The molecular basis of Huntington's disease | 91TV
Transcript
- So I'd like to thank the Medals and Awards Committee very much for giving
- me the opportunity to give this lecture, and to David for his kind words. So when
- I started working on Huntington's disease, all we knew was that it
- was a gene within the last 4 million base pairs of the short arm of chromosome 4,
- and a couple of weeks ago, we heard of the first disease-modifying treatments for this disorder,
- so we have come a long way. I think one of the things that you will take home from this lecture,
- I hope, is how important mouse models have been in that journey, and I think without mouse models,
- we really wouldn't have the understanding of the disease that we do have today. So as David said,
- Huntington's disease gene, it was the first gene to be mapped to a chromosome for which there was
- no prior knowledge of its location in 1983 by the Gusella group, and then it took ten years
- to clone the gene. Oh, I'm so sorry - I was going to stop elsewhere - I've just jumped a slide.
- Let me begin by reminding you of the cardinal features of Huntington's disease. So this is an
- autosomal dominant disorder. It can start at any time in life from very early-age onset in early
- childhood to very late older age. It doesn't matter when it starts, it's a very complex
- disorder. So there are problems with movement, psychiatric disturbances and also cognitive
- problems. So the adult-onset disorder generally starts in midlife, and the movement disorders can
- be quite complex, but the choreiform movements are what the disease was originally named for,
- and here you can see a lady with these dance-like movements from Sarah Tabrizi' clinic, but there
- are other movement disorders too, and the name was changed to reflect the complexity of the disease.
- So she has dystonic posturing problems with fine motor coordination, problems with eye movements.
- In addition to the movement disorders, there are psychiatric disturbances. Many
- patients have depression, and in a smaller proportion will develop psychotic symptoms, and
- they'll develop also a cognitive decline, problems with information processing, executive function.
- Something also we don't really understand is that, even with a fairly high calorific intake,
- Huntington's disease patients can lose weight. So the onset of the disorder is defined as the onset
- of movement symptoms, but actually the depressive or cognitive problems can start many years before
- that. So as I've mentioned, the disease can start in childhood or adolescence, and in this case,
- the disease can look very different. Here we have again a patient from Sarah's clinic,
- and this boy was 12 or 13 when he developed the disorder, and he was 16 in this video. You can
- see he has very, very slow movements. He can develop rigidity, tremors, and a proportion
- of these young patients develop seizures. So the Huntington's disease gene was the first
- gene to be mapped without any prior knowledge of its location, in 1983, and it took ten years to
- clone the gene. This was done by a collaborative group of scientists, and this was a meeting
- hosted by Dennis Shea in 1989, and this was led by Jim Gusella, Hans Lehrach, David Housman, John
- Wasmuth, Francis Collins, and Peter Harper under the umbrella of the Hereditary Disease Foundation,
- and the president was Nancy Wexler. So this gene was then published as this collaborative
- research paper without a first or last author. So what did this gene look like? It's a large gene,
- 180 kb of DNA with 67 exons. The mutation is a CAG repeat expansion within exon 1.
- If you have 35 CAGs or less, you will not develop Huntington's disease, but if you have 40 or more,
- this is a fully-penetrant mutation and will cause disease within a normal lifespan,
- and with this CAG expansion of 36 to 39, it gives an increasing chance of developing the disorder.
- So if we look at the relationship between the age of onset of the disorder. Let's see
- if I can get this. Here, against the CAG repeat size, you can see that there is a relationship,
- but it's most pronounced in these very early-onset cases with the large CAG expansions. Within this
- sort of 40 to 55 range, a CAG repeat expansion of any size can result in an age of onset that varies
- by decades. So there are clearly other factors contributing to this, and we'll talk about those a
- little bit more later. So it encodes a very large multifunction scaffold protein, and the CAG repeat
- encodes a polyglutamine tract, and the mutation imparts a gain of function to this protein.
- So as David mentioned, after I started my own lab, we were trying to develop a mouse model,
- and the first mice that we generated were transgenic mice. These mice were the R6 lines
- of mice. They were transgenic for human exon 1 of huntingtin. There were four lines. They had
- CAG repeat expansions varying from 115 up to about 150, and they developed neurological phenotypes.
- Over the next two or three years, quite a number of discoveries were made using these mice,
- and I'm just going to mention a few of them now, because these are phenotypes that are going to
- feature throughout the lecture. So Steve Davis, using an antibody from Marian DiFiglia's group,
- showed that there were inclusion bodies deposited in the neuronal nuclei in the mouse brains, and
- here, also, you can see there's an ultrastructure of the cell nucleus, and this here is the
- nuclear inclusion. The darker structure is the nucleolus. We were discussing these results and
- collaborating with Marian DiFiglia's group, and around the same time, they showed these structures
- were also present in post-mortem brains and that they could only detect them with an antibody to
- the very N-terminus of the protein. If she used an antibody that detected around 500 amino acids into
- the protein, you didn't see these structures. We were working with Erich Wanker in Berlin,
- and Erich set up fusion proteins with this exon 1 protein and was able to show that,
- when you had expansions in the mutant range, these proteins could form amyloid fibrils,
- and this really pleased Max Perutz, because Max had predicted three years before,
- using X-ray diffraction and glutamine peptides, that they would form cross-beta sheet structures,
- that is amyloid, by hydrogen bonding between the main chain and side chain amides. So if
- we compare a post-mortem brain from somebody who's died from late-stage Huntington's disease
- with a normal brain, you can see in this picture there's a dramatic difference. The Huntington's
- disease brain weighs a lot less, 200g, 400g less. Every structure is smaller. There's a generalised
- atrophy. There's also a specialised cell death on top of this, and so you can see that structure,
- CP, is the caudate putamen, that's the striatum, and in the Huntington's disease brain that has
- been atrophied really to a rim of tissue. It's the medium spiny neurones make up about 90 per cent of
- the neurones in that structure, and those are the neurones that have atrophied and died.
- Then, of course, now we have this aggregate pathology on top of this,
- and this is from Claire-Anne Gutekunst's paper of 1999, showing that you could see these inclusion
- bodies scattered throughout the neuropil, lining up in processes, and also present in nuclei. Then,
- I just want to make the point that, really, again, the extent of pathology is also related
- to the length of the CAG repeat. So the longer the CAG repeat, especially in those childhood cases,
- the pathology is much, much more widespread. We also look to see what happened to this CAG repeat
- expansion in brain, and found that in the mouse brain it continued to expand with age. So this is
- the R6/1 line of mice, and we have these are mice at three weeks of age. This is the CAG repeat that
- you would see in those tail samples. There's a little bit of stutter. These are mice at six weeks
- of age looking at some brain regions, and then by 30 weeks there's a lot more expansion has occurred
- in these CAG repeats and also in the liver. Another phenotype that was identified at
- this time, was Jang-Ho Cha and Young's lab published the first paper to suggest that
- part of this transcriptome might be dysregulated in Huntington's disease, and that group went on
- to show in these mice that, in fact, thousands of genes are dysregulated in the striatum of the
- R6/2 mice and also in muscle. Then, together with Lesley Jones' group, they went on to show that
- this was also true in the post-mortem brain. So the R6/2 line was the line that was used
- most extensively throughout the community. So this has two copies of normal huntingtin,
- normal mouse huntingtin, and this small transcript of human exon 1. There are hundreds of papers
- actually published on these mice. They're extremely deeply phenotyped, but just sort
- of basically, over this time scale, the colony that we worked with for many years in our lab,
- had about 200 CAGs in the mutant transgene. So huntingtin aggregation would appear before
- four weeks of age. Transcriptional dysregulation followed that at around six weeks of age. Weight
- loss was a feature that came later, but neurological phenotypes you didn't really
- see until about 12 weeks, and end-stage disease in those mice was around 14 weeks of age.
- So we were also working with knock-in mice, and these particular mice were generated by Peter
- Detloff from Alabama, and he had just put a CAG expansion, a large expansion, into the mouse gene,
- so we could work with them as heterozygotes or as homozygous-mutant. We did some fairly deep
- phenotyping on these, and we when we realised that, when the CAG repeat size and the strain
- background of the mice were controlled, that the phenotypes that were appearing
- in the knock-in mice were very similar to the ones in the R6/2mice, but just over a longer
- time scale. In these mice, end-stage disease was about 22 months of age in the homozygotes,
- but if you compared mice at end-stage disease in both lines, they looked really, really similar.
- This observation would be published in Brain Research Bulletin back in 2007, but it's the basis
- of everything that we've done since then. So when you compare the R6/2 mice with
- the knock-in mice, where repeat size, strain background and stage of disease are controlled,
- they both have weight loss, they both develop very similar neurological phenotypes. When you look at
- the aggregate pathology throughout the brain, it's completely widespread. They both develop almost
- identical transcriptional dysregulation, impaired heat-shock response. They both deposit aggregates
- in the same peripheral cells, ones that are terminally differentiated, like pancreatic
- islets or muscle fibres. Both get skeletal muscle abnormalities, cardiac dysfunction,
- innate immune activation, and altered glucose homeostasis. So this made us question, is it a
- fragment that is causing the disease in these knock-in mice? This wasn't a new idea. People
- had been - in fact Marian had proposed this a long time ago - and people had been working and looking
- at caspase and calpain fragments, but we decided to take an unbiased approach and just see if we
- could identify all of the proteolytic fragments that were present in the Q150 knock-in mice.
- So this work was done by Christian Landles, and we published it in 2010.
- So what we did was accumulated a series of antibodies across the protein,
- and then immunoprecipitated the protein with those antibodies, and used a classical mapping strategy,
- used N-terminal antibody, to detect it. So what you saw was this characteristic pattern of about
- 14 fragments that we always saw in the knock-in mice, and you were able to map the end of the
- fragment to lying between antibody-binding sites. The interesting thing was, and I won't
- go into how we knew that just at the moment, but we found out that this smallest fragment,
- which doesn't run fastest, was an exon 1 protein. It was just encoded by the exon 1 of the gene,
- and it was basically the same protein that is present in the R6 lines of mice. So given that it
- was an exon 1 protein, we questioned, well, maybe it's not a proteolytic fragment. Maybe it's arisen
- through some splicing mechanism. So we looked into that, and two or three years later published that,
- indeed, that was what was happening. As I've mentioned, we have a gene of 67
- exons. All of the introns can be spliced out to produce a mature message, and a distal polyA site
- put at the 3-prime end, but we found that in the presence of a long CAG repeat, you can have
- activation of cryptic polyA sites in intron one, and termination of transcription in that intron
- resulting in this small transcript that's just exon 1, where this part of the intron, basically,
- becomes 3-prime UTR. In human huntingtin intron 1 is 12 kb, in mouse huntingtin it's 20 kb. In both
- cases, there are two cryptic polyA sites that are activated at 2.7 and 7.3kb, or at 680 and
- 1145 base pairs in mouse, and this transcript is called huntingtin 1a. So 20 years after the gene
- had been cloned, we'd found that there's another transcript generated from the mutant allele.
- So I'm going to leave huntingtin 1a for the moment, and I'll come back to it later in the
- talk, and just segue a little bit back into this somatic CAG repeat expansion. So as you saw from
- that graph on one of the first slides, although the CAG repeat does influence age of onset,
- it's not the only determinant. There are clearly other factors involved, and in fact, it makes
- up about 60 per cent of the variation in age of onset, and the rest of it has some heritability.
- So for many years, people had started to look at trying to identify genetic modifiers for this
- heritability, but it wasn't really until 2015 when the genetic modifiers consortium
- really had the number of samples available, and the power of well-characterised individuals,
- to start to generate really consistent results. Over the next few years, from the GeM consortium,
- and from Sarah Tabrizi's lab, a very consistent pattern of genetic modifiers has emerged. This is
- showing you one from a paper this year, where they've looked at genetic modifiers of total
- functional capacity, so now really specific phenotypes, not just the age of onset.
- The striking thing in all cases, was that a large proportion of these genetic modifiers did mismatch
- repair genes, and that really caught everybody's attention. Right at the beginning in 2015,
- this was really interesting, and it was really interesting because we knew that in mice,
- if you knock out the mismatch repair genes, you prevent somatic expansion. In fact, the very
- first cross that was ever done with our mice was by Anne Messer's group, published in 1999, where
- she showed if you knock out Msh2, you prevent this striatal somatic expansion, and that's in
- the in the wild-type, and this is in the Msh2 knock-out. Vanessa Wheeler's group and Chris
- Pearson's group have gone on to show that that's true for other mismatch repair genes over the last
- few years. So this really focussed attention now, again, back onto somatic CAG repeat expansion.
- Interestingly, Peggy Shelbourne's group had shown, back in 2003, that in post-mortem brain, if she
- used small pool PCR to try and identify PCR across rare alleles with long CAG repeats, that she could
- identify these rare alleles in post-mortem brain with expansions up to 1000 CAGs. So, as I say,
- with a new focus on somatic expansion, more modern techniques have been applied. So Nat Heintz' group
- have sorted nuclei from caudate putamen of post-mortem brains, used MiSeq to sequence the
- CAG repeat lengths, found that in their case they were specific to these medium spiny neurones that
- die in the striatum, and also of one interneuron class, but not present in the other interneurons
- or the other glial cells. So these expansions were fairly - this is up to about 100 CAGs.
- Then Steve McCarroll's lab have taken a slightly different approach to a similar problem. Again,
- they've sorted nuclei from the caudate of post-mortem brain, but here they've used
- PacBio sequencing from the same nucleus used for RNA sequencing to look at the transcriptome.
- They've been able to identify expansions of as much as 800 CAGs in some of these cells.
- They've found that expansions more than 100 CAGs are pretty much restricted to these medium spiny
- neurones, and when they had looked at expansions above about 150, these correlated with a cell
- autonomous transcriptional dysregulation. This is the plot that they've showed from
- how the distribution of these CAG repeats would be in a post-mortem brain. This is from
- computer modelling, and this would be from somebody with a CAG expansion of, say,
- 4243 in their blood. So you can see that, even though most of the alleles have expanded, in
- most cases they're still less than 100, but it's this two to five per cent of rare alleles that are
- containing these very, very large expansions. So altogether, putting this together, the
- genome-wide association studies and the studies of somatic instability in brain, we really now accept
- that it's somatic CAG repeat expansion that is driving the age of onset and the rate of disease
- progression in Huntington's disease. So this really suggests that the pathogenic CAG repeat
- length in brain is not known, and it may not be the same in every cell, and also it suggests,
- as raised by McCarroll's group in their paper, that mutant huntingtin might actually remain
- benign for a long time until you hit a very long, a much longer CAG repeat length. So if
- this is step one in the pathogenic cascade, what is step two? So could step two be the alternative
- processing of the huntingtin transcript that we've identified? Just to recap, we have 67 exons in
- huntingtin. You can have complete processing of mature message and RNA. You produce this
- large protein that has a number of proteolytic fragments. When there's a long repeat, we can
- develop this small huntingtin 1 mRNA. This encodes huntingtin 1a protein, or the exon 1 huntingtin
- protein, and people have been working on exon 1 huntingtin protein for 20 years, or more than
- 20 years, and we know that it's very aggregation prone, and it's very, very highly pathogenic.
- So is huntingtin 1a the effector through which CAG repeat expansion exerts its pathogenic
- consequences? In order to model what might be happening as the CAG expands in a neurone,
- we've used an allelic series of knock-in mice. So these knock-in mice have one copy
- of mouse huntingtin has been replaced by human huntingtin with a different CAG repeat expansions.
- The expansions range from 20 through 50, 80, 111, 140 up to 190 CAGs, and these mice were
- generated by Marcy MacDonald and Scott Zeitlin's group. So we've used these lines to model what
- happens during this expansion on huntingtin transcripts and huntingtin protein isoforms
- and the molecular pathogenesis of the disease. So when we're looking at the huntingtin transcripts,
- we used wild-type heterozygous and homozygous mice from each of these lines, and they were all at 11
- weeks of age. So what happens to the huntingtin 1a transcript? This shows you that, as the CAG repeat
- is expanding from 20 up to 190, so the blue line is the amount of huntingtin 1a in the wild-type
- mice, that's just background levels, the orange is in heterozygous and the silver in the homozygous
- mice, so it increases with increasing CAG repeat length. There's twice as much in the homozygotes.
- When you look to see what happens to the full-length huntingtin transcript, it doesn't
- change much. It's pretty constant until you get to these long CAGs, and then it starts to decline
- in the heterozygous and homozygous mice. Trying to look at the huntingtin 1a protein isn't too
- straightforward, because there are no amino acids present in this small protein that are not present
- in the full-length protein, complete protein. So we have to rely on antibodies that will
- detect - that are specific to the C-terminus. So it ends in a proline, and these antibodies will
- only detect huntingtin 1a when that proline is a free proline at the end of this small protein.
- So we've developed a number of assays using these specific antibodies, and again, when
- you look to see what happens to the huntingtin 1a protein, it's increasing with CAG repeat length,
- but when we look at full-length huntingtin we find it's decreasing. You can visualise that
- in these Western blots that Christian's done. So here, the first six samples are from wild-type
- mice, the next six from HETs, and the last six from homozygotes. We're looking at 50,
- 111 and 75 with about 190 CAGs. So you can see when you have 50 CAGs in these lines, then the
- amount of full-length huntingtin is the same in wild-type mice as in homozygote mice, essentially,
- when you quantify it. As they increase, you can also see that the long polyglutamines result in
- this retardation as the proteins migrate through the gel, but the amount of full-length huntingtin
- is decreasing. So in a homozygous zQ175 mice, there's only about 10 to 12 per cent of
- full-length huntingtin in those mouse cells. This is [?cortex/core text 0:25:08.1], and
- that's the quantification. So can we detect this small transcript and protein in human cells?
- We spent quite a long time trying to see if we could find it in iPSCs and medium spiny neurones,
- without any success, but very recently the Tabrizi lab have generated an isogenic series of
- iPSCs with CAG repeat expansions ranging from 12 up to 175 CAGs, and they can differentiate
- these into medium spiny neurones and leave them to age for about 100, a little more than 100 days.
- Using a digital droplet PCR assay developed by Mike Flower, they found that when you get to these
- very long CAG repeat expansions, that's when you can start to detect huntingtin 1a. So Mike can see
- it in 140 CAGs, about three per cent of the level of full-length huntingtin, and then in 175 CAGs it
- goes up to about four-and-a-half, but there's a lot more in the medium spiny neurones. So we can
- detect huntingtin 1a in medium spiny neurones in a dish, but only with very long expansions. Andreas,
- when he was in the lab, had a go at looking for it in human post-mortem brain, and in bulk RNA it's
- really only obvious in juvenile cases with very long CAG repeats. This is actually what you might
- expect, based on the trajectory of those curves with how much it increases with CAG repeat length
- in mice, because, based on McCarroll's data, most of the CAG expansions in those post-mortem brains
- from somebody with the adult-onset form of the disease, are still less than 100, and those very
- long expansions are quite rare. Whereas, if you're already working with expansions in this sort of
- range, then the distribution is going to be much more skewed to longer CAG repeats, and it's more
- likely that you can detect this transcript. Finally, another way we can look for it is to
- use RNA scope - this is from Sandra Fienko in the lab - where you use a probe to look for
- transcripts. This is from a post-mortem brain from a CAG repeat of 63, and this is looking at
- full-length huntingtin, and we can see transcripts within the nucleus within the DAPI-stained nuclei,
- but also single transcripts outside the nucleus of processed message. Again,
- when we look at huntingtin 1a, it's much rarer, but again we see transcripts in the nucleus,
- and we can see transcripts outside the nucleus where it has been translocated for translation
- into the protein. So can model the effects of the expanded CAG repeat in a neurone
- as the CAG expands through somatic CAG repeat expansion. The amount of huntingtin 1a message
- in proteins increases. Full-length huntingtin levels decrease. We know it's very aggregation
- prone. So is huntingtin 1a driving aggregation? So what does huntingtin aggregation look like in
- these mice? So the zQ175 mouse is the knock-in mouse with the fastest phenotype, so we've
- used it a lot, and we actually characterised aggregation in these mice in great detail. Also,
- even though it's the fastest mouse, there's no obvious overt neurological phenotypes before a
- year of age. So looking at aggregated huntingtin in the striatum, we don't really see anything
- by immunohistochemistry at one month, but at two months we can already see a considerable amount,
- and then this increases over this period up to six months. Each of these disc-like, filled-in discs,
- each of these is a cell nucleus, and we can see a lot of extra nuclei aggregation as well. So
- how do we know this is aggregated huntingtin? If we use an antibody that detects the glutamine,
- or the polyglutamine tract, that is buried in the aggregate so the epitope is masked. So if you use
- this 4H7H7 antibody from Steve Finkbeiner, on a two-month section from zQ175 mice,
- you don't see anything, but if you pre-treat those samples with formic acid, then you start
- to reveal the aggregated huntingtin. If we go in to higher power, these are these nuclei, and they
- have - so probably every medium spiny neurones in the striatum of a two-month-old zQ175 mouse
- is full of aggregated huntingtin. You can already see also huntingtin aggregates in the cytoplasm.
- So to answer this question, does huntingtin 1a nucleate aggregation we tried to make
- knock-in mice that don't make huntingtin 1a. So in order to do this, we tried to prevent
- termination within intron one, and so we removed all of the cryptic polyA sites from intron one.
- So we're going back again to this knock-in mouse, the Q150, a large deletion out of the
- intron to remove all the cryptic polyA sites. We had a lot of founders with the deletion on
- the wild-type allele, and one founder with the deletion on the mutant allele, and the CAG repeat
- expansion was very similar to our Q150 colony. So had we succeeded? So these are, again, these
- immunoprecipitation Western blots from Christian. Here we have wild-type, the knock-in mice, and the
- mice the deletion, and your immunoprecipitating with antibody to the polyglutamine. So it only
- brings down the mutant protein, so you're not seeing anything in the wild-type lanes.
- If you probe with this antibody just to exon 1, that's just within exon 1 polyclonal, we
- can see there's this pattern of fragments that we discussed previously, but you can see that there's
- a lot less of this fragment, and here we're using two antibodies that are specific to huntingtin 1a,
- and we can see that we haven't got rid of it completely, but we've really diminished it.
- This is to exon 2, so now we don't see that band, and this is just pulling down the protein with an
- antibody [?for the/further 0:31:44.1] C-terminal, and you can see it's a genotyping assay. So it
- looks like we've really diminished the amount of huntingtin 1a in these mice. What effect has
- that had on aggregation? Here, we're looking at striatum cortex two months through to 17 months
- of age. On the top, for each brain region, we have the knock-in mice, and you can see the appearance
- of aggregates over that time scale. We see the diffuse staining, and then more and more obvious
- inclusion bodies, similarly in the cortex. If we look in the mice where we've done the
- deletion, and there's very little huntingtin 1a by comparison ,the aggregation is very,
- very much delayed, and this antibody is specific to huntingtin 1a, so we know even though
- there's a lot less huntingtin 1a, it's the huntingtin 1a that's aggregating. Similarly,
- if we look at the hippocampus, the CA1 region here, and the dentate gyrus and hilus, again,
- this is how it would develop in the knock-in mice, and in the ones where we have depleted huntingtin
- 1a, it's very, very much delayed. So we can see that using a huntingtin 1a-specific assay that
- Christian developed again and again. So now we have the wild-type mice with the delta intron.
- There's absolutely no aggregation in those in the cortex, striatum, and hippocampus. We can
- see increasing aggregation in the knock-in mice, and it's very much diminished when
- we really reduce the amounts of huntingtin 1a. So what effect does this have on transcriptional
- dysregulation? You'll see these sorts of plots again. So here we're comparing the mice with the
- deletion and not making much huntingtin 1a with the knock-in mice, striatum at six months of age,
- 12 months, and hippocampus at 12 months. This is the total number of dysregulated genes in that
- structure at that age, about 1200 at six months, almost 3000 when we get to 12 months, and this is
- the extent to which this dysregulation has been delayed. So these genes are not dysregulated at
- all, and these are only partially dysregulated. So it's been slowing it down from somewhere
- between 25 to 50 per cent. Really strikingly, one of the other things we've been looking at,
- are biomarkers in the CSF from these mice, because these are biomarkers that are used in the clinic.
- So neurofilament light chain is used frequently, and something called YKL-40, which is an
- inflammation marker. Again you can see, if we're looking at 12-month-old and 17-month-old mice,
- these are the wild-type, wild-type delta intron, knock-in mice, the ones with the deleted intron,
- that although we're seeing these biomarkers increasing in the CSF in the knock-in mice,
- they're at wild-type level. So we've completely normalised these markers of disease
- in the ones that don't have huntingtin 1a. So the reduction in huntingtin 1a profoundly
- delays huntingtin aggregation throughout the brain, partially delays transcriptional
- dysregulation, and it completely rescued the CSF biomarkers. So what have we put this together as
- a model? We have the huntingtin mutant gene, which has a CAG repeat expansion, and the size
- of that increases with age, and then this can be transcribed to produce full-length huntingtin,
- and the longer the CAG repeat, the more huntingtin 1a we're producing. So the transcript translocates
- to the cytoplasm, where it generates the huntingtin 1a protein, and we think that
- this is critical in initiating aggregation. So aggregation is a concentration-dependent process,
- and whether or not it occurs will depend on whether you hit the critical concentration
- for that process. So that's going to depend on the glutamine length, because the longer the
- glutamine the lower the concentration required, the level of the huntingtin 1a, and possibly we
- think huntingtin fragments, small fragments of full-length huntingtin. We know post-translational
- modifications have been shown to affect this, interacting proteins, RNAs, the subcellular
- localisation, it may not be the same concentration in all parts of the cell, and of course, the
- protein folding homeostasis of that cell, which is trying to keep these proteins soluble. So if
- the concentration is reached, then you'll have oligomeric and fibrillar aggregates forming,
- and that's causing cellular dysfunction. So we know huntingtin 1a is a small protein.
- It can diffuse into the nucleus, but it doesn't stay there, because it has a very
- potent nuclear export signal at its N-terminus, and it gets excluded again very, very quickly,
- very dramatically. So it stays in the nucleus, because it's aggregated there. I don't have
- time to show you the data at all, but we're not sure what happens to other huntingtin fragments,
- but we know that it's aggregated huntingtin in the nucleus that causes transcriptional dysregulation.
- We've also seen that, in some cases, the full-length huntingtin and huntingtin RNA can form
- RNA clusters in the nucleus, and at this moment we are further investigating what the implications
- of this might be. So if we want to try and prevent this cascade, this pathogenic cascade,
- we could prevent the somatic CAG repeat expansion and prevent this production of huntingtin 1a,
- or we can try and get rid of the full-length huntingtin and huntingtin 1a transcripts,
- or just the huntingtin 1a transcript maybe, and prevent this cascade from occurring.
- So there's an awful lot of work at the moment, a lot of attention, in terms of developing
- therapeutics to somatic CAG repeat expansion in industry and in academia, and we're working
- in that area, too, but I don't have time to tell you much about that today. So what I'm going to
- focus on is the huntingtin-lowering strategies that are being employed.
- So huntingtin-lowering. Huntingtin-lowering is something that has been pursued for the last 20
- years, really, and the very first successful approach has been antisense oligonucleotides.
- They were developed by Ionis Pharmaceuticals and then taken into the clinic with Roche,
- in partnership with Roche. So there was a safety trial, a Phase I/IIa safety and tolerability
- trial, and it looked very exciting. Tominersen was administered intrathecally into the CSF
- every four weeks for a total of four doses, and when they looked to see the level of huntingtin
- knock-down, there was a dose-responsive reduction in huntingtin in the CSF.
- This was very, very encouraging, and Roche took this to a Phase III clinical trial,
- and this was a huge trial. It had more than 800 patients worldwide. There were
- three cohorts. So the dose in all case was 120mg given every two months, every four months,
- or a placebo arm. But it was devastating for the community when, after 69 weeks, the Independent
- Data Monitoring Committee halted the trial because of safety concerns, and essentially because they
- couldn't see any improvement. So one of the primary endpoints was the consolidated United
- Huntington's Disease Rating Scale. These were the data. This is starting from baseline. If things
- were being improved, you'd expect them to go in that direction. So the grey is the placebo group,
- the orange is the four-monthly treatment, and the blue is the two-monthly treatment.
- So it looked as though individuals on active ASO were performing worse than those on placebo.
- So why was this? Could it be a safety issue? It certainly could. There was an increased
- lateral ventricle size. There was a CSF spike in neurofilament early on, which could be a marker of
- neuronal damage. Or could it be the wrong target? The position of these ASOs was around exons 36,
- or something like that, so they're only targeting full-length huntingtin. They're knocking down both
- mutant and wild-type full-length huntingtin, so lowering wild-type huntingtin would be
- detrimental. That was something that has been a concern for the community for a long time,
- but also, they would not be altering the levels of the pathogenic huntingtin 1a fragment. So could
- that be the problem? For huntingtin-lowering, what we need to know is what is the relative
- pathogenesis of huntingtin 1a and full-length huntingtin, and should we target huntingtin alone,
- huntingtin 1a alone, or both transcripts. So what I'm going to do next is show you data from two
- studies that are currently preprints. The first is from Jeff Carroll's lab at the
- University of Washington. And Jeff was using an ASO from Wave Life Sciences. So I told you earlier
- on - so this was in the Q111 knock-in mice, and this is the wild-type allele - I told you before
- that mouse exon 1 had been replaced with human exon 1, but this had also replaced part of intron
- one. So the first part of intron one was also humanised, and the ASO that Jeff got from Wave
- Life Sciences targeted this humanised part of the intron. So it was specific to the mutant allele
- and it knocked down mutant full-length huntingtin, but it was also before the first cryptic polyA
- site, and so it was also knocking down huntingtin 1a. He also tested the equivalent of tominersen,
- which here is targeting exon 42, so the equivalent of the ASO that Ionis and Roche had used.
- So what they did, was they injected mice into the lateral ventricles at three months of age, again
- at six months of age and took them at nine months. This is showing the aggregation in the striatum.
- The green colour is aggregated huntingtin, and the purple colour is [?from the NeuN 0:43:22.5],
- so that's showing you the neurones. So this is the mice treated with saline.
- These are the mice treated with the equivalent of tominersen, that's just knocking down full-length
- huntingtin. It's had absolutely no effect. This is the effect of the mice that were treated with the
- ASO that's taking out huntingtin 1a full-length huntingtin. There's a dramatic reduction in
- aggregation, and that's the quantification. When they looked at the transcriptome, if they compare
- the transcriptome of mice that have been treated with saline, with mice that were treated with the
- tominersen equivalent, there's basically no difference. Huntingtin is different because
- they've knocked down huntingtin and one other gene, but when they compare the ones treated with
- saline with the one that's targeted huntingtin 1a and full-length, then there's a dramatic
- change. When they look at the specific signal of transcripts that are altered in the striatum,
- they found that about 50 per cent of them were rescued in terms of their dysregulation.
- So when they've just knocked down full-length huntingtin, they see no effect on these molecular
- measures, but when they knock down huntingtin 1a and full-length, then they have a dramatic effect.
- The other study I want to tell you about is one that we've done in collaboration with
- Anastasia Khvorova. Anastasia is director of the RNA Therapeutics Institute at UMass,
- and she has developed a divalent siRNA chemical scaffold that results in the sustained modulation
- of expression throughout the CNS, and some years ago she developed an siRNA that would target
- full-length huntingtin. It's a very potent siRNA. It targets the 3-prime UTR. So she did a screen
- for siRNAs that would target intron 1 before the first polyA in the mouse, and her lab,
- together with Cat in my lab, identified two siRNAs to take forward. So siRNA is targeting
- mature transcripts in the cytoplasm. So here, we can compare the consequences of targeting either
- full-length huntingtin or huntingtin 1a. So the one targeting full-length huntingtin
- decreased full-length huntingtin by about 80 per cent, and the ones targeting huntingtin
- 1a decreased by about 50 per cent. So the one targeting full-length was much more
- potent. We set up a fairly complicated trial, where we treated mice at two months of age,
- and took at three months of baseline to look at potency; two months, take at six months;
- two months, treat again at six months, take at ten months; and six months, and take at ten months.
- The potency and durability was greatest in the hippocampus, so that's what I'm going to show you.
- Again here we're looking at the effects in the CA1 region of these untreated mice. It's the CA2
- region pf the hippocampus. This is what it looks like in six-month mice, and this is in ten-month
- mice. This is the double and late cohorts, so again you have more inclusion body formation.
- The non-targeting siRNA had no effect. Full-length huntingtin had no effect.
- [?Take out/Knock down 0:46:46.3] huntingtin 1a, even though it's much less potent,
- then it has a dramatic effect on aggregation, but it's not so black and white. I don't want
- to give you that impression. So if we look at the CA3 region of the hippocampus, here we have again
- untreated mice. This is six months and ten months of age. We can see the nuclear inclusion at those
- times. We've also got a letter of extra nuclear aggregation here. Again, the non-targeting had
- no effect at all. Full-length huntingtin, now we have we've slowed down the appearance of these
- inclusion bodies a little bit, and we've removed a lot of these extra nuclear inclusions. So even
- though the effect of knocking down huntingtin 1a was more pronounced, though it was less potent,
- if you really knocked down full-length huntingtin, we are seeing some effect, but the effect if you
- start later, if you don't knock down until six months of age, then that is much less effective.
- When we again look at these transcriptional dysregulation profiles here - this is the early
- cohort, the double and the late - ad the second line in all cases is knocking down huntingtin 1a,
- and again, we see a greater improvement with that approach. So lowering the levels
- of huntingtin 1a can have a profound effect on huntingtin aggregation, and our mouse data
- indicate that lowering huntingtin 1a levels is more efficacious than lowering full-length
- huntingtin. So it really suggests that approaches should include huntingtin 1a, and that they should
- start before the aggregation process is well established. I should point out, these mouse
- models all have a basal CAG repeat length that's already very long, so we're actually asking a
- lot to treat these mice in terms of the amount of huntingtin 1a that's been produced, and the rapid
- aggregation phenotypes, because they're already, all cells, are containing CAG repeats that might
- be at or close to the pathogenic threshold. So what is available in the clinical trials
- at the moment? This isn't an exhaustive list. I just wanted to list the range of treatments
- that are being developed. So we can see that we have in the nucleus, we've got the pre-mRNA
- full-length pre-mRNA and huntingtin 1a, and then transported to the cytoplasm. Here we have
- the mature message and huntingtin 1a message, and these produce these proteins. So in terms
- of targeting full-length huntingtin, there's the ASO for total huntingtin. that's tominersen that
- we've heard about. Wave life Sciences also have one that just targets mutant huntingtin. Again,
- in the it's not in the region that would hit huntingtin 1a. PTC Therapeutics and Novartis
- have a small molecule, again that will lower full-length huntingtin, but not huntingtin 1a.
- The approaches that target full-length huntingtin and huntingtin [?wild/1a 0:49:44.5] because they
- are directed to exon 1, are uniQure, a microRNA, the AMT-130, and Alnylam have an siRNA, again
- targeting in the cytoplasm, targeting the exon 1. So they will hit both of these transcripts.
- We have worked with uniQure to make sure that the microRNA is doing what they wanted it to do. So
- these microRNAs are packaged into AAV viruses. They use AAV5. They did bilateral intrastriatal
- injections into, again, these zQ175 mice. They had a wild-type vehicle and then zQ175 mice with
- low dose and a high dose. These were treated at five months - so this is quite late - and taken
- at two months later. We used our assays to show that they were knocking down both full-length
- huntingtin and huntingtin 1a. This is the this is looking at soluble huntingtin 1a protein.
- The striatum is decreased up to around 50 per cent in full-length, and also in the cortex,
- because even though the virus is going into the striatum, it will spread also to the cortex.
- When you look at soluble mutant huntingtin, again similar reduction. Surprisingly,
- when we looked at aggregated huntingtin 1a, there was even a reduction in striatum, given how late
- it was that they did this in those mice. UniQure already had a clinical program in
- place when we did that work with them. They are using AMT-30 a Phase I/II clinical trial.
- This is a single administration of AMT-130. They use MRI-guided convection enhanced stereotaxic
- surgery to both sides, both striata. This is a very complicated and long surgical process, that
- Sarah told me takes 12 hours minimum I think. Yes, 17 hours to begin with. They had 17 in the high
- dose and 12 in the low dose, and the 36-month data was compared to what they call a propensity score
- matched external control from the CHDI enrol database, where they were comparing to large
- numbers of individuals, both for the high dose and the low dose. Of course, this was great news,
- two weeks ago, because it has had a dramatic effect, and it's the first time we have seen
- a treatment for Huntington's disease that works. So the results were that at 36-months there was a
- 75 per cent slowing of disease progression, as measured by the composite UHDRS, which was the
- primary endpoint; a 60 per cent slowing of disease progression, as measured by the total functional
- capacity; favourable trends in other secondary endpoints. They had elevated NFL in their CSF
- and there was a reduction from baseline, which was remarkable, and it was well it continued to
- be well tolerated. So the first disease-modifying treatment for Huntington's disease, and I think it
- underlines the importance of targeting huntingtin 1a in this case. So I'm going to leave it there,
- and I think anything else can come out in the discussion. I just want to thank all of the
- funding agencies over the years, particularly the CHDI Foundation, because without them, we
- would have never been able to do all this work on huntingtin 1a. I thank all of my collaborators and
- the many collaborators that have gone before them. This is my current research group. I want
- to thank all the alumni. I don't have pictures of everybody, unfortunately. Our Huntington's
- Disease Centre was opened in 2017, and this is the centre at our symposium last month.
- Thank you. I went on a bit long.
- inspirational. Thank you so much. Do we have any questions, or should I say,
- how many questions do we have? Yeah?
- and good timing with the uniQure... That's right on time. So I was wondering, when you
- look at that Htt exon 1a transcriptopathy, did you compare it to Msh3 lowering transcriptopathy from,
- for example, William Young's paper? Do you have the same kind of profile when you look...?
- Oh, so the transcriptopathies. I think that is quite complicated to compare them, actually. Then,
- I think this comes back to the fact that these mice are already at a point where the CAG repeat
- is quite long. So even in the Q111 mice, I think there's enough huntingtin 1a there to start that
- aggregation in the nucleus. It's very low level, but it's there, and you don't need
- much aggregation in the nucleus to start the transcriptional dysregulation. So you're never
- going to get it to nothing because it's already there. You kind of stop - if you block somatic
- expansion, you can hold it at that point, but you don't get... So when you compare these
- percentages, it's hard to, really, because there's different numbers of genes that are dysregulated
- and what have you. But I think, in the patients we won't necessarily have that many alleles at that
- length, and that's why I think, especially with the uniQure thing, I think it's making us rethink
- it a little bit and how we interpret our mice. I didn't say - I mean, there's so much to say
- about it, but one of the striking things about it was that it was given to symptomatic patients.
- And we've all been saying for so long, 'Oh, we're going to have to go early. We're going to have to
- go really early. We're going to have to go before symptoms, because it's going to be too challenging
- to slow anything down,' but it hasn't. So I think we need to think about the cytoplasmic
- aggregation a bit more. We focus so much on the nuclear aggregation because transcription
- is so easy to measure. We have a ton of data which I didn't put in - I had to take out,
- I had to take out so much - that shows that actually we think it's the extranuclear pathology
- that is more related to disease progression than the nuclear pathology. I think that it may
- be that, if you start to lower the precursor of aggregation, then you've got more chance to clear
- in the cytoplasm than you have in the nucleus. So I think the fact that we have this transcriptomic
- data might be a little bit misleading, but it's very hard to compare one study with
- other just based on these percentages.
- So this is a question from our online audience, from Sophie, asking, what surprising or unexpected
- findings within recent years have changed how you think about Huntington's disease pathogenesis?
- Well, in a way, it's a good question, because you always have to follow those unexpected
- results. The thing that doesn't make sense at the time. So speaking from your own lab,
- you know, we couldn't understand why we only saw one band on a Western blot,
- and then realised that this smaller fragment was huntingtin 1a, when we were doing all of
- that stuff in 2010. Surprising results? I think the fact that the somatic expansion was so...
- Those alleles are so expanded in the brain. I think, early on, we didn't really expect that
- to be the case, even though we saw quite large expansions in the mice very early on. Again,
- you were starting from a high baseline. You were starting from something more than 100. So I think
- at the beginning we would never have expected that you might see 500 CAGs in a cell in the brain,
- and It's turned out to be probably the first step in the pathogenic process.
- A question at the front here from your collaborators.
- That was an amazing talk, a real tour de force. It was fantastic. So I actually have a question,
- and I know I could ask you any time, but I actually... Why do you think the expanding
- CAG results in more of huntingtin 1a production? What's the mechanism?
- We don't understand the mechanism.
- Well, yes, but not ones that.... Not anything I have any real evidence for. It's probably a
- structural thing, but we haven't really got - any hypotheses we have developed,
- we've generally disproved. So we haven't got anything really active at the moment.
- So maybe I could ask one? So is there any evidence for a
- prion-like effect in Huntington's disease?
- to be able to take up aggregates from the media and from - but there hasn't been a convincing
- experiment in mice. I don't know whether that's because people haven't tried hard enough. We did
- some fairly simple experiments. I wouldn't say that they were particularly conclusive,
- because I didn't want to go down this rabbit hole, which I thought might be something that
- was going to waste somebody's time for a couple of years or something. We did simple experiments,
- of the sort that they did in Alzheimer's disease, or in the prion diseases didn't
- really - weren't anything obvious.
- just on the left here.
- Gill. I had a thought about this sequence, or the structure of the huntingtin protein,
- which you know very well is a modifier of onset, but it doesn't seem to be linked to
- somatic instability or some of the evidence that's coming out recently. Do you think it
- influences the exon 1 transcript production, and are you going to look into it?
- I'm not sure I really actually follow what it is that you're bringing up.
- So the fact that if they had a loss of CAA or...
- Oh, sorry, so we're talking about the...
- Yes, so some of the most dramatic modifiers are the CAG repeat itself. So the canonical sequence
- is a CAG tract, and then a CAA and a CAG, and they all encode glutamine, so it doesn't change
- anything in the protein. If you if you lose that interruption, then you accelerate onset,
- and if you have additional interruptions, you delay it. It doesn't seem to be related to somatic
- instability, so it could be related to huntingtin 1a production, but the experiments that would
- really test it haven't been done.
- Thank you. I may have missed it. The somatic expansion, is that restricted to the vulnerable
- neuronal population, or it's more diffuse within the brain? Then second, has this
- phenomenon has been demonstrated for other dynamic mutations like spinocerebellar ataxia, etc.?
- So in terms of the striatum, it's definitely restricted to the population that are vulnerable.
- I don't know whether that would be true throughout the whole brain, because I'm not sure that the
- cells that die necessarily die through a cell autonomous process. It could be elsewhere. Yes,
- it is, it is something that's common to all of these CAG repeat expansion disorders,
- and so that is one of the very exciting things for these colleagues of ours who are developing
- treatments that are targeting the mismatch repair proteins, because if they find one that works
- for Huntington's disease, it will probably be applicable to all the spinocerebellar ataxias and
- many other repeat expansion disorders as well.
- I first came across Huntington's chorea 40 years ago, when I was a journalist in South Wales,
- so I'm just amazed that you've achieved so much. It's just wonderful. I wanted to know the extent
- to which your breakthrough research is helping people with Huntington's chorea, Huntington's
- disease now, and how widespread treatment is.
- uniQure trial is quite small, and because it's a gene therapy, it's going to be very expensive, so
- it's unlikely that it's going to be available for all Huntington's disease patients. I don't think
- we can say at the moment what would be the case in the United Kingdom, but there will certainly
- be areas of the world where this is never going to be available to the people, because of the
- complexity of the surgery and the expense in this treatment, but it's opened the door. I mean,
- that is the really important thing. We know now that if you target exon 1 of full-length
- huntingtin and you knock down huntingtin 1a as well, there's enough mouse data, and now
- the clinical data, to say this is important. So it's not the only modality that's doing that. You
- know there are there's the Alnylam siRNAs are already being tested in the clinic, and there
- will be many people now will be shifting some of the work they're doing more into this region.
- So I think we've got a lot more to do to really understand the levels of knock-down and whether
- you need huntingtin 1a and full-length, or whether you could just do huntingtin 1a, because there's
- always going to be a little bit of a concern about knocking down the wild-type allele. So there's
- lots more for us to do, because we need treatments that are going to be available for everybody.
Join us for the Royal Society Ferrier Prize Lecture delivered by Professor Gillian Bates FRS.
Professor Gillian Bates FRS is awarded the Ferrier Medal 2025 for her work in understanding the molecular basis of Huntington’s disease and consistently producing highly impactful findings which have moulded the course of this field.
This talk covers how in 1993, the mutation that causes Huntington’s disease, a devastating neurodegenerative disorder, was found to be an extra-long CAG repeat in exon 1 of the huntingtin gene (HTT) that encodes a polyglutamine tract in the HTT protein. Over the last 30 years, the Bates lab has been unravelling the first molecular steps by which the mutation causes neuronal dysfunction and neurodegeneration. The mutant CAG repeat is unstable, expanding in specific neuronal cells with age, and the rate of this somatic expansion is known to drive the age of disease onset and rate of disease progression. The Bates lab have found that the mutant HTT pre-mRNA is alternatively processed to generate the small HTT1a transcript, the longer the CAG repeat, the more HTT1a is produced. This encodes the aggregation prone HTT1a protein, that they have shown to be highly pathogenic, representing the second step in the pathogenic cascade. Their recent preclinical studies demonstrate that lowering the HTT1a transcript is much more effective than lowering full-length HTT and may explain the failure of a large huntingtin-lowering clinical trial. Their data indicates that huntingtin-lowering strategies should be designed to target the HTT1a transcript.
About the Royal Society
91TV is a Fellowship of many of the world's most eminent scientists and is the oldest scientific academy in continuous existence.
/
Subscribe to our YouTube channel for exciting science videos and live events.
Find us on:
Bluesky:
Facebook:
Instagram:
LinkedIn:
TikTok: