Abstract: TH-PO533
Topological Analysis of High-Resolution Spatial Transcriptomics Reveals Immune Glomerular Architecture of Lupus Nephritis
Session Information
- Glomerular Diseases: From Inflammation to Fibrosis - I
November 02, 2023 | Location: Exhibit Hall, Pennsylvania Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1401 Glomerular Diseases: From Inflammation to Fibrosis
Authors
- Bull, Katherine R., University of Oxford Nuffield Department of Medicine, Oxford, United Kingdom
- Benjamin, Katherine, University of Oxford Mathematical Institute, Oxford, Oxfordshire, United Kingdom
- Bhandari, Aneesha, University of Oxford Nuffield Department of Medicine, Oxford, Oxfordshire, United Kingdom
- Shang, Zhouchun, BGI Group, Shenzhen, Guangdong, China
- Xing, Yanan, BGI Group, Shenzhen, Guangdong, China
- An, Yanru, BGI Group, Shenzhen, Guangdong, China
- Zhang, Nannan, Bejing Genomics Institute-Qingdao, Qingdao, China
- Tillmann, Ulrike, University of Oxford Mathematical Institute, Oxford, Oxfordshire, United Kingdom
- Harrington, Heather A., University of Oxford Mathematical Institute, Oxford, Oxfordshire, United Kingdom
Group or Team Name
- Oxford Kidney Pathology Research Group and Oxford Mathematics Institute.
Background
To develop targeted treatments for immune complex diseases such as lupus nephritis (LN) we need to understand immune and renal cell interactions. The advent of spatial sub-cellular resolution whole transcriptome technologies offers the potential to study cell distribution within glomeruli, but requires new mathematical approaches. Conventional ‘fixed-window’ approaches bin data across spots, discarding the granularity of a high-resolution platform and missing small or rare cells such as immune infiltrates.
Methods
Murine kidneys treated with topical imiquimod or vehicle were prepared for single nuclei RNA sequencing (snRNA-Seq, 10x Genomics) and 200nm spot spatial transcriptomics (STOmics, Beijing Genomics Institute) and analysed (Seurat, RCTD). We developed topological automatic cell type identification (TopACT), combined with multiparameter persistent homology (MPH) to quantify multiscale spatial cell organization. TopACT independently classifies and annotates spot level cell type using a dynamic local neighbourhood.
Results
On synthetic data imputed from renal snRNA-Seq, TopACT produced high accuracy spot-level cell type annotations in comparison to fixed-window approaches. In murine kidney, TopACT spatially resolved individual immune cells in LN, enriched within glomeruli, where the average MPH landscape indicated large loops of immune cells (Fig 1A). This leads to the prediction, driven by spatial data, and confirmed by CD45 immunofluorescence, of a peripheral ring structure of glomerular immune cells in LN (Fig 1B and C).
Conclusion
Our multiscale method for topological automatic cell classification improves accuracy of cell-type information for subcellular resolution spatial transcriptomics, and detects the spatial arrangement of glomerular immune cells in LN. TopACT is generalisable, flexible and has potential for further application to 3D or spatiotemporal data.
Funding
- Commercial Support – Funding from UKRI/ Medical Research Council and Kidney Research UK. Beijing Genomics Institute collaborated on the research - no direct financial support /funding to research team.