Deconvolving the recognition of DNA shape from sequence.

Area of Science:

• Molecular Biology
• Genetics
• Biophysics

Background:

• Protein-DNA binding relies on recognizing DNA base chemistry and 3D shape.
• Disentangling sequence-based and shape-based recognition is challenging.
• Hox proteins are crucial transcription factors involved in development.

Purpose of the Study:

• To investigate the independent role of DNA shape in Hox protein binding specificity.
• To determine if DNA shape recognition is a distinct mechanism from sequence recognition.

Main Methods:

• Mutagenesis of Hox protein residues involved in DNA shape recognition.
• Assessing binding preferences of mutant Hox proteins to DNA sequences.
• Transferring shape-recognizing residues between Hox proteins to observe functional changes.
• Employing statistical machine learning to analyze sequence and shape features for binding prediction.

Main Results:

• Mutant Hox proteins lacking shape-recognition residues lost specific DNA shape preferences.
• Transferring shape-recognizing residues altered Hox protein binding specificity in vitro and gene regulation in vivo.
• Machine learning models showed improved accuracy in predicting binding specificity when DNA shape features were included.

Conclusions:

• DNA shape readout is a direct and independent component of Hox protein binding site selection.
• Shape recognition contributes significantly to the specificity of Hox-DNA interactions.
• Understanding shape readout provides new insights into gene regulation by Hox proteins.