Posted 30 November 1998
Interviewed by Keith A. Crutcher
To what extent do you believe that AD is a genetic disease?
I think for patients under the age of 60 that 5% or so is probably fully genetic. The toughest thing to do in that population is prove a sporadic case. You'll get a case, once in a while, that's 50 years old with AD and no apparent familial history but then if you sequence presenilin 1, in our experience, most cases we sequence, essentially all that are under age 50 year old at onset have a presenilin 1 mutation. Under 60, there is still a strong genetic component but not as much as with PS1. But basically, for cases under 60, (that's roughly 5% of AD and you have to watch out because people interchange "early-onset" with 60 cut-off or 65 cut-off; with 65 as a cut-off that's about 15-17% of AD), just about all are genetic or "familial", with the latter defined traditionally as having two or more first-degree relatives who are affected. You can also define it in a single case as by seeing a presenilin mutation or some other early-onset mutation. I tend to like to call the early onset cases where you only see one affected person with no family history a "singleton" rather than a "sporadic". "Sporadic" is a very big statement and usually not backed up.
Because it implies a mechanism that is somehow different....?
Yeah, it implies that this was truly out of the blue. Somehow the guy just got AD and it had nothing to do with genetics. And I think, in that same vein, that even in the later onset cases, It will be difficult to show that a case was completely sporadic without the influence of at least one or a handful of genetic risk factors. One person in response to certain insults may get the disease by 70 whereas another person with the same insults might not get the disease until 90. That's probably a genetically predisposed case. And we have to think more along the lines of not only how many genetic risk factors and whether AD is genetic or not, but also what are the different types of genetic risk factors and how do they interact and work together to determine overall accumulated risk of the disease. Some risk factors may only affect whether you're going to get the disease or modify some other cause of the disease, genetic or environmental. Others might simply influence progression or your age of onset.
Do you think its useful to draw a distinction between genetic factors that are causative versus permissive?
"Permissive" is not a bad term. It's usually reserved for lower organisms but it might not be a bad term to introduce because risk factors really just influence your reaction to some other insult whether it be genetic or environmental. So you can have a risk factor that is acting like a modifier and, in essence, genetic modifiers or genetic risk factors regulate permissivity. So it's not a bad term. But I think, it may be that in late onset AD there's no single cause that's being modified and that there are a bunch of risk factors all working together to determine whether you're going to get the disease before you die otherwise within the normal life span of a human being. There may be plenty of variations in genes that appear to be harmless and innocuous right now but once we beat AD away so that it doesn't hit on average by 100 years old, then you're going to have new ones that previously made no difference at all but suddenly determine whether you're going to make it to 110 without AD. And it's going to go on and on. As life span gets longer and longer it's going to be a constant struggle with morbidity based on genetic variations which in one given life span are innocuous, but in a longer life span make all the difference in the world as to whether you're going to make it to 120 without AD, or cancer, or diabetes. These public variations which people often say don't matter because they're not unique and they're not exclusive for the disease are probably the most important ones. And they also can be manipulated more readily because their influence is subtle. They have a modest effect. It's a lot easier to imagine how you might do something about one of these gene variations than something that might be untouchable like a presenilin gene defect.
Do you have a favorite hypothesis?
It's not really that original but basically I would say that genetic and environmental factors, but clearly genetic factors, contribute to your life-long production and accumulation of Aß and that if you are a human being, and not a mouse, and you accumulate enough Aß, neurons begin to die. Aß turns out to be a very tight molecular and pathological correlate of the disease, even if it's not directly causing neuronal cell death. Determine the genes that influence the production of Aß (like presenilin and APP) or its accumulation versus degradation and clearance (like, I would say, apoE and A2M), and by working on how to mimic the beneficial aspects of these genes, especially the late-onset risk factors that affect degradation and clearance, you're probably going to come up with pretty good drugs for the disease. Whether that's going to bring us to the root cause of AD, I don't know, because I think there's also room for apoptosis and it could be that presenilin mutations increase sensitivity to apoptosis by simply increasing steady-state levels of Aß in the brain or in neurons. Maybe the Aß just follows along.
Any favorite alternative hypotheses?
Just to offer one alternative hypothesis, we have data in our unit from Dr. Ashley Bush that Aß can act like a superoxide dismutase. Aß reduces copper. It takes reactive oxygen species and converts them to hydrogen peroxide. So you have a molecule that's using copper, reducing copper, binding copper to generate hydrogen peroxide. You effectively have a mini version of SOD. In fact, if the reaction goes too far and glutathione doesn't come and take hydrogen peroxide out of the equation you can get corruption of Aß just like you can corrupt SOD in ALS. So one of the things we've been thinking is maybe presenilin mutations and APP mutations are driving apoptotic events and then the increased production of Aß may be a rescue response. This is total heresy, but for all we know, it could be that Aß could be produced as a quick response for the need for SOD, as an antiapoptotic measure; a rescue response on the cell that's about to undergo apoptosis. And then, in what I would call a case of molecular antagonistic pleiotropy, the production of Aß (which was meant to be a good thing to deal with ROS that are being overproduced in apoptosis, where the Aß will come in and reduce copper and generate hydrogen peroxide hoping there's glutathione and catalase around that can get the hydrogen peroxide out of there) could turn out to be a bad thing if the reaction goes too far. Just like it could happen for SOD. We use this term "antagonistic pleiotropy" for genes in aging that give you a survival advantage and reproductive fitness early in life but with a trade-off because later in life they become detrimental to survival. This would be an example of Aß being in a group of molecular antagonistic pleiotropy, maybe just a peptide with good intentions. But the road to hell is paved with good intentions. So that's an alternative hypothesis we are also exploring along the way to figure out how apoptosis fits into the equation.
I see some analogy there to the arguments that some people make for the role of inflammation. That it may be that you're triggering a response that is normally beneficial but that the side effects of the response are actually contributing to the pathology.
Yes, there are many cases where there are molecules that are trying to rescue a cell and can in a situation of excess be corrupted by their own environment and then cause more problems than were originally there to begin with.
Can you think of any critical experiment that still needs to be done or do you feel the "smoking gun" data are already in pointing to Aß?
The experiment that hasn't been done, in terms of getting to the cause of the disease -- and let me preface this by saying, getting to the cause of this disease isn't necessarily the fastest way to a drug. I think the fastest way to the drugs really is through the modifiers, the genetic risk factors which contribute to just why you get garden variety AD. And it may be it's more of a basic science exercise to look at the cause of the disease. But if we want to get to the cause of AD as a basic question, we know the presenilin mutations do two things. They increase Aß -- and I'm not even totally convinced they increase the Aß42/Aß40 ratio. Because you have to remember that when you make Aß, if you make more Aß40 and Aß42, the 42 has a longer half-life in the media. So all this ratio of 42/40 increase might just be because you made equally increased amounts of 40 and 42 but the 42 lasted longer. So I'm not so convinced that there's really a ratio change. In any event, we know presenilin mutations increase Aß and for one reason or another, perhaps due to differential half-life, we see increased 42/40. The other thing we know is presenilin mutations sensitize cells to be more vulnerable to apoptosis and we know this by a variety of apoptotic measures. Cells that express low levels of the mutant presenilin, sufficiently low levels to not cause spontaneous apoptosis, still make the cell more susceptible to apoptosis when challenged with an apoptotic insult like staurosporine or serum withdrawal. So basically those are the two things we know regarding effects of presenilin mutations: enhanced susceptibility to apoptosis and increased production of Aß.
We also know that Aß induces apoptosis. So the question is: Does susceptibility to apoptosis, conferred by presenilin mutations, depend on increased production of Aß? And the answer is: we don't know. No one's done this experiment. The thing to do really, and what we're planning on doing (Dora Kovacs in our lab is going to do this) is cross mutant presenilin mice with APP null mice. Take these neurons and then see if they are still more sensitive to apoptosis than the ones producing wild-type presenilin. That way you don't have to use antisense strategies to knock out APP. Just take a system where APP is not there, where Aß can't be made and answer this question once and for all. Do mutant presenilins really matter for apoptosis or is the effect mediated by increased Aß? If it's only based on Aß then if you really take APP out of the equation, Aß out of the equation, with an APP null mouse crossed with a presenilin mutant mouse, and suddenly the increased sensitivity goes away then you can say, well, Aß still acts before increased sensitivity to apoptosis. If, however, there's still sensitivity without Aß present, then I think you have to start speculating even more that maybe the production of Aß is a secondary response and a reaction of a cell that's about to undergo apoptosis. And, I would guess it might even be a rescue response based on the fact that this is a peptide mimicking SOD.
Very interesting. I knew that you had been working in this area but I hadn't put it together in terms of this idea.
When Aß starts to aggregate and the half-life starts to get into days rather than hours, or even longer, then you may quickly find yourself with way too much SOD-like activity and there's no way glutathione and catalase can keep up and get rid of your excess hydrogen peroxide and so, bang, what are you going to get?, you get Fenten chemistry. You get your hydrogen peroxide with your reduced copper, both mediated by Aß. These things get together and make hydroxyl radicals and things start to fall apart.
Do you think there's any area of AD research that's not being explored sufficiently? Either in terms of scientific areas that are under-represented or politically in terms of scientists who are marginalized because they are not working on the favored hypothesis?
I think it's getting better. I think for a while we were probably too amyloidocentric. The best thing that could have happened to the tau people is frontotemporal lobe dementia mutations. I think the transgenic APP mice have also shown us that amyloid production in a mouse is not sufficient. So we really have to think about what is missing there. Is it the form of tau? Is it the microglial response? So, I would say if anything I would see more work, and it is happening now, on tau but more work also on the role of microglial activation, especially as it differs between mouse and man. I was very intrigued by this talk that showed that the COX2, which mediates the inflammatory response, has a different set of promoter elements between mouse and man. Man has a TATA box and the mouse doesn't. The mouse just has a couple of AP1 sites. It could be a little difference like that, in terms of how inflammation is triggered, that allows a mouse to survive, and then die of other causes with tons of amyloid, whereas the human response to amyloid, with maybe one little promoter difference in COX2...bang, you've got an inflammatory response that is driving most of your neuronal cell death. So I think differences between mouse and man and why the model isn't perfect, especially with regard to microglial activation, need to be looked at.
Do you think there are even more extreme, or fringe, hypotheses that deserve attention?
Even after presenting A2M as an AD gene, because this wasn't done with a standard case-control study and something that people are familiar with, and it was done with a complex genetic analysis that involves family-based comparisons, I now have to be patient and just wait for a forum to assemble. It just takes time.
Of course you also already have the credentials.
But for A2M I can be attacked as robustly, if not more, than others have been attacked for their work. When you get a new idea and it's on the fringe -- I mean mine isn't really on the fringe...A2M binds Aß, clears it through LRP and degrades it, it's not that hard to imagine how it's involved in AD -- but I think we just have to be patient. Keep increasing the sample size because eventually, when the sample size is in the hundreds...hundreds, and it's holding up over and over again, people have to listen. Any time you're doing an association type of study, whether it's a genotype like apoE or a phenotype, like presence of HSV by PCR in the brain, the only thing that's going to convince people is sample size. Otherwise, people can still yell over and over... "Sample size is too small, type I error." And that's the problem with most of the case-control studies involving new Alzheimer genes. The original sample sizes are small and you just have to be patient and wait to see if it's going to be replicated.
But this also assumes that the system is going to allow you to have the resources to increase the sample size.
Well, if I'm reviewing a grant that has the preliminary data I would say give them 5 years and let them do it. Set out to test the hypothesis. Either show it beyond a shadow of a doubt or exclude it. Just get it done. Now, I agree, there may be people in the review system who don't see it that way and who would say, philosophically "I don't believe this." There's really no room for that type of thing. If someone has preliminary data and they know how to do the experiments and it's an intriguing hypothesis, let them finish it. It's good the Alzheimer's Association is also around. People can use those funds to generate more preliminary data to get an RO1. In the end, I find that with study sections it's really preliminary data that drives a successful proposal. If you set out to rigorously test the hypothesis, with the sole intent to exclude it, and that includes all the appropriate controls, then you're going to get the money. But you need to have the preliminary data first before you can have the compelling hypothesis that you set out to exclude.
Rudy Tanzi's research focuses on identifying and characterizing the
genetic and enviromental factors involved in neurodegeneration in
Alzheimer's disease, aging, and Down syndrome. In addition, he is
interested in addressing the mechanisms underlying the etiology and
pathogenicity of the genes responsible for Alzheimer's disease through
the application of molecular, cell biological, and biochemical
strategies. He is also interested in determining the environmental
factors that interact with these genes as well as those that govern the
formation of beta-amyloid in the brains of AD patients and elderly
individuals. Beyond AD, he is interested in identifying genes that
influence longevity (gerontogenes) and healthy aging of the brain.
One topic of particular interest in Dr. Tanzi's lab is the question
of how susceptibility to apoptotic cell death may play a role in AD. "We
are currently attempting to determine whether the abnormal generation
and accumulation of caspase-derived presenilin protein fragments
contribute to increased neuronal cell death and the generation of
A-beta," he explains. "Our investigations of the normal function of the
AD genes have also led to the identification of specific proteins that
interact with APP and the presenilins (e.g. FE65,
discovered to interact with APP by Suzanne Guenette, and sorcin,
discovered to interact with presenilin 2 by Tae-Wan Kim). Currently, we
are investigating how these proteins and influence the processing and/or
trafficking of APP and the presenilins. An in-depth understanding of the
normal function of the AD genes and how mutations and variants of these
genes can lead to beta-amyloid deposition, neuronal cell death, and
dementia will be invaluable for the development of novel treatments for
Dr. Tanzi majored in history and microbiology as an undergraduate at the
University of Rochester, then went on to earn a Ph.D. in neurobiology at
Harvard. He has held postdoctoral appointments at Harvard and is
currently Associate Professor of Neurology (Neuroscience), Harvard
Medical School and Massachusetts General Hospital.
At MGH, Tanzi directs the Genetics and Aging Unit which consists of nine
laboratories including those of Drs. Wilma Wasco, Ashley Bush, Tae-Wan
Kim, Dora Kovacs, Suzanne Guenette, Robert Moir, Craig Atwood, Xudong
Huang, and his own. In his spare time, Dr. Tanzi enjoys playing jazz
piano, scuba diving, skiing, basketball, and tennis.