Work

Defining transcriptomic niches in human fibrotic lung with multi-sample spatial transcriptomics analysis using the MAPLE algorithm

Year: 2022

Type: preprint

Abstract: Idiopathic pulmonary fibrosis (IPF) is an age-related lung disease marked by patchy parenchymal fibrosis that disrupts gas exchange and leads to respiratory failure. Single-cell RNA-seq studies have revealed cellular heterogeneity in IPF, but the spatial contexts underlying alveolar epithelial cell loss, honeycomb cyst formation, and fibroblastic foci remain unclear, motivating comparisons based on multi-sample spatial transcriptomics (ST) across clinical and anatomical conditions. Here, we present a comprehensive pipeline for analyzing multi-sample lung ST data that accounts for experimental design, featuring the MAPLE algorithm to jointly identify cell-spot subpopulations and discover spatial biomarkers, with validation using 10X Xenium ST data. Our analysis reveals that IPF progression involves epithelial-to-mesenchymal state transitions and region-specific fibrotic programs, characterized by extracellular matrix remodeling in the upper lobes and B-cell-enriched immune niches in the lower lobes. We identify fibrotic regions with senescent cell (SnC) enrichment and demonstrate their spatial colocalization with aberrant basaloid niches. Author Summary Idiopathic pulmonary fibrosis is a progressive lung disease in which normal lung tissue is replaced by scar tissue in uneven and patchy patterns. Understanding these patterns requires methods that compare multiple tissue samples while preserving information about where genes are expressed. We developed and applied MAPLE, a statistical framework for jointly analyzing multiple spatial transcriptomics samples while accounting for differences between clinical groups and spatial relationships within each tissue section. Using lung tissues from healthy donors and patients with idiopathic pulmonary fibrosis, we identified shared tissue regions across samples and linked them to cell composition, gene expression, and pathological features. We observed a stepwise decrease in alveolar epithelial and endothelial cells across the analyzed conditions, an increase in mesenchymal states, extracellular matrix remodeling in upper-lobe fibrotic regions, and B-cell-associated programs in lower-lobe regions. We also found that senescence-associated signals were spatially concentrated near aberrant basaloid cell niches. These results demonstrate how multi-sample spatial analysis can reveal disease-associated tissue organization that may be missed when tissue samples are analyzed separately. MAPLE provides an interpretable framework for studying spatially heterogeneous diseases across multiple patients and anatomical regions.

Source: bioRxiv (Cold Spring Harbor Laboratory)

DOI: 10.1101/2022.02.28.482296