December 4-7, 2001
Friiberghs Manor, Örsundsbro and Stockholm University
by George P. Smith
Division of Biological Sciences, Tucker Hall, University of Missouri, Columbia, MO 65211-7400, USA
When I first started work on phage display 18 years ago, I was quite unaware of the revolution in biologically oriented chemistry that was taking place around me: the emerging field I'm going to call "combinatorial biotechnology." The key idea that unites the field is synthesis of large, complex collections ("libraries" in the field's parlance) of chemical compounds through combinatorial assembly from simple building-blocks. To the extent possible, assembly is accomplished by simple, generic, easily-repeated chemical reactions; the researcher's ingenuity is focused, not on sophisticated chemistry, but rather on devising clever schemes for surveying the libraries for those few compounds with the desired target activity—for example some useful pharmaceutical effect. Often this field is called "combinatorial chemistry," but here that term will be reserved for the branch of combinatorial biotechnology in which libraries are constructed non-biogenically through traditional chemical synthesis. I'm going to focus instead on the other branch, which I'll call "combinatorial biology," and in which the members of the library are nucleic acids or polypeptides assembled through ordinary biological replication, transcription or translation. From a synthetic chemist's point of view, these three biological assembly processes are anything but simple; but from a practitioner's point of view they are very easy reactions because living organisms—especially microbes—can be enlisted to do the synthetic work for us. The key feature of these biologically produced chemicals is that they replicate, and therefore can be amplified and cloned at will. Even polypeptides—not themselves replicating molecules, of course—fit this description, as long as they are attached or otherwise associated with the nucleic acids that encode them.
The schematic diagram below emphasizes the close parallelism between combinatorial biology and natural selection in the living world. From this vantage point, a library of nucleic acids or polypeptides is like a population of evolving organisms. The researcher imposes on the population some sort of artificial selection to favor the desired property—for instance, affinity for some target receptor of pharmaceutical interest. From this selection emerges a subpopulation of the initial population that is "fit" according to the artificial selection criterion. Because of the replication property of biological libraries, these fittest survivors can be amplified to large numbers, yielding an amplified subpopulation that can be subject to yet another round of selection. The parallelism with natural biological evolution can be made even more complete if amplification is accompanied by mutagenesis, generating additional diversity for selection to act upon. The entire artificial in vitro evolutionary process takes place en masse with billions or even thousands of trillions of molecules simultaneously. Only at the end, after several rounds of selection, are individual members of the final selected subpopulation cloned and amplified individually for analysis.
Phage display exemplifies combinatory biology. Each member of a phage-display library is a polypeptide fused genetically to a coat protein on the surface of a bacterial virus. The displayed polypeptide is specified by the coding sequence of the recombinant coat protein gene inside the virus particle. Thus when the virus is propagated by infecting fresh bacterial host cells, the displayed polypeptide replicates concomitantly. An individual virus particle or viral clone displays one or more copies of a single polypeptide, but a typical large phage display library would comprise about 10 billion clones altogether, and therefore represent 10 billion different polypeptides. Because the peptides are available on the surface of the virus particles, peptides that bind some target receptor can be specifically isolated by standard affinity purification techniques. And because each affinity-selected peptide remains attached to the phage particle that contains its coding sequence, it is in effect replicable.
The power of phage display to tackle practical goals in new ways is exemplified by Galina Kouzmitcheva's selection of diagnostic peptides for Lyme disease (LD). She started with a dozen libraries of random peptides (encoded by degenerate synthetic oligonucleotides) displayed on phage, and had available serum samples from LD patients and control subjects. She used antibodies from the LD sera to affinity-select binding peptides, and then screened the selected peptides for the desired pattern of reactivity: strong binding to serum antibodies from LD patients, no binding to sera from control subjects. It is noteworthy that although the spirochete pathogen for LD is well known, she did not exploit that knowledge in any way. It is not clear, even in retrospect, how her artificially evolving population of peptides managed to meet the demands she imposed on them. In particular, none of the diagnostic peptides she selected bears any recognizable similarity to any pathogen proteins at the amino acid sequence level. Yet in many respects, her blind, "irrational" quest for LD diagnostics was more successful than "rational" research programs based on detailed knowledge of the spirochete. This dispensing with advance knowledge is of more than academic interest: it means that her program of discovery can be applied without change to any infectious disease—perhaps even to non-infectious diseases—even when no pathogen is known or even suspected.
In a modest way, the LD project also adumbrates the central promise of combinatorial biology: by emulating evolution in the natural biosphere, we can hope to create artificial biospheres that encourage the creation of chemicals able to meet ever more exigent goals, through evolutionary pathways whose details could never be anticipated in advance. It seems inevitable that this process will at least occasionally outperform rational, knowledge-based design.
Biologically constructed libraries can be propagated indefinitely and distributed among an unlimited number of users at very little cost. Furthermore, their use calls for the most part on simple, generic microbiological methods that are easy to master. Combinatorial biology thus represents a profound democratization of chemical engineering, lowering the barrier that academic penury or chemical naiveté might otherwise throw up before talented competitors.