Research on presenilins presented at this year's meeting centered around two major questions: 1) What is the role of presenilins in Aβ production and APP processing, and 2) what are the other biological actions of presenilins? A major debate in the presenilin field revolves around the suggestion by Wolfe, Selkoe and colleagues that PS1 is a γ-secretase. Different presentations addressed this issue by investigating three different subjects: inhibitors of APP processing, the effect of mutations on presenilin function and the distribution of presenilins.

Inhibitors of APP Processing: Wolfe and colleagues (122.3) have designed a series of peptides that mimic the site on APP cleaved by γ-secretase. These peptides contain difluoro ketone moieties embedded in the peptide which are known to inactivate aspartyl proteases. Wolfe tested these peptides against the cysteine protease calpain, and observed no effect on calpain activity. Each of these APP-mimetic peptides inhibited Aβ production. Based on this data, Wolfe and colleagues suggested that the γ-secretase is an aspartyl protease. The data are quite appealing but the conclusion contradicts a paper by Robakis' group [Figueiredo-Pereira, 1999 #1362] showing that three other cysteine protease inhibitors, Z-VF-CHO, Z-VL-CHO and E-64d all inhibit Aβ40 production, although they were less effective on Aβ42 production. An essential question is the specificity of the different inhibitors used by Wolfe's and Robakis' groups. Each group of peptides ostensibly acts specifically on the classic aspartyl of cysteine proteases, but if PS1 is a protease, it is clearly not a classic protease. The activities of these inhibitors against proteases that don't fit the classic structural definitions is unknown.

Murphy, Golde, and colleagues (219.10 and 219.11) addressed this issue by examining the action of peptide inhibitors on mutant APP constructs. They have observed that the aspartyl protease inhibitor, pepstatin, inhibits Aβ40 and 42 production in cells expressing wild-type APP. However, when they introduce mutations into APP that shift the predominant cleavage of APP from Aβ40 to other sites, such as Aβ43, they observed that pepstatin does not inhibit Aβ43 production from the mutant APP constructs. The fact that pepstatin does inhibit γ-secretase with one substrate (wild-type APP) but does not inhibit it with another substrate (mutant APP) suggests that there are some proteases that can cleave Aβ that are not classic aspartyl proteases. However, this approach does not rule out the possibility that the mutant APP outcompetes the pepstatin for the enzyme or that the mutations cause the APP constructs to be processed by a different set of proteases than normally carry out γ-secretase activity.

Mutations affecting PS function: An important part of the argument that PS1 is γ-secretase lies in the discovery that mutation of the aspartates at positions 257 and 385 (to alanine) reduces the production of Aβ. Many groups replicated this finding. Haass and colleagues have now extended this work to PS2 (219.5) by showing that mutation of the homologous aspartates in PS2, at positions 257 and 366, also reduce Aβ production. Selkoe and colleagues also examined the PS2 mutations (219.6). Previously, they had observed that expressing the PS1 D385A construct reduced Aβ expression by about 65%, but did not block Aβ production. Now Selkoe and his team have shown that cells expressing both PS1 D385A and PS2 D366A constructs produced no Aβ. Their interpretation of this data is that PS1 is either a direct regulatory cofactor for γ-secretase or the actual γ-secretase. Most of this work has been performed using autosomal cells, such Cos7 cells. Kim, Sisodia and colleagues evaluated the function of the D257A mutation in N2A cells and in primary neurons from PS1 -/- mice (291.8). They observed no loss of Aβ production in the N2A cells, and a small amount of rescue Aβ production in the primary PS1 -/- neurons (~30%). However, the expression level of the mutant construct in the N2A cells was not sufficient to completely replace wild-type PS1 function, which leaves open the possibility that the remaining Aβ activity resulted from residual wild-type PS1 activity. In contrast to Kim and Sisodia's findings, we observe that the D257A and D385A mutations inhibit APP processing, as judged by impaired stimulation of APPs secretion by protein kinase C and elevated levels of APP C-terminal fragments, although we have not yet examined Aβ production (29.10).

It appears that the D257A and D385A mutations in PS1 are not the only mutations capable of blocking PS1 activity. Golde and colleagues (219.11) produced two novel dominant negative PS1 constructs. One construct has a 3 amino acid displacement of the aspartate at position 385 (achieved by inserting three amino acids ahead of the aspartate), while the other PS1 construct lacks transmembrane domains 1 and 2 (but maintained the N-terminal signal region). Expression of either construct inhibited Aβ production and inhibited the ability of protein kinase C to stimulate secretion of APPs. Haass and colleagues (219.5) showed that a naturally occurring PS2 variant that lacks exon 8 inhibits NOTCH processing but does not inhibit Aβ production, although it does increase the levels of APP C-terminal fragments in a manner similar to the PS2 D366A dominant negative construct. This indicates that there are multiple different mutations in PS1 that can lead to losses of function.

Signal transduction: One of the most exciting findings in the field of presenilin research is discovery of the importance of presenilins in cell biology, which is exemplified by the lethal phenotype of the PS1 knockout. The discovery that PS1 is required for cleavage of Notch has captivated the field. Several groups in the meeting replicated this finding. For instance, De Strooper and colleagues (219.1), Ray Goate and colleagues (219.2) and Song, Yankner and colleagues (219.3) showed that expression of the D257A PS1 construct inhibits Notch cleavage in a dominant negative manner and is unable to rescue Notch cleavage in fibroblasts from PS1 knockout fibroblasts. Over-expression of PS2, however, could rescue Notch cleavage in the PS1 knockout cells, showing that PS2 can complement PS1 function.

In contrast to the work on Notch, which appears to be very consistent among laboratories, studies of PS1/beta-catenin interactions appear to be more variable. It appears that interaction of PS1 with β-catenin changes with the state of the cell and, possibly, the type of cell. This issue is exemplified by studies investigating the cellular localization of PS1. De Strooper and colleagues (219.1) showed that most of the PS1 localizes to the golgi/endoplasmic reticulum in dispersed cells. Ray Goate and colleagues (219.2) showed that surface biotinylation of cells did lead to detection of PS1 at the cell surface. The distribution is important because the PS1/ γ-secretase hypothesis requires that PS1 co-localize with all sites of γ-secretase activity. Because Notch is cleaved at the cell surface, the identification of PS1 at the plasma membrane fulfills an important prediction of this hypothesis.

Work by Marambaud, Efthimiopoulos, Georgakopoulos, Robakis, and colleagues (523.3 and 523.4) presented one of the most striking observations about PS1 function. They observed that PS1 localized to the surface of cells at points of contact with other cells. They immunostained cells with anti-PS1 antibodies. In dispersed cells they observed that PS1 was localized mainly in the Golgi and endoplasmic reticulum, however in confluent cells, anti-PS1 antibodies produced bright staining at the plasma membrane at sites of cell to cell contact. Immunocytochemistry with E-cadherin and β-catenin produced a similar pattern in confluent cells, and Robakis' team followed up this work by showing that PS1 co-immunoprecipitated with E-cadherin, α- and β-catenin, which have all previously been implicated in regulating cell to cell adhesion. By quantitating cell adhesion, they were also able to demonstrate a functional phenotype. Cells overexpressing PS1 exhibited increased cell adhesion while the D257A mutant PS1 constructs inhibited cell adhesion.

This work validates a prior observation by Dewji and Singer showing that PS1 can localize to the plasma membrane and regulates cell to cell adhesion. They also found examples in vivo in which PS1 localized to points of cell to cell contact, including synapses in the brain and corneal epithelial cells. Robakis speculates that PS1 might exist as a large complex that includes Notch in addition to E-cadherin and α-/β-catenin. Although extremely interesting, the localization of PS1 to sites of cell to cell contact does not directly agree with prior in vivo studies of PS1, which show that most PS1 primarily localized in the endoplasmic reticulum/Golgi compartment-despite the fact that most of the cells in our body are in contact with other cells.

Other studies showed that PS1 can regulate β-catenin signaling. Koo and colleagues report that PS1 increases beta-catenin signaling in PC12 cells (309.3). In contrast, we observe that both wild-type and the D257A and D385A PS1 mutants inhibit β-catenin signaling. Expression of these mutant constructs in N2A cells inhibited β-catenin signaling using a hTcf4 luciferase reporter or by measuring nuclear translocation of β-catenin (29.10). The difference in results between the two laboratories might lie either in the methods or in the cell types. Koo uses PS1 expressed transiently from an ecdysone-inducible promoter in PC12 cells, while we use NSA cells stably transfected with the PS1 constructs.

Several reports were also presented examining the interaction of presenilins with other proteins. One interesting protein being investigated is calsenilin. Choi, Buxbaum, Wasco and colleagues (523.2) showed that calsenilin binds to the C-terminal fragment of PS2 that is generated by caspases during apoptosis but not to the C-terminal fragment generated by the basal processing of PS2. Leissring, LaFerla and colleagues (641.7) showed that overexpression of PS1 increases calcium fluxes into the endoplasmic reticulum and that calsenilin inhibits the PS1-dependent calcium fluxes. Tezapsidis and colleagues (219.9) identified another protein that binds to PS1, CLIP-170, which is a protein that links membranous organelles to microtubules or intermediate filaments. As with calsenilin, the interaction appears connected with calcium metabolism because binding to PS1 is calcium-dependent. PS1 and PS2 have also been shown to regulate proteins involved in apoptosis. Alberici, Binetti, and colleagues showed that PS1 binds to Bcl-2 during apoptosis (641.4) and Sato, Takashima and colleagues (641.6) showed that PS1 binds to JNK, an important kinase in apoptosis and the stress response.

In Vivo Studies: Clarification of the functions of PS1 might come from in vivo studies. Shen and colleagues (426.1) showed that brains from mice lacking PS1 show a reduced neural progenitor cell population, region-specific neuronal loss and disrupted cerebral laminar organization. They focused on the stem cell defects and observed that the reductions in neural progenitor cells was due to premature differentiation of the neural progenitor cells. In situ studies implicated Notch in this process because the brains from PS1 knockout mice showed decreased expression of Hes5, a downstream effector of Notch, and increased expression of the Notch1 ligand DII1. Notch1 transcript and protein were unchanged. These results suggest that loss of PS1 interferes with Notch signaling. One of the challenges in studying PS1 function in vivo is the lethal phenotype of the PS1 knockout. Zheng, Qian, and colleagues (426.2) approached this issue by rescuing PS1 knockout mice with PS1 constructs driven by a Thy1 promoter which restricts PS1 expression to the brain in adults. Interestingly, these animals develop epithelial hyperproliferation and tumors consistent with increased activation of β-catenin. This suggests that PS1 might normally inhibit activation of cell growth by bet-catenin.

Summary: After much confusion in the presenilin field, a clearer picture of PS1 and PS2 is emerging. PS1 is a transmembrane protein that binds to a group of molecules known to regulate cell contact. These contact-related molecules include Notch, GSK-3, α-, β- and γ-catenin and E-cadheren. PS1 might bind to these molecules as part of one large complex that mediates cell to cell contact or signaling. Part of the function of PS1 in this complex might be as a scaffold, holding these proteins together, while another part of the function is to facilitate cleavage of Notch in response to engagement of Notch by ligands. PS1 might either bind the critical protease or be the protease.

PS1 exists both in the endoplasmic reticulum/golgi and at the plasma membrane. Its function might depend on its localization. In the endoplasmic reticulum PS1 might function in regulating calcium metabolism in a process that is regulated by calsenilin. PS2 appears to have functions that overlap PS1.

The cell appears to have borrowed PS1 function and used it to regulate APP processing, possibly because APP also plays a role in regulating cell to cell contact. It is clear that PS1 is required for APP and Notch proteolysis, PS1 co-localizes to sites of APP proteolysis and that inhibiting PS1 function inhibits APP and Notch proteolysis. This suggests that PS1 is either a γ-secretase or is a scaffold protein required for γ-secretase activity. A definitive answer to this question can really only be shown by demonstrating that recombinant PS1 can catalyze APP cleavage in vitro or by identifying a protease that binds to PS1 carries out the cleavage of APP. Because several groups are actively prusuing these questions, it seems quite possible that this question will be answered within the year.—Benjamin Wolozin

Presentations Cited
29.10: Presenilin 1 and presenilin 2 are necessary for the regulation of amyloid precursor protein secretion. - J.J. PALACINO*, B.E. BERECHID, C. ECKMAN, S. YOUNKIN, J.S. NYE AND B. WOLOZIN Loyla Univ. Chicago.

122.3: Peptidomimetic probes suggest a large S1 pocket for Alzheimer's γ-secretase. - M.S. WOLFE*, C.L. MOORE, T.S., DIEHL, T. RAHMATI, W. XIA AND D.J. SELKOE Univ. of Tennessee, Memphis, Brigham and Women's Hosp. and Harvard Med. Sch.

219.1: Notch and APP proteolytic processing controlled by presenilin 1 and 2. - B. DE STROOPER*, W. ANNAERT, P. CUPERS, D. HARTMANN, P. SAFTIG AND R. KOPAN K.U. Leuven, Belgium, Christian Albrechts Univ. Kiel and Molec. Cell Biol., Göttingen, Germany and Washington Univ.

219.2: The presenilin-1-Notch1 complex forms in the secretory pathway and is transported to the cell surface. - W.J. RAY, M. YAO, J. MUMM, E. SCHROETER, M. WOLFE, D. SELKOE, P. SAFTIG, B. DE STROOPER, R. KOPAN AND A. GOATE* Washington Univ. Med. Sch., Harvard Med. Sch., Univ. of Gauottingen, Germany and KU Leuven, Belgium.

219.3: Proteolytic release and nuclear translocation of Notch-1 are induced by presenilin-1 and impaired by pathogenic presenilin-1 mutations. - W. SONG*, P. NADEAU, M. YUAN, X. YANG, J. SHEN AND B.A. YANKNER Children's Hosp. and Brigham and Women's Hosp., Boston.

219.4: Differential effects of the presenilin-1Δexon 8 splice variant on amyloidogenesis and notch signalling. - A. CAPELL*, H. STEINER, S. KECK, H. ROMIG, B. PESOLD AND C. HAASS Ludwig Maximillians Univ., Munich and Boehringer Ingelheim KG, Germany.

219.5: A functional role of presenilin-2 in amyloidogenesis and notch signaling. - C. HAASS*, A. CAPELL, J. WALTER, H. ROMIG, M. CITRON, K. FECHTELER, U. LEIMER, J. HARDY, R. BAUMEISTER, K. DUFF AND H. STEINER Univ. of Munich, Boehringer Ingelheim, Germany, Amgen Inc., Mayo Clin. Jacksonville and Nathan Kline Inst., Orangeburg, NY.

219.6: Evidence that presenilins 1 and 2 mediate intramembranous cleavage of amyloid precursor protein rather than acting via a protein trafficking role. - D.J. SELKOE*, W.T. KIMBERLY, B.L. OSTASZEWSKI, T. RAHMATI AND W. XIA Brigham and Women's Hosp., Harvard Med. Sch.

219.8: Evaluation of a role of presenilin-1 in the proteolytic processing of amyloid precursor protein. - S.H. KIM*, Y. ZHANG, H.H. SLUNT AND S.S. SISODIA Univ. of Chicago and Johns Hopkins Univ.

219.10: FAD-linked presenilin: effects on pepstatin sensitive and insensitive γ-secretase activities. - M.P. MURPHY*, R. WANG, S.N. ULJON, T.E. SMITH, H.A. LOOKINGBILL, K. FINDLAY AND T.E. GOLDE Mayo Clin. Jacksonville and Rockefeller Univ.

219.11: Presenilins function as regulators of multiple γ-secretase activities. - T.E. GOLDE*, R. WANG, S. ULJON, T.E. SMITH, H.A. LOOKINGBILL, K. FINDLAY AND M.P. MURPHY Mayo Clin. Jacksonville and Rockefeller Univ.

309.3: The Neurobiology of Presenilins and Associated Signaling Molecules in Development and Disease. - E. H. KOO

426.1: Studies of presenilin-1 function in neurogenesis and the adult brain. - J. SHEN*, H. YU, M. HANDLER, X. YANG AND J. KESSLER Brigham and Women's Hosp. and Harvard Med. Sch.

426.2: Reduced expression of presenilin 1 leads to epithelial hyper-proliferation in transgenic mice. - H. ZHENG*, Y. WU, A. FLETCHER AND S. QIAN Baylor Col. of Med. and Merck Res. Labs., Rahway, NJ.

523.2: Calsenilin preferentially interacts with the caspase-derived presenilin 2 C-terminal fragment and cleaved by caspase during apoptosis. - E.K. CHOI*, N. ZAIDI, A.C. CROWLEY, J. MILLER, C. LILLIHOOK, J.D. BUXBAUM AND W. WASCO Harvard Med. Sch., Charlestown and Mount Sinai Sch. of Med.

523.3: Presenilin1 forms Ca++-dependent complexes with the E-cadherin/catenin cell adhesion system. - P. MARAMBAUD, J. SHIOI, A. GEORGAKOPOULOS, S. EFTHIMIOPOULOS AND N.K. ROBAKIS* Mount Sinai Sch. of Med.

523.4: Presenilin 1 forms complexes with cell surface E-cadherin and regulates cell-cell aggregation. - S. EFTHIMIOPOULOS*, W. CUI, A. GEORGAKOPOULOS, J. SHIOI, P. MARAMBAUD AND N.K. ROBAKIS Mount Sinai Sch. of Med.

641.4: Modulation of presenilin-1 and Bcl-2 interaction during apoptosis. - A. ALBERICI, L. BENUSSI, D. MORATTO, L. GASPARINI, R. GHIDONI, J.H. GROWDON, R.M. NITSCH AND G. BINETTI* IRCCS Ctr. San Giovanni di Dio, FBF, Brescia, Italy,

641.5: Presenilin 1 and nerve growth factor modulation in cultured hippocampal neurons. - S. DI LORETO, R. MACCARONE, A. CRESTINI, L. MALVEZZI CAMPEGGI AND A.M. CONFALONI* CNR, L'Aquila, Italy and Inst. Superiore di Sanità, Rome.

641.6: Regulation of JNK signal transduction by presenilin-1. - S. SATO, X. SUN, O. MURAYAMA, K. KAMINO*, Y. SAKAKI AND A. TAKASHIMA Brain Sci. Inst., RIKEN, Wako and Nippon Med. Sch., Japan.

641.7: Presenilin-2 modulates intracellar Ca2+ release and clearance.-M.A. LEISSRING*, I. PARKER AND F.M. LAFERLA Univ. of California, Irvine.


  1. Wolozin states above that our presentations validate earlier results by Dewji and Singer, which have been criticized on the basis of both antibody specificity and suggested PS1 topology (see for example discussion by Li and Greenwald). Since this has been a hotly debated and confusing point, I received several inquires about it since Ben's commentary appeared in Alzforum. I would like to state that we have no data to support the Dewji and Singer conclusion that the PS1 amino terminus and large loop regions are extracellular or extracytoplasmic.—N.K. Robakis.

    I read Ben Wolozin's commentary Presenilins: An Overview, in Alzforum. Under Signal Transduction (third paragraph), he writes about our work showing PS1 localized at cell-cell contact: "Although extremely interesting, the localization of PS1 to sites of cell to cell contact does not directly agree with in vivo studies of PS1, which show that it is primarily localized in the endoplasmic reticulum/golgi compartment- despite the fact that most of the cells in our body are in contact with other cells". Ben apparently missed the presentation by Georgakopoulos et al from my laboratory where we clearly showed that PS1 is localized at cell-cell contact sites of mouse cornea epithelial tissue. More important, we showed that in mouse brain PS1 concentrates at synaptic junctions. The detection of PS1 at cell-cell contact/communication sites both in vivo and in vitro shows that there is no Space Paradox (Annaert and De Strooper et al., 1999), as PS1 probably processes Notch1 at cell-cell contact sides. This work is in Press at Mol. Cell.


    . Presenilins: molecular switches between proteolysis and signal transduction. Trends Neurosci. 1999 Oct;22(10):439-43. PubMed.

Make a Comment

To make a comment you must login or register.


No Available References

Further Reading

No Available Further Reading