The exceptional value of gene targeting technology to generate mouse models of human disease exists under the shadow of potential genetic errors. We previously observed an unexpected brain-behavior phenotype that resulted from a gene-targeting experiment designed to delete the Zfa gene. Given that the transcription of Zfa is restricted to the germ cell lineage of adult testis, it was both a surprise and a concern when the resulting mice had a phenotype present in both sexes that included abnormal brains and violent behavior. We hypothesized that an unrelated mutation may have been responsible for the unexpected phenotype. Here we show that the single gene mutation, Nr2e1(frc) (fierce), which was responsible for the brain-behavior phenotype, existed in the embryonic stem (ES) cell even before the derivation of the Zfa knockout mice. Our work thus highlights a concern in gene targeting, namely, that ES cells can harbor unexpected mutations, which can lead to genotype-phenotype misattribution. Based on our findings, we caution the gene-targeting community to use low-passage ES cells, to characterize mice derived from more than one independently targeted ES cell clone, and to backcross mice to allow for segregation of distant but linked mutations.
Copyright 2004 Wiley-Liss, Inc.