Chris Weihl, with Bruce Yankner, Benjamin Wolozin, Eddie Koo, and Kenneth Kosik led this live discussion on 13 August 1999. Readers are invited to submit additional comments by using our Comments form at the bottom of the page.
Live discussion held 13 August 1999 and moderated by Chris Wheil.
Participants: Chris Wheil, Bruce Yankner, Ben Wolozin, Ken Kosik, Eddie Koo
Note: Transcript has been edited for clarity and accuracy.
Chris Wheil: Welcome Drs. Kosik and Wolozin. Thank you for joining us. We will try to start at 12pm EST.
Chris Wheil: We are waiting on Dr. Yankner and then we can begin.
Edward Koo: This is Eddie Koo, I'm here too.
Chris Wheil: Welcome Dr. Koo.
BWolozin: Hi Eddie!
Chris Wheil: Now we can begin. I want to welcome everyone to this discussion of the role of PS1 in beta catenin signalling. I thought that we might open the forum with a brief discussion about why papers have demonstrated different results in regard to PS1 function of beta-catenin. In particular what is the best measurement of beta-catenin function?
Bruce: I would say that the best measure is its biological activity in vivo.
Chris Wheil: Dr. Yankner suggested the best measure is in vivo function in animal models. He mentioned some preliminary data. Perhaps Bruce would elaborate.
Bruce: That remains to be determined
Edward Koo: I agree with Bruce to a large extent. However, the in vivo results have not all been consistent depending on the organism. So I think other evidence needs to be brought in to go along with an in vivo outcome.
Bruce: Our group and Rich Carthew's group have results which suggest that PS1 can potentiate the biological activity of beta catenin in Xenopus and Drosophila. Although these results are consistent with a stabilizing effect of PS1, the precise mechanism remains to be determined.
BWolozin: It looks like there is no response to the question about in vivo models, so perhaps we should try to figure out which are the best methods when using cell culture (acknowledging that in vivo is the gold standard).
Chris Wheil: Dr. Wolozin. Your data with the luciferase reporter seemed superior in gauging the function of beta-catenin compared with stability assays.
BWolozin: Perhaps we should focus on cell culture models, since there is more familiarity with that.
Kosik: Bruce, how did you do the experiment regarding the potentiation?
Bruce: We collaborated with Xi He to inject PS1 constructs into Xenopus and assayed axis duplication as an indicator of the Wnt/catenin activity
Chris Wheil: Bruce what were the results?
Bruce: Wild-type PS1 induces complete and partial axis duplication. FAD PS1 mutations show loss of function.
Chris Wheil: Dr. Koo would you please comment on this data as it is contrary to your theory of beta-catenin and mutant PS1.
Edward Koo: I'm trying to absorb Bruce's results. I would agree that our results did not have a functional read out, such as Bruce's xenopus injections. On the other hand, we've always looked at endogenous catenin levels, not co-expressed catenin. We now have some data using nuclear reporter assays that confirm our catenin data. So the differences to Takashima's papers may rest with cell type and method of expression, rather than different biologic effects per se.
BWolozin: Loss of function appears consistent with what many have observed - even the Aβ effects could be due to loss of function.
Edward Koo: Ben, what do you mean by loss of (presumably beta-catenin) function and Aβ effects?
zhenglab: I do not think the Aβ effect is a loss of function since in PS1+/- mice where the PS1 expression is reduced by half, there is no increase in Aβ.
Bruce: In all fairness to Eddie, our results are not completely at odds with his points about catenin pools because this assay required coexpression of catenin. However, Rich Carthew has selectively inhibited PS1 expression at different stages in Drosophila development and observes a partial Wingless (loss of catenin ) phenotype.
BWolozin: If PS1 is controlling degradation of proteins, then having half as much PS1 would still allow correct processing of Aβ and of catenin, hence no change.
Edward Koo: Rich Carthew's results surprise me. I asked Mark Fortini about his studies. So far, he tells me that none of his drosophila PS mutants look anything like Wnt/armadillo phenotype. They all look more like notch phenotypes.
zhenglab: But in any case, Aβ in PS1/FAD and PS1+/- is not the same.
BWolozin: Bruce, what is the role of catenin in axis duplication - is it due to adhesion effects? And........can you rule out effects on other signaling systems - like Notch?
Kosik: Regarding the expression of h-PS1 in other systems we have expressed h-PS1 in fly and see a lethal phenotype that is predominantly an adhesion defect.
Edward Koo: I think the adhesion versus signaling is very complicated. If you over-express cadherins, you can sop away the catenins from signaling, but it also increases catenins at cadherin sites. So it is not obviously an either/or situation.
Bruce: Eddie, Fortini knocks out PS1 expression from the start. Rich used a new method, double-stranded RNA interference, to selectively inhibit PS1 expression at specific developmental stages. In the Fortini experiments the Notch phenotype could have obscured the catenin phenotype - but I am certainly not an expert on Drosophila development. Regarding Ben's question, Wnt induces axis duplication through a primarily transcriptional mechanism. I don't know whether the effector phase involves adhesive interactions.
Zhuohua: Just curious about zhenglab's comment. PS1+/- does not really explain the loss of function. since a mutant ps1 expression may replace the wild-type ps1, as we all know PS1 expression is regulated by some unknown limited factor. It would be interesting to know what percentage of mutant PS1 is in the patient v.s. the wild type PS1 (I mean the protein).
Chris: What role does beta-catenin have in AD pathology?
Chris: Does increased or decreased Wnt signalling increase BAPP production?
Kosik: We have data that expressing Wnt-1 in PC12 cells will increase Aβ, but cannot assume that it is direclty related to the effect of Wnt or some downstream effect.
Bruce: Ken, do you know the relative effects of Wnt-1 on Aβ 40 and 42?
Kosik: Bruce, we found the effect on 40 an 42 remains indeterminate because its levels were below the threshold of the assay.
BWolozin: Ken, Do you think that the adhesion defect is mediated by GSK and APC (or could it be explained by another adhesion system)? So far there hasn't been much talk about the connection between APC and PS1.
Kosik: Ben, my first impression was that the adhesion defect was due to titration of beta-catenin from junctions by PS1.
Chris: Dr. Koo, did you find an interaction between APC and PS1?
Edward Koo: We only looked at this once but could not co-IP APC and PS1. But the catenin complex is getting larger (axin, catenin, APC, dsh, GSK, among others). Maybe we looked at the molecule in the incorrect phosphorylation state.
BWolozin: Ken - Chris brought up an interesting connection relating to APC. So far there has been little talk of connections with APC. Do you think that the PS1/APC/Aβ connection might be operative here? My sense is that the junctions represent APC/cat binding. Is this correct?
Edward Koo: I don't think anyone has shown how the stabilizing or destabilizing effects of PS1 on beta-catenin works. It could in fact be through GSK, APC, or other molecules. So the speculations are very reasonable.
Chris: Iva Greenwald found that SEL-12 associated with SEL-10, a slimb homologue involved in ubiquitin ligase. This could be a mechanism too. Dr. Yankner has done some work in this area.
Bruce: We do not yet know whether the Fbox protein beta Trcp is also present in the PS1 complex.
BWolozin: Since I use a lot of cell culture, I was wondering if people could give their opinions on the relative merits of various assays - cat levels, nuclear cat, nuclear translocation, PS1 binding, reporters, etc.
Chris: Dr. Wolozin this a key issue to understand why there are different results.
Bruce: Ben, none of them work.
BWolozin: Bruce - can you elaborate?
Bruce: Only joking - but I think at the end we have to understand the activity in vivo.
BWolozin: Agreed, but in the meantime....
Chris: Murayama's data came very close to functional data (use of reporter luciferase).
BWolozin: Chris - the odd thing about the luciferase data is that WT PS1 reduced Tcf activity. However, the loss of function in mutant PS1 relative to WT was consistent with what everyone is saying.
Chris: Perhaps this relates to transient overexpression?
Bruce: Regarding Ben's question, one important technical issue when comparing cell lines stably overexpressing PS1 constructs is to control for expression levels. This is not trivial in the case of the PS1 exon 9 deletion in which the protein is not metabolized to fragments.
BWolozin: That would be my bet. We see differences between transients and stables.
Chris: Ben- differences in what way?
BWolozin: For instance, looking at Elk (sorry to switch systems) we see large effects on reporters when the PS1 is expressed transiently, but less significant effects in stables.
Chris: Dr. Koo mentioned that there are problems with overexpression of PS1 and the increase in artifactual full length PS1? In addition, it appears that mutant PS1 fragments may be more stable as well. (Mike Lee's Nature Med paper). Dr. Koo please comment.
Edward Koo: We've noticed different results depending on transients or stables. In addition, in our inducible system, the length of time of induction and level of induction also provides varying results. Our suspicion is the full length and fragments. That's the obvious culprit.
chenel: Dr. Koo, do you see differences in the complex formations with PS1 depending on transient or stable transfection?
Edward Koo: As to stability of fragments, we haven't looked carefully. Mike Lee's paper used Bruce's favorite model, i.e. in vivo from transgenic mice. So most of the in vitro studies may not be comparable. The del x9 is obviously more stable than WT PS1.
Bruce: We found that the stabilization of catenin by wt PS1 was less in stables than transients, but the mutants still appeared to destabilize in our hands. We used inducible Heks from Sam's lab, and stable constitutive overexpressor's from Dennis Selkoe's lab.
Chris: Ben- do you think that PS1 affects several signalling systems?
BWolozin: No question of that. It affects many systems - the hard part is to figure out which effects are direct.
Edward Koo: In our hands, the complex are much more variable in transient transfections. In fact, they are very difficult to detect.
dekang: I would like to add to Eddie's comments. In our experimence, transient PS1 transfection does not augment the endogenous beta-catenin/PS1 complex.
Edward Koo: By complex, I mean PS1 with GSK and beta-catenin. We can't see APC.
chenel: Dr. Koo, which proteins appear/disappear in the stable vs transient transfections?
Edward Koo: In our hands, it's very difficutl to increase the levels of PS1 NTF and CTF in transients. The full-length protein obviously comes up readily, but not the stable fragments. On the other hand, in the stables or conditional stables (ecdysone inducible), the fragments come up much more, in parallel with the full-length protein.
Kosik: Bruce, we have tried several times unsuccessfully to see beta-catenin instability in Jie's KO animals, but beta-catenin always looks like the controls on western blots. Please comment.
Bruce: Ken, when we examined Jie's cells we saw no significant change in full-length beta catenin, but observed increased appearance of lower MW catenin fragments. However, in Bart's PS1-KO cells, we see both the fragments and reduced full-length catenin. Interestingly, Bart's cells also appear more severely affected by the KO phenotype in other assays as well, like Aβ production.
Edward Koo: I'm concerned about Bart's KO cells when differences come up since his KO is at a more downstream exon. I don't [think it has been excluded that an N-terminal PS1 fragments is produced.
BWolozin: One of the things that I am wondering is whether PS1 connects directly to the proteosome, since all of the effects of PS1 seem to ultimately funnel through the proteosome. Anyone have any data or thoughts on this?
Chris: Could the problem with transients relate to apoptosis and caspase cleavage of PS1 and disruption of beta-catenin binding?
Edward Koo: Chris, that is a possibility but I don't think PS1 WT should be killing cells normally, even if caspase results in loss of catenin binding, as reported by Rudy's lab.
Chris: Ed even if it is overexpressed?
Edward Koo: In our hands, overexpression of WT PS1 does not induce apoptosis. But admittedly, it's another of the in vitro type experiments. Let me ask a question for a change: Has anyone looked at PS2 and beta-catenin or GSK or any other part of the complex?
Kosik: Eddie we cannot find a PS2 d-cat or beta-catenin partnership.
BWolozin: It seems that PS2 has a stronger connection to things that directly affect cell death - like D'Adamio's Bcl-X results.
Bruce: Rudy has told me that his lab has not observed PS2: catenin co-recipitation.
Kosik: PS2 mutations are also not fully penetrant
Chris: Ben- Is there more neuronal loss in PS2 FAD families?
BWolozin: Not to my knowledge - even though there is a catenin connection, I still think the important thing is that PS1/2 affects Aβ.
Edward Koo: Ben, are you implying that the PS2 catenin connection is at the cell death level, a la your studies with Luciano?
Bruce: Eddie, we see no PS1 (or PS1 fragments) in Bart's cells using a very high titre antibody to the extreme N-terminus.
BWolozin: Eddie I think that there are two issues. How does PS1 cause AD and how does PS1 affect catenin. The latter question is exceptionally interesting, but unfortunately, probably not relevant to AD.
Chris: Ken-What role does PS1 play in the metabolism of other catenin?
Kosik: Chris--no data on that one yet.
Chris: Ken- do you think that PS1 role in delta-catenin will be more important that beta catenin, because delta is more neuronally expressed?
Kosik: Chris-many of the same doubts about the role of beta-catenin in AD apply to delta-catenin.
Chris grins evilly.
Chris: Does beta catenin relate to AD?
BWolozin: However...I'm not implying a PS2 catenin connection. PS2 probably affects cell death via BclX or JNK.
Bruce: Ben, what do you think about the absence of a phenotype in the PS2-KO given your point about the difference in PS1/PS2 functions regarding cell death?
BWolozin: Bruce - there are other regulators of Bcl-X. PS2 might regulate Bcl-X or JNK, but might not be the major regulator.
Chris: Dr. Koo-- does mutant PS1 affect GSK-3beta activity on beta catenin or tau?
Edward Koo: I think we tried some phosphorylation experiments a while ago and they weren't very clean. So it is possible that stability or destablity of PS1 on catenins is via GSK. Ben or his colleagues in Japan may have more on this than I do.
Bruce: Eddie, we have recently evaluated catenin phosphorylation using antibodies specific for catenin phosphorylated at the GSK3 sites in collaboration with Xi He, and find that PS1-/- cells have higher steady state levels of PO-catenin than controls.
BWolozin: Perhaps we should think of the term scaffold or complex - that includes GSK and catenin and PS1.
Edward Koo: Ben is absolutely correct. The signaling or stabilizing complex of catenin is getting larger. To think only about PS1-cat or PS1-GSK is almost certainly incorrect.
BWolozin: Okay, now for the messy question - to reiterate Chris' question. Can anyone come up with a cogent explanation for why the catenin connection would impact on AD?
Chris: Apoptosis? and beta catenin destabilization? Dr. Yankner?
chenel: Have there been any FAD cases that are related to mutations in catenin?
Edward Koo: I guess they (the catenin mutations) would die of cancer before getting AD.
BWolozin: Judah Folkman might allow us to address that once he cures cancer!!
Bruce: Ben, that would depend on whether the catenin mutations are gain of function, i.e the phosphorylation sites (the mutations causing colon cancer) or loss of function, i.e. C-terminal mutations.
Chris: Bruce what about FAD mutants and beta-catenin phosphorylation? I like that theory: An increase beta-catenin stabilty = cancer and decrease in beta-catenin stability = neurodegeneration? Please comment.
Bruce: Chris, we are just beginning to address that issue with the phosphorylation-dependent catenin antibodies. No definite results yet.
BWolozin: Good thought Bruce (as always).
BWolozin: I have to run off soon, but I would like people's input onto the merits of various catein assays. Any last thought - other than frogs and flies.
Edward Koo: Ben, my take is that the nuclear translocation is least informative since the complex does not have to signal in the nucleus. Levels of catenin need to be correlated with reporter assays (and biological effects, if possible) for the best in vitro readout at this time.
Chris: Well the hour is almost up. One last comment from our four participants regarding the role of PS1 in beta-catenin signalling. Drs. Koo, Yankner, Kosik and Wolozin....
BWolozin: Ciao to everyone.
Kosik: Re the request for a last comment, I would look to dysfunction within synaptic and adherens junctions for a possible role of these proteins in AD.
Bruce: Perhaps the most interesting biological question is how the same molecule, PS1, can affect the function or trafficking of a soluble cytoplasmic protein like beta catenin as well as proteins in the secretory pathway such as APP and Notch.
Edward Koo: I agree with Bruce. If PS1 is not gamma-secretase, then the most cogent explanation at this time is that PS1 functions as an escort protein of some kind.
Chris: Thank you again to all of our participants. We are still far away from any definitive answers. Maybe at our next discussion well have the problem of AD solved!
By Chris Weihl
Originally cloned and identified in late 1995, mutations in presenilin-1 (PS1) and presenilin-2 (PS2) lead to an aggressive and early onset form of familial Alzheimer's disease (AD). To date >50 familial AD causing missense mutations have been identified in the presenilins. The putative structure and localization of the presenilins is presumed to be an 8 transmembrane domain, endoplasmic reticulum(ER)/golgi membrane resident protein with a large hydrophilic loop exposed to the cytosol (reviewed in Price, 1998).
Despite intense efforts over the past 4 years, the function of the presenilins remain unclear. Early data suggested the presenilins play a key role in development, since they are functionally homologous to a C. elegans protein, SEL-12, involved in Notch signal transduction. Moreover, PS1 knockout embryos demonstrate embryonic lethality and developmental disruptions reminiscent of Notch knockout mice (e.g. defects in somite segmentation). Notch signaling involves the transduction of a developmental signal from one cell via the surface ligand, Delta/Serrate, and its subsequent binding to the Notch receptor on an adjacent cell. Intracellularly-associated Notch then releases from the cell membrane and enters the nucleus turning on a specific developmental pathway.
The pathological consequences of the PS1 mutations may involve an increase in the misprocessing of the amyloid precursor protein, APP, resulting in the secretion of the more amyloidogenic Aβ1-42 peptide and enhancing amyloid plaque formation in FAD patients. This may be due to PS1's role in protein trafficking and perhaps its "proposed" secretase function. Others have suggested that mutations in PS1 enhance neuronal susceptibility to apoptosis.
Very recent data has emerged suggesting that PS1 plays a role in another developmental signaling pathway, Wingless, via PS1's binding to beta-catenin. This association has the potential of elucidating PS1's role in development and Notch signaling (as wingless and notch pathways may interact). In addition, FAD mutations in PS1 may disregulate beta-catenin signaling, thereby potentiating neuronal apoptosis. Finally, while evidence exists suggesting PS1's role in Aβ deposition, PS1's association with beta-catenin and its regulatory kinase, GSK-3 beta, (also known as tau protein kinase-1) link PS1 to the hyperphosphorylation of tau and neurofibrillary tangle formation(NTF).
PS1 Associates with Armadillo Repeat Proteins
Several studies have been aimed at identifying proteins that interact with the presenilins. Initial evidence suggested that PS1 associates with APP; however some papers disagree with this finding. In order to screen for proteins that might interact with PS1, Zhou and colleagues used the hydrophilic loop domain of PS1 as bait in a yeast-two hybrid screen. Yeast two hybrid studies express one protein fused to the DNA binding domain of a transcription factor (GAL4) as "bait" and screen a cDNA library with fused sequences to the GAL4 transactivation domain. Positive interactions are assayed by the ability of "bait" and "prey" to associate and lead to the expression of a reporter protein (beta-galactosidase). This screening strategy identified a novel armadillo repeat protein, delta-catenin, that was highly expressed in brain. Subsequent studies identified that PS1 could also associate with a delta-catenin homologue, beta-catenin in HEK293 cells (Zhou, et al., 1997).
Since this study, PS1 has been shown to associate and colocalize with several other armadillo repeat proteins including beta-catenin, delta-catenin (also termed NPRAP) and p0071 but not to associate with alpha-catenin or gamma-catenin (Levesque, et al. 1999; Stahl, et al., 1999). The interaction has been mapped to the C-terminal hydrophilic loop domain of PS1 (a.a. 263-407). In addition, the PS1:beta-catenin association is abrogated upon caspase mediated cleavage of PS1 (at a.a. 329 and 341) during apoptosis (Tesco, et al., 1998). FAD mutations in PS1 do not appear to disrupt the association of PS1 with beta-catenin, delta-catenin or p0071. The loop domain of PS2 has been reported to associate with beta-catenin (Levesque, et al., 1999); however several reports fail to see this interaction.
What Are Beta-catenin and Armadillo Repeat Proteins?
An armadillo repeat is a 42 amino acid motif involved in protein-protein association and is found in proteins, such as beta-catenin. Armadillo repeat proteins are a growing class of diverse signaling proteins involved in cell to cell adhesion, protein-protein interactions and signal transduction. Mutations in some armadillo repeat proteins lead to distinct pathologies ranging from cancer to neurodegeneration (reviews include Dale, 1998; Willert and Russe, 1998; Barth, et al., 1997).
The best-characterized armadillo repeat protein is beta-catenin. Beta-catenin is a component of the WNT signaling pathway of early development. WNT signaling allows for cell to cell communication during important developmental decisions such as cell-fate determination and vertebrate CNS pattern formation. The pathway of WNT signaling in a developing neuron is as follows:  a secreted or cell associated WNT ligand binds to an adjacent cells Frz receptor.  Frz activation transduces a signal to inactivate the serine/threonine kinase GSK-3 beta via an unidentified pathway mediated by another protein, Disheveled.  GSK-3 beta is responsible for regulating the stability of beta-catenin. When GSK-3 beta is active it phosphorylates beta-catenin at multiple sites targeting beta-catenin for ubiquitination and proteasomal degradation. When GSK-3 beta is inactivated (as in WNT signaling), beta-catenin is stabilized.  unphosphorylated and stable beta-catenin can then enter the nucleus where it co-activates the Tcf/LEF family of transcription factors setting the cell on its programmed developmental path of gene expression.
In the absence of WNT ligands and Frz receptors after development, the role of beta-catenin in cell signaling is more unclear. One potential role involves trophic factor stabilization of beta-catenin. After development, growth factor stimulation can activate the PI3 kinase/Akt survival pathway. Akt is serine/threonine kinase that inactivates GSK-3 beta resulting in stabilized beta-catenin. The role of beta-catenin in cell survival remains unclear. However, it is known that point mutations that change the serine/threonine regulatory residues in beta-catenin increase its stability and lead to some forms of cancer. In addition, the tumor suppressor, APC (adenomatous polyposis coli) and another protein, axin, tether beta-catenin to GSK-3 beta, enhancing GSK-3 beta mediated phosphorylation of beta-catenin and decreasing beta-catenin stability.
Mutations in or Deletion of APC Results in an Increased Stability of Beta-catenin and Tumor Formation
Another role of beta-catenin and perhaps its homologues, delta-catenin and p0071, involves cell to cell adhesion by their association with cadherins. Cadherins are a large family of cell adhesion molecules involved in cell to cell interactions at sites known as desmosomes and adherins junctions. Beta-catenin associates with the cytoplasmic domains of E-cadherin and alpha catenin linking them to the actin cytoskeleton. Tyrosine phosphorylation of beta-catenin during cell migration or neuronal process formation releases beta-catenin from E-cadherin and the cytoskeleton allowing for a change in cell morphology or cell adhesion. This role is distinct from beta-catenin's role in signal transduction, as beta-catenin mutants can be generated to disrupt gene expression but not alter its ability to bind E-cadherin and alpha catenin. It has been suggested that beta-catenin's role in both cell adhesion and signal transduction may be mediated through APC since it competes with E-cadherin at the same site on beta-catenin.
Role of PS1 in Beta-catenin Metabolism and Signaling
Several studies demonstrate that PS1 associates with beta-catenin in vivo. However, the function of this interaction is unknown. Recently, four papers from independent groups have shown that PS1, in addition to binding with beta-catenin, plays a role in regulating beta-catenin metabolism and FAD mutations in PS1 may perturb this regulation. Unfortunately, these papers use a wide range of differing techniques and show conflicting results regarding the effects of PS1 on the "stability" beta-catenin. We intend to review the theoretical, as well as technical differences in these recent reports. Hence, the focus of this discussion is intended to highlight the "putative roles" that PS1 and FAD mutations may have on beta-catenin's function.
Zhang and colleagues demonstrated that when PS1-WT and myc-tagged beta-catenin were co-expressed in HEK293 cells, myc-tagged beta-catenin showed an increase in stability. On the contrary, FAD mutations in PS1, when co-expressed with beta-catenin, failed to enhance beta-catenin stability. Moreover, transgenic mice expressing mutant PS1 showed an increase in beta-catenin immunoreactive degradation products, suggesting a decrease in beta-catenin stability. PS1-deficient fibroblast cultures also had an increase in beta-catenin degradative products. Finally, brain homogenates from FAD patients with PS1 mutations showed a significant reduction in the levels of beta-catenin.
A report by Murayama and colleagues demonstrated a somewhat different set of results. Transient transfection of PS1-WT and mutant PS1 caused a dramatic decrease in the levels of "cytoplasmic" endogenous beta-catenin. In addition, using a sensitive reporter assay for beta-catenin function, Murayama and colleagues further demonstrated that beta-catenin stability was decreased significantly more in PS1 mutant expressing cells. The reporter assay measured the amount of beta-catenin that is able to transactivate a Tcf/LEF promoter upstream of a luciferase reporter gene. This technique assesses beta-catenin function, stability and its entry into the nucleus.
Nishimura and colleagues assayed the effect of PS1 on beta-catenin function by directly immunostaining human FAD patient-derived fibroblasts for nuclear localized beta-catenin. By inhibiting GSK-3 beta (the regulatory kinase of beta-catenin) and thus increasing the stability of beta-catenin, they assayed the amount of beta-catenin that was translocated into the nucleus. FAD patient fibroblasts with PS1 mutations had a significantly reduced amount of nuclear localized beta-catenin. In addition, one cell line with a PS2 mutation showed a similar result. Nishimura and colleagues propose that this phenomenon is not due to a decrease in beta-catenin stability but instead suggest that mutant PS1 affects the trafficking beta-catenin from the cytoplasm to the nucleus. The basis for these observations results from experiments looking at the stability of beta-catenin in these cells following GSK-3 beta inhibition and proteasomal inhibition.
Kang and colleagues show different results in regard to PS1's function upon beta-catenin stability. They demonstrate that PS1-WT expression in HEK293 cells decreases the stability of endogenous beta-catenin, whereas FAD mutant PS1 increases the stability of beta-catenin as compared with untreated controls. Moreover, FAD PS1 transgenic mice demonstrate an increase in beta-catenin levels that correlates with PS1-mutant expression levels. Finally, PS1-deficient fibroblast show an increase in the stability of beta-catenin as compared with normal fibroblasts.
Kang and colleagues lend more insight into the role of PS1 in beta-catenin metabolism by confirming a previous study by Takashima and colleagues who demonstrated that PS1-WT associates with GSK-3 beta in addition to beta-catenin. In contrast to Takashima who demonstrated an increase in GSK-3 beta binding to mutant PS1, Kang and colleagues failed to find an association between GSK-3 beta and mutant PS1.
These reports are in agreement that PS1 associates with beta-catenin, and that PS1-WT and FAD mutants affect the metabolism of beta-catenin. However, it is still unclear from these papers whether, PS1 and FAD PS1 mutants positively or negatively regulate beta-catenin (see Table). One problem is the diverse range of techniques presented in these papers to assess beta-catenin activity. Zhang et al pulse-labeled transfected beta-catenin from PS1 expressing cells and observed beta-catenin degradative products in brain homogenates. Murayama et al measured the levels of cytosolic endogenous beta-catenin and assayed beta-catenin's ability to activate a reporter construct. Nishimura et al calculated the number of beta-catenin positive nuclei to assess beta-catenin functionality. Finally, Kang et al pulse labeled total endogenous beta-catenin in order to arrive at their conclusions. Before any conclusions can be made regarding PS1's function, one must take into to account the varied methods of experimentation.
Beta-catenin stability and/or entry into the nucleus compared to mock
|Zhang, et al||increased||no change||decreased|
|Murayama, et al||decreased||more decreased||n/a|
|Nishimura, et al||no change||decreased||n/a|
|Kang, et al||decreased||increased||more increased|
Questions for the Panel
1) What is your current model of how PS1 and FAD PS1 mutants affect beta-catenin function? How do resolve your model in light of the differences seen by fellow investigators?
- Reply by Eddie Koo: We have no updates to our model as to how PS1 affect beta-catenin beyond what’s proposed in our paper, i.e. PS1 may be a scaffold whereby beta-catenin and GSK may complex to. We also discussed in that article possible explanations to differences in results with the other published articles. The difficulty, as was pointed out in the terrific summary of Chris Weihl, is that the assays were very different. So if we all did the experiments the same way and expressed the same mutations, maybe the differences will disappear. One concern that I have is how reliable are the results obtained from transient transfections. For example, in transients, there is a high expression of full length PS that rarely exists normally. Supposing that the N- and C-terminal fragments are the active molecules, then those are not really increased that much and the full length molecules can give misleading results.
Reply by Bruce Yankner: Our model is based on evidence that a cytoplasmic complex of axin with beta catenin, GSK3 and APC mediates the phosphorylation and degradation of beta catenin. Our results are consistent with the formation of an alternative PS1-beta catenin complex that inhibits catenin phosphorylation thereby increasing catenin stability. This complex would also faciliteate beta catenin translocation to the nucleus, possibly by promoting the interaction of beta catenin with the Lef-1/Tcf family of transcription factors. It is also possible that PS1 complex formation with GSK3 inhibits the ability of this kinase to phosphorylated beta catenin, which would also result in increased catenin stability. Our results suggest that FAD PS1 mutations reduce the ability of PS1 to stabilize beta catenin, increasing catenin degradation and reducing catenin signaling.
The disparate results on the effect of PS1 on beta catenin stability from several groups most likely reflect technical issues inherent in the experiments. Everyone agrees that PS1 complexes with beta catenin; the point of contention is the functional consequence of complex formation. There are several key technical issues that may account for the disparities, which will hopefully be addressed in the upcoming forum. One central issue is the use of SV40-transformed fibroblasts from PS1-null mice. At the Keystone meeting, I presented results from my lab which show that SV40 transformation downregulates PS1, and that this may occur through the induction of p53 by SV40. We found no significant difference in steady state catenin levels or degradation products in SV40-transformed fibroblasts from PS1-KO and wild-type mice. By contrast, non-transformed fibroblast cultures obtained Jie Shen's and Bart DeStrooper's PS1-KO animals showed increased beta catenin degradation. Moreover, several groups have recently informed me that they have been unable to obtain good expression levels of exogenous PS1 in SV40-transformed PS1-KO fibroblasts. Thus, results from these cells, which have been used in some studies, may be problematic.
A key piece of information which has not yet been published is the effect of PS1 on the biological activity of beta catenin in vivo. This information could provide some clarity by bypassing the confounding technical issues in the in vitro experiments. Our group and another group now have reults which suggest that PS1 can potentiate beta catenin signaling in vivo in Drosophila and Xenopus, findings which have not yet been published. These results are consistent with a stabilizing effect of PS1 on beta catenin.
2) Do changes in beta-catenin stability/function contribute to FAD? Saura and colleagues demonstrated that deletion of the PS1 loop region associated with beta-catenin binding is not necessary for the phenotypic changes in Aβ production seen in all FAD mutants (Saura, et al., and personal communication with Drs. Saura and Thinakaran).
- Reply by Eddie Koo: There is as yet no evidence that changes in beta-catenin stability/function contribute to FAD that I’m aware of. There is really only the apoptosis study of Bruce Yanker that relate to possible AD pathogenesis. But his experimental outcome may not be a direct consequence of changes in catenin stability. I am not aware of anyone having any direct evidence linking this pathway to AD pathophysiology. I am discounting the GSK/tau connection for the moment. Whether the catenin stability, catenin translocation, etc. directly influences AD pathology is up in the air. For that matter, there is no data that Notch signaling has any bearing on AD pathophysiology either. Maybe we will hear differently on Friday.
I would explain Gopal’s data this way: the loop may not be the only site where catenin interacts with PS1. One of the difficulty in attributing PS1 to AD pathogenesis is what to do with PS2. If we assume they play a similar role, then PS2 also ought to interact with catenins. Since the loop is so different, perhaps there are other sites of interaction between PS and catenin. Alternatively, PS may interact with other partners that in turn influence catenin. Remember that any direct association by 2-hybrid assay has only been shown with PS1 loop and delta-catenin and p0071. The evidence linking PS1 to catenin in a complex is only by co-IP. Whether there is another binding partner in the complex remains unknown. Peter Hyslop’s paper showed that PS1-catenin complex appears to migrate heavier than the sum of the molecular weights, assuming 1:1 stoichometry and no dimerization. Finally, I would mention a recent paper for which Bruce is a co-author. This deals with the beta-trp F box protein that also modulates beta catenin degradation. So there are certainly other potential binding partners that have not been reported.
Reply by Bruce Yankner: Although several studies suggest that FAD PS1 mutations alter beta catenin stability or nuclear translocation, the case for a role of beta catenin in the pathogenesis of FAD is not yet compelling. The unpublished results of Saura and colleagues suggest that the PS1-catenin interaction may not be involved in the elevation of Aβ42 production by FAD PS1 mutations. However, it has not been established that this is the mechanism by which PS1 mutations cause FAD. Although a substantial body of evidence suggests that Aβ42 is involved in AD pathogenesis, a role for increased neuronal vulnerability to apoptosis must also be considered, as suggested by the work of Ben Wolozin and Mark Mattson and colleagues on PS1 mutations. Our recently published findings suggest that PS1 mutations could increase neuronal vulnerability to apoptosis by impairing beta catenin signaling (Zhang et al., 1998). However, the in vivo relevance of this mechanism remains to be established.
3) Does PS1's role in beta-catenin metabolism explain the presence of hyperphosphorylated tau and neurofibrillary tangle formation in FAD patients? Conflicting reports suggests that mutant PS1 affects the activity of GSK-3 beta (Takashima, et al. 1998; Nishimura, et al. 1999; Irving and Miller, 1997).
- Reply by Eddie Koo: We have not looked at tau phosphorylation. However, I find it hard to believe that the small changes we see with PS1 mutations can have a big influence on GSK activity. The percentage of GSK that is actually bound to PS is very small.
Reply by Bruce Yankner: The role of PS1 in neurofibrillary tangle (NFT) formation in FAD is at present unclear. Our group and others have demonstrated PS1 colocalization with a subset of NFTs in AD. Takashima et al. and Kang et al. report that PS1 complexes with GSK3, and Takashima et al. find that FAD PS1 mutations increase tau phoshorylation. However, other groups have not detected an effect on tau phosphorylation. An important difference between the Takashima report and the others is that Takashima evaluated the effect of PS1 mutations on tau phosphorylation by endogenous kinases, whereas the other groups utilized GSK3 overexpression systems which could potentially swamp out a PS1 effect (Irving and Miller, 1997; Nishimura et al., 1999). None of the groups have yet examined this issue in a neuronal system. I believe that the best approach would be to analyze endogenous tau phosphorylation in PS1 wt and mutant transgenic and knock-in mice.
4) Does PS1's role in beta-catenin metabolism explain the developmental phenotype seen in PS1 knockout embryos? Recent data suggest that the wingless pathway may interact and negatively regulate Notch signaling.
- Reply by Eddie Koo: Our hypothesis is that PS1 role’s in beta catenin metabolism underlies some (unknown) aspect of the developmental phenotype seen in PS1 KO animals. On the other hand, the evidence is certainly against a major role for catenin at this time. This is because the notch results are quite impressive. Having said that, Chris is absolutely correct to point out that there is a lot of cross-talk between the notch and catenin/armadillo pathways, or signaling network. So in a roundabout way, disturbances in catenin function may ultimately influence notch function.
Reply by Bruce Yankner: The developmental phenotype in PS1-KO mice has been largely attributed to impaired Notch signaling, but it is unclear whether this is the only contributory signaling pathway. The impaired development of the paraxial mesoderm in PS1-KO mice is consistent with loss of Notch signaling. However, loss of Wnt signaling can also affect paraxial mesoderm development (Yoshikawa et al., 1997, Dev. Biol. 183:234). Morevoer, the reduced number of neural progenitors in PS1-KO mice reported by Shen and Tonegawa could also be consistent with impaired Wnt signaling (Ikaya et al., 1997, Nature 389:966). I do not believe that the developmental phenotype of PS1-KO mice will be entirely due to Notch. For example, Paul Saftig has recently described a cortical migration defect in PS1-KO mice which is not a known Notch-related phenotype.
5) Expand on the significance of PS1 interactions with other armadillo repeat proteins (e.g. delta-catenin and p0071). Will these interactions be more important to FAD pathogenesis since they are expressed neuronally instead of ubiquitously like beta-catenin?
- Reply by Eddie Koo: So little is known about the other catenins that it is hard to even guess where the other catenins come in. These other catenins may appear to interact with PS only in our artefactual systems because of the presence of the armadillo repeats. On the other hand, if the other catenins were to play a role, then it is likely not through signaling because they have rather low signaling activity, at least through the known pathways. In which case, what’s left are the cadherin interactions.
Reply by Bruce Yankner: There is not enough information about PS1 interactions with other armadillo proteins in the context of FAD to address this question. The point that beta catenin is ubiquitously expressed whereas delta catenin is neuron-specific, while interesting, does not address the role of these proteins in AD. APP and presenilins are ubiquitously expressed, yet these proteins can still give rise to brain-specific pathology in AD.
Barth, A. I., Nathke, I. S., and Nelson, W. J. Cadherins, catenins and APC protein: interplay between cytoskeletal complexes and signaling pathways. Curr Opin Cell Biol. 1997 Oct;9(5):683-90. Review. Abstract.
Dale, T. C. Signal transduction by the WNT family of ligands. Biochem J. 1998 Jan 15;329 ( Pt 2):209-23. Abstract.
Irving, N. G., and Miller, C. C. J. Tau phosphorylation in cells transfected with wild-type or an Alzheimer's disease mutant Presenilin-1. Neurosci Lett. 1997 Jan 31;222(2):71-4. Abstract.
Kang, D. E., Soriano, S., Frosch, M. P., Collins, T., Naruse, S., Sisodia, S. S., Leibowitz, G., Levine, F., and Koo, E. H. Presenilin-1 facilitates the constitutive turnover of beta-catenin: differential activity of alzheimer's disease-linked PS1 mutants in the beta-catenin signaling pathway. J Neurosci. 1999 Jun 1;19(11):4229-37. Abstract.
Levesque, G., Yu, G., Nishimura, M., Zhang, D. M., Levesque, L., Yu, H., Xu, D., Liang, Y., Rogaeva, E., Ikeda, M., Duthie, M., Murgolo, N., Wang, L., VanderVere, P., Bayne, M. L., Strader, C. D., Rommens, J. M., Fraser, P. E., and St. George-Hyslop, P. Presenilins interact with armadillo proteins including neural-specific plakophilin-related protein and beta-catenin. J Neurochem. 1999 Mar;72(3):999-1008. Abstract.
Murayama, M., Tanaka, S., Palacino, J., Murayama, O., Honda, T., Sun, X., Yasutake, K., Nihonmatsu, N., Wolozin, B., and Takashima, A. Direct association of presenilin-1 with beta-catenin. FEBS Lett. 1998 Aug 14;433(1-2):73-7. Abstract.
Nishimura, M., Yu, G., Levesque, G., Zhang, D. M., Ruel, L., Chen, F., Milman, P., Holmes, E., Liang, Y., Kawarai, T., Jo, E., Supala, A., Rogaeva, E., Xu, D. M., Janus, C., Levesque, L., Bi, Q., Duthie, M., Rozmahel, R., Mattila, K., Lannfelt, L., Westaway, D., Mount, H. T., Woodgett, J., Fraser, P. E., and St George-Hyslop, P. Presenilin mutations associated with Alzheimer disease cause defective intracellular trafficking of beta-catenin, a component of the presenilin protein complex. Nat Med. 1999 Feb;5(2):164-9. Abstract.
Price, D. L., and Sisodia, S. S. Mutant genes in familial Alzheimer's disease and transgenic models. Annu Rev Neurosci 21, 479-505. Abstract not available.
Saura, C. A., Tomita, T., Davenport, F., Harris, C. L., Iwatsubo, T., and Thinakaran, G. Evidence that intramolecular associations between presenilin domains are obligatory for endoproteolytic processing. J Biol Chem. 1999 May 14;274(20):13818-23. Abstract.
Stahl, B., Diehlmann, A., and Sudof, T. C. Direct interaction of Alzheimer's disease-related presenilin-1 with armadillo protein p0071. J Biol Chem. 1999 Apr 2;274(14):9141-8. Abstract.
Takashima, A., Murayama, M., Murayama, O., Kohno, T., Honda, T., Yasutake, K., Nihonmatsu, N., Mercken, M., Yamaguchi, H., Sugihara, S., and Wolozin, B. Presenilin 1 associates with glycogen synthase kinase-3beta and its substrate tau. Proc Natl Acad Sci U S A. 1998 Aug 4;95(16):9637-41. Abstract.
Tesco, G., Kim, T. W., Diehlmann, A., Beyreuther, K., and Tanzi, R. E. Abrogation of the presenilin 1/beta-catenin interaction and preservation of the heterodimeric presenilin 1 complex following caspase activation. J Biol Chem. 1998 Dec 18;273(51):33909-14. Abstract.
Willert, K., and Nusse, R. Beta-catenin: a key mediator of wnt signaling. Curr Opin Genet Dev. 1998 Feb;8(1):95-102. Review. Abstract.
Zhang, Z., Hartmann, H., Do, V. M., Abramowski, D., Sturchler-Pierrat, C., Staufenbiel, M., Sommer, B., van deWetering, M., Clevers, H., Saftig, P., De Strooper, B., He, X., and Yankner, B. A.. Destabilization of beta-catenin by mutations in presenilin-1 potentiates neuronal apoptosis. Nature. 1998 Oct 15;395(6703):698-702. Abstract.
Zhou, J., Liyanage, U., Medina, M., Ho, C., Simmons, A. D., Lovett, M., and Kosik, K. S. (1997). Presenilin 1 interaction in the brain with a novel member of the Armadillo family. Neuroreport, 1997 Apr 14;8(6):1489-94. Abstract.
No Available Further Reading