MIT model could improve drug efficacy

An MIT research team has developed a computer model which could improve a class of drugs based on antibodies, essential to a healthy immune system.

According to the team, the model can predict structural changes in an antibody and thus improve its efficacy.

Antibodies, which are produced by the human immune system to defend the body from infection, are often used for diagnostics and therapeutics. Starting with a specific antibody, the MIT model looked at many possible amino-acid substitutions which could occur in the antibody. It then calculates which substitutions would result in a structure that would form a stronger interaction with the target.

The MIT model has already been used to create a new version of cetuximab for the treatment of colorectal cancer, helping the drug to bind to its target with 10 times greater affinity than the original, thus reinforcing the treatment's market presence.

The team also used the model with an anti-lysozyme antibody called D44.1. They were able to achieve a 140-fold improvement in its binding affinity. The authors expect the model will be useful with other antibodies as well.

The recent research, which will be published in the 23 September advanced publication of Nature Biotechnology, stems from a collaboration using both laboratory experiments and computer simulations, between Professors Dane Wittrup and Bruce Tidor of MIT.

Researchers developing antibody-based drugs have traditionally used an evolutionary approach. They remove antibodies from mice and further evolve them in the laboratory, screening for improved efficacy, leading to improved binding affinities. However, the process is time-consuming, and it restricts the control that researchers have over the design of antibodies.

In contrast, the MIT computational approach can quickly calculate a huge number of possible antibody variants and conformations, and predict the molecules' binding affinity for their targets based on the interactions that occur between atoms.

Using the new approach, researchers can now predict the efficacy of mutations that might never arise by natural evolution.

Shaun Lippow, who is lead author of the Nature Biotechnology paper, said: "Combining information about protein (antibody) structure with calculations that address the underlying atomic interactions allows us to make rational choices about which changes should be made to a protein to improve its function. Protein modelling can reduce the cost of developing antibody-based drugs."

Janna Wehrle, who oversees computational biology grants at the National Institute of General Medical Sciences, which partially supported the research, said: "Dr Tidor's new computational method can predict which changes in an antibody will make it work better, allowing chemists to focus their efforts on the most promising candidates. This is a perfect example of how modern computing can be harnessed to speed up the development of new drugs."