Postdoctoral researcher Skirmantas Kriaucionis sat at the computer in the lab of molecular biologist Nathaniel Heintz at the Rockefeller University in New York, looking at an unidentified spot that hung mysteriously on the screen in front of him. He had been watching as, line by line, the scanned image of separated neuronal nucleotides took shape. Fifteen minutes later, the image was complete, but the picture was less clear than ever.
He immediately started repeating the experiment, reproducing the same result until he had convinced himself that the spot was real, and not just a mistake or an artifact of botched methodology. When Heintz came by his bench to see how the experiment was progressing, Kriaucionis showed him the unexpected results...
“As soon as I saw the spot, I thought it was important and interesting, [but] I didn’t believe it,” Heintz says. “[It] wasn’t A, G, C, or T, or methyl C”—the methylated form of cytosine (5-methylcytosine, or 5mC) and the only known modified base in mammalian DNA—“[so] we immediately knew there must be some other modified nucleotide in the genome of these neurons.”
The neurons in question were Purkinje cells, some of the largest neurons in the mammalian brain. Like some other large nerve cells, these cells contain giant nuclei—five to 10 times larger than the average nucleus—that have very little heterochromatin, the more tightly packed form of DNA. Curious about the reason for this peculiar structure, Kriaucionis and Heintz isolated these nuclei, as well as the nuclei of the smaller, heterochromatic granule cells for comparison, to quantify the amount of DNA methylation, which is associated with DNA packing. As suspected, they found less 5mC in Purkinje cells than granule cells, but surprisingly, no compensating increase in normal, unmethylated C. Instead, it appeared that a sixth nucleotide was taking up the slack.
Another series of tests revealed the mystery nucleotide’s identity as 5-hydroxymethylcytosine (hmC), a modified 5mC that had previously been identified in bacteriophages, but never conclusively in mammals. “It’s very exciting,” says molecular biologist Timothy Bestor of Columbia University, who was not involved in the work. “For the last 60 years we thought there was only one modified base in mammalian DNA. Now we know there’s at least two”—both of which are modified cytosines.
At the same time as Heintz and Kriaucionis were puzzling over the identity of the mysterious shape on the screen, at Harvard University chromatin biologist Mamta Tahiliani and her colleagues stumbled upon an enzyme, known as TET1, that converts 5mC to hmC (Science, 324: 930–35, 2009). She was actually looking for proteins that might demethylate 5mC, she says, joining in the “long hunt for an event that could modify 5mC.” The known DNA demethylases remove a methyl group from a nitrogen atom, Tahiliani explains, but in 5mC, the methyl is attached to a carbon, which makes for a much stronger bond.
Tahiliani began her search by looking for mammalian homologs to the proteins that alter thymine to create the modified “base J” found in trypanosomes, parasitic protozoa that infect insects. Using multiple sequence alignments and iterative searches, the team narrowed in on the TET family of proteins. When they overexpressed TET1 in mouse embryonic stem (ES) cells, they saw a decrease in 5mC, suggesting that the enzyme might, in fact, be altering the already modified base. And just like Kriaucionis and Heintz, thin-layer chromatography revealed that the decrease in 5mC was accompanied by an unidentified spot: hmC.
The discovery of the enzyme that appears to convert 5mC to hmC is “critical,” Heintz says. “It says that this modification is not a random accumulation of chemical damage without any biological process controlling it.” To the contrary, it implies that hmC is performing some sort of biological function.
“The fact that [hmC bases] exist means they have some beneficial function,” Bestor says. “In principal it could do anything. Every modification increases the information content of DNA.”