Deep learning-guided design of cell type-specific AAV promoters

Area of Science:

• Molecular Biology
• Genetics
• Bioengineering

Background:

• Precise cell type targeting is crucial for adeno-associated viral (AAV) vector applications in clinical and experimental settings.
• Engineering AAV vectors with cell type-specific activity remains a significant challenge in the field.

Purpose of the Study:

• To compare strategies for designing cell type-specific AAV promoters using single-cell chromatin accessibility data.
• To evaluate a deep learning-based method for generating de novo regulatory sequences for AAV vectors.
• To assess the in vivo performance and translational potential of deep learning-designed synthetic promoters.

Main Methods:

• Leveraged single-cell chromatin accessibility data to design AAV promoters.
• Employed a deep learning-based method to generate de novo regulatory sequences.
• Tested synthetic promoters in mouse retina targeting retinal ganglion cells and horizontal cells.
• Validated promoter activity in human retinal organoids.

Main Results:

• Deep learning-guided promoter design outperformed rational approaches in vivo.
• Synthetic promoters exhibited stronger and more specific expression in targeted retinal cells.
• The designed promoters supported diverse transgenes for cell recording and ablation.
• Promoter activity was conserved in human retinal organoids, indicating translational potential.

Conclusions:

• Deep learning offers a powerful approach to synthesize cell type-specific AAV promoters.
• The developed platform provides a versatile tool for cell type targeting in gene therapy and research.
• Findings suggest broad implications for advancing gene therapy and basic neuroscience research.