Predicting Macrophage Spatial Localization from Single-Cell Transcriptomes to Uncover Disease Mechanisms

Area of Science:

• Genomics
• Computational Biology
• Immunology

Background:

• Single-cell RNA sequencing (scRNA-seq) loses crucial spatial information when cells are isolated.
• Organs with compartmentalized anatomy, like the kidney, present unique challenges for spatial reconstruction.
• Understanding cell location is vital for interpreting cellular function and disease pathology.

Purpose of the Study:

• To develop and validate an algorithm (MERLIN) for reconstructing positional information from scRNA-seq data in compartmentalized organs.
• To investigate the predictability of immune cell localization within the kidney.
• To explore the functional and disease-specific transcriptomic signatures associated with cell position.

Main Methods:

• Generation of independent immune cell scRNA-seq datasets from distinct renal compartments.
• Training and evaluation of machine learning algorithms, including a modified multi-layer Perceptron.
• Application of MERLIN to predict cell positions in healthy and disease models (crescentic glomerulonephritis, acute kidney injury, diabetic nephropathy).
• Analysis of positional transcriptomic fingerprints and enrichment pathways.

Main Results:

• MERLIN achieved over 75% accuracy in predicting the positions of resident macrophages in murine and human kidneys.
• Motile immune cells, such as lymphocytes, were not accurately predictable by position.
• Positional transcriptomic data revealed enrichment in microenvironmental response and adaptation pathways, with a notable sex bias.
• MERLIN successfully predicted macrophage localization in kidney disease models and aligned with known pathology and therapeutic efficacy in specific compartments.
• The algorithm was also trained to predict spatial distribution of brain microglia.

Conclusions:

• MERLIN enables the spatial interpretation of scRNA-seq data in anatomically defined organs.
• The algorithm enhances mechanistic insights into disease processes by linking cell location to function and response.
• MERLIN provides a powerful tool for spatial transcriptomics in complex tissues.