Formulation: ANTIBODIES

© PHOTOGRAPHER: SEBASTIAN KAULITZKI | AGENCY:

Build Better Therapeutic Antibodies

Chemical modifications can improve target binding five-fold

Monoclonal antibodies (mAbs), biotechnology'S greatest medical achievement, are one of the great successes of modern medicine. Of the 500 protein drugs in clinical testing today, no fewer than 300 are mAbs. In 2005, Business Communications Company estimated the worldwide market for mAbs at $15 billion, nearly one-fourth of the world protein therapeutics market. Sales of therapeutic mAbs are projected to grow to $26 billion by 2010, an average growth rate of 11% per year. Interestingly, mAbs enjoy a long product life cycle. According to a report by Arrowhead Publishers, sales of the oldest three mAbs on the market, Rituxan, Remicade, and Synagis, all approved in the late 1990s, are still growing.

Antibodies are very large proteins that contain a constant region and a variable region. The variable region possesses chemical affinity to specific antigens or epitopes. For medicinal mAbs, the epitope target is usually a molecule implicated in disease. In cancer therapy, the mAb binds to cancer cell surface antigens and, through any one of several complex mechanisms, induces cell death. In the case of autoimmune disease-fighting mAbs, the antibody binds to and inactivates disease-mediating inflammatory molecules.

Among the leading blockbuster mAb treatments are the oncology drugs Rituxan (rituximab/non-Hodgkins lymphoma; Genentech, Biogen Idec), Herceptin (trastuzumab/breast cancer; Genentech), Avastin (bevacizumab/colorectal cancer; Genentech), and Erbitux (cetuximab/olorectal cancer; Bristol-Myers Squibb, Imclone). Each of these drugs works by targeting and binding to a disease-specific protein on cancerous cells. But mAbs are not limited to cancer treatment. One of the earliest approved antibody treatments was Orthoclone OKT3 (muromonab-CD3/transplant rejection; Ortho Biotech). At least five other mAb products have been approved in the United States for a wide range of autoimmune diseases, including Humira (adalinumab; Abbott), Remicade (infliximab; Centocor), and Raptiva (efalizumab; Genentech).

MAbs are increasingly viewed as "targeted" or "personalized" therapies because of their specificity. For example, the blockbuster breast cancer drug Herceptin is administered only to women who are high expressers of the HER2neu gene, which is present in most breast tumors to varying degrees. The more copies of the gene a tumor expresses, the more susceptible it is to treatment with Herceptin. Only about 25% of women with breast cancer respond to the drug. Similarly, only 48% of non-Hodgkins lymphoma patients respond to Rituxan, which targets the CD-20 antigen.

Making mAbs Better

Historically, a problem with mAb treatments has been immunogenicity directed at foreign proteins, even those that are beneficial. Early therapeutic antibodies patterned on mouse proteins gave rise to human-anti-mouse antibodies, which limited the number of times a patient could be dosed with the mAb. Subsequently, chimeric-or partially humanized antibodies-which appear more human-like to the immune system, were developed. Today at least two companies, Medarex and Abgenix, claim to possess manufacturing technology that generates fully human antibodies.

There has been significant interest in modifying protein therapeutics to improve pharmacokinetics, safety, and efficacy. In the early 1990s, Altus Biologics introduced cross-linked enzyme crystals, which were dimers of common enzymes hard-wired together by covalent chemical bonds. This chemical linking of two identical proteins was never successfully applied to therapeutic agents. As we will see, however, the idea of mAb molecules aggregating-but only at the active site-can be a powerful strategy for improving the therapeutic properties of antibodies.

Perhaps the most significant modification to therapeutic proteins to date has been PEGylation-the attachment of very large polyethylene glycol residues that greatly improve the circulating half-life of a protein. Among the blockbuster PEGylated protein products are Amgen'S Neulasta (pegfilgrastim), a PEGylated version of granulocyte colony stimulating factor used to boost white blood cells, and Roche'S Pegasys PEGylated alpha interferon for treating hepatitis. In both cases, PEGylation improves the efficacy of the protein and reduces dosing. Most importantly, PEGylation demonstrates that chemical modification can make complex protein drugs more effective.

The most significant modification to mAbs has been the introduction of fully humanized antibodies, which are much less likely to cause adverse immunologic responses than are murine antibodies. Generation of fully humanized antibodies is, however, a ground-up approach that involves serious molecular biology, years of product development, and a high risk of failure. It turns out that the combination of chemical modification and aggregation can significantly improve the activity of disease-fighting antibodies.

Factors Affecting Potency

Safety aside for a moment, the effectiveness of a mAb treatment depends on several factors, including the natural affinity of the antibody for the target and the avidity of that interaction. Although the terms are sometimes used interchangeably, there are subtle differences between affinity and avidity. Affinity relates more to the native binding strength between one binding site and one antigen, whereas avidity takes into account the "valency" of binding.

Perhaps the most significant modification to therapeutic proteins to date has been PEGylation-the attachment of very large polyethylene glycol residues that greatly improve the circulating half-life of a protein.

A mAb with four binding sites will have greater avidity for the target than a mAb with only one binding site, which suggests that multi-valency of binding will cause antibody and target to bind more strongly. For antibody treatments, that means a higher level of efficacy per unit of antibody. Antibody-target interactions of higher avidity could achieve the desired therapeutic effect at much lower antibody dose, which is highly desirable for reducing side effects and lowering the cost per manufactured dose.

Figure 1. Cross linking can improve an antibody'S avidity for its target; this image shows enhanced antibodies beginning to cross link (arrows).

SOURCE: INNEXUS BIOTECHNOLOGY

Avidity enhancement could also change current views and approaches towards personalized, or targeted, therapies. Herceptin has become the poster child for such treatments, which seek to match patients possessing specific geno-typically defined diseases with drugs that target those genotypes. Along with numerous benefits, personalized medicine presents a unique ethical dilemma: What can be done for patients whose genotypes suggest no treatment will work? In the case of small-molecule drugs, the answer is probably to do nothing, because ineffective treatments can cause more harm than good.

The good news is that identifying drugs that are effective for certain geno-types should goad companies into revisiting the vast number of compounds that have failed in clinical trials due to less-than-stellar efficacy or unacceptable toxicity. Resurrecting rejected drugs will not be an exercise for the faint-hearted, because clinical development (initially rejected drugs would need to be tested at least in Phase II and Phase III) is extremely expensive. Yet, for some diseases, it might be more cost effective than beginning from scratch.

An Alternative

Luckily for mAb drugs, an intriguing alternative exists that would most likely entail a far less costly route to expanding a drug'S label. One possible way to improve efficacy of an antibody treatment is to improve target binding, but that involves basic discovery of new antibody molecules, which is fraught with risk. A more attractive possibility involves increasing the number of effective binding sites on the target, which would raise a mAb'S avidity and make it more effective per administered dose.

Such a strategy would achieve several goals. More effective antibodies would require lower dosing to achieve the therapeutic effect for which the drug was initially approved. Conversely, patients could benefit from significantly higher efficacy at the same or a higher dose. In both situations, developers of therapeutic mAbs would provide more highly effective treatments at lower or equal production cost per dose.

Within the context of targeted or personalized medicine, improved avidity would broaden the numbers of patients considered candidates for a drug. For example, cancer cells in approximately 75% of women with breast malignancies do not express enough HER2neu antigen to make treatment with Herceptin worthwhile. Improving the avidity of the Herceptin-target interaction several-fold could generate many more candidates for Herceptin treatment. This would also increase Herceptin'S market share from 25% to perhaps 80% or 90% of all breast cancer patients.

Figure 2. This antibody was modified using cross linking technology (arrows); its initial response is approximately two-fold higher than for the unmodified antibody.

SOURCE: INNEXUS BIOTECHNOLOGY

The implications of avidity-enhancing strategies for future mAb-based treatments are immense. Although approval rates have historically been higher for mAbs than for small-molecule drugs, antibodies do indeed fail in clinical trials.

Fully humanized mAbs have an approval rate of 25%, more than 10 times that of early-stage small-molecule drugs. That is the "glass one-quarter full" view. The other way to look at these approval rates is that, historically, 75% of these drugs fail for either toxicology or efficacy. How many mAbs might be rescued by appropriate modification to higher avidity is anyone'S guess. Given the cost of Phase III failures, many sponsors would likely attempt to salvage antibody medicines by pairing them with an appropriate genotyping test or with a suitable technology for improving avidity, if one existed.

Amid the euphoria over Avastin, which analysts recently predicted would soon enjoy worldwide annual sales of $7 billion, one should remember that the drug failed to achieve clinical endpoints in its breast cancer Phase III trial. Herceptin was almost not approved because of less-than-stellar efficacy during Phase III testing. Were it not for the genotyping test for the HER2neu gene, Herceptin would have been a failed drug instead of a blockbuster with sales of $1.5 billion projected for this year. Similarly, BMS/Imclone'S Erbitux had more than its share of approval troubles. A more effective form of this mAb product might have hit the market up to one year earlier and achieved a larger number of indications (colorectal, head, and neck cancers) than it now enjoys.

Other mAb products were not so lucky. Roche'S R1549, a radiologic-antibody combination drug, failed to show any efficacy in Phase III testing in ovarian cancer. And the multiple sclerosis drug Tysabri (natalizumab; Biogen Idec) was voluntarily withdrawn and then re-approved-with severe "black box" restrictions due to safety issues. Antibodies also fail during clinical trials because of unwanted side effects. For example, the anti-tumor necrosis factor antibodies Enbrel (etanercept; Immunex) and Humira (adalimumab; Abbott) may cause immune suppression. The anti-CD3 mAbs, such as the anti-rejection drug OKT3 (which has also been tested in new-onset type 1 diabetes), attack all T-cells, not only those involved in disease.

Enhancing Avidity

Dynamic cross linking (DXL), under development at InNexus Biotechnology, improves an antibody'S avidity for its target without affecting critical binding or immunogenicity factors. DXL is based on the discovery that certain naturally occurring antibodies bind to one another as dimers after they attach to a target. Researchers noticed that these antibodies contained a peptide sequence of about 24 amino acids consisting of two regions at opposite ends of the sequence and separated by several amino acids. When these antibodies bind to their target, the first sequence on antibody "A" binds to the second sequence on antibody "B" through typical electrostatic and hydrophobic interactions (see Figure 1, p. 35). The sub-sequences may also loop around and bind to one another.

Investigators named this interaction inverse hydropathy and found that they could create antibody dimers by inserting the identical sequences into non-binding regions of other antibodies. In most cases, the added amino acids did not alter the antibody'S affinity for the target, even when more than two antibodies bridged relatively distant antigens.

In effect, an antibody bound to a target cell now provides a second point of attachment to the target, in addition to the antigen. Multiple, cross-linked antibodies thus provide long-lived binding that gives a wider operational window to the cell death mechanism that prevails for a lone antibody. The effect has been observed for many antibody-antigen pairs and several key cell-killing mechanisms, including complement-dependent cytotoxicity, antibody-mediated cytolysis, cellular internalization of the antibody-antigen complex, and apoptosis.

Compared with unmodified mAbs, DXL-modified antibodies maintain a larger therapeutic mass on the target antigen for a longer time period, effectively increasing the half-life of the drug and providing the various mechanisms of cell death with a longer time period in which to act.

Antibody Amplification

DXL amplifies the normal effect of antibodies by clumping the antibodies at the target; circulating therapeutic antibodies are only bound to their target in the presence of the target at any one time. Because they enjoy another point of attachment, DXL-modified antibodies concentrate the therapeutic antibody dose on the target, where it belongs, resulting in a several-fold improvement in binding.

For example, in one apoptosis experiment, 50% of cells were killed in three days using an anti-CD20 mAb modified with the affinity sequence, whereas only 12% of cells treated with unmodified anti-CD20 died. The effect of DXL is, therefore, to improve the avidity of an antibody treatment rather than to improve the innate binding to the antigen. Interestingly, DXL-modified antibodies do not dimerize in solution but only when they are close enough together or when one is immobilized.

Antibody-target interactions of higher avidity could achieve the desired therapeutic effect at much lower antibody dose, which is highly desirable for reducing side effects and lowering the cost per manufactured dose.

InNexus recently announced its first DXL-modified antibody to enter pre-clinical development, DXL625 (CD20). The company is evaluating several additional therapeutic mAbs in their pre-clinical program, including anti-CD19 and anti-CD20 targeted antibodies. These antigens have been implicated in cancers of lymphoid tissue, autoimmunity, and neurodegenerative diseases. Both induced significantly higher tumor killing in mouse xenograft models than comparison antibodies. Additionally, the company is conducting research programs for HER2/neu, caspases, epithelial cellular adhesion molecule, tumor necrosis factor, epidermal growth factor receptor, and others.

The CD20 antigen for DXL625 demonstrates improved binding compared with the native anti-CD20 mAb (see Figure 2, p. 36). The initial response for DXL625 (CD20) is approximately two-fold higher than for the unmodified antibody. The duration of binding, or off rate, exhibited by DXL625 was approximately 10-fold higher than other anti-CD20 mAbs. This increase in binding avidity translates to a concrete functional improvement. Further, DXL625 (CD20) inhibits DHL-4 tumor cell growth by approximately 40% compared with unmodified anti-CD20.

These physical-chemical measurements translate directly to improved cell-killing in vitro through apoptosis, complement-dependent cytotoxicity (CDC), and antibody-dependent cell-mediated cytotoxicity (ADCC). Using standard apoptosis assays on Raii and Ramos cell lines, DXL625 killed at least twice as many cancer cells as native anti-CD20. Results with CDC as-says on Raii cells were even more encouraging. At antibody concentrations ranging from 1 to 20 mcg/ml, DXL625 out-performed anti-CD20 by up to seven-fold. With JOK-1 cells, DXL625 killed between two and three times as many cells as anti-CD20. Improved killing in the 40% to 250% range was also observed in ADCC as-says for Raii and Ramos cells.

Hurdles, Future Development

Manufacturing DXL-modified mAbs is straightforward and can be accomplished either through chemical attachment or through recombinant techniques in which the mAb is expressed with the sequence in place. The latter method will probably be preferred for clinical-grade material because it is straightforward, predictable, and worthy of good manufacturing practices.

The fact that DXL mAbs are entirely new molecules is both a blessing and a curse of sorts. The good news is that as entirely new molecules, these products carry no intellectual property restrictions. Because InNexus owns all the relevant patents, the entire portfolio of current mAb blockbusters is open to DXL modification. The bad news is that because these are new compositions of matter, regulatory authorities and good clinical practice suggest that each must be tested as such, through a complete set of Phase I through III clinical trials.

Toxicity and side effects are another set of unknowns for DXL antibodies. The toxicity of mAbs is a function of immune responses to the mouse-derived component of the protein and cross-reactivity with antigens on normal cells. DXL should reduce antigenicity-related side effects, because dosing per unit of effect will almost always be lower with cross-linked MAbs. Unfortunately, the higher avidity of DXL-mAbs for antigens on non-diseased cells might cause a problem, unless the improved therapeutic effect outweighs the higher toxi-city. Each DXL-modified antibody obviously needs to be assessed independently to determine the relative intensification in efficacy and toxicology.

One possible way to improve efficacy of an antibody treatment is to improve target binding, but that involves basic discovery of new antibody molecules, which is fraught with risk.

In addition to rescuing molecules that fail during clinical trials, this technology could help create therapeutic-grade antibodies from those not considered worthy of clinical development. These might include reagent- or diagnostic-grade molecules that, for one reason or another, were never subjected to preclinical or clinical testing. Because the DXL modification creates a new molecular entity, companies embarking on a discovery program based on cross-linked mAbs would be afforded patent protection for the full 20 years. This business strategy is similar to the chiral switches of the 1990s, in which even third-party developers were able to patent chiral forms of drugs previously patented as race mates.

The potential of this technology for improving diagnostics and reagents is similarly high. By improving the avidity of diagnostic mAbs, DXL, in effect, amplifies the signal, thereby improving the accuracy of a test. We expect an immediate result of this magnification to be a drastic drop in false positives and more accurate detection of low-level antigens. ?