mice might be due to compensation by other genes, 
as suggested by several lines of biochemical evi- 
dence. To test the hypothesis that other kinases in 
the same family as src may be playing overlapping 
roles, mice carrying mutations in the genes encod- 
ing the fyn and yes kinases have been derived by 
gene targeting in ES cells. Both mutant mice fail to 
exhibit an overt phenotype, again probably because 
of functional compensation. However, further stud- 
ies of the fyn mutants reveal signaling defects with 
respect to the T cell receptor in thymocytes, and to a 
lesser extent in mature T cells. In addition, prelimi- 
nary evidence suggests that the fyn~ mice may be 
impaired in their ability to undergo negative selec- 
tion in the thymus. These studies provide definitive 
evidence for an essential role of fyn in T cell recep- 
tor signal transduction. 
Work done in collaboration with Dr. Eric Kandel 
(HHMI, Columbia University) has shown also that 
fyn~ mice exhibit specific defects in long-term po- 
tentiation (LTP) in the hippocampus, with accom- 
panying abnormalities in learning and memory. 
Other collaborative work with the laboratory of Dr. 
Harold Varmus (University of California, San Fran- 
cisco) is aimed at understanding the role of tyrosine 
kinases in macrophages in src families. To this end, 
mice deficient for the hck and^r kinases have been 
produced as well. 
Mice deficient in more than one kinase have been 
generated by intercrossing src, yes, and fyn mutant 
mice. Consistent with the hypothesis that these ki- 
nases are able to compensate for one another, the 
expected number of double mutants is not recov- 
ered from crosses involving two kinase mutants. In 
some of these crosses, particularly those involving 
src, over 90% of the double mutants die at birth or 
during gestation. Crosses involving both fyn and yes 
have allowed the recovery of a larger fraction of dou- 
ble homozygotes. However, these animals become 
sick and die of membranoproliferative glomerulone- 
phritis, usually between the ages of two and four 
months. 
Because fyn may be involved in negative selection 
in the thymus, and yes is also expressed by thymo- 
cytes and T cells, it remains to be seen if these symp- 
toms are due to an autoimmune disease. In another 
line of work, interactions between nonreceptor tyro- 
sine kinases and neural cell adhesion molecules are 
being investigated by intercrossing kinase-mutant 
mice with mice, produced in the laboratory, that 
carry a mutation in the neural cell adhesion mole- 
cule LI . These studies are in progress. 
It has been known for some time that src and re- 
lated kinases are negatively regulated by phosphory- 
lation at the carboxyl-terminal tyrosine. Previous 
studies from this laboratory and that of Dr. Joan 
Brugge (HHMI, University of Pennsylvania) had 
demonstrated that another kinase was responsible 
for this activity. A candidate kinase, csk (c-src ki- 
nase) , has been shown to be functional on src and 
several other members of the src family. To examine 
the consequences of dephosphorylation of the 
carboxyl-terminal tyrosine, and therefore simulta- 
neous activation of several, if not all, members of 
the src family. Dr. Soriano's laboratory has produced 
mice carrying a disruption of the csk gene. Prelimi- 
nary observations indicate that this mutation leads 
to a recessive lethal phenotype. Because neoplasms 
have not been detected yet in heterozygous animals, 
it remains unclear whether csk acts as a tumor- 
suppressor gene. 
Promoter Traps in ES Cells 
The laboratory has been interested in identifying 
new genes that play critical roles in early develop- 
ment, based on the phenotypic consequences of 
their inactivation in transgenic mice. Because only 1 
out of 20 transgenic strains on the average exhibits a 
mutant phenotype, a screen based on "promoter 
traps" was devised. In this screen a reporter gene is 
placed downstream of a splice acceptor sequence in 
a retroviral vector. The reporter gene encodes a fu- 
sion protein, /3-geo, with both iS-galactosidase and 
neomycin phosphotransferase activities. Following 
infection of ES cells, administration of G4 1 8 selects 
for insertions downstream of a promoter active in 
such cells. These ES cells are then used to derive 
transgenic mice in which the activity of the mutated 
gene can be traced by |S-galactosidase expression. 
Fifty transgenic lines so derived have now been 
generated. These lines exhibit different staining pat- 
terns during development, ranging from restricted 
to specific tissues (e.g., CNS or liver) to ubiquitous. 
Among 42 strains with /3-galactosidase activity, 18 
exhibit a recessive lethal phenotype due to pro- 
moter trap events, indicating the efficiency of this 
type of screen. 
The first gene identified by this approach is tran- 
scriptional enhancer factor 1 (TEF-1), yielding a 
broadly expressed transcription factor. Analysis of 
mutant mice with a tagged TEF-1 gene has shown 
that the provirus insertion leads to a null mutation 
and that homozygous embryos die at midgestation. 
Further analysis of this and other mutant strains is 
under way. (The project described above was sup- 
ported by a grant from the National Institutes of 
Heahh.) 
262 
