Skip to Content

Computers Boost Antibody-Based Drugs

Physics-based modeling may accelerate the drug-development process.
October 1, 2007

A computer program developed at MIT could vastly improve the design of antibody drugs. The software identified improvements in the anticancer drug cetuximab that increased its binding affinity by a factor of 10 in subsequent laboratory tests–a change that could lead to lower drug dosages and increased drug efficacy.

Better antibodies: Cetuximab, shown here binding to its target, a fragment of epidermal growth factor receptor, is an antibody-based drug used to treat cancer. MIT researchers used a computational model to redesign the antibody, allowing it to bind more tightly to the target.

“Through mathematical modeling of complex biological systems, you can make predictions about how to improve them in a directed, designed way,” says Bruce Tidor, a computational biologist at MIT who led the study. “That’s usually not what’s done in medicine.”

Our immune system constantly produces antibody molecules, which identify and disarm foreign particles, termed antigens. To treat conditions ranging from psoriasis to cancer, doctors take advantage of the body’s natural defense system using antibody-based drugs. For example, anticancer therapeutics such as cetuximab bind to malfunctioning receptors on the cell surface, stopping them from triggering uncontrolled tumor growth.

However, developing antibody-based drugs is a complex and time-consuming process. Scientists harvest antibodies from mice that have been injected with the target antigen. They then try to optimize the molecules by growing them in yeast and searching for random mutations that make the antibody bind to its target more tightly. Tidor and his colleagues aim to improve this process by transferring it to a computer. “The implication is that by having less trial and error and more design,” he says, “there’s much more control over what comes out.”

The system starts with a template antibody molecule and then swaps in each of our body’s 20 amino acids at 60 different positions. The molecule’s ability to bind to its target is calculated mainly from the electrostatic forces acting on its chemical components. Using this approach, the team can analyze far more sequence combinations than they could ever test in the laboratory.

Tidor and graduate student Shaun Lippow, now at Codon Devices, in Cambridge, MA, tried out their system on several antibody drugs, generating candidate molecules that were then tested in the lab with collaborator Dane Wittrup. The breakthrough success came with cetuximab, when they discovered multiple mutations that could generate a molecule with 10 times greater binding efficiency. Although there are no guarantees that higher binding affinities make for better drugs, “All the evidence so far is that tighter is better,” Tidor says. The findings were published last week in the journal Nature Biotechnology.

Researchers in the field hailed the paper as a major success for computational biology. “The exciting thing about doing this thing computationally is you can save a lot of time in the wet lab,” says Bruce Donald, a computational biologist at Duke University who studies protein structures.

Protein engineer Katarina Midelfort of Pfizer, based in Groton, CT, was impressed by the multiple successes of the team’s approach. She says that in many cases, improvements in binding affinity require that all the mutations work in concert, but the MIT team has been able to identify multiple mutations that provide stepwise improvements to molecules.

So why can Tidor’s model produce antibodies that seem to be superior to those generated by Mother Nature? Tidor explains that “evolutionary stumbling blocks” make it difficult to simultaneously evolve the different mutations needed to optimize a molecule’s binding affinity. In the MIT study, ten out of the twelve mutations that the team identified required such evolutionary leaps.

It’s still too early to know how soon patients will be able to reap the benefits of computational biology in the form of better medicines. But Tidor and his colleagues hope to make that process as easy as possible. They have made the sequences of their improved antibodies freely available, and are developing new versions of the system that are easier and faster to use. They also plan to develop similar models to improve small molecule drugs.

Keep Reading

Most Popular

DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.

“This is a profound moment in the history of technology,” says Mustafa Suleyman.

What to know about this autumn’s covid vaccines

New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.

Human-plus-AI solutions mitigate security threats

With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure

Next slide, please: A brief history of the corporate presentation

From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.