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.

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-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.

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.

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.

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.

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 log₂ fold-change (log₂FC) 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.

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 inflammatory 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.

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.