Structure-based prediction of C2H2 zinc-finger binding specificity: sensitivity to docking geometry

Area of Science:

• Computational biology
• Structural biology
• Genomics

Background:

• Transcription factor (TF) binding specificity prediction is vital for understanding gene regulation.
• Computational identification of cis-regulatory elements relies on accurate TF binding site prediction.
• Protein-DNA structural information offers a promising avenue for improving binding specificity predictions.

Purpose of the Study:

• To assess the accuracy of predicting TF binding specificity using protein-DNA structures.
• To investigate the impact of docking geometry on prediction accuracy.
• To establish a quantitative measure for structural similarity (Interface Alignment Score - IAS) to ensure accurate binding specificity predictions.

Main Methods:

• Utilized protein-DNA structures to predict TF binding specificity.
• Employed a molecular-mechanics force field for predicting high-affinity nucleotide sequences.
• Quantified docking similarity using the Interface Alignment Score (IAS).
• Tested predictions using the second zinc-finger (ZF) domain from Zif268 and other C2H2 ZF domains as templates.

Main Results:

• Demonstrated a strong correlation between IAS values and the accuracy of binding specificity predictions.
• Defined a specific range of IAS values that are expected to yield accurate structure-based predictions.
• Showcased the potential to predict Position Weight Matrices (PWMs) for entire protein families from limited structural data.

Conclusions:

• Accurate structure-based prediction of TF binding specificity is achievable when docking geometries meet defined similarity thresholds (IAS).
• The IAS provides a reliable metric for evaluating the suitability of structural templates for binding specificity prediction.
• These findings support the feasibility of large-scale, structure-based prediction of PWMs for diverse transcription factor families.