TL;DR
This study presents a comprehensive single-cell multi-omic analysis of the primate and rodent neocortex, revealing how gene regulatory programs evolve across species. By integrating gene expression, chromatin accessibility, DNA methylation, and 3D genome conformation data from over 180,000 cells, the researchers identified conserved and divergent regulatory elements, highlighting the significant role of transposable elements in human-specific regulatory innovation and demonstrating the power of deep learning to predict regulatory elements from DNA sequence alone.
What Problem Does This Paper Address?
Evolutionary changes in cis-regulatory elements are known drivers of species-specific traits, but understanding how these molecular changes translate to the evolution of complex structures like the neocortex at a single-cell level has been a significant challenge. Previous studies often relied on bulk tissue analysis, masking cell-type specific differences and evolutionary dynamics. This paper aims to bridge this gap by providing a high-resolution, multi-omic view of gene regulation across different species and cell types within the primary motor cortex, enabling a deeper understanding of how molecular evolution shapes brain complexity and function.
Methods
The study employed a multi-modal single-cell approach, integrating four key molecular modalities:
- 10x Multiome: Simultaneously profiled gene expression (RNA) and chromatin accessibility (ATAC) from individual nuclei.
- snm3C-seq: Captured 3D genome conformation (Hi-C) and DNA methylation (mCG and mCH) at single-cell resolution.
These assays were performed on the primary motor cortex of human, macaque, marmoset, and mouse. For analysis, the researchers developed a new framework for assessing evolutionary conservation and divergence of gene regulatory features. Machine learning, specifically deep learning models (LegNet), was used to predict chromatin accessibility from DNA sequence alone, demonstrating the conserved "grammar" of regulatory elements.
Datasets
The study utilized single-cell multi-omic data from:
- Human: Primary motor cortex (M1)
- Macaque: Primary motor cortex (M1)
- Marmoset: Primary motor cortex (M1)
- Mouse: Primary motor cortex (MOp)
A total of over 180,000 cells were profiled across the four modalities. The deep learning model was trained on data from three species and evaluated on a fourth unseen species, utilizing syntenic regions to minimize data leakage.
Key Insights and Ideas
- Epigenetic states evolve faster than promoters: Distal cis-regulatory elements (cCREs) show faster epigenetic evolution compared to promoters, suggesting a dynamic regulatory landscape.
- Transposable Elements (TEs) are major drivers of human-specific innovation: TEs contribute to a substantial portion (nearly 80%) of human-specific cCREs in cortical cells, highlighting their role in generating novel regulatory functions.
- Genomic regulatory syntax is highly conserved: Despite rapid evolution of epigenetic states, the underlying genomic regulatory grammar and syntax are remarkably preserved from rodents to primates, as demonstrated by deep learning models.
- Epigenetic conservation aids disease variant interpretation: Epigenetic conservation at cCREs is correlated with the conservation of their target genes' expression and is more enriched for genetic variants associated with neurological diseases, suggesting a powerful tool for interpreting non-coding disease risk variants.
Key Results
- The study profiled over 180,000 single cells across four species and four molecular modalities.
- Cell type fractions were highly species-specific, with notable expansions/reductions (e.g., oligodendrocytes expanding from mouse to human).
- Approximately 20% of orthologous genes showed conserved expression patterns across all species, while 25% exhibited species-biased expression.
- Mammal-level conserved cCREs were more likely to act as enhancers and harbor disease risk variants.
- Deep learning models achieved high accuracy in predicting cell-type specific chromatin accessibility from DNA sequence alone, demonstrating a conserved regulatory grammar.
- Leveraging epigenetic conservation significantly improved the enrichment of GWAS variants for neurological traits compared to using all cCREs.
Limitations and Future Work
- Limited species scope: While covering rodents and primates, the study could be expanded to include a broader range of species to further elucidate evolutionary trajectories.
- Experimental limitations: Despite the large cell numbers, some rare cell types or subtle regulatory differences might be missed.
- Future directions: The authors suggest expanding the repertoire of species in epigenomic datasets to overcome challenges in achieving highly accurate sequence-based predictions of gene expression. Further investigation into the precise mechanisms by which TEs contribute to species-specific regulation and the functional consequences of epigenetic divergence on neurological traits are also promising avenues.
Summary updated