Okay, so now, we're going to start thinking about these epigenetic aberrations that we see in cancer and we're going to delve into detail about DNA methylation. This is where we really know the most about the epigenetic mistakes that happen. We'll start off, so, by mentioning that there are two types of DNA methylation abnormalities and I really showed you these in a picture in the last lecture. First of all, there's DNA hypermethylation that is locus specific, and secondly, there's genome-wide hypomethylation. So to show you this in another way, we went through this very briefly in the first week, when we thought about DNA methylation. So in a normal cell, what you see is a hypomethylated CPG island, so you don't see methylation in general at CPG islands. But you find that the genome in general is methylated in their repetitive elements through the intergenic regions and indeed even in the introns of genes as shown here. In contrast, if you look in general at a cancer cell, now the CpG islands are more likely to be methylated. They're not all methylated, but they're more likely to be methylated than in a normal cell. And the rest of the genome in general, including the repetitive elements in these intergenic elements, and indeed, the introns, are hypomethylated. And so we see this kind of swap or reciprocal cross, swap in terms of where the DNA methylation is found. So we're going to start out by thinking about this methylation at CpG islands that's found in cancer. So this is, tends to be called CpG island hypermethylation, or CGI hypermethylation. And in general when you look at where these CpG islands are found, they're in the promoters of tumour suppressor genes. So this is because if you think about the way that a cancer cell is able to inactivate tumour suppressor genes, it could do so genetically, or it could happen epigenetically by silencing the tumour suppressor gene by locking in a silent inactive state. And since DNA methylation is mitotically heritable, this means that this is a very good locking down mechanism in terms of epigenetic silencing. So, we now know that actually DNA methylation can be the second hit, if you like, or one of the hits when in terms of the Knudson hypothesis. So Knudson proposed that you would have to have multiple hits in order for cancer to occur. And this is because if you just hit one tumour suppressor and take out one tumour suppressor, or perhaps one copy of the two alleles of a tumour suppressor, this is not sufficient to cause tumourogenesis. But rather that you need to either take out both copies of the tumour suppressor. Or both copies of a tumour suppressor and in fact multiple other hits that would need to occur. For example, you'd need to also activate an oncogene. So you need to have many insults that will mean that the, a cell will end up rather than dying, will end up actually going on and forming a tumour. So it's recognised that DNA methylation and, consequent epigenetic silencing of tumour suppressor genes can be, indeed, one of these hits. So we know this CpG island hypermethylation and associated gene silencing occurs really frequently in tumours. And it has been found in essentially all tumours that have ever been studied. So this is probably because, as I said, DNA methylation is mitotically heritable. So it's a very effective way of silencing a tumour suppressor gene, and these epimutations can indeed be rapidly selected for. So if you think that actually silencing a tumour suppressor is going to allow that particular cell to have a competitive advantage over the others in the surrounding tissues, then over a period of time, the cells that have this epimutation are more likely to divide more rapidly, or to not die as much, and, therefore, they will take over. And this is indeed how we think cancer's eventuate. So the critical difference between hyper-methylation of a CPG island as a way to inactivate a tumour-suppressor gene, in comparison to a genetic mutation is that these epimutations, the DNA hypermethylation, is reversible. And this is really important because if this, hypermethylation, this DNA methylation, is occurring frequently in cancer and has some consequence for the cell, but it is reversible, this brings up the option of being able to perhaps remove some of that methylation therapeutically, and have some good outcome for the patient. And we'll come back to this in later lectures. So what are the key features then of the hyper-methylation of CpG islands? Well, actually if you look at the how frequently it occurs, tumours suppressor hypermethalyation in general is more frequent than genetic mutation. This isn't true for every tumour, but as a globalised rule it is extremely frequent and can be more frequent than mutations. We know that hypermethylation of particular CpG island or CGIs varies by tumour type. So while one tumour or one tumour type may hypermethylate one set of CpG Islands that belong to tumour suppressor genes, a different tumour type might hypermethylate a different set of tumour suppressor gene promoters, or CPG islands. And this has been able to be discovered because of new technology, which has enabled us to look at DNA methylation spread throughout the genome, so genome wide. And by doing this, they could really compare and have a look and say well overall what's the level? So now we're not just going to look at one gene or two genes or a small handful of genes. Let's look at all of the genes and see what we can find out. And what you find is that the DNA methylation profiles are unique to each of the particular tumour sub-types. We also know that CPGI methylation as I mentioned before, progresses with time. And so I showed you this picture that's shown at the bottom again here. And so this, over time as the cancer progresses we get increased methylation of CpG islands that are associated with tumour suppressors. So, this can be very useful diagnostically and prognostically. However, it's also interesting to note that as people age, you get this same sort of change in DNA methylation within the genome. And this has really raised the ideas that maybe cancer is just a consequence of ageing. Maybe we're all kind of in a slightly preneoplastic state as we age. And indeed it's known that most people at some point in their life time will suffer from some cancer and that if they very old, if they are very elderly patients that may die at the age of 100 probably have some minor cancer in their body that they weren't aware of. So potentially it is true that these epigenetic mistakes accumulate over your lifetime and it's only some times that they might actually result in cancer. But probably in the end, we may all have one of these mistakes made so that we have some small tumour whether or not it's life threatening or not. So let's think a little bit about the sorts of genes that are found to be hypermethylated. So, from the 90s onwards there have been a slew of publications reporting hypermethylation of tumour suppressors. So, I'm just going to, briefly mention a few of them. So if we think about these single gene examples they really highlight that hypermethylation can do the same job as mutation. So say for example in retinoblastoma, retinoblastoma is just as it sounds a type of cancer of the eye. And it's frequently found to be a hereditary cancer, so you will inherit from your parents one mutation in the retinoblastoma gene, or it could be sporadic. And then, you'll usually find that the other retinoblastoma gene, your second copy, is either mutated or it's lost genetically. But similarly, what can also happen is that the RB gene, this is the gene responsible for this disorder when mutated can also be hypermethylated, and so, it does the same job as if it were mutated. I mentioned last week in the lectures considering transgenerational epigenetic inheritance. We spoke about DNA methylation of the MLH1 locus in colorectal cancer in Lynch syndrome. So here, the MLH1 activity can be taken out genetically, so by mutation, or epigenetically, and I told you about the hypermethylation of the MLH1 CPG island promoter. In that case it was about constitutional epimutation in other words, spread throughout the person. But equally in other case of Lynch syndrome it will just be within the tumour, so it's the second way of being able to get rid of the function of MLH1. The same is true in breast cancer. We know BRCA1 and BRCA2 are responsible for a fairly large proportion of the hereditary breast cancer, and they seem to be mutated in sporadic cases as well. Similarly BRCA1 can be hypermethalated rather than mutated. And finally, MGMT can be hypermethylated, is found to be hypermethylated in glioma. It's a type of brain cancer, but also in colorectal tumours, and, I'll come back to this cause it has relevance to therapy. So, each of these single genes could be used as single biomarkers. So what do I mean by a biomarker? Well, a biomarker is really a way that we can distinguish between cancer cells and normal cells in the very same sample. So that same sample in one case could be, for example, if you had a lesion on your skin that you are worried about could be, a skin cancer, you would have it excised, And then they send it off to the lab and then get back to you to say whether or not they believe this was a skin cancer, or whether is was something benign or something not to worry about. Some of the ways they could do this is by knowing that there are particular genes, particular CpG islands that are hypermethylated in the case of skin cancer, but not in the case of benign normal skin. And so in this way they'd be able to tell tumour from normal skin, but all of the sample would be skin. Similarly in prostate, so you could tell normal prostate from tumour prostate. So was it a normal overgrowth of prostate? Or was it a tumour prostate, tumour cancer, sorry a prostate tumour? And so this is really important to think about because this changes the outcome for the patient. Do they have skin cancer that needs to then go on and being treated or is it just a normal outgrowth which, which you don't need to worry about? You might also be able to distinguish cancer cells from cells that are of a completely different origin. This is really useful if you think about what tissues are most accessible for us to screen. So our blood is probably the easiest samples, one of the easiest samples to take and to be able to screen. And so with biomarkers, if you know that there's a particular gene which is hyper-methylated only in the tumour tissue or only in the DNA from the tumour tissue, then you should be able to detect that as different to what's normally found in the blood. So, we know that some tumour cells passage through the blood. And so, if you could detect those tumour cells, you might be able to say whether or not somebody is either has a cancer or, maybe, you can talk about how the treatment is going. But not only do we find some tumour cells in blood, you can also find some DNA that's free of cells. It's just DNA floating around from the tumour also in the blood. So this cell-free DNA can also be detected in this way. And this is very, very useful, as I said, for monitoring how well tumour clearance is going upon treatment or tumour recurrence is happening after treatment. Finally, you can use biomarkers to identify a specific feature of a cancer. Say for example you know that a particular gene becomes hypermethylated or you see hypermethylation as associated with metastasis, and again this is a useful thing to know clinically. So at the moment the reason that I'm bringing up these biomarkers when we're talking about hypermethylation, is it's because these hypermethylation biomarkers are the ones that are favoured clinically. The reason for that is that first of all the techniques are more able to detect heavy methylation, but also if you think about what we know about CpG islands, CpG island spread throughout the genome in general are hypomethylated. And so what's unusual to find is one that's hypermethylated, so it's much more sensitive to be able to detect something that is hypermethylated in a sea of things that are hypomethylated, so heavily methylated in a sea of things that are really not methylated compared with the reciprocal. So, if we think about one example of how these DNA Methylation biomarkers are found, this is a slide from a research project where they were looking for ovarian cancer biomarkers. So, then began by looking at about 12, over 12,000 markers. So they looked genome wide with particular techniques to be able to do this. And they looked at a large number of ovarian tumours and compared with normal peripheral blood leukocytes, that's what PBL means. First of all they sorted them and they said okay there's a large number, these ones here, that are unmethylated in the blood. They're the ones we want to look at, and then we want to ask again which are hypermetholated in ovarian tumours. So they then re-sorted this information and now they've got at the very top up here, they've got those that are still hypometh, less, not un-methylated or very low methylated in terms of this key in blood, but are hyper-methylated and are red in ovarian tumours. So they've expanded that out here in Panel C. And you can see at the very top they have, have found 15 genes so they think are potential biomarkers that are heavily methylated in ovarian tumours but not methylated in blood. And then we follow up studies they were able to go through each of these 15 genes that are hypermethylated based on these tumours in the ovarian, in the ovarian tumours, and go through them one by one and use the appropriate techniques to validate these findings with further, looking in more tumours and looking in more peripheral blood samples. And they found that from all of these, just one of these is a useful biomarker. So this is the approach that can be used to take an un-biased look at which genes might be a useful biomarker in ovarian tumours. So to recap then on what I've said about CpG island hypermethylation and its usability when we're talking about single biomarkers. It's useful both for the diagnosis of disease, and prognosis, and we've discussed these sorts of things already. So, for example, GSTP1 hypermethylation is found in prostate cancers, but not in benign prostate growths. So, if you found a sample that had hypermethylation of GSTP1 then this would indicate it was more likely be a malignant than something that was not so sinister. For prognosis, we know for example that hypermethylation of this cluster of micro RNA genes miR-34BC cluster, is associated with metastasis. And this tells you about the patient outcome, and that's why it's a prognostic marker. If somebody has a metastatic disease then clearly it's going to have a worse outcome than somebody that does not. But finally, these single biomarkers can also be used to inform upon how to treat a patient, the best course of treatment for a cancer patient. So the example here you'll practice, one of the best examples is MGMT hypermethylation. So, if the MGMT gene is found to have hypermethylated, and therefore the MGMT protein is not made, then they suggest that these patients will actually perform well with temozolomide treatment. And that's because temozolomide is an alkylating agent. And the way that it actually brings about death of cancer cells is that it adds acyl or methyl residues to bases in the DNA. And these are then recognised as DNA damage, and what ensues is cell death. However MGMT normally, what it norm, what its normal function is, is repair these alkylated guanine residues. So if MGMT is being expressed in the cancer cells they can repair that DNA damage and they're less likely to die. However if MGMT is hypermethylated, then what will happen is that the DNA repair will not ensue and so you end up with cell death. And so this is why MGMT hypermethylation actually is a good prognostic indicator, if you use temozolomide treatment, and therefore it has this third function, we can use this biomarker to inform upon treatment. So next week what we will think about is how we can use not just single biomarker, single hypermethylation biomarkers, but rather how you can use sets of biomarkers or panels of biomarkers in a similar way and their usability in the clinic.