Background/Purpose: Lupus nephritis (LN) is a heterogeneous disease driven by diverse immune and tissue cell types. We defined the cell states in the tissue and blood and tested their association with histopathological indices.

Methods: We obtained 538K single-cell and 143K single-nuclear profiles from kidney biopsies of 156 LN patients and 30 controls from pre-implantation transplant biopsies. We also obtained 327K single-cell blood profiles from a subset of 119 cases and 19 controls (Figure 1). All LN patients were >18 years, had urine protein/creatinine ratio >0.5 at biopsy, and were International Society of Nephrology (ISN) class III, IV, V, or combined III or IV with V.

Results: We built an atlas of kidney tissue cell types in LN encompassing 55 immune and 36 tissue states. Integrating tissue immune single cell data with blood data enabled classification of tissue-specific immune cell states (Figure 2). For example, we identified autoimmune-associated B cells (ABCs, B5) as a cell state shared between tissue and blood while plasma cell populations (P6-8) were tissue-specific. In the myeloid compartment, we identified plasmacytoid dendritic cells (DC20) as a shared population, while macrophage cell states were tissue-specific (M5-14). Using Covarying Neighborhood Analysis, we identified LN pathological features associated with cell state changes (Figure 3). Increasing chronicity tracked with the most dramatic changes in both tissue and immune states. The proportion of proximal tubule cells that were injured or degenerating (PT0, PT7) expanded with chronicity, reflecting ongoing, irreversible tissue damage. Increasing chronicity was associated with an expansion of tissue-specific CD56 bright NK cells (NK1), GPNMBhigh NUPR1high Macrophages (M5), and CLEC10Alow cDC2 (DC17). Most expanded immune populations were tissue-specific states. It was essential to identify activity-associated populations since they may be driving reversible inflammation in LN. Adjusting away chronicity, we identified specific activity-associated cell states in the myeloid compartment. Highly tissue-specific macrophage populations were expanded, including M5, GPNMBhigh LYVE1low (M11), C1Qlow SPP1high, SPP1high FABP5high (M7), and MERTKhigh FABP5high (M9). We speculate that M5 may be a profibrotic pathogenic population that may be targeted for therapeutic benefit. After adjusting for chronicity and activity, ISN classes were not independently associated with specific changes in cell states. Blood cell states minimally associated with activity or chronicity. We observed case-status associated with interferon signatures in both blood and tissue.

Conclusion: Our results define key cell states in the tissue and immune compartments and argue that profound changes in kidney tissue and immune cell states are associated with increasing chronicity. Specific myeloid compartment changes are associated with activity. These features are not associated with substantial changes in the blood. Our results argue that myeloid populations may be targeted, in addition to B cell therapies, to prevent progression of tissue inflammation and ongoing tissue damage in LN.

Figure 1. a, Schematic overview of the data collection and analysis pipeline. Blood and kidney samples were collected from LN patients and healthy controls (HC). Histopathology and clinical scoring was performed on renal biopsies. CITE-sequencing was performed on blood samples, and scRNA-sequencing and snRNA-sequencing were performed on tissue samples. Both tissue modalities were integrated with Harmony (PMID: 31740819). Broad cell types were annotated separately in tissue and blood. Fine grain clustering for cell state annotation was performed on each broad cell type. For each tissue, covarying neighborhood analysis was performed within each cell type to define cell state associations with various clinical variables. b, UMAPs of all single cell or single nuclear annotations in tissue (left) and blood (right), colored by broad cell type annotation.

Figure 2. a-b, Joint embedding of tissue (a) and blood (b) myeloid single-cell populations, colored by their original fine-grain annotations obtained in a separate embedding. c, Tissue specificity metric (TSM) scores stratified by cell state annotation. TSM was calculated as the proportion of a given cell’s 50 nearest neighbors that are from tissue, weighted to account for larger amount of total myeloid blood than tissue myeloid cells.

Figure 3. a, UMAP of myeloid cell states in tissue, colored by fine grain cell state annotations. Annotation boxes are located at the centroid (mean) of each cluster. b, UMAP of per-cell correlation coefficient associations with activity index, as quantified in CNA. c, TSM for cells significantly expanded or depleted with activity index in CNA. d-e, CNA correlation coefficients for cells in each myeloid cluster, indicating cell state correlations with activity index (d) and chronicity index (e).

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/deconstructing-lupus-nephritis-kidney-tissue-at-single-cell-resolution/