Science Editor

In the July 20, 2008, online edition of Nature Biotechnology, researchers from South San Francisco biotechnology firm Genentech Inc. described a way to conjugate drugs to specific amino acid residues on antibodies that leave folding and binding of the antibodies unaffected. The method allows them to make antibody-drug conjugates that are controlled more precisely, and have unchanged efficacy but reduced toxicity relative to classically produced conjugates.

Antibody-drug conjugates possibly could be the peanut butter and chocolate of the drug world. The targeting power of antibodies theoretically should make it possible to deliver highly toxic drugs to very specific sites in the body, making it possible to treat patients without getting the proverbial cure that's worse than the disease.

In practice, the best days of antibody-drug conjugates seem to still lie ahead. In the U.S., though companies including Curagen, Seattle Genetics and Genentech itself have such conjugates in clinical trials, only one such agent, Genasense and Wyeth's Mylotarg (gemtuzumab ozogamicin), has been approved by the FDA to date.

One reason is that conjugation itself is currently often imprecise. Classical methods for conjugation link the drugs to either lysines or cysteines, but can not control specifically amino acids in an antibody, or how many in total, the drug attaches to. The result is a mixture of chemical reaction products that can encompass scores of different conjugates, "each with distinct in vivo pharmacokinetic, efficacy and safety profiles," as the researchers pointed out in their paper.

Additionally, when higher numbers of drug molecules are conjugated to one antibody, toxicity appears to rise more quickly than efficacy.

The reasons for the steep increase are unclear; "higher conjugation species may clear more readily through organs that are sensitive to the drugs," senior author William Mallet told BioWorld Today. Alternately, it also is possible that when antibodies have a number of drugs conjugated to them, some of the chemical bonds linking drug molecules to the antibodies are unstable, leading to a greater loss of payload in the bloodstream rather than at the target site.

In their paper, the scientists from Genentech first developed a method to make the conjugation process more predictable. By chemically altering certain cysteines that were located in places where the change would not interfere with antibody folding or antigen binding, the scientists were able to direct the drugs to specific sites on the antibody.

Because the drugs only would bind to the altered cysteines, the Genentech team also was able to control how many drug molecules bound to each antibody, limiting the number to two in the conjugate described in the paper.

In their paper, the researchers named the antibodies Thiomabs and described the Thiomab-drug conjugates they were able to manufacture with their method as "nearly homogeneous."

Mallet, first author Jagath Junutula, and their colleagues demonstrated the utility of their approach by conjugating an antibody to MUC-16, an ovarian cancer antigen, with two molecules per antibody of the drug monomethyl auristatin E, a tubulin inhibitor, which is also part of Seattle Genetics' antibody-drug conjugate cAC10-vcMMAE.

The researchers found that their Thiomab-based conjugates were much more homogeneous than MUC-16-monomethyl auristatin E conjugates produced by standard methods, which, at anywhere from zero to eight drug molecules per antibody, can generate up to 100 different conjugates. When they tested their Thiomab-based conjugates in a xenograft mouse model of ovarian cancer, their efficacy was comparable to standard antibodies, but both rats and monkeys tolerated higher doses of Thiomab conjugates, giving the Thiomab-based approach a broader therapeutic window.

Asked whether the technique was fundamentally a way to make better drugs or better antibodies, Mallet replied, "I think of it as improving drugs." The antibody, to him, is more of a targeting device, although the paper pointed out that the approach also could be used to improve the therapeutic efficacy of antibodies that bind very specifically, but are only weakly effective in doing anything to the cells when they do.

The technique that Mallet and his team described is proprietary; Mallet said the company would not comment on commercial plans either for possibly outlicensing the technology or for the specific conjugate of monomethyl auristatin E to anti-MUC16 used in the paper.

But he did explain that the method is not limited to the specific examples his team described in the paper. "The technology is general," Mallet said. "There's nothing about the technology that's limited to this antibody and this drug."