28 March 2000

Interviewed by Chris Weihl

ARF: What is the primary hypothesis that guides the research in your lab?

Wolozin: My laboratory's research is focuses on two different projects: presenilins (PS) and alpha-synuclein. With regard to presenilins the primary hypothesis that we are following is that the Aβ processing activities can be separated from the signal transduction activities of PS. Therefore it might be therapeutically possible to inhibit the pathogenic properties of PS1 without perturbing the other signaling functions associated with PS.

I guess more fundamentally, I essentially believe in the Aβ hypothesis. Presenilins cause familial Alzheimer's Disease (FAD) primarily by elevating Aβ1-42. Although, I think that there is room for the possibility that other effects of the mutations in PS on signal transduction might impact on particular FAD courses, perhaps by determining the exact age of onset.

ARF: So, your hypothesis suggests that mutations in PS cause FAD by increasing Aβ42. This effect is independent of PS effects on signal transduction cascades. Are these two functions inter-related functionally?

Wolozin: My sense is that a general hypothesis for what PS does in the cell, is that PS is an integrator between signal transduction and proteolysis. PS plays a key role in the proteolytic regulation of Notch and beta-catenin as well as APP. PS1 may be the secretase that cleaves APP to generate Aβ and may also be the protease that cleaves Notch. However, in the case of beta-catenin, PS1 is not directly cleaving beta-catenin but controls the degradation of beta-catenin possibly by bringing GSK-3-beta in to contact with beta-catenin to regulate its phosphorylation or, alternately, by simply determining whether beta-catenin is bound or free.

ARF: So then is AD a disease of protein catabolism resulting in protein accumulation and aggregation?

Wolozin: Yes. I teach a course in neurodegeneration here at Loyola and what I stress is that in disease after disease the same theme appears to be repeating itself. That is the aggregation phenomenon. The specifics of a particular disease vary depending on the particular protein that aggregates. I separate the process of aggregation from the function of the protein by analogy to a car door. You can see how the car door works, and understand how the door functions, but when the door is broken because it has a rusted hinge, the process of rust accumulation has little to do with the function of the door.

ARF: It is interesting then that PS itself does not aggregate but instead causes other proteins to accumulate and aggregate. This is not like other disease-causing proteins such as APP, prion protein or triplet repeat proteins.

Wolozin: I have a review in press in the Archives of Neurology (co-authored with Christian Behl) and in this review the term use to describe these diseases is "gain of aggregation" (GAG diseases). People typically think of autosomal dominant genetic diseases as resulting from a gain of function — there must be some new function that causes the disease — but I think that the way to conceptualize the disease-causing phenomenon is that the new process is the gain of an increased tendency for protein to aggregate, which is often separate from the normal biological function of the particular protein. In the case of PS, I agree, it is not the direct protein aggregating, but it is increasing the production of a pro-aggregative species. In that sense it is somewhat indirect.

ARF: So how would a gain of aggregation cause a neurodegenerative phenotype and lead to cell death?

Wolozin: The actual mechanism of death is not known in many of these diseases. However, in the case of prion protein mutations that cause familial CJD, the mutations increase the propensity for the protein to aggregate. We are doing similar work with synuclein. The mutations in synuclein that cause familial Parkinson's disease also increase the tendency of synuclein to aggregate.

Actually in the case of Huntington's disease, the whole picture isn't there; aggregation of the Huntingtin protein occurs but it is unclear whether the aggregates cause the toxicity. So in general one can say gain of aggregation, but it is not entirely clear at this moment that the final aggregate is what precipitates the disease. The same is true in ALS where when you over-express mutant SOD you get an aggregate. But that doesn't actually appear to be what causes the disease. So there are specifics that are not known.

ARF: In the case of these diseases could there be undetectable, microaggregates that are causing the disease?

Wolozin: We are actually looking at that. In fact in most of these diseases what happens is a preliminary nucleation event. Then you develop smaller aggregates that increase in size. Ultimately they lead to the final inclusion that we associate with the disease. But I suspect that what is actually causing the damage is these microaggregates. One example is Bill Klein's ADDLs, where these Aβ oligomers are very toxic (see Panel Discussion).

The simple kinetic analogy which may or may not be true is as follows: if you take the example of Aβ or synuclein, they bind to alot of different proteins. Some of the things that they bind to are toxic. But it's not important because the affinity, for example to the p75 receptor or RAGE receptor is low. However, when you make a dimer, trimer or tetramer of Aβ, you essentially increase the number of binding sites. So the affinity for the receptor can potentially increase. This becomes important when aggregation increases the affinity of Aβ for molecules associated with cell death.

Another possibility takes into the account the work of Ashley Bush who has shown that these aggregates bind metals, leading to increased toxicity . This may be also be the cause of cell death.

ARF: So if these diseases are all caused by a gain of aggregation, have you seen similar mechanisms of cell death in your cell models of AD and PD? For example, Aβ-mediated cell toxicity and synuclein-mediated cell toxicity?

Wolozin: Yes we have. When dealing with cell culture models it always depends upon the cell type you are using. What we have seen is that Aβ certainly increases oxidative stress. We have also observed that synuclein increases oxidative stress, especially the mutant forms. So I think that there is some similarity, although the mechanism of cell death induced by synuclein is not that well understood.

ARF: That leads me into another question regarding cell models vs. animal models. Your lab uses primarily cellular models to gain insight into the pathogenesis of AD. What advantages do cell models afford considering the field seems biased toward animal models?

Wolozin: I think that both are important. Cell models allow you to test different conditions and essentially modify the system more extensively than you can with an animal model. Animal models are best for determining how closely the model correlates with the pathology of the disease. But animal models may not be the ultimate answer in determining therapies for the disease. In summary, the animal models are best at determining whether what you are looking at is truly similar to the human disease. However, in terms of actually manipulating a system in order to find inhibitors or block enzyme reactions, I think that cell models are more effective.

ARF: Do you think that we have a good animal or cell model for AD? What else needs to be done?

Wolozin: Well everyone knows that answer. The mouse models are of course disappointing. People place a lot emphasis on the fact that the APP over-expressors don't develop significant cell death or pathology. They say that therefore shows we should question the Aβ hypothesis. I personally don't find that a cogent criticism because there is no question that the APP mutations are capable of causing the disease in humans. Yet when you take these same proteins and place them into a mouse background, you don't get as much cell death. This brings up the question of animal backgrounds. These mice are not humans and that is the reality. The models are not perfect. They don't model in vivo neurodegeneration of AD. What they do model, in particular the APP/PS1 transgenics, is neuritic accumulation of Aβ. They do that well. They don't model actual neurodegeneration.

ARF: So we have models of Aβ accumulation and plaque formation but not AD?

Wolozin: Right. These transgenic mice accumulate Aβ but the amount of neurofibrillary pathology or cognitive loss does not compare to that occurring in AD.

ARF: Can you identify some therapeutic targets that you think will be promising?

Wolozin: In the case of FAD it is very easy to point at the amyloid cascade, where production of Aβ and accumulation of Aβ leads to neurodegeneration. In that case, strategies focused at inhibiting Aβ production such as gamma-secretase inhibitors, BACE inhibitors and perhaps the amyloid vaccine will in all likelihood work. I think that I am one of the few people who has high hopes for the amyloid vaccine.

In the case of sporadic AD or late onset AD, it is true that blocking amyloid accumulation may be an effective therapy. It will prevent the tau and synuclein inclusions and subsequent neurodegeneration. The catch is, that in these sporadic diseases, it is not clear what is driving the Aβ accumulation. So in that case it is possible that there is an underlying pathophysiology which is driving the accumulation of Aβ. This detrimental physiology might continue even if we block Aβ production with inhibitors. However, you might gain a number of years of intact cognition by blocking Aβ production with inhibitors or increasing Aβ clearance with vaccines.

ARF: So it is not as clear with sporadic AD whether it is a disease of increased production like in FAD?

Wolozin: Right. I think with FAD, it appears to be relatively straightforward, although there are some questions about the age of onset. In the case of sporadic AD, the accumulation of Aβ is likely to be a major aspect of the disease, if you believe the Aβ hypothesis. But since we don't know what causes the accumulation of Aβ, it is possible that there is an underlying stress. For example oxidative stress, that will become manifest once you remove the major source of neurodegeneration (i.e. Aβ).

ARF: A recent paper of yours in JBC uses knockout PS1 cell lines to demonstrate that PS is important not only in Aβ processing but also a variety of other cleavages of APP (Palacino et al., 2000). Do you think that any of these other APP cleavage products is also important in FAD?

Wolozin: Once again, based upon what we know, the simplest answer, but not the most interesting answer, is that increased Aβ40/42 is what is driving the illness. I know that for example Luciano D'Adamio has views where he feels that the C-terminal fragment of APP maybe important in cell death. But I think that is not required in our current model for the disease pathogenesis. Especially when the simplest explanation of Aβ40/42 accumulation, leading to neurodegeneration, can suffice.

ARF: You have some previous extensive work on the role of PS in apoptosis. Do you feel that PS's role in apoptosis or Aβ production is important in FAD?

Wolozin: Yes, I was very enamored with that when I initially found it. I think that PS1 and PS2 both play important roles in signal transduction, development and apoptosis. We are focusing currently on the interaction between PS and beta-catenin, and plan to compare these results with those seen with Notch processing. The involvement of PS1 and PS2 in cell biology is spectacular and fascinating. Once again I come back to the analogy of the car door. Beta-catenin, Notch, filamin and all of these other proteins that bind PS are essential to understanding PS function and the potential side effects that may ensue if we block PS function. But the problem is with the rust and that is separate from the primary function of the protein.

ARF: Do you think that there is link between neurodegeneration and development? Considering PS1 binds to developmental ligands (i.e. Notch and beta-catenin)? Maybe these disease are developmental and manifest at a late age independent of Aβ formation.

Wolozin: Back in the '80s I published a paper demonstrating that Alz-50 is a developmentally expressed epitope that is evident in fetuses . We brought up that developmental hypothesis. But I don't think that they are the same thing. In development you don't have the accumulation of oxidized aggregates of proteins. So it is actually quite different than a developmental disease.

However, there at least one example of a neurodegenerative disease that does have a link with development. Mutations in the Parkin gene which cause a loss of function (reduced expression) produce a juvenile onset form of Parkinson's disease, termed autosomal recessive juvenile Parkinsonism disease. Even though this disease can appear in adulthood, I agree that it might be a delayed developmental disease. That is quite different from diseases of neuronal inclusion which define AD, Parkinson's, or Lewy body dementia.

ARF: Those comments remind me of Yankner's work using aged primates vs. young primates and finding differences in injected amyloid pathology. What is different between young brains and aged brains?

Wolozin: That's a good question. Obviously one difference would be that the aged brain has much more damage in it. It is much less robust owing to mitochondrial deletions, oxidation products and non-enzymatic glycosylation products that accumulate. All of that leads to reduced function of the cell. When we think about aging in non-dividing cells, it comes back to the accumulation of irreversible chemical reactions. You have to separate the chemical reaction from the cells' response to the chemical reaction. If you damage DNA the cell can go into apoptosis. If you take an inhibitor of apoptosis and treat the cell while it is undergoing DNA damage and then allow the cell to recover, it will recover but it still has damaged DNA. I think while we can't prevent the damage that occurs due to oxidation and aging, we can dampen the cell's response to that damage. That may be beneficial. In the case of these diseases, we are trying to block one of the problems associated with aging (i.e the accumulation of Aβ), whereas other people are working on blocking the cells' response to Aβ and other insults.

ARF: So if we were to keep the aged brain young and prevent oxidative damage, our brains would be able to tolerate the Aβ accumulation? Or is Aβ accumulation contributing to the aging?

Wolozin: I think that the young brain would be able to tolerate much more Aβ than the old brain because it potentially has less damage to start with. There is no question though that even the young brain is prone to cell death. So it is always levels of gray and not an absolute.

ARF: While working with Peter Davies you discovered Alz-50, an antibody that is now widely used and detects neurofibrillary tangles in AD brains. Could you describe when in your career this occurred and how it came about?

Wolozin: That happened when I was a graduate student. I guess that it is every graduate student's dream to have some work that captures a lot of attention. It was certainly thrilling and great to be involved with. I think in some ways it spoiled me. If the attention comes early in your career, there are personal expectations that are hard to achieve later on. It was great to be involved with but I think that there are also benefits to achieving success later on in ones career.

ARF: Tell me about the steps that led up to the discovery of Alz-50.

Wolozin: This was at a time when the use of antibodies was just emerging. Peter and I were discussing projects for me to do. One project was to identify proteins that accumulate in the AD brain. Another set of projects was identifying proteins that were lost in the AD brain. I actually wanted to look at proteins that accumulate, but Peter thought that was more risky for a thesis project. So we decided I would look at things that were lost and a technician would look for things that accumulate. So we generated antibodies. It turns out the project looking for lost antigens didn't work out that well. Alec Pruchnicki had taken over the antibodies from the technician who was doing the gain of protein project. He also didn't have much success. But I was very interested in finding antigens that were increased in the AD brain. I thought there might be something there. So Alec had a bunch of antibodies that were unsuccessful. So I asked him if I could take a look at them too.

ARF: So how were you screening these antibodies?

Wolozin: We were using an ELISA. Essentially we took brain homogenates, dried them on ELISA plates and screened the monoclonals. We were looking for monoclonals that discriminated between control homogenates and AD homogenates. I still remember the green well that appeared and ended up being Alz-50. It was very exciting.

ARF: How many antibodies did you screen?

Wolozin: In total as a group we probably screened thousands of antibodies. The ones in the particular batch from which Alz-50 was found might have been 20-30.

ARF: So was that the only antibody that lit up?

Wolozin: Yeah, actually on that particular day, it was. It was very striking. One dark green well and the rest of them are blank. And you think "Oh my God."

ARF: Had tau been isolated by other labs at this point?

Wolozin: Yeah, actually Khalid Iqbal was doing work on tau and phosphorylated tau at this time. We initially didn't realize that tau was the same thing as the A68 antigen (recognized by Alz50). Soon after I found A68, I actually ran tau and A68 side by side and found in fact that they co-migrated. It was actually an argument that Peter and I had. He said "that doesn't mean A68 is tau." Now I realize that I should have just immunoprecipitated the proteins in order to confirm that they were the same protein. But I was young and naive—so I didn't. But now there is no question that Alz50 recognizes tau. I actually think we should have published that early on. It would have saved alot of controversy in the field. At the same time Peter always thought that there was more to A68 than just phosphorylated tau. That has also been borne out to be true. Alz50 has been shown to bind with conformation-dependent sites in phosphorylated tau. So there is more to A68 than a simple phosphorylation event.

ARF: You have been in the field over 15 years. Have you seen alot of changes in attitudes about the pathogenesis of AD?

Wolozin: When I was originally at Albert Einstein University, Bob Terry and Bob Katzman were on staff. They had just come through an era in the 1970s where they had defined Alzheimer's disease as a disease. So in the early '80s it was actually new to consider AD a disease. Now it is accepted that AD is a disease. So that is one major change. Having defined AD as a disease, people now find the disease interesting. When I first came into the field, it was dominated by studies based on classic neuropathology. In some ways the field seemed much less inviting when compared with cancer research which had marvelous tools from the fields of molecular biology, biochemistry and cell biology. And now look at where we are. The difference between AD research and cancer research is the difference in the tissue that we study. The methods are very similar and equally exciting in either case. There has been a tremendous advancement in the methodology.

The other change that I have observed is the confluence mechanisms between different neurodegenerative diseases, much like in cancer. In cancer, you have a confluence of mechanisms in which dysregulation of protein function leads to the disease. In AD there is a dysregulation—not so much a dysregulation—but the accumulation of these aggregates that cause disease. It is the type of aggregate that determines the disease.

In the case of cancer it looks like one of the solutions may have come out of left field. Who would have guessed that use of angiogenesis inhibitors, which don't directly interact with the cancer cells, would produce such promising results. So one wonders if there is some kind of left field solution for AD as well. Maybe the amyloid vaccine is the left field solution. So I think that we should always be open to new ideas and theories.

I have always noticed that scientists can say very definitively with supreme confidence, "that's not true because of A, B and C." I feel that one makes a much stronger position when they hear a hypothesis and consider how it might be true because of A, B and C.

Recently in the case of synuclein, many investigators have questioned whether synuclein is really involved in neurodegeneration. Even though Mucke and Masliah have a mouse that develops aggregates (Masliah et al., 2000), other people have not seen it, so they say, "I don't believe it." To me that sounds very reminiscent of the early mouse work in AD, where some investigators couldn't get aggregates. Once people understood the requirements to get aggregates it became very easy and everyone accepted it. In the case of synuclein, I imagine it will be the same evolution. As we show the mechanisms by which the aggregates are causing the disease and determine ways of reproducing these inclusions—then it will become routine. I have heard some skepticism in the field and I would encourage people to remember the history of AD research. Perhaps the recent work showing that Drosophila can develop alpha-synuclein inclusions will persuade people (Feany et al., 2000).

ARF: What advice do you have for young investigators and graduate students in the field of AD?

Wolozin: I have a couple of points of advice that have been handed down to me over the years. One is do what you enjoy. I think that is true. It is important to pursue things that you enjoy doing. Second, you should not discount the importance of good training. Third, it is important to look ahead and try to read the future of the field—look at where the field is going and anticipate directions. Then don't be afraid to test your hypothesis.

And, finally one bit of practical advice that I think people have a tendency not to follow. Many times when you do experiments, you see irregularities in your data that you presume are due to mistakes that you made. In science it is important to look at these irregularities in your data and to question whether they actually represent an underlying clue to the mechanism that you're studying. A lot of times these irregularities are not mistakes you made, but instead are important clues. If you don't pay attention you may overlook them. All too often we overlook these irregularities and then we hit our heads like Homer Simpson when a paper is published examining the same "mistake" that we had ignored earlier.