Peter Nelson, with Dennis Selkoe, John Hardy, and Alain Israel, led this live discussion on 5 May 1999. Readers are invited to submit additional comments by using our Comments form at the bottom of the page.
Live discussion with Dennis Selkoe, John Hardy, and Alain Israel, moderated by Dr. Peter Nelson, was held on 5 May 1999 at 12:30 p.m. (EST).
Participants: Peter Nelson, Dennis Selkoe, John Hardy, Alain Israel, June Kinoshita.
Note: Transcript has been edited for clarity and accuracy.
Peter Nelson: I'd like for people with questions to ask them
privately to me. People with comments should make them, but preceded by
COMMENT, which do not require response. Otherwise, I'll give
questions and have people on the panel answer those questions. Is that all
right? I thought, though, it would be good to have some introductory remarks.
Peter Nelson: One thing that's a bummer is that we can't CUT
Selkoe: Peter, do you want me to give an introductory remark?
Peter Nelson: Yes, please Dr. Selkoe.
Selkoe: My colleagues and I believe that PS will be shown
to be an intramembrane-cleaving aspartyl protease or a unusual diaspartyl
cofactor for the enzyme. We think current data is more consistent with this
model than with a principal role for PS in membrane trafficking.
Peter Nelson: Any other panelists with a remark? Otherwise,
we'll go on....
Hardy: COMMENT The Nature paper reported provocative and interesting
experiments which have revitalized the field. They are hypothesis-driven
experiment which paid off and it is always difficult to argue with that.
I am agnostic about whether presenilin is γ-secretase. What would we expect
of γ-secretase; binding to substrate (c99) and binding to inhibitors: neither
have yet been shown.
Peter Nelson: I have a question for the panelists, just to
get the ball rolling. Is there anyone, knowing the current experiments,
who does NOT think that presenilin will turn out to be, itself, THE γ-secretase?
Peter Nelson: So Dr. Hardy is the only skeptic?
Aisrael: Comment : I agree with J. Hardy : none of the presented
data unambiguously proves that PS is the secretase (although the hypothesis
is clearly exciting).
Peter Nelson: (to Drs. Hardy and Israel) Would binding/association
data be what you would seek for proof?
Hardy: Yes: both to substrate and to an inhibitor. That would
be close to being proof in my view.
Aisrael: Better would be in vitro cleavage with recombinant
PS (might be tricky because of the requirement for membrane).
Peter Nelson: Okay. To the believers: are such
experiments under way/ doable?
Ccweihl: COMMENT: PS exists primarily as stable cleaved fragments
yet the Selkoe model suggest that the holoprotein is active. What are the
fragments doing that they remain stable in the cell?
Weimingxia: If the substrate (C99) transiently interacts with
enzyme and is immediately cleaved to generate A-β, it would be hard to
catch the complex.
Selkoe: Yes, both are good experiments and are underway. The
binding of inhibitor to PS is an important experiment that is feasible.
We are trying this now. The reconstitution of Aβ generation with pure
PS and pure C99 is beginning but will clearly take quite a lot of work and
luck. I don't imagine we'll have an answer very soon for that one. The field
may first need to discover the critical cofactors that bind to and stabilize
Hardy: Why should PS form a complex with APP, but not (yet)
Selkoe: We think C99 has to form a transient complex with
PS, but as Weiming just said, it may be hard to capture it in the act.
Weimingxia: If the inhibitor is potent enough and acts as
an irreversible inhibitor, it will be easier to detect this binding.
Peter Nelson: There is an extensive scientific literature
regarding presenilin to work with. Chris Weihl asks how the processing of
β-catenin fits into this scheme. Also, there was some evidence that a
g-protein (pertussis-blocking function) is involved. Dr. Selkoe and Dr.
Hardy mentioned substrates and cofactors. Can you comment specifically on
these, especially β-catenin?
Selkoe: I doubt that catenins will turn out to be critical
for the role of PS as γ-secretase, because catenins bind to PS1 alone
yet PS2 has the same function as PS1 vis a vis γ-secretase. I think
that only conserved regions of both PS proteins will turn out to be important
in regulating APP and Notch cleavages.
Peter Nelson: Any comments on this?
Ccweihl: Comment: The role of catenins in PS fits more into
the scheme of trafficking (Nature Medicine paper, Hyslop).
Hardy: Specifically, though, why does PS (apparently) interact
with full length APP? Doesn't this too, fit more with a trafficking role?
Selkoe: I think we can capture full length APP with PS because
that is the beginning of the catalytic complex. Thereafter PS endoproteolysis
needs to occur and so does β-secretase cleavage of APP. Then the complex
is ready to fire and C99 is probably not left intact very long.
Aisrael: concerning Notch, the Fortini's paper introduces
a note of doubt by showing that a constitutive form of Notch apparently
can signal in a PS-minus background. Since it requires processing to work,
it seems to suggest that processing takes place in the absence of PS. This
is quite hard to reconcile with the other papers (especially the Struhl/Greenwald),
unless you admit that signaling might take place in the absence of processing.
Peter Nelson: All the papers seem to demonstrate normal processing
of NOTCH/BAPP without presenilin. Is this not true?
Peter Nelson: I mean, prior to proteolysis....
Selkoe: This is not true. PS must be present and wild type
to enable Notch intramembranous cleavage. Fortini's result could
be explained by some of his construct not remaining membrane-bound. It could
then go directly to the nucleus without requiring cleavage.
Aisrael: No, I think the idea is that Notch processing (at
least the step that takes place in or near the membrane) does require PS
Peter Nelson: That's what I meant. But, significantly, PRIOR
steps are not effected.
Aisrael: As Struhl/Greenwald used basically the same construct,
it should behave the same in the 2 papers. However the 2 authors assay different
Hardy: I note that Dennis' comment above gets quite close
to the idea that PS is involved in trafficking (“beginning of catalytic
Selkoe: No, I mean a physical complex of PS with its substrates
in one submembranous domain. I don't think this relates to the usual way
we use the term membrane trafficking. So I don't think that PS trafficks
membrane subdomains in general.
June Kinoshita: It seems that Pete's system has crashed. Dr.
Hardy or Dr. Israel, do you have any additional comments or questions to
raise regarding the trafficking vs. processing issue?
Ccweihl: COMMENT: What about the Narusse/Sisodia paper demonstrating
that several proteins are misstrafficked in PS1 knockout fibroblasts.
Selkoe: I am skeptical that that paper is pinpointing direct
trafficking functions of PS. Instead, I think they correctly observed several
downstream effects of completely and permanently deleting PS1 from the mouse.
Ccweihl: Comment: Will we identify new γ-secretase substrates
other than Notch/APP and what is the consensus sequence or determinants
Weimingxia: In PS1 knockout fibroblasts, APP and C99 are not
misstrafficked. The subcellular distribution of APP and C99 is not altered
in PS1 KO fibroblasts.
ToddGolde: Comment: Dennis you and others have shown that
γ-secretase cleavage of A-β 40 and A-β 42 are somewhat pharmacologically
distinct-indicating that multiple proteases may contribute to the γ-activity.
This is somewhat incongruous with the observation that PS KO or asp mutant
expression decrease A-β production equally for all species examined.
Also what about the effect of PS KO on inducible α-secretase activity?
It seems PS regulates multiple proteolytic events of membrane bound substrates
making it unlikely they are the secretase themselves.
Selkoe: Yes, I suspect there will be other substrates and
also other intramembranous proteases in the fullness of time.
Selkoe: I think the different IC50's of a compound required
to inhibit 40 vs 42 cleavage can be accommodated by a role of PS in a catalytic
complex. I think some C99 molecules are oriented with the 42-43 bond facing
the two PS asps while most are oriented with the 40-41 bond facing them.
I don't know of clear evidence that con. α-secretase activity is affected
by the asp mutations in PS.
Ccweihl: Comment: To Selkoe what consequence will γ-secretase
inhibitors have pharmacologically, if we inhibit cleavage of other proteins,
Selkoe: Like hmg co A reductase inhibitors we hope to only
inhibit PS 30-40% and allow Notch and APP cleavages to still occur. Otherwise
we face likely marked toxicity.
Peter Nelson: A question to Dr. Selkoe: Why didn't the delta-nine
PS in your experiments change the output of c83 and c99? Wouldn't it be
expected to increase the production of those peptides?
Selkoe: Peter, No. Delta E9 acts like a wild-type PS heterodimer
(except for the new missense mutation that it contains). Therefore, delta
E9 should allow normal amounts of C99 and C83 fragments to occur.
ToddGolde: COMMENT: to Weiming, the amount of APP that goes
to A-β is quite small, therefore I do not find it surprising that no
effects on subcellular distribution of CTF are distinct in PS KO. This will
have to be looked at in much finer detail. Not the rather crude
subcellular fractionations that have been employed.
Hardy: Comment from Todd to Dennis: The data on α-secretase
is from KO lines it has not been done with the asp mutants, yet, However
Roger Nitsch and colleagues also showed that FAD-linked PS1 mutants fail
to augment inducible α-secretase activity, while overexpression of wt
PS holoprotein does.
Peter Nelson: Okay, then, on to γ-secretases. I'd like
a variety of input here. How much hope do they provide? What is their likelihood
of success, toxicity, impact, and cost? And, have they been tested in animal
Peter Nelson: ...recognizing that there's a lot at stake here,
and not all beans can be spilled in this context.
Selkoe: We don't know until some published data on γ-secretase
inhibitors in vivo come out. They may have some toxicity as so many compounds
do, but hopefully not so much that they will no longer be useful clinically.
Ccweihl: Comment to Selkoe. Will any method to stop PS1 cleavage
block γ secretase (proteasome inhibitors).
Selkoe: I would think preventing PS heterodimer formation
would be like inhibiting γ-secretase.
Peter Nelson: Anybody else with γ-secretase thoughts?
Vangool: Back to trafficking: I think the trafficking hypothesis
does not necessarily imply that APP or its derivatives are misrouted in
the absence of PS. In fact isn't it so that the present data can be explained
by misrouting of the secretase or regulatory factors thereof?
Peter Nelson: Asks June: how early would one have to start
take a γ secretase inhibitor to prevent AD?
Selkoe: I think that we will need to start inhibitors in people
as early as possible, even as a preventative ultimately.
Peter Nelson: As prevention just for people at risk, or everyone?
Selkoe: I think in the early years, after proof of efficacy,
it will be used just for people at risk. But later on could be used in the
wider elderly population, assuming its quite safe, like a statin drug for
Ccweihl: Comment: Will overexpression of PS1 cause an increase
in γ secretase activity even though the holoprotein is not cleaved and
then how are we getting more A-β in transgenics.
Selkoe: I don't think overexpressing PS1 will increase γ
secretase activity because only the stable amounts of heterodimers likely
Ccweihl: If overexpression of PS doesn't increase γ secretase
activity then why do we see more A-β deposition in transgenics? Is this
solely because of the increase in 1-42?
Peter Nelson: Okey-doke, then. If γ-secretase inhibitors
do not interest people directly, could we have some responses on other topics?
Perhaps a role for ApoE? The connection of Alzheimer's disease and developmental
paradigms? The generalizableness of the intramembranous domain proteases?
Ccweihl: I am still curious what role β catenin plays in
PS1's function. And is this trafficking or proteolysis?
Peter Nelson: ...and, a question to Hardy. Do you have any
remarks to Dr. Selkoe about HIS remarks about your article? For example, about ApoE's
Hardy: I strongly think that the data relating ApoE to age
of onset in families with APP mutations (and Down syndrome) suggests that
ApoE is downstream directly of A-β in those families, but the absence
of an effect in PS families suggests that it is not downstream in those.
To my mind, this suggests there are two A-β routes to AD: one ApoE dependent,
and one ApoE independent.
Selkoe: As you know John, I disagree. We have very good evidence
that APP and PS mutations operate at the molecular level by a very similar
mechanism. The lack of clinically detectable effect of ApoE4 in PS gene
carriers could have a number of other explanations and does not need to
be interpreted as a genetic upstream vs. downstream issue.
Peter Nelson: To Hardy and Israel: how to best resolve the
ApoE question experimentally?
Hardy: Of course, Dennis is right; this is not definitive.
Finally: it's a good paper, but leaves space for Nature to publish conclusive
data either way.
Peter Nelson: Perhaps that is a good note to wrap up on unless
another topic/question is to be asked. Thank you all VERY much for being
here. It was extremely kind of you. The transcript of this discussion will
be posted and additional comments can be added as anyone wishes, through
the Alzheimer Research Forum web site.
Selkoe: Thanks very much Peter. It worked well. Bye.
June Kinoshita: Thanks to all, especially Dennis, John, Alain
Aisrael: It was an interesting experience. Bye.
June Kinoshita: Problem with this virtual stuff is I can't
invite you all out to lunch afterward. Bye!
Aisrael: You mean dinner.
June Kinoshita: Pardon! For you, yes, dinner.
June Kinoshita: Ciao!
A classic hallmark of Alzheimer's disease is the
accumulation of amyloid β peptide deposits in the
cortex. Generating β requires cleavage within the
transmembrane domain of its precursor, APP. An enzyme,
dubbed γ-secretase, has been posited as the agent
responsible for this cleavage, but this enzyme has managed
to elude every effort to isolate and identify it. Now
four reports published in the 8 April 1999 issue of
Nature appear to be closing in on this long-sought goal.
Three of them show that presenilin-1 is necessary for
the cleavage of Notch, which mediates cell-cell interactions
in determining cell fate during development. To carry
out its signalling function, Notch requires cleavage
in its transmembrane domain in a manner reminiscent
of β production. The fourth paper reports on mutations
in PS-1 that completely abolish APP cleavage and suggests
that PS-1 itself is γ secretase. A News and Views
article (from which we cribbed the title for this discussion)
comments on the four reports and raises some issues
for further discussion.
Note: Full text is available only to subscribers
- Two transmembrane aspartates in presenilin-1 required for presenilin
endoproteolysis and -secretase activity. Michael S. Wolfe, Weiming Xia,
Beth L. Ostaszewski, Thekla S. Diehl, W. Taylor Kimberly & Dennis J.
Selkoe. Full text.
- A presenilin-1-dependent -secretase-like protease mediates release
of Notch intracellular domain. Bart de Strooper, Wim Annaert, Philippe
Cupers, Paul Saftig, Katleen Craessaerts, Jeffrey S. Mumm, Eric H. Schroeter,
Vincent Schrijvers, Michael S. Wolfe, William J. Ray, Alison Goate &
Faphael Kopan. Full text.
- Presenilin is required for activity and nuclear access of Notch in
Drosophila. Gary Struhl and Iva Greenwald. Full text.
- Neurogenic phenotypes and altered Notch processing in Drosophila Presenilin
mutants. Yihong Ye, Nina Lukinova & Mark Fortini. Full text.
- News and Views: In Search of γ-secretase. John Hardy and Alain
Israël. Full text.
Points Raised in News and Views by
John Hardy and Alain Israël
- "The results of the mutagenesis experiment (Wolfe, et al.) are
provocative, but they do not conclusively show whether the presenilins
are important in trafficking or in cleavage. They could be causing
defects upstream of the cleavage events, either by altering
trafficking of the substrates (APP and Notch), or by altering trafficking
or activation of the protease or proteases involved in γ-secretase-type
- " ...it is notable that, in humans, the apolipoprotein E genotype
modulates the age of onset of Alzheimer's disease encoded by APP-717 mutations...
but not that of presenilin-encoded disease. From this it would seem that
these events are genetically distinct, a conclusion supported by other
lines of evidence" [see below].
- "At Keystone, R.
Nixon [Real Audio file] reported that presenilin and APP mutations have different effects
on the vesicular trafficking of APP...."
- ".... the function of spe-4, the 'forgotten' presenilin, seems
more closely connected to vesicle trafficking in C. elegans testes than
- " ... Most (but not all) studies in mammalian cells indicate that
presenilins are located in the endoplasmic reticulum or in the Golgi,
yet the processing events described above probably occur at the plasma
- "Both views remain viable - that presenilins are indeed γ-secretase,
or that they instead directly traffic APP and Notch to the right cellular
compartment for γ-secretase processing. Direct biochemical experiments
will be required to distinguish between them."
- "Finally, a word of caution. Drugs targeting γ-secretase...
may have unwanted immunosuppressive effects [due to role of Notch in the
- Jackson GS, Hosszu LL, Power A, Hill AF, Kenney J, Saibil H, Craven CJ, Waltho JP, Clarke AR, Collinge J. Reversible conversion of monomeric human prion protein between native and fibrilogenic conformations. Science. 1999 Mar 19;283(5409):1935-7. PubMed.