Tackling Production Issues for Bispecifics

Complex Purification Challenges Prove Not Always Amenable to Platform Technologies

First introduced during the 1980s, bispecific antibodies fell from favor due to issues with efficacy and manufacturability. But 30 years is a long time in biotechnology. Bispecifics are now making a comeback.

Nearly every large biopharma and many mid- to small-stage companies have bispecific or multispecific antibody programs. A presentation by MicroMet’s Patrick Baeuerle, shortly before his company’s acquisition by Amgen, described bispecifics as en vogue.

Bispecifics are in the same position, in terms of business potential, as conventional antibodies were 25 years ago. But due to strides in expression, manufacture, and purification, they are far more advanced scientifically than were the monoclonals of the early 1990s.

Technologies for discovering and manufacturing bispecifics have proliferated. GEN readers have seen such terms as Triomab, Tribody, BiTE, DART, DuoBody, TandAb, DVD-Ig, Nanobody, CrossMab, Dock and Lock, and approximately 30 others—all involving techniques for combining two or more affinity functions into a single molecule.

Shivani Singh, who authored “Bispecific Antibody Therapeutics Market, 2013–2023” (Roots Analysis consultancy), calls bispecifics “one of the fastest growing markets in the field of antibody therapeutics.” Singh’s prediction for 2023 sales of bispecifics, $4.4 billion, perhaps underestimates the impact of this therapeutic class.

Although just one bispecific, Removab (catumaxomab), has been approved (in the EU), at least 100 are in various stages of clinical development. Removab, which binds to two well-characterized cancer antigens, CD3 and EpCAM, is illustrative of efficacy strategies based on affinity. Neovii Biotech now markets the product after Fresenius sold its biotech business last year.

Other biospecifics undergoing R&D include those that involve the neutralization of two ligands or inhibition of two receptors by one molecule, the crosslinking of different receptors on single cells, and the recruitment of immune system cells to the vicinity of tumors.

One-Step Expression

From manufacturing and regulatory perspectives, chemical or enzymatic linking of two or more antibodies or antibody fragments post-expression is risky. Writing in the Journal of Antibody Conjugates (June 6, 2014), Ruta Waghmare, Ph.D., of EMD Millipore notes that chemical linking results in purification issues that are not amenable to platform technologies: “The purification of these molecules can be complex, and a platform approach is not always feasible for the downstream processing.”

In addition, Dr. Waghmare summarizes the difficulties posed by these molecules. They may be unstable, diverse in their make-up, and liable to aggregate. Also, they may bind to Protein A and be of varied sizes, with varying impurity profiles.

The excitement in multispecifics in no small part results from molecules that are produced and purified similarly to conventional antibodies. Xencor’s approach, for example, relies on the company’s Fc engineering technology. “We express two different Fc chains together in the same cell, and they assemble preferentially with each other,” explains Bassil Dahiyat, Ph.D., the company’s president. “We can then attach different variable domains, or anything else we like, on top of the individual Fc chains.”

Stability and manufacturability issues have plagued previous attempts at using a heterodimeric Fc region to direct the composition of the variable region. Xencor fell back on its huge library of Fc engineering expertise, and the thousands of Fc variants they’ve studied, and repurposed that knowledge into building a rock-stable heterodimeric Fc.

Xencor adds an extra twist aimed at streamlining characterization and downstream processing. Adding one or two amino acids to the Fc chains provides a means of separating homodimeric impurities from the desired bispecific molecule, based on the molecules isoelectric point (pI), through ion exchange chromatography.

Like most leading bispecific companies, Merrimack Pharmaceuticals produces its molecules intact, in this case in CHO cells, without post-expression chemical or enzymatic manipulation. The company has bispecifics in Phases II and III, with a third entering the clinical pipeline in 2015.

Merrimack uses a systems biology platform to identify biological targets but designs each molecule individually. “Our bispecifics are unique in that they target two different receptors on the surface of a tumor cell,” asserts Ulrik Nielsen, Ph.D., the company’s CSO. In some instances, one of the receptors provides an anchor, while the second—which may also exist on noncancerous cells—is the actual target.

“Our transfection and manufacturing approaches are completely analogous to those of our monoclonal antibody programs,” notes Dr. Nielsen. For example, the bispecific gene is introduced as a single plasmid, and downstream processing involves capture and ion exchange chromatography. Even two- to three-week dosing, according to Dr. Nielsen, is “very antibody-like.”

Drug Platforms
Evolution of multispecific platforms: first-generation monospecifics (constant regions in green); second-generation bispecifics (additional antigen-binding site and maintenance of heavy-light chain pairing); third-generation constructs (more flexible formatting into a single genetic construct with multiple valencies and/or target specificities). Representative third-generation constructs include Ablynx’ Nanobodies®. If desired, the half-lives of these constructs may be extended via traditional means, such as incorporating a serum albumin-binding VHH domain (yellow oval).

Smaller Multispecifics

Nanobodies, a specialty of Ablynx, are expressed in CHO cells, bacteria, and yeast as appropriate. The nanobody concept arose from the observation that camels, in addition to producing conventional antibodies consisting of both light and heavy chains, also make heavy-chain-only molecules. It was discovered that these heavy chains, or VHH domains, are by themselves extremely stable, are capable of daisy-chaining genetically, and possess full binding capability, but weigh in at about 1/10th the size of conventional antibodies.

These benefits translate directly to manufacturing. Since they are expressed intact as bispecific, tri-specific, etc.—Ablynx has successfully strung together up to five, each with unique affinity. No chemical post-expression processing is needed.

Along with suitability to most expression systems, nanobodies pose no particular purification problem and in most cases do not require glycosylation for their activity.

Ablynx has been quite successful in getting its bispecifics to clinic. Of the 20 currently in human testing worldwide, Ablynx owns or shares 5. “Nanobodies are a straightforward, modular path to multispecifics,” says Tony De Fougerolles, Ph.D., chief scientist at Ablynx.

Volumetric productivity for nanobodies stands at between 2 and 15 g/L, about the average for platform monoclonal antibody processes. But since they are only 1/10th the size of monoclonal antibodies, nanobodies provide 10 times the therapeutic punch on a molar basis.

Thus, all things being equal, a 2,000 L antibody process could reduce to 200 L. Improved efficacy through bispecific binding could reduce dosing and manufacturing volumes even further.

All bispecific antibodies attempt to “hit” two antigens with one molecule. The valency of nanobodies is such that an anticancer nanobody may be designed to target three, four, or even five cancer pathways, and operate only on those expressed in the patient—a type of personalized medicine by force. Ablynx has even used one slot on a multitarget nanobody to bind albumin, to improve circulating half-life.