Researchers utilize the Hicformer transformer model to link genome folding and transcriptomic shifts in Alzheimer’s disease, revealing a new regulatory layer of neurodegeneration.
A collaborative research team has identified a fundamental link between 3D genome architecture and neurodegenerative progression in Alzheimer’s disease. By applying a multimodal analytical framework to postmortem prefrontal cortex tissue, investigators demonstrated that genome folding patterns serve as a critical, previously underappreciated regulatory layer in disease pathology, with the transformer-based AI model Hicformer providing the predictive architecture necessary to decode these complex interactions.
The study, published in Science, employed GAGE-seq to capture simultaneous measurements of gene expression and physical genome contacts within individual cells. Researchers from Carnegie Mellon University, the University of Pittsburgh, and the University of Washington integrated these single-cell multiomics data with spatial transcriptomic maps to construct a comprehensive view of cellular states. This approach allowed the team to correlate structural genomic shifts with specific transcriptional programs across diverse brain cell types, establishing a new baseline for understanding how the physical structure of the nucleus influences disease progression.
Central to the analytical pipeline was Hicformer, a transformer-based AI model designed to integrate DNA sequence data with 3D genome features. The model effectively predicted cell-type-specific gene activity by accounting for chromatin accessibility and spatial organization, demonstrating that 3D genome features provide information beyond DNA sequence alone. By enabling the prioritization of distal regulatory elements whose effects are mediated through chromatin contacts, the model offers a high-resolution view of the regulatory landscape.
The data revealed significant structural reorganization within the genome of affected cells, characterized by a reduction in short-range interactions and an increase in longer-range contacts. This shift suggests a breakdown in the segregation of active and inactive genomic compartments, leading to widespread dysregulation of gene expression. These findings indicate that the spatial arrangement of chromatin is not merely a byproduct of cellular state but a primary driver of disease-relevant pathways.
The investigation also highlighted specific cellular impacts, including evidence of senescence-related activation in microglia and sex-dependent dysregulation of X-linked genes in female subjects. Spatial transcriptomic analysis further confirmed that these molecular changes manifest as altered cellular neighborhoods within the brain. The integration of these datasets provides a unified view of how genome structure, gene regulation, and tissue organization collapse during the progression of Alzheimer’s disease.
Researchers also observed a weakening of promoter-proximal interactions, which suggests that the physical proximity of regulatory elements to gene promoters is compromised in diseased tissue. This structural degradation appears to disrupt the fine-tuned coordination of gene programs, potentially explaining the loss of cellular identity observed in neurodegeneration. By mapping these interactions at the single-cell level, the team identified specific signatures of 3D genome remodeling that correlate with the severity of the disease.
Yang Zhang, PhD, a project scientist in Carnegie Mellon’s Computational Biology Department, noted that the paired measurement of gene activity and chromosome structure provides a consistent signature of 3D reorganization. This methodology allows researchers to move beyond static observations and identify specific regulatory regions for future mechanistic validation. The ability to directly link chromosome structure to disease-related gene programs remains a primary advantage of this single-cell approach.
Jian Ma, PhD, the Ray and Stephanie Lane Professor of Computational Biology at Carnegie Mellon University, emphasized that the complexity of Alzheimer’s disease necessitates a multi-layered analytical perspective. The research demonstrates that genome folding acts as a fundamental regulatory layer connecting DNA sequence to functional gene activity. By synthesizing genome folding, cell state, and tissue context, the team aims to transition from cataloging changes to identifying actionable biological mechanisms.
The study establishes a framework for future experiments designed to isolate which structural alterations contribute directly to neurodegeneration. By prioritizing distal regulatory elements through chromatin contact mapping, the researchers have identified new potential targets for therapeutic intervention. Ongoing work will focus on testing these structural mechanisms in experimental models to determine their causal role in disease development and to evaluate whether these genomic signatures can be modulated to mitigate neuronal damage.
