Computational prediction resolves thousands of homooligomeric phage protein structures.

Area of Science:

• Structural biology
• Computational biology
• Virology

Background:

• Bacteriophages (phages) are crucial in microbial ecosystems, but their proteins are largely uncharacterized.
• Understanding protein structure, especially quaternary structure, is key to determining protein function.
• Many phage proteins form homooligomers (complexes of identical subunits), making their prediction a significant interest.

Purpose of the Study:

• To develop and validate a computational framework, PHLEGM, for predicting homooligomeric states of phage proteins directly from their amino acid sequences.
• To assess the accuracy and utility of PHLEGM by experimental validation and benchmarking against other methods.
• To explore the diversity of phage homooligomeric protein complexes within a large phage protein database.

Main Methods:

• Developed PHLEGM, integrating AlphaFold-Multimer modeling with interface quality assessment for homooligomer prediction from protein sequences.
• Experimentally validated two predicted homooligomers (a dimer and a trimer) using size exclusion chromatography and hydrodynamic techniques.
• Applied PHLEGM to over 22,000 phage protein sequences from the PHROGs database and benchmarked against protein language model predictors.

Main Results:

• PHLEGM successfully predicted homooligomeric states, with experimental validation confirming predictions for a dimer and a trimer.
• Application to the PHROGs database revealed significant diversity in phage homooligomeric protein complexes.
• PHLEGM demonstrated superior accuracy compared to protein language model-based predictors, particularly for trimers and higher-order complexes.

Conclusions:

• PHLEGM is a robust, structure-based computational tool for predicting phage homooligomeric states, advancing the understanding of viral protein function.
• The study highlights the value of experimentally benchmarked computational predictions in deciphering the vast phage sequence space.
• Publicly released predicted structures and functional inferences will facilitate future research on phage proteins.