also shown cytogenetically to lose this band. Several 
years ago Dr. Feinberg's laboratory first applied 
RFLPs (restriction fragment length polymorphisms) 
to show loss of allelic heterozygosity of Hp in WT, 
molecular evidence for a tumor suppressor gene on 
this chromosome. In collaboration with Drs. David 
Schlessinger (Washington University, St. Louis) and 
Bryan Williams (Hospital for Sick Children, To- 
ronto) , the laboratory cloned the 11 p 1 3 WT gene 
region in yeast artificial chromosomes (YACs) . 
Interestingly, they found that this region contains 
at least two genes {WIT- 1 and WIT- 2) expressed spe- 
cifically in developing kidney. WIT-2 (also known 
as WTl) is a transcription factor that is mutated in 
some WTs. These mutations occur infrequently, 
however, and the laboratory is investigating 
whether reduced expression of the gene may be a 
more common mechanism for tumorigenesis. Both 
WIT-1 and -2 showed reduced or absent expression 
in ~30% of sporadically occurring WTs. These tu- 
mors are of a particular histologic type that reflects 
the earliest stages of renal development, and WTs in 
WAGR patients are exclusively of this type. 
An Additional Wilms' Tumor Gene on llpl5 
One of the predictions of Knudson's model is that 
some families should develop Wilms' tumor as an 
autosomal dominant trait, by transmitting one mu- 
tant copy of the gene. Previously Dr. Feinberg's 
laboratory mapped such a disorder, Beckwith- 
Wiedemann syndrome (BWS), to 11 pi 5, and not 
llpl3 (site of WTl^, by genetic linkage analysis. 
BWS is characterized by multiorgan overgrowth, 
other developmental malformations, and predispo- 
sition to a variety of "embryonal" tumors, such as 
WT, rhabdomyosarcoma, and liver and adrenal tu- 
mors. Supporting the idea of genetic heterogeneity 
in WT, BWS-associated tumors, in contrast to WAGR- 
associated tumors, are of a histologic type reflecting 
later stages of renal development. They also show 
normal WTl expression. 
In collaboration with Drs. Marcel Mannens and 
Jan Hoovers (University of Amsterdam) , the labora- 
tory has found that several germline balanced chro- 
mosomal rearrangements in BWS patients lie within 
a small region of 1 lpl5. To localize this gene, the 
laboratory has isolated 23 YACs from llpl5. These 
were found to include six of the chromosomal 
breakpoints from BWS patients. The breakpoints lie 
within a 1 -Mb region, and the laboratory is now look- 
ing for candidate genes within the YACs spanning 
them. 
One of the most intriguing aspects of BWS is the 
possible role of genetic imprinting (allele-specific 
modification). At least two mouse homologues of 
1 lpl5 genes are imprinted, and some BWS patients 
show paternal uniparental disomy for most of 
llpl5, suggesting a difference between maternal 
and paternal alleles. The laboratory found that six of 
six balanced germline rearrangements were mater- 
nally derived, and six of six unbalanced duplica- 
tions of llpl5 were paternal in origin, also consis- 
tent with an imprinted gene. 
A Novel Strategy for Isolating 
Tumor Suppressor Genes 
Although the WTl gene is on 1 Ipl 3, Dr. Feinberg 
and his colleagues found that genetic loss in WT, as 
reflected by loss of allelic heterozygosity, specifi- 
cally involves llpl5, site of the BWS gene, rather 
than 1 1 p 1 3 . Furthermore, 1 1 p 1 5 is also commonly 
lost in other embryonal tumors and in ovarian, lung, 
and breast cancers. Thus WT is more complex than 
investigators had previously believed, and an 1 Ipl 5 
gene may also be important in the progression of 
common cancers. 
A fundamental problem in the identification and 
isolation of tumor suppressor and other growth-in- 
hibiting genes is the fact that one loses the power of 
genetic complementation at the subchromosomal 
level. Thus, while the existence of tumor suppres- 
sor genes was first demonstrated by cell fusion — 
and suppression can also be detected by mono- 
chromosome transfer into tumor cells — direct 
expression cloning of suppressor genes in manage- 
able vectors is usually not possible, since suppres- 
sion is normally selected against. Furthermore, 
while YACs have been transferred to mammalian 
cells, success has been limited to small genes and 
specific cell types, and assaying for tumor suppres- 
sion with the thousands of YACs needed for a chro- 
mosome is impractical. The laboratory therefore 
sought to develop a strategy for transferring 
subchromosomal fragments intermediate in size be- 
tween YACs and chromosomes. 
This strategy involves three steps: 1) transfection 
of a mammalian selectable marker gene into mouse 
cells containing a single independently selectable 
human chromosome; 2) chromosome transfer by 
microcell fusion followed by double selection for 
both the human chromosome and the marker gene, 
yielding a panel of hybrids that carry the human 
chromosome with the marker gene integrated ran- 
domly within it; and 3) irradiated microcell transfer 
of the pooled hybrid panel to isolate individual 
marker-containing chromosomal subfragments from 
the remainder of the chromosome. The result- 
ing "subchromosomal transferrable fragments," or 
STFs, unlike conventional radiation hybrids, can 
then be transferred individually to any mammalian 
GENETICS 1 87 
