After a tumultuous early history, during the course of which they were alternately hailed as magic bullets and derided as expensive failures, antibodies have become pretty much a standard part of the therapeutic repertoire.

Since the approval of the first monoclonal antibody in 1986 (Orthoclone OKT3, an anti-transplant rejection drug from Ortho Biotech Products LP), a number of antibodies have been approved, mainly for the treatment of cancer and autoimmune diseases.

Unfortunately, the most recent headlines made by an antibody were less than stellar. In February, Biogen Idec Inc. and Elan Corp. plc announced they were suspending marketing of Tysabri only three months after receiving FDA approval, due to serious side effects. (See BioWorld Today, March 1, 2005.)

But the antibody market as a whole is large enough that such individual problems no longer reverberate through the therapeutic class. Compared to small-molecule drugs, antibodies have their own strengths and weaknesses, but, fundamentally, a successful antibody needs the same characteristics as a successful small-molecule drug: strength and specificity. And as with small molecules, the best way to find such a compound is to start with the greatest possible diversity.

In research appearing in the March 2005 issue of Nature Biotechnology, scientists from Cambridge, Mass.-based Dyax Corp. described a method for increasing such diversity. The paper is titled "Generation of high-affinity human antibodies by combining donor-derived and synthetic complementarity-determining region diversity."

When Mother Nature does it, the generation of high-affinity antibodies is a two-step process. During development, both B cells and T cells generate diversity through a gene segment shuffling process known as gene rearrangement. After an infection, memory B cells go through a process called somatic hypermutation to generate additional diversity.

Dyax, which has an early corporate history characterized by broad development of the uses of phage display technology, began focusing on antibodies about five years ago.

"One of the remarkable observations from the early work was that often, the antibodies [identified by phage display] don't have high affinity," Clive Wood, chief scientific officer at Dyax, told BioWorld Today. So after antibodies are identified, they need to go through affinity maturation. There are several methods for affinity maturation, but all are lengthy and necessitate in vivo steps; the goal of the research described in the current publication was to create an in vitro method of mimicking the affinity maturation process.

The scientists started by tapping two types of antibody donors: normal individuals and individuals with autoimmune disorders, who will make antibodies to proteins that normal donors ignore. When asked whether that could lead to the production of antibodies that were diverse but did more harm than good, Wood pointed out that there still is a lengthy selection process to weed out undesirable antibodies, including those with autoimmune effects.

In antibody structure, there's variable, and then there's variable. The variable region of an antibody has three hypervariable regions, also known as complementarity-determining regions (CDRs), separated by four framework regions. The researchers used the entire light chain, as well as the hypervariable or complementarity-determining region 3 of the heavy chain, from donors to create the library.

"In the heavy chain, the CDR3 diversity is enormous," Wood said. Not just its amino acid composition, but its overall size varies nearly by a factor of 10, varying from four to 35 amino acids, though the median size is about 13. Heavy-chain CDRs 1 and 2 were made synthetically.

The scientists then further increased the diversity of those antibodies by inserting point mutations at selected amino acid positions in CDRs 1 and 2. The number of mutations introduced at each position varied, and Wood explained that "we introduced the mutations you'd expect to see during normal maturation from primary to secondary response." In some cases, all amino acids except for cysteine were substituted for the original. (Cysteine was not inserted to prevent the formation of disulfide bonds, which in turn can cause misfolding.)

The researchers tested their antibodies against four therapeutically relevant targets. In each case, some antibodies had dissociation constants in the nanomolar range, which are indicative of very high affinity. The authors noted in their paper that they identified antigen-binding fragments "against multiple therapeutic targets that have higher affinities than approved therapeutic antibodies."

Abbott Laboratories' Humira (adalimumab) is the first example of an approved therapeutic antibody derived from phage display. Wood named higher throughput as the main advantage of using phage display rather than letting affinity maturation run its course in mice.

"We can screen thousands of hits in parallel and very quickly," he said. "Immunizing a mouse is a longer-term strategy, and has a less predictable outcome."