Polycomb discoveries mark new way of thinking in the field
A major discovery from two groups in the Biochemistry Department could flip the understanding of a key developmental gene regulatory system on its head.
Super-resolution 3D-SIM images of cells stained with histone modifications (green and red) and DAPI (blue)
Published in Cell (1) and Cell Reports (2), the complementary papers come from the labs of Dr Rob Klose and Professor Neil Brockdorff together with colleagues from Oxford, London, Japan and the Babraham Institute.
Using different approaches, the researchers demonstrate the existence of an unexpected pathway in cells for targeting polycomb proteins, which establish the earliest transcription patterns in development.
Polycomb proteins play a key role in developmental gene regulation in most multicellular organisms, from plants to mammals. They were discovered in Drosophila as factors which allow cells to ‘remember’ patterns of gene expression, established through differentiation and development, through successive cell divisions.
The proteins accomplish silencing by chemically modifying histones around which the DNA is entwined. The two protein complexes in which polycomb proteins are generally found have different activities: Polycomb repressive complex 1 (PRC1) places a specific ubiquitin mark (H2AK119ub1) and Polycomb repressive complex 2 (PRC2) places a specific methylation mark (H3K27me3).
'Recruitment of polycomb proteins has traditionally been viewed within the context of a hierarchical pathway,' explains Dr Neil Blackledge, one of the post-docs in Dr Klose's lab who carried out much of the work. 'It has been widely proposed that PRC2 first places the methylation mark, then PRC1 recognises this and places its mark.'
Almost all work to date has focused on how PRC2 gets to its target. But some recent findings, including those from the Klose and Brockdorff labs, have revealed that the molecular chain of events that leads to formation of polycomb domains in vivo is more complicated than thought.
The two new papers have now gone on to show that in ES cells and mice, bringing in PRC1 and its ubiquitin mark is sufficient to recruit PRC2 – the complete opposite to the prevailing view.
The paper from Dr Klose and colleagues describes their two-pronged approach. They devised a system where they could target PRC1 to specific sites in chromatin in ES cells and then manipulated PRC1 to see what the effect on PRC2 binding would be.
'We deleted the catalytic core of PRC1 and saw that all H2A ubiquitylation is lost from cells,' Dr Blackledge explains. 'When we looked at PRC2 genome-wide using ChIP sequencing, we found that almost all sites show a significant loss.'
Collaborating with a group in Japan, the researchers went on to target PRC1 in mice. They engineered a mouse in which they could knock out the DNA-binding activity of a protein called KDM2B. The Klose lab had previously shown that this subunit of PRC1 specifically recognises non-methylated CGs – regions which most mammalian polycomb target sites are associated with.
Schematic to illustrate placement of H2A ubiquitylation by PRC1 complex (with its specific subunits) leading to PRC2 binding and H3K27 methylation, at a non-methylated CG site (CpG island) (Click to enlarge)
'When we deleted the DNA-binding domain of KDM2B, we lost a significant amount of PRC1 from its target sites and also saw loss of PRC2 at these sites,' says Dr Blackledge. This loss was reflected in the phenotypes of the mice, with the few surviving heterozygous mice showing abnormal body patterning typical of polycomb mutations.
The findings provide evidence that the KDM2B/PRC1 pathway contributes significantly to recruitment of PRC2 and the setting up of functional polycomb domains.
The paper from Professor Brockdorff's lab describes work that reaches a similar conclusion about the molecular chain of events at polycomb sites but using a different approach.
Building on recent studies showing an association between non-methylated CGs and polycomb recruitment, the group wanted to find out where polycomb would go if there was no DNA methylation in the cell.
To their surprise, they could visualise polycomb proteins being recruited to pericentric heterochromatin (areas flanking the centromeric region) in methylation-deficient ES cells. This is not a normal polycomb target, suggesting that the region's high non-methylated CG content is sufficient to bring in polycomb.
More detailed analysis using super-resolution imaging in the Micron facility revealed that the polycomb protein did not go to all pericentric heterochromatin. It was prevented from going to a subset of sites by specific chromatin modifications.
The group then used ChIP to look genome-wide at how polycomb complex distribution was altered by the absence of DNA methylation, tapping into bioinformatics expertise from the CGAT group at the University.
'When there was no DNA methylation we could see polycomb going away from its normal sites and going to new sites in the genome which are high in CG content,' explains Dr Sarah Cooper from Professor Brockdorff's lab who carried out the majority of the work. 'But chromatin modification associated with active gene expression stopped it going to a subset of these new sites.'
Exploiting the ability to follow polycomb recruitment at pericentric heterochromatin sites, the group developed a system to target specific factors there and establish the sequence of events.
'By bringing in H2A ubiquitylation, we could see recruitment of PRC2 and methylation of H3K27,' says Dr Cooper.
The two sets of findings challenge the long-held view about PRC1 and PRC2 recruitment by providing evidence for a different sequence of events in vivo. 'What we've both seen in these two papers is that if you can get H2A ubiquitylation being placed, then this can bring in PRC2,' adds Dr Cooper.
But the researchers agree that it is very unlikely that this could explain all the polycomb recruitment in a cell. For example, loss of PRC2 following deletion of the DNA-binding domain of KDM2B is incomplete. But for the time being, the problem of disentangling what is going on at complex endogenous polycomb sites means that researchers must use approaches like the ones used here to tease apart the system.
Schematic to illustrate potential models of PRC2 recruitment by H2A ubiquitylation (H2AK119u1), either by direct/indirect protein interactions, shown by the “reader” at the top, or by “chromatin state” at the bottom (Click to enlarge)
With the discovery of an unexpected interaction between PRC1 and PRC2, the findings raise an intriguing question: what is the molecular link between PRC1 (and its ubiquitylation mark) and PRC2?
There could be a number of possible explanations, suggests Dr Blackledge. 'Perhaps there is some component of PRC2 that recognises the H2A ubiquitylation, or maybe the modification is changing the chromatin in a way that facilitates binding of PRC2.'
He and Dr Cooper point to a paper recently published by Jürg Müller's group that uncovers evidence for the potential molecular basis of this recognition (3).
The group's work developed from an interest in understanding what H2A ubiquitylation does. Their biochemical studies to pull out proteins interacting with this histone mark show that known components of PRC2 can recognise it.
The discovery came as a complete surprise to the Klose and Brockdorff groups but fits well with their own findings. 'The work was done independently from both of ours but provides a possible molecular explanation for our observations,' says Dr Blackledge.
Professor Brockdorff and Dr Klose add that their findings open up important avenues. 'Our discoveries have revealed an unexpected new mechanism for polycomb group protein targeting and in doing so provide a sound footing for new work aimed at discovering how these complexes regulate normal gene expression.'
'This is an important step towards uncovering how and why perturbation of the polycomb systems can lead to cancer and other human diseases.'