Transcriptional and spatial profiling of the kidney allograft unravels a central role for FcyRIII+ innate immune cells in rejection

Abstract

Rejection remains the main cause of premature graft loss after kidney transplantation, despite the use of potent immunosuppression. This highlights the need to better understand the composition and the cell-to-cell interactions of the alloreactive inflammatory infiltrate. Here, we performed droplet-based single-cell RNA sequencing of 35,152 transcriptomes from 16 kidney transplant biopsies with varying phenotypes and severities of rejection and without rejection, and identified cell-type specific gene expression signatures for deconvolution of bulk tissue. A specific association was identified between recipient-derived FCGR3A+ monocytes, FCGR3A+ NK cells and the severity of intragraft inflammation. Activated FCGR3A+ monocytes overexpressed CD47 and LILR genes and increased paracrine signaling pathways promoting T cell infiltration. FCGR3A+ NK cells overexpressed FCRL3, suggesting that antibody-dependent cytotoxicity is a central mechanism of NK-cell mediated graft injury. Multiplexed immunofluorescence using 38 markers on 18 independent biopsy slides confirmed this role of FcγRIII+ NK and FcγRIII+ nonclassical monocytes in antibody-mediated rejection, with specificity to the glomerular area. These results highlight the central involvement of innate immune cells in the pathogenesis of allograft rejection and identify several potential therapeutic targets that might improve allograft longevity.

Fig. 1. Identification of the cell subtypes in kidney allograft biopsies by single-cell RNA seq.

a Experimental Approach. Kidney allograft biopsies were used for scRNAseq (n = 16 biopsies from 14 patients). b Uniform Manifold Approximation and Projection (UMAP) plot of 35,152 cells passing QC filtering. The main kidney cell types are represented, including loop of Henle (LOH), podocytes, vascular smooth muscle and pericytes (vSMp), proximal tubule (PT), intercalated cells (IC), three endothelial cell subsets comprising vasa recta (ECvr), glomerular (ECg) and peritubular capillaries (ECptc), myeloid cells, and lymphoid cells. c Schematic of a juxtamedullary nephron showing relevant cell types and associated vasculature. d Dot plot showing average gene expression values of canonical lineage markers (log scale) and percentage of major cell types represented in the whole dataset and UMAP plot. e Relative proportions of the 18 different cell types identified in the kidney transplant biopsies. f Polar plot of the 16 biopsies. The biopsies were reclassified into 4 groups: non-rejecting donor-specific antibody (NR DSA)-negative (NR DSA-), NR DSA-positive (NR DSA+), T cell-mediated rejection (TCMR), DSA-positive antibody-mediated rejection (ABMR DSA+). g Spearman’s correlation coefficient represented by a dot with 95% confident interval represented by the error bars of the correlation between inflammation severity (radius on the polar plot) and the frequency of the different immune cells as a proportion of the total population. Panels a and c were created using biorender.com.

Fig. 2. CIBERSORTx deconvolution confirms enrichment of FCGR3A+ cells in biopsies with rejection.

a Experimental Approach. scRNAseq-derived signature matrix KTB18 was used for deconvolution of the dataset GSE147089 encompassing 224 transcriptomes from kidney biopsies. Created using biorender.com. b Polar plot of the 224 biopsies. The biopsies were reclassified into 6 groups: non-rejecting donor-specific antibody (NR DSA)-negative (NR DSA-), NR DSA-positive (NR DSA+), T cell-mediated rejection (TCMR), antibody-mediated rejection DSA-positive (ABMR DSA+), ABMR, DSA-negative histology of ABMR (ABMRh DSA-), and mixed rejection. c Pearson’s correlation coefficient is represented by a dot, with 95% confident interval shown as error bars, for the correlation between inflammation severity and frequency of different immune cells as a proportion of the total cell population (n = 224 biopsies). d Correlogram of the indicated immune cell proportions and Banff histological lesions using Pearson’s correlation. Colors indicate correlation coefficient. i, interstitial inflammation; ptc, peritubular capillaritis; v, vasculitis; mvi (g+ptc), microvascular inflammation; t, tubulitis. e, f Frequency of the indicated cell subsets measured by deconvolution and stratified according to clinical outcome. The difference between groups was assessed by a two-tailed Kruskal-Wallis test and multiple comparisons using the Dunn’s test. Data were obtained from the GSE147089 test set f Data derived from the public GSE36059 validation set.

Fig. 3. The cellular composition of rejection characterized at the protein level using Multiple Iterative Labeling by Antibody Neodeposition (MILAN).

a Experimental Approach. MILAN was used for characterization of infiltrating cells in kidney biopsies at the protein level in 18 kidney biopsies. Created using Biorender.com. b Representative sample illustrating the MILAN analysis with hematoxylin and eosin (H&E) staining (left), composite fluorescence image of 5 markers (plus autofluorescence, AF) after image processing (center), digital reconstruction of the sample highlighting the main cell populations of interest. c Uniform Manifold Approximation and Projection (UMAP) plot of 7804 representative cells. d Relative proportions of the 555,479 cells identified in the kidney transplant biopsies (n = 18). e Polar plot of the 18 biopsies. The biopsies were reclassified into 6 groups: non-rejecting donor-specific antibody (DSA) negative (NR DSA-) and DSA positive (NR DSA+), T-cell mediated rejection (TCMR) and antibody-mediated rejection DSA positive (ABMR DSA+) and DSA negative (ABMRh DSA-) and Mixed rejection. f Spearman’s correlation coefficient represented by a dot with 95% confident interval represented by the error bars of the correlation between inflammation severity and the frequency of different immune cells as a proportion of the total cell population (n = 18 biopsies). g Correlations between the frequency of indicated cell types and inflammation severity. Spearman’s correlation coefficient and two-tailed p-value are indicated.

Fig. 4. Distinct monocytes populations are enriched in inflamed biopsies.

a Reintegration of DC, neutrophils and macrophages populations and Uniform Manifold Approximation and Projection (UMAP) plot of the 4795 corresponding cells. b Spearman’s correlation coefficient represented by a dot with 95% confident interval represented by the error bars of the correlation between inflammation severity and the frequency of different immune cells as a proportion of the total cell population (n = 18 biopsies). c Correlations between the frequency of indicated cell types and inflammation severity. Spearman’s correlation coefficient and two-tailed p-value are indicated.

Fig. 5. Immune cells are differently compartimentalized in the kidney allograft.

a Radar plots showing the percentage of indicated cell populations as a proportion of the total cell population measured in the glomerular, tubular, vascular, large vascular and interstitial compartments. b Percentage of indicated cells measured in the glomerular vs. non-glomerular area. The mean ± SEM is depicted.The difference between groups was assessed by the two-tailed p value calculated with Mann-Withney test. c Representative case showing a composite fluorescent image of 3 markers (plus autofluorescence, AF) after image processing (top), and the resulting digital reconstruction highlighting the cell types of interest (bottom). The reader should note that FcγRIII+ NK cells and FcγRIII+ nonclassical monocytes infiltrate the glomerulus whereas classical monocytes are located outside of it. d Enrichment of 6 different immune cell types based on their distance to the closest glomerulus. The y-axis represents the percentage of cells (from the total cell population) belonging to a given cell type at a maximum given distance (x-axis) from the closest glomerulus. Different samples from the same Banff classification were pulled to get a single line per category.

Fig. 6. FCGR3A+ monocytes are involved in innate myeloid allorecognition.

a All myeloid cells identified in the 16 scRNAseq samples were reintegrated in a UMAP plot of 1,168 myeloid cells. b Dot plot demonstrating average gene expression values of canonical lineage markers (log scale) and percentage of myeloid cell types represented in the UMAP plot. c Correlations between inflammation severity (the radius on the polar plot) and the proportion of each myeloid subset relative to the total number of cells assessed in each biopsy. Spearman’s correlation coefficient represented by a dot with 95% confident interval represented by the error bars. d Pseudotime trajectories for monocytes and macrophages based on Slingshot, showing the common branch of CD14+ monocytes differentiating into either FCGR3A+ monocytes or CD68+, then into CD163+ macrophages. The light grey error band represents the SEM of the gene expression e Profiling of marker genes along these trajectories.

Fig. 7. Transcriptional changes in FCGR3A+ NK and FCGR3A+ monocytes during rejection.

a, b The differentially expressed genes (DEGs) of FCGR3A+ cells were identified by the FindMarker function in Seurat using Wilcoxon test, according to the diagnostic phenotype of the samples (n = 16). a The fold change-fold change (FC-FC) plot compares the transcriptional differences within FCGR3A+ monocytes, between non-rejecting donor-specific antibody (NR DSA-) and NR DSA+ (y-axis) and antibody-mediated rejection (ABMR) (x-axis). The highlighted DEGs represent transcripts with log FC > 1.5 or log FC < −1.5 in ABMR cases compared to NR DSA-. b FC-FC plot comparing the DEGs in FCGR3A+ NK cells between NR DSA- and NR DSA+ (y axis) or ABMR (x axis). The highlighted DEGs represent transcripts with log FC > 1.5 or log FC < −1.5 in ABMR. c, d Dot plots demonstrating average gene expression values of selected genes (log scale), and percentage of FCGR3A+ Monocytes expressing indicated genes (c) or FCGR3A+ NK cells expressing indicated genes (d), according to inflammation severity (the radius on the polar plot) represented by the rank of the biopsies in the plot.

Fig. 8. FCGR3A+ monocytes present specific communication patterns.

a All the significant communications contributing to secreted signaling inferred by CellChat analysis are depicted. b Ligand-receptor pairs that significantly contribute to the secreted signaling from FcγRIII+ NK to all other cell types. c Ligand-receptor pairs that significantly contribute to the secreted signaling from FCGR3A+ monocytes to all other cell types. d The inferred LGALS9-HAVCR2 signaling network. e, f The inferred CXCL10 signaling network encompassing CXCL10-CXCR3 (e) and CXCL10-ACKR1 ligand-receptor pairs (f).

Fig. 9. FCGR3A+ monocytes neighborhood analysis.

a All the significant communications contributing to cell-to-cell signaling inferred by CellChat analysis are depicted. b Ligand-receptor pairs that significantly contribute to the cell-to-cell signaling from FCGR3A+ NK (left panel) or FCGR3A+ monocytes (right panel) to all other cell types. c Neighborhood analysis results showing the enrichment of cell-cell interactions between FCGR3A+ monocytes and FCGR3A+ NK cells or CD8 T eff for the different Banff classification groups. The y-axis represents the number of random cases with a higher (if above 0) or smaller (if below 0) number of interactions than the observed data (see methods). The x-axis represents the distance at which one cell is considered to be in the neighborhood of another cell.

Fig. 10. FCGR3A+ cells mainly interact with the endothelium by direct contact but also through secreted mediators.

a FCGR3A+ cells were reintegrated with endothelial/epithelial cells in a UMAP plot. b UMAP depicting the concomitant expressions of two markers of injury: VCAM1 and HAVCR1. c Dot plot demonstrating select average gene expression values (log scale) and percentage cell types represented in the UMAP plot. d CellChat analysis was performed and the number of incoming and outgoing ligand-receptor interactions is plotted per cell type. e All the significant communications contributing to cell-to-cell signaling inferred by CellChat analysis are depicted: ligand-receptor pairs that significantly contribute to the secreted effectors (left panels), or cell-to-cell direct contact from FCGR3A+ NK or FCGR3A+ monocytes to all other cell types (right panels).