Engineering bispecific antibodies is challenging, creating unwanted side products. CrossMAb technology, designed as a simple platform for complex molecules, solves those problems.

By Laura Fraser, New York Times bestselling author

More than thirty years ago, not long after scientists began developing therapeutic antibodies—breakthrough medicines that precisely bind to single targets —they began thinking about the potential advantages of medicines that could hit two targets at once: bispecific antibodies.

If one molecule could hit two targets, patients might be able to take one medicine instead of two, with potentially fewer side effects, enduring fewer injections. Pharmaceutical companies might get a drug combination to market faster by producing one molecule instead of two, with one set of clinical trials, and without having to test a regimen in combination. Most intriguingly, it seemed there were instances where delivering two functionalities in one molecule could accomplish what the same functions in two medicines could not, such as by simultaneously binding to tumour cells and delivering toxic payloads, or recruiting immune cells into close proximity to cancer cells in order to kill them.

“Bispecific antibodies provide access to new modes of action that have not been possible with single antibodies,” says Stefan Weigand, Roche’s Global Head of Large Molecule Research. But the problem with bispecific antibodies is that they are very hard to engineer. The beauty of single-target antibodies is that they are close to molecules made by nature and so the body often accepts them with few side effects; but the more you add to the molecules, the less the body likes them, which can mean more unwanted side effects. “These days there are literally hundreds of different ways of making bispecific antibodies,” says Stefan Weigand, “But few have made it into later-stage clinical trials, partly because while many of them work on paper, they are very far away from the original antibodies.” That can affect everything from how difficult the molecules are to manufacture and their stability in that process to how well patients will eventually tolerate the medicines.

The essential problem with engineering bispecific antibodies is how to guarantee correct assembly of the parts to deliver only the desired molecule, while omitting unwanted side products. Classic antibodies have two identical heavy chains and two identical light chains. But combining the function of two different antibodies into one bispecific antibody requires two different heavy and two different light chains. The cell then randomly assembles these chains, creating ten different antibodies—only one of which is the desired product. The rest are unwanted side products that need to be purified. Some of these side products are easily separable and not critical, says Weigand. But others are hard to separate and critical in terms of toxicology and side effects, and make large-scale production difficult, if not impossible.

“This is the key issue that delayed the entire field of bispecific antibodies for years,” says Janice Reichert, Executive Director of the Antibody Society. “People couldn’t efficiently make a molecule of high enough quality to do clinical studies.”


In an important first step to simplifying the production of bispecific antibodies, in 1997 Genentech scientist Paul Carter was the main inventor of the “knobs-and-holes” technology that solved the problem of the two heavy chains being paired correctly. The light chains, however, still randomly connected with the heavy chains. Now instead of ten antibodies produced, there were four. Since then, several companies have developed diverse techniques for creating bispecific antibodies without unwanted side products. “It took years for people to work out their technology and manufacturing,” says Reichert.

There are a lot of viable technologies, but the hard part for any of the bispecific antibodies is to get them through phase three trials and have them be safe and effective in the patient population.

While there are many in various phases of development, only three bispecific medicines have been approved worldwide, one of which was later withdrawn from the market. However, the interest and promise in these molecules is clear.


Bispecific antibodies currently are in clinical trials.


of bispecific antibodies in clinical trials are bringing immune cells closer to cancer targets.

A new technology to produce just one bispecific molecule

Now, Roche has taken bispecific technology a step further and developed the next generation in bispecific antibody engineering: CrossMAb, a technology that produces one bispecific molecule—not ten, not four, but just what the scientists need. The engineers had to solve the problem of preventing the light antibody chains from binding to the heavy chains in ways they didn’t want. Weigand describes the process as being similar to putting together LEGO™ blocks: by exchanging the molecular blocks between the heavy and light chain, each of the two different arms binds to a specific light chain. With this elegant engineering, Roche has developed a technology that can combine two antibodies and create the bispecific antibody they desire—with practically no side products.

“It’s one of the few methodologies that generates a molecule that is essentially indistinguishable from a natural antibody, except that it binds to two antigens,” says Weigand. “That’s the trick about it—in hindsight, it’s very simple.”

The fact that the molecule is similar to a natural antibody means that the body’s defense system is less likely to recognise it as foreign. Since the molecules don’t have to go through a lengthy process of purification, they’re much easier and faster to tinker with. There’s one cell line, and one production process; it’s just one molecule from the start. “We can very rapidly prepare hundreds of bispecific antibodies for laboratory testing to select the best combination possible.” The simplicity of the technology will make it faster to produce and scale—which will ultimately get the medicines to patients more rapidly.1

The CrossMAb platform is so robust and simple that it can also be used to develop trispecific antibodies—with three targets.2 It can also support a variety of complex additions to classic antibodies. The CrossMab technology is a powerful tool to extend the therapeutic protein format landscape. “This creates an additional diversity to address biology that is unprecedented in other methodologies” Weigand says.

The CrossMAb molecules are the basic building blocks to which other functions can be added in the future. One is the brain shuttle—a bispecific antibody that crosses over the blood brain barrier that could deliver antibodies to target3 neurological diseases such as Alzheimer’s or Parkinson’s, allowing for lower doses of medicines than systemic small molecules. Another is an extra arm that would help activate or dampen immune cells, depending on the disease, by combining a cancer-targeting antibody arm with an antibody arm targeting immune cells. Ultimately, the simple, strong CrossMAb building block will make it much easier to develop personalised health care—medicines that are designed to fit a specific biological need, rather than trying to fit a broad number of diseases, such as cancers, to one medicine.

Meanwhile, the first CrossMAb molecule, for ophthalmology, is about to enter phase 3 trials, and, if the treatment for macular edema proves to be significantly more effective than the current standard, will be submitted to the FDA for approval. At Roche, the large molecule portfolio is currently comprised of approximately 80% complex molecules, including bispecifics based on the CrossMAb technology.

There are still many engineering challenges in the way in the future of these flexible antibodies. “Despite the beauty of the technology, our molecules are getting more and more complex, and the more they deviate from a standard antibody as it is produced by the body’s defense, the higher risk of side effects that are hard to test and predict,” Weigand says. But starting with a simple technology will help.

“Though there are a hundred different ways of making bispecific antibodies, they’re not all equal,” says Janice Reichert. “The CrossMAb technology has a leg up in being versatile, robust, and the molecules are developable and manufacturable. Once you solve the technical problems and can make the molecule in large quantities, then the world is your oyster.”

Tags: Science