Background/Purpose: We recently identified novel immune cell states in the kidneys of lupus nephritis patients (Arazi et al, Nature Immunology 2019). To determine the similarities with lupus mouse models, we compared human immune cell transcriptomes to those from Sle1.Yaa and NZBW mouse kidneys. We focused on macrophages and DCs since intrarenal myeloid abundance may drive tissue damage in lupus patients.

Methods: We collected myeloid cells from dissociated kidneys from pre- and post-nephritic Sle1.Yaa and NZBW mice (3-4 mice per group). Using 10x Genomics we performed single cell transcriptomic profiling and analyzed droplets that contained >500 genes and UMIs after doublet removal. To identify intrarenal mouse immune cells equivalent to those in humans, we leveraged immune cell transcriptomes collected from the kidneys of lupus patients in AMP Phase 1 to generate state-specific gene signatures from each human myeloid cell cluster composed of: (i) the top 10 to 40 most highly and ubiquitously differentially expressed (DE) genes, (ii) the top 4-10 DE transcription factors, or (iii) modules of co-varying genes from non-negative matrix factorization. After clustering mouse myeloid cells (Seurat v3.0), we calculated the average of the scaled expression for each human signature across mouse clusters to identify those most similar to human. We also calculated the Pearson correlation coefficient between human and mouse clusters using homologous genes with identical names between species.

Results: We mapped gene signatures from the human myeloid states: ‘inflammatory’ macrophages (CM0) that likely enter from blood and transition to ‘phagocytic’ (CM1) and ‘reparative’ (CM4) states; resident macrophages (CM2); dendritic cells (CM3). We identified the human CM0 signature in clusters from NZBW (and weakly in Sle1.Yaa) post-nephritic pro-inflammatory macrophage/DC clusters that infiltrate from blood, similar to CM0 cells in humans. The human CM1 signature mapped to clusters in both mouse models that resembled DCs and that were enriched for transendothelial migration. We identified in both models the CM4 signature in pre- and post-nephritic mouse clusters that resembled resident macrophages and were enriched for antigen presentation and lysosomal processing. Like humans, the post-nephritic CM4 mouse equivalent cluster exclusively expressed Ccl8, a leukocyte chemoattractant and activator, and Gas6, a mediator of apoptotic cell clearance and was skewed towards an alternative phenotype. Finally, we mapped the human CM3 signature to mouse CD103+ DCs and identified gene programs shared with humans for immunoregulation and antigen presentation.

Conclusion:

From the kidneys of Sle1.Yaa and NZBW mice we identified human myeloid gene signatures: ‘inflammatory’ macrophages (CM0), ‘phagocytic’ (CM1), ‘reparative’ (CM4) states; dendritic cells (CM3); we did not find the human resident macrophage (CM2) signature. We identified putative functions shared between mouse and human clusters including antigen presentation, phagocytosis, Ccl8 expression. These analyses indicate that we may be able to study in mice some of the myeloid subsets and genes that are highly relevant to human disease.

« Back to ACR Convergence 2020

ACR Meeting Abstracts - https://acrabstracts.org/abstract/single-cell-transcriptomics-of-mouse-and-human-lupus-nephritis-identifies-conserved-myeloid-populations-across-species/