Tracing the Shared Foundations of Gene Expression and Chromatin Structure

Area of Science:

• Genomics and Molecular Biology
• Computational Biology
• Gene Regulation

Background:

• Chromatin's three-dimensional organization into topologically associating domains (TADs) influences gene regulation by mediating long-range gene interactions.
• Technical limitations in defining TAD boundaries and assessing their direct impact on gene expression have hindered a full understanding of TAD function.

Purpose of the Study:

• To develop novel methods for defining TADs and their gene composition, enabling new functional interpretations and capturing species-specific organizational differences.
• To leverage a generative AI foundation model for contextual similarity to better represent functional gene relationships than traditional co-expression methods.
• To investigate the role of TADs in gene co-regulation, cellular plasticity, development, aging, and disease, and to develop predictive tools for cancer drug response.

Main Methods:

• Developed consensus TAD maps for human and mouse using a "bag-of-genes" approach.
• Utilized a generative AI foundation model trained on 33 million transcriptomes to define contextual similarity.
• Created "TAD signatures" for enhanced statistical analysis of single-cell transcriptomic data.

Main Results:

• Demonstrated that TADs actively facilitate gene co-regulation, potentially via transcriptional condensates.
• Found that TAD-mediated transcriptional enhancement is strongest during early development and diminishes with age.
• Observed distinct TAD usage patterns in cancer cells that change with chemotherapy, indicating roles in cellular plasticity.
• Developed "TAD signatures" to improve prediction of cancer cell-line drug response from transcriptomic data.

Conclusions:

• Chromatin organization via TADs plays an active role in gene co-regulation and cellular plasticity.
• TAD function and impact vary across developmental stages, aging, and disease states like cancer.
• TADs operate through probabilistic mechanisms, offering new avenues for understanding and treating diseases.