All tags Epigenetics The Role of Polycombs in Cancer webinar

The Role of Polycombs in Cancer webinar

Review Polycomb group (PcG) proteins and their role in cancer

Join Professor Bracken as he discusses the connections between Polycombs and cancer, by considering a number of early and more recent studies on the subject, and provides a perspective on the future and how this knowledge could be harnessed for improved cancer therapies.    

Webinar Topics:

  • Introduction: Polycombs are repressors of developmental genes
  • Polycombs initially presumed to be oncogenes: The INK4A and DNA methylation links
  • The NGS revolution: Polycombs are both oncogenes and tumour suppressors
  • Future cancer therapy and perspective

About the Presenter:

Adrian Bracken is an Assistant Professor of Medical Molecular Genetics at Trinity College in Dublin, Ireland. Previously, he was a Post-doctoral research fellow in the lab of Professor Kristian Helin at Copenhagen University, Denmark and at the European Institute of Oncology in Milan, Italy.

Since 2009, Dr. Bracken and his lab have focused on understanding the role of Polycomb group Proteins in stem cells and cancer.

Webinar Transcript:

Hello. Welcome to Abcam's webinar on the Role of Polycombs in Cancer. Today's principal speaker is Adrian Bracken, Assistant Professor of Medical Molecular Genetics at Trinity College, Dublin in Ireland. Adrian was a post-doctoral research fellow in the lab of Professor Kristian Helin at Copenhagen University and at the European Institute of Oncology in Milan. Since 2009, his research has been focused on understanding the role of Polycomb group proteins in stem cells and cancer. Joining Adrian today will be Miriam Ferrer, Product Manager for cellular assays at Abcam. Miriam completed her Biology degree at the University of Barcelona, and has a PhD from the Vrije University in Amsterdam. After completing her PhD she joined MRC Laboratory of Molecular Biology in Cambridge. I will now handover to Adrian who will start this webinar.

AB:    Thank you, Vicky. It's a pleasure to talk about the role of Polycombs in cancer. Today I'm going to break my talk up into four sections, so we're going to start off with an introduction about Polycombs, then we're going to move on to some of the early connections between Polycomb and cancer. Then really to the more recent revolutions and next-generation sequencing we found a lot of mutations in genes that encode Polycomb group proteins, both activating and inactivating mutations. We'll talk about that and then we'll move on towards the end to how we can use all this information towards new cancer therapies, and we'll give a little bit of a perspective on what I think will happen in the next few years.

So Polycombs, what are they? Well, they were identified in Drosophila as being repressors of HOX genes, homeotic genes. So here's a classic experiment and you can see on the left a wild-type embryo, and you can see its staining for this particular HOX gene, AbdB. So, of course, it's expressed this particular region of the embryo, and if you look at the right side of the slide you can see here is a Polycomb mutant. So it lacks the gene or it's mutated in the gene that encodes the Polycomb protein, and you can see the HOX gene is still expressed where it should be expressed, but of course, it's upregulated in regions of the embryo where it shouldn't be expressed. So the classic phenotype of Polycomb is shown on this slide, and that is that it is required for the repression of HOX genes, the maintenance of the repression of these HOX genes during development.

So Polycombs, as I said, are repressors of HOX genes and when these early genetic studies were done, the Drosophila geneticists wanted to find out other genes that could rescue this phenotype. Here's another classic experiment where these scientists discovered kind of like the opposite of Polycombs, which are a group of genes called trithorax group gene. So just to take you through this slide, there are stainings for the Scr HOX gene and these are leg imaginal disks, so you can see the imaginal disk of leg one, two, three. The main points in the wild-type stainings here is that you'll see that the particular HOX gene Scr is highly expressed in this leg, and much less expressed in this leg and this leg. Again, when you knockout a Polycomb you can see the HOX gene is getting upregulated here and here, so that's the classic, again, Polycomb phenotype of derepression of HOX genes. So what they did was they screened many, many genes to see if they could rescue this phenotype. You can see here they actually found many genes, and they called these genes trithorax group genes and you can see here. So if you have a double mutant of a Polycomb and a trithorax, you can again more like go back to wild-type where you have lower levels of this particular HOX gene in leg two and leg three.

To put that into perspective and, of course, there were many, many other studies, but what we now know is that trithorax and Polycomb group proteins are probably, in my view, and of course, I'm biased, but they're probably two sets of the most interesting epigenetic regulators in terms of developmental biology. What we think of them as being is very important for cellular memory, and what happens, of course, in early development is that transcriptional activators and transcriptional repressors set gene transcription effects.

But it's actually the trithorax and Polycombs that maintain these either on or off states during subsequent cell division, so we can see that depicted here. Here is the transcriptional activator activating the gene, and then through many, many subsequent cell divisions the trithorax group proteins that maintain the active stage of this gene here, and you can see that the transcriptional activator is often not even expressed in these cells.

Similarly so with Polycombs, a transcriptional repressor will repress a particular gene and a particular cell at a particular time, but it's actually Polycombs that will take over and maintain the repression of that gene during many subsequent revisions. If you remove Polycomb early in development you get the upregulation of HOX genes, so I would look at Polycombs as really being maintainers of a particular gene transcription stage, repressed state and not actually active repressed as such.

How do Polycombs work? What are they? What do the genes encode? Of course they encode proteins and these proteins form multi-protein complexes, and most famously the Polycomb repressive complex 1, as depicted here, and Polycomb repressive complex 2. We'll probably talk a little bit more about Polycomb repressive complex 2 today. What we know is that in Drosophila we have a protein called EZ and it has two homologs in mammals: EZH1 and EZH2, and we'll talk a lot about EZH2 today. EZH2 is in the PRC2 complex together with EED, at least in mammalian cells it's called EED, and Suz12. We'll talk a lot about Suz12 and EED later on.

Just to say, there's another complex, the PRC1 complex and this contains, again, many more proteins in mammals compared to Drosophila, but if you think about it through evolution, in terms of the multi-protein complexes, we have a similar sort of a structure in that we have a PRC2 and a PRC1 in Drosophila right up to mammals. So these multi-protein complexes are conserved through evolution. Well, we have another PRC1 complex called a non-canonical PRC1 complex, and we won't talk too much about that today.

How does it work? Well, probably the most important discovery in terms of the biochemistry, the PRC2 complex was the discovery of its activity. What this slide is depicting is the biochemistry to do histone methyltransferase enzymatic assay. So you can see here, here's the PRC2 complex purified, and what they do is they mix it with histone H3, but this is a recombinant histone H3. Either wild-type or recombinant histone H3 mutated at the various lysines along the tail, so lysine 4, 9, 27, 36 and 79. You can see it's the purified PRC2 complex can methylate all of these except one, so it's not able to methylate recombinant histone H3 if you mutate lysine 27. So this tells us, of course, that the PRC2 complex is remarkably specific, it only methylates at lysine 27. Whereas the Suv enzyme, which is famous at being a histone H3K9 methyltransferase, it will not methylate the recombinant histone H3 if it's mutated at lysine 9. So these enzymes, these histone methyltransferases are specific. EZH2, of course, is the actual methyltransferase enzyme, but it needs its friends Suz12 and EED, and so on for this activity. So to look at the EZH2 protein, and this will actually become important later when we look at some of the EZH2 mutations in cancer. So, importantly, I'll draw your attention to the SET domain, so this SET domain is required for the histone methyltransferase activity of EZH2 and EZH2, of course, in the PRC2 complex.

So H3K27me3, what does that do? As I said, the PRC2 complex mediates the trimethylation of lysine 27me3, and what we think it does is it brings in the other complex PRC1 via the Polycomb proteins that it's called in Drosophila, or sometimes it was called HPC123. Now, we call it actually CBX8, CBX7 and CBX4. It has many names, and we only have one Polycomb group protein in Drosophila; we've about four to five in mammals. All of these proteins: Polycomb and the CBX proteins in mammals, they have a domain called a chromodomain. This chromodomain actually binds to and reads the H3K27me3, and then it brings in the PRC1 complex. Then what we think is that this PRC1 complex has - well, we know it has another activity - and what we think is that the monoubiquitylation of histone H2A at lysine 119, we think that contributes to the gene repression. The truth is that we actually don't fully understand how Polycombs mediate gene repression, and that's something that's a very active area of research at the moment.

Now we know how the complexes presumably get the chromatin, is that what we see if we look at the location of Polycombs and the location of the H3K27me3 genome-wide, and it is in fact the case. So the first genome-wide mapping studies of Polycombs were done about eight years ago now, and what they showed up was that, indeed, when you have PRC2 members, for example, Suz12, or when you look at K27me3, or the CBX8, again, a member of the PRC1 complex. Essentially, you see them in the same places, so you see PRC2, you see the modifications and then you see the PRC1 mark. So, again, strongly supporting the idea that the PRC1 complex gets to chromatin by binding to the mark H3K27me3, and mediated by the PRC2 complex.

So what are the target genes? Well, reassuringly, that Abd-B HOX gene I mentioned a little bit earlier, well we know in humans and mammals that there are four homologs: HOXA9, B9, C9 and D9 in mammals. All of those, reassuringly, are Polycomb target genes in mammals, and what we're saying here in this slide is that Polycombs directly bind to the HOX genes both in Drosophila and in mammalian cells. Just to broaden it out a little bit, it's not just HOX genes. Basically, the genes that are bound by Polycombs are a who's who of stem cell biology and differentiation biology, and that will make sense in a couple of more slides. So it's not just HOX genes, it's basically just developmentally genes are important in developmental biology.

So is the K27 methylation important for all of this? Well, the dream experiment would be to see if you could mutate the lysine 27 and see if that phenocopies the loss of a Polycomb. This dream experiment was actually published, to my delight, last year and I think this is a very, very elegant study. In short, what they show is they can knockout EZH2's homolog in Drosophila, but, cleverly, just in certain parts of the developing lava here, so you can see in red when the knockout in EZH2's homolog in Drosophila you get upregulation of various HOX genes, so that's very nice. They had to go to a huge lot of trouble to be able to knockout the histone H3, because, of course, there's 12 genes that encode histone H3 in Drosophila, even more in mammals. So they actually knocked out the HOX cluster and replaced it with 12 HOX genes that are histone H3 genes that are mutated at lysine 27, and you can see here that that phenocopy loss of the Polycomb EZH2. A very elegant study showing that, indeed, you can knockout a Polycomb repressive complex 1 or a Polycomb repressive complex 2 protein, and you will get upregulation of HOX genes. But you can even single this down to just one lysine residue in histone H3 that's required for this biochemistry to work.

What happens if you knockout Polycomb group proteins in mice? The quick answer is it's very lethal, and early post-implantation fatality and you can see that here; this is just a representative example of when we knocked out Suz12. But what happens when you look at embryonic stem cells? Fortunately, you're actually able to get embryonic stem cells lacking EZH2, EEDs as well, they go as fast as wild-type ES cells. They look pretty much normal, nothing really to say about the embryonic stem cells, but when you differentiate them that's when things go wrong.

So when you differentiate them you - in this assay we do what's called an embryonic body assay, so, essentially, you take off a key molecule called LIF and the embryonic stem cells grow into football-like structures. If you cut these football-like structures and look inside them, you'll see various features of differentiation. What actually happens is you have undirected, random differentiation going on. However, if you knockout Suz12 and in this case you still form the embryonic bodies, but they fail to differentiate. What we figured out was that this was due to a failure to properly repress embryonic stem cell genes, such as the famous NANOG gene, Oct-4 and so on. This study and many other studies helped us to formulate what we consider is the model of how Polycombs work during cell fate positions.

So, as I said, we found out that Polycombs bind a lot of developmental and stem cell genes, genes involved in differentiation and so on. Our lab and many other labs have tracked Polycombs during differentiation. This model that I am presenting here is essentially a sum up of a lot of work and what it tells you is that on your slide here you can see four different types of cells, so a stem cell and three different types of differentiated cells. I'm representing four genes: a stem cell gene on the left and then lineage A, B, C and these are differentiation genes. So you need, for example, lineage A gene to turn into differentiated cell type A. Let's look at differentiated cell type A, of course, that gene has to be on in that cell type gene, lineage B gene is on and in the differentiated cell B. You can see the stem cell gene and all of the stem cells.

So, actually, if you were to do a genome-wide ChIP-seq of Polycombs in any of these four types of cells, the pattern would be different. You could actually do a ChIP-seq of Polycombs, not know the cell type and be able to protect the cell type it is. Intrinsic in all of this model is that you can see that Polycombs are recruited to the stem cell genes during the differentiation, but they're also displaced from certain lineage-specific genes during the differentiation. So we kind of have a dance of Polycombs that they're both recruited to and displaced from key lineage-specific genes. I should point out that if we were to say we've 25,000 genes in a mammalian genome or the human genome, about 2,000 to 2,500 of these genes would be Polycomb-target genes. If you were to do a GenomOncology analysis of these genes, again, they're stem cell genes, differentiation genes, developmental biology genes. So it's pretty clear that these Polycombs are particularly important repressors of developmental genes, so that's what makes them very, very interesting from a developmental biology perspective.

But what's going on in cancer? So there were some very, very early links, actually, between Polycombs and cancer, probably the first link, well, definitely the first link was BMI which is a member of the PRC1 complex. It actually was identified as a gene that can, and I've emboldened the word 'can', so it can cooperate with c-Myc to induce lymphomas in mice. I underlined or emboldened 'can', the word 'can' because if the experiment doesn't say it does, it doesn't say in actual human beings that lymphomas, that BMI is collaborating as an oncogene with Myc, it just means it can. That's an important point, and we'll come back to that a little bit more. What's the evidence that BMI is deregulated in cancer per se? Not very much, I mean, it has been shown to be amplified in - as you see here in this bottom paper - in a few lymphomas, but this hasn't been looked at in a very thorough way, in my view. Certainly, BMI, as far as I'm aware, has not popped up in some of the next-generation studies, but it could be next week we find out that BMI is amplified in certain cancers - we'll have to watch this space!

So BMI potentially could be an oncogene in this mass model. If it is an oncogene, how is it an oncogene? Well, some of the early studies by Maarten van Lohuizen's lab, established that BMI is upstream of the pRb and p53 pathways. Of course, the pRb and p53 pathways are deregulated in practically all, if not all cancers. So how is BMI the upstream of these two pathways? Well, there's a gene locus called Ink4a/Arf which encodes two proteins: p16 being upstream with the Rb pathway, and p19 in mouse, so p14, again, with Arf which is upstream of p53 pathway, and that’s a pathway that's regulating both apoptosis and G0 and G1 proliferative arrest. So this Ink4a/Arf focus is repressed by BMI and you can see, for example, when you overexpress BMI here it'll repress the transcription of p16 or the Ink4a gene, and the protein levels of p16 are downregulated here. So we made an antibody for BMI, and we were able to show that it's direct, so BMI, like other Polycombs is on the Ink4a/Arf locus and directly repressing its transcription. So you can see here the control, not on BMI, not on the cyclin-A2 promoter, it is on the Ink4a promoter here. Actually, if you overexpress BMI you will see more being recruited there, and that correlates with its repression.

My skepticism on BMI being an oncogene, and I've stated that already, but certainly BMI could still be a potential therapeutic target, because what we do know is BMI is required for stem cell self-renewal. So if you knockout BMI what'll happen is it'll upregulate p16 and Arf, and this will lead to apoptosis and cellular senescence. So p16 is a key regulator of cellular senescence, so a permanent proliferative arrest. Similarly so, you can do this in normal hematopoietic stem cells, but you can also knockout BMI in cancer stem cells and what will happen there is essentially the same thing. You will upregulate p16 and Arf, and this will lead to apoptosis and cellular senescence. So this will tell you, of course, that it's required for the repression of the Ink4a/Arf locus both in normal and cancer stem cells. But I'd draw a line and say it doesn't tell you that BMI is actually the initial driver of the cancer itself. Of course, there would be a risk if you were to inhibit BMI, that maybe you could also affect normal stem cells. But it certainly is a link between Polycombs, and one of the earlier links between Polycombs and cancer.

Initially, EZH2 is also linked with cancer, and we'll go through some of that work. So, very dramatically, EZH2 was shown to be highly expressed in prostate cancers, breast cancers and many other types of cancers. You can see here, for example, staining the EZH2 protein on benign prostate cancer or malignant, and as you can see it's pretty self-evident that EZH2 is very highly expressed in the bad form of the cancer, the malignant type.

You can do survival analysis, and this is the Kaplan-Meier curve here and the take-home point is that if you have low levels of EZH2, the patients are surviving after surgery. Most of the patients are surviving, whereas if they have moderate or strong levels of EZH2, more patients are unfortunately dying of the cancer. So it tells you that not only is EZH2 high in cancer, but the cancer that has high EZH2 are worse; the patients will more likely die.

Does that mean that EZH2 is causing the cancer? Well, possibly not, I mean, so we established that EZH2 is an E2F regulator gene, so all of these experiments here are northern blots, so we can activate each with one, two and three. You can see here, like the well-known E2F target gene cyclin E; EZH2 is strongly upregulated when you activate E2F-1, 2 or 3.

What are E2F-1, 2 and 3? They're actually key transcription factors that regulate cell cycle genes, and genes that you need to get into S-phase and so on. Eventually, if you're an E2F target gene it means you will be highly expressed in highly proliferative cells, both normal or cancer. So just to tease this out a little bit more, we do an experiment again in normal cells here. These are cells that are in a state called G0 or quiescence that are non-growing, and you can see there's low levels of EZH2-RNA, and low levels of, or no CCNA2 RNA. So then you add serum to these quiescent cells, allow them to go into the cell cycle and you see them going into S-phase after 24 hours. You can see the EZH2, but also EED and we subsequently showed Suz12. So core members of the PRC2 complex are both E2F regulated and their expression status changes when you go from non-growing to growing cells. There's these types of results that have been shown from many labs around the world in the last ten years or so. But that raises a doubt, so you have to ask the question, well - and this is the question we were asking is: EZH2, okay, it's high in a lot of cancers, but that's just the side-effect of the cancers that they grow a lot faster than normal cells; and we'll come back to that in a second.

What happens if you knockout EZH2 or other PRC2 complex members? Well, the simple answer is both normal fibroblasts, but also cancer cell lines they stop growing. Now, there are some cancer cell lines that don't stop growing when you get rid of EZH2, but generally when you knockout PRC2 members you have a proliferative block. You knock them out for a long time, up one to two weeks, you will also get an upregulation of Ink4a p16, and you can get a lot of Polycombs. You can see here when you get rid of, for example, getting rid of Suz12 or EZH2, you'll have loss of EZH2 when you have chromatin immunoprecipitation. You'll have loss of the whole PRC2 complex on the locus, and you will have cellular senescence when you score for that.

But the question comes back to, well, is EZH2 a mover or a marker in cancer? Based on everything I've shown you so far, you would have to say it's more likely a marker of cancer. In other words, it is high like Kr67; the most famous marker of cancer, a gene that is an E2F target gene and just simply have high levels in highly proliferative cells, both normal and cancer. This is the way we were thinking about things about four years ago, and we were thinking that, well, elevated levels of EZH2 probably wouldn't lead to elevated levels of K27me3, or increased Polycomb group function. Because, of course, you need its friends Suz12 and EED, these proteins form multi-protein complexes for a reason, and you would need equally to upregulate Suz12 and EED to comparable amounts. So that's where we were until about four years ago, and then, of course, the revolution in next-generation sequencing came about. Well, we'll return to that in a slide or two.

Just to point out that there was actually a lot of people thinking that maybe because p16 is a Polycomb target gene, and because p16 is epigenetically found in cancer, there is still an open hypothesis that maybe high levels of EZH2 or PRC2 members in cancer, or even deregulated in the BMI and so on, they could lead to altered DNA methylation on a tumor-suppressor gene, such as p16. You can see here that there's many ways to inactivate a tumor-suppressor in cancer, you can mutate the gene itself, you could just delete it, or you can epigenetically silence it here.

So supporting the idea that Polycombs could contribute to DNA methylation deregulation in cancer, well, these here, in this circle here are the Polycomb target genes. Essentially, what they did is they looked at these genes and asked the question: Are they more likely to be epigenetically silenced in cancer? The answer is yes, in fact, if you were a Polycomb target gene you are 12 times more likely to become epigenetically silenced in cancer. That's quite a striking finding, however, the question I would ask is, well, was that caused by deregulation of Polycombs? The answer is I don't know, and certainly we need to look into that still. It could be just a coincidence of the fact that Polycombs like a thing called CpG islands. These CpG islands are unmethylated, normally CGs generally around the promoters of genes. Polycombs like those, they bind to those, and I won't go into the specifics of how, but these CpG islands are not normally methylated, but they become DNA methylated in cancer. One possibility is that these two events: the Polycomb and the DNA methylation are unrelated and just coincidental. Again, it's an open story.

So far it's a lot of hand waving, and what's the actual evidence that Polycombs are deregulated in cancer? Well, the evidence is really, really strong and this has really come about in the last five years. Actually, just in the last few weeks we've had some even more advances on this. So it's a bit of an explosion, and I'll just try and take you through all these very exciting findings now in the next few slides. This is a slide I made, which is adapted from an excellent paper published in Science by Bert Vogelstein. So they have a supplementary table where they list all, or about 130 genes that are statistically mutated in all human cancers. Of course, some of them are tumor-suppressors, some of them are oncogenes. Unsurprisingly, we have our famous genes like pRB, p53, p16 that I mentioned earlier on, encoded by CDKN2A and so on, apoptosis genes, DNA damage genes. But probably the most striking finding from these next-generation sequencing studies would be mutation of chromatin regulators, enzymes such as EZH2, and, believe it or not, histone proteins themselves. So we're going to come back to that and take you through it probably more from a Polycomb perspective.

So showing you again this structure of the EZH2 protein, and here is the protein itself. We discovered - well, I didn't discover it - many labs discovered mutations in the SET domain. So, again, this is a histone methyltransferase domain of EZH2, and you can see these mutations here. So, importantly, these are recurrent mutations which means they occur in many patients with B-cell lymphomas. Analogous to conserved mutations in the Raf oncogene, so we can pretty much say these are activating, or we can say these are mutations that caused the cancer because they recur in different patients. So, immediately, with this slide what we can say is, well, EZH2 is an oncogene in B-cell lymphomas.

The next question is then, how is an oncogene, what does it do? This is where we go back to our histone methyltransferase assays. What we know about EZH2, one detail I didn't specify earlier on is that both EZH2 when it's in the PRC2 complex, it likes a substrate which is unmethylated. So it likes histone H3me0, so that means histone H3 unmethylated at lysine 27, as you can see here. It likes that substrate, it's not as fond of the substrate for this monomethylated lysine 27, and it doesn't like histone H3 when it's dimethylated at K27. The mutant forms of EZH2 have either gained a function or changed a function, so this mutant here is, again of function, so it does what the wild-type can do, but it gains the ability to convert H3K27me2 into H3K27me3. So that's a gain of function, and then change of function, this particular mutation the EZH2 mutant cannot, can no longer convert unmethylated histone H3 into histone H3K27me3, but it can use the substrate H3K27me2. But, remember, these mutants exist in the heterozygous base, so they will have both the wild-type and the mutant at the same time. So because it has the wild-type EZH2 present, the cancer cells with this mutation in heterozygous state will be able to use the wild-type to convert the unmethylated histone H3.

What's the net effect? Well, if you overexpress EZH2 in cells, it doesn't change the profile of H3K27me3, but any of these mutants here either change or gain function, mutations will increase the levels of K27me3. As I outlined with the biochemistry, they will reduce the K27me2 because they are, of course, able to increase or they have the ability to turn me2 into me3.

So having spoken about that, are there other ways to activate the activity of PRC2 in cancer? It turns out there's at least three different ways, possibly four. Here's another example of the wonderful work that's coming out of next-generation sequencing, so the gene UTX it encodes a histone demethylated-specific for lysine 27. It removes K27me3, so, in fact, this UTX gene encodes the protein which is the exact opposite to EZH2. So this gene is deleted in many cancers and the net effect will be that you, presumably, increase K27me3. Although that hasn't been absolutely entirely looked into yet, but you see in the model here we can either have gain of function or change of function in the EZH2 mutations. We can have UTX deletions and mutations, EZH2 overexpression, or what I think are probably more likely to be important results as opposed to EZH2 overexpression, is the overexpression of substoichiometric components of the PRC2 complex, such as PHF19 and PCL3.

Here's a very big surprise, it turns out that EZH2 can also be a tumor-suppressor, so clearly it's an oncogene, but it also can be a tumor-suppressor. More so, actually, so can EED and Suz12; EED and Suz12 are also deleted and mutated in various cancer types here. So this is very, very recent work just in the last few months coming out in Nature and Nature Genetics, and this is a little bit older. So EZH2 in these studies is shown to be mutated and deleted in myeloid malignancies. There's been some nice mouse genetics on this, so we now know that you can actually phenocopy this in a mouse. You can conditionally delete EZH2 in mouse hematopoietic stem cells, and these will lead to T cell leukemia, so you can see that here.

This is actually even a bigger surprise, so we now know that histone H3 at lysine 27 is mutated in pediatric glioblastomas, so you're changing the lysine to methionine. Just to tell you a little bit about histone H3, so we've essentially three variants of the histone H3 protein in our cells, and these three variants are encoded by multiple different genes. So all of these genes would make exactly the same protein in this case here, and these two genes highlighted in red are mutated.

How does this work? Well, dramatically, if you look at these cancers with the K27M mutations, so you've a western blot for that variant there. What you can see is that it reduces the overall levels of K27me3, and you can even look at the tumors themselves, so this is the cancer with the Y-type histone H3. You can see that there's K27me3 positivity in the cancer cells here, but when you change the K27 to an M the K27me3 is gone. But if I revert to the previous slide you'll that this is remarkable, because just mutating one gene blocks the ability of all the other histone H3 proteins made from all the other genes, to be methylated at lysine 27.

So how could this possibly be? It's very striking and surprising results. So excellent work from David Allis’s group who proposed this model here, is that the K27m itself acts, as you can see here, almost like a sponge in that it binds to the PRC2 complex and thereby blocks it from going on, and methylating other histone H3 tails. So it essentially acts like a dominant negative to block PRC2 function.

What does this all mean in terms of therapy, what can we do about it? Well, just to go back to those cancers that have elevated levels of PRC2 activity, so, again, just to remind you again, those with the gain and change of function mutations here, or with the UTX mutations, or which would lose the histone demethylase, or with overexpression of EZH2 or PCL3. So all these scenarios would lead to elevated levels of K27me3.

How can we treat that in terms of drugs? Well, several companies and labs have come up with various different EZH2 inhibiting drugs, and here's just one example here. You can see, you can add increasing amounts of this drug to cells and you get the specific reduction of K27me3. You can get B cell lymphoma cell lines, put them into mice and you can see here that these lymphomas will grow. But if you treat the mouse with this particular drug, then it specifically inhibits EZH2 you can see the tumor growth has dramatically reduced. Similarly so, you can look at the lifespan of these mice on a Kaplan-Meier curve, and you can see here that the mice will die very quickly, in less than two months. But if you treat with high levels of this drug, or various levels of this drug you can actually keep them alive. So these drugs are actually in clinical trials and there's several companies that are manufacturing and making, and innovating in this case.

So could we use the EZH2 inhibitors for cancers in terms of a wider spectrum of cancers? Well, I'm going to take you back to the idea of trithorax group proteins being the opposite of Polycomb group proteins, so that's earlier slides in the introduction about Drosophila genetics. I'm going to tell you about the SWI/SNF complex, which contains many trithorax group proteins. Dramatically, these complexes - well, they contained many genes that are mutated in cancer. This complex is a very large multi-protein complex, and it's thought to have ATPase activity and it can move nucleosomes around. I've highlighted some of the genes in here that are mutated in many, many different types of human cancer.

If you go back to the Drosophila genetics and look at this a little bit more carefully, when they were looking for genes that would rescue Polycomb defects in Drosophila, they uncovered many trithorax genes and I've highlighted Brahma and Osa here. So if you look down at the bottom here, Osa, the homolog of the Drosophila gene so, for example, ARID1A, ARID1B, Brahma and BRG1, these are the genes that are frequently mutated in human cancers. We've only learnt this in the last three years, so actually a lot of trithorax group genes are mutated in human cancers.

So if we go back to our Drosophila genetics and think about it again, could it be that if you were to lose a trithorax group protein or gene, that you would have a possibility to treat these cancers with inhibitors of Polycomb? So it's going back to these rescue experiments that were done many, many years ago. So, in fact, my prediction is that not just cancers that have elevated levels of EZH2, they would not be the only cancers responsive to EZH2 inhibitors, but also maybe cancers that lose the activity of trithorax in proteins.

So just to wrap up, dramatically, EZH2 has emerged in the last four to five years of being both an oncogene and a tumor suppressor. The deregulation of K27me3 occurs in many, many different cancers, it can come about by mutation of EZH2, or other PRC2 components such as Suz12 or EED. It also can come about by mutation of UTX, which removes K27me3, or it can come about by mutation in the histone H3 itself. We already have small molecule inhibitors available to inhibit the activity of EZH2, they're available for lab use but also they're being clinically trialed at the moment.

We have, or I do know that there are many companies innovating in terms of making inhibitors of BAF complex members, and I'd just like to say in the fourth point here that perhaps inhibitors of EZH2 or PRC2 complex will be effective not only in cancers with elevated PRC2, but also in cancers that lose BAF complex function. The complete reverse, could it be that inhibitors of BAF complex members could be effective at treating those cancers that lose PRC2 function? So I think this is a good example of how Drosophila genetics in the early-90’s and much before has led to really important concept paradigms that are really important, and how we're going to frame cancer therapy in the next few years. However, lots of work still needs to be done in this space before we really understand how PRC2 deregulation contributes to cancer. Thank you all very much, that's the end of my talk. I'll be back again in about 5 to 10 min to answer some of your questions, so I'll now hand you over to Miriam. Thank you.

MF:        Thank you Adrian for such an interesting and comprehensive presentation on Polycombs. I will now talk to you about some Abcam products that might be interesting. We have just launched a high-sensitivity ChIP kit that has been specifically designed to be used when there is a limited amount of sample available to perform chromatin immunoprecipitation, for example, when working with patient material, transgenic mice or stem cells. With the high-sensitivity ChIP kit, you can obtain enrichment of your sequence of interest by using as little as 2,000 cells of half microgram, half milligram of tissue per reaction. You can get your results in only five hours, so the experiment can be done in one working day, and the eluted DNA can be used straightaway in sequencing, microarray or qPCR. Our high-sensitivity ChIP kit is part of our range of high-sensitivity products designed to be used when there is a limited amount of starting material. If you want to create a DNA library with a limited amount of DNA, our high-sensitivity DNA library preparation kit for Illumina® sequencing allows you to create a DNA library for only 0.2 ng of DNA.

Our ChIP-seq high-sensitivity kit combines the benefit of the two kits I've just mentioned. This kit is designed to carry out a successful ChIP-seq starting directly from as little as ten to the five cells. We also have a range of MeDIP and hMeDIP kits for detecting a specific cytosine modification. Immunoprecipitation of 5-mC or 5-hmC residues can be ready in less than four hours. Chromatin preparation kits are an essential tool to prepare the samples that will be used later on to perform the detection and quantification of DNA methylation or histone modifications. Our chromatin extraction kit has been optimized to prepare chromatin for ChIP experiments in less than one hour. Bisulfite sequencing is the use of bisulfite treatment to determine the pattern of methylation. To help you take your methylation studies further, we now offer a couple of products to prepare post-bisulfite DNA library through Illumina®.

With our post-bisulfite DNA library and preparation kit you can prepare a library using pre-treated DNA in only five hours. If you haven't treated your DNA, then we recommend our bisulfite-seq high-sensitivity kit, which has all the reagents to perform bisulfite modification followed by the DNA library preparation steps. Your library is modified and ready in only six hours. Adrian has mentioned the role of DNA methyltransferases and demethylases as potential drugs for cancer. We also have available TET and DNMT activity quantification kits, these are colorimetric ELISA-like assays that can detect the activity of a specific enzyme by only using a half microgram of nuclear extract in only four hours. Without further delay, I will pass the microphone back to Adrian who is ready to answer some of your questions.

AB:         Thank you. So I've got several questions there and I'll start with this one, so the question is: Wouldn't PRC2 or BAF complex inhibitors have side-effects? Of course they quite possibly would, and, of course, as I showed you, PRC2 function - and I didn't show you, but the BAF complex function, of course, they're both important in developmental biology, and also for homeostasis. So inhibiting important enzymes like this could clearly have side-effects, so it's an excellent question. My answer would be, well, of course, like anything we need to do clinical trials, safety trials and they're ongoing, at least for EZH2 inhibitors, I can tell you that. There is a possibility that the cancers will be more sensitive to, for example, an EZH2 inhibitor than, for example, a normal cell. So I guess part of the trick will be to find the sufficient dose to negatively affect the cancer cells, but hopefully not to affect the normal cells.

This is a paradigm that we've known for a long time that cancers can become so-called oncogene-addicted, so one would imagine, for example, the B cell lymphomas with their gain of function or change of function EZH2 mutations that have elevated K27me3, you would imagine that they would be particularly sensitive to sustained EZH2 activity, to proliferate and continue to go. One could imagine ongoing need to see this, if this is the case, but one could imagine or hope that these cancers will be particularly sensitive to EZH2 drugs.

Another question: Are other PRC2 complex members deregulated in cancer? Yes, so, obviously, EZH2 is both an oncogene and a tumor-suppressor, and we talked about that today. But often the detail possibly there was the recent discovery that EED and Suz12 are deleted and mutated in cancers so, presumably, tumor-suppressors. We also know that the PHF-1 or PCL-1 Polycomb, so its name is Polycomb like-protein one, we know that that's translocated in endometrial cancers. There are some evidences of other Polycombs also deregulated in other cancers.

Another question; this is more of a mechanistic question: How are Polycombs recruited to target genes? Funnily enough, we don't exactly know the answers to that question. My lab showed in 2012 that Polycomb-like proteins are certainly required to recruit the PRC2 complex to target genes. But they're not sufficient and so, in other words, the PRC2 if you - the PRC2 complex doesn't go everywhere across the genome, it's certainly targeted to these developmental genes. We don't know exactly how it gets there, and some of the ongoing ideas are that they require non-coding RNAs to recruit them in there, or DNA binding transcription factors. Another idea is that they might even need a block of transcription, and that's just the default mechanism for which they go in then to repress these genes. Another idea is actually the K119 ubiquitination on histone H2A that that - we know now that that has an affinity to the PRC2 complex, recent work has shown that. So that might be a way that, actually, it's the other way around that non-canonical PRC1 complex members may be the means to personally mediate K119 ubiquitination, and then that brings in PRC2 complex. Then, obviously, the K27me3 and that leads to the canonical PRC1 complex getting in there.

There's a lot of detail I gave you there, but the take-home message is that we're still figuring it out and there's lots of things going on. No doubt it is an important question.

This will be my last question that I'll answer, so just we're running out of time. The question is: How is EZH2 - how could it possibly be both an oncogene and a tumor-suppressor? I guess the question is from a biological perspective, how could that be? The truth is we don't know, as I showed you, you can knockout EZH2 in the hematopoietic stem cells and maybe that study is the best study so far, because it told us this leads to T cell leukemia. So, certainly, there's a tumor-suppressor in that context, and I think the - I've underlined the word 'context'. My idea of it is, is that the loss or the gain of EZH2 function will in the end lead to the same thing, and that is the inability to differentiate. So you'll stay in a proliferating early progenitor-like state, you won't terminally differentiate and that's my hypothesis, best hypothesis at the moment. Remember, as I showed you, Polycombs during lineage specifications, they're both silencing stem cell genes in differentiated cells, and are silencing differentiation genes in stem cells.

I guess, depending on when you gain or lose EZH2 function, you could have a deregulation of differentiation, and we know from many different cancer types that the inability to differentiate is a key feature of cancer. I'd like to wrap up there, and thanks for all the interest and the questions, and I'll pass you back to Vicky. Thank you very much.

Thank you, Adrian and Miriam for presenting today. If you have any questions about what has been discussed in this webinar or have any technical enquiries, our scientific support team will be very happy to help you and they can be contacted at We hope you have found this webinar informative and useful to your work. We look forward to welcoming you to another webinar in the future. Thank you again for attending, and good luck with your research.