Design of a Labile RNase A Using Protein Language Models

Area of Science:

• Biochemistry
• Protein Engineering
• Molecular Biology

Background:

• Protein language models (PLMs) are advanced tools for generating functional protein sequences.
• Current research prioritizes enhancing protein stability for industrial uses, overlooking the potential of designing less stable proteins.
• Easily inactivated enzymes can simplify molecular biology workflows by eliminating the need for physical removal after use.

Purpose of the Study:

• To explore the engineering of functional, yet less stable proteins using Ribonuclease A (RNase A) as a model.
• To develop a protein variant with reduced stability that retains enzymatic activity but is easily inactivated.
• To demonstrate the concept of engineering "worst of the best" enzymes.

Main Methods:

• Utilized protein language models (PLMs) to sample sequences from the embedding space near wild-type RNase A.
• Engineered a variant, TempRNase, designed for reduced stability while maintaining RNA degradation function.
• Employed a fluorometric RNA degradation assay to assess TempRNase stability under heat and reducing conditions.

Main Results:

• Successfully engineered TempRNase, a variant of RNase A with significantly reduced stability.
• Demonstrated that moderate heat and reducing treatments permanently inactivate TempRNase, with minimal impact on wild-type RNase A.
• Sequence and structural analyses provided insights into the mechanisms of stability modulation and protein dynamics.

Conclusions:

• Established the feasibility of engineering functional but less stable proteins, termed "worst of the best" enzymes.
• RNase A serves as an effective model system for quantitatively tuning protein stability.
• The developed TempRNase offers a promising tool for simplifying molecular biology protocols through controlled inactivation.