Extended stop codon context predicts nonsense codon readthrough efficiency in human cells.

Area of Science:

• Molecular Biology
• Genetics
• Computational Biology

Background:

• Protein synthesis termination relies on stop codons, but near-cognate tRNAs can cause readthrough.
• Understanding stop codon readthrough is crucial for developing therapies targeting genetic disorders.

Purpose of the Study:

• To investigate the regulatory mechanisms of stop codon readthrough using machine learning.
• To assess the conservation of readthrough factors across species and predict therapeutic responses.

Main Methods:

• Machine learning analysis of ribosome profiling data from human (HEK293T) and yeast experiments.
• Comparative analysis of readthrough efficiency based on sequence context and genetic factors.
• Model validation using data from drug-treated cells and cystic fibrosis nonsense mutations.

Main Results:

• Identified conserved factors in readthrough regulation, including stop codon identity, context, and 3'-UTR length.
• Found evidence suggesting a role for the P-site tRNA in readthrough regulation, which is not conserved.
• Developed predictive models that accurately forecast readthrough of premature termination codons in cystic fibrosis.

Conclusions:

• Machine learning models can predict stop codon readthrough efficiency based on sequence context.
• The predictive models show potential for guiding the development of personalized nonsense suppression therapies.
• This approach may help predict patient response to readthrough-promoting drugs for genetic diseases.