Considerations for Choosing an Antibody Platform

Antibody-based drugs now rival or surpass the small-molecule drugs in terms of effectivity, specificity, minimal side effects, number of ongoing clinical trials, and sales revenues. But which platform will find the best antibody in the shortest time—best in regard to specificity, affinity, and scalability?

It is generally agreed that the antibody must be human, and nowadays there are many ways to select target-specific human antibodies. The antibodies can come from humans, rabbits, rats, chickens, llamas, yeast, mammalian viruses, and bacteriophages.

In infectious disease, one may isolate the (memory) B lymphocytes from patients who have recovered from disease. Antibody (gene) libraries from these individuals are screened for the antibodies that best neutralize the etiologic agent. This is an effective way, albeit dependent on an anecdotal approach rather than on a broad experimental approach. And it does not lend itself easily to applications in cancer and autoimmune disease.

For a broad experimental approach, large antibody libraries from many patients can be constructed and represented in yeast and microbes. Constructed as a general tool for antibody discovery to any given target, such libraries by necessity are “naïve”; i.e., they represent the primary immune repertoire. Because this is done in vitro, the subsequent scalability (stability, solubility) is not guaranteed. But probably more important, such a library mostly contains antibodies of low affinity and modest specificity. Although rare antibodies of higher affinity can be isolated, one usually has to improve the affinity by a reiterative mutation and selection process. Most immunologists would agree that this in vitro process has not yet succeeded in replicating the in vivo hypermutation and selection that happen in the germinal center of higher vertebrates.

Although invented forty years ago, the hybridoma technique is still unsurpassed in its effectiveness and simplicity. It works best with mice and rats, but also with rabbits. So why not use regular mice, for example, to then “humanize” the isolated antibodies by grafting the specificity-determining regions onto human frameworks? Computer programs and cloning procedures for doing so are simple, and the patents for this process have expired. Indeed, this route is perfectly reasonable. It does, however, come with a cost in time of more than half a year, plus the uncertainty of not knowing whether the final product will be scalable and retain the same affinity. For each potential antibody, several humanized derivatives have to be tested.

Mice that have replaced their immunoglobulin loci with the human counterparts provide the best resource for discovery of human antibodies to any antigen, affinity maturation of the antibodies, in vivo testing of correct protein folding, and fast and easy identification and isolation of the most promising candidates. However, owing to their genesis by 1990s technology, the “pioneer human antibody mice” have practical issues. Because of their limited antibody repertoire, the discovery of several important antibodies was possible only with a massive-scale effort.

Today, there are six other transgenic rodents with human antibody genes: one rat and five mice. Only three of the mice contain the full human repertoire at the endogenous mouse loci. One of these is the Trianni Mouse. In contrast to the other two mice, which contain large fragments from human genomic DNA, the Trianni immunoglobulin loci were constructed in silico and chemically synthesized. In the Trianni Mouse, the mouse immunoglobulin exons of the variable (V) gene segments are replaced by the human exons, but the control elements remain of murine origin. That is, the V gene segments are neither human nor mouse, but chimeric. However, the V regions of the antibodies that the mouse synthesizes are entirely human.