Integration of spatial transcriptomics and immunofluorescence staining to enable colocalized multiomics analysis in chronic liver disease

Le résumé fourni par la source

Spatial transcriptomics (ST) and multiplexed immunofluorescence (mIF) imaging offer complementary insights into disease biology, but integrating these modalities remains challenging. This study developed an automated pipeline to colocalize and analyze mRNA and protein expression at the ST spot level in liver tissue, focusing on hepatocytes in chronic liver disease. Eight liver samples (6 diseased, 2 healthy) underwent ST and mIF data acquisition. Image co-registration was performed using affine and non-rigid (B-Spline) transformations, achieving a registration error of 9µm (IQR: 5–17μm). Cell segmentation on mIF images utilized the Cellpose model, leveraging DAPI and NaK-ATPase signals to define cell boundaries. Hepatocyte phenotyping was performed using an adapted UNI pathology foundation model, fine-tuned for three-class biomarker classification (Negative, mIF Weak, mIF Strong). The model was trained on manually labeled data generated using MONAI Label and QuPath, incorporating 2.1k mIF Strong, 1.7k mIF Weak, and 15k Negative hepatocytes, with selective sampling to address data imbalance. The model achieved >80% agreement with manual annotations for Negative and mIF Strong classes and ⁓50% for mIF Weak spots. Our findings reveal spatial heterogeneity and discrepancies between mRNA and protein expression, with some spots exhibiting mRNA positivity despite the absence of detectable protein, suggesting biological processes such as protein secretion or degradation. This highlights the value of integrating ST and mIF data for a more comprehensive understanding of disease mechanisms. The developed pipeline enables large-scale, spatially resolved molecular profiling while reducing manual effort, offering a powerful tool for studying chronic liver disease and other tissue-based pathologies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.