Protein codes promote selective subcellular compartmentalization

Area of Science:

• Cell Biology
• Molecular Biology
• Bioinformatics

Background:

• Cells must precisely localize billions of proteins to specific compartments for proper function.
• Proteins with shared functions need to efficiently assemble in designated subcellular locations.

Purpose of the Study:

• To investigate if amino acid sequences contain codes for subcellular protein localization.
• To develop a predictive model for protein compartment destination.
• To explore the role of this code in protein assembly and disease.

Main Methods:

• Development of a protein language model, ProtGPS, trained on human protein sequences.
• Prediction of subcellular localization for proteins not included in the training set.
• Generation of novel protein sequences guided by ProtGPS for targeted compartment assembly.
• Identification of mutations affecting protein localization codes.

Main Results:

• Proteins with shared functions utilize common amino acid sequence codes for compartment targeting.
• The ProtGPS model accurately predicts subcellular localization of human proteins.
• ProtGPS successfully guided the creation of novel proteins that assemble in specific compartments.
• Pathological mutations altering this code were identified, leading to mislocalized proteins.

Conclusions:

• Protein sequences encode a previously unrecognized 'distribution code' for subcellular localization, in addition to the 'folding code'.
• This code is crucial for the assembly of functional protein complexes within cells.
• Alterations in this protein localization code can contribute to disease pathogenesis.