To be or not to be: predicting soluble SecAs as membrane proteins

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• SecA is crucial for bacterial protein translocation, existing in soluble and membrane-integrated forms.
• Existing membrane prediction tools inconsistently identify SecA's membrane-associated segments despite high sequence homology across species.

Purpose of the Study:

• To develop a novel, consistent method for predicting integral membrane domains of bacterial SecA proteins.
• To improve the accuracy and reliability of membrane protein prediction for SecA.

Main Methods:

• Developed and applied a new prediction method, PSSM_SVM, utilizing a PSSM encoding scheme.
• Evaluated PSSM_SVM accuracy and consistency on a blind test set and compared it with existing methods (TMpred, Top-Pred).
• Analyzed predicted membrane segments for 223 known SecA sequences.

Main Results:

• PSSM_SVM demonstrated superior accuracy (Q2) and consistency in predicting SecA's integral membrane domains across bacterial species.
• The method consistently predicted eight to nine residue embedded membrane segments within SecA sequences.
• These predicted segments align with known 12-residue helical structures in SecA crystal structures.

Conclusions:

• The PSSM_SVM method offers a reliable approach for predicting bacterial SecA membrane integration.
• Findings contribute to understanding SecA's structural and conformational properties within lipid bilayers.
• This method has implications for studying other soluble proteins that integrate into lipid bilayers.