A trio of papers published March 16 in the journal Science answers a decades-old debate about a fundamental question in developmental biology: How do cells destined to form a particular tissue or structure remember what they’re supposed to be?
As multicellular organisms develop, their cells become progressively more specialized. Because every cell carries all the same genes, this specialization requires turning some genes on whilst turning others off, committing each cell to giving rise to a particular tissue such as skin, muscle, gut, or brain. Importantly, once cells decide which genes to express, all their descendants must remember the same special combination. This overlay of information, which controls how the same genes are used in different ways, is called epigenetics, and glitches in it can lead to everything from birth defects to cancer.
All three papers address the question of how cells keep such epigenetic memories. This question has been answered for some genes that must be kept on. Such genes make proteins that control their own expression and cells pass these regulatory factors down to their daughters to tell the gene to stay on. However, says Gary Struhl, PhD, professor of genetics & development at Columbia University, “it’s much harder to explain how cells remember to keep genes off, since silent genes do not make proteins that can be passed down to their daughters.” Consequently, biologists have debated for more than 30 years how genes are epigenetically silenced.
One camp argues that each gene is silenced by its own special network of regulatory proteins made by other genes and inherited by their descendants. Dr. Struhl calls this the “smart” model, but notes that it might not actually be that smart, as every silenced gene would require its own regulatory circuit. Others, however, posit a “dumb” system that attaches a generic chemical tag to genes that are supposed to stay off, and then copies that tag, together with the DNA, every time a cell divides. In this case, no special circuit is required — only the capacity to initially tag the gene and then to mindlessly inherit the mark from one cell generation to the next.
The latter idea got a boost with the discovery of enzymes that attach methyl groups to the histone proteins that coat DNA and package it into chromosomes. Indeed, numerous studies have shown that silenced genes are often coated by histone proteins that bear either of two specific methylation marks. However, advocates of smart silencing mechanisms have argued that this correlation doesn’t prove causality. Instead, the presence of these marks may merely reflect the silenced state of the gene which is inherited by some other, more fundamental process. And until now, there has been a conspicuous lack of evidence to counter this argument.
The new studies, one from Dr. Struhl’s group and two others from independent laboratories at the Max Planck Institute of Biochemistry in Germany and Harvard Medical School, now provide the evidence that the inheritance of these specially methylated histones can carry the memory of the silenced state.
That evidence was a long time coming and for Dr. Struhl began with his thesis work almost 40 years ago, when he discovered a gene encoding what we now know to be the first component of Polycomb Repressive Complex 2 (PRC2). PRC2 is the enzyme that catalyzes one of the two methylation marks associated with gene silencing. By mutating this gene, Dr. Struhl was able to show that PRC2 is required for epigenetic silencing of a special class of master regulatory genes called HOX genes that control developmental fate in the fruit fly Drosophila. Since then, PRC2 has been shown to be critical for the epigenetic silencing of HOX and other master control genes across the animal kingdom, including in humans. However, the question of how PRC2 activity maintains such epigenetic memories has remained unanswered.
Although Dr. Struhl’s subsequent research shifted to the study of secreted factors that organize cell pattern and organ growth, he has periodically returned to the question of how genes are heritably silenced as new discoveries and more powerful experimental approaches have come along. One such discovery was that HOX genes carry cis-acting DNA elements, called Polycomb Response Elements (PREs), which recruit PRC2 and are just as essential for epigenetic silencing. A second discovery was that methylated parental histones that are inherited during DNA replication serve as triggers to induce PRC2 to copy the mark to newly incorporated histones.
These discoveries posed two key questions. First, are PREs necessary only to initiate the methylated off state, after which it is propagated from one generation to the next by the ability of PRC2 to bind to and copy the mark? Or, must PRC2 be recruited to PREs to restore the mark after each replication cycle? The second, and more fundamental, question is whether the copying mechanism is correct, and, if so, does the ability to transmit and copy the mark carry the memory of the silenced state?
To answer these questions, Dr. Struhl used a molecular device he developed called a Flp-out cassette, which allows researchers to remove a defined DNA segment from the fly’s genome at any time. He then created a HOX gene that carries its PRE inside a Flp-out cassette and asked if removing the PRE cassette would lead to a loss of the off state. Using this approach he found that the PRE is, indeed, required continuously, but surprisingly, the loss of memory following PRE removal was not immediate. Instead, cells could undergo several rounds of cell division before silencing was lost, raising the possibility that PRC2 might be able to propagate the off state for multiple division cycles despite the absence of a PRE anchor. It also left unresolved whether inheritance, per se, of the methylation mark is responsible for the memory of the silenced state.
Around this time, Rory Coleman, a graduate student, joined Dr. Struhl to explore the molecular basis of epigenetic silencing. In a first series of experiments, they asked what happens to methylated histones following removal of the PRE and discovered that the percentage of methylated histones is diluted by 10 percent to 12 percent after each round of cell division. This result is consistent with a model in which DNA replication normally dilutes methylated histones by 50 percent, and PRC2 anchored at the PRE efficiently copies the mark from parental to newly incorporated histones to restore it to the 100 percent level. At the same time, it also showed that residual “free” PRC2 can copy the mark, but not well enough to restore it to the 100 percent level, explaining why silencing is eventually lost after several cell divisions.
They then performed a second set of experiments in which they blocked cell division after removing the PRE and discovered that doing so permanently stopped any further dilution of methylated histones. This result established that PRE anchored PRC2 activity is specifically required to restore the full quota of methylated histones following their dilution during replication. Finally, and most critically, they manipulated the ability of PRC2 to copy the methylation mark after PRE excision and demonstrated a causal relationship between the capacity to restore the mark and the ability to remember the silenced state.
Taken together, these findings establish a precedent that a histone tag that is inherited during replication can propagate a permanent state of gene expression. As such, it resolves a central challenge to the proposed role of histone modifications as carriers of epigenetic memory.
The investigators at the Max Planck Institute used somewhat different functional and molecular assays in flies to reach the same conclusion while posing further questions about the extent to which PRC2 can propagate the histone methylation mark on its own. The team at Harvard obtained equivalent findings in yeast, focusing on the other histone tag associated with heritable gene silencing and probing the requirement for anchoring of the enzyme complex that copies the mark.