One of the most common ways to investigate the role of a gene in human physiology is to delete its equivalent from a mouse genome and to observe the effect. The use of one enzyme in particular, the recombinase 'Cre', has revolutionized the study of gene function in mice. The technique allows researchers to introduce mutations and gene deletions in a tissue or cell type at any stage.

Hundreds of studies using this technology have been published since it was introduced more than ten years ago, shedding light on areas such as important developmental processes and the role of numerous genes in, for example, the immune or nervous systems, or in various diseases.

Briefly, it works by introducing the target DNA sequence used by the Cre enzyme, known as a loxP site, to either end of the gene sequence in question. By subsequently introducing the Cre enzyme, the sequence is excised. Gene targeting can be regulated by controlling where Cre is expressed or activated.

But the technology is not without its pitfalls. A number of issues have been described in a recent overview (M. Schmidt-Supprian and K. Rajewsky Nature Immunol. 8, 665–668; 2007). Readers, authors and editors alike need to be alert to one particular problem: the potential toxicity of Cre expression to cells.

The induction of cell death as a consequence of Cre activity, unrelated to the targeting of any specific gene, is thought to occur when Cre targets sites similar to loxP that are present in genomic DNA, thereby inducing mis-recombination and DNA damage. Most mice strains in which Cre is expressed seem to develop normally and do not show any overt signs of Cre toxicity, and it is somewhat unclear exactly under what conditions it arises. It has been suggested to result from long-term expression of high levels of the enzyme.

Regardless of the exact mechanism and circumstances, Cre toxicity is clearly a potential problem, yet in the view of some researchers it has been neglected or played down in the community. In fact, one study has systematically analysed studies using a particular Cre mouse strain and found that in more than half of the cases the appropriate control for potential Cre toxicity — the use of the same mice without the loxP-flanked target gene — was not included (J.-Y. Lee et al. J. Biol. Chem. 281, 2649–2653; 2006). Nature is aware that it has in the past published papers in which such controls were lacking, although many will no doubt have been independently validated with other techniques at the time or subsequently.

It can be argued that potential toxicity due to Cre expression becomes pertinent only when the observed phenotype resulting from gene targeting involves cell death, but the complexity of biological processes probably warrants attention to the issue in all experiments. Researchers planning experiments should take into account the need for additional mice as controls. Editors at Nature will consider the issue and the appropriate controls with referees during the assessment of submitted papers.

No technology is without caveats, and — as the Nature Immunology article concludes — there will always be a degree of uncertainty with which researchers have to live. But in the interest of best scientific practice, everyone involved would be wise not to neglect the dangers and subtleties at play even in routine experiments.