DEVELOPMENTAL REGULATION OF GENES 
Shirley M. Tilghman, Ph.D., Investigator 
I. Transcriptional Control of the a-Fetoprotein- 
Albumin Locus. 
This laboratory is investigating genetic and mo- 
lecular mechanisms that govern cell type specifi- 
cation during embryogenesis in the mouse. One 
approach to this problem has been to study in de- 
tail the transcriptional activation of the mouse 
a-fetoprotein (AFP) and albumin genes; the expres- 
sion of these genes is restricted to a subset of endo- 
dermal cells in the mouse embryo. Molecular ge- 
netic approaches have been used to identify the 
DNA regulatory signals that are required for their 
activation, and a combination of genetic and bio- 
chemical approaches have been applied to the 
study of the proteins that interact with those 
signals. 
A. Identifying DNA regulatory elements. The albu- 
min and AFP genes constitute a small, linked gene 
family on mouse chromosome 5. Studies in many 
laboratories had delineated the positions of two 
sets of transcriptional regulatory elements sur- 
rounding the genes. Their roles in development 
were assessed by introducing segments of the locus 
into the mouse germline by microinjection of zy- 
gotes. These experiments demonstrated that three 
enhancers that lie between the two genes are es- 
sential for the activation of both genes early in de- 
velopment in liver and that a fourth enhancer 5' to 
the albumin gene exhibited little activity before 
birth. This dependence of both genes on the same 
regulatory signals provides an explanation for the 
maintenance of their linkage, since they arose from 
a duplication approximately 500 million years ago. 
Although the genes are co-activated in three cell 
types— the fetal liver, gut, and visceral endoderm of 
the yolk sac— the neonatal liver continues to tran- 
scribe the albumin gene at a high rate while it re- 
presses transcription of AFP. The element responsi- 
ble for the selective repression of the AFP gene in 
neonatal liver was localized to a small segment of 
DNA, between the enhancers and promoter of the 
gene, that acts as a dominant silencer sequence. Its 
removal from the gene results in its constitutive ex- 
pression in adult liver, indicating that the positive 
trans-acting factors necessary for AFP transcription 
are present in adult liver. 
Removal of the silencer has a less profound effect 
on the repression of the AFP gene in the adult gut. 
In situ hybridization studies have shown that AFP 
mRNA is synthesized in all of the epithelial cells lin- 
ing the rudimentary villi in the fetal gut. AFP mRNA 
in the adult gut, however, is restricted to a small 
number of enteroendocrine cells, which represent 
—1% of the cells of the adult villi. Thus in the adult 
gut the stem cells of the crypts are continuously 
generating two populations of mature cells, the ma- 
jority of which have the AFP gene in a silent config- 
uration and a small minority of enteroendocrine 
cells that transcribe the gene at a high rate. The re- 
pression in the majority of cells must proceed 
through a mechanism that is not identical to that in 
the adult liver, as the silencer defined for liver does 
not appear to play such a central role in the gut. 
B. Trans-acting factors that interact with the AFP 
gene.The tissue-specific behavior of the AFP gene 
promoter has been attributed to the action of an 
endoderm-speciflc factor, HNF-1. Its importance in 
AFP transcription was demonstrated by generating 
specific mutations in one of the two HNF-l-binding 
sites of the gene. This mutation has a deleterious 
effect on transcription in both liver and gut-derived 
cell lines. 
Genetic analysis in inbred strains of mice iden- 
tified a locus, raf that encodes a liver-specific cell- 
autonomous function that is necessary for the 
postnatal repression of the AFP gene. By crossing 
transgenic mice carrying various mutations in the 
AFP gene silencer with BALB/cJ mice, which harbor 
a mutation in the raf gene, Dr. Tilghman's labora- 
tory showed that the raf gene product does not in- 
teract with the silencer but requires one or more 
elements in the promoter or gene body itself 
II. Unusual Properties of the H19 Gene. 
The mouse H19 gene was originally identified in 
a screen of fetal-specific liver cDNAs for genes that 
were under the regulation of raf. DNA sequencing 
revealed that the longest open reading frame in the 
gene was contained entirely within the first of its 
five exons, sufficient to encode a protein of only 14 
kDa. The likelihood that this reading frame was 
used in vivo was reduced when the sequence of 
the human homologue was determined. Like the 
mouse gene, no long open reading frame was pres- 
ent. More striking, no open reading frame was held 
in common with the mouse gene, despite the fact 
Continued 
319 
