The knockout mouse: six years old and growing stronger.

Abstract

Site-specific alteration of the mouse genome has become a routine procedure. The past six years have witnessed the generation and characterization of several hundred novel animals, each harboring an independent mutation in one of the 50 to 100,000 estimated murine genes (1). Genes previ­ ously predicted to be involved in normal development or in the maintenance of adult homeostasis have been targeted for mutation, as have murine homologues of genes implicated in human disease, including some affecting the respiratory system. The phenotypes resulting from these mutations have ranged from the predictable to the surprising-witness the unexpected respiratory defects in animals lacking a func­ tional endothelin-l gene (2), or the relatively benign pulmo­ nary pathology associated with a number of mutant alleles of the CFTR gene (reviewed in 3). The phenotypic charac­ terization of these animals has helped to define biological functions of individual genes and to identify networks of genetic interactions. Rather than attempt to summarize a vast accumulation of data thus far generated, this Perspective will address some of the limitations inherent in this technology, emphasizing how an understanding of these limitations is contributing to a more thorough genetic description of a mammalian organism and its development. The protocol for mutagenesis is by now straightforward: cloned DNA, first mutated in vitro, is introduced into a popu­ lation of pluripotent embryonic stem (ES) cells grown in cul­ ture; those cells in which recombination has replaced the tar­ get, endogenous gene with its mutant homologue are isolated and transplanted into a blastocyst stage mouse embryo; this embryo is implanted and allowed to develop to term, and the resulting chimera (containing cells from the blastocyst as well as ES cells) is then bred to propagate the mutant se­ quences (4, 5). If the mutant allele is a recessive lethal, it can be maintained in the heterozygous state, and a simple genetic cross is used to generate homozygosity. The majority of the currently available mutant animals carry loss-of-func­ tion, or null, alleles of the gene of interest. Not only are such mutations technically the simplest to construct, but the phenotype of the ensuing knockout mouse establishes a baseline requirement for the target gene. Initial expectations of this approach are that the time and place of gene action will be defined, that development may