Single-cell spatial multi-omics and deep learning dissect enhancer-driven gene regulatory networks in liver zonation

Area of Science:

• Molecular Biology
• Genomics
• Systems Biology

Background:

• Hepatocytes in the mammalian liver display distinct metabolic and functional characteristics based on their position within the liver lobule, a phenomenon known as zonation.
• The underlying gene regulatory mechanisms that establish and maintain this spatial zonation pattern remain incompletely understood.

Purpose of the Study:

• To map enhancer-gene regulatory networks across mouse liver cell types.
• To elucidate the regulatory code governing hepatocyte identity and zonation.
• To investigate the role of specific transcription factors and enhancers in driving these spatial states.

Main Methods:

• Single-cell multiomics and spatial omics technologies were employed to profile gene expression and chromatin accessibility.
• Massively parallel reporter assays were utilized to functionally test enhancer activity.
• A hierarchical deep learning model, DeepLiver, was developed to analyze enhancer-gene relationships.

Main Results:

• Zonation significantly impacts gene expression and chromatin accessibility in hepatocytes and other liver cell types.
• Key transcription factors, including TCF7L1, TBX3, HNF4A, CEBPA, FOXA1, and ONECUT1, were identified as drivers of hepatocyte identity and zonation states.
• The study successfully mapped enhancer-gene regulatory networks underlying these spatial variations.

Conclusions:

• This research provides a multimodal understanding of the regulatory code controlling hepatocyte identity and liver zonation.
• The findings offer insights into the complex interplay of transcription factors and enhancers in establishing spatial gene expression patterns.
• The developed deep learning model and identified regulatory networks can potentially be used to engineer enhancers with specific activity and zonation patterns.