Scientists eager to understand the function of a gene often get stuck in drudgery: putting the desired alleles into mouse embryonic stem cells can easily take the better part of year. Now, scientists at the Wellcome Trust Sanger Institute have created a welcome resource (Skarnes et al., 2011). In addition to making mouse embryonic stem cell lines for interrogating 9,000 genes, they designed vectors that can help researchers quickly put additional genes under experimental control. The work is part of an international effort to target each of the 20,000 or so mouse genes.

The vectors used in the project incorporate what is called a 'knockout-first' design. The engineered allele is initially inactive, but can be reactivated and inactivated again by a series of recombinases, allowing researchers to choose the developmental stage or the tissue in which to study the allele. “You make the derivative alleles you want by crossing the mice,” explains William Skarnes, project leader for the high-throughput gene knockout program. For example, to inactivate gene function in hippocampal neurons, mice carrying the knockout-first allele would be bred with mice expressing the appropriate DNA recombinase in hippocampal neurons.

What is more, the alleles can also be converted to reporter genes, and vectors enable mix-and-match elements: researchers can easily replace the default lacZ reporter gene with any other reporter. A selection cassette can be swapped with others to permit multiple alleles to be targeted in the same cell. Barry Rosen, a coauthor, conceived the modular strategy, says Skarnes. Not only does it make the vectors flexible in the present, it also “future-proofs the resource,” he says, allowing researchers to incorporate genetic tools that have yet to be invented.

To design the vectors, the researchers began with a computer program that scoured the mouse genome, figuring out where to place selection cassettes and other machinery as well as identifying critical exons whose removal would shift a gene's reading frame. This approach worked for about 60% of protein-coding genes but not for smaller genes containing only a single exon. However, says Skarnes, his team has since developed a new design that overcomes this limitation.

But designing vectors that work for thousands of genes is not the same as making thousands of vectors and putting them into thousands of embryonic stem cell cultures. For that, the researchers created an extremely efficient vector-assembly process so that multiple steps could be conducted in 96-well plates without the need to grow and select clones between steps. They subjected a bacterial artificial chromosome to three rounds of modification, purified the resulting plasmid and passed it through recombinase-expressing bacteria to generate an intermediate targeting vector. The final vector is assembled, conveniently, in vitro. Finally, they electroporate the vectors into a mouse embryonic stem cell line that contributes strongly to the germline and other tissues in chimeric mice (Pettitt et al., 2009)

The technique should work for other mouse strains and even other species, says Skarnes. He and his colleagues are currently using this system to eliminate both copies of genes in embryonic stem cells to study gene function in a model cell.

But the real payoff will come not from engineering cells but from studying mice derived from them. That is why Skarnes is particularly excited that several international government bodies are supporting this task. “Once the cells are converted into mice,” he says, “we, [scientists,] can start the real work, which is to understand gene function.”