Posted 22 December 1998
Interviewed by Brian J. Cummings
We invite you to submit comments and questions
on Huntington Potter's remarks.
HP: One of the few things about Alzheimer's disease on which everyone
basically agrees is that it starts 20 or 30 years before the first symptoms.
That means, when we look in an AD brain, we are looking at the very last
stages of the process. Most of the current work is designed to try to identify
the earliest stages of the disease, with the assumption that preventing
the earliest steps in the pathogenic pathway will be the best approach to
therapy. That assumes that we are going to have, in parallel, people working
on diagnostics that can identify those pre-symptomatic AD patients.
ARF: So that the 'diagnosis' is early enough for the therapy
to have an effect?
HP: Exactly. And of course that brings up ethical questions about
how extensive the screening ought to be. I tell lay people that in a room
full of reasonably healthy individuals, probably 30% of are going to have
some symptoms of AD by the time they die. They may not die of the disease,
but they are certainly going to have some symptoms. That means that everyone
ought to worry about being pre-symptomatic.
ARF: And does that mean that we risk putting everyone on a drug
for 15 years that has potential liver problems or other toxic side effects?
HP: That's the next level of pharmacological intervention: to
develop drugs that have no side effects, that are as simple to take as Vitamin
E and as harmless, hopefully. Then people who are really worried will be
able to take it. But that means we really have to understand the pathogenic
pathway from the beginning to the end, with all of its steps. And I think
that most people are directing their attention at that in various different
ARF: What, according to you, are the best hopes for understanding
the molecular progression of the disease. What is your primary focus?
HP: The cell cycle is disrupted in AD. There are substantial
data to support this idea. For example, there are present in the Alzheimer's
brain several mitotic enzymes and phospho-epitopes that are only present
in mitotic cells and have no business being in neurons of any brain. Finding
them in the AD brain is highly suggestive. Then there's our data, which
have been repeated and extended in a number of labs, that presenilins are
involved in the cell cycle. There is also a recent paper by Roperch et
al. in Nature Medicine that came out in July that shows that p53, a classic
cell cycle protein involved in tumor suppression, changes the way presenilin
functions. They then showed that if you suppress presenilin expression
by an antisense transfection, then you change the cell cycle features of
the cell and keep it from becoming a tumor upon injection into susceptible
mice. These kinds of results and our results are very suggestive that there
is a cell cycle problem in AD. This is a completely novel way to think
about AD, and I would suggest that people pay more attention to it -- and
we certainly will -- but one of the questions that comes up is: is the
cell cycle disruption an early event or a late event?
ARF: Is it a consequence of other pathology or does it lead to
HP: This brings up the tremendous power that we have in AD, which
is the use of genetics. Because when we identify a gene, which when mutant,
causes familial AD, we know that whatever that gene and its accompanying
products do, it has to be one of the very first steps of the disease. And
so presenilins and APP, and possibly ApoE as well, are probably part of
the early steps in the pathogenic pathway. And this brings us to the question
of what do these proteins do and how are they different in AD?
What I find tremendously surprising is that there isn't more interest
in finding out what the normal function of APP is. This is a protein that
is made in every single cell throughout development and almost nobody, is
interested in finding out what APP's normal function is. And that's because
we have always assumed that the function of APP is irrelevant to AD and
its only function in AD is to make Aß. And I think that's a little
premature. And even if it is true, it's highly unintellectual to ignore
the function of a major protein in AD.
ARF: And what would you say the odds are that the real problem
is a loss of function in APP as opposed to an Aß dysfunction?
HP: I think that it is only about 20%, but we can't ignore that
ARF: And the same for the presenilins?
HP: Now, PS1 and PS2 are anybody's guess. There are ideas about
what the normal function of the proteins are that are related to cell cycle
in our lab, and there are ideas that are being proposed this week (at the
Society for Neuroscience meeting) that the presenilins make proteases involved
in their own processing and also the processing of APP. There are suggestions
that they are cell surface receptors. Since the presenilins don't make
Aß, I think we should keep our eyes as open as possible as to what
the function might be. The data are coming very fast from all sides and
they're suggesting what the presenilins might do. I would hope that there
would be more work on the genetics of the presenilins in drosophila and
nematodes, because there you have a combination of biochemistry and genetics
that would tell you what the normal function of the proteins is.
That brings us to another question as to whether the presenilins have
more than one function. Our work suggests that in dividing cells, the presenilins
are localized to the centrosome, the microtubule organizing center, the
nuclear membrane, and the interface kinetochores. But in some cells, such
as neurons, the majority of presenilin is in the cytoplasm, presumably associated
with some kind of membranous vesicles or organelles, such as the endoplasmic
reticulum. There was also a poster yesterday that said that presenilins
were associated with the cytoskeleton in the growth cone. That is very
interesting because if the presenilins are involved in movement of membranous
vesicles along the cytoskeleton, it would fit many of the apparently contradictory
data being presented at this meeting. For instance, in neurons the presenilins
may be involved in vesicle transport because that's what neurons mostly
do. They don't have to divide, so they don't have to worry about cell division
and the cell cycle. Whereas in dividing cells, the presenilins can be both
in the cytoplasm and be involved in vesicle transport and also be associated
with the nuclear membrane and the kinetochores and thus involved in the
movement of other structures, i.e. chromosomes, along the microtubule spindle.
That needs to be investigated.
ARF: What do you think about Dranovsky's talk about the phosphorylation
of nucleolin in early neurofibrillary tangles?
HP: Nucleolin is a very interesting phosphorylation epitope,
which he has identified as different in AD. It is going to be a little
while before we know whether this is a phospho-epitope which is reflecting
a change in the general phosphorylation of proteins in the AD brain or whether
it is one of the key parts of the pathway. It is exactly that kind of search
and discovery approach that we are still involved in. I am a little unhappy
about the total devotion to APP processing and presenilin processing by
some of the larger labs because, although it's probably important, it's
not guaranteed to be important, and there's a lot of effort being put on
that. I think that it could be a little bit less.
ARF: What are the primary hypotheses that guide your group's
HP: We have 3 major projects, one which I just talked about which
is to generate a transgenic-knockout combination mouse that makes human
forms of Alzheimer-like tau. The way we do that is to make a knockout mouse
that has hyper-phosphorylated tau due to the phosphatase calcineurin being
knocked out and then mate with it a transgenic mouse that has human tau
so that you can get hyper-phosphorylated human tau in an in vivo situation.
The other approach is to find out if inflammation is important
in AD. We first found that inflammation is possibly involved by finding
the first acute phase/inflammatory molecule that is actually produced in
the AD brain -- antichymotrypsin. We then found out that IL1, which is
an inflammatory cytokine that is upregulated in microglia in AD, as found
by Sue Griffin, is responsible for inducing the astrocytes to make antichymotrypsin
in the AD brain. And we have just found that IL1 will also increase the
amount of APP protein produced by astrocytes, not at the transcription level
of RNA synthesis, but at the translational level of protein synthesis.
This is a completely new connection between inflammation to AD, through
the APP protein. There's a stem-loop in the 5'-untranslated region APP that
is necessary and sufficient for sensitivity to IL-1-induced translation.
So, the reason why we think inflammation is important is because it induces
the expression of APP and of amyloid promoting proteins (antichymotrypsin
and ApoE), all of which proteins have genetic evidence in favor of their
involvement in the disease process.
ARF: So you've got IL1, ApoE, and antichymotrypsin -- all of
those are feeding back into the same pathogenic pathway.
HP: Exactly. We know from recent in vivo experiments
from Lilly by Kelly Bales and Steve Paul and their colleagues and from the
previous in vitro experiments, that ApoE is definitely an amyloid
promoting factor and that is probably why it is a risk factor for developing
late-onset AD. The AD risk associated with the antichymotrypsin-A allele
is becoming more and more clear. ACT-A is a risk factor, and ACT is an
amyloid promoting factor. Furthermore, we have recently found that the
antichymotrypsin-A allele increases the secretion efficiency of antichymotrypsin,
as was predicted by Kamboh and DeKosky when they first found that the antichymotrypsin-A
allele polymorphism was associated with AD.
So now we have an inflammatory cascade which turns out to be
absolutely essential for the conversion of Aß into amyloid. And many
data suggest that ß-amyloid is an essential part of the disease process
or of the early neurodegeneration process. So, without the inflammation,
you're not going to have the AD. And that's a tremendous discovery because
it means that a whole area of potential therapeutic targets has been opened
up. There was a hint of it in Joe Roger's original work with the anti-inflammatory
agents suggesting that they reduce the proficiency of AD. But now that
we have a pathway, we don't have to use general anti-inflammatories, we
can target much more specifically, which would reduce the possibility of
side-effects. And in a situation where you would want to potentially treat
people for 20 years, side effects are very, very important. Targeted anti-inflammatory
therapy, I think, is going to be very promising.
Finally, we have just recently become interested in the cell cycle and
the involvement of the presenilins in the cell cycle as an explanation for
AD. Our original idea was that the cell cycle defect might lead to chromosome
mis-segregation and an accumulation of cells with three copies of chromosome
21. Since Down syndrome individuals all get AD, we thought that maybe if
you have 10% cells with trisomy 21 that would give you AD too, but at a
later age. I believe that is still a viable hypothesis. The data have
been consistent with that and there has been not much in the way of data
directly against it. The other possibility is that the cell cycle defects
that we and others have seen in AD are causing a physiological problem in
the cells which leads to their production of Aß 1-42. In this light,
it is interesting that APP and tau are both cell cycle regulated proteins.
Specifically, APP is a phosphoprotein whose phosphorylation changes during
the cell cycle: it goes up during mitosis and down during interphase. Tau
is a well-known phosphoprotein whose phosphorylation goes up during mitosis
and goes down during interphase. So, these two key AD players are cell
cycle regulated proteins. It is highly likely that and their phosphorylation
affects their function -- which we already know about tau, and it wouldn't
be at all surprising that the phosphorylation that occurs during mitosis
changes the processing of APP. That means that if you mess up the cell
cycle in any way so that the phosphorylation of APP that is so carefully
controlled through the cell cycle is also messed up, then you're likely
to have a change in the APP processing. So, maybe the cell cycle differences
that we're seeing are causing a general physiological defect in the cell
whose output is a change in APP processing. That would be fine with us.
ARF: And does that open up the avenue that there are multi-factorial
possibilities to how you get AD?
HP: Oh yes. It is certainly possible that the pathogenic pathway
has a number of branches in it. We have already identified one, which is
that inflammation is necessary in parallel with Aß production to get
AD. And there might be other parallel parts of the pathways that we haven't
ARF: How strong a role does genetics play in getting AD. Do
you believe that genetics will ultimately explain everyone who gets AD?
HP: I was raised as a geneticist, so I have certain prejudices.
I think it is clear that at least 50% of AD is partially genetic now-if
you count the 10 to 20% of the early AD cases caused by presenilin and APP
mutations, and another 30 to 50% due to antichymotrypsin-A and ApoE alleles.
I just saw a poster that suggested that the IL1 receptor antagonist gene
has a polymorphism associated with an increased risk of AD, and that makes
sense with the inflammatory process that we were just talking about. We
are certainly going to see other new risk factors being discovered as well.
As soon as you get more and more risk factors, then it becomes a very complex
genetic question as to how much of a person's getting AD is due to any one
particular gene And I wouldn't be at all surprised if we see the number
of cases influenced by genetics rise above 50%. These individuals will
be quite different than the 20% who develop early onset, autosomal dominant,
100% penetrant AD. The majority of AD patients are going to have a combination
of genetic risk factors. And those risk factors are going to combine together
to give a higher or lower chance of somebody getting AD, but there will
be no guarantee of anything one way or the other. And that brings up a
whole slew of ethical questions that we are going to be faced with, with
respect to predicting what the chances are of somebody getting AD. People,
even scientists, are not very good at dealing with life choices based on
statistics and that's going to be difficult from a public education point
ARF: Instead of having early diagnostic screening in terms of
some physiological test or enhanced resolution MR, are we're just going
to take a gene sample and calculate the odds ratio and say that if you're
at a certain threshold we're going to treat you with this drug?
HP: If the drug were harmless, I would think that would be a
likely scenario, yes. I would say that if I had a few risk factors for
AD and I didn't have any fear of side effects, I would probably take it.
But that's an ideal world, and these drugs almost always have some side
effects (ARF: And cost). Well, from an individual point of view, we may
decide to shoulder the cost. From a public health point of view, this could
literally break the bank. If there are 30% or 40% of people who have a
risk of getting AD, if they live long enough, that's a lot of treatments,
and the drug would have to be very, very, cheap. I suspect we might use
a combination of a genetic profile and some early diagnostic markers of
actual AD pathology. The genetic profile would be used to identify people
at risk, and then potentially more invasive screening, like CSF, or other
biological markers, might be used to identify those among the genetic risk
factor group who should receive early therapy.
ARF: In this question of therapy, do you see our interventions
be able to "cure" the disease or just be able to slow it down?
HP: Cure essentially involves taking a situation that has clear
pathology and behavior deficits and then reversing it. I think we have
some chance at that. That's not to say that we'll be able to replace all
the neurons that have already died by some kind of miraculous stem cell
rejuvenation -- although the stem cells may already be there. More likely,
we will be able to stop the process from getting worse; the cells that are
left may then be able to recover a lot of function. They will be able to
sprout without getting their fingers nicked, and they will be able to send
out longer processes than they had before. With the training of individuals
to make use of this sprouting capability we will be able to do a lot. From
a functional point of view, people may be able to get better, and that's
sort of like "curing" the disease. How we would do that is going
back to the pathogenic pathway and the same kind of drugs we discussed before.
ARF: What if you're completely wrong regarding you cell-cycle
hypotheses and inflammation mentioned earlier; what is your 'runner-up'
hypothesis for the cause of AD? Or, what is the direction we ought to be
hunting down (no pun intended)? If you had an extra post-doc and a graduate
student on the side in a room down the hall, what would you have them working
on in secret just to cover yourself?
HP: I would say it's axonal transport or dendritic transport
mechanisms. I don't think there's any data that I can think of that are
inconsistent with it. The problems with presenilin biology and APP biology
would still be important in trying to address that question. And we still
would do some of the same kinds of experiments, but we would focus our mind
on transport processes. ALS is an interesting parallel here where there
is dying back of the axons, but there are also inclusions in the cells which
include the SOD protein that is mutant in the inherited forms of the disease.
These inclusions could be responsible for killing the cells. And it's
not clear whether they are killing the cells directly and the processes
are dying in response to that cell death or whether those inclusions, SOD
inclusions for instance are messing up the axonal transport, which then
allows (or forces) the axons to die back. Of course, such dying back would
reduce the neurotrophic factor support for the cells and the cell finally
dies. I think it has been useful in the past to consider parallels between
different neurodegenerative diseases and I would consider the parallel between
ALS and AD worth looking at.
I also don't think that too many people would completely disagree that
presenilins may be involved in transport. APP is definitely transported
out into the axons and the dendrites (it is suggested that presenilin may
also be). If presenilins are microtubule binding proteins -- which they
might be because there has just been report that presenilin will bind to
CLIP 170. CLIP 170 is a minus-end microtubule binding protein.
Thus a transport or microtubule defect is not inconsistent with everything
we know about AD that I can think of. And such an model would just mean
refocusing on that particular question. Now why hasn't it been done before
or as much as it could be? Matsuyama and Jarvik proposed a microtubule
defect model many years ago. Now, I think we have become so focused on
Aß and extracellular amyloid deposits that we may have become blind
to other ideas. If amyloid didn't exist in AD, we would be focused on axonal
transport. The APP mutations basically support Aß as being an essential
part of the disease. That doesn't necessarily mean that amyloid is an essential
part of the disease. Okay, it kills neurons, Yankner first showed that.
We have repeated it, other people have repeated it, some people had trouble
repeating it initially and then it was discovered that it took the filaments
to kill the neurons. And it may be that amyloid is an important part of
the cell death chain process. Or maybe amyloid is a result of changes in
the physiology of the brain that lead to its deposition, and which are reflective
of the disease, but not causative.
ARF: Assume you had unlimited funding and technical support in
the future, you have to imagine some future technology so that now you can
look at individual neurons in the living brain, physiology, fMRI, 3-D imaging
in real time, etc... What future thing would most help us in understanding
AD and what do you think that thing is?
HP: What kind technical advance would make this possible...?
ARF: Well, yes. It's up to you to pick what would be more important,
but basically, do we not know the "answer" to AD because of technical
limitations where you can't measure mRNA in living cells simultaneously
or we don't know the answer because even if we could do that, or some other
futuristic technical thing, it still wouldn't help with what we need to
be able to focus on.
HP: Okay, well the miracle machines that we want and that every
pharmaceutical company wants is a way to determine, absolutely for sure,
the function or functions of a particular protein in an in vivo,
physiologically relevant situation. And then we would be able to see how
that might change during the disease process. And this is what has to be
added to the Human Genome Project. We can identify all the genes, we can
find out what the 3-D structures are if we can get crystallographers to
do it, but that doesn't tell us what the function is. The reason why it
is so hard is because the proteins may have many in vitro activities
- they may even have many in vivo activities. But their exact localization
within the cell, the exact proteins that are co-localized in the same place,
and which are therefore given a chance to interact with the proteins of
key interest, are a lot of work to try to find out. And then you have to
find all the potential interactions that occur: Are there enzymatic reactions,
binding reactions, transport reactions? Are they energy-dependent, or
oxidative, redox potential dependent? Each one of these questions has to
be asked independently for each protein and you can have hundreds and hundreds
of personnel hours spent just asking about one particular protein -- and
you can still miss something. Thus, the big black box that we call "the
cell" has lots of little players, and we need to know what those proteins
do. I don't see an easy way to get at that problem, but in the past, it
has been through genetics. You make mutations, we see how it changes the
physiology of the cell. Any one mutation can be misleading because you
can get a gain of function that is irrelevant to the overall function of
that protein. But if there is a long list of mutations for each protein,
placed back into the cell, in a knockout background, and you can then ask
what changes in a systematic way? This is the approach that has been used
in systems where there is a the possibility of combining genetic and biochemical
approaches. That is why I mentioned early on that I would like to see more
effort expended on the traceable genetic systems: the drosophila and the
ARF: Do you think we're about to break out of the APP, presenilin
HP: Well, we might. It is very helpful to know when you have
gain of function that is absolutely, 100% penetrant, because then you know
that you really have got your hands on the right molecule. There may be
other such moleculaes, but, for now, there's going to be a fair amount of
attention paid to APP and presenilin, because, again, you can be very sure
with autosomal dominant genetics. There should also be work on ApoE, antichymotrypsin,
and alpha-2 macroglobulin, because, again, genetics is really the only
test we have of whether something is relevant in vivo for a given
ARF: My second to last question would be your suggestions for
NIA to fund what type of projects in AD research?
HP: There would be two. One would be the study of transport
along microtubules and other cytoskeletal stuctures in AD and the other
would be cell cycle studies. Those would be my two because I think they're
not appreciated, they need an RFA to get them going and that would be good.
How I would change the NIH, I could spend another hour on that....
ARF: What would be your advice to graduate students and post-docs
at this stage in your career? Is there anything you want to say to the
HP: Have faith in yourself and your ideas. Ideas are the only
thing that distinguishes an average graduate student and scientist from
a great scientist. I wouldn't say all ideas are good. Each one has to
be tested, and it has to be subject to testing. But the way we think about
the world using the data we have collected in the laboratory is what makes
those data useful and important. You can learn how to collect data, but
learning how to think is harder. Unfortunately, you will find that many
scientists are against ideas or they're against a specific idea (they don't
happen to like it). There's a common phrase that "ideas are cheap".
That's the worst possible thing to say to a young scientist. When that
happens it's usually because the person who's saying it is going to steal
the idea and in that case, of course it's cheap. The other possibility
is that they're just trying to be nice because they think the idea is lousy.
That is no more helpful. The purpose of an idea is to stimulate thinking,
to provide a framework in which to view current data, and to suggest specific
new experiments. Those are real benefits. In fact, in my own experience,
ideas can be very costly. The wrong ones can dominate a field and result
in a tremendous waste of time and energy; the right one can give us the
code by which to connect a confusing array of seemingly unconnected facts.
So choose your ideas carefully. Then, as long as their light holds, pursue
them through hell and high water.