Therapeutic enzyme engineering using a generative neural network.

Area of Science:

• Biochemistry
• Molecular Biology
• Bioengineering

Background:

• Enhancing messenger RNA (mRNA) therapeutics is crucial for treating rare genetic disorders, potentially allowing for reduced dosing frequency and lower doses.
• Enzyme engineering can improve mRNA therapeutic potency by optimizing the expression, half-life, and catalytic efficiency of encoded enzymes.
• Current methods for mapping protein sequence to function are often inaccurate, time-consuming, or labor-intensive, hindering rapid enzyme engineering.

Purpose of the Study:

• To develop a broadly applicable, rapid engineering method for identifying metabolic enzyme variants with improved thermal stability and catalytic activity.
• To enhance the potency of mRNA therapeutics by optimizing enzyme function for rare disease treatment.

Main Methods:

• A novel engineering approach combining deep latent variable modeling of sequence co-evolution with automated protein library design and construction.
• Application of the method to engineer ornithine transcarbamylase (OTC), a key enzyme in the urea cycle.

Main Results:

• Identification of metabolic enzyme variants exhibiting both increased thermal stability and enhanced catalytic activity.
• Successful application of the novel engineering strategy to improve the potency of ornithine transcarbamylase (OTC).

Conclusions:

• The presented method offers a rapid and broadly applicable strategy for engineering enzymes with improved stability and activity.
• This approach has significant potential for advancing the development of potent mRNA therapeutics for rare metabolic diseases.