Transcription Apparatus Seen to Uncoil—and Recoil—DNA

Studies Begin to Integrate Various Known Histone Modifications

Anyone who has ever made a necklace knows that it is much harder to get beads on a string than to get them off. Might the same be true for chromatin? Because this classic bead-on-a-string structure of nucleosomal DNA must be taken apart every time a gene is transcribed, scientists have wondered for decades about the molecular jewelers that wind DNA back onto the histone proteins that form the chromatin core. But over the last few years, work by several HMS faculty members, including Kevin Struhl, the David Wesley Gaiser professor of biological chemistry and molecular pharmacology; Stephen Buratowski, HMS professor of biological chemistry and molecular pharmacology; and Fred Winston, HMS professor of genetics, has helped to completely refashion the understanding of this dynamic restoration. Their groundbreaking studies reveal that the very machinery that requires chromatin to be unraveled, the transcription apparatus, is also the one that ensures it is faithfully reassembled.

Photo by Graham Ramsay

Seminal contributions from the labs of Fred Winston (front), Kevin Struhl (back), and Stephen Buratowski (left) have substantially changed the understanding of genetic regulation, revealing how the transcriptional machinery not only unwinds, but rewinds DNA around core histones. In their latest findings, Struhl, Buratowski, research fellow Michael Keogh (right) and their colleagues show how RNA polymerase–dependent deacetylation of histones plays a key role in reassembly of the chromatin structure.

New installments in this developing story were recently described by Struhl and Buratowski in the Dec. 22 Molecular Cell and the Nov. 18 Cell, respectively. Independently, the two labs arrived at the same conclusion—that deacetylation of histones is inextricably linked to the progression of RNA polymerase II along the DNA template. These findings give further support to the idea, developed independently by Winston and Struhl, that restoring repressive chromatin is essential to prevent aberrant transcription at internal start sites. “In short, the work confirms that the transcription elongation machinery is sophisticated enough to guarantee transcriptional fidelity,” said Struhl.

Rapid Recruitment
Deacetylation of histones strengthens their affinity for DNA, while acetylation has the opposite effect. One might predict, therefore, that quiescent genes might be tightly packed around deacetylated histones. In fact, things are not so simple. In most yeast genes there are more acetylated histones at the promoter regions and more deacetylated histones in the coding regions: this makes sense because the chromatin needs to open up first at the promoter to allow access to the DNA by transcription factors. In 2004, Struhl and colleagues showed that this pattern depends on the activity of a protein called Eaf3. Because others showed that this protein associates with a histone deacetylase called Rpd3, the suggestion was that Eaf3 might preferentially attract the deacetylase to the coding regions, but it was unclear how. Now, Struhl, together with postdoctoral fellow Amita Joshi, and Buratowski, together with research fellow Michael Keogh and colleagues, demonstrate that Rpd3 recruitment depends on a different posttranslational modification of histones—methylation. Because methylation of histones has been linked to RNA polymerase II activity, their finding indicates that deacetylation, too, must be linked to transcription.

Methylation of histones is not uniform across genes, either. Several years ago, Buratowski and others showed that histone H3 can be methylated at lysine 36 by a protein called Set2. This takes place mostly in the coding regions where histone H3 is primarily deacetylated. Figuring that these two patterns are probably not coincidental, Joshi used chromatin immunoprecipitation experiments to see what happens to histone acetylation in yeast that lack Set2. She found that when histone H3 is not methylated at lysine 36, the normal pattern of more H3 acetylation at the promoter and less in the coding regions breaks down. In addition, Joshi found that when Eaf3 is missing its “chromodomain,” a motif that interacts with methylated lysines on histones, then deacetylation of histones in coding regions is compromised. The findings indicate that the methylation at lysine 36 must occur before histone H3 can be deacetylated.

Three Is Not a Crowd
Meanwhile, Buratowski, Keogh, and collaborators in California and Toronto had been trying to figure out how the Rpd3 deacetylase represses different stages of transcription. They found that the deacetylase actually exists in two different complexes, but that only the smaller one binds to Eaf3. In order to pinpoint a role for Eaf3, the group analyzed yeast strains that lack this subunit. They found that these expressed almost the same subset of genes as Set2-negative yeast, a strong indication that Eaf3 and Set2 function in the same biochemical pathway. Indeed, Keogh and colleagues went on to show that point mutations that abolish the chromodomain activity of Eaf3 prevent the Rpd3 complex from binding chromatin, as did loss of Set2 activity or mutation of histone H3 lysine 36. This work shows for the first time there are two distinct Rpd3 complexes that are specifically targeted to coding and promoter regions.

Together, the two papers were also the first to show that histone methylation is directly related to histone deacetylation. But the papers also link deacetylation to transcription because the methyl transferase Set2 is recruited to chromatin by a complex that travels with RNA polymerase II. In fact, Set2 recruitment is dependent on the phosphorylation status of the polymerase, which, as Buratowski showed five years ago, changes as it moves away from the transcription initiation site. The amazing interdependence of three different posttranslational modifications— phosphorylation, methylation, and acetylation—is even more remarkable given that it is also spatially controlled to ensure that deacetylation occurs preferentially in the coding region. Furthermore, Struhl’s lab previously discovered that a different methyl transferase, Set1, methylates a different histone H3 amino acid, lysine 4, and that this occurs at the 5' end of coding regions. This methylation depends on yet a different phosphorylation of RNA polymerase II, indicating that the polymerase, as it travels down the gene, gets phosphorylated in at least two distinct ways, corresponding to two distinct methylation events.

All these findings fit well with the bigger theme of restoration of repressive chromatin structure. “It may seem paradoxical to link transcription to transcriptional repression, but it allows genes to be turned off as rapidly as they are turned on,” said Buratowski. This may be important for regulation of gene activity in yeast—and mammals, which have homologs of all these factors—but it also ensures that the coding regions are not left exposed to wayward initiation factors. In fact, both papers report that loss of Eaf3 leads to enhanced transcription. Buratowski and colleagues found that without Eaf3, transcription elongation factors Bur1 and Bur2 are no longer needed, while Joshi and Struhl demonstrated that transcription initiation occurs within coding regions. This confirms findings reported by Jerry Workman and colleagues at the Stowers Institute for Medical Research in Kansas City in a Cell paper that ran back-to-back with Buratowski’s.

Of course, there is more to rethreading the chromatin structure than deacetylation of histone H3. Winston and colleagues demonstrated two years ago that special chaperones work rapidly to restore all four histones onto the DNA once the RNA polymerase has passed. These chaperones may be the true jewelers, while deacetylation may serve as the glue that keeps the histone beads in place.

—Tom Fagan

top