Deep learning based prediction of reversible HAT/HDAC-specific lysine acetylation

Area of Science:

• Molecular Biology
• Bioinformatics
• Cancer Research

Background:

• Protein lysine acetylation is crucial for cellular processes and implicated in diseases like cancer.
• While many acetylation sites are known, enzyme-specific regulation by histone acetyltransferases (HATs) and histone deacetylases (HDACs) remains poorly understood.

Purpose of the Study:

• To develop a deep learning tool, Deep-PLA, for predicting HAT/HDAC-specific protein lysine acetylation.
• To investigate the role of protein-protein interactions in acetylation regulation.
• To analyze the impact of acetylation-associated mutations in cancer signaling.

Main Methods:

• Curated experimentally identified HAT/HDAC substrates and sites from literature.
• Integrated protein sequence features with a deep neural network.
• Optimized hyperparameters using particle swarm optimization.
• Analyzed protein-protein interaction networks and cancer mutation data.

Main Results:

• Deep-PLA achieved high performance in predicting enzyme-specific acetylation sites, outperforming previous methods.
• Protein-protein interactions were found to enrich enzyme-specific acetylation regulatory relationships.
• Acetylation regulation is significantly disrupted by mutations in cancer, impacting cancer signaling pathways.

Conclusions:

• Deep-PLA provides a valuable tool for understanding enzyme-specific protein acetylation regulation.
• The findings highlight the disruption of acetylation in cancer and its role in signaling.
• This research offers insights for cancer prognosis and treatment strategies.