Researchers from Los Alamos National Laboratory are helping restore the efficacy of Evusheld, a therapeutic used to treat COVID infections. The results were published today in Nature.
“As COVID mutated, many therapeutics became less effective,” said Antonietta Lillo, of Los Alamos’ Biochemistry and Biotechnology group and coauthor of the paper. “We used our long-standing antibody development capability to validate the restored therapeutic molecules that could renew the potency of drugs such as Evusheld.”
Evusheld is an antibody-based therapeutic approved for immunocompromised COVID patients. Omicron variants, which emerged after the drug was developed, acquired mutations to the spike protein binding domain, making it more difficult for Evusheld therapeutic antibodies to attach to the virus, rendering the drug far less effective.
Los Alamos is part of a collaboration that includes Lawrence Livermore and Sandia National Laboratories, which used a computationally driven approach, called Generative Unconstrained Intelligent Drug Engineering (GUIDE), to address the efficacy of the antibodies. GUIDE — led by the by the Joint Program Executive Office for Chemical, Biological, Radiological, and Nuclear Defense — combines high-performance computing, simulation and machine learning (an application of AI) to co-optimize binding affinity against multiple antigen targets.
The GUIDE scientists scanned tens of thousands of antibody variants to identify mutations that could improve binding without incurring additional problems. The team also sought to maintain the antibodies’ efficacy against other circulating SARS-CoV-2 variants.
Los Alamos scientists have an established capability to develop high-affinity antibody fragments using yeast cells. This capability allows scientists to create a library of antibodies, each displayed on a yeast cell, that enables high-throughput screening to rapidly identify the antibodies that bind with the highest affinity and specificity to a target of interest. Furthermore, the scientists can identify the exact genetic sequence that encode the best antibodies, so these antibodies can be reliably reproduced in large scale.
The Los Alamos team used their high-throughput approach to evaluate 450 of 700 antibody sequences computationally redesigned by their GUIDE collaborators. The Los Alamos set contained sequences encoding antibodies whose behavior was difficult to predict (long-shot candidates). The high-throughput yeast display-based screening significantly reduces the time and cost needed to validate the AI-developed antibodies.
“Producing and testing each long shot antibody as single molecules would have been too risky an investment, whereas en masse analysis through yeast display technology was a much less costly endeavor” Lillo said.
The GUIDE program, which launched in 2022, is an interagency effort between the Department of Defense, Department of Energy, academia, and industry. Through this program, GUIDE partner institutions integrate their computational and experimental capabilities to accelerate drug development, preparing for and potentially preventing the next pandemic. As part of the overall GUIDE effort, the Los Alamos partners are using their high-throughput selection capability to validate AI-developed therapeutics, and to improve AI models by providing experimental data on efficacy, manufacturability and toxicity.
Paper: “Computationally restoring the potency of a clinical antibody against SARS-CoV-2 Omicron subvariants.” Nature. DOI: 10.1101/2022.10.21.513237
Funding: The GUIDE program is executed by the Joint Program Executive Office for Chemical, Biological, Radiological, and Nuclear Defense, Joint Project Lead for Enabling Biotechnologies on behalf of the Chemical and Biological Defense Program. This effort was in collaboration with the Defense Health Agency COVID funding initiative.
LA-UR-24-24420