July 15, 2004
Stanford Team Develops New Way To Generate Potential Drug Compounds

A team in the Pehr Harbury lab at Stanford has developed a method that may allow the automated generation of a large number of organic compounds as drug candidates through molecular breeding.

Traditionally, developing small molecules for research or drug treatments has been a painstaking enterprise. Drugs work largely by binding to a target protein and modifying or inhibiting its activity, but discovering the rare compound that hits a particular protein is like, well, finding a needle in a haystack. With a specific protein target identified, scientists typically either gather compounds from nature or synthesize artificial compounds, then test them to see whether they act on the target.

The birth of combinatorial chemistry in the early nineties promised to revolutionize this laborious process by offering a way to synthesize trillions of compounds at a time. These test tube techniques have been refined to "evolve" collections of as many as a quadrillion different proteins or nucleic acids to bind a molecular target. These techniques are called molecular breeding, because like traditional livestock and crop breeding techniques, they combine sets of genotypes over generations to produce a desired phenotype. Molecular breeding has been restricted to selecting protein or nucleic acid molecules, which have not always been the best lead compounds for drugs. Conventional synthetic organic chemistry, which has traditionally been a better source of candidate drugs, has not been amenable to this type of high throughput molecular breeding.

But this bottleneck has potentially been overcome and is described in a series of three articles by David Halpin et al. in this issue of PLoS Biology. By inventing a genetic code that acts as a blueprint for synthetic molecules, the authors show how chemical collections of nonbiological origin can be evolved. In the first article, Halpin et al. present a method for overcoming the technical challenge of using DNA to direct the chemical assembly of molecules. In the second, they demonstrate how the method works and test its efficacy by creating a synthetic library of peptides (protein fragments) and then showing that they can find the "peptide in a haystack" by identifying a molecule known to bind a particular antibody. The third paper shows how the method can support a variety of chemistry applications that could potentially synthesize all sorts of nonbiological "species." Such compounds, the authors point out, can be used for drug discovery or as molecular tools that offer researchers novel ways to disrupt cellular processes and open new windows into cell biology. While medicine has long had to cope with the evolution of drug-resistant pathogens, it may now be possible to fight fire with fire.

The first, second, and third articles are available on-line for reading. All PLoS Biology are available to be read without any cost to the reader.

Peptides (which are just are just a sequence of amino acids and serve as components of larger protein molecules) and DNA are hard to get into the body because they tend to get broken down before absorption. Even if they are injected into the bloodstream they stand a pretty good chance of being broken down before they reach a desired target. Whereas a lot of synthetic compounds can be absorbed and reach their targets more easily without getting broken down by enzymes. So the most interesting aspect of these papers is the claim (at least as far as I understand it) that they can use this technique to generate chemical compounds that are not DNA or peptides.

The punch line is in the third article.

Beyond the direct implications for synthesis of peptide–DNA conjugates, the methods described offer a general strategy for organic synthesis on unprotected DNA. Their employment can facilitate the generation of chemically diverse DNA-encoded molecular populations amenable to in vitro evolution and genetic manipulation.

The need that they are trying to solve is the generation of a large number of different compounds to test more rapidly as potential antibiotics against bacteria. The sort of Holy Grail would be a method to do high volume automated means of generating compounds, testing against pathogens, and then feeding that back into the generator mechanism to make more variations most like those variations that had the strongest effects against the pathogens. The hope is that when a new drug-resistant pathogen pops up then by sheer brute force so many compounds could be tried against it so rapidly that in a relatively short period of time antibiotics effective against the new pathogen strain would be identified.

Share |      Randall Parker, 2004 July 15 04:46 PM  Biotech Advance Rates

Raj said at July 21, 2004 8:22 AM:

I wonder if this "Molecular breeding" is a "New Way" discovered by David Halpin et al.

A process known as "DNA breeding" was invented by Willem "Pim" Stemmer in 1997. " .. almost any gene product can be used as a potential therapeutic. Unfortunately, many native cellular proteins are ill-adapted for commercial use—they're unstable at a different pH, or they need greater specificity, or their activity is diminished—creating a roadblock to their actual use as drugs. To get around this roadblock, biochemists have often used a form of simulated evolution to modify existing compounds into new and more useful forms. This has traditionally been done with a "brute force" approach, in which genes and gene products are randomly altered and tested for improved efficacy or specificity. Unfortunately, such strategies are extremely labour- and time-intensive. In "Survival of the Fittest Molecule," the authors discuss a more efficient approach: A form of directed evolution called "DNA breeding," which shuffles bits of DNA that have already been sculpted by countless generations of natural selection. This strategy has dramatically decreased the time and effort required to develop new drug candidates, and it has already produced a range of new molecules, including immune system enhancers and a dengue fever vaccine."

In what way I wonder is this "molecular breeding" similar to "DNA breeding"

Seth said at August 16, 2004 1:50 PM:

Harbury's methods derive from a combination of nucleic acid chemistry (SELEX) and split--pool organic synthesis. The idea is ultimately to make small molecule drugs. The protein affinity-maturation methods are similar but involve biological protein synthesis with ribosomes. One can't make small molecules that way, and small molecules are the orally bioavailable, high volume (profitable) drugs we generally want. The real question (IMHO) is whether or not affinity maturation or in-vitro evolution provides an important advantage for small molecule drug discovery. It's an open question for the time being. Some might argue that synthetic organic chemistry is not sufficiently well-developed to suffficiently populate chemical space so that evolutionary principles can be used to advantage. I tend to the opinion that in the long run chemistry is bound to improve, and that this and similar technologies will become increasingly attractive to the mainstream.

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright ©