© PASIEKA/SCIENCE SOURCE

After selling his first company to Merck in 2006 for $400 million, Dartmouth College bioengineer Tillman Gerngross faced a tough decision about starting another company: choose a field with minimal competition but high risk of failure, or enter a space that was already clinically and commercially successful but crowded with rival technologies. He chose the latter when he chose antibodies. “I hypothesized that if we had a truly better way of discovering antibodies, there was business to be had,” says Gerngross.

Still, at the time, the pharmaceutical industry was already saturated with companies looking to identify new antibodies and engineer existing ones to be more effective. From a distance, pharma’s obsession with these Y-shaped proteins is baffling. Big and clunky proteins, antibodies take twice the time and money to develop into a marketed drug that small molecules do, are terribly complicated to manufacture, cannot be taken...

Yet antibodies have something that small molecules, growth factors, and cytokines lack—exquisite precision. Unlike any other class of drug, an antibody gloms onto a target like a magnet latching onto a needle in a haystack. As therapeutics, antibodies can be used to specifically bind a diseased cell, ignoring healthy cells, and cause the diseased cell to self-destruct, for example, or activate the immune system against it. The large molecules can also suppress the immune overactivity that characterizes autoimmune diseases, bind rogue proteins, deliver toxic payloads to sites of illness, and more. In the late 1990s, antibodies finally lived up to their promise as companies learned how to successfully engineer them. First there was Rituxan, then Herceptin, then Remicade. All are billion-dollar blockbuster drugs, for non-Hodgkin’s lymphoma, breast cancer, and Crohn’s disease, respectively.

Gerngross joined the fray in 2007, cofounding the small New Hampshire-based biotech Adimab, and spent the next 2.5 years building a yeast-based antibody-discovery platform designed to be less complex than existing technologies. And it worked: in 2009 the company signed five partnership deals with pharmaceutical companies; the following year, they inked 10 more. Last year, the first of the company’s antibody therapies headed into clinical trials.

Gerngross’s success is not only a tribute to his company’s technology, but an indication of the industry’s continued thirst for new antibodies. “There’s been a sea change,” says Bassil Dahiyat, president and CEO of Xencor, a California-based antibody biotech. “Because there has been a lot of success with antibodies, the industry now feels that this is actually viable—that there are many multibillion-dollar drugs to be had.”

Today, biotech companies left and right, far more than can be described in a single article, are inventing ingenious new ways to discover and engineer antibodies by tweaking every part of the proteins, from the tips of the Y to the bottom of the stalk. “It’s a great time to be in this space,” says Gerngross. “I’d do it all again.”

The art of discovery

Monoclonal antibodies, first discovered in the 1970s, were successfully made into human drugs in the 1990s, but it wasn’t easy. Antibodies are far too complicated to simply synthesize de novo by even the cleverest chemist. Instead, scientists have developed two major techniques to produce new antibodies. In one method, they expose a mouse to an antigen to elicit an immune response, extract antibody-manufacturing B cells from the spleen of that mouse, fuse those B cells with fast-growing cancer cells, then isolate the antibodies the hybrid cells produce.

Alternatively, researchers can use viral systems to produce antibodies. By inserting bits of antibody sequences pulled from gene libraries into bacteriophage DNA, then transfecting that DNA into E. coli, researchers can quickly express an antibody’s component parts, then piece the whole thing together afterward. After either of these two processes to produce novel antibodies, the proteins undergo further in vivo testing to zero in on molecules that have the desired biological effect.

A DRUG CLASSIC: Monoclonal antibodies, first discovered in the 1970s and the basis of many of today’s blockbuster drugs, are large, highly specific Y-shaped proteins. The tips of the two short arms, called the variable regions of the antibody, bind to an antigen of interest, while the long arm, called the Fc (fragment crystallizable) region, interacts with the host immune system, recruiting macrophages to digest a foreign cell, for example.Both production techniques are well-tested and oft-used in antibody work, but they require many steps, including the removal of undesirable mouse or phage segments, which are immunogenic in humans, from the resulting proteins. So biotech companies have been seeking better methods of antibody discovery. Adimab’s technology produces human antibodies—constructed entirely from human genes—straight out of yeast cells. And thanks to built-in quality-control systems that cause yeast to excrete only properly folded proteins, the end product is a high-quality, fully human antibody.

Tarrytown, New York-based Regeneron is also making fully human antibodies, but does so using genetically engineered mice. After years of making drugs based on receptors for growth factors and cytokines, the company moved into the antibody field when its researchers developed their “VelocImmune” mouse, designed to produce human antibodies. Most mice engineered to produce human antibodies are immunocompromised—and therefore not as efficient at making new antibodies—because of the inclusion of human antibody Fc (fragment crystallizable) regions, the stems of the Y that interact with the host immune system. (See illustration below.) But the mouse’s immune system is critical for the proper working of Regeneron’s technology.

“There are millions of antibody sequences made stochastically in vivo, and the mouse immune machinery picks out the very best ones to bind to the antigen,” says Neil Stahl, the company’s senior vice president of research and development sciences. The researchers therefore engineer the mice to express human antibody variable regions—the tips of the Y-shaped structure that bind a target antigen—and mouse Fc regions, allowing the mouse immune system to select for the best antibody for a given target, such as the sought-after protein PCSK9 (see Sidebar: Eyes on the Prize). Once they have a well-fitting human variable region, they clone that region and attach a human constant region of choice. The company currently has 12 antibodies in clinical trials, seven of which were created with Big-Pharma partner Sanofi.

Yet even when a company discovers a new antibody that firmly binds a desired target, that does not mean the antibody will have the desired biological effect. “Anybody can get an antibody to anything,” says Richard Lerner, an immunochemist at the Scripps Research Institute in La Jolla, California. “The hot new ticket is direct selection for function.” Lerner, who has spent many years developing methods to identify antibodies, has now done exactly that—created a technique to select antibodies for a desired biological function. And the method is already bearing fruit.

Lerner developed an automated system with millions of individually cultured test cells designed to emit fluorescent light if an antibody causes a specific functional result. The cells are infected with lentiviruses, each delivering one antibody gene per cell, so each cell then produces a unique antibody. Then, if an antibody elicits the desired function—such as activating or inhibiting a receptor—the corresponding cell glows under the microscope. The team swoops in, isolates the antibody, sequences it, and immediately has a potent antibody for the desired function. “We can sort through 2 million events in an hour,” says Lerner.

He’s already used the technique to identify potent antibody mimics of erythropoietin (EPO) and thrombopoietin (TPO), hormones that control the production of red blood cells and platelets, respectively. Such antibodies could be used to treat patients lacking these essential blood components, including cancer patients after chemotherapy. Lerner’s team has also identified an antibody that can turn stem-like cells into neurons and an antibody that blocks a virus from killing cells. Numerous pharma companies have expressed interest in licensing the system, says Lerner.

The art of engineering

BUILDING A BETTER DRUG: Many researchers are looking to boost antibodies’ tumor-killing capabilities, and they’re using a number of different tactics. Some scientists are designing “bispecific” antibodies, whose variable regions (the molecule’s two short arms) target both a tumor cell and an immune cell, bringing the cells into close proximity (1). Another strategy is to make multispecific antibodies that bind more than one receptor on a tumor cell, allowing the molecule to inhibit or activate several of the cell’s biological pathways at once (2). Other researchers are attaching tumor-killing drugs to antibodies, designing what are known as antibody-drug conjugates (3). Still others are altering the amino acids of the Fc region—the long arm of the antibody that interacts with the immune system—to make more potent drugs?(4) THE SCIENTIST STAFFIn addition to new ways of producing antibodies, biotechs are pushing the limits of chemical engineering, endowing antibodies with superpowers, such as binding to multiple sites at once, carrying and delivering toxic payloads, and activating or suppressing the immune system with unprecedented precision. (See illustration here.)

One of the simpler techniques is to include two or more mutually exclusive antibodies in a single drug. Each antibody binds the same receptor but in a different way, like using multiple hoses at different angles to put out a fire. “Five years ago, people would have thrown up their hands because of the regulatory issues” of getting multiple antibodies into a single treatment, says Gerngross. But today, “concerns about cocktails are over.” Crucell, for example, a Netherlands-based pharmaceutical company, currently has a two-antibody combo for rabies virus that has completed Phase 2 trials and is being fast-tracked by the US Food and Drug Administration.

Companies have also recently had success in attaching a toxic payload to “arm” the antibody, creating what’s called an antibody-drug conjugate. (See “Tumor Snipers,” The Scientist, November 2012.) Another major push has been to develop antibodies that can bind to two different cells and bring them together. Called bispecific antibodies, these elusive constructs have long been pursued in the industry, but with little success. Stitching together the arms of two different antibodies tends to result in heavily engineered, Rube Goldberg-like conglomerations that are unstable and broken down quickly in the body. Only one such antibody has made it to clinical trials: Amgen’s blinatumomab, which binds both tumor cells and B cells, coupling the immune cells with the cancer they kill. The drug has such a short half-life, however, that leukemia patients in a current Phase 2 trial have to wear an infusion pump to constantly push the antibody into their system.

But the idea of binding two targets at once is so tantalizing that many companies, including Adimab and Regeneron, are now applying their technologies to make bispecific antibodies. “It went from a graveyard to a very hot and contested field overnight,” says William Strohl, vice president of biologics research at Janssen R&D, the research arm of Johnson & Johnson.

Zyngenia, a small biotech based in Gaithersburg, Maryland, is taking that idea one step further. Because most diseases are complex, “it seems to make sense that a drug that can target multiple elements within a disease should be more effective in stopping disease progression,” says David Hilbert, chief scientific officer at the company. The researchers start with a conventional monoclonal antibody, then fuse short peptide sequences—only 20 to 50 amino acids long—to all three tips of the antibody, creating a supercharged molecule that can bind up to five different targets at once. Hilbert says Zyngenia’s peptide additions add only 4 percent to the mass of the antibody, as opposed to larger structures added by other companies that can grow the molecules by up to 40 percent—limiting the ability of the proteins to move freely in the body. The company has had preclinical success with an antibody that binds three different “Her” receptors, overexpressed in many cancers and targeted by the blockbuster antibody drug Herceptin; the combination antibody is better at inhibiting those receptors than Herceptin alone.

While most companies focus on ways to produce antibodies with the right variable region at the tips of the Y, one company has made a splash by turning to the base of the antibody. Motivated by the overcrowded industry, executives at California-based Xencor saw engineering the antibody’s constant region as the one area of “green field,” recalls company chief Dahiyat.

The Fc region on the Y’s stem is a beacon to the immune system: when the variable region binds a tumor cell or pathogen, for example, the Fc region sticks out and activates or deactivates the immune system against the target. Through sophisticated protein-algorithm software, developed by Dahiyat and Xencor cofounder Stephen Mayo of Caltech, the researchers have literally tested every single amino acid of the Fc region, one after the other, to see how small changes affect the constant region’s ability to bind to various immune-system receptors. “What you can accomplish therapeutically [with an antibody] is so much richer and more powerful if you’ve got this toolkit of Fc domains,” says Dahiyat. Today, numerous drug companies, including Pfizer and Merck, approach Xencor with their tough nuts—antibodies that may bind a target but aren’t potent enough at attracting the immune system. Xencor pops on a new Fc domain to optimize the antibody to carry out its task.

“Antibody engineering as a whole is maturing,” says Janssen’s Strohl. Now, he adds, it’s just a matter of time to see who gets to the big targets fastest. “That’s where the future is: taking these nice, engineered antibodies that we’ve all worked so many years on and applying them to really important targets to treat diseases that we simply cannot treat today.” 

SIDEBAR: EYES ON THE PRIZE

“All large pharma now are into biologics; all are into monoclonal antibodies; all have smart people working on these things,” says William Strohl, Janssen R&D’s vice president of biologics research. “The big difference now is who can go after the right targets at the right time, and who is going to make lots of money off of those.”

Here are some of today’s hottest antibody targets. If you can bind and activate one of these receptors, you’ve got a very bright future.

The target The allure Who's pursuing it
PD-1
Programmed cell death protein 1
Anti-PD-1 antibodies appear to strongly boost the immune system against cancer. Merck, Bristol-Myers Squibb
PCSK9
Proprotein convertase subtilisin/kexin type 9
Anti-PCSK9 antibodies may significantly lower “bad” LDL cholesterol. Amgen, Regeneron, Genentech/Roche, Merck
NGF
Nerve growth factor
Anti-NGF antibodies appear more potent than narcotics for pain, and without the negative effects. Janssen, Pfizer, Regeneron
c-MET
Hepatocyte growth factor receptor
Anti-MET antibodies appear to inhibit tumor growth and prevent metastasis. Genentech
IL-6
Interleukin 6
Anti-IL-6 could treat numerous inflammatory diseases and cancer. Janssen, Regeneron/Sanofi
CD38
Cyclic ADP ribose hydrolase
Found on the surface of many tumors, CD38 is a promising oncology target. Janssen, MorphoSys

 

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?