Embryo uses chemistry to tell genes apart.

When male and female germ cells unite to create a new life, their genetic contributions come with tags labeling them as paternal or maternal. "Reading" these tags, the developing embryo sometimes selects the mother's gene for expression, while the father's version of the same gene stays idle. In other cases, the embryo draws on the paternal gene, leaving the maternal one unused. Activating both parental genes -- or neither -- can cause fetal death or a serious disorder.

Now, researchers at the Whitehead Institute for Biomedical Research in Cambridge, Mass., report that this discrimination, called genetic imprinting (SN: 5/2089, p.312), relies on a chemical mechanism. Methylation, the addition of molecules called methyl groups to the DNA molecule, is crucial for enabling the embryo to tell the parents' genes apart. They report their findings in the Nov. 25 NATURE.

Geneticist En Li, now at Massachusetts General Hospital-East in Charlestown, says he and his co-workers showed for the first time that an embryo needs DNA methylation to selectively activate imprinted genes. Researchers had suspected that sine methylation is an inherited DNA modification, it might help genes "remember" their origin. Yet no one had ever directly linked the two.

"This study goes a long way towards proving [that link]," comments M. Zaim Surani, a geneticist at the Wellcome/CRC Institute of Cancer and Developmental Biology in Cambridge, England. "It provides very strong correlative evidence, although we still need more experiments to nail it down completely."

The new evidence comes from knockout mice, animals altered to disrupt the gene for the DNA methylation enzyme. "We created two mutations," explains Li. "The less severe one reduces the degree of meythlation by 60 to 70 percent; the more severe one prevents methylation almost entirely. This way, we can study genetic effects and also quantify them."

Analyzing three genes, the Whitehead group found that in mutant embryos the parental imprints were erased. Moreover, mutant embryos didn't follow the normal pattern of activating only one parent's copy of a gene. It appeared as if they could not distinguish which gene came from which parent, Surani says.

Take, for example, H 19, a gene of still mysterious function. Normally, only the maternal copy gets expressed, while the parental one is repressed. In the mutant embryos, however, the lack of DNA methylation lifted that repression, causing both genes to become active.

That finding fit nicely with an assumption about DNA imprinting: that it represents a way to inhibit a gene. It turned out not to be that easy, though. The other two genes, one encoding a protein called insulin-like growth factor 2 (Igf-2) and one encoding the Igf-2 receptor, showed just the opposite behavior.

Rather than activate the normally silent gene, the mutation silenced the normally active gene, thus thwarting the production of the corresponding proteins in the embryo, Li says. This finding indicates that methylation activates some imprinted genes but keeps a lid on others, Li explains.

Adding another twist is the finding that DNA methylation seems to act on some genes directly but uses genetic go-betweens, called gene silencers, to act on others. Depending on their methylation status, these silencers could determine the fate of imprinted genes. However, this is speculation, Li admits. "Nobody has evidence yet to pin down a silencer and prove what it's doing."

Not all genes are equally sensitive to the loss of DNA methylation, he notes. The imprint of the Igf-2 receptor gene vanished only in embryos carrying the more severe mutation, perhaps because this gene enjoys preferential treatment, Li suggests. The less severely mutated embryos might rally their remaining methylation capacity to secure adequate methylation of the Igf-2 receptor gene, while somehow deeming other genes less important, he speculates.