Protein Engineering Method May Lead To More Exact Treatments

Enzymes called proteases act like molecular scissors for proteins in the body and play a role in therapies to stop viruses from replicating and to kill cancer cells. The development of these medicines, however, has been slowed by the difficulty of predicting how individual proteases will behave.

Researchers at The University of Texas at Dallas have developed a machine-learning model that analyzes how proteases have evolved over time to predict whether a specific protease will carry out an intended task before testing it in the lab.

To calculate how a specific protease will cut a target protein substrate, the interdisciplinary team built a model called ProSSpeC (protease substrate specificity calculator). ProSSpeC was able to suggest engineered synthetic proteases that outperformed widely used enzymes, demonstrating its potential to lead to more targeted therapies. Their research was published online Feb. 26 in Nature Communications.

"Our approach to predicting protease activity opens the door to designing new enzymes that could be used for more precise and effective treatments across a range of diseases," said Dr. P.C. Dave Dingal, assistant professor of bioengineering in the Erik Jonsson School of Engineering and Computer Science and co-corresponding author of the study.

"Our approach to predicting protease activity opens the door to designing new enzymes that could be used for more precise and effective treatments across a range of diseases."

Dr. P.C. Dave Dingal, assistant professor of bioengineering in the Erik Jonsson School of Engineering and Computer Science

Dingal collaborated with Dr. Faruck Morcos, associate professor of biological sciences in the School of Natural Sciences and Mathematics and co-corresponding author.

Their research combines bioengineering with evolutionary biology - the study of how living organisms change and adapt over time - and computational biology, which uses computational techniques to analyze large amounts of biological data.

"Our model essentially learns from millions of years of evolution. By analyzing how similar enzymes have changed over time, we can predict which engineered variants will function and which won't - without having to test thousands of random variations in the lab," said Morcos, Fellow, Cecil H. and Ida Green Professor in Systems Biology Science. "It's like letting nature guide us in building better molecular tools."

Protein engineering is similar in concept to genetic engineering, in which scientists manipulate DNA to achieve desired outcomes. Protein engineering is a newer field and focuses directly on designing the proteins themselves to create enzymes that work better, faster or in new ways. Developing medications to block specific proteases has often involved a slow process of trial and error in the lab.

"Our model essentially learns from millions of years of evolution. By analyzing how similar enzymes have changed over time, we can predict which engineered variants will function and which won't - without having to test thousands of random variations in the lab."

Dr. Faruck Morcos, associate professor of biological sciences in the School of Natural Sciences and Mathematics

ProSSpeC compares thousands of related enzymes - each with different amino-acid sequences - from the Potyviridae family of plant viruses and identifies which parts of a protease are essential to its function and which parts can be changed.

After building the model, the researchers validated its predictions in the lab by producing and testing new proteases that the model indicated would be effective. The researchers have filed a provisional patent for some of the enzymes, which are more effective than the tobacco etch virus protease commonly used to purify proteins in research labs and pharmaceutical production.

Biomedical engineering doctoral student Medel B. Lim Suan Jr., co-first author of the study and a Eugene McDermott Graduate Fellow, worked closely with undergraduate students involved in the project. He said he appreciated the opportunity to gain experience in interdisciplinary research.

"It was very cool learning all about modeling and how it translates to the experimental side of things," Lim Suan said. "So, while doing the experiment, I also was learning how the computational work fit into the biology."

Co-authors of the study include Cheyenne Ziegler MS'20, PhD'25; Zain Syed BS'25; molecular biology senior Arjun Sai Yedavalli; Jaimahesh Nagineni BS'25; Rodrigo Raposo; Ajay Tunikipati BS'25; and Jaideep Kaur BS'24.

The research was supported by grants (R35GM150967, R35GM133631) from the National Institutes of Health's National Institute of General Medical Sciences and a National Science Foundation Faculty Early Career Development (CAREER) Award that Morcos received in 2019. In addition, Dingal and Morcos received support from the UT Dallas Office of Research and Innovation through a New Faculty Research Symposium Grant.

Media Contact: Kim Horner, UT Dallas, 972-883-4463, [email protected], or the Office of Media Relations, UT Dallas, (972) 883-2155, [email protected].