DALLAS--A team of developmental biologists examining how tumor
growth factor (TGF)-beta sends messages into cells has come up with
the first animal model for colon cancer that closely mimics the human
The "knock-out" mouse lacks the gene for the intracellular
protein Smad3, which transduces signals from the TGF-beta receptor to
the cell nucleus and alters gene expression. Mice without Smad3
cannot respond to TGF-beta regulatory signals. They also develop
colorectal cancers remarkably like those in humans. Such tumors are
deeply invasive, occur at both ends of the bowel, and metastasize to
The new mouse model "should provide a powerful tool for
identifying the genetic and molecular alterations that lead to
malignancy" as well as providing a useful model for trials of
drugs to prevent or treat human colorectal cancers, the researchers
said in their report in the September 18th issue of Cell.
The research group is led by Luis F. Parada, PhD, and Jonathan M.
Graff, MD, PhD, and includes James A. Richardson, DMV, and graduate
student Yuan Zhu, who made some of the key observations and is first
author on the Cell paper. All are at the University of Texas
Southwestern Medical Center.
"We have a particular interest in how external signals are
transmitted to the cell nucleus, resulting in modification of gene
expression and leading to organized embryogenesis," Dr. Parada
told Oncol-ogy News International. "Jon Graff had discovered and
studied Smad genes in the frog, and we decided to investigate these
genes further in the mouse."
The investigators expected that homozygous Smad3-negative mutations
would be lethal during embryonic development and that heterozygotes
(having one normal Smad3 gene) might eventually develop some form of
Mr. Zhu mutated the mouse Smad3 gene to test this idea (see figure).
The gene was inactivated in mice by homologous recombination.
"To our surprise (and initial chagrin), the homozygous mice were
born and appeared normal," Dr. Parada said. Smad3 was clearly
not required for normal embryonic development. However, Mr. Zhu kept
the mice and examined them regularly to see what would happen as they
aged. "To our delight, colon cancer developed in these mice
after several months," Dr. Parada said.
The final surprise was that colon cancer developed in all 129 of the
inbred mice who were homozygous for the Smad3 mutation. Heterozygous
mice are viable and fertile, and have been observed for up to 9
months with no apparent signs of disease.
The current model for development of colorectal adenocarcinoma
suggests that the disease progresses from normal to hyperplastic
colonic epithelium, then to benign adenomatous polyps and to invasive
and then metastatic carcinoma.
"It would appear that this Smad3 knock-out mouse resembles the
histo-pathology of the human condition with considerable
fidelity," Dr. Parada said. "If this is the case, given the
100% incidence of cancer within a defined window of time, these mice
may be an ideal substrate for therapeutic and pharmacological
studies." Among the similarities to human disease: The mouse
tumors are invasive and metastatic, limited to the colorectum, found
mostly at the end of the lower intestine and at the boundary with the
small intestine, and exhibit most of the stages of colon cancer progression.
To produce the homozygotes, the chimeric mice (lacking a Smad3
allele) were bred both with the mouse strain from which the
blastocyst came, to produce an inbred strain of Smad3-negative
heterozygotes (see figure), and with a different strain, to produce
Mating heterozygotes from the inbred strain to each other resulted in
the Smad3 homozygous mice who all developed colon cancer. However,
homozygotes produced by mating heterozygotes from the hybrid strain
were less likely to develop colon cancer, and, when they did, the
tumors developed much later.
"This raises the new question of what are the
modifier genes present in the other mouse strain that so
dramatically reduce the incidence of cancer," Dr. Parada said.
Finding out why these homozygous mice are less susceptible to colon
cancer could open the door to new preventive approaches in humans.
The findings confirm a central role for the loss of TGF-beta
signaling in at least some forms of colon cancer. TGF-beta has a
variety of functions, including inhibiting the proliferation of cells
such as colonic epithelium. A TGF-beta receptor mutation has also
been reported in a patient with colorectal cancer.
Three receptors are involved in transducing TGF-beta signals: Smad2,
Smad3, and Smad4. Previous attempts to produce homozygous Smad2 or
Smad4 knock-out mice were unsuccessful because the mice all died in
utero or shortly after birth, long before tumors could form.
Successful production of Smad3 knock-outs that develop colon cancer
constitutes the "proof of principal" that loss of the
ability to respond to TGF-beta sets up the colon for neoplastic