Posted 15 March 1999

Interview with John Hardy by June Kinoshita.

ARF: What is the primary hypothesis that guides your laboratory?

JH: Overall the theme of the lab is to use genetics to find what causes disease and to model the disease in cells and animals. With respect to AD, we believe that that approach has been successful and pointed very clearly at genes that modify amyloid production.

ARF: What molecular and cellular changes account for the initial symptoms of AD, and what might be the earliest pathological changes?

JH: What the mutations tell us -- that is in APP and presenilin -- is that they alter APP processing, and of course they alter APP processing presumably from birth and before. If that happens from birth, why do we only see symptoms late in life? The real answer is we don't know. But certainly part of the process is an accretion process. Part of it depends on amyloid building up. But possibly there are other components that we only dimly understand that has to do with the changing brain environment as we age. We have a very nebulous idea of what's going on in the aging brain.

ARF: But if you were to speculate?

JH: Of course I don't know, but continuing to be nebulous, I suspect that amyloid and APP have to do with synaptic plasticity, and there are changes in synaptic plasticity with age. But really I can't answer that.

ARF: How strong a role do you think genetics plays in AD?

JH: Of course the simple genetic variance of the disease is well known. [In FAD cases] it seems virtually determined entirely by genetics. For example, in the French-Italian family, individuals living in France, Italy and the U.S., all get the disease at the same age. That's a presenilin mutation family. Leaving aside these rare kindreds, I think in sporadic cases genetics has a very important part to play. If you have a sibling with AD, your chances of getting AD are 2-3 times that of the rest of the population. ApoE is also a factor. I think this nature versus nurture debate is kind of sterile. Usually it's an interaction of their genes with their environment. There's been nice work on head injury showing that people with the E4 allele with head injury are more likely to have problems relating to that head injury.

ARF: Outline for us the progression the disease takes.

JH: That's a difficult question. This is overall what I think is happening. Let me answer it specifically with respect to a family with an APP or PS mutation. Those individuals are producing more A-beta 42 right from birth. At some point, what happens is the amount of A-beta starts to cause problems in the cell. There's been very nice work, especially from Charlie Glabe's group, suggesting that at a certain critical point, the amount of A-beta starts to mess up the metabolism of the APP in a particular cell (see BA Bahr, et al). At that point, you get something that changes from a fairly benign overproduction to a runaway event. The cell continues to make APP but gets tripped up, and this becomes a self-perpetuating event and the cell produces massive amounts of A-beta. Like something smoldering that breaks into flames. One presumes that the cell then begins to produce tangles, although the relationship between A-beta and tangles is not at all well known. Then they begin to spread to adjacent cells. There's good data from Carl Pearson that shows an element of AD pathology travels down neuronal pathways. Why does it start? The Braaks have shown the disease starts in the hippocampus, but why there we don't know. Perhaps it has to do with that part of the brain producing more of APP and thus A-beta, and there are parts of neuronal selectivity we don't understand.

ARF: What do you think accounts for the specific anatomical pattern seen in the progression of the disease?

JH: That really deserves a discussion. We really don't understand it. What's really striking in the transgenic animals is that the specificity is pretty much the same as in humans. Very impressively the Athena people have shown that whatever level of amyloid production you reach, you never deposit amyloid in the thalamus. That's independent of APP production. Even if through transgenesis, you produce lots of APP in the thalamus, you don't get plaques. Clearly this has something to do with the neuronal architecture or extracellular matrix. We have no clues at all right now.

ARF: How useful are the current mouse models? What would you do to improve them?

JH: Our group has always believed they're the key to understanding the disease process. At the moment the mice we have are useful for looking at the amyloid deposition process and for testing drugs that interfere with amyloid deposition. We can do lesioning experiments to see how that changes the deposition process. What they don't do is they don't show massive cell loss, and they don't show neurofibrillary tangles. So they are really limited in terms of studying drugs that interfere with those processes. That's a deficiency we're working to correct. Karen Duff, Mike Hutton and I are trying to use findings on the tau gene to push the disease process further. One possible problem is that mouse tau is different from human tau, so Karen Duff has been putting human tau into an amyloid producing mouse, and we're waiting to see whether she's successful or not.

ARF: Do mice have the right kinases and phosphatases [for tau]?

JH: Generally they have the right enzyme machinery. The difference between the mouse and human has more to do with splicing. All tau has 4 microtubule binding motifs in it. In the mouse, it's always 4-repeat tau, but in humans, by alternative splicing, you produce 3-repeat tau. In adult humans, about half is 3-repeat, and half is 4-repeat. Mike Hutton has shown a stem loop structure in the intron that is responsible for this difference. We've shown that people with a mutation in this structure develop a frontal dementia. We're now putting into mice the normal human tau and tau with mutations. We're gradually humanizing the mouse neuronal genes so eventually we'll be able to mimic the full pathology. It's just a question of how much genetic manipulation we'll have to do before we're successful.

ARF: What about differences between mouse and human immune system?

JH: I'm not expert in that area. It's possible that the immune system may be involved, and there are differences between mouse and human immune systems. I'm quite prepared to believe that dampening down the immune response could help slow down the pathology, but I don't believe it's involved in initiating the pathology.

ARF: How are A-beta and tangles related?

JH: I don't think the relationship is particularly tight. First of all, there are several syndromes which resemble AD where you get plaques of different sorts and tangles. For example, Indiana disease cases have prion disease and tangles, and in Worster-Drought syndrome in England, there's amyloid deposits and it's a dementing disease with tangles as well, but the amyloid is a different peptide. Here are three completely different diseases in which you get amyloid deposits of different sorts and tangles, which suggests to me there isn't a direct biochemical link between A-beta and tangles, but rather that plaques or amyloids in general cause a nonspecific type of damage that leads to tangles. I wouldn't suspect that A-beta triggers a specific kinase, but that it's cruder than that.

ARF: What form of A-beta is bad for you?

JH: I think plaques are bad, because they act as a reservoir for more A-beta -- a reservoir of problems, if you like. As for the toxic species, I think it might be a fairly small molecule. It comes back to Glabe's work. What he really has shown is that A-beta is binding the A-beta part of APP, thus tripping up the metabolism of APP. You'd expect it to be the monomer that trips up the APP molecule. That leads to misprocessing and the cycle repeats itself. I suspect that's the likely explanation.

ARF: There's been growing interest lately in small A-beta derived oligomers, and some work by Bill Klein's lab at Northwestern showing that they activate an apoptotic pathway, so that's a different kind of mechanism than the one you've described (see Lambert MP, et al). What do you think of that theory?

JH: I'm very suspicious of apoptosis, actually. Apoptosis is a cell death program that takes place over a few days, and what you have here is a chronic neurodegenerative process that goes on over years, and tangles develop over years. I've been very suspicious that apoptosis plays any role in AD. I'm never quite sure what apoptosis is, to begin with. A politically correct name for cell death. There are studies that purport to show apoptosis in AD brain, but I think by its nature, it's impossible to show that in human brain, because it's dead. All the cells you're looking at are dead, so there are huge confounding variables. Apoptosis could be triggered in the half hour or hour before that person died.

ARF: What key experiment is needed to convince skeptics that your hypothesis is correct? Let's assume there are no technical, ethical or financial constraints.

JH: I'm not sure who these skeptics are. I'd like to know who it is that doesn't believe A-beta is a key initiating event. There's a very nice paper in Annals of Neurology showing that the reason Down's syndrome individuals get AD is explicitly the triplication of the APP gene (see Prasher VP et al). You have six or seven different molecular causes of AD that all lead to overproduction of A-beta 42. If you're going to have any other theory, you've got to do better than the A-beta hypothesis. The ideas that are proposed don't get even close to clearing that high hurdle. If someone came up with a hypothesis that explains everything.. but... Newtonian law explained the movement of the planets, and Einstein theories of relativity explained not only the movement of planets but the movement of electrons as well. People who question the hypothesis don't even try to explain how all these mutations cause more A-beta 42.

ARF: What about convincing yourself 100% that it's right? What's missing right now?

JH: There are huge amounts we don't know. We mentioned neuroanatomy. The amyloid hypothesis has no neuroanatomy. It could almost happen in the liver. That's a massive problem. We clearly want to model it in mice, and we're only halfway there. We don't understand the link between A-beta and tangles, but that could be worked out in the next year or two. The key has been the identification of mutations that cause tangle formation. That points us in a direction. I also think Michel Goedert's paper showing if you surround tau with negative molecules, you get tangle formation might be very important (see Hasegawa M, et al). We might be close to clearing that hurdle. Those are the two main things. Of course, what we really want to see, both to prove the hypothesis but also from the clinical point of view is a drug which reduces the production of A-beta and causes a remission in the disease. Then people would really be happy both from an intellectual point of view and a human, practical point of view.

ARF: What should NIH be funding? Should the focus be on extending the A-beta hypothesis, or in addressing some of these major unknowns?

JH: Working on making transgenic models more effective and more easily distributed is one thing NIH is doing and should continue to do. What it's best at doing is funding research that's close to fruition. There are good experiments that can be done now to link A-beta to tangle formation. That's something they should be doing. The NIH in comparison to the European funding agencies have done a very good job. They've been proactive. I started working on AD in 1979 and it was kind of a scientific joke then. But now it's really leading into areas of basic science -- e.g. the role of presenilin in Notch signaling -- and it has really flourished because of NIH's policies.

ARF: What would cause you to doubt your hypothesis?

JH: My mind has changed over the last two years on the importance of plaques. Eighteen months ago, I would have said they were key. But we have an FAD family with no neuritic plaques, so that was important in changing our minds. Colin Masters also said he no longer thinks plaques are where the action is. My mind has also been changed on tangles. I thought they were kind of a side show, but the mutations that cause tangles show us tangles are absolutely on the path of neurodegeneration. Now there are things that don't quite fit into the hypothesis at the moment. For example, it's worried me for a long time that ApoE modifies age of onset in families with amyloid mutations, and also in Down's syndrome, but not in presenilin families. Now, why not? We're saying the effects of APP mutations and PS mutations are identical, yet clearly they're not modified by ApoE in the same way, so clearly we're missing something. There was a nice talk by Randy Nixon at Keystone which indicated that although the biochemical effects [of the mutations] are the same, they occur in different cellular compartments. APP occurs in vesicles going to the cell surface or at the membrane, while PS is in vesicles that never leave the cell. That fits with the ApoE data, because A-beta produced in the one would come in contact with ApoE, whereas the other kind would not.

ARF: There has recently been some interesting theoretical work by Paul Ewald arguing that diseases like Alzheimer's can't be primarily genetic because they cause enough of a negative impact in reproductive fitness that any genetic mutations causing the disease should be selected against. He thinks that a pathogen is involved. As a geneticist, what do you think of that argument?

JH: There have been a couple of papers about viruses and genetic susceptibility to AD. I'm not convinced by any means. ApoE4 is present in 15 percent of the population and is implicated in many bad things. So it must be doing something right. I suspect that might be something like enhancing survival on a low-calorie diet. I would suspect there's a plus side as well as a minus side. I suspect with ApoE it's got something to do with food intake.

ARF: I've also become interested recently in the link between cardiovascular disease and Alzheimer's.

JH: That's basically around ApoE.

ARF: And perhaps also things like estrogen, oxygen radicals and so on.

JH: That's clearly very powerful data. The nice thing about genetics is that the facts are both simple and very hard. It gives you hard pieces of data. The trouble with risk factors and epidemiology is it gives you very large amounts of soft data. I like small amounts of hard data. I'm not very good at judging [the epidemiological data].

ARF: Where do we fall short in tools or resources to enable us to understand this disease?

JH: The further you get away from DNA, the more chaos there is. That's definitely the case. I certainly don't understand cells to anything like the extent I feel I should. Just think how many people, maybe 2,000, have been working on APP processing over the past ten years. That's maybe 20,000 people-years, and what we know can be written on one side of 8 by 11 paper. That shows you how difficult cell biology is. How we are doing this is start with DNA and RNA and gradually move to protein. As the human genome project sweeps through, we'll hopefully get more information about the basic components of cells. When you look at a protein going through a cell, you only have a vague idea of what else it's bumping into along the way. We're moving in the right direction. In the next 5-10 years, we'll know all the components of a cell, and then we'll be able to do experiments properly.

ARF: Let's turn to the issue of therapies. At what stage in the pathogenic pathway do you think intervention is going to be most successful?

JH: You're dealing with something that starts simply and grows in ever widening events, so the best place to stop it is as close to the beginning as possible. I think your drug target should be to reduce A-beta 42. Not necessarily to knock it down. I think it's a threshold event, so even a modest reduction could shift the balance. You want to intervene as close to the top of the cascade as you can, basically.

ARF: Are there other important issues you want to address that we haven't covered so far?

JH: Yes. What's really struck me in the last year is the close relationship between AD and other neurodegenerative diseases. [See J. Hardy and K. Gwinn-Hardy] There are great similarities between AD and triplet-repeat diseases. Even more strikingly, the relation between the prion diseases and frontotemporal dementia and Parkinson's and Alzheimer's is really striking. Clearly there's very close pathogenic relationships between formation of tangles and formation of Lewy bodies. In my view, all form one family of disease. That's very exciting. We understand after the work on tau mutation the causes of progressive supranuclear palsy. Clearly it's a disease of the tau gene. It's very interesting that Lewy Bodies occur in AD and prion diseases. That's very exciting and very unexpected.

ARF: Is there a neuroanatomical pattern to prion diseases?

JH: The prion diseases are a mess actually. Very varied. GS syndrome is initially cerebellar, Jakob-Creutzfeld is cortical, and fatal familial insomnia is thalamic. The distribution of pathology in Parkinson's is confusing. The clinical features of PD are movement problems, and are explicitly to do with substantia nigra. If you had a disease that led to Lewy bodies in the cortex, you'd never call it Parkinson's because you don't get movement disorders. You get Lewy Body dementia. You have to be very careful because diseases are defined by their neuroanatomy.

ARF: Is there anything else you want to say?

JH: I think we've covered quite a lot of ground already.

ARF: Well, that was most interesting. Thank you very much for your time.

Selected References

Bahr BA, Hoffman KB, Yang AJ, Hess US, Glabe CG, Lynch G. Amyloid beta protein is internalized selectively by hippocampal field CA1 and causes neurons to accumulate amyloidogenic carboxyterminal fragments of the amyloid precursor protein.J Comp Neurol 1998 Jul 20;397(1):139-47. Abstract.

Johnson-Wood K, Lee M, Motter R, Hu K, Gordon G, Barbour R, Khan K, Gordon M, Tan H, Games D, Lieberburg I, Schenk D, Seubert P, McConlogue L. Amyloid precursor protein processing and A beta42 deposition in a transgenic mouse model of Alzheimer disease. Proc Natl Acad Sci U S A 1997 Feb 18;94(4):1550-5 . Abstract.

Hardy J, Gwinn-Hardy K. Genetic classification of primary neurodegenerative disease. Science 1998 Nov 6;282(5391):1075-9. Abstract.

Hutton M, Lendon CL, Rizzu P, Baker M, Froelich S, Houlden H, Pickering-Brown S, Chakraverty S, Isaacs A, Grover A, Hackett J, Adamson J, Lincoln S, Dickson D, Davies P, Petersen RC, Stevens M, de Graaff E, Wauters E, van Baren J, Hillebrand M, Joosse M, Kwon JM, Nowotny P, Heutink P, et al. Association of missense and 5'-splice-site mutations in tau with the inherited dementia FTDP-17. Nature 1998 Jun 18;393(6686):702-5. Abstract.

Prasher VP, Farrer MJ, Kessling AM, Fisher EM, West RJ, Barber PC, Butler AC. Molecular mapping of Alzheimer-type dementia in Down's syndrome. Ann Neurol 1998 Mar;43(3):380-3. Abstract.

Spillantini MG, Bird TD, Ghetti B. Frontotemporal dementia and Parkinsonism linked to chromosome 17: a new group of tauopathies. Brain Pathol 1998 Apr;8(2):387-402. Abstract.

We invite you to submit comments and questions on John Hardy's remarks.