Bench philosophy: Epigenetic editing
Overwriting Nature's Errors
by Steven Buckingham, Labtimes 06/2016
Epigenetics works at a coarse-grained level, affecting broad stretches of the genome. Surprisingly, it turns out that the new discipline of epigenetic editing could provide a better alternative to genome editing using CRISPR.
Targeted rewriting of epigenetic marks, e.g. histone modifications, is at the heart of epigenetic editing. Photo: NHGRI
Epigenetics is the study of inheritable changes in gene expression and, therefore, of phenotypes, without changes in the DNA sequence. There has been huge research interest aimed at understanding how such gene expression can be altered and inherited in this way. Although many fundamental questions remain to be answered, this new line of research has begun to unravel how epigenetic changes are brought about.
There are two ways you can mark the genome epigenetically. One way of doing it is to do something to the histones. Remember that the DNA in the nucleus is wrapped in a highly ordered way around a number of histones. How else would you pack a two-metre-long sugar molecule into a 10 μm-wide nucleus? In humans, there is a small family of histones, and they can be modified in two basic ways: by methylation or by acetylation. And marking the histones associated with a regulatory element of a gene can have one of two effects: it can either up regulate its expression or down regulate it.
Another way you can mark a gene for regulation is to methylate the CpG residues (a cytosine sitting next to a guanosine). In fact, if you read through the genome, you will come across a lot of stretches, where there is an abundance of these CpGs. These are called “CpG islands” and are the site of intense epigenetic activity. As with histone modification, methylating the DNA directly can cause either up- or down-regulation of gene expression.
In vivo, marking a gene or histone is inherited. During DNA replication, regulatory mechanisms make sure that the epigenetic marks are copied to the newly-synthesised strands, so ensuring that the daughter cells inherit the marks. Epigenetic marks are also reversible, and the complex activity of marking and unmarking (methylation/demethylation, acetylation/deacetylation) is under precise control.
Just how this precision control of epigenetic marking is brought about is not exactly clear, which is perhaps one reason why it has occurred to so few researchers that it can be exploited in controlling gene expression at the bench or in the clinic. After all, if you don’t know how something is controlled, you can’t imagine how you would go about hijacking those control mechanisms to your own ends.
So, given that we know so little about how epigenetics really works, how can we possibly do epigenetic editing? The answer is simple – use the same toolbox as we use for DNA editing.
Until recently, epigenetic editing was done using Zinc Finger Proteins (ZFPs) and Transcription-Activator-Like Effectors (TALEs) fused to an enzyme that performs methylation or acetylation. The TALE or ZFP part provides the specificity of targeting by binding to the DNA sequence specified by the protein’s DNA-binding domain. In the case of ZFPs, each ZFP protein recognises a unique three base-pair sequence, whereas a single TALE recognises a single base pair.
So, to modify a gene’s expression, it is a matter of fusing six ZFPs or 18 TALEs to cover 18 base pairs on the genome – a length almost guaranteed statistically to be unique. Once you have got the specificity of your DNA-binding moiety, then you need to glue on an effector, such as a methyl transferase. Your artificial transcription factor will bind to its recognition site, while the effector flaps about, marking any histone nearby.
Sure, making artificial transcription factors in this way is not difficult but it is laborious and time-consuming. You have to make one for every sequence you want to target, for instance. But if we drop ZFPs and TALEs for CRISPR, it’s a whole different story. Recall how CRISPR works – you have a Cas9 that chews up DNA, guided to a specific site by a guide RNA (gRNA), whose sequence matches the target sequence (yes, I know, there is slightly more to it than that but the principle is right). So, if you want to cut out a gene, you just transfect cells with Cas9 along with a construct for your gRNA. And if you decide you want to hit a different gene, all you have to do is to design a different gRNA, which, thanks to online design and ordering, is a piece of cake. Now, that’s a lot easier than having to make new, fused protein constructs, wouldn’t you agree?
So much for CRISPR. But how does this help with epigenetic editing? First, you knock out the catalytic activity of the Cas9 – a simple point mutation will do that. Next, you fuse the catalytically-dead Cas9 with an effector enzyme.
Now, all that remains to be done is to generate the gRNA and you can use your fused Cas9-methyltransferase to target any gene/promoter/enhancer/repressor you like – just change the gRNA.
Sounds great but does it work? The answer appears to be an unguarded “yes”. Surprisingly, modifying epigenetically actually turns out to be more efficient and more effective than editing the genome with CRISPR, or knocking a gene down with RNAi.
In a recent review (Nature Methods 13: 127), Pratiksha Thakore from Charles Gerbach's group at the Duke University, USA, summarised the findings of five papers as reporting “near perfect specificity”. Interestingly, when you look for off-target effects using Chromatin Immunoprecipitation (ChIP), it seems that there is some binding of the construct to off-target regions but they aren’t functionally activated.
Even the genes immediately adjacent to the target appear unaffected. You can also get better knockdown than with other methods, although you may find you have to hit the target with a number of effectors simultaneously.
It is important to remember that epigenetic editing is, unlike genome editing methods, bidirectional – you can get up regulation as well as down regulation, although it is harder to achieve. And don’t forget the heritability feature that comes free with it. Oh, and another thing: remember it is reversible, so you can reverse the effect at will.
But like the electric car that runs for 500 miles on a half-hour charge, the promises of epigenetic editing seemed a bit too good to be true.
Marianne Rots adopts epigenetic editing to mimic and reverse cancer-associated epigenetic mutations. Photo: University of Groningen
So Lab Times spoke to Marianne Rots, Professor of Molecular Epigenetics at the University of Groningen. Rots has been an ardent promoter (pardon the pun) on the promise of epigenetic editing for years.
First of all, Lab Times asked her if the claims were true.
Is epigenetic editing as specific as the salesman says?
Rots: There are two ways I look at the issue of specificity. First, there are times when you actually may not want too much specificity. When you target a gene, to get the best effectiveness you don’t want to hit just one genomic region. You want to hit a number of CpGs, not just one. So, having some spread around the target actually is probably an advantage. On the other hand, you don’t want a cluster of genes being down regulated; you want just the gene of interest. And with CRISPR/Cas you can get that single gene specificity.
So, bringing CRISPR/Cas into the epigenetic field has made a big difference? Is that what suddenly makes epigenetic editing exciting?
Rots: In the old days, we used to rely on zinc finger proteins and TALEs to make a designer DNA-binding domain. It is really laborious to do it that way and also it is not very specific. With CRISPR/Cas, it is definitely easier and more specific. But still, with every new target, you still have to check every combination of gRNA and effector domain to make sure if the effect is really localised. All the same, the promise offered by CRISPR/Cas and epigenetic editing is really very high.
Do you see epigenetic editing getting into the clinic?
Rots: Epigenetic editing therapy is actually already well accepted. In fact, there are FDA-approved therapies in clinic trial now, based on inhibitory epigenetic enzymes for certain haematological cancers. There are therapies that affect the whole genome –but that is not what you really want. Most diseases are associated with certain specific genes, not the whole genome. Genes that are either shouting too loudly or perhaps are asleep when they should be active. What we need to do, is to quiet the noisy genes down or wake the sleeping ones up. So, gene targeting with very precise interference is arousing a lot of clinical interest, and my experience tells me that it is realistic and within our reach. The first clinical trials of a therapy using CRISPR/Cas will take place in 2017. It is a safe and attractive approach – what is to stop it?
Is epigenetic editing as easy to use as they say? To use it, do you need to be a specialist lab or can anybody with basic molecular biology do it?
Rots: You know, the techniques are actually really very simple. Making the gRNA is straightforward and there is plenty of software to help you do it. The rules for making gRNA are better understood these days and you can just buy ready-optimised effectors, fused with cas9 from commercial suppliers. They use fully-validated effector domains. All you have to be able to do, is a transfection and the read-out!
I remember similar promises about RNAi...
Rots: Sure, but epigenetics is offering more than RNAi. Epigenetics is about reprogramming the genome, whereas with RNAi the construct has to be continuously present to lower the RNA in cell. OK, I’ll admit that we are not quite “there” yet for epigenetic editing but we can’t ignore the promise that it can result in sustained gene repression or activation. We are making rapid progress and more will come as we get to understand epigenetics better.
I get the impression that a lot hinges on getting the right effector and that no-one really understands the rules behind what a given effector does. Can you predict beforehand, what effector is right for the task at hand?
Rots: Yes, in general if you want to down regulate a gene, you know what effector to use. Same with up-regulation. There are exceptions but, in general, it is easy to know, which class of effector to use for your particular application. What is more difficult to predict is, which effector domain will turn out to be potent enough for a given heterochromatin context and which one to use, to keep the epigenome being reprogrammed over the long term.
Are epigenetic edits really completely inheritable?
Rots: In theory, yes. All daughter cells should inherit the marks, assuming the maintenance machinery is in place. When DNA is copied, DNA methyltransferases search for methylation and when they find it on one strand, they put methylation marks on other strands as well.
That’s the theory. But in practice?
Rots: There are actually only two papers that have looked at the sustainability. One found methylation was inherited, the other not – so it seems to be context dependent. What we need at this point, is systematic research into heritability in different heterochromatin contexts.
CRISPR/Cas was chosen by the journal, Science, as the Method of the Year in 2015 and it clearly is having a huge impact. Will epigenetic editing be as hot as genome editing?
Rots: Yes, although of course you are not asking the right person! But I can honestly say that interest in epigenetic editing is growing fast. When I first published on editing in 2012, and even before I got my first grant in 2006/7 on this topic, it was a hard job to convince people of its importance. People thought that epigenetics is not instructive and that it is only a cell’s way to remember its expression profile. There was also the view that if a gene is not expressed, it is because it is not accessible. I had to show that if the reason the gene is not expressed is because of its heterochromatin context, then we could actually get it to re-express. That was not thought possible at the time. Now, there are strong labs publishing big impact papers in journals like Nature Biotechnology.
If any of our readers wants to have a go, what hints and tips can you give them?
Rots: Just have a go! The protocols are all settled. gRNA is so cheap, you should just order 20 or so and see which ones work the best. I was going to say it is just trial and error but, truly, it is so successful that error doesn't even enter into it!
Last Changed: 28.11.2016