Glycosylation and the Demands of Antibody Engineering

Engineered antibodies are the quintessential success story of the biopharmaceutical industry today. Several approved antibodies have significantly advanced the treatment of cancers and autoimmune diseases and are currently garnering immense profits for their parent companies. It is estimated that approximately 30% of new drugs likely to achieve approval in the next decade will be antibodies or antibody fragments, and the number may run even higher as engineering technologies improve.

Improvements in analytical technology and identification of subtle modifications in glycosylation patterns have created a demand for a higher level of understanding defining the interactive mechanisms of these molecules with their targets.

A leading figure in the study of glycosylation is Roy Jefferis, PhD, of the University of Birmingham, UK. Armed with formidable analytical tools, Jefferis and his colleagues are now able to ask extremely detailed and precise questions concerning how sugars drive the functions of antibodies and how they affect their performance as therapeutics.

In a recent interview, Jefferis elaborated on some of the themes presented in his extensive reviews on the art and science of antibody glycosylation.

KJM: Antibodies and other proteins are perhaps the most expensive drugs on the market. How will technical advances in glycosylation and the legal wrangling over their intellectual property rights affect their pricing?

RJ: Antibodies developed for treatments in cancer depend on: 1) specificity for antigens expressed on the tumor cells and 2) effector functions that destroy the target cell by activating downstream biological reactions. Protein and glycosylation engineering is employed to generate antibodies with enhanced effector functions. The presence or absence of one sugar residue—fucose—can result in a two-orders-of-magnitude difference in the ability to kill cancer cells by antibody-dependent cell cytotoxicity (ADCC). A consequent impact on dose, and hence cost, is anticipated.

KJM: Recombinant antibodies have proven to be immunogenic, at least in a proportion of patients. What properties influence immunogenicity?

RJ: Rituxan and Herceptin are not fully human antibodies—the region conferring specificity is of mouse origin—so it's not surprising that they raise an immune response in certain cases. But even fully human antibodies can raise an immune response. A very important factor in their production is that the preparations be aggregate-free. This puts a lot of pressure on downstream quality control, since a bad batch could immunize a patient for life, and exclude him as a candidate for this therapy. Glycosylation can enhance the solubility of protein molecules and may protect against aggregation.

Quick Recap

KJM: Is IgG1 always the best choice for a therapeutic antibody? What role does glycosylation play in that decision?

RJ: No, it depends on the task that you have selected—i.e., the disease indication. In oncology, we want to kill cells, and an IgG1 antibody bearing non-fucosylated oligosaccharides may be optimal. For other jobs a Fab fragment may be sufficient. For example, an anti-tumor necrosis factor Fab fragment has clinical efficacy; it is pegylated to confer an extended half-life.

KJM: Can one reliably extrapolate performance from in vitro studies and in vivo animal models to humans?

RJ: No, definitely not! FDA insists on animal data before investigators can move ahead with human trials, but simply because the drug appears benign in animal trials, that still doesn't mean that it's entirely safe. Improved animal models are being developed using genetic engineering, e.g., mice and other animals in which endogenous Fc receptors are "knocked out" and human Fc receptors are "knocked in" to the genome. This provides a more realistic model, but in the final analysis, extensive human trials are essential, including Phase 4 monitoring.

KJM: Will changing the site of glycosylation from ASN 297 to another position have a radical effect on antibody performance?

RJ: Yes, but in a negative way. In one series of experiments, the glycosylation site of IgG1 was removed and an IgA glycosylation site was introduced. This resulted in a total loss of biological function (ADCC and complement fixation). This reflects the fact that the oligosaccharide of IgG makes multiple non-covalent interactions with the protein surface and determines the precise three-dimensional structure required for activation of ADCC and other responses.

KJM: Why is Herceptin not effective in all cases of breast cancer? Is it because some patients lack receptors which bind the antibody?

RJ: Yes, in part. Herceptin has specificity for a molecule (Her 2) expressed on the surface of breast cancer cells. However, the level of expression determines the amount of antibody that can bind and consequently the efficacy of kill. In the UK, patients are rated for receptor density on their cancer cells on a 1 to 5 scale. If they fall into the 1–2 range, they simply won't get treated because the probability of response is so small. We hope that the next generation of non-fucosylated Herceptin drugs will extend the range of efficacy. There are a number of other biological considerations related to the biological heterogeneity of the tumor cells.

KJM: Antibodies seem to work well in some situations as cancer therapeutics, but they often fail. Is it because resistant cell populations arise? If the therapy is bound to eventually fail, how can the immense cost be justified?

RJ: It must be appreciated that tumor cells are very heterogeneous, both between and within patients. Some cancers are more aggressive than others; some may have their origin in a precursor cell which is undifferentiated and won't respond to the drug, but will continue to throw off mature tumor cells.

Although the overall benefit of a particular therapy may seem to be marginal to an outside observer, for an individual cancer patient the value of the life extension is immeasurable. The National Health Service in the UK struggles with the question and refers to the National Institute for Clinical Excellence, which provides an independent evaluation of clinical data on experimental drugs and makes recommendations on a cost–benefit basis.

KJM: It has been stated in the press that the devastating side effects in patients treated with the anti-CD28 recombinant monoclonal antibody, which elicited a catastrophic cytokine cascade in a Phase 1 trial, "could not have been predicted." Why could the effects not have been predicted? Can they now be predicted?

RJ: Opinions are divided among the experts. The government appointed Expert Group concluded:

"Although Fc receptors for IgG4, the particular antibody type of TGN1412, are thought to be rare, such cross-linking activity of TGN1412 in vivo cannot be excluded on current evidence."

This statement is in error; IgG4 is fully capable of engaging and activating FcγRI and the non-fucosylated form activates FcγRIII. There is high homology but not identity between human and non-human primate IgGs and FcRs.

KJM: Would it make sense to design aglycosylated antibodies, devoid of effector functions, for use as carriers of low molecular weight toxins, to deliver a poison to a cell?

RJ: Yes, and this is already underway.

KJM: How far are we from engineering yeast and bacterial strains that effectively glycosylate proteins?

RJ: These developments are well advanced. "Knock-out" and "knock-in" technology has been applied to plants, moss, yeast, etc., to suppress non-human glycoforms and generate human glycoforms. It is hoped that these vehicles will allow production of selected homogeneous antibody glycoforms and reduce upstream costs. Downstream costs may not be reduced and new protocols for the removal of host proteins will have to be developed.

KJM: Do you foresee a lot of advances in fundamental and applied glycosylation science on the horizon?

RJ: These developments can be anticipated to have a significant impact on antibody performance and ultimately on the cost of goods. They will also be applied for the improvement of other biologics. Thus, glycosylation has profound effects on the performance of erythropoietin (EPO). Amgen has recognized this and has re-engineered the molecule to introduce two extra glycosylation sites. The product has improved biologic efficacy and a longer half-life.

The next challenge is the development of biosimilars and competition in the marketplace resulting in a lowering of costs. Competition will come from low-cost economies; even if not licensed within the US, they will compete since they will be available from the web.

K. John Morrow, Jr., PhD, is the president of Newport Biotech, Newport, KY, 513.237.3303; kjohnmorrow@insightbb.com