling in both macrophages and epithelial cells. Five RM subsets display distinct transcriptional programs-profibrotic, regulatory, inflammatory, remodeling, and surveillance with shifting proportions during injury and repair. Concurrently, PTC subclusters undergo transient loss and recovery, exhibiting stress adaptation, metabolic, and cytoskeletal changes. The emergence of two PTC populations (Suppl. Figure 1) suggests PTC1 represents injured epithelium with lower transporter expression and higher inflammatory signaling driven by immune-epithelial cross-talk.
Dynamics of sub-RM-PTC2 cell-cell interactions after RM ablation
At the major cell level, macrophages remained the primary signal senders after RM ablation, coordinating epithelial crosstalk, autoregulatory loops, and macrophage recruitment. PTC2 was a major recipient of RM-derived signals and showed self-regulation. To test whether subtype-level interactions differed, we focused on seven populations: Fn1^+^Cd11b^+^ RM, Pglyrp1^+^Ace^+^ RM, Ccr2^+^Ly6c2^+^ RM, and four PTC2 subtypes (PTC2_S1, PTC2_S1S3, Slc22a6_low_PTC2_S1, Slc25a25_high_PTC2_S1) (Fig. 7a-d). Circle plots of the top 10% differential interactions revealed strong temporal shifts. At Day 1, residual RMs showed increased communication compared with Day 0, with Fn1^+^Cd11b^+^ RM as the dominant sender; Pglyrp1^+^Ace^+^ and Ccr2^+^Ly6c2^+^ RMs acted as both senders and receivers with autoregulation. At Day 3, Ccr2^+^Ly6c2^+^ RM became the main signaling hub, interacting strongly with itself and Pglyrp1^+^Ace^+^ RM. By Day 7, Fn1^+^Cd11b^+^ RM again dominated, targeting both RM subsets and PTC2 subtypes (PTC2_S1S3 and Slc25a25_high_PTC2_S1), while Ccr2^+^Ly6c2^+^ RM retained autoregulatory signaling (Fig. 7a). Pathway-level analysis highlighted subtype-specific programs (Fn1, Spp1, Ccl) (Fig. 7b). In Ccr2^+^Ly6c2^+^ RM, Fn1 signaling peaked at Day 3, Spp1 was prominent only in Day 1 incoming signaling, and Ccl remained stable; Apoe emerged as a Day 3–7 incoming pathway. In Pglyrp1^+^Ace^+^ RM, Fn1 showed strong outgoing signaling at Days 1–3, Spp1 incoming signaling was high at Days 1 and 7, and Sirp increased from Day 3 to Day 7 an effect not captured at the major cell level. Ligand–receptor mapping confirmed Fn1-Cd44 as the dominant axis from Fn1^+^Cd11b^+^ RM, primarily targeting itself and Pglyrp1^+^Ace^+^ RM, while App–Cd74 dominated Pglyrp1^+^Ace^+^ RM with Ccr2^+^Ly6c2^+^ RM interactions at Day 1 (Fig. 7c-d). Fn1-integrin interactions peaked at Day 3 from Ccr2^+^Ly6c2^+^ RM, and Sirpb signaling was consistently enriched in Pglyrp1^+^Ace^+^ RM, indicating a subtype-restricted program. Overall, RM–PTC2 communication follows a temporal hierarchy: Fn1^+^Cd11b^+^ RM leads early and late, whereas Ccr2^+^Ly6c2^+^ RM dominates mid-phase signaling, together coordinating niche restoration through Fn1, Spp1, Ccl, and App driven networks.
Discussion
In this study, we used scRNA-seq analyses to monitor RM dynamics, renal global and single-cell transcriptomic changes in the response to acutely empty RMs niche in ILY-injected hCD59 labelled RMs mice [5, 6]. First, scRNA-seq faithfully recapitulated these dynamics, resolving all major renal cell lineages and confirming the specificity and reproducibility of RM loss and recovery in line with established kidney single-cell atlases [9, 12, 48]. Second, we confirm that Cx3cl1 is a pivotal epithelial-derived signal that triggers and sustains RM regeneration (Liu et al., 2020) and documents that Cx3cl1 is mainly produced in PTC1 and PTC2 after depletion. These results are consistent with the previous transcriptomic analysis by others showing that CX3CL1 is highly expressed by tubular epithelium in injured kidneys and drives monocyte/macrophage attraction and retention in both homeostasis and disease states, including acute injury and immune nephritis models[49, 50]. The sustained epithelial release of CX3CL1 may further support macrophage survival and differentiation, as CX3CR1 signaling has been implicated in macrophage longevity and function in other tissues[51]. Third, data self-validation of global and single-cell transcriptomic comparisons at three different time points showed highly reproducible gene-expression changes in macrophages, PTC1 and PTC2 and pathway changes in macrophages (Fig. 2c). Only Ly6i and Treml4, established markers of tissue-resident (yolk-sac–derived) RMs, were consistently downregulated across Macro, Cx3cr1^+^, and bulk datasets, confirming selective depletion of resident macrophages. Macrophages displayed induction of DAMP and proliferative genes, consistent with inflammatory monocyte repopulation, alongside sustained upregulation of Cxcl10, a key chemokine driving monocyte-derived macrophage recruitment in kidney injury[52, 53]. Shared upregulation of S100g and Cd5l (AIM) across macrophage populations reflected conserved injury-responsive programs, with CD5L implicated in macrophage remodeling and renal fibrosis across multiple kidney disease models[54-56]. Proximal tubules diverged into a metabolically rewired PTC1 state and an immune-stressed PTC2 state, the latter marked by Ly6e and Pfkfb3, linking epithelial stress to immune-metabolic pathways known from inflammatory macrophages[57]. Collectively, these data highlight tight macrophage–epithelial interdependence and establish ILY-ihCD59–mediated ablation with integrated transcriptomics as a robust framework to study RM niche regulation and epithelial homeostasis. Transcriptomic analysis showed that regenerating RMs do not immediately return to a homeostatic state but instead transiently acquire an injury-adaptive program marked by heightened metabolic activity, proliferation, and stress responses. Enrichment of glycolysis, oxidative phosphorylation, MYC- and E2F-driven cell-cycle programs, and DNA repair pathways indicate that RM regeneration is metabolically demanding and tightly linked to proliferative control. In contrast, canonical inflammatory pathways, including TNFα–NF-κB and IL-6–JAK–STAT3 signaling, were relatively subdued, consistent with a reparative rather than proinflammatory macrophage phenotype. This profile aligns with prior studies demonstrating that tissue-repair macrophages depend on metabolic reprogramming and proliferative capacity to restore organ integrity[1, 58, 59], and with kidney-specific evidence linking macrophage metabolic states to epithelial recovery and fibrosis limitation[60, 61]. Our findings extend this paradigm by defining a transcriptional program uniquely associated with macrophage niche reconstitution following acute depletion. Collectively, our findings highlight a spatiotemporally coordinated epithelial-immune axis that orchestrates macrophage niche regeneration, with implications for targeted modulation of chemokine signaling in renal repair and chronic kidney diseases.
Cell-cell communication analysis revealed that RM ablation triggers a dynamic, time-dependent rewiring of intercellular signaling networks. RM-epithelial and RM-immune interactions peaked at days 1 and 7 post-ablation, with a transient reduction at day 3, consistent with macrophages acting as signaling hubs that coordinate early inflammatory responses followed by reparative programs[1, 62-64]. Ligand–receptor analyses identified Spp1-Cd44 and Fn1-Cd44 as dominant pathways, implicating macrophage-driven ECM remodeling, fibrosis, and epithelial–immune communication[65] [66, 67] Spp1 signaling peaked at day 7, paralleling expansion of Spp1^+^ macrophages linked to chronic kidney injury and fibrotic repair[68, 69], while Fn1-Cd44/Sdc4 interactions supported immune retention and matrix remodeling[70, 71]. Functional analyses revealed that SPP1 is essential for macrophage regeneration following depletion, while being largely dispensable for steady-state homeostasis. Specifically, Spp1^−/−^ mice exhibited a pronounced delay in macrophage repopulation after clodronate-mediated ablation, indicating a critical role for SPP1 in injury-induced macrophage renewal. Consistent with these findings, SPP1 has been implicated as a key mediator of immune cell-vascular interactions, particularly under inflammatory and reparative conditions[72, 73]. Together, these findings support a model in which macrophage- and epithelial-derived SPP1 establishes CD44-dependent autocrine and paracrine loops that drive macrophage recruitment, survival, and niche restoration during kidney repair[74, 75]. Single-cell transcription factors regulatory network analysis identified 18 TFs selectively activated at days 1, 3, and 7 following RM ablation, implicating them in niche regeneration rather than homeostatic maintenance. Integration of these TFs with ligand-receptor pathways (Spp1, Fn1, App, Ccl6, Ccl9) revealed strong protein-protein interaction enrichment and highlighted IL-1–responsive and differentiation-related programs, consistent with IL-1 signaling as a key regulator of macrophage activation and myeloid differentiation during injury and repair[9]. Prominent TFs, including Hif1α, Myc, Cebpα, Klf family members, and BHLHE40/41, are well established drivers of macrophage metabolism, proliferation, and reparative polarization under inflammatory and hypoxic conditions[76-80]. TF clustering further revealed functional modules governing metabolic–proliferative reprogramming, stress and inflammatory responses, and chemokine-mediated leukocyte recruitment. Together, these data support a conserved differentiation program underlying RM regeneration, in which TF-driven induction of ligands such as Spp1 sustains autocrine and paracrine signaling essential for macrophage niche reconstitution.
After RM ablation, most macrophages detected from day 1 onward represent newly recruited cells, resulting in increased heterogeneity compared with the baseline RM pool. Single-cell analysis identified five RM subsets and ten proximal tubule cell (PTC) subtypes across time points, revealing greater epithelial diversity than previously appreciated. Among RMs, Fn1^+^Cd11b^+^ cells displayed a profibrotic remodeling phenotype[39], while Pglyrp1^+^Ace^+^ and Ccr2^+^Ly6c2^+^ subsets reflected antimicrobial and inflmmatory monocyte-derived states. Mmp13^+^Ccl12^+^ macrophages were associated with ECM remodeling, whereas Ly6c1^+^Adgrf5^+^ cells represented niche-surveillance populations. Temporally, remodeling subsets peaked early, followed by expansion of inflammatory monocyte-derived populations, consistent with stepwise monocyte recruitment and differentiation during repair[15, 81].
PTC sub-clustering revealed ten distinct epithelial states, with three PTC1 subtypes transiently reduced at day 1 and recovering by days 3–7, consistent with macrophage-dependent epithelial repair mediated by CSF-1 (Alikhan, 2011). These PTC1 subsets reflected stress responses, metabolic adaptation, and cytoskeletal remodeling. Together, these data demonstrate coordinated temporal remodeling of macrophage and epithelial compartments after RM ablation, extending prior studies of RM heterogeneity and highlighting macrophage-epithelial cross-talk as a central driver of kidney regeneration[82-86]. Subtype-resolved communication analysis further demonstrated that RM-PTC2 crosstalk follows a temporal hierarchy. Fn1^+^Cd11b^+^ macrophages dominated early and late signaling through Fn1-Cd44 and integrin pathways, consistent with fibronectin’s established role in macrophage retention and tissue remodeling[39]. In contrast, Ccr2^+^Ly6c2^+^ macrophages became the principal signaling hub at the mid-phase, coinciding with peak monocyte recruitment and differentiation. Pglyrp1^+^Ace^+^ macrophages engaged distinct pathways, including App–Cd74 and Sirpb signaling, suggesting immunomodulatory roles not captured at the major cell level, in line with emerging evidence for APP-related signaling in macrophage immune regulation[87]. These data align with the view of macrophages as regulators of repair via structured networks[88, 89]. The shifting roles of RM subsets mirror established macrophage plasticity during renal injury[20, 47]. Furthermore, the prominent FN1 and SPP1 axis signaling accords with emerging models of macrophage-epithelial communication[20]. Our findings reveal a specific communication hierarchy where RM subsets sequentially regulate PTC2 signaling to coordinate epithelial repair. This highlights RM regeneration as a dynamically orchestrated inter-lineage process.
Conclusion
Acute depletion of renal macrophages (RMs) initiates a highly coordinated regenerative program characterized by rapid reconstitution of the macrophage niche through the sequential recruitment and differentiation of discrete macrophage subsets, including Ccr2^+^Ly6c2^+^, Pglyrp1^+^Ace^+^, and Fn1^+^Cd11b^+^ populations. Integrated single-cell transcriptomic and cell-cell communication analyses identify post-ablation macrophages as dominant signaling hubs that engage in dynamic bidirectional crosstalk with immune and proximal tubule epithelial (PTE) cells via key ligand-receptor pathways such as Spp1-Cd44 and Fn1-integrin. Collectively, these findings establish renal macrophage repopulation as a tightly regulated, transcription factor-dependent and communication-driven process that synchronizes immune-epithelial networks to restore kidney niche integrity, thereby providing mechanistic insight into regenerative repair while highlighting therapeutic opportunities to enhance regeneration and limit fibrosis. Despite the mechanistic clarity afforded by murine depletion-repopulation models, the transcriptional regulation of macrophage-tubular epithelial interactions in human kidney disease remains poorly defined. In particular, how injury-induced transcriptional programs govern macrophage recruitment, fate specification, and niche re-establishment through evolving intercellular signaling networks, and how macrophage heterogeneity integrates with epithelial stress responses to promote adaptive repair versus maladaptive fibrosis, remain unresolved. Addressing these gaps will require future studies employing chronic injury models, metabolic perturbations, and integrative multi-omics approaches in human biopsies and organoid systems to evaluate conservation, biomarker relevance, and therapeutic tractability of macrophage-epithelial communication pathways in kidney disease.
Supplementary Material
This is a list of supplementary files associated with this preprint. Click to download.
• SUPPLEMENTARYMATERIALS.docx
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Wynn TA, Vannella KM. Macrophages in Tissue Repair, Regeneration, and Fibrosis. Immunity. 2016;44:450–62.26982353 10.1016/j.immuni.2016.02.015PMC 4794754 · doi ↗ · pubmed ↗
- 2Kratofil RM, Kubes P, Deniset JF. Monocyte Conversion During Inflammation and Injury. Arterioscler Thromb Vasc Biol. 2017;37:35–42.27765768 10.1161/ATVBAHA.116.308198 · doi ↗ · pubmed ↗
- 3Li L, Huang L, Sung SS, Vergis AL, Rosin DL, Rose CE Jr., Lobo PI, Okusa MD. The chemokine receptors CCR 2 and CX 3CR 1 mediate monocyte/macrophage trafficking in kidney ischemia-reperfusion injury. Kidney Int. 2008;74:1526–37.18843253 10.1038/ki.2008.500PMC 2652647 · doi ↗ · pubmed ↗
- 4Nelson PJ, Rees AJ, Griffin MD, Hughes J, Kurts C, Duffield J. The renal mononuclear phagocytic system. J Am Soc Nephrol. 2012;23:194–203.22135312 10.1681/ASN.2011070680 PMC 3269181 · doi ↗ · pubmed ↗
- 5Liu F, Dai S, Feng D, Qin Z, Peng X, Sakamuri S, Ren M, Huang L, Cheng M, Mohammad KE, Distinct fate, dynamics and niches of renal macrophages of bone marrow or embryonic origins. Nat Commun. 2020;11:2280.32385245 10.1038/s 41467-020-16158-z PMC 7210253 · doi ↗ · pubmed ↗
- 6Islamuddin M, Evangelista J, Kabir ME, Vidal AKN, Liu S, Qin X. Rapid Depletion of Renal Macrophages Using Human CD 59/Intermedilysin Cell Ablation Tool. J Vis Exp 2025.
- 7Lee S, Huen S, Nishio H, Nishio S, Lee HK, Choi BS, Ruhrberg C, Cantley LG. Distinct macrophage phenotypes contribute to kidney injury and repair. J Am Soc Nephrol. 2011;22:317–26.21289217 10.1681/ASN.2009060615 PMC 3029904 · doi ↗ · pubmed ↗
- 8Wang W, Wang X, Chun J, Vilaysane A, Clark S, French G, Bracey NA, Trpkov K, Bonni S, Duff HJ, Inflammasome-independent NLRP 3 augments TGF-β signaling in kidney epithelium. J Immunol. 2013;190:1239–49.23264657 10.4049/jimmunol.1201959 · doi ↗ · pubmed ↗