Male Wistar and Sprague-Dawley (SD) rats aged 6–8 weeks (body weight 200–250 g) were obtained from the Animal Experiment Center of the Air Force Medical University. An acute renal allograft rejection model was established by orthotopic transplantation of kidneys from male Wistar rats into SD recipients (24), Wistar→SD pairs were selected for their defined MHC disparity. Both donors and recipients were maintained under anesthesia with Zoletil 50 (0.1–0.12 ml/100g). The recipient’s native left kidney was excised. The left kidney from the Wistar rat was transplanted into the left abdominal cavity of the SD rat, The renal artery and vein of the graft were anastomosed end-to-end with the left renal artery and vein of the SD recipient, respectively. The ureter of the graft was anastomosed to the bladder of the SD rat. All rat experiments were conducted in specific pathogen-free (SPF) facilities in accordance with the guidelines of the Animal Care and Use Committee of Air Force Medical University. Postoperatively, animals were euthanized under Zoletil 50 anesthesia at designated time points (0, 1, 3, and 7 days) for sample collection. The experimental protocol was approved by the Animal Ethics Committee of Air Force Medical University(KY20223099-1).

At each time point, transplanted kidneys were harvested from 3 rats per group. Following euthanasia, kidneys were perfused via the abdominal aorta with ice-cold phosphate-buffered saline (PBS) for blood clearance. The three kidneys per group were pooled, minced into 1–2 mm³ fragments, and digested in collagenase (17018029/17104019, eBioscience™) at 37 °C for 45 min. The digestate was filtered through a 40-μm cell strainer. Erythrocytes were lysed using 1 ml ice-cold RBC Lysis Buffer (C3702, Beyotime) for 5 min, followed by PBS washing to obtain a single-cell suspension. The prepared single-cell suspension was counted and examined by microscopy. Cells were incubated with anti-CD45-APC antibody (17-0461-82, eBioscience™) at 4 °C for 30 min. During the last 5 min of incubation, DAPI (1:1000 dilution, 62248, eBioscience™) was added and mixed. After incubation, cells were washed once with RPMI-1640 medium (12633020, eBioscience™) via centrifugation (300g, 5 min), resuspended, and filtered through a 40-μm cell strainer. The final suspension was transferred to flow cytometry tubes and kept protected from light on ice. Pre-prepared collection tubes containing RPMI-1640 with 10% FBS were loaded into the sorting chamber for CD45⁺ cell sorting. After sorting, the fluidics system was flushed with sterile ultrapure water. A subset of sorted cells was re-analyzed for quality control, and flow cytometry data were recorded. The target cell suspension was transferred from flow tubes to 15-mL centrifuge tubes, pelleted by centrifugation, counted, examined microscopically, and processed for single-cell library preparation.

The cell suspension was loaded into Chromium microfluidic chips with 3’ (v2 or v3, depends on project) chemistry and barcoded with a 10× Chromium Controller (10X Genomics). RNA from the barcoded cells was subsequently reverse-transcribed and sequencing libraries constructed with reagents from a Chromium Single Cell 3’ v2(v2 or v3, depends on project) reagent kit (10X Genomics) according to the manufacturer’s instructions. Sequencing was performed with Illumina (Hi-Seq 2000 or Nova-Seq, depends on project) according to the manufacturer’s instructions (Illumina).

Cell Ranger (v7.1.0, 10x Genomics) was used to align raw FASTQ files to the rat reference genome mRn7 (GCA_015227675.2). Seurat (v5.0.3) was employed for quality control, data preprocessing, and dimensionality reduction. Specimens at each time point (0, 1, 3, 7 days) were derived from transplanted kidneys of 3 rats per group. After obtaining the gene-cell matrices from transplanted and control groups, all matrices were merged. To filter low-quality cells, we applied the following criteria: Cells with >500 genes detected and <5,000 genes; Cells with mitochondrial gene percentage <5%; Genes detected in ≥5 cells with ≥1 feature count.

The high-quality single cells were re-processed through the full Seurat workflow to generate the final dataset for all downstream analyses. Major cell types were identified using standard Seurat (v5.0.3) clustering at a resolution of 0.6. Cellular clusters were visualized via uniform manifold approximation and projection (UMAP). Cluster-specific marker genes were selected from differentially expressed genes (DEGs) using Seurat’s FindAllMarkers function. Six major cell types in the final dataset were manually annotated based on canonical marker gene sets.Visualizations were generated using Seurat plotting functions: VlnPlot; FeaturePlot; DotPlot; DoHeatmap.

We employed the MiloR package (v1.12.0) to detect differential abundance of cell populations across experimental conditions by modeling cell counts within k-nearest neighbor (KNN) graph neighborhoods. To visualize results from differential analysis on neighborhoods, we construct an abstracted graph, where nodes represent neighborhoods and edges represent the number of cells in common among neighborhoods. The size of nodes represents the number of cells in the neighborhood. The position of nodes is determined by the position of the sampled index cell in the single-cell uniform manifold approximation and projection (UMAP), to allow qualitative comparison with the single-cell embedding. Graphical visualization is implemented using the R packages ggplot and ggraph.

Gene Ontology (GO) enrichment analysis was performed using the enrichGO function from the R package clusterProfiler (v4.10.0). KEGG pathway analysis was conducted with the Kyoto Encyclopedia of Genes and Genomes database (KEGG release 110.1). Results were visualized using R packages ggplot2 (v3.5.0) and enrichplot (v1.26.0). Volcano plots generated via ggplot2 displayed significantly upregulated/downregulated genes (|log₂FC| > 1, adj. p < 0.05).

Subclustering analysis of target cells was performed using Seurat (v5.0.3) at a resolution of 0.6. Cellular subclusters were visualized via Uniform Manifold Approximation and Projection (UMAP). Marker genes for each subcluster were identified from differentially expressed genes (DEGs) using Seurat’s FindAllMarkers function (min.pct = 0.25, logfc.threshold = 0.5). Subclusters were manually annotated based on canonical marker genes and known biological functions as described above.

Enrichment scores were calculated using Seurat’s AddModuleScore function. This function computes the mean-centered expression of a gene set relative to control gene sets, which represents the average relative expression. Gene sets were curated based on previously published signatures for macrophage lineages and phenotypic functions (Supplementary Table 2).

Pseudotime trajectories for each cell type were constructed from Seurat objects using the R package Monocle 2 (v2.26.0). Pseudotemporally variable genes were identified with Monocle 2’s differential-Gene-Test function (q-value < 0.01). Branch-dependent expression patterns were assessed using Branch Expression Analysis Modeling (BEAM) and visualized via plot-genes-branch-heatmap.

Interaction weights were calculated as the product of ligand fold-change in sender cells and receptor fold-change in receiver cells. Cell-cell communication networks were predicted from single-cell RNA sequencing data using the CellChat library (v1.6.0). Ligand-receptor interactions were inferred with CellChat, considering only ligands and receptors expressed in >5% of cells within corresponding cell types.

Tissues were fixed in 4% paraformaldehyde (PFA), embedded in paraffin wax, sectioned at 5 μm thickness. Paraffin sections were dried overnight on a heating plate at 40 °C. Deparaffinization and rehydration were performed through sequential immersion in xylene and ethanol gradients. Sections were stained with Mayer’s hematoxylin for 3 min, followed by three washes in distilled water. Counterstaining was performed in eosin Y solution for 2 min. Dehydration was achieved through graded ethanol series (70%, 95%, 100%), cleared in xylene for 5 min, and mounted with synthetic resin under cover slips.

Renal tissues were dewaxed with xylene. Antigen retrieval was then performed in 0.01 M citrate buffer for 0.5 hours. After incubation in blocking solution containing 0.1% Triton X-100 for 1.5 hours at room temperature, the tissues were incubated overnight with CD3 antibody (1:200 dilution; ab11089, Abcam) and KIM-1 antibody (1:200 dilution; PA5-79345, eBioscience™). After washing away excess antibodies, sections were incubated with secondary antibodies, followed by DAB staining (PR30018, Proteintech), and images were captured.

Immunofluorescence analysis was performed on renal sections. After antigen retrieval, slides were incubated with primary antibodies:Anti-CD3 (1:500 dilution; GB151137, Servicebio); Anti-CD68 (1:200 dilution; GB113109, Servicebio); Anti-ISG15 (1:200 dilution; 15981-1-AP, Proteintech). Following sequential washing, sections were labeled with species-matched secondary antibodies conjugated to Alexa Fluor^® dyes (Invitrogen). Nuclei were counterstained with DAPI (1μg/mL, D9542, Sigma-Aldrich). Autofluorescence was quenched with 0.1% Sudan Black B for 30 min. Sections were mounted with anti-fade medium (ProLong Gold, P36930, Thermo Fisher) and imaged using a confocal microscope (Nikon Eclipse C1, Japan) with whole-slide scanning (Pannoramic MIDI, 3DHISTECH).

A Ccr5 inhibitor maraviroc (H-13004, MedChemExpress) was dissolved in 10% DMSO(1000μL)+ 40%PEG300(4000μL)+ 5% Tween-80 (500μL)+ 45% saline(4500μL) and mixed until clear. Kidney transplant rats were intraperitoneally injected daily at a dose of 50 mg/kg starting from 1 day before transplantation and continuing until day 7 after transplantation. In the control group, only 10% DMSO was injected. In addition, venous blood samples were collected from transplanted rats for comprehensive biochemical analysis of serum creatinine and blood urea nitrogen (BUN).

Data are expressed as the mean ± standard deviation (SD) and were analyzed by using GraphPad Prism 10.2 software. Two-tailed unpaired Student’s t-test was used for comparisons between two groups. One-way ANOVA was used for comparisons among multiple groups. The P value for graft survival was determined by the Mann–Whitney test. The significance level was set at P < 0.05.

To delineate the immune reconstitution of renal allografts and uncover early mechanisms of acute rejection, we established an allogeneic transplantation model using Wistar rats as donors and Sprague-Dawley (SD) rats as recipients (Supplementary Figures S1 A, B), Allografts were harvested at postoperative days (POD) 0, 1, 3, and 7. CD45⁺ immune cells were isolated by fluorescence-activated cell sorting (FACS) and processed for single-cell RNA sequencing (10x Genomics) (Figure 1A). Concomitant histopathological analysis via H&E staining revealed progressive features of acute rejection over time, including escalating interstitial infiltration of mononuclear inflammatory cells, prominent tubulitis and intimal arteritis, along with increasing intracapillary inflammatory cell accumulation within glomerular tufts (Figure 1B), demonstrating the development of acute allograft rejection(Supplementary Table 1); Partial renal tubules exhibited hydropic degeneration of epithelial cells, disappearance of epithelial nuclei, and epithelial cell sloughing, culminating in necrosis of both glomeruli and tubules. Immunohistochemical staining revealed marked T-cell infiltration (CD3) and robust expression of KIM-1 in the renal tubules (Supplementary Figure S1C), while biochemical analyses of serum creatinine and blood urea nitrogen (BUN) demonstrated significantly impaired renal function (Supplementary Figure S1D). Flow cytometric analysis corroborated these findings (Figure 1C), demonstrating a significant time-dependent increase in the proportion of infiltrating immune cells within renal allografts: 11.7% (Day 1), 20.9% (Day 3), 50.7% (Day 5), and 78.5% (Day 7).

We obtained a total of 37,159 cells from all samples for single-cell RNA sequencing (scRNA-seq). Following rigorous quality control and robust batch effect removal/integration (Supplementary Figures S2A, B), cell populations were visualized in a Uniform Manifold Approximation and Projection (UMAP) plot using the Seurat R package. Unsupervised clustering revealed 19 distinct cell clusters (Supplementary Figure S2C). By annotating cell clusters through marker gene expression profiling and integration with published cell atlas datasets, we identified six major cell lineages across the four samples (Figures 2A, B), including B cells (Ms4a1, Cd79b), Neutrophils (S100a8, S100a9), Macrophages (Cd68, Cd86, C1qa), Dendritic cells(DCs) (Clec9a, Xcr1), Natural killer cells(NK cells) (Klrb1b, Nkg7) and T cells (Cd3d, Cd3e)(Figures 2C–E). Stream plot visualization revealed that macrophages constituted the predominant proportion among all immune cells. Innate immune cells, including NK cells and DCs, maintained relatively consistent proportions during the early phase (days 0, 1, and 3) but declined to minimal levels by day 7. In contrast, T cells and B cells—core components of adaptive immunity—progressively expanded following transplantation (Figure 2F). Differential abundance analysis using the MiloR package further revealed substantial increases in neutrophil, T cell, and B cell populations at day 7 post-transplantation during acute rejection (Figures 2G, H).

Collectively, our data clearly delineate the early-stage immune cell landscape in renal allograft acute rejection, revealing that macrophages may act as early “responders” in transplant immunity and play an important role in acute rejection.

Prior studies have established that T cells play a central role in mediating acute rejection, and their reactivity dictates both short- and long-term outcomes following solid organ transplantation (25). Therefore, we performed further subtyping of T cells and identified eight distinct T cell clusters (Figures 3A, B). By screening differentially expressed genes (DEGs) in each cluster and comparing them with previously reported T cell subsets and canonical marker genes in the literature, these eight clusters were defined as: CD4+ T cells (Cd4, Cd40lg); CD8+ T cells (Cd8a, Gzma, Gzmb); Natural killer T cells (Klrd1, Klrb1c); Treg cells (Foxp3, Ctla4, Lag3); Tissue-resident memory T cells(Trm);(Cd69, Itgae [CD103]); and Naive T cells (Sell [CD62L]). Cluster 3 was defined as “Homeostatic Regulatory T Cells” (H-Treg) due to its co-expression of genes associated with: negative regulation of signaling pathways-Ptpn22 (a lymphoid tyrosine phosphatase that suppresses TCR signaling) and Spred2 (an inhibitor of the Ras/MAPK pathway); And metabolic adaptation-Foxo1 (regulating Treg metabolism and survival while inhibiting the PI3K-Akt pathway) and Inpp4b (a phosphoinositide phosphatase that dampens PI3K signaling); Cluster 6 was designated as “Tissue-Remodeling Regulatory T Cells” (TR-Treg) based on its co-expression of genes implicated in: inflammation modulation -Il1b (a pro-inflammatory cytokine), Tgfbi (mediating TGF-β signaling), and Cd274 (PD-L1); And tissue remodeling-Vcan (versican) and Fn1 (fibronectin) (Figures 3C, S3A).

Following the definition of T cell subsets, we constructed a river plot depicting the proportional abundance of each subset across time points (Figures 3D, S3B). This analysis revealed that CD4+ T cells characterized by an initial increase followed by a decline. In contrast, CD8+ T cells showed a marked expansion at Day 7, becoming the predominant T cell subtype. Furthermore, comparative analysis of differentially expressed genes in T cells at post-transplant day 7 versus day 0 via volcano plot revealed significant upregulation of cytotoxicity-associated genes, including granzyme A (Gzma), granzyme B (Gzmb), and Tnfrsf9 (a co-stimulatory receptor; CD137) (Figure 3E). These findings collectively indicate that with the occurrence of acute rejection, effector T cells begin to activate and release cytokines as well as cytotoxic mediators, further leading to graft damage.

Given the predominance of macrophages within the immune cell compartment, we performed macrophage subclustering to comprehensively investigate their heterogeneity and delineate the immune landscape during acute rejection, ultimately partitioning them into seven distinct subsets (Figure 4A). Each subset exhibited distinct transcriptomic signatures. Through differential gene enrichment analysis of macrophage subsets (Figure 4C), we generated bubble plots displaying subset-specific marker genes and visualized their expression patterns projected onto UMAP embeddings, successfully annotating these seven macrophage populations (Figures 4B, S4A). Additionally, functional profiling of each subset was performed using gene signature scoring (Supplementary Table 1 (10)).

We defined cluster 1 as Isg15+ Mac because it expresses interferon-related genes (Isg15) and pro-inflammatory genes (Nfkbiz, Tnfrsf1b, Xdh), while simultaneously exhibiting the highest functional scores for phagocytosis and pro-inflammatory activity (26)(Figure 4F); Cluster 2 was defined as C1qc+ Mac (27), characterized by its expression of complement component genes (C1qa, C1qc); Cluster 3 was defined as Arg1+ Mac due to its expression of classical M2-type genes (Arg1) and tissue-repair genes (Vcan, Fn1, Vegfa) (28), while concurrently exhibiting the highest functional score for tissue repair promotion (Figure 4F); Cluster 4 was defined as Cxcl11+ Mac based on expression of the chemokines Cxcl11 and Cxcl9 (29); Cluster 5 was designated Spp1+ Mac due to its specific expression of Spp1 (30), Scoring results indicate that this subset may promote tissue fibrosis;Cluster 6 characterized by high proliferative activity and elevated expression of proliferation-associated genes (Stmn1, Mki67, Spc24), was designated Stmn1+ Mac (26). Cluster 7 defined by Lyve1 expression, was termed Lyve1+ Mac (31) (Figure 4D).

Finally, to provide an intuitive visualization of the dynamic changes in macrophage subsets following acute rejection, we generated time-stratified UMAP projections (Figure 4E) and stacked bar plots quantifying the proportional abundance of each subset across four post-transplantation time points (Figure 4G). We observed that C1qc+ Mac constituted a major proportion of macrophages in non-transplanted kidneys. Their abundance progressively decreased following acute rejection, suggesting a homeostatic role. Conversely, four subsets—Isg15+ Mac, Arg1+ Mac, Cxcl11+ Mac, and Spp1+ Mac—exhibited differential expansion during rejection. Notably, Isg15+ Mac demonstrated the most pronounced increase, with proportions surging within the first post-transplant day and sustaining elevated levels thereafter (Figure 4G). This indicates that the shift toward a pro-inflammatory macrophage phenotype occurs most rapidly following transplantation, manifesting within 24 hours.

To validate the substantial presence of Isg15+ Mac post-transplantation, multiplex immunofluorescence staining for Isg15 and CD68 was performed. This analysis revealed significantly increased Isg15 expression in renal allografts, with co-localization observed between Isg15 and CD68, suggesting its expression on macrophage surfaces (Figure 5A). Isg15+ Mac also demonstrated high expression of genes critically involved in pro-inflammatory responses(Il17ra (32)、Tnfrsf1 (33)、Map4k1 (34)、Itga4 (35)、Irak3 (36)和Xdh (37))(Supplementary Figure S4B). We further performed GO enrichment analysis on macrophages(Supplementary Figure S5A-B), revealing that Isg15+Mac showed significant enrichment of genes involved in immune response activation—including activation of the immune response, lymphocyte proliferation, and T cell activation (Figures 5B, C). In summary, the Isg15+ Mac subset expands in renal allografts during acute rejection and demonstrates high expression of pro-inflammatory genes, indicating its critical involvement in acute rejection pathogenesis.

Next, trajectory analysis of macrophages was performed using Monocle 2, integrating functional annotations and differentially expressed genes across all subsets. This analysis designated State 1 as the initial state, while identifying C1qc+ Mac and Lyve1+ Mac as two macrophage subsets in a resting state (Figures 5D–F). This finding aligns with our observation that C1qc+ Mac predominates among macrophage subsets in Day 0 kidneys, reinforcing their role as a baseline population. Pseudotime analysis across temporal samples confirmed State 1 as the initial state, with cells progressively differentiating toward distinct trajectory endpoints (Figure 5G). Further stratification of the pseudotemporal trajectory by macrophage subsets revealed preferential localization of Isg15+ Mac within the State 2 branch (Figure 5H).

Finally, differential gene expression analysis along the macrophage differentiation trajectory, coupled with GO functional enrichment, delineated the functional shifts during this process. We revealed that early in acute rejection, macrophages undergo a binary fate divergence: a pro-inflammatory trajectory and a pro-repair trajectory (Figure 6A). Specifically, Arg1+ Mac—representing the pro-repair branch—dominates post-injury tissue restoration, while Isg15+ Mac drives inflammation and rejection through the pro-inflammatory trajectory.

Using CellChat, we analyzed intercellular communication between macrophages and other immune cells. The intercellular communication heatmap revealed that macrophages primarily interact with T cells and NK cells (Figures 6B, S6A), which aligns with established literature reporting that macrophages infiltrate early post-transplantation and establish a pro-inflammatory microenvironment to activate T cells. Meantime, we validated the Ccl3-Ccr5 ligand-receptor pair as a critical signaling hub in macrophage-T cell communication (Supplementary Figure S6B). Next, we specifically investigated intercellular communication events with macrophage subsets as signal senders and T cells as receivers. This analysis revealed that Isg15+ Mac exhibited more prominent interactions with T cells—both in quantity and intensity—compared to other macrophage subsets (Figures 6C, S7A–C). Furthermore, the Ccl3-Ccr5 signaling axis played a dominant role in Isg15+Mac-T cell crosstalk, with signal strength progressively strengthening over time (Figures 6D, S7D). Co-localization of Isg15+ Mac and T cells in post-transplant renal allografts further supports their potential interactions (Figure 6E). Consequently, Isg15+Mac cells may act as early orchestrators of the immune response, influencing T-cell activation and recruitment following kidney transplantation through Ccl3–Ccr5 ligand–receptor interactions, and, concomitantly, releasing chemokines to recruit circulating peripheral T cells that infiltrate the graft and amplify the rejection response. Furthermore, by analyzing human renal biopsy specimens with rejection and acute kidney injury (AKI) (38), we also identified the abundant presence of Isg15+Mac, as well as the communication between Isg15+Mac and CD8+ T cells via the CCL3-CCR5 ligand-receptor pair (Supplementary Figures S8A–D).

As Ccr5 serves as the receptor for chemokines Ccl3, Ccl4, and Ccl5 and is highly expressed on T cells, we further validated the role of the Ccl3-Ccr5 axis in early renal allograft rejection. Rat kidney transplant recipients were randomized into two groups: the treatment group received Maraviroc (an FDA-approved Ccr5 antagonist) administered 1 day preoperatively and during the first postoperative week (Figure 7A), while the control group received an equivalent volume of DMSO. Maraviroc treatment significantly improved allograft survival in the treatment cohort (Figure 7B). Immunofluorescence confirmed reduced T cell infiltration in allografts (Figures 7C, S9A), Histological and immunohistochemical analyses of renal allografts demonstrated that renal injury was alleviated in the experimental group(Figure 7D, E), and renal function parameters showed partial recovery compared with the control group (Supplementary Figure S9B).

Collectively, we propose that Isg15+Mac may serve as critical contributors to acute rejection by activating T cells via the Ccl3–Ccr5 axis, thereby amplifying the process. Our results further indicate that targeting the Ccl3-Ccr5 axis represents a promising therapeutic strategy for preventing acute allograft rejection.