Directed Evolution for Protein Engineering

Directed evolution is a method for engineering proteins that involves mimicking natural selection in a laboratory setting. We interviewed experts from several companies to understand where this approach—which originated in the 1960s—is heading.

What is directed evolution?

According to Lance Encell, Senior Research Scientist at Promega, “Protein engineering enables us to improve enzyme function—including thermostability, specific activity, processivity, and resistance to inhibitors. In most cases, the goal is to optimize multiple properties at once.”

He notes that the two main approaches to protein engineering are rational design and directed evolution. “Rational design is a hypothesis-driven approach that requires an understanding of the role that individual amino acids play in structure and function—something that we often do not understand.” This strategy typically focuses on making single amino acid substitutions at specific locations and then testing variants for desired properties.

In contrast to rational design, directed protein evolution involves the introduction of gene mutations in a process called mutagenesis. “The great thing about directed evolution is that you don’t need a lot of knowledge about protein structure. Instead, you create gene mutations and then screen the resulting protein variants.”

Directed evolution to perfect bioluminescent enzymes

Promega has been applying directed evolution to engineer novel protein sequences for nearly 30 years. One notable example is the company’s bioluminescent luciferase enzymes, which are used in many of its technologies that detect binding events in live cells—including NanoBiT and NanoBRET assays.

“Although natural bioluminescence is remarkable, the deep-sea shrimp whose enzyme we started with did not possess all the properties needed for our assays.” Therefore, Promega employed directed evolution methods for many years to engineer enzymes with critical features such as enhanced signal intensity and stability.

Promega used a multi-step, iterative process involving random mutagenesis and high-throughput screening. “Random mutagenesis is a form of directed evolution that does not target specific sequence positions,” Encell explains. “Instead, you are looking for beneficial mutations that happen to accrue over time.” He notes that advances in DNA synthesis, molecular cloning, lab automation, and machine learning have improved the feasibility of directed evolution research at Promega over time.

Synthetic DNA to introduce variation

Emily Leproust, CEO and Cofounder of Twist Bioscience, says that in contrast to random mutagenesis approaches, Twist’s DNA synthesis platform harnesses the semi-rational use of synthetic DNA. She notes that the platform is advantageous when the researcher has some knowledge of the protein and an idea about where to introduce variation.

Leproust explains that adding sequence variation within a specific region allows the researcher to focus on a particular function. This then reduces the number of sequences that must be screened to discover desirable variants.

“Compared with random mutagenesis, our techniques ultimately lead to a more sophisticated and efficient engineering workflow for testing multiple variants and selecting the best proteins for advancement.” She adds that emerging computational approaches (including structural prediction tools based on large language models) can also help to select the best proteins before screening begins.

More efficient approaches to directed evolution

New England Biolabs provides a comprehensive set of DNA assembly and mutagenesis tools and products to assist in constructing and expressing variant gene libraries, according to Dr. Jennifer Ong, Scientific Director of Research.

For example, the company has a product (called the NEBuilder HiFi DNA Assembly) that can create site-specific diversity on the protein of interest using degenerate oligonucleotides. Another product (the NEBridge Golden Gate Assembly) is a single-tube DNA assembly method that can cut and paste multiple DNA fragments. Ong notes that this latter product is ideal for swapping or combining protein elements.

NEB also aims to reduce the timeline needed to construct libraries. Ong says that the company has integrated automation throughout its platform, enabling the high-throughput screening of protein variants.

The company is also pioneering cell-free protein synthesis with in vitro systems like the NEBExpress^® Cell-free E. coli Protein Synthesis System. Ong adds that the combination of automated methods and cell-free protein expression has reduced the time for constructing and screening variant libraries from seven to three days.

Introducing directed engineering

“Our approach involves using machine learning to efficiently search sequence-function space and navigate to optimal solutions,” Mark Welch, Vice President of R&D at ATUM, explains. “The goal is to simultaneously consider multiple criteria—such as stability, protein yield, quality, and function—for optimal protein product performance.” He adds that ATUM uses proprietary informatics technologies to analyze phylogenetic and other protein function and structure data to inform engineering strategy. Candidate proteins can then be more carefully screened using high-quality assays.

But is ATUM’s machine learning approach a form of directed evolution? “Directed evolution means something different to different people,” replies Welsh. “We like to think of our platform more as directed engineering, where we determine the relationship between protein sequence and properties of interest and use that to drive engineering of improved variants.”

As an example of ATUM’s services, Welch relates a story about how a client approached the company with the desire to improve the affinity of an antibody without any deleterious impacts on product titer, aggregation, stability, and, most importantly, pharmacokinetics.

Despite trying various engineering strategies, the client had been unable to meet baseline requirements. “However, we were able to use our platform to engineer the antibody for improvements in all of these desired respects without any immunogenicity issues.”

Advice from an expert

Ong emphasizes the importance of choosing the appropriate assays for screening. “As a young scientist working in directed evolution, I was always told that you get what you screen for.” She adds that the biggest challenge in commercial protein engineering often involves developing high-throughput enzyme assays that closely align with the final application.

Ultimately, Ong stresses that protein engineering is a field that encourages creative thinking. “However, it is easy to get carried away with the excitement and creativity of a new approach. The key is to stay focused on the end goal, which typically involves finding a practical or simple solution. This balance between creativity and practicality is what makes protein engineering both challenging and rewarding.”