Some snippets of DNA that have been highly conserved throughout evolution pose a puzzle for researchers: What, exactly, does the DNA do?

When the genetic ciphers are deleted, individual laboratory animals appear to go on living just fine.

"The true function of these regions remains a mystery, but it's clear that the genome really does need and use them," said Gill Bejerano, assistant professor of developmental biology and of computer science at Stanford University.

In an article published in the Oct. 2 issue of Genome Research, Bejerano and graduate student Cory McLean report that these "ultraconserved" regions are about 300 times less likely than other regions of the genome to be lost during mammalian evolution. Previous research has shown that mice missing each of four regions behave perfectly normally.

"It's very surprising that none of the four has any observable phenotype," said Bejerano. "In some ways it just doesn't make sense. When we tried to determine whether similar deletions occur in the wild, we found that this is almost never seen in nature."

The regions are not intrinsically resistant to change: they are mutated in about one in 200 healthy humans. But the changes are smoothed by the process of evolution, and Bejerano and McLean believe that something similar - though too subtle to observe in the experimental conditions -- may be happening in the laboratory mice.

After establishing how rarely the ultraconserved segments are deleted, the researchers investigated whether the degree of homology (which is the percentage of nucleotides shared between species) or the extent of conservation (the evolutionary distance between species that share the sequence) correlated most closely with the possibility that it would be lost in primates or rodents.

In the evolutionary process, sequences shared among many distantly related species are likely to be older than sequences found only in closely related species. In this case, the researchers found that less-highly conserved sequences shared among several distantly related species - including opossum, platypus, chicken, frog and fish - are more likely to also occur in humans than are more-homologous sequences that occur in only a few closely related species. The likelihood that a sequence will be found in humans increases as the evolutionary age of the sequence grows.

"Interestingly," said Bejerano, "the longer the sequence has been in us, the less likely it is to be lost. It's almost like the bricks in the foundation of a building, which hold up the rest of the structure."

The upcoming availability of several additional well-sequenced mammalian genomes will give the researchers even more data.

And putting the laboratory mice missing the ultraconserved regions through a variety of conditions, such as changes in diet or living habits, may emphasize any differences between them and the mice without changes.