ARF: What is the hypothesis driving the research in your lab?

PL: It has always been that there is something bad about fibril formation. Now our work is focused on exactly where along this pathway we need to stop things and, in addition, how intermediate species cause cell death. That's a question many in the field avoid.

We are trying a chemist's approach: Eventually, we want to write out a precise mechanism that can be tested easily by two strategies. One is by making amino acid changes in the aggregating protein that affect certain steps but not others in the aggregation pathway. We are searching for extreme cases where fibril formation doesn't start at all, where it stops halfway, where it goes very fast. Then we want to put these into flies and see what happens in terms of neurodegeneration.

The second strategy we are pursuing is to find small molecules that stop fibril formation-again, stop at the beginning, stop in the middle, stop at the end-and treat flies and/or mice with those to look for an effect on the disease. So we are trying to find ways to disprove our hypothesis instead of trying to find ways to support our hypothesis, an approach many other labs are taking.

ARF: Is this the major way in which your approach is different from what is happening in the field of neurodegeneration?

PL: An obvious difference is that we're really reductionists here. We oversimplify the problem to a huge extreme. The reason is that we want to be very precise in the questions we can ask. That's a difference between the chemistry approach-and we're no longer the only lab taking that approach-and the predominant approach, which is really cell biology. It is that we can be very precise about what the status of Ab or a-synuclein is at a particular point if our test tube contains that protein and nothing else. Our problem is linking this to the real disease, and the way we want to do that is with the drugs, or small-molecule probes.

ARF: What do you say to critics who argue that the oligomers or protofibrils of Ab or a-synuclein have never been shown to exist in vivo, so that your hypothesis is not provable?

PL: First of all, that's a negative argument. I think we know that protofibrils are an obligate intermediate to get a fibril in vitro, so I believe with my soul, as a chemist, that they have to be there in vivo. Other people with different, medical training are taught to go with what's observed, so I understand where that comes from.

I think our work is slowly getting to the point where we will be able to start to address that issue. For instance, consider transthyretin amyloidosis. There is very nice literature clearly showing early-stage familial disease, with symptoms showing up prior to fibrils. These transthyretin protofibrils, or things that resemble protofibrils, have activity in cell culture, also. I think Eliezer Masliah's mouse model for Parkinson's is another great example, where there are no fibrillary inclusions.

There are species that are not fibrillary. We do not know that they are the same things we observed in vitro, but we strongly suspect it. Recently, we have isolated discrete protofibril structures; we are trying to purify them now. Besides, going back into the literature on brain-derived Lewy bodies and looking at electron micrographs, we can see these structures clearly. They have been ignored because, when you have different types of structures on a picture, and one of those is a giant fibril, it's very easy to see the rest of it fading into the background. So we actually have material from Lewy bodies and collaborations and we're trying to address that particular problem.

ARF: What are these structures that you refer to?

PL: Right from the beginning, we saw protofibril structures that looked to be spheres, although it's very hard to find those. All of this, by the way, is common to Ab and a-synuclein. We also see those spheres anneal into what we call chains. At some point, when those chains reach a certain length, we very rapidly see fibrils. So we believe that those chains sort of wrap around each other, a bit like a double helix. The dimensions of those are consistent with that model. Once the chains reach a certain point, there is enough interacting surface to drive that process and you see fibrils immediately. Once you see fibrils grow, all these smaller species just disappear.

All of these things are oligomeric, none are fibrillar. All of them have ordered secondary structure in contrast to the monomers, which are disordered. They already have b-sheet structure, which is seen in the fibril.

We have done a lot of analytical ultracentrifugation to measure their sizes, and they all have discrete sizes. Really, we don't see any evidence for a populated species until you get to about 15 to 20 in the case of a-synuclein and around 40 in the case of Ab. Of course that doesn't mean that smaller oligomers don't exist. It means that as this protein oligomerizes, it is not stable enough to accumulate until it hits maybe 40. At that point, for some reason, there is an energy minimum in which 40mers accumulate. Upwards of the 40mers, we clearly see a progression, a non-random smear of intermediates.

ARF: What technique are you using for these observations?

PL: That data is mostly from atomic force microscopy. We now focus on this population of spheres and chains. Working at low temperature allowed us to separate these species by gel filtration. It looks to us like the spheres form chains, which can essentially bite their own tails to form rings. We reported on such annular species last year (Volles et al., 2001) and have many more data on them now.

They actually have some stability. It turns out you can isolate, let's say, a spherical species, measure their height, and find their height falls into a Gaussian distribution. When you re-chromatograph those spheres, they stay the same. Now if you add a lot of monomer, they will progress to the next stage. We can work with these things independently to make them progress to fibrils very rapidly.

Now that we have these discrete species, we are trying to determine their structure at a higher resolution. With a more homogeneous group of structures, you can do things like image reconstruction by electron microscopy. We are averaging the electron micrographs of thousands of a given species to get a better resolution. We see clearly that this sphere of 20 to 40 molecules is retained in all the subsequent structures, it looks like beads on a chain.

Now that we think we can structurally characterize these, the next question is: what do they do? In a simplified, in vitro model of vesicle permeabilization, we have observed that these species can act like pores (Volles and Lansbury, 2002). Unlike apolipoproteins, for instance, which bind on the surface and can be washed off easily, these subsequent species insert into vesicles. They allow permeants to move from the inside to the outside. This pore-like activity is sensitive to the size of the permeant. Small things pass rapidly, but as you increase the molecular weight of the material trapped inside, there is a sudden point where it won't flow out anymore, so the protofibrils make a hole of a discrete size. We are now trying to connect in-vitro permeabilization activity and in-vitro structures. Obviously, if that were happening in a neuron, that could be very detrimental.

ARF: Are you suggesting a toxicity mechanism as simple as poking holes into neuronal membranes and disturbing membrane potentials, ion imbalances, and the like?

PL: Yes, there are a million consequences you can imagine: puncturing a mitochondrial membrane and causing all sorts of ion- and ATP-related problems, to name just one. I don't focus on that part because I believe many events flow from this initial one. For example, mitochondrial pores might lead to activation of caspases. There are probably other damaging consequences here. I'm more interested in how it all starts.

The next step for us is to microinject discrete structures. In a fascinating paper, Andrea LeBlanc showed Ab, when microinjected into cortical neurons, is vastly more toxic per mole than it is when applied to the medium extracellularly (Zhang et al., 2002). Also, that toxicity depended on Ab being oligomerized, but not Ab being fibrillized. Plus she had electron micrographs that led us to believe she was looking at the structures we also see. So we will microinject discrete fractions of Ab and synuclein oligomers that we can characterize structurally and assay their in-vitro toxicity.

Yet even that will not silence the criticism I hear routinely. To do that, we must find molecules that block the formation of that discrete oligomer, and such molecules can then be tested in an animal model of the disease. I believe we may never directly observe these intermediate structures or extract them from brain; there are many reasons why that will be extremely difficult. But if we can deploy a panel of maybe 20 compounds to show that we can predict how a particular compound will affect the progression of disease in an animal model based on how it affects the oligomerization of the protein in a test tube, then I don't think there'll be much doubt.

ARF: If you could pin toxicity on, say, the annealed ring structure, and had an inhibitor against its formation, in which systems would you use these to test if you can induce, and then block, death of the right types of neurons?

PL: In Parkinson's, we will use standard midbrain cultures, where you have dopaminergic neurons of various types mixed with non-dopaminergic neurons. We'd like show that we can rescue this selective toxicity. Our approach is the reverse of that many other people are taking, where they over-express the protein in an animal or a cell and then work backwards to understand its role in pathogenesis. There are many problems in dealing with the protein introduced in this way, so we are complementing that approach by making a protein in a test tube and working our way forward.

ARF: Outline for us what you think are the steps leading to Parkinson's and Alzheimer's, and where you see overlaps.

PL:  I want to keep an open mind but if I had to bet, I'd say both are initiated by these small protofibrillar species. By that I simply mean structured oligomers, whose formation is obligate in the fibrillization pathway. An interesting parallel is that mutations in the aggregating proteins themselves can promote this initial process. Currently, I am very interested in the arctic mutation in Ab, where the mutation lies right in the sequence that can promote this.

Another way of promoting the formation of these species is to make more of the protein. Most of the AbPP and presenilin mutations have that effect, and there also is a polymorphism in the synuclein promoter that hasn't been fully tracked down but may well affect the expression level of synuclein. In both diseases, if aggregation is at the root, making just a tiny bit more could be a severe problem. That's a key point I am trying to make about aggregation: it is not a linear thing.

Then there is the degradation flipside. Some people are interested in IDE, neprilysin, and other Ab-degrading enzymes. In Parkinson's, both parkin and ubiquitin C-terminal hydrolase (UCH), which are the two other gene products clearly involved besides a-synuclein, have a role in degrading a-synuclein. If you make too much, degrade it insufficiently, or have a mutation that changes its critical concentration, all of those things would allow oligomerization to occur.

I think what probably causes the cell death in both of these diseases is a species that may represent around 1 percent of the total protein. This appears logical. a-Synuclein is a very highly expressed protein, so if you follow the bulk of it, you are following its normal function. I think one must follow a very small percentage of it. To touch on a related debate, I actually think Ab cytotoxicity might also be intracellular, or there may be several sites of toxicity, one being cytoplasmic. There may be mechanisms by which AbPP is incorrectly inserted or misfolded, then it is pulled back into the cytoplasm by the quality control system and the proteasome takes over. In terms of sites of toxicity, there is perhaps not a direct analogy. Even so, I feel strongly that in both diseases, a really small oligomer is subject to this critical concentration that is driving the subsequent steps. The subsequent steps eventually kill the neuron in many different ways.

I believe these diseases are just accidents. They are not evolved pathways for cell differentiation. It isn't cancer. It is a degenerative disease invisible to evolution, so there are probably many different pathways set off by the trigger. I am really interested in therapies. There, I think targeting those many pathways is a losing proposition, because once you've blocked one, five others will take over. If you nip that first step by lowering the protein's expression, preventing it from oligomerizing, or increasing the cell's degradative capacity, you can treat both diseases.

ARF: What similarities and differences do you see between prion diseases and protein aggregation in Alzheimer's and Parkinson's?

PL: The big difference is the transmissibility, a separate issue. What makes the prion protein able to transmit information may be different from what makes it pathogenic. If you just look at genetic forms of Creutzfeldt-Jacob disease that are analogous to familial AD or PD, there again you have a protein that aggregates. It is possible that the mechanism of pathogenicity of the familial prion disease may be related. Many people are starting to think membrane insertion may be the problem. Susan Lindquist has data in prionic yeast on the mechanism whereby misfolded proteins are meant to be secreted. When the secretion pathway for a misfolded protein fails, that material is then targeted back to the cytoplasm for degradation. If there was a process of aggregation competing with degradation, you would start to have a problem.

There is also a paper now on the prion protein itself being capable of permeabilizing membranes (Sanghera and Pinheiro, 2002). I believe these people are studying oligomers of the prion protein without being aware of it, just as there have been 50 papers on Ab permeabilizing membranes. Very few, if any, recognize that the reason it's permeabilizing membranes is because there is a small amount of the oligomer in their solution and they are unable to detect it because it is not fibrillar nor does it scatter light. It could even be that PRP scrapie--the fibrillar, protease-resistant, diagnostic form--may not be the pathogenic entity. Stan Prusiner has unpublished data that PRP scrapie does not correlate with infectivity. That is different, because you have infectivity and pathogenicity. But this notion that the form you see at the time of death-the most stable, fibrillar form-might not be the one that you should be studying is out there in the prion community, too.

Finally, there are transgenic models where cells start to die before any PRP scrapie becomes evident. David Harris has done this beautifully, showing that PRP in the brain was changing its form; it was clearly becoming more resistant to protease and more insoluble (to use this word in a purely operational sense), but it wasn't PRP scrapie yet even while cells were dying (Chiesa et al., 2000; Chiesa and Harris, 2001). This is analogous to Lennart Mucke's studies in his mice, which I think are well done, and Eliezer Masliah's studies on synuclein mice. So to really push it, I say protein aggregation into ordered aggregates is a really bad thing--ordered being the key because then the aggregates are likely to bind things that you don't want them to bind.

ARF: So that's the commonality, but subsequent mechanisms of cell death are cell-specific and diverse?

PL: They could be highly cell-specific. I use the analogy of breaking a bone. Your kid falls down the stairs, and then the phenotype of what happens depends on what bone they break. There may be things that are just different about certain cell populations. As I said, this is partly accidental.

ARF: Do you have an opinion, then, on what accounts for the anatomical progression of AD? You know that cell death starts in layer 2 of the entorhinal cortex and then "spreads" to other brain areas. This varies from patient to patient but, overall, there is a common pattern.

PL: No, I don't have an answer. To me, it's just clear that this process is going to be favorable under some conditions but less so under others, and each cell in the brain will fall along a spectrum. If, for argument's sake, you imagine initiating the aggregation process in every cell in the brain to an identical extent, it would not kill every cell at the same time. (In reality, initiation events may well be cell type-specific.)

This argument always comes up. But would anyone expect every cell to die at the same time? Every neuron is different; that's the whole beauty of the brain. So why would you expect that they die at the same rate? I think this pattern is just a rate issue. The most vulnerable neurons die first, and as you progress through the disease cell types die that are less vulnerable.

ARF: So this differential vulnerability is not an issue that interests you very much?

PL: I think it has to be explained in every model. But the fact that all neurons are not affected equally does not strike me as unusual. To me, selective vulnerabilities are a secondary issue.

ARF: Where does tau come into your picture? Not to revisit the tau versus Ab debate, but ultimately, they are both going to be important in some way. How do you think they interact?

PL: That is right, ultimately they are both going to be important. Isn't it odd that, of all the flavors of neurodegenerative disease, there is a tau-aggregate-only flavor, there is a synuclein-only flavor, but there is no such thing as an Ab-only flavor.? Ab either comes with tau in AD or with a-synuclein in the Lewy body variant of AD. This opens up the possibility that Ab can be secondary. But I think it is more likely that these pathways are synergistic. Masliah and Mucke did beautiful studies where they crossed the respective transgenic mice and showed synergy (see related news), and now people have crossed Ab transgenics with tau transgenics (see related news). I think all of these independent pathways can combine in different ways to produce many different consequences in the cell. Again, you would expect each of these combinations to have different cell-type selectivities, because they're slightly different processes.

ARF: Have you studied oligomeric, protofibrillar tau aggregates?

PL: We have looked at a piece of tau called K19, which has, I believe, three microtubule-binding domains. The Mandelkows first made it, and I feel like I don't have to worry about tau with them around because they have done such great work. We do see pre-fibrillar oligomeric species but have not pursued that further. I decided to focus the lab on synuclein and the Parkinson's system for the next couple of years partly because there is a clean animal model, the fly, which we can use to test our hypothesis. Except for a peripheral interest in Ab and some other things, most everyone in the lab is working on synuclein aggregation. Ultimately, our case is going to be circumstantial, based on getting probes and variants that all are predictive. So the more we get and the faster we get them, the better we'll be able to prove our point.

ARF: What key bits of evidence are still missing that would convince you of the correctness of your hypothesis? You explained isolating the toxic intermediate species and using inhibitors for its aggregation to induce, and then block, neurotoxicity? Anything else?

PL: I think what you want to see is the whole intermediate, everything between the unstructured monomer and the fibril. For instance, people using other approaches-like Mel Feaney using fly genetics to look for enhancers and suppressors-should see things where they can separate fibrillization from disease. Enhancers should block all fibrils from forming, and suppressors should promote fibrillization. There are isolated examples of those in all of these diseases. And, of course, all of this should translate back into the test tube. We have done many experiments with mixtures of b- and a-synuclein that make me very confident. You cannot prove a mechanism, you can only disprove it, and so that's what we're really trying to do. Our drug discovery group should help us find compounds that inhibit aggregation.

ARF:  Is it difficult to publish papers if you structure your experiments as trying to disprove your experiments?

PL: That can be a problem. There is a perception in the field that editors want something incredibly novel, so people create a straw man around their results. This happens in the aggregation field all the time, it drives me crazy. But if we can do the right experiments, and if we demand as a group that experiments to disprove hypotheses are done, the true data will eventually stand out. I also think, though, that journal editors should be more careful not to promote this kind of politicking.

ARF: Is there existing data that you have trouble incorporating into your hypothesis?

PL: {laughs} Trouble-I have no trouble doing it. Where I have to wave my hands...?

ARF: that would tend to disprove your hypothesis, that just doesn't fit in, but that nevertheless you take seriously?

PL: Okay, I'll tell you one. I believe my hypothesis is a simple explanation for all neurodegenerative diseases, including ALS, but why don't we see inclusions in patients with sporadic ALS? That's a problem. In familial ALS patients, you see inclusions that are SOD-positive, and SOD is a mutant protein, so that fits well. But, in all these other diseases, you also see inclusions containing the wild-type protein in patients who have sporadic disease. Why aren't those in ALS? It could be that they haven't been looked for in the right way, or that the epitope is shielded...there is a bunch of explanations, but I am stumped on this one, because inclusions should be there.

Also, I have wondered why there isn't a sporadic form of Huntington's? In the other diseases, there is a rare, highly penetrant genetic form, and then there are some risk factors that predispose, that exaggerate the tendency. In Huntington's, there aren't. Of course this depends how you define it. Repeat lengths between 34 and 37 glutamines are less than 100 percent penetrant, so I guess you could call that a susceptibility factor for a sporadic form. It would be interesting to look more carefully at those patients who are right on the border of the pathogenic repeat length. But that doesn't really fit well.

ARF: What kind of evidence would convince you that your hypothesis is wrong?

PL: If I had a sequence variant that did not oligomerize at all in vitro yet was extremely pathogenic. Such non-oligomerizing variants are exactly what we are looking for because we believe they will be curative. But if they are not curative, that would kill the hypothesis. That is why I think those are good experiments. In fact, we are collaborating with Susan Lindquist to have synuclein inserted into the yeast prion sequence in order to identify sequence variations that have extreme properties. So if we found a synuclein that was really solubilized, prevented all aggregation, but was extremely pathogenic, that would be deadly. Or a compound that did the same thing.

On the flip side, if I had a mutation or small molecule that allowed oligomers to form and accumulate but blocked fibril formation, now if that turned out to cure the disease, that would bother me, too. We would argue those compounds will induce the disease or make it more severe. The things that will convince me are experiments I am hoping we will do.

ARF: It's a nicely staked-out approach. What approach to drug development do you consider most promising at this time?

PL: I hope our lab will contribute hypothesis testing, perhaps even more, to drug development. One approach would be to prevent formation of these ordered oligomers. The industry dislikes it, because it's unlike any other target they have shot at. They prefer receptors or enzyme active sites where you get a one-to-one interaction, can measure dissociation constants, model a complex, and then look at it on a screen at a managerial meeting and be confident that pushing ahead is the right decision. This is different; the species you target is some sort of amorphous, evolving structure in some sort of blob, and it's a difficult thing to visualize. The stoichiometry is unclear, the binding coefficient is unclear, and it is going to be an intracellular inhibition. But we are pushing it. You can also inhibit oligomerization indirectly. For instance, in the dopamine adduct case, we find that if you introduce a form of synuclein with a dopamine covalently attached to it, that form acts as an endogenous inhibitor. Every crystallographer knows that even a tiny impurity makes it hard to get that final, ordered thing. So we thought about adjusting the level of cytoplasmic dopamine.

ARF: Do you think it is practically feasible to use dopamine toward modifying a-synuclein aggregation? Obviously this is appealing because L-dopa is already a therapy, but maybe you can't tinker much with dopamine levels because you need it to be just right for synaptic transmission.

PL: Right. We think cytoplasmic dopamine is the one that would promote protofibril formation. The idea would be to inhibit dopamine biosynthesis somewhat and at the same time promote its export, so it is not lingering in dopaminergic vesicles in the synapse and leaking back out into the cytoplasm. One might be able to design molecular sponges that chemically sequester cytoplasmic dopamine. Glutathione would be a good example. As you know, glutathione levels drop as you age and are way down in Parkinson's, so there are many ways of thinking about that strategy, too.

If you can find out why substantia nigra neurons are selectively sensitive and then just took away this sensitivity, you would be fine until you had developed diffuse Lewy bodies. The cortical neurons would be next, but at least you could delay onset of nigral degeneration. We are thinking about that strategy. We are also thinking about promoting degradation. I like UCH because we have a compound promoting its expression. If we could get such a promoter for parkin also, that would be nice. Those are usually not attractive therapeutic targets because it is easier to inhibit something than to promote it. So both parkin and UCH might be tough in that respect.

As another method, we are also thinking about adjusting the ratio of phosphorylated synuclein to non-phosphorylated synuclein by focusing on an extracellular receptor phosphorylation, because that may affect the oligomerization pathway in interesting ways. I think Masliah's b-synuclein work is about exactly that: b-Synuclein looks sufficiently like synuclein that b-synuclein prevents all oligomerization of a-synuclein. So adjusting the expression of b-synuclein would be the strategy, and we have the b-synuclein promoter cloned in a system that can be screened by high throughput and are ready to go with that. We are screening for compounds that promote expression of b-synuclein. Instead of delivering your drug in there, try to use an extracellular or nuclear approach to…

ARF: ... induce b-synuclein expression as an endogenous inhibitor of a-synuclein aggregation?

PL: Yes. Take hydroxyurea: it induces expression of fetal hemoglobin, a gene that is silent after birth. In a test tube, fetal hemoglobin inhibits the fibrillization of mutant sickle-cell hemoglobin. This strategy was discovered by accident, but we are really interested in the mechanism behind it-using an endogenous protein to act as the inhibitor rather than trying to get the inhibitor into the cytoplasm itself.

ARF: Though generally, the pharmaceutical industry goes after an enzyme activity that it tries to inhibit with a small molecule, there is precedent also for small molecule drugs that induce genes expression? At a recent meeting I just saw one that was made against an NMDA receptor and turned out to induce BDNF expression.

PL: I am very excited about BDNF. I have sketched out the simple core of our model, but a huge number of things affect the rate of progression of these aggregation steps, and all of those are useful therapeutic targets. We will find that many of the other leads that people have are going to play into the same thing, like they have been in the Alzheimer's amyloid area. There you look at head trauma and find it induces Ab expression; homocysteine, estrogen, cholesterol, they all appear to affect Ab levels. I think many of these other things that are now hovering out there seemingly unconnected might actually all work in similar ways. Our lab is trying to focus on the central things right around that synuclein pathway.

ARF: Anyone entering the neurodegeneration field-Ph.D. students, postdocs looking to build a career-must be forgiven for being initially confused by the myriad different research directions. Any advice for novices on how to pick a promising niche?

PL: Don't listen to me. {Laughs} Don't listen to anybody who thinks they know the answer, because if they did, it would be over. No one really knows. It is important not to follow the crowd, especially in this field where there's a giant favorite. It's like my daughter's soccer games: the ball squirts over and then all these kids run there and it's just a big pile of dust! So I am the last person to give advice, except to say think of different ways to go, different explanations, and follow your own ideas.

ARF: Finding a rich area is a challenge for someone just starting out.

PL: I am hoping in a few years I will do that again and move out of this field. I am happy so many people are interested in protein aggregation now. At the same time, I feel like my job is done. There are enough good people who are going to figure this out.

Many practical issues make it hard for young people to be bold, different, and creative. But if you don't do it when you are starting, you are doomed to a life of never doing it, because the system will train you to do experiments like everybody else. So if you listen to what people say, what study sections say (especially when it is "Well, that's not a productive....") you will never get anywhere. Better to try to do something completely wild and make a go of it!

ARF: Some researchers say Alzheimer's can have many causes, or even that familial and "sporadic" AD are different diseases. Do you think there is going to be one or a very few key pathways in Alzheimer's pathogenesis that will explain the disease and make good, non-redundant targets?

PL: I think Ab aggregation might be the one. However, if you take clinical Alzheimer's disease, I would never argue that 80 percent of patients will be responsive to an Ab-lowering agent. That's crazy, and not true of any disease. So the answer to your question is probably no, there will be many flavors. Increasingly that is so for most diseases, most notably cancer. Therein I see a big revolution coming: There will probably be, let's say, 10 targets, each will have 10 drugs, and then some combination of those targets will work for you. Hopefully, we will be able to find out to which you will respond best by typing you in some way, rather than by trial and error. I saw an article in the Boston Globe today, where scientists used EEG to predict which patients would respond to antidepressants (Cook et al., 2002). I think that will happen, too, in Alzheimer's. It would be great to be able to image the response to experimental drugs soon after people begin taking them, rather than having to administer them for months before performing the Mini-Mental-State and similar tests.

ARF: Going downstream from aggregation, many scientists feel that the ensuing pathogenic pathways ought to be clarified, both to generate more targets drug companies can go after, and to generate biomarkers that epidemiological studies, for example, can include in their observation of people. On this point, do you think there will be a few key pathways?

PL: No, I don't. I think there will be many targets. Some of them could be useful therapeutically, especially in affecting the progression rate. I choose to focus on the start because I think it is the optimum place to interfere. As things diverge, you have more and more targets, and as you block one pathogenic pathway, another one may take over. The simpler part of the pathway may be at the start, making it easier to hit the home run than to hit all those singles. Even so, figuring out downstream pathways is incredibly useful and important, because many people will not respond to Ab therapy for some reason. The clinical diagnosis of AD still has a 20 or 30 percent error rate. All those patients could have no problem with aggregation, and that is a lot of people. I think all of those things need to be looked at.

ARF: What sorts of research or technology areas outside of Alzheimer's or Parkinson's would have much to offer if they were brought to bear on the problem of neurodegeneration, if people in the field embraced them more?

PL: I am glad you asked this question. I believe the big thing in Alzheimer's is presymptomatic diagnosis. There is incredible technology out there to do the opposite of hypothesis-driven biomarker research. Hypothesis-driven biomarker research measures tau levels, or phosphorylated tau in CSF, or a combination of that and Ab in plasma. That is useful and could be coming.

But I am saying methods now exist to look at, let's say, the proteome in urine. Suppose you had a quantitative measure of the proteome in urine and then, by using only informatics technology, analyzed that to recognize patterns. Those patterns will be completely outside of our ability to understand in the mechanistic way that we like to understand things. But all the same, the mere patterns might give away incipient AD. By analogy, take the clinical neurologist, who has a bunch of things not defined at the molecular level that nevertheless go into the diagnosis.

So I would like to see the Alzheimer's community reach deeper into the proteomics community. May of these guys come and talk to me. Various aspects of Alzheimer's disease turn them off, such as that the clinical diagnosis is only 70 to 80 percent accurate, so you have to have pathology in order to confirm a diagnosis.

I think the Alzheimer's community must work together to solve the problems of getting tissue collection done with very strong clinical histories, and also to look at more homogeneous patient populations.

There also ought to be an effort to reach out to facial-recognition technologies and surface-size technologies, where people can array things and observe changes in arrays based on some unknown. I think that would be huge, and it could be huge really quickly.

By contrast, the search for a biomarker or surrogate marker for Alzheimer's is going to be unbelievably slow. Maybe there is no single biomarker, just like for most Alzheimer's patients there is no single gene. For most patients, it is a complex combination of 20 variables, and no one in our field can permutate 20 in that space.

I know many groups that have high-tech diagnostic technologies and drug discovery technologies, but Alzheimer's disease scares them off. Part of what repels them is an impression of the field being divided and contentious. I think that it is not necessarily true. If you look at the good papers, it is not so divided. This impression is created by people who spend their time trying to make it be more contentious because they think this helps them publish papers and distinguish themselves. So people from the outside look at it and say "My goodness, this is a mess, I am not going to get involved in this." The field is shooting itself in the foot because it is preventing fresh people from becoming interested, who instead go to work on cancer, inflammation, or anti-infectives.

ARF: What can AlzForum do to bring these kinds of technology people together with willing Alzheimer's researchers?  

PL: A lot, it just has to be done in a big way. First you need to get people to share and say it is more important that my samples go to a bigger cause than it is for them to serve me alone. That is a difficult argument to make. It has to be done on a large scale so that enough people participate for the selfishly driven people so say: "Jeez, it's in my best interests to participate also." When I was just starting out, a very famous biologist advised me to stay away from Alzheimer's disease at all costs, that getting involved in it was suicide. His reasoning was clear: It is a mess, the biology is complicated, there is no clear progression, there is not a will in the community to make things work and get a convergent hypothesis off the ground. You know that is true. Instead, there is a drive to make things more complicated, or to set up straw men, or to split hairs more and more so that everyone can have their own version of the hypothesis. Instead we should be saying "As a first approximation, let us think that maybe everybody is sort of right."

But I think this convergence is going to happen. For instance, the Harvard Center for Neuroregeneration and Repair is confronting these issues actively in discussion. You realize the whole culture of academic science is really the antithesis of the culture of getting to an answer. There are people who don't want to get to the answer unless it is theirs. And, of course, if 49 of 50 people are working against the truth, well, the poor guy who knows what is going on will get squashed. I don't know how to solve that.

One thing the Parkinson's community has going for it, and in the Alzheimer's community it is an obvious problem, is that we have involved patient activists. There are many young-onset Parkinson's patients, and they really turned around my thinking of how I wanted to do science. I have patients calling me regularly to ask how the research is going and chat about how they are doing. That concentrates the mind of people who, like me, don't see patients and miss that inspiration. In Alzheimer's, that is a catch-22.

So active patients can drive people to do the right thing. I think institutions can. Our dean, Joe Martin, is terrific at that. He wants it to happen and understands the problem. But people must be rewarded for working together.

ARF: And yet, universities still make tenure and other important decisions in the old way...

PL: Yes. It is so different in industry. There they love to kill a project because they hate for projects to linger inefficiently in their pipeline. Needing to focus their resources, they love it when people show something is not going to work. But in academia, if I read a paper and think there is something wrong with it, there are so many pressures to not be open about it. There is nothing in it for me scientifically to go out and reallocate my resources in order to tackle that problem. I frequently approach the author, but that becomes a political headache. Even if I do it gingerly with an offer to provide resources to solve certain technical problems, say, it makes you enemies and there's no reward. It's a balancing act: On the one hand, institutions and funding agencies must recognize when someone is truly an innovator on a creative path, and let them totally go. On the other hand, the community has to crack down on people who are just propagating erroneous ideas without asking hard questions. For one thing, the field ought to spend three times longer reviewing papers, and really force people to clean up their act.


Make a Comment

To make a comment you must login or register.

Comments on this content

No Available Comments

Comments on Primary Papers for this Article

No Available Comments on Primary Papers for this Article


External Citations

  1. mouse model
  2. Volles et al., 2001
  3. Volles and Lansbury, 2002
  4. Zhang et al., 2002
  5. Sanghera and Pinheiro, 2002
  6. Chiesa et al., 2000
  7. Chiesa and Harris, 2001
  8. see related news
  9. Cook et al., 2002

Further Reading

No Available Further Reading