DNA shuffling in the genetics game

Dr Willem (Pim) Stemmer has spent the past decade demonstrating, in the most graphic manner possible, the value of biodiversity, and the existence of a massive genetic treasure trove in nature, far greater than the sum of its individual components.

Stemmer, vice-president and co-founder of the US biotechnology company Maxygen, is the inventor of DNA shuffling. He described recent advances in the field to today's session of the 28th annual Lorne Conference on Protein Structure and Function in Victoria.

The powerful new techniques he and his colleagues have pioneered take the best genes nature can offer, and improve on their protein products, by amplifying the already rich genetic diversity found in living organisms.

DNA shuffling doesn't merely speed up protein evolution, it opens up a new universe of possibilities. It 'breeds' novel proteins that mutation and natural selection might never have conceived, in the genetic sense, in billions of years.

DNA shuffling promises to short-circuit the development of new biopharmaceuticals, to enhance the speed and efficiency of slow, low-yield industrial biochemical processes, and to speed the breeding of new crops and livestock species that will eclipse the productivity of any in the field today.

Since the middle of last century, a host of microbes and at least 2000 modern plant varieties have been developed by chemical or radiation mutagenesis.

They include CSIRO's flax-derived oilseed crop Linola, a red variety of the pink hybrid tea rose, 'Queen Elizabeth', and many microbes used in industrial processes to convert raw materials into industrial materials like plastics.

Radiation and chemical mutagenesis has been likened to performing major surgery with a shotgun. The technique produces a host of random changes in genes, nearly always with broadly deleterious effects on the organism -- including, in animals, cancer.

While Stemmer's DNA shuffling techniques can exploit induced mutations, as a means of amplifying genetic adversity, his team normally works with naturally occurring genes.

A typical example might involve a gene that varies between four closely-related species of bacteria, or four different strains of the one species.

Conceptually, the four genes are laid out in parallel, so their homologous regions are juxtaposed. Corresponding segments of the gene are then randomly mixed-and-matched across the four strains. The result is a panel of thousands of bacteria, in which each cell contains a novel 'hybrid' gene consisting of some unique combination of segments from the four species or strains.

Cells carrying the best-performing variants of the hybrid gene are then selected on the basis of their superior performance in a desired biosynthesis reaction, such as protein production, or transformation of a raw feedstock into an industrially useful material.

The elite cells can then be subjected to a second or third round of DNA shuffling, to produce even better-performing variants.

"The goal of our research is to create a process with a short cycle time that can be applied to single genes, whole biochemical pathways, even to whole genomes, by exploiting one to multiple selected parental types, and applying focused selection pressure," he said.

"We've now completed around 100 gene projects. We've obtained higher expression levels, increased activity, new activity, new selectivity for particular enantiomers of molecules, improved pH tolerance, improved interaction with co-factors, and reduced toxicity or immunogenicity.

"The success rate is very high, if you are careful in selecting your starting organisms or genes. Close to 100 per cent of the shuffled hybrids show significant improvements in activity, ranging between threefold to 30,000-fold."

For example, said Stemmer, two decades of efforts to improve the efficiency of the rubisco enzyme, which performs the initial carbon-capture and synthesis reactions in plants and cyanobacteria, had resulted in less than a 1.5 per cent improvement in the carbon-capture reaction.

By shuffling rubisco genes from several strains of cyanobacteria, sharing about 80 per cent homology at the peptide level, his team has doubled carbon dioxide fixation in the cyanobacterium Synechocystis.

Taking several strains of a bacterium isolated from soil contaminated by the herbicide atrazine, his team performed DNA-shuffling and then tested the recombinant strains for their ability to break down 15 different triazine-based compounds.

From an initial two strains that were able to dechlorinate atrazine, or deaminate aminoatrazine, the project had yielded strains that could 'crack' eight other triazine compounds; only five remained recalcitrant.

The new strains were up to 200 times more efficient than the original wild-type strains in breaking down atrazine and aminoatrazine, and up to 10,000 times more efficient in breaking down other triazine variants.

DNA-shuffling had been used to reduce the immunogenicity of five native human proteins -- proteins that tend to aggregate in human tissues because they fold inefficiently in the absence of accessory proteins.

Stemmer said large protein aggregates tend to be immunogenic. The 'reshuffled' variants fold more efficiently, and don't aggregate. He believed the technique could be useful for producing non-immunogenic variants of pharmaceutical proteins, such as vaccines.

Shuffling genes from eight different mammals, for the cytokine interleukin-12, a key player in the immune response, produced hybrid IL-12 molecules with greatly increased activity.

By shuffling genes from the four major strains of the dengue fever virus, Stemmer's team produced a quadrivalent antigen that, tested on mice, produced cross-protection against all four strains.

Dengue fever can cause severe illness, even death. Infection by one strain does not provide cross-protection against other strains, and a second encounter with a different strain is notorious for inducing a much more virulent infection.

Stemmer told the conference that two new microbe strains developed by DNA shuffling are already entering industrial production.

Cargill-Dow is building a huge factory in the American mid-West, which will use a new strain of Lactobacillus to produce polylactic acid, an industrial polymer, from lactic acid. The new strain is three times more efficient than previous strains.

The pharmaceutical giant Eli Lilly is using a shuffled strain of Streptomyces fradiae to produce higher yields of the antibiotic tylosin.

"Thousands of people are now involved in sequencing genomes of bacteria and higher organisms. Our research shows that you can breed greatly improved strains of industrial microbes like plants, without even knowing which genes are involved," Stemmer said.