CREATING MICE WITH SPECIFIC MUTATIONS BY GENE TARGETING 
Mario R. Capecchi, Ph.D., Investigator 
Homologous recombination between DNA se- 
quences residing in the chromosome and newly in- 
troduced, cloned DNA sequences (i.e., gene target- 
ing) allows the transfer of any modification of the 
cloned gene into the genome of a living cell. Fur- 
thermore, if the recipient cell is a pluripotent, 
mouse embryo-derived stem (ES) cell, it is possible 
to transfer that modification (created in a test tube) 
to the germline of a living mouse. Thus the poten- 
tial exists for the defined modification of any mouse 
gene and evaluation of the phenotypic conse- 
quences of that modification. The ability to generate 
specific mouse mutations via gene targeting should 
have a major impact on many phases of mammalian 
biology, including development, cancer, immunol- 
ogy, neurobiology, and human medicine. 
I. Disruption of the hprt Gene. 
Mammalian cells can mediate recombination be- 
tween homologous DNA sequences, but they dem- 
onstrate an even greater propensity for mediating 
nonhomologous recombination. Identification of 
homologous recombination events in a vast pool of 
scattered, nonhomologous recombination events is 
the resulting problem. The hprt (hypoxanthine 
phosphoribosyl transferase) gene is an ideal model 
system for developing the technique of gene target- 
ing in ES cells, because the targeting event can be 
selected directly. Since this gene is on the X chro- 
mosome, only one mutant copy is needed to yield 
the recessive hprt~ phenotype in male ES cells. The 
hprt~ cells are selected by growth in the presence 
of the base analogue 6-thioguanine (6TG), which 
kills hprt^ cells. 
Two classes of targeting vectors were tested for 
their ability to disrupt the hprt gene: sequence re- 
placement and sequence insertion vectors. With 
yeast as a paradigm it was anticipated that se- 
quence replacement vectors would replace endoge- 
nous DNA with exogenous sequences, whereas se- 
quence insertion vectors would insert the entire 
vector DNA sequence into the endogenous locus. 
Each class of vectors contains a neomycin resistance 
(neo") gene within an exon of hprt. This arrange- 
ment not only disrupts the coding sequence of hprt 
but also provides a selectable marker for cells con- 
taining an integrated copy of the recombinant vec- 
tor (resistance to the drug G418). 
After the introduction of the targeting vectors 
into ES cells by electroporation and selection for 
resistance to G418 and 6TG, all survivors were 
found to have lost hprt activity as a result of tar- 
geted disruption of the hprt gene. Replacement 
vectors and insertion vectors were equally efficient 
at disrupting the endogenous hprt gene. Further- 
more, both vectors showed the same strong depen- 
dency of the targeting frequency on the extent of 
homology between the targeting vector and the en- 
dogenous DNA sequence. Over the range tested 
(2.9-14.3 kb) a 5-fold increase in DNA sequence 
homology resulted in approximately a 100-fold in- 
crease in the targeting frequency. 
II. Nonselectable Genes. 
The hprt gene was chosen as a model in the ini- 
tial studies because direct selection could be used 
to isolate cells in which a homologous recombina- 
tion event had occurred. However, in the vast ma- 
jority of cases a selectable cellular phenotype is not 
associated with the inactivation of both copies of a 
gene or with the more frequent single-copy inacti- 
vation event. Therefore it is desirable to have some 
means for identifying the rare ES cell in which a 
nonselectable gene has been inactivated. This can 
be achieved by using indirect enrichment and/or 
screening procedures. Pursuit of enrichment proce- 
dures rather than screening procedures was chosen 
because, if successful, they should be less labor in- 
tensive and permit identification of rarer events. 
Recently an enrichment procedure was described 
that is independent of the function of the target 
gene and of its expression in ES cells. This positive- 
negative selection (PNS) procedure uses a positive 
selection for cells that have incorporated the target- 
ing vector anywhere in the ES genome and a nega- 
tive selection against cells that have randomly inte- 
grated the vector. The net effect is to enrich for 
cells containing the desired targeted mutation. 
PNS was used to enrich for ES cells containing 
disruptions of the hprt gene. After introduction of 
the ^/?r?-PNS-targeting vector into ES cells, virtually 
all (19/24) of the selected colonies contained targeted 
disruptions of the hprt gene, even though direct se- 
lection for the hprt~ phenotype was not done. 
III. The Mouse hox Genes. 
Recently molecular genetic analysis of early de- 
velopment in Drosophila has revealed a network of 
genes that control the formation of its metameric 
pattern. Many of these genes share a DNA se- 
Continued 
167 
