Identification of Critical Phosphorylation Sites Enhancing Kinase Activity With a Bimodal Fusion Framework

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• Phosphorylation is a key cellular process regulating protein function, particularly kinase activity.
• Critical phosphorylation sites (phos-sites) enhance kinase activity, but many remain undiscovered.
• Existing methods lack a systematic computational approach to identify these crucial phos-sites.

Purpose of the Study:

• To develop the first systematic computational method for identifying critical phos-sites on kinases.
• To leverage both phosphorylation mass spectrometry and protein sequence data for enhanced prediction.
• To discover novel phos-sites that significantly enhance kinase activity.

Main Methods:

• Introduced PhoSiteformer, a transformer-inspired model for generating phos-site embeddings from mass spectrometry data.
• Developed CSPred, a classification model using a bimodal fusion strategy.
• Combined PhoSiteformer embeddings with protein language model (ProtT5) embeddings.

Main Results:

• Successfully identified 77 critical phos-sites on 58 human kinases.
• Experimentally validated two novel sites: T517 on PKG1 and T735 on PRKD3.
• Demonstrated the efficacy of the bimodal fusion strategy in predicting critical phos-sites.

Conclusions:

• Presented the first systematic computational approach to identify kinase-activating phos-sites.
• PhoSiteformer and CSPred offer a powerful tool for exploring kinase regulation.
• The identified sites provide new targets for understanding and manipulating kinase function.