June 6, 2000 Scientists at NIDDK have discovered a key mechanism that controls the expression of Igf2, a gene that codes for insulin-like growth factor 2. Adam C. Bell and Gary Felsenfeld reported in the May 25 issue of Nature that a protein called CTCF prevents Igf2 from making the growth factor, which acts in fetal development. Their findings show that the control mechanism for this imprinted gene is novel.
Most of the thousands of genes a person inherits come in two copies, and both function the same way. Imprinting is a process in which one of the two copies of a gene is switched off. Scientists have suggested that this is done to protect organisms that would otherwise be harmed by receiving a double dose of a particular protein. In fact, overexpression of Igf2 in humans is associated with fetal overgrowth and in some cases the development of tumors.
But functioning is different for the few genes that get imprinted, or marked, when eggs and sperm develop. Either the copy from the mother or the father will be active. Which one functions depends on the presence of markers called methyl groups that chemically modify the DNA near the gene. In the case of Igf2, the copy inherited from the father is functional and has methylated DNA nearby, while the copy from the mother is inactive and is unmethylated.
Bell and Felsenfeld had been studying the protein CTCF because it is part of a boundary element between the loosely configured DNA where genes are active and the more tightly packed DNA where genes are inactive. "The protein is interesting because it works as an insulator, or roadblock, in other systems," says Felsenfeld, whose main work is in chromatin structure. The Felsenfeld laboratory had shown that CTCF binds to specific sites on DNA and insulates enhancers (gene-activating elements) from the genes they act on. Bell and Felsenfeld wanted to learn whether CTCF played a similar role in keeping the maternal copy of the Igf2 gene turned off.
In the work described in Nature, the researchers report that DNA near the Igf2 genes of mice and humans contain CTCF binding sites. These lie between the Igf2 gene and a distant enhancer. Using cells in culture, they next showed that these binding sites can only block the action of an enhancer on the gene when the binding sites lie between enhancer and gene. This blocking ability is a signature of insulator activity.
The CTCF binding sites are found in the same region of DNA that is usually methylated on the paternal copy of the Igf2 gene and unmethylated on the maternal copy. In their final experiment, Bell and Felsenfeld showed that adding methyl groups to the CTCF DNA binding sites prevents the binding of the CTCF protein. The finding strongly suggests that methylation abolishes the insulating capability of CTCF on the paternal copy of the gene, says Felsenfeld. As a result, the enhancer can activate Igf2 expression on that copy. In contrast, insulation on the unmethylated maternal allele prevents Igf2 from being expressed.
Medically, methylation matters. When the insulator is lost because methylation occurs on the maternal copy of Igf2, there is an overexpression of Igf2, which is associated with Beckwith-Wiedemann syndrome. This fetal overgrowth disorder predisposes children to malignancies such as Wilms tumor.
Scientists have long known about the importance of methylation, but many of the mechanisms governing imprinting are still being worked out. Felsenfeld's study supports an important hypothesis proposed by Princeton University researcher Shirley Tilghman several years ago. She suggested that the Igf2 region might contain an insulator that prevented the maternally inherited gene from producing the growth factor for which it codes and that methylation must block this insulator.
"Our work and hers confirms the hypothesis," says Felsenfeld. The Tilghman lab's recent Igf2 inactivation studies also appear in the May 25 issue of Nature.
Felsenfeld wants to know what other roles CTCF plays. "The fact that CTCF binding, and therefore insulator activity, can be turned on and off by changing the methylation state of DNA suggests that there may be many other places in the genome which use the same mechanism," says Felsenfeld.
Citation: Nature, Vol. 405, 25 May 2000, pp. 482-485.
# # #
CONTACT: Anna Maria Gillis
(301) 402-7875
