So, hi everybody.

What we are going to do today is to have a discussion with Doctor Andrew Keniry

about DNA demethylation and also reprogramming.

So, these were two topics that we've covered within the course but,

not in great detail yet and yet, they're very active areas of research and

Andrew knows an awful lot about both of these areas, given where he did his PhD.

So, Andrew, active demethylation was first known to occur because

people saw there were periods in development when you would lose

methylation at a really rapid rate without having any cell division.

And so I guess it was striking for

quite a long time that there must be something going on,

there must be an enzymatic process going on.

>> Sure. That's right.

So in the one cell embryo, we can see both potential mechanisms for

the loss of DNA methylation because we have both the male and

female pronucleus in that one cell

and soon after fertilization we see

both of those nuclei staining very brightly for DNA methylation.

And then, very quickly, before any cell division in the male pronucleus,

we see a very rapid loss of DNA methylation, staining for DNA methylation

and so, this means that there has been an active process of DNA demethylation.

Whereas in the female pronucleus, it happens more slowly

over multiple rounds of division,

and essentially, the DNA methylation is being lost passively through

multiple rounds of cell division,

and essentially just being diluted through cell division.

>> Right, so I guess that's because DNMT1 is not doing it's job,

it's not maintaining the methylation, and in fact DNMT1 is excluded from

the nucleus for some of that critical period, during early development.

>> That's exactly right.

So if DNAMT1 is excluded from the nucleus the mechanism is not there to

maintain DNA methylation through cell division and

you just get a dilution of the DNA methylation through time.

>> So, what's always struck me is that it seems like even from the early 90s,

but certainly at least from about 2000,

there was very strong evidence that we had active demethylation

but it took quite a long time before the enzymes,

the TET enzymes that were involved in the active process were discovered.

So I guess, why did it end up taking so long?

Do you have an opinion as to why it took

a relatively long amount of time to find these guys?

>> Yeah, well I think there's a lot of reasons for why that happened

and I guess the first one is because the stages of development where

DNA demethylation happens are quite hard to study, so

we'd mentioned the one cell embryo and

the other place where it happens naturally is in the primordial germ cells.

And so each of the cell types are rare and don't provide lot of material to study

technically in the lab so- >> So

biochemistry is a bit impossible then?

>> Exactly. >> Because you've only got a few cells

>>Biochemistry is very hard.

>> Yep

>> And then the other problem was well, we've had

recent advances in the techniques so with advent of next gen sequencing now,

we're able to look more closely for more subtle differences

and so this has allowed us to see enzymes that are perhaps acting more subtly.

And then the major reason for why it took so

long to find it is we haven't yet a DNA demethylase.

The methyl group is attached to the cytosine through a carbon bond,

covalently carbon bond, which is very, very hard to cleave.

>> So, you've then got a carbon-carbon bond, because the methyl group is a CH3,

so you have carbon-carbon linkage which is incredibly stable.

>> Exactly, exactly

and we still have not found an enzyme that is capable of cleaving that bond

and perhaps we never will because it doesn't exist.

Instead we found enzymes that either degrade the methyl group from the carbon

or degrade the base, which then feeds into the base excision repair pathway and which

will flip out the methylated cytosine and replace it with an unmethylated cytosine.

>> Right, so the TET enzymes, they end up performing an oxidation reaction.

>> Exactly.

>> So what are the intermediates in that reaction?

>> So first of all,

the TET enzymes will oxidize the methyl group to a hydroxymethyl group

and from the hydroxymethyl to a formal methyl cytosine,

and from that to a carboxymethyl-cytosine,

and a carboxymethyl-cytosine base is able to feed into the base excision repair

pathway then, and can be replaced with a completely unmethylated base.

But potentially it's possible a de-carboxilation reactions are technically

possible, feasible, but we haven't found an enzyme that will directly do that yet.

>> Okay.

So the other pathway that leads into base excision repair is,

as you said, by degradation.

So, you have a deamination for AID, the other enzyme that can perform some, or

lead into base excision repair so that, in the end, you have a demethylation event.

So it was found a little bit earlier.

Do you think that was just by chance,

or it was- >> [COUGH] Well, people had the theory

that the base excision repair pathway would be involved for quite a long time.

Whereas they went looking for oxidation reactions,

so people weren't actively looking for

genes that could modify the base so that it would feed into

the base excision repair pathway and deaminases were prime candidates for it.

So people were specifically looking for

AID in the places where they assumed it would work.

>> Well, that makes sense.

So, I guess the hydroxymethylation, as you said there are been advances in our

ability to look for, with using next generation technology.

So we actually cannot only perform bisulphite sequencing which we

discussed in other lectures.

But also you can also look to find hydroxymethylation in somewhat

more indirect measures I guess in a way, chemically.

>> Sure.

>> We weren't going to the details of that

but I guess now what seems to be happening is that people are now able to perform

this more widely.

And they're looking for

hydroxymethylation throughout the genome at different stages of development

and using this as an indication of where active demethylation occurs.

And it's quite surprising to me that it's not only during these stages we've

already mentioned

so preimplantation development or primordial germ cell development, but

it also seems to be quite common in the brain and most recently,

also, shown in various immune subsets, so B cells and T cell development.

>> Sure.

Sure, yes, so that's a really interesting question.

It kind of raises the possibility that, perhaps, hydroxymethylcytosine

is an epigenetic mark in its own right, with its own function and

well, we already know that there are proteins that specifically bind

hydroxymethylcytosine in preference to methylcytosine or unmethalated cytosine,

so this really suggests that it may really operate in its own pathways.

>> So, it's not perhaps well,

that we're using it at this stage to say that it's active demethylation, and

of course it is part of the active demethylation process by the TET enzymes

maybe it's meaning something else too.

So it's interesting that it's been found in the brain

and this was found a couple of years before the finding of hydroxylation, or

active demethylation in B cells and T cell subsets

and that kind of fits, I guess.

Why would you need active demethylation

rather than passive? Well most of the brain, the neurons are postmitotic, so

they're not dividing.

So you can't dilute out DNA methylation I guess.

>> Sure.

Yep, that's right,

and yeah,

I guess you need a rapid sort of, ways to evolve, or differentiate in the brain.

And it might be that we have a DNA demethylation assisting with

that in the brain, in more somatic cells.

>> Yep

and the other thing when we think about active demethylation,

is that it in some ways it goes against what we've thought

especially, actually the fact that it occurs, it would appear,

more commonly than we originally thought.

Would be, it somewhat challenges the idea of having DNA

methylation as a very stable epigenetic mark.

I mean, it's the one that was first described to be how it was mitotically

heritable through the action of DNMT1

and most commonly the canonical view of DNA methylation is that it's a very stable

epigenetic mark, that can be maintained for the lifetime of an organism.

And while I still think that's true, we know of many cases where DNA methylation

is established and then maintained,

for example, on the inactive X chromosome.

Maybe, now, with active demethylation, it means that in some instances,

DNA methylation shouldn't be thought of as being an extremely stable epigenetic mark.

>> Yeah, well, I think that's an interesting question, too, and

it's entirely possible.

We see the TET proteins come on in us promiscuously in somatic cells.

And if they're expressed in somatic cells,

I mean that really challenges what we think of DNA methylation doing,

because presumably if they're on, they're demethylating to somatic cells.

>> So it would be really interesting to see in the future which regions

they are demethylating at those times.

So one can imagine that there are going to be regions of the genome that

are prone to demethylation, or that are being actively demethylated, and

regions that maybe are protected.

So maybe there are regions that are protected from the demethylases,

and they are retained in a stable fashion.

Perhaps like CPG islands on the inactive X chromosome.

And then there are regions where they are actively being demethylated

and the reasons as to why this occurs, we could postulate on lots of them.

I mean, it could be that you need to maintain demethylation

of most CPG islands.

We certainly find that most CPG islands aren't methylated,

so maybe there's some process that keeps them unmethylated by

removing any methyl marks that are added there.

>> Sure, well there certainly are processes that

control the DNA methylation state of particular regions of the genome.

We already spoke about the one cell embryo having the male pronucleus behaving

slightly different to the female pronucleus

and we know that this is because of an oocyte specific factor

called STELLA which comes from the maternal oocyte and

selectively binds the DNA of the maternal pronucleus

and then offers protection to the maternal genome from demethylation.

>> From that active demethylation.

>> From active DNA methylation, right?

>> So we also know the DMR's or the imprint control regions that are found,

they are protected from any demethylation, active or passive,

during that pre-implantation period,

so there are proteins, I believe that also protect those regions at that time.

>> Yeah, that's right, too.

So, STELLA is believed to do that actually.

To bind two DMR's in the paternal genome,

whereas it leaves the rest of the paternal genome unbound,

and to protect from DNA demethylation.

There's also a protein called zinc finger proteins 57

which will bind the methylated DMR of an imprinted region

and then it recruits other factors, including the denovo DNA methyltransferases

to make sure these imprinted regions stay methylated.

>> Stay methylated.

>> Through rounds of global DNA demethylation.

>> So perhaps, just like we know about these ones for these really critical

regions for the imprinted genes, maybe there are also proteins that bind to other

regions in the genome that you really need to keep faithfully methylated

and that's how we get the stability.

>> Sure, definitely,

and there are regions such as repeat regions that we know have to remain

silent through DNA methylation

and indeed they maintain their DNA methylation

so there's still a lot to learn there.

>> Yeah, I guess the interesting thing as well would be to think how does this

happen?

Why does it happen?

And so there's going to be a really active field in years to come.

It's already an active field and I imagine over the next few years lots of

the mechanisms that we're discussing will hopefully be revealed.

>> Sure that's what makes it still an exciting field I guess.

>> [LAUGH] Exactly, that's why we both work in it.

So I guess one of the interesting things that demethylation has also been studied

in, other than in the brain and in primordial germ cell development and

in early embryonic development, has also been in reprogramming.

So this is where we've discussed in the other lectures in week four that we have,

and week five I think, where we take a somatic cell and

it gets reprogrammed back to an induced pluripotent state.

And this is really important for therapy I guess.

People are really interested in doing this so

that you can create a pluripotent cell line from a particular patient

and then create differentiated cells, for example blood cells or new epithelial

cells of different kinds so that they can have them for transplantation.

So to be self transplantation with their very own cells, rather than having

the problems that you have with graft versus host disease, normally.

But, reprogramming hasn't just existed or

been known about since this discovery was made about ten years ago, in about 2006.

>> So, that's right.

The first reprogramming was first done in the late 1950s,

by a young scientist called John Gurdon

who had a fairly simple question by today's standards,

but really wanted to know whether a somatic cell contained all

the genetic information to produce a whole new organism.

>> So, by that you mean that he thought that there was a potential differentiation

as we know it might actually be a result of throwing genes out to feel that.

>> Exactly, regressing, loss of- >> So you're not having him any more.

>> Loss of genes, exactly.

And so his question was if the somatic nucleus contained all the genes required,

then it should form a new organism.

So what he did was to take a somatic nucleus from a frog and

put it into the oocyte of a frog, and see if it produced an embryo,

and sure enough it did, and that really nicely answered his question I guess.

That there really are all the genes required as we know now.

But yeah and I guess suggested for one of the first times that it was,

rather than a loss of the genes, it's a silencing of the genes.

So it was one of the first times we noticed epigenetic silencing in action.

But I guess what John Gurdon didn't know when he did these experiments was that

there had to be complete reprogramming of that somatic nucleus to completely

remodel the epigenome in the way that we understand reprogramming now.

>> So then a long time later,

50 years later,

then Yamanaka's group came along, and they used four transcription factors.

In fact, they screened for all sorts of different factors that might,

the combinations of different factors that might enable reprogramming of

a somatic cell back to something that looked like an embryonic stem cell,

which they called an induced pluripotent cell.

And they happened upon four particular transcription factors,

that are now called the Yamanaka factors.

And they got this reprogramming to work, but it's incredibly inefficient.

Maybe, less than 1% of cells will do this even if they all have

these four transcription factors.

So extremely inefficient.

>> That's true.

>> When you just add these four factors.

>> That's true, but we know it can be a lot better or made a lot more efficient

because the reprogramming by somatic cell nucleotransfer happens

with almost 100% efficiency, compared to 1% efficiency.

It also happens much faster, and the reprogramming that you get appears

to remodel the epigenome more faithfully than by the [INAUDIBLE] method.

So if you compare a somatic cell nuclear transfer reprogrammed cell,

it very closely resembles an ES cell and

in a more homogeneous way than what will a Yamanaka factor

reprogrammed cells.

>> Yeah. So that,

normal biology does it better than we do with.

[LAUGH]. >> Yeah.

>> This does not recapitulate no more situations.

>> Exactly.

Which doesn't surprise me at all.

>> Right. >> But

it means that there are more factors to find

and we can improve the, improve this process.

And there's a lot to learn from whatever is contained in that oocyte.

>> Absolutely. As an interesting note, because in

the early 2000s, there were groups and probably still are groups that wanted to,

for livestock purposes, they wanted to be able to reproduce their bull, if you like,

that had just the right genetics or

a particular deer that had just the right genetics for their livestock purposes.

And so before the Yamanaka factors and

actually still probably the best way to do it.

They used to use an oocyte and of course in livestock, it's okay to get more oocytes,

that's not really the case for humans, of course.

There are ethical issues and problems of getting enough of them.

>> That's right. >> But

that's not an issue with livestock so they, or not such an issue, I guess.

And what they would do to try and increase the efficiency because

while it can be close to 100% efficient I think this is organism specific so

I think some organisms are more efficient than others.

And the frog being more efficient than most because they have

maybe because they have a massive oocyte.

But what they used to do was sometimes fuse together two oocytes and

take out their nuclei and put in the nucleus and

figure well twice as much is going to help,

and actually it probably did so yeah, biology happens normally

and if we could do this relatively well without knowing anything about the factors

that are actually involved,

but that doesn't help us for the human therapeutics angle because we still

really want to identify what's going on,

because we're never going to be able to use human oocytes in the same way.

>> No, no, that's right.

>> Nor should we of course.

>> Exactly.

>> So, what's interesting to me, was that these Yamanaka factors,

the transcription factors, when it first came out in 2006,

I remember having what we call a journal club

and we would have one of the scientists discuss this particular new and

exciting article

and what struck me,

was that they weren't the factors that I thought would be involved.

They were transcription factors rather than epigenetic regulators and

to me I thought well, but it's all about epigenetic reprogramming,

why aren't there epigenetic factors in there?

But it's interesting that now the factors that increase the efficiency of

reprogramming seem to be a sweep of epigenetic factors.

>> Yeah, that's right.

So it's true that reprogramming is an initiated by the transcription factors,

but what happens after that is largely epigenetic,

epigenetic changes.

So the reprogramming happens in a sort of three phases

that we term the early phase, and the intermediate phase and the late phase

and the early phase is initiated by the introduction of these exogenous

transcription factors, which we call OSKM for short, but they're genes.

Oct4, Sox2, KLF4 and C-Myc

>> Mm-hm.

>> And the early phase is this sort of stochastic phase where some

cells will apoptose and some cells will syness and

other cells will undergo mesenchyme to epithelial transition.

>> Um-hm.

>> And it's these which we call MET.

>> [LAUGH].

Yeah.

>> So these these cells that undergo MET

are then the cells that are able to enter into the intermediate phase

and they have a range of different transcriptional profiles,

these epithelial cells

and the intermediate phase is sort of characterised by adjustments

to the transcriptome via the epigenome of these epithelial cells

and this is really where it's an inefficient process at this stage because

only some of these cells will pass through the intermediate stage and

it's the cells that end up expressing what we call early pluripotency factors that

will make it through the intermediate stage

and these early pluripotency factors tend to be factors that were originally marked

with H3K27 methylation so

that's the first histone mark to be reprogrammed in the process.

So the cells then that are expressing these early pluripotency factors then go

into what is the late phase, which is a deterministic, and very highly ordered

process where the later

stage pluripotency factors become activated, so that includes Oct4, Sox2.

So, once they're expressed,

the cell is no longer dependent on the exogenous transcription factors

and then the process is kind of unstoppable at this stage,

and so this is when we get most of our epigenetic remodeling so,

the genes that were marked by H3K9 trimethylation and

dimethylation are then reprogrammed and activated

we get the remodeling of DNA methylation at this stage, and

the resetting of the demethylation for the pluripotent state.

And one of the final things we get is the reactivation of the silent X chromosome,

which is really indicative of the process having worked at a lot of levels.

>> And I guess the inactive x has H3K9 methylation and

DNA methylation as striking features.

>> That's right. >> And so

those are some of it is one of the large bodies that would remove those and so

it's representative in many ways.

>> Sure exactly.

>> That's just it.

>> Exactly.

>> Yeah.

>> So that you would expect this to be one of the last things to happen in

reprogramming.

>> One of the interesting things is that the induced pluripotent cells or

the IPS cells that you take and compare between are a really quite variable,

more variable than what the underlined genetic difference would explain.

>> Mm-hm.

>> And so this is telling us that well we can get back to that point,

all those inefficiencies along the way, even when you do get back to

the final point, perhaps we still don't do it entirely correctly

and perhaps we can still improve the situation.

Now only in the efficiency of getting there but

also in how well they represent ground-state pluripotency maybe.

>> Yeah, sure so it's not perfect and there's a lot of,

the epigenome isn't remodeled in exactly the way that we would hope,

and this is one of the reasons why adding extra

epigenetic remodellers into the process often aids the process.

>> Yeah, so that, yeah, we can increase the efficiency, perhaps of stage two and

stage three, but certainly stage three.

The ones that have been described so

far seem to in Increase the efficiency of this final stage.

>> Yes.

>> And really getting it back maybe reprogram your last final steps.

>> Yeah, sure.

And we get this idea of epigenetic memory so, this comes from the fact that if you

take the parent's cell, reprogram the parent's cell and that parent's cell

is more able to differentiate back to- >> What it came from.

>> The IPS cell produced from that parent cell,

more able to differentiate back to what it came from.

>> So, what you mean is if you started out with a B-lymphocyte, and you'd

de-differentiate back to an IPS cell, you're more likely going to be able to

make another B-lymphocyte than you are to make skin or muscle or other cell types.

>> That's right,

that's right, and

it seems to be because of some kind of epigenetic memory.

>> Yep.

>> Often- >> Or inefficient

clearing I guess is another way of saying.

>> Inefficient clearing.

Exactly.

So you're left with remnants of the DNA methylation profile of the parent cell.

>> Yep.

Which could be helpful potentially if you actual want to take a B cell to create

more B cells.

But is not very helpful if you want to take maybe something that's easily accessible

like a swab from the cheek of the patient and end up producing red blood cells.

That's not going to be helpful to have a memory that you came from a cheek cell.

What we want to do really, is remove any memory that there is.

>> That's exactly right.

Although for workarounds, it would be possible to take,

to make sure you were using cell types that were related.

>> Yeah.

Potentially.

Maybe sometimes genetically not possible.

Because one of the ideas now is that, there have been some recent developments

with CRISPR Cas9 genome editing, which you can read about,

and we usually have discussions on this in the discussion forum.

We might be able to, people that have a genetic mutation which means that

they're unable to create a particular type of cell, you could create an

induced pluripotent cell

and then from any cell type, repair its DNA and

then reproduce the cell type that that patient was otherwise not able to produce.

>> Sure.

>> Which would be wonderful in theory.

One of the dangers of this at the moment is that if you have any remnant

induced pluripotent cells around,

these can really cause havoc if you transplant them into a patient.

So and that's because a pluripotent cell can create all sorts of different

cell types,

for example, they can create hair, eye,

tooth, liver, skin, all sorts of different cell types

and so if you transplant these pluripotent cells

what can happen is that you have what's called a teratoma,

and that's a tumor that actually can create all sorts of different types.

So normally in a person, these teratomas would occur maybe based on

an oocyte for example, and so there can be a tumor that occurs in the ovary and

when they section and they do indeed find hair growing inside the ovary

[LAUGH] and some tooth and some other bits of tissue, so

they're quite distressingly horrible

tumor types, but we don't want to have that sort of thing occur in patients

so this tells us that of course we want to get it right in terms of

the differentiation again,

so we're talking about dedifferentiation but other people work on trying to get

this reprogramming or differentiation down to a particular path perfect and

get rid of these pluripotent cells.

But the other way to manage this is to never make them to an induced potent state

in the first place.

And so there's a new field, a relatively new field again, although it's existed for

a very long time as well, where they performed transdifferentiation.

So they might take for example, a cell from your cheek and

try to directly make it into a red blood cell

and so if you don't,

if you avoid going back to this pluripotent state then you avoid that problem.

>> Sure. >> And

there's some progress happening in terms of being able to get this to work,

it seems like you still go through some sort of a progenitor like state.

But you're not forcing it back to that stage then differentiate again,

you're instead trying to get it to morph straight into that other cell type.

>> Yeah, it's getting that cell to do the minimum possible amount of reprogramming

and then send it back out.

>> Yep, so there is still going to be reprogramming and

it's still going to be an epigenetic event.

And so it will be interesting to see in the future how epigenetic modifications

and epigenetic regulators contribute to this process just like they do for

reprogramming back to the induced pluripotent state.

>> It will be.

>> Okay, thanks, Andrew

and so hopefully you've enjoyed this chat that we've had on these two topics and

you'll find also reviews that we've given you to read on this area and

hopefully you'll find it an interesting area to discuss.

Thanks.

>> Thanks.

5.6 Extension lecture: Interview with Dr Andrew Keniry

Epigenetic Control of Gene Expression

与讲师见面