New approaches to antibody engineering were presented at the “European Antibody Congress” held in Geneva, Switzerland last month. At the meeting, companies grappled with long-standing issues of glycosylation, antibody structures, alternative structures, and novel purification approaches.
While the mechanism of tumor-cell destruction by therapeutic antibodies is not completely understood, they appear to function in part by binding to target cells and triggering antibody-dependent cell cytotoxicity or complement-based killing. It has been observed, however, that direct cell killing by cross-linking cellular receptors through antibody bridges is another mechanism that can increase the rate of apoptosis, according to Scott Glaser, Ph.D., director of molecular engineering at Biogen Idec.
Dr. Glaser and his colleagues reasoned that a multivalent antibody would be more efficient at cross-linking receptor targets, and therefore, the team designed a tetravalent antibody that engaged a target antigen commonly expressed on the surface of leukemia cells.
In order to engineer a satisfactory candidate, it was necessary to build an IgG-like antibody of exceptional stability using engineered single-chain variable fragment domains as building blocks. This decision was driven by the observation that bispecific antibodies, engineered with unstable single-chain antibody components, tend to be susceptible to chemical and physical degradation.
Dr. Glaser and members of his team, including Brian Miller, Steve Demarest, and Alexey Lugovskoy combined three complementary design methods—statistical-computational, structure-based, and knowledge-based—to generate a library of single-chain variable domains from the parent single-chain antibody.
The expression library was then screened in a microplate assay to select those single-chain domains with the highest thermal stability while still retaining binding activity. Iterative rounds of mutagenesis and screening produced an optimal candidate. With these high-stability scFv domains defined, they could be engineered into the tetravalent antibodies that display the expected valency and increased avidity.
In a series of experiments conducted in the lab of Ann MacLaren, Ph.D., a researcher-scientist at Biogen Idec, the tetravalent antibody was shown to display single-agent apoptotic activity in a cell-culture system that was enhanced by cross-linking. As predicted, the induction of apoptosis in these cells marks their eventual demise.
Finally, the tetravalent antibody combined with alemtuzumab (Campath, marketed for the treatment of chronic lymphocytic leukemia) induces a powerful apoptotic response in leukemic cells drawn from patients.
To date, antibody therapies have shown the ability to slow the progress of metastatic cancers, but they do not constitute a cure, and they must be combined with conventional chemotherapy regimes. Tetravalent antibodies, with their ability to bring about cross-linking, will be carefully considered as investigators seek to hone the performance of antibody therapeutics and break away from conventional protocols.
“The mechanism of action of naked therapeutic antibodies is profoundly associated with their specific carbohydrate residues at the ASN 297 position in the antibody constant region,” states Qun Zhou, Ph.D., principal scientist in the department of therapeutic protein research at Genzyme. Dr. Zhou concurs with Dr. Glaser that widely employed recombinant antibodies, including Rituxin, Campath, and Herceptin, attach to specific tumor antigens and then attract either complement or natural killer cells by way of their carbohydrate residues, which are then able to destroy the targeted cancer cell.
Indeed, experiments have shown that in antibody molecules that have been shorn of their carbohydrates there is no antibody-dependent cell cytotoxicity (ADCC) or binding to the Fc gamma receptor (FcgR), while in the absence of terminal galactose or N-Acetylglucosamine (GlcNAc) there are minor effects on ADCC and FcgR binding.
Addition of a bisecting GlcNAc to antibody molecules lacking this structure resulted in a 20- to 100-fold increase of ADCC. On the other hand, absence of a core fucose moiety brought about by expressing antibodies in engineered cell lines or transgenic systems drove a 50- to 100-fold increase ADCC activity.
There are a number of approaches for production of antibodies that are appropriately glycosylated in a fashion that will optimize their therapeutic potential, including engineering of cell lines, either mammalian, plant, or yeast. Dr. Zhou and his colleagues reasoned that a more rapid and cost-effective approach would be through the use of metabolic engineering, that is, the use of enzyme inhibitors that block steps in the pathway of glycosylation.
It is possible to use inhibitors for enzymes in the glycosylation pathway such as kifunensine to make antibodies with oligomannose-type glycans lacking fucose. Dr. Zhou and his team have observed that treatment of existing antibody-producing cell lines at low concentration results in antibodies with higher affinity for FcgRIIIA and greater ADCC, without affecting antibody production, antigen binding, and antibody pK. When the Genzyme team examined the properties of the antibodies produced through kifunensine treatment, an inverse correlation between fucosylation and ADCC was demonstrated.
The targeted modification of the antibody carbohydrate structures holds promise for future development; it may allow the design of more effective anticancer treatments and permit the conservation of resources, avoiding treatment of unsuitable patients.