Posted 12 December 2002
Interviewed by Gabrielle Strobel
Susan Lindquist and colleagues published two papers in Science last month, which provide the backdrop and introduction to this interview. See news summary.
ARF: You have studied protein folding of prion and other proteins extensively. Can you summarize some findings that might relate to Alzheimer's?
Lindquist: We don't yet have results directly relating to Alzheimer's; however, this is one of the directions I'd like to move into. My lab has been interested in protein-folding problems for a long time. It started out as interest in the heat shock response, a stress response in which proteins start to unfold when the temperature get a little high, and you need new proteins to help them refold or help take care of the protein-folding crisis. They're called protein chaperones or heat shock proteins (Hsps). Since then, we've become very interested in the biological consequences of protein misfolding in many different contexts, from signal transduction, to evolution, to human disease.
We're also interested in alternative protein-folding states that can actually serve as mechanisms of inheritance. These proteins are called prions. There is an obvious difference between yeast prions, which are a mechanism of inheritance, and mammalian prions, which are a mechanism of disease. In the one case it's useful, in the other case it's deadly. But there seems to be an underlying similarity in the process in that you have a protein that can exist in different stable conformational states. One of those states is rare, and it has the capacity to influence other proteins to enter into that same stateto join up for the ride, so to speak. In yeast, that actually acts as an element of inheritance, because when the protein changes its conformation, it also changes its function. In this altered structural and functional state, some of that protein is passed from a mother cell to the daughter cell. When the daughter starts to make her own proteins, they also will make this change in function. It's a self-perpetuating mechanism of passing on a new biological trait through many cell divisions. It's inherited not through a change in DNA; it's inherited through a change in protein conformation.
My initial interest in protein-folding diseases began with this same quality of protein misfolding that starts a cascade of consequences that inexorably follow from it. This seemed to be characteristic of several protein-misfolding diseases. Moreover, in some cases, there even seem to be some direct physical relationships. For example, a lot of these yeast prions, as we call them, are very rich in glutamines. The glutamine expansion, of course, makes one immediately think about Huntington's disease. Our ideal was to take the proteins deemed responsible for some of these diseases, put them into yeast, and use yeast as a living test tube to start to understand the mechanisms that govern this change in folded state. Because the yeast the quality-control systemsthe chaperones, the ubiquitin systems, the proteasomesall those things are really very, very highly conserved. So instead of trying to duplicate things in a test tube, or doing everything in a very expensive animal like a mouse, this was an intermediate place to be. We could take advantage of genetics and cell biology in yeast.
So, we started studying huntingtin and are actively pursuing that. We believe that the kinds of transitions in state that take place when huntingtin misfolds bear many similarities to these self-perpetuating yeast prion states. We're also studying a-synuclein quite intensively, too, though we haven't yet published anything on that.
The third mammalian disease protein we are working on is the mammalian prion protein, PrP. What we've found here points to a new view of the toxic mechanism, and also to a new view of how the self-perpetuating, transmissible forms of prion diseases might get started in the first place. Our findings all involved changes in quality-control systems, which constantly recognize a misfolded protein and dump it into the cytosol, where the proteasome will get rid of it. If the proteasome's capacity to chew up proteins doesn't keep pace with what's coming at it, you wind up accumulating cytosolic PrP. We found that was extremely toxic to neurons.
A question we are very interested in now is: what is the relationship of that toxicity to other kinds of neurodegenerative diseases? So little of cytosolic PrP is required to kill the neurons that you would never see it unless you're deliberately looking for it. We are interested in exploring the extreme of thinking that there might be a similar mechanism going on with Alzheimer's disease, maybe not just involving Ab but also other fragments of the amyloid precursor protein. They could be signaling inside cells that there's a problem, they could be involved in creating toxic species inside cells, or it could be the misfolding that is generated when Ab starts being clipped in the wrong place. There's a misfolding that indicates a more general misfolding event, and PrP could even be displaced and put into the cytosol and be signaling. We don't know how many of these things are related and how many aren't.
That's one of the places we'd like to go in terms of looking at Alzheimer's disease. We have not done a single experiment in Alzheimer's; we're just thinking hard about it. It's a difficult and complex problem. We started initially with other protein-folding diseases, which are also difficult and complex, but at least more tractable in terms of knowing that the protein is misfolding inside the cell. It was a good place to start for us because we understand protein folding inside the cell really well.
ARF: ...you mean the prion diseases?
Lindquist: First huntingtin and a-synuclein. We understand these proteins a little better because they clearly misfold within the general cellular environment rather than outside, unlike Ab. We started to study PrP because the yeast prion system is so similar, so we initially were taking advantage of mutants that act at different places in the maturation pathway to see what happens to PrP in yeast. We had a couple of surprising findings there that moved us to test our hypothesis in mammalian cells and transgenic mice. The current Science papers, with the previous PNAS study, were really our first foray into that world. It's very exciting to think that you can find something unexpected in yeast, form a hypothesis that might have broader significance, and then take several steps into mammalian cells and find out that, in fact, you might be right.
ARF: Does your recent work gel in your mind into a broader, unifying theory for neurodegenerative diseases?
Lindquist: Not clearly, yet. In my wildest fantasies, I think this could be related and that could be relatedand it could actually all be working through a similar mechanism. But I am still cautious. There is a very strong temptation to simplify things. It's so satisfying when disparate pieces of information that were otherwise confusing all of a suddenclick, click, clickcan be put together and make sense. I don't want to have that siren sing to me too sweetly and get carried away by it. I need to learn more about neurodegenerative disease.
ARF: I did want to encourage you to speculate a bit to help stimulate interest in protein misfolding among the AD field. How would misfolding occur and become pathogenic in AD?
Lindquist: I've speculated about it as far as I want to take it. This is pure, absolute speculation, and we intend to test it. I think it's important in this very complex, difficult disease to have new, fresh ideas and not be subsumed by dogma. Remember how difficult this disease is. A lot of really beautiful work has been done on Ab and on tau. Those were hard studies to do. There's a lot of interesting biology out there but it doesn't yet answer everything. So the question is: is there something about those two systems and those two views of the Alzheimer's world that just hasn't been put together yet and that will make amyloid and tau sufficient to explain Alzheimer's? Or is there actually some new, additional player that has to be taken into account?
ARF: And you would introduce a "procedural" new player, i.e. misfolding, rather than a physical character?
Lindquist: Procedural misfolding and/or some aspect of the quality-control system that's no longer processing this right. That could be the prion dumped into the cytosol, or it could be something that's happening in the endoplasmic reticulum. Misfolding there could lead to the change in processing. So many of these diseases have to do with aging neurons, and that makes you suspect that there's at least some common elements that might contribute to them all, perhaps a lapse in protein folding quality control.
ARF: Could you help us understand the relationship between the misfolding-proteasome connection as you see it on the one hand and endoplasmic reticulum (ER) stress and the unfolded protein response on the other? There has been quite a bit of work in Alzheimer's research on the latter two topics, much of it focused on alterations in AbPP processing and trafficking.
Lindquist: That's still unclear. I think there's no question that there is a change in AbPP processing. The question is, what brings that about? I would be foolish to speculate much about that because I just haven't, unfortunately, had time yet to steep myself in the literature. It does seem to me that we really do not understand what causes the change in processing in sporadic, late-onset Alzheimer's. It conceivable that, rather than being the initial cause, misprocessing is a consequence of some earlier thing that's gone wrong in the ER with in the protein-folding quality-control system. Even if that is the case, it doesn't mean that Ab itself is not the major player. But the reason this change in processing takes place could be because something about the quality-control system is no longer handling things well. That could back up into a problem in the ER or the cleavage site for Ab. The cell is a continuum; it's constantly sensing what's happening, and when something goes wrong in the ER, for example, a signal goes out into the nucleus to say: let's turn up the synthesis of the proteins we need in the ER. They are not separate worlds, they're intimately connected and constantly sensing each other. So I think that problems with protein quality control that might occur even in the cytosol would wind up feeding back into the ER.
ARF: If, indeed, there were to be misfolded cytoplasmic Ab at work in dying neurons in Alzheimer's, analogous to what you have found in the prion mouse, how could that be uncovered, given that such tiny amounts of at least PrP kill the mouse?
Lindquist: You'd have to do the kind of experiments we did with the prion protein. It is not a bandwagon that a lot of people want to hop onto because it's only got a small chance of being correct. However, if it were to play a role it would be very important to know.
ARF: A number of studies exist in the AD field that may bear on this. Frank LaFerla described apoptotic neurodegeneration in mice expressing intracellular Ab (LaFerla et al., 1995). There are transgenic mice by Lennart Mucke's group expressing a PDAPP minigene inside neurons, and they saw synaptic toxicity in the absence of plaques. And Gunnar Gouras had a paper last month showing that he detected aggregating Ab42 in the outer membrane of small endosomal vesicles called MBVs of AD-vulnerable neurons in human and rodent brain (see Takahashi, 2002, and live discussion). These MBVs apparently are linked to proteasome degradation. So there seem to be hints that�
Lindquist: ...our ideas might not be so outlandish. One of the reasons why I'm encouraged to think along these lines is that we have been continuously surprised by protein folding and the consequences of misfolding. It has led to so many different kinds of unexpected findings, and this is from our work in Drosophila, in Arabidopsis, and in yeast, as well as now in mammalian cells and whole mice. It is such a complex problem and there still is very little understanding of the real consequences of protein misfolding.
Just think of the evolutionary dimension. The cell is jammed full of proteins at the incredibly high concentration of 300 mg per ml. At 37 degrees, they are bumping into each other and jostling around at a high rate with a lot of kinetic energy, so the prospect for things going wrong are tremendous. Life couldn't have evolved without devoting a substantial fraction of total protein in the cell to helping other proteins fold and also helping them out when they misfold. Heat shock proteins, chaperone proteins and the ubiquitin and the degradation systems are ancient and have, therefore, been intimately tied up with many aspects of our normal biology. As you evolve new processes, those have to evolve in that same environment, and there are so many interconnections within this web of players that it's just a constant surprise right now. I think in another 10 years, we'll have a much clearer picture of it. Right now, you tweak on one thing or do a little something here, you very frequently wind up with a big surprise.
ARF: Chaperones are cropping up more frequently now in neurodegeneration. Just recently, a paper in the October Nature Genetics reports that a mutated tubulin chaperone causes motor neuron degeneration in a mouse model for a childhood form of spinal muscular atrophy (see related news story). This July, Chris Link reported that intracellular Ab interacted with several chaperone proteins in a C. elegans model of AD (Fonte et al., 2002). Could you elaborate a bit about chaperones?
Lindquist: Yes, there are going to be more and more examples like that. The reason lies in the way chaperones work. They recognize proteins that have a variety of problems with protein folding, and each chaperone seems to recognize a different kind of thing. For example, Hsp70 seems to recognize an extended polypeptide chain, where there would be some sticky hydrophobic residues that should normally be folded up and buried inside the core of a protein. Hsp90 seems to recognize more of a whole protein domain that is wobbly. The Hsp60 (GroEL) protein seems to recognize whole proteins or domains that are in a molten state.
And then, many of these chaperones work together in a variety of different complexes that probably use different kinds of recognition systems to recognize different kinds of protein folding problems in a combinatorial fashion. What they do when they interact with a protein that's folding or misfolding can vary depending upon the circumstances. One thing they can do is just bind to it and then let go again a little while later, and the protein will fold up on its own. They've bound to the sticky surface at a time when it could be dangerous. They let go when the protein folds up, and then everything's fine. If, however, the chaperones keep doing that and the protein is not behaving properly, there is what's believed to be a kinetic partitioning mechanism, by which the protein is more likely to wind up in the proteasome the more it is associated with the chaperones. So if you keep having futile cycles of interaction with Hsp 70letting go, interaction, letting gothen that protein's going to be shunted off for degradation.
There are all kinds of really interesting ways in which these quality-control mechanisms play against each other and monitor each other's activities. Proteins that don't fold well will wind up being degraded, and that can be part of their normal biology, as well as the abnormal biology of disease. Misfolded proteins that are not degraded will wind up being sequestered away in larger aggregates, and these might be dangerous as well as protective. There's one step after another after another. There are many different ways of taking care of a protein that is not folding well. That's what the chaperones are about and it is why they are beginning to figure in so many different diseases. If a protein is misfolding, a chaperone is going to be there. So, it's got to be involved.
ARF: Are you aware of other diseases that are, in a sense, protein-folding diseases?
Lindquist: Huntington's has many connections with chaperones. It is clear that the mechanisms for clearing misfolding huntingtin protein are very dependent upon chaperones. With a-synuclein things are not yet clear.
ARF: I vaguely recall reading that some hearing disorders are considered protein deposition diseases, and they are, of course, extremely prevalent in old age.
Lindquist: Yes, that is a very interesting story. I heard about it, too, but have not read up on it in detail.
ARF: Where does the b-sheet conformation come into the picture here?
Lindquist: Very good questionvery poorly understood right now. Is it the toxic fold or is it the fold that acts to sequester something else that's toxic? The appearance of that stuff is certainly associated with toxicity, but is it associated with toxicity because proteins that are passaging through a less structured toxic state often have a tendency to end up in b-sheets? Other possibilities are that the b-sheet itself could be directly toxic, because it could then interact with other proteins in abnormal ways. Or interactions between the same b-sheet forms could produce more and more plaques, and the plaques might be what kills you. In Huntington's, it is thought that same sort of b-sheet-like structure can wind up attracting proteins that don't belong in these aggregates. You lose the specificity of the aggregation process and wind up aggregating critical cellular factors, and when you titrate them out the cell dies.
One of the complexities may be that there might be different answers for different diseases. One always wants to unify, and one of my main motivating forces is to always look for a common element in things. But it just might be that in some diseases, the b-fold is a really bad thing, and in other diseases it is not.
ARF: Do you have suggestions for young investigators who are starting out and are looking around for a really rich question in Alzheimer's?
Lindquist: Boy, there are so many! I would say that having a much clearer picture of the way in which protein folding and quality-control systems interface with AD biology is extraordinarily important. Is it that these mechanisms are not doing their job, or is it that they're doing their job and then doing the wrong thing? Sometimes, the consequences of grabbing onto a protein and interacting with it can actually lead to the wrong state. Very, very big questions. Why aren't these things being recognized as being abnormal and just degraded? Could it be that the chaperones and quality-control systems are their own worst enemy? That they actually wind up causing problems as well as taking care of problems? It's just an incredibly rich field out there.
ARF: Thank you very much.
Lindquist: You're so welcome!