ARF: What’s the primary hypothesis that guides your lab group?
RN: We’ve said for many years that we think that the carboxy terminal fragment of APP, that we call C100, which would result from cleavage at the N-terminus of the A-beta sequence, causes the disease. We still do, but our hypothesis is becoming more developed in that we’ve cloned a couple of proteins that bind to this fragment and to the C-terminus of APP. Both of them are proteins that are involved in cell cycle regulation. So we’re starting to join growing numbers of people who are suggesting that, in fact, AD is a cell cycle disorder. Our idea is that the normal function of APP, or at least one of its normal functions, is regulation of cell cycle or of apoptosis (apoptosis and the cell cycle are very closely related). We think that C100 builds up in the brain in AD. We think that when it’s on its own, and not attached to the rest of the APP molecule, it becomes ligand-independent and can cause apoptosis.
ARF: As an intracellular process?
RN: Yes. Ikuo Nishimoto showed a number of years ago that FAD mutants of APP cause apoptosis. We showed that if you express those mutants in neurons you get accumulation of C100. We published that in JBC last year. So we think C100 is responsible for the apoptosis caused by FAD mutants of APP. The central thing is that I believe that it’s a dysfunction of a normal function of APP that is causing the disease, rather than, e.g., a nonspecific death caused by A-beta.
ARF: So would you characterize this as a loss of a normal function or the gain of a negative function, or both?
RN: Probably a gain of a negative function. I think normally APP is involved in preventing apoptosis or modulating it. It’s a ligand-dependent function. We don’t know what the ligand for cell-associated APP is, although we are going after it. Normally it’s involved in cell cycle regulation as well, or at least whatever the cyclins do in neurons. If you get C100 accumulation in neurons either as a result of FAD mutations or perhaps of aging you have the buildup of a fragment that probably is ligand-independent and thereby interferes with the normal function of APP. There’s a good analogy with it, which is the EGF receptor. The EGF receptor normally is ligand-dependent, but there are truncated mutants (C-terminal fragments) of it that cause cancer because they are ligand-independent. I’m thinking in those terms.
ARF: So, along that line, to what extent would you characterize AD as being primarily a genetic disease?
RN: I think that the cause(s) of sporadic AD are environmental, with genetic predisposing components. Again, just to go back to the cancer analogy, probably it’s environmental causes coupled with aging. I think that processes that cause inflammatory reactions, and other aging processes that we don’t know about, result in inhibition of catabolism of C100 to smaller fragments such as A-beta, resulting in the gradual accumulation of C100 in the neurons. I think C100 is generated normally at very low levels, but that the cell takes care of it very rapidly. I don’t think it has a normal function. It’s just probably a product of the breakdown of APP.
ARF: So it’s the accumulation over time?
RN: Yes, very gradual.
ARF: In terms of the genetics then, are all these other genetic "causes" of the disease only feeding into this as a final common pathway?
RN: Yes. I think that the way that FAD mutants of APP cause the disease is by causing intracellular accumulation of C100, because we’ve shown that they do this in neurons. For the presenilins, I think it’s going to be another mechanism, but I think it’s going to feed into cell cycle/apoptosis regulation.
ARF: How would you outline the progression of the steps the disease takes?
RN: We think that C100 activates a signaling pathway that is mediated by the proteins that have been found to bind to the C terminus of APP. I think that they’re all going to play a part in this pathway: Go, which Nishimoto’s group found and the FE65 family of adaptor proteins, that may be involved in pulling together the different components of the machinery that centers around Go and APP. One of our two APP-binding proteins is a P21 activated kinase (PAK). Actually it’s PAK3, the one that has been shown, in a report in Nature Genetics, to be involved in a form of X-linked mental retardation. So it is known to be involved in cognitive functions. The other protein that we’ve shown binds to the C terminus of APP, we just call APP-BP1 (for binding protein 1). The really exciting thing that’s happened with that recently (it hasn’t been published; it wasn’t our work), is that it has been found that that particular protein in hamster cells regulates cell cycle progression. It regulates negatively the entry into S and regulates positively the entry into mitosis. Mutations in it cause the cells to stay in the S phase and to replicate their DNA repetitively without going into mitosis.
ARF: The thing I don’t understand, because I talked to Hunt Potter about this idea, is where in the disease progression would you see this primary effect occurring? Do you see this as a developmental process that then is played out at the end of life as a pathology?
RN: I would say that the initial event is when you start to have buildup of C100 in the neuron, causing dysfunction of this pathway. Our hypothesis is different from Hunt’s hypothesis. The only way that we mesh is that I think that it involves cell cycle proteins. Ann Cataldo and Randy Nixon have shown aberrations in lysosomes and endosomes, and we’ve just recently found that APP and PAK3 and this other cell cycle protein are found in recycling endosomes. They are all there together. So I think that the initial event is C100 building up in the cell, which affects trafficking and apoptosis, which are very closely related actually. Regarding the anatomical specificity, if we look at the distribution of these two proteins that interact with the C terminus of APP, the PAK3 and the APP-BP1, in the brain, we see them at very high levels in regions of the brain that are involved in AD. There are high levels in the CA1 region of the hippocampus and the entorhinal cortex and certain associative regions of the cortex. And if we do in situ hybridization in AD brain to look at the expression of the genes encoding these C terminal interacting proteins, when we look at the CA1 region of the hippocampus, we see that the cells that have died or dropped out are those expressing the mRNAs of these proteins, suggesting again that the presence of these proteins kills the cells. We’ve also shown that if we inhibit the expression of these APP-interacting proteins in neurons, we can inhibit C100-mediated death of the neurons.
ARF: Has that been published?
RN: No, not yet, but it’s been submitted.
ARF: Is there any experiment you can think of that would allow you to refute your hypothesis? Assuming it could be refuted, what would your back-up hypothesis be?
RN: I think about that a lot. I think the experiment that would refute the hypothesis would be if it were shown by the beta-amyloidophiles in the field that we were right about C100 neurotoxicity only insofar as C100 is a source of A-beta. In that case, it would in fact be A-beta that’s killing, perhaps by a different mechanism. Dennis Selkoe and his colleagues are just starting phase I clinical trials with gamma secretase inhibitors, and if he’s able to show that that does indeed impede the process of the disease, then our hypothesis would be refuted and probably the beta-amyloid hypothesis would be my next hypothesis of choice. The clinical trials they are doing with gamma secretase inhibitors are something I’m keeping an eye on, because to me that’s the ultimate test of our hypothesis. We know that if we make mutations in C100 so that it cannot be broken down to beta-amyloid, so that it stays intact, neuronal apoptosis caused by C100 is enhanced. So our hypothesis is that gamma secretase inhibitors will increase the amount of C100.
ARF: So the patients might actually do worse?
RN: Yes. Right. That’s what I would predict. But if the patients actually get better I’ll concede.
ARF: Well, that’s rather nice to have such a clear outcome, either way, strongly supporting your idea or going the other way. The TINS review you wrote was both timely and I thought very useful in terms of trying to put the A-beta hypothesis in perspective and I thought it was really a nice review of the literature.
RN: Thanks. Nick [Robakis] and I worked hard on it.
ARF: What kind of response have you gotten, or have you gotten any response, to it, particularly from the A-beta camp?
RN: Well, Dennis Selkoe sent us a letter outlining the reasons why we’re probably wrong and why the A-beta hypothesis is right. They’re his usual arguments, such as the argument that the increase in A-beta1-42 is one common consequence of all known FAD mutations, either of the presenilins or APP. That’s the strongest argument he put forward. But another argument relates to the relationship between amyloid deposition and cognitive impairment. Bob Terry, for example, has always said that amyloid doesn’t cause AD because there’s no correlation of amyloid deposition or amount of it with cognitive dysfunction. But I understand that Brian Cummings and others have more sophisticated methods for looking at A-beta load and have shown that there actually is a correlation. So that was another argument that he put forth. And then there was this recent Nature paper, in which a group in Europe showed that there is cell death in some animal models.
ARF: Although my understanding is that this is still somewhat contentious.
RN: Isn’t there still contention among pathologists about whether there is cell loss in AD brains? Isn’t that amazing?
ARF: Well a lot of this comes down to a lot of technical issues about how you quantify cell loss. I’ve talked to Mark West about this and I know its really difficult to come up with a method that everyone agrees on as being the most sensitive measure of cell death.
RN: That’s right. And another response I got was a letter from Bob Terry pointing out diplomatically that he said fundamentally a lot of the same things in an earlier paper. I knew that, and I wished I’d cited it because I have a lot of respect for Bob. You wonder why sometimes you overlook the obvious. You try to cover everything.
ARF: So where do you see this heading now in terms of your hypothesis to a treatment? What would be the appropriate target? How do you envision moving on?
RN: We’d like to prevent the binding of APP to PAK3, but of course you worry then that that’s also going to disrupt the normal functional interaction of these two molecules. If there was some way we could specifically affect the binding of PAK3 to C100 but not to APP, that is what we would like to do. It may be that C100 goes to a place in the cell that APP isn’t—maybe it disrupts things by being where it’s not supposed to be. It turns out that the subcellular localization of PAK3 is very important for its normal function in apoptosis. And maybe C100 can inappropriately get to where PAK3 is in the cell. I think that if we can prevent the interaction of the C100 with these downstream proteins, that’s our best bet. Of course, you could think in terms of preventing the formation of C100 with beta-secretase inhibitors, but again that’s dangerous because we don’t know what effect that’s going to have on cellular function as a whole. But then who knew that calcium channel blockers could be used for migraines? That they wouldn’t just shut down the body? So those are the two things I’m thinking about, beta-secretase inhibitors or preventing the binding of C100 to PAK3 or APP-BP1 by generating another molecule that will competitively bind to it, to tie it up.
ARF: What about the mechanism by which C100 is killing cells? Do you see any chance of intervening farther down in the pathway?
RN: You want to get as far downstream as you can get, because that gives you more specificity than you have earlier in the pathway from which it may have multiple branches. And I really haven’t thought that far ahead yet. One might try to generate apoptosis inhibitors that work specifically in the APP/PAK3/APP-BP1 pathway. I think that Mark Mattson’s paper showing elevation of PAR4 in AD brain was really important, because before that it just wasn’t clear whether apoptosis was involved in AD or not. There were many people, including ourselves, doing work on apoptosis, but it wasn’t clear that that was worthwhile because it wasn’t clear that that was the kind of death that was occurring in the AD brain. So the PAR4 work was really important because it showed very clearly that at least one kind of death occurring in AD brain is apoptosis. So maybe generalized inhibitors of apoptosis might also be effective.
ARF: In terms of the initial insult, and this goes back to the question about where you see the disease initiating, does everyone get AD if they live long enough?
RN: Right, so if we slow it down, that's the main thing? I’m operating on the assumption that whatever normal aging processes are giving higher incidences of many cancers and other diseases, are also operative in AD. I’m assuming that if we learn which nutrients or which drugs can help with management of free radical production and oxidative mechanisms, aging in general or anti-inflammation, we can use those agents to slow down the development of AD. It may be as simple as that. I have my parents on antioxidants, and I’ve been taking them for years, because I really think that whatever is operational in helping to prevent other forms of age-related diseases will also help with AD.
ARF: Let me ask you one other thing. In terms of the field, do you think there is any area of science related to AD that is not being looked at closely enough, either in terms of the funding priorities or just in terms of the big labs?
RN: Yes. Probably the inflammatory mechanisms. It’s still pretty much the same labs, the McGeer lab, Joe Rogers’ lab, and a few others whom I forget, who are working in this area. It’s clear that people who chronically take anti-inflammatories have a much lower incidence of AD. So I think that that’s a big area that needs to be concentrated on, because that is going to be the area where preventative measures may be taken. I think that inflammatory mechanisms are probably very much involved in the neurodegeneration that occurs in AD.
ARF: Do you see any issues regarding how receptive the scientific community is to novel ideas?
RN: The field is very exclusive. To me, Ikuo Nishimoto’s work showing that FAD mutants of APP cause apoptosis that is independent of beta amyloid formation is incredibly important, but his work is completely ignored. In a recent commentary in PNAS on making mice smarter with APP, the authors reviewed all the work that’s been done on the normal function of APP but failed to cite Nishimoto’s work showing a Go-mediated signaling pathway initiated by APP. That pathway probably will be critical for understanding AD. People tend to ignore the question of what the normal function of APP is. When a new hypothesis comes in and people don’t know how to put it into their little corner, they often just ignore it. That’s what happens with findings about the normal function of APP, which a lot of people think isn’t going to be important in the disease. But I think it will be. I think it’s great that the injection of the secreted forms of APP makes the mice smarter, because AD is loss of cognitive function so maybe people will start thinking in those terms, that it is a loss of APP-mediated cognition. But it is a very exclusive field.
ARF: I think that Mark Mattson was one of people arguing early on that the loss of APP functions might be as important, if not more important, than the production of A-beta.
RN: Yes. From way back. I have a lot of respect for Mark’s work. He’s good and he hung in there and I think that he was ignored for many years but finally his work became so compelling that people couldn’t ignore it any more. The most nagging criticism of our hypothesis is that most people think that C100 is merely a source of A-beta, so that our hypothesis is really the beta-amyloid hypothesis. There was a Nature paper from Shapiro’s group, that came out in June of ‘97, that showed that C100 transgenics have impaired LTP and impaired spatial learning, and considerable AD pathology, and their interpretation was that this was a model of beta-amyloidosis because C100 is a source of beta- amyloid. Karen Duff came to give a talk at McLlean—she’s made numerous mouse models for AD, some of which express FAD mutants of APP—and in her talk she was talking about the oddity that they see massive amyloid deposition without other aspects of AD pathology in the mice. She said, "We think that we may have to go with expression of C100 to get a beta-amyloid model that also shows neurodegeneration." She suggested that perhaps the N-terminus of APP is protective, so that we have to get rid of it by expressing C100. To me, that’s the most nagging criticism. We’re doing a lot of work now with mutants of C100 from which beta-amyloid can’t be generated.
ARF: Has this translated into any difficulty in convincing study sections of the viability of your approach?
RN: No, fortunately it hasn’t.
ARF: Thanks for a most provocative interview.
No Available References
No Available Further Reading