Dynamic cellular programs of human cardiac allograft rejection revealed by spatial transcriptomics

Figures

• 

  Fig. 1. High-resolution spatial mapping of the cellular landscape in human cardiac allograft rejection. a, Study overview. b, Alluvial plot demonstrating rejection grading at the time of diagnosis in the left column, immunomodulatory therapies in the middle column, and biopsy grading following therapies in the right column. B cell-targeted therapies include combinations of intravenous immunoglobulin, plasmapheresis, rituximab, bortezomib and eculizumab. Cytolytic therapies represent thymoglobulin. Combination therapies correspond to combinations of cytolytic and B cell-targeted therapies. c, UMAP representing cell types identified. A total of 10,343,560 high quality transcripts were measured among 162,638 cells. d, Bar plot demonstrating the 28 cell types identified and their respective cell numbers measured. e, Dot plot heatmap of select marker genes used for annotated cell types in the dataset. See Extended Data Fig. 1 for the expanded version. f, Cell type composition across rejection types. DCs, dendritic cells; CMs, cardiomyocytes. Schematics in a created in BioRender; Banovich, N. https://biorender.com/n62l518 (2025). Source data

• 

  Fig. 2. Spatial colocalization in acute cardiac allograft rejection. a–d, Cell proximity results across ACR pretreatment samples (a and b) and AMR pretreatment samples (c and d). The heatmaps (a and c) show the log odds ratio (OR) where positive values (red) represent cell-type pairs that are proximal to each other while a negative value (blue) represents cell-type pairs that are less likely to be found near each other. Nonsignificant results (FDR >0.05) are coded as 0. Highlighted examples of proximal cell types are overlaid on H&E stainings for ACR (b) and AMR (d) biopsies. Cells are colored by cell types, and arrows show specific significant source cell–neighbor pairs. In the ACR biopsy, the following source cell–neighbor pairs are highlighted: mDC–CD8+ T cell, mDC–Treg, NK–endothelial and NK–pericyte. In the AMR biopsies, the following source cell–neighbor pairs are highlighted: macrophage–fibroblast, proliferating pericyte–fibroblast, POSTN+ fibroblast–fibroblast, and BMX+ activated endothelial–fibroblast. Source data

• 

  Fig. 3. Substantial transcriptional heterogeneity exists within and across histologic rejection grades. a, Representative examples of rejection types and severity. Clinically meaningful rejection grades warranting augmented immunomodulatory therapies include 2R and 3R within ACR and pAMR1-i and pAMR2 within AMR. Diagnosis of AMR required the presence of C4d deposition (bottom portion of a) in ≥50% of capillaries within the biopsy sample. b, Heatmap showing transcriptional similarity within and between biopsy grades. A positive log(OR) (red) represents biopsy grade pairs that are more transcriptionally similar, while negative log(OR) (blue) represents biopsy grade pairs that are less transcriptionally similar; extreme values of −Inf and Inf correspond to odds ratios of 0 and infinity, respectively, where biopsy grade pairs exhibit zero or absolute overlap in transcriptionally similar cell states. Asterisks denote significant (FDR <0.05) biopsy grade pair associations. We excluded the nearest-neighbor biopsy grade 0 for visualization purposes, as none of the comparisons was significant. c, PCA plot of all three rejection types based on pseudobulk gene expression demonstrates lack of clear…

• 

  Fig. 4. Cell-type-resolved transcriptomics defines molecular signatures of acute rejection type and therapyresistance. a, Volcano plot representing genes with significant differences in expression between ACR and AMR. A positive log2FC means upregulated in ACR. A negative log2FC means upregulated in AMR. The color of each dot represents its corresponding cell type. Transparency reflects adjusted P values, with transparent dots below the dashed black line representing those with adjusted P values ≥0.05. Differential expression was assessed using a two-sided Welch’s t-test, and P values were adjusted for multiple comparisons using the Benjamini–Hochberg FDR. Supplementary Table 3 contains the full DEG results. b, Bar plot demonstrating the number of DEGs among cell types. c, GSEA depicting results of hallmark pathways that showed significance at an FDR <0.1. A positive normalized enrichment score (NES) means enriched in ACR, while a negative NES means enriched in AMR. Supplementary Table 4 contains full results of the GSEA. d, Overlap between DEGs (adjusted P value <0.05) in Xenium compared with DEGs (Benjamini–Hochberg-corrected P value <0.05 and |log2FC| >1.5) in a publicly available EMB…

• 

  Fig. 5. Cell-type-resolved expression profiles during acute rejection and following antirejection therapiesare associated with CAV. a, Coronary angiogram of a patient included in this study with moderate CAV (CAV grade 2). White arrows point to diseased portions of the left anterior descending artery and a diagonal branch. b, Association between cell-specific gene expression across biopsies spanning all three types of rejection and following immunomodulatory therapies. The x axis represents the β from the GLMM, while the y axis represents the −log10(P value). The color of each dot represents its corresponding cell type. Transparency reflects adjusted P values, with transparent dots representing those with adjusted P values ≥0.05. Associations were estimated using a GLMM (Gaussian family) with a patient-level random intercept, and significance was based on a two-sided test; P values were adjusted for multiple comparisons using the Benjamini–Hochberg FDR. Supplementary Tables 9–13 contain the full results of the GLMM models across all biopsies. c, The top 50 unique genes (arranged by absolute value of their β in the GLMM models) and cell types within those top betas are depicted. d, Dot…

• 

  Extended Data Fig. 1. Marker genes for all cell types. Dotplot heatmap of a combination the top five expressed marker genes for each cell type and canonical marker genes to demonstrate cell type annotation in our data set. Source data

• 

  Extended Data Fig. 2. Spatial organization of immune cells in an ACR patient biopsy. Spatial dim plot for a biopsy from an ACR patient. Non-immune cell types are colored in grey, immune cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of immune cells. Transcripts are colored in green.

• 

  Extended Data Fig. 3. Spatial organization of immune cells in an AMR patient biopsy. Spatial dim plot for a biopsy from an AMR patient. Non-immune cell types are colored in grey, immune cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of immune cells. Transcripts are colored in green.

• 

  Extended Data Fig. 4. Spatial organization of endothelial cells in an ACR patient biopsy. Spatial dim plot for a biopsy from an ACR patient. Non-endothelial cell types are colored in grey, endothelial cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of endothelial cells. Transcripts are colored in green.

• 

  Extended Data Fig. 5. Spatial organization of endothelial cells in an AMR patient biopsy. Spatial dim plot for a biopsy from an AMR patient. Non-endothelial cell types are colored in grey, endothelial cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of endothelial cells. Transcripts are colored in green.

• 

  Extended Data Fig. 6. Spatial organization of mesenchymal cells in an ACR patient biopsy. Spatial dim plot for biopsy from an ACR patient. Non-mesenchymal cell types are colored in grey, mesenchymal cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of mesenchymal cells. Transcripts are colored in green.

• 

  Extended Data Fig. 7. Spatial organization of mesenchymal cells in an AMR patient biopsy. Spatial dim plot for biopsy from an AMR patient. Non-mesenchymal cell types are colored in grey, mesenchymal cell types are colored based on the key provided. Overlaid transcripts (green) on H&Es for set of genes used for cell type annotation of mesenchymal cells. Transcripts are colored in green.