Live Discussion: Deficiency of Presenilin 1 Inhibits the Normal Cleavage of the Transmembrane Domain of Amyloid Precursor Protein

Deborah Watson and Kenneth Kosik led this live discussion on 30 January 1998. Readers are invited to submit additional comments by using our Comments form at the bottom of the page.

Paper Under Discussion: De Strooper B, Saftig P, Craessaerts K, Vanderstichele H, Guhde G, Annaert W, von Figura K, Van Leuven F. Deficiency of Presenilin 1 Inhibits the Normal Cleavage of the Transmembrane Domain of Amyloid Precursor Protein. Nature. 1998 Jan 22;391(6665):387-90. Abstract

View Transcript of Live Discussion — Posted 15 February 1998

Background Text
By Deborah Watson and Kenneth S. Kosik. Dr. Bart De Strooper's responses are listed below each question.

Many independent observations have converged around the conclusion that mutant presenilin molecules alter APP processing at the gamma-secretase site. In particular, increased production of Aβ ending at residue Ala-42 appears to contribute to Alzheimer's disease pathology in patients carrying any one of the numerous identified hereditary mutations in the PS1 or PS2 genes. The mechanism by which mutant presenilin causes this increase is a matter of great interest, and the demonstration that APP co-precipitates with presenilin suggests that a significant interaction between the PS and APP in intracellular membranes - with the possible involvement of other binding partners such as β or delta-catenin - may be occurring.

This concise and timely paper by de Strooper et al demonstrates that presenilin is required for gamma-secretase processing of APP (both the 40- and the 42-cleaving activities). The authors examined both endogenous mouse APP and virally-introduced human APP in neuronal cultures of presenilin 1 knock-out mouse embryos and showed that levels of both Aβ and p3 dropped sharply. The precursors of these two fragments (the p10 and p12 membrane-associated C-terminal stubs of APP) accumulated 2- to 5-fold in cell membranes, consistent with an arrest of gamma-secretase cleavage of APP. Both alpha- and β-secretase-cleaved forms of secreted APP (APPs) were present at normal levels.

Discussion Questions and Dr. De Strooper's Responses
Question 1 Intriguing as the paper is, it still does not identify the mechanism by which mutant presenilin causes increased Aβ42 production. Some hypothetical roles for PS include: trafficking molecule; positive regulator of gamma-secretase activity; or gamma-secretase itelf. What are others? How would they tie in to the findings of De Strooper et al?

Bart De Strooper replies: See my answer to Dr. Robakis's comment

Question 2 Previous work has shown that the over-expression of human mutant PS1 in mice increases Aβ42 levels even in the presence of the endogenous mouse PS1. For this reason and others PS mutations have been considered gain of functions. How do these new observations fit with the gain of function idea and does the observation that neurons from +/- heterozygous knockout animals show no change in APP processing offer any additional insights?

Response: Since absence of the PS1 gene inhibits gamma secretase activity, it seems logical to conclude that the increased Aβ42 production observed with the clinical mutations is a gain of function. However, it is a selective gain of function, since the mutations only increase Aβ42, not Aβ40. It could be that the mutations cause a slight change in the way PS1 is presenting/exposing APP to gamma secretase. It seems more difficult to understand this result if the PS1 knockout effects are interpreted as being the result of an APP transport deficiency. The statement that in the +/- animals no effects are observed with APP processing is not entirely correct, since we see a small decrease in amyloid peptide production in these cultures as well.

Question 3 Knockout of presenilin 1 reduces gamma-secretase processing of APP, but does not abolish it completely. Is it reasonable to expect that endogenous PS2 might be responsible for the residual activity?

Response: That is what we suggested. The PS2 knockout, and double PS1/PS2 knockout will give us an answer to this question. However, it is also possible that we are looking here at the basal activity of gamma secretase

Question 4 Though multiple labs have demonstrated an interaction between APP and PS, other labs have contested this finding. Does the decrease in gamma-secretase processing of APP in PS-knockout neurons reflect on this issue?

Response: In my opinion our results at least suggest that there is a temporary/ functional interaction of PS1 with APP(fragments). On the other hand our data demonstrate that PS1 is not needed for the normal maturation of APP and that it is only later in the biosynthetic pathway (after alfa- and β-secretase cleavage) that PS1 plays its role. The cited publications indicate that it is mainly "immature" APP that binds to PS1 which is somewhat difficult to conciliate with the findings in our publication.

Question 5 If we conclude that PS1 can in some way regulate gamma-secretase activity, what is the role of the endoproteolytic cleavage in this activity?

Response: Interesting question, no answer.

Question 6 Both 40- and 42-cleaving activities were inhibited equally in the cells lacking presenilin. Does this shed any light on whether the activities arise from the same or different enzymes?

Response: The data suggest at least that PS1 influences both gamma secretase activities.

Question 7 The search for proteins that interact with presenilin will now surely intensify. What technologies are available for discovering proteins which interact within their transmembrane domains?

Response: I would like to refer to a recent paper by Rawson, R.B., Zelenski, N.G., Nijhawan, D., Ye, J., Sakai, J., Hasan, M.T., Chang, T.Y., Brown, M.S., Goldstein, J.L. (1997) Complementation cloning of S2P, a gene encoding a putative metalloprotease required for intramembrane claevage of SREBPs. Molecular Cell, 1, 1-20, 1997

Question 8 Could inhibitors of presenilin activity serve as an effective therapeutic for Alzheimer's disease?

Response: We should be careful here:

1. We do not know whether PS1 function is essential in adult brain.
2. We do not know whether it is sufficient to simply decrease amyloid peptide production to stop Alzheimer's Disease (this is still a hypothesis, not a fact).
3. As we demonstrated, the inhibition of PS1 results not only in a decrease in amyloid peptide, but also in an increase in carboxyterminal APP fragments. I refer to the review of Dr. R. Neve on this site for further reflections on the possible role of these fragments in the pathogenesis of AD.

Nikolaos Robakis responds: The report by De Strooper et. al., (1),and the accompanying commentary by Haas and Selkoe (2), conclude that PS1 may directly modulate g-secretase cleavage and Ab production, and therefore, inhibition of its function in adult brain may serve as a suitable therapeutic target in AD. The controversy over the role of Ab peptides in the etiology of AD notwithstanding (3), a more parsimonious explanation of the observed partial inhibition of Ab production in the PS1-/- cells (1), is that in these cells transport of C-terminal fragments, (the substrates for g-secretase) to the g-secretase compartment is inhibited. This explanation would be in agreement with a suggested role of PS1 in membrane protein trafficking(4), and with its presence in transport vesicles(5). It also suggests that in addition to APP, PS1 may modulate transport of a host of membrane proteins, consistent with the embryonic lethality of the PS1-/- phenotype. If so, then inhibition of PS1 function may do more harm than good for AD patients.

Nikolaos K. Robakis
Professor
Mount Sinai School of medicine
New York, Ny 10029

Response to Dr. Robakis: I agree with the comments of Dr. Robakis. We, and also Dr. C. Haass and Dr. D. Selkoe, discuss explicitly this alternative interpretation of our results . The analogy with SREBP/SCAP is however striking and lessons could be drawn from this system that are very interesting for the Alzheimer field. The recent identification of the S2P proteinase which cleaves SREBP in the transmembrane domain shows us for instance how a gamma-secretase could look like.

Ken Kosik asks: Your paper shows a suggestive change in Fig. 1 g and it could be important for the mechanism.

Bart De Strooper replies: There is indeed a suggestive effect in the PS1+/- cultures. We did not do sufficient experiments to demonstrate that in a statistically significant way.

Ken Kosik asks: I'd like to revisit a previous question -- Bart, why is there the discrepant responses of Aβ in the supernatant where you see a nice effect vs the Aβ in cell extracts where the effect looks minimal?

Bart De Strooper replies: This is an interesting question, and we are puzzled about it. we are investigating this further using different APP mutants.

Ken Kosik asks: "Bart, could you reiterate why you believe PS1 mutants are gain of function?"

Bart De Strooper replies: Well, simply spoken: when PS1 is absent, amyloid peptide production goes down; when the mutations are there, amyloid peptide production goes up. Of course it is a "selective" gain of function, since only the 42 peptide goes up.

Ken Kosik asks: Isn't it unusual that so many different mutations in PS1 all lead to gain of function?

Bart De Strooper replies: This is indeed surprising, and I would suggest that the mutations all cause a small conformational change in PS1 which changes slightly the way in which PS1 presents APP to gamma secretase.

Carmela Abrahams asks: Do Bart's results mean that gamma secretase cleavage occurs in ER or Golgi, thus Aβ is completely formed there?

Bart De Strooper replies: I am not sure. An alternative interpretation of our data is that PS1 controls a transport step, eg the transport of gammasecretase from the endoplasmic reticulum to an undefined compartment downstream in the secretory or endocytic pathway, were interaction with the APP fragments could occur.

Jfloring asks: Is there any reason to imagine that the overexpression of APP may have led to the PS-1 effects? That is, what's going on with normal APP processing? Is the double knockout plus transgene on anyone's list?

Bart De Strooper replies: On the first part of the question: we confirmed most of our results in a qualitative way on the endogenously expressed mouse APP (see first figure of our paper). On the second part of the question: the double knock out is certainly on our list. I do not see directly whether the rescue with a clinical PS mutant could learn us more than what we know already.

Selected References
Xia, W. et al (1997) Interaction between amyloid precursor protein and presenilins in mammalian cells: Implications for the pathogensis of Alzheimer disease. Proc. Natl. Acad. Sci. USA 94: 8208-13. 1997 Jul 22. Abstract.

Weidemann, A. et al (1997) Formation of stable complexes between two Alzheimer's disease gene products: Presenilin-2 and β-amyloid precursor protein. Nature Medicine 1997 Mar; 3: 328-332. Abstract.

Scheuner, D. et al (1996) Secreted amyloid β-protein similar to that in the senile plaques of Alzheimer's disease is increased in vivo by the presenilin 1 and 2 and APP mutations linked to familial Alzheimer's disease. Nature Medicine 1996 Aug; 2: 864-870. Abstract.

Lemere, C. et al (1996) The E280A presenilin 1 Alzheimer mutation produces increased Aβ42 deposition and severe cerebellar pathology. Nature Medicine 1996 Oct; 2: 1146-1150. Abstract.

Zhou J, Liyanage U, Medina M, Ho C, Simmons AD, Lovett M, Kosik KS:Presenilin 1 Interacts in Brain with a Novel Member of the Armadillo Family. NeuroReport 8:1489-1494, 1997 (abstract) and erratum published 8: 2085-2090, 1997 (abstract).

Participants: Deborah Watson, Ken Kosik, Bart De Strooper, Kieran Breen, Steven Barger, June Kinoshita, Ben Wolozin, Wilma Wasco, Carmela Abrahams, Dmitry Goldgaber, Weiming Xia, Taisuke, Rong, Walker McGraw, Minami, Jean Floring, Chris Weihl.

Note: Transcript has been edited for clarity and accuracy.

Deborah Watson: I'd like to start by asking Dr. DeStrroper about the effects of one copy of PS on APP processing. Is there an intermediate effect between +/+ and -/-?

Bart De Strooper: We did not do extensive experiments on the +/- but there might be a decrease in amyloid peptide production in the +/- cultures as well.

Ken Kosik: But your paper shows a suggestive change in Fig. 1 g and it could be important for the mechanism.

Steven Barger: Steve Younkin's group presented data at Neuroscience from a PS antisense model. Their results were essentially the opposite of this knockout model -- any speculation on the resolution?

Ken Kosik: Responding to Steve Barger: could you tell us how the effect was the opposite?

Bart De Strooper: Hello everybody (with a lot of delay). First again on the PS1+/- .There is as I told, a decrease in amyloid peptide production in the PS1 +/- cutltures. We focussed however until now mostly on the PS1-/-

Steven Barger: They showed that antisense of PS1 INCREASED Aβ 42(43). One caveat is that it was in HEK cells.

Ken Kosik: Did they prove that they completely suppressed the expression of PS1? That's for S.Barger.

Ken Kosik: Weiming, do you have any comments about the heterozygotes?

Weiming Xia: In what aspect of heterozygotes?

Ken Kosik: Regarding Aβ production and CTF production.

Weiming Xia: We have not tried Aβ detection on hetorozygotes.

Steven Barger: No, I don't think there was complete annihilation of PS1; antisense is almost never that complete. So, you're right, there's a quantitative issue to deal with. But this might be a mechanistic hint, as well (?).

Deborah Watson: Do you have any more ideas, Bart, about the residual Aβ production in PS 1 null mice?

Weiming Xia: Is there any linear relationship regarding the decrease of PS1 expression and changes of Aβ, particularly Aβ 42?

Ken Kosik: Bart, could you reiterate why you believe PS1 mutants are gain of function?

Bart De Strooper: to Deborah Watson: either it is PS2 or it is residual activity of γ-secretase.

Rong: Can you explain why the decrease of intracellular Aβ is not so significant Taisuke: Did you see the expression of PS2 in PS1 -/- mice?

Ken Kosik: Bart, do you agree that the decrease applies to both intracellular and secreted a-β?

Weiming Xia: to Bart, any comments on p3?

Ken Kosik: Rong, do you think there are specific secretases for the other fragments?

Taisuke: Did you see the expression of PS2 in PS1 -/- mice?

Bart De Strooper: To everybody, It is impossible for me to really participate in the discussion. I get disconnected every few minutes. To Wxia: p3 behaves like β A4

Rong: To Kosik: At this moment, I don't think so. I may argue that same γ-secretase is responsible for generating Aβ42,40,38,37,..

Ken Kosik: To Rong: does the same secretase create all these fragments? Bart, sorry for the problems you are having, but we do appreciate your responses.

Steven Barger: The gain-of-function slant is satifying since it is consistent with the ability of PS mutants to restore development of somites, spine, etc. in the PS KO's.

Ken Kosik: Steve, are your referring to the Sisodia data about rescue? Wilma, do you have any thoughts about whether decreasing amounts of presenilin increase or decrease Aβ levels?

Steven Barger: Yes, to my knowledge, Sisodia's are only publicly presented data on this issue.

Rong: To Kosik: I don't have definite answers for that, but it is possible.

Ken Kosik: Steve and others: isn't it unusual that so many different mutations in PS1 all lead to gain of function?

Carmela Abrahams: Do Bart's results mean that γ secretase cleavage occurs in ER or Golgi, thus Aβ is completely formed there?

Ken Kosik: I'd like to revisit a previous question -- Bart, why is there the discrepant responses of Aβ in the supernatant where you see a nice effect vs the Aβ in cell extracts where the effect looks minimal?

Steven Barger: Yes, I certainly agree that the number of mutations doesn't fit Occam's razor. One thing that initially occurred to me is that the function we talk about gaining or losing may be two different functions: 1) a developmental role and 2) APP processing.

Ken Kosik: Bart, are you still with us?

Bart De Strooper: Ken, would it be possible to pool the questions and to send them to me by email. I get only part of the discussion on my computer.

June Kinoshita: Apologies to all. Apparently Bart's server is having problems. I will go through the questions for Bart and email them to him. The questions and his responses will be posted on the newsgroup.

Ken Kosik: Thank you June! Steve, even applying gain of function just to APP processing we have a problem with regard to Occam's razor.

June Kinoshita: We can use this time for discussion of these issues among ourselves. I'm sure many interesting questions will grow out of the discussion.

Ken Kosik: Rong, do you then believe the γ secretase is not highly specific for a particular amino acid and are any of these fragments seen in vivo (37, 38...)?

Weiming Xia: Rong, do you think there is one γ-secretase producing Aβ ending at different positions?

Jean Floring: Steve, would you repeat the paradox (vs Younkin's results). My computer crashed.

Bart De Strooper: To all, I am leaving and will do as June suggested.

June Kinoshita: Thank you so much, Bart. Sorry this Brave New Cyberworld isn't always what it's cracked up to be!

Weiming Xia: Thank you, Bart.

Ken Kosik: Thanks Bart.

Deborah Watson: Carmela, do you believe there are other sites besides ER/Golgi where Aβ is produced--if you say endosomes could you also be more specific? (From Kosik)

Carmela Abrahams: Nice work. Thank you Bart.

Steven Barger: Younkin used antisense to decrease PS1 levels in HEK293 cells and showed an increase in Aβ 42(43) (in contrast to the decrease that occurred in these KO cultures).

Carmela Abrahams: Debra, I don't think at this moment we can tell whether Aβ is formed also in endosomes or gets there from the ER. Does anyone who is plugged in know more?

Jean Floring: Thanks, Steve..I'll reserve my opinion about antisense. Anyone: Is there any reason to imagine that the overexpression of APP may have led to the PS-1 effects?

That is, what's going on with normal APP processing? Is the double knockout plus transgene on anyone's list?

Rong: Yes, all of these Aβ fragments have been detected in vivo.

Ken Kosik: Rong, does the concentration of the fragments differ greatly?

Rong: The are more or less associated with Aβ40 and Aβ42.

Steven Barger: To Ken: Actually, having two functions for PS might explain why AD would involve so many mutations: the developmental role would select (in terms of lethality) for mutations that were mild enough to retain the developmental role. But ...

Taisuke: To Jfloring: I transfected only PS2 mutant cDNA into Neuro2a cells, and endogenous Aβ42 was increased.

Steven Barger: ... But once this developmental selection is satisfied, there might be many routes to loss-of-function in the APP processing role.

Ken Kosik: Steve, are you suggesting that the PS1 mutations are loss of function with regard to APP processing?

Rong: I have to leave. It is very nice to discuss with everyone here. Thanks.

Steven Barger: Yes, Ken, I think that's a possibility.

Ken Kosik: I'd like to thank everyone for participating. Please make any last comments now that can be posted to Bart.

June Kinoshita: Thank you Ken for moderating and for putting together the thoughtful discussion.

Steven Barger: Very intriguing work, congratulations!

June Kinoshita: Again, please check the newsgroup for questions and responses from Bart.

Ken Kosik: Thanks everyone. Good-bye.

June Kinoshita: Bye Ken! Thanks to all for participating!

Jean Floring: Steve: what's your email address?