Cellular Mechanisms of Alzheimer's Disease: Neurodegenerative Diseases 2006

Diseases Cellular Mechanisms of Alzheimer’s Disease Editors Christian Haass, München Gerd Multhaup, Berlin 51 figur...

Author: C. Haass | G. Multhaup

94 downloads 1021 Views 5MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Diseases

Cellular Mechanisms of Alzheimer’s Disease

Editors

Christian Haass, München Gerd Multhaup, Berlin

51 figures, 13 in color, and 3 tables, 2006

Basel • Freiburg • Paris • London • New York • Bangalore • Bangkok • Singapore • Tokyo • Sydney

S. Karger Medical and Scientific Publishers Basel • Freiburg • Paris • London New York • Bangalore • Bangkok Singapore • Tokyo • Sydney

Disclaimer The statements, options and data contained in this publication are solely those of the individual authors and contributors and not of the publisher and the editor(s). The appearance of advertisements in the journal is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements. Drug Dosage The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any change in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.

Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

All rights reserved. No part of this publication may be translated into other languages, reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, microcopying, or by any information storage and retrieval system, without permission in writing from the publisher or, in the case of photocopying, direct payment of a specified fee to the Copyright Clearance Center (see ‘General Information’). © Copyright 2006 by S. Karger AG, P.O. Box, CH–4009 Basel (Switzerland) Printed in Switzerland on acid-free paper by Reinhardt Druck, Basel ISBN-10: 3–8055–8183–1 ISBN-13: 978–3–8055–8183–7

Vol. 3, No. 4–5, 2006

Diseases

Contents

191 Preface St. George-Hyslop, P. (Toronto)

247 Control of Amyloid--Peptide Generation by

Subcellular Traﬃcking of the -Amyloid Precursor Protein and -Secretase

Walter, J. (Bonn)

Introduction 193 Six Years of Collaborative Alzheimer’s Disease

255 -Secretase Activation – An Approach to Alzheimer’s

Research in Germany

Disease Therapy

Haass, C. (Munich); Multhaup, G. (Berlin)

Fahrenholz, F.; Postina, R. (Mainz) 262 Ectodomain Shedding of the Amyloid Precursor

Original Papers 197 Spectroscopic Approaches to the Conformation of Tau

Protein in Solution and in Paired Helical Filaments von Bergen, M.; Barghorn, S.; Jeganathan, S.; Mandelkow, E.-M.; Mandelkow, E. (Hamburg) 207 Signaling from MARK to Tau: Regulation, Cytoskeletal

Crosstalk, and Pathological Phosphorylation Timm, T.; Matenia, D.; Li, X.-Y.; Griesshaber, B.; Mandelkow, E.-M. (Hamburg) 218 Subcellular Traﬃcking of the Amyloid Precursor

Protein Gene Family and Its Pathogenic Role in Alzheimer’s Disease Kins, S.; Lauther, N.; Szodorai, A.; Beyreuther, K. (Heidelberg) 227 Presenilin Function in Caenorhabditis elegans Smialowska, A.; Baumeister, R. (Freiburg) 233 Functional Role of the Low-Density Lipoprotein

Receptor-Related Protein in Alzheimer’s Disease Waldron, E.; Jaeger, S.; Pietrzik, C.U. (Mainz) 239 The Functions of Mammalian Amyloid Precursor

Protein and Related Amyloid Precursor-Like Proteins Anliker, B. (Frankfurt); Müller, U. (Frankfurt/Heidelberg)

Protein: Cellular Control Mechanisms and Novel Modiﬁers Lichtenthaler, S.F. (Munich) 270 Amyloid Precursor Protein and BACE Function as

Oligomers Multhaup, G. (Berlin) 275 Assembly, Traﬃcking and Function of -Secretase Kaether, C.; Haass, C.; Steiner, H. (Munich) 284 Cellular Functions of -Secretase-Related Proteins Haffner, C.; Haass, C. (Munich) 290 Modulators and Inhibitors of - and -Secretases Schmidt, B.; Baumann, S.; Narlawar, R.; Braun, H.A.; Larbig, G. (Darmstadt) 298 -Secretase Modulation with A42-Lowering

Nonsteroidal Anti-Inﬂammatory Drugs and Derived Compounds Czirr, E.; Weggen, S. (Mainz) 305 Role of Amyloid Precursor Protein, Amyloid- and

-Secretase in Cholesterol Maintenance Hartmann, T. (Heidelberg)

312 Author Index 312 Subject Index

© 2006 S. Karger AG, Basel Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Access to full text and tables of contents, including tentative ones for forthcoming issues: www.karger.com/ndd_issues

Diseases

Neurodegenerative Dis 2006;3:191–192 DOI: 10.1159/000095255

Preface

When members of the International External Peer Review Committee were first asked to sit on the review panel for the Deutsche Forschungsgemeinschaft (DFG) collaborative priority program project, we all assumed that the outcome would be much like any other major, focused research initiative sponsored by national research councils (which in North America are typically referred to ‘RFA’s). The typical product of such national focused research programs is usually a collection of independent, unrelated research applications which range in quality from simply excellent to mediocre, with the good and excellent proposals being the ones then chosen for funding. Interaction between these excellent, funded applications is often hoped for, but rarely actually happens. Our expectation that this would be the likely outcome also for the DFG program was immediately shattered at the first meeting of the review panel. With each subsequent review panel meeting and at the annual meeting of the successful applicants in Eibsee, it has become increasingly clearer that this program is a massive exception to the usual models. The program from its inception has been stellar in both its intellectual capacity and in its productivity. It is truly an example of the merits and the powerful utility of a national strategic research designed initiative to bring together experts in a given field. What has emerged is a series of laboratories, each of which individually is simply outstanding, not just on a national scale, but on an international scale. More importantly, there are multiple cross-cutting interactions within this group. There has been massive exchange of interests, expertise, reagents and, helpful advice. It has been a fertile ground for training students and fellows.

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0191$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

The net result is that the German Alzheimer Disease Program, on a per capita basis, is by far and away the most productive in the world in terms of papers per investigator. It is superseded on a total numerical basis only by the US program, which of course, has far greater numbers of investigators and far greater funding. However, more important than the simple metric of numbers of papers per investigator, there is a more important measure, which is the measure of impact. This group has provided profound novel understandings on basic questions of the biology of Alzheimer’s disease as well as normal physiology. This group has made incredibly important observations, for instance on the biology of APP, tau, BACE, presenilins, APOE, on medicinal chemistry and on more general questions of biology such as the process of regulated intramembranous proteolysis and the functional role of homologous proteins related to nicastrin and the presenilins. They have set up experimental model systems in

Peter St. George-Hyslop, MD, DSc University of Toronto Tanz Neuroscience Bldg., Rm 118, 6 Queen’s Park Crescent Toronto, ON M5S 1A8 (Canada) Tel. +1 416 978 7461, Fax +1 416 978 1878, E-Mail [email protected]

cells, mice, worms, flies and zebrafish which have been examined with a host of modern, powerful technologies. This infrastructure, know-how, and expertise will be an invaluable resource for future work. A fundamental question still needs to be asked: ‘Why has this group been so outstandingly successful’? The answer is complex and likely to be slightly embarrassing to the self-deprecating nature of the scientists in the German AD program. It is clear that the leaders of this program are themselves simple outstanding scientists who lead by example rather than by coercion and direction. Second, it is clear that the body of scientists within the program are themselves superb, with a genuine and abiding interest in science. They have that important sense that science is fun, difficult, often frustrating, but that in the end, true and correct observations will stand on their own, and do not therefore need to be overly strongly proselytized. It is apparent that members of this group genuinely like and respect each other, and can clearly see that the greater goal is an advancement in the understanding of nature, which is best achieved through the concerted application of many different skills and in-

192

Neurodegenerative Dis 2006;3:191–192

terests. Extraordinarily, this group appears to be able to address this greater goal, and to set aside the more tedious personal agendas common in many other programs. Finally, and of equal measure in terms of importance, the strong financial support provided by the DFG has brought these elements together. This generosity is a sine qua non for successful science. No matter how clever, no matter how willing to work together, science and scientists cannot produce new knowledge without the ability to purchase the necessary equipment, personnel and reagents. In summary, this program has been incredibly successful. It is certain that its success will continue in time beyond the completion of the original funding period. It is already clear that its success transcends beyond the geographic bounds of the program project. There are numerous collaborations between scientists in this program with other scientists in Germany, elsewhere in Europe and elsewhere in the world. My colleagues on the review committee, our colleagues in the field of neurodegenerative research, and patients with this disease commend the DFG and all of the scientific members of this highly successful program.

Preface

Introduction

Diseases

Neurodegenerative Dis 2006;3:193–196 DOI: 10.1159/000095256

Six Years of Collaborative Alzheimer’s Disease Research in Germany Christian Haass a Gerd Multhaup b a

Laboratory for Alzheimer’s and Parkinson’s Disease Research, Department of Biochemistry, Adolf Butenandt Institute, Ludwig Maximilians University, Munich, and b Institut für Chemie/Biochemie, Freie Universität Berlin, Berlin, Germany

This special issue of Neurodegenerative Diseases is devoted to a 6-year program of the Deutsche Forschungsgemeinschaft (DFG), which supported up to 13 projects on cellular mechanisms of Alzheimer’s disease (AD) with 7.5 million euros. Since the maximum funding by the DFG for such initiatives is 6 years, the program came to an end in June 2006. The members of this program have used this occasion to present their projects, their major findings and the collaborative network in the form of summarizing articles. At the International Conference on Alzheimer’s Disease in Amsterdam in 1998, Volker Herzog (University of Bonn) and C. Haass discussed for the first time the idea to initiate a nationwide priority program on AD research with a special focus on cellular mechanisms leading to amyloid- (A) peptide production and tau aggregation. We both felt that this was about time, since a collaborative research program on this hot topic did not exist in Germany. For us this was quite surprising, since we were used for many years to run collaborative projects with scientists throughout Europe and the United States. Moreover, we had the strong feeling that by this time Germany’s AD researchers were split into two opposing groups, the ‘Baptists’ and the ‘Tauists’. This dogmatic separation of the field prevented coordinated and collaborative research for decades in Germany. Moreover, we also found it surprising that in Germany only a small

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0193$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

number of research teams existed. This was even more surprising, knowing that Germany looks back to a fantastic tradition of excellent AD research. Needless to say, that it all started in Germany, when Alois Alzheimer described for the first time the brain pathology of a heavily demented patient exactly 100 years ago (November 1906). Furthermore, Germany is also the place where the -amyloid precursor protein (APP) gene was cloned. This major breakthrough was possible due to the pioneering work of Konrad Beyreuther, who is now regarded as the father of modern AD research. A number of the members of this network are his ‘children’ or ‘grandchildren’. The idea behind our initiative was to bring together molecular biologists working in AD research from complementary fields and opinions and to provide a novel platform for collaborations. In addition, interdisciplinary collaborations should be initiated by including cell biologists. Finally, we wanted to create an opportunity for young investigators to start their own and independent research. The DFG-sponsored priority program represents an optimal opportunity to support such collaborative nationwide research on novel ideas, which allows all interested scientists throughout the country to work in the field and to benefit from a collaborative network. The program runs for 6 years, with international reviews at the beginning and after every other year.

Prof. Dr. C. Haass Laboratory for Alzheimer’s and Parkinson’s Disease Research Department of Biochemistry, Adolf Butenandt Institute Ludwig Maximilians University, DE–80336 Munich (Germany) Tel. +49 89 218 075 471, Fax +49 89 218 075 415, E-Mail [email protected]

Fig. 1. Alzheimer laboratories of the Priority Program SPP 1085

in Germany (only members of the last funding period are shown). Newly and previously established laboratories are in red and green, respectively. Fig. 2. The 6th Eibsee Meeting. Fig. 3. Impressions from the Eibsee V and VI conferences. A Left to right: Eckhard Mandelkow (MPI, Hamburg), Raymond Kelleher (MIT), Rudi Tanzi (MGH and Harvard Medical School), Tobias Hartmann (Center for Molecular Biology Heidelberg), Dora Kovacs (MGH and Harvard Medical School), Jee Shen (Harvard Medical School), and Eva Maria Mandelkow (MPI Hamburg). B Peter St. George-Hyslop (University of Toronto). C Christoph Hock (University of Zürich). D Charles Glabe (University of California, Irvine). E Frank LaFerla (University of California, Irvine). F Michael Shelanski (Columbia University). G Gang Yu (University of Texas). H Zugspitze, Eibsee, and the lake at sunrise. I Fred van Leuven (University of Leuven). J Peter Kloetzel (Charite, Berlin) and (right) Gerd Multhaup (Free University Berlin). K Peter Breuer announcing the first Award of the Hans and Ilse Breuer Foundation. L Harald Steiner (Ludwig Maximilians University Munich) receiving the Hans and Ilse Breuer Award for Alzheimer’s Disease Research. M Dietrich Reinhardt (Dean of the Medical Faculty of the Ludwigs Maximilian University of Munich) holding the laudatio on Harald Steiner.

1

2

194

Neurodegenerative Dis 2006;3:193–196

Haass /Multhaup

A

B

E

D

G

C

F

H

I

J

K

L

M

3

Six Years of Collaborative Alzheimer’s Disease Research in Germany

Neurodegenerative Dis 2006;3:193–196

195

After a first and very successful review, the program started under the name ‘Priority Program SPP1085 Cellular Mechanisms of Alzheimer’s Disease’ in spring 1999. We covered many aspects of modern AD research centered around the cell biology of APP, the three different secretase activities and tau. These included in vivo analysis of APP transport, functional analysis of presenilin, identification and reconstitution of -secretase, assembly of the -secretase complex, identification of novel secretase substrates, -secretase function in animal models (Caenorhabditis elegans and Drosophila), cellular analysis of -secretase complex components, life imaging of -secretase trafficking, identification of -secretaselike proteases, functional analysis of APP and its homologues in mice and primary neurons, effects of cholesterol on A peptide generation, functional analysis of BACE and its dimerization, phosphorylation of secretases, the role of -secretase in the prevention of A production, tau aggregation, and effects of abnormally phosphorylated tau on cellular transport. In addition, projects related to novel therapeutic strategies based on the cellular and molecular mechanisms discovered in our program were initiated. Drugs were screened for their potential to block tau aggregation and its detrimental effects on cellular transport, -secretase modulators were investigated for their potential to selectively inhibit A peptide generation without affecting Notch signaling, and novel compounds were designed to block and investigate secretase activities. The outcome of these projects will be summarized by the members of the SPP1085 in their articles in this special issue of Neurodegenerative Diseases. In addition, a list of publications originating from the SPP1085 projects can be found below. By going through these articles one can immediately appreciate that AD research in Germany grew together to a group of internationally competitive scientists. Tauists are now even working with Baptists! Moreover, we established a growing number of novel research teams led by young investigators (Drs. Weggen, Kaether, Lichtenthaler, Kins, and Steiner). Furthermore, several members of the SPP1085 obtained major positions at German Universities (Drs. Baumeister, Hartmann, Pietrzik, Schmidt, Multhaup, Walter, Müller, and Haass), and laboratories focusing on AD are now distributed throughout Germany (fig. 1).

196

Neurodegenerative Dis 2006;3:193–196

Besides funding the research projects, the DFG also supported annual meetings to allow discussion and exchange of data. All members and their co-workers, the reviewers, and selected speakers from all over the world were invited to these meetings. The meetings took place at a rather remote place in the Bavarian Alps and are now known as the Eibsee Meetings. Six such meetings were held throughout the funding period including satellite meetings of the project leaders in Hamburg and Berlin where the co-speaker of the SPP has accepted the chair for biochemistry during the funding period (G. Multhaup, Free University of Berlin). These conferences turned out to be an excellent platform for generating new collaborations and for bringing together German AD researchers. This meeting developed to an internationally well-recognized conference. Importantly, the very special ‘Eibsee atmosphere’ allowed detailed informal discussion during hiking and at the fireplace in the evening (fig. 2, 3). Here, many novel ideas and collaborations developed, frequently with inclusion and advice of our international guest speakers and reviewers. The Eibsee Meeting is now the German platform for AD research. We are extremely excited that the Hans and Ilse Breuer Foundation will generously support this meeting in the future. At this year’s Eibsee VI conference (the last meeting sponsored by the DFG), the foundation already invited Dr. Dennis Selkoe as a keynote speaker. Moreover, the Hans and Ilse Breuer Foundation will not only help to keep German AD researchers together by supporting the Eibsee conference, but will also directly support AD research in Germany by offering a major research award (EUR 100,000) and several stipends for PhD students. It is our great pleasure to congratulate Dr. Harald Steiner as the first recipient of this award. Finally, we want to thank the DFG (Drs. Golla and Schmidtmann) for the strong support throughout these years. We also greatly acknowledge the tremendous work and important scientific advice from the international board of reviewers (Drs. St. George-Hyslop, Glabe, Heppner, Kloetzel, van Leuven, Saftig, and Aguzzi). Many of them not only had to review numerous applications but also travel long distances for the final board meetings and our conferences. We are looking forward to fruitful and long-lasting excellent AD research in Germany based on this successful DFG program!

Haass /Multhaup

Original Paper

Diseases

Neurodegenerative Dis 2006;3:197–206 DOI: 10.1159/000095257

Spectroscopic Approaches to the Conformation of Tau Protein in Solution and in Paired Helical Filaments M. von Bergen S. Barghorn S. Jeganathan E.-M. Mandelkow E. Mandelkow Max Planck Unit for Structural Molecular Biology, Hamburg, Germany

Key Words Paired helical filament Tau protein Alzheimer’s disease Microtubule-associated proteins

Abstract The abnormal aggregation of the microtubule-associated protein tau into paired helical filaments is one the hallmarks of Alzheimer’s disease. This aggregation is based in the partial formation of -structure. In contrast, the soluble protein shows a mostly random coil structure, as judged by circular dichroism, Fourier transform infrared, X-ray scattering and biochemical assays. Here, we review the basis of the natively unstructured character of tau, as well as recent studies of residual structure and long-range interactions between different domains of the protein. Analysis of the primary structure reveals a very low content of hydrophobic amino acids and a high content of charged residues, both of which tend to counteract a well-folded globular state of proteins. In the case of tau, the low overall hydrophobicity is sufficient to explain the lack of folding. This is in contrast to other proteins which also carry an excess charge at physiological pH. By tryptophan scanning mutagenesis and fluorimetry we found that most of the sequence is solvent exposed. Analysis of the hydrodynamic radii confirms a mostly random coil structure of various tau isoforms and tau domains. The pro-

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0197$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

teins can be further expanded by denaturation with GdHCl which indicates some global folding. This was substantiated by a FRET-based approach where the distances between different domains of tau were determined. The combined data show that tau is mostly disordered and flexible but tends to assume a hairpin-like overall fold which may be important in the transition to a pathological aggregate. Copyright © 2006 S. Karger AG, Basel

The Microtubule-Associated Protein Tau

Tau is a neuronal microtubule-associated protein whose expression is strongly upregulated during neuritogenesis [1]. Its function is the stabilization of neuronal microtubules for neurite outgrowth and for their role as tracks for intracellular transport of vesicles, organelles, and protein complexes by motor proteins [for review, see 2]. In Alzheimer’s disease (AD) and other ‘tauopathies’, including frontotemporal dementias with parkinsonism with mutations in the tau gene [for review, see 3], tau is altered in several ways, e.g. by hyperphosphorylation, aggregation, missorting from the axonal into the somatodentritic compartment, glycation, and amino acid modifications. Current hypotheses about tau’s pathological functions include three distinct aspects:

M. von Bergen Max Planck Unit for Structural Molecular Biology Notkestrasse 85 DE–22607 Hamburg (Germany) Tel. +49 40 8998 2810, Fax +49 40 8971 6822, E-Mail [email protected]

htau40 1

I1 I2

P1 P2

1

P1 P2

1

2

3

4

441

3

4

441

3

4

372

3

4

372

htau23 1

K18

244 1

K19

244 1

2

[8]. Clusters of positively charged amino acids in the repeats have recently been discovered to be involved in microtubule binding [9]. They are neutralized by the negatively charged C-terminal half of tubulin. The paired helical filament (PHF) assembly function of tau is also centered in the repeat domain since this domain contains elements promoting -structure [10]. Indeed, the core of PHFs from Alzheimer brains and from PHFs assembled in vitro contains the repeat domain while the N- and C-terminal domains contribute to the ‘fuzzy coat’ [11, 12].

Fig. 1. Isoforms and constructs of tau. The longest isoform of tau

(htau40 = 2N4R, top) contains two N-terminal inserts and four repeats, whereas the shortest naturally occurring isoform, htau23 (0N3R) lacks the two N-terminal inserts and also the second repeat. For further analysis, shorter constructs comprising only the four or three repeats of the microtubule binding domain (K18 and K19) were used. In the shorter constructs, the wild-type sequence is preceded by an additional methionine for recombinant expression.

• tau aggregates may obstruct the intracellular distribution and movement of cell components, • hyperphosphorylated tau in AD cannot stabilize microtubules which therefore break down and no longer serve their function in intracellular transport, • elevated binding of tau to MT could lead to overstabilization and to a blockade of transport due to interference between tau and motor proteins. The tau gene contains 15 exons; exons 1–5, 7, 9–14 are expressed in the CNS, leading to six isoforms generated by alternative splicing [4, 5] (fig. 1). A further exon (4a) leads to a ‘big tau’ isoform expressed in the PNS. The six CNS isoforms are distinguished by the presence of absence of two near-N-terminal inserts (29 residues each, N1, N2, coded by exons 2 and 3), and the second of four repeats (31 residues, R2, exon 10). This generates isoforms ranging from 0N3R (smallest tau, 352 residues) to 2N4R (largest, 441 residues). Tau has overall a basic character, but the N-terminal third is acidic (up to residue 120) and therefore projects away from the acidic microtubule surface (‘projection domain’). The microtubule-binding function is centered in the repeat domain (R1–R4, residues 244–368, fig. 1) but requires the basic proline-rich flanking domains for tight binding. The repeat sequences are highly conserved in tau and also show significant homology to microtubule-binding repeats of other microtubule-associated proteins [6, 7]. Tau binds along and across protofilaments on the outer surface of the microtubule 198

Neurodegenerative Dis 2006;3:197–206

Primary Structure of Tau and Prediction of Folding

The number of recognized natively unfolded proteins or proteins with unfolded domains has increased rapidly in recent years [13–15]. The list contains proteins with different functions, such as entropic chains like the nucleoporin Nup2p [16], which is involved in gating the nuclear pore complex, inhibitors of proteases like calpastatin [17, 18] and proteins which promote assembly like microtubule-associated proteins in the case of microtubules and caldesmon in the case of actin [19]. The most important feature of unfolded proteins is a low content of hydrophobic amino acids and thus a predominance of hydrophilic amino acids. In addition, a high net charge is often found in unfolded proteins [15]. The charge causes electrostatic repulsion, which prevents the molecule from forming close intramolecular contacts. Tau fits well into the scheme, because it contains only a very small content of hydrophobic amino acids (for the longest isoform the content of the hydrophobic amino acids I, W, L, F, V accounts to 15% altogether). The net charge of tau at pH 7 differs between the isoforms and ranges from +8 (2N4R) to +15 (2N3R). This indicates that the repeat sequences in the microtubule binding domain contain more positively charged amino acids and the two alternatively spliced inserts at the N-terminus bear an excess of negatively charged amino acids. The original studies of tau from brain tissue already revealed its unusual features because the protein was devoid of secondary structure (by circular dichroism, CD) and retained its biological function even after harsh treatment (boiling, acid) [20, 21]. Cloning of the gene [22, 23] revealed a composition rich in hydrophilic and charged residues, and poor in hydrophobic residues. In the meantime, a large number of such disordered proteins or protein domains have been discovered and classified, and several algorithms exist that predict their ocvon Bergen/Barghorn/Jeganathan/ Mandelkow/Mandelkow

currence. There are two types of algorithms, the first is based on the statistical analysis of the abundance of amino acids in sequences showing disorder [24, 25] and the second takes into consideration the ratio between net charge and hydrophobicity in a window of observation [26, 27]. Figure 2 shows the prediction for the longest isoform of tau (2N4R) and as a comparison the results of -tubulin, known to be a well-folded globular protein. Based on the prediction of PONDR (prediction of natural disordered regions, http://www.pondr.com) which is based on statistical distribution of amino acids in sequences of disordered proteins, the probability of disorder was predicted with an algorithm built from a neuronal network that was trained to find determinants of disordered sequences from a set of well-folded and disordered proteins [28]. The original output varies from 0 to 1, where the higher values indicate a higher probability of folding. We transformed this range into a –0.5 to 0.5 scale for better comparison. In such a plot, positive values indicate a high likelihood for order, negative values hint at disordered sequences. The black line in figure 2b represents the PONDR prediction of tau. The probability for order is low for most of the sequence except for the repeat sequences in the microtubule binding domain. Thus, although these predictions show a general agreement with the experimental data (which reinforce the view of an unfolded protein, see below), they disagree for the repeat domain because the predictions suggest a folded state, whereas the domain is in fact also natively unfolded, as judged by different spectroscopic methods [10, 29, 30]. The apparently misleading prediction of an ordered structure in the repeat domain might stem from two types of motifs. The first comprises the hexapeptide motifs at the beginning of the second, third and the fourth repeat, which have a more hydrophobic character and an enhanced propensity for -structure, in agreement with NMR evidence [9]. The second reason for the prediction of seemingly ordered structure arises from the PGGG motifs at the end of the four repeats. These PGGG motifs have a strong prediction to form a turn structure and thereby are classified as folded by the PONDR algorithm. Finally, the prediction for the C-terminal tail shows a partially folded structure which is consistent with the prediction of an amphipathic -helix in the range 421– 441 [31], and with NMR results obtained TFE (trifluoroethanol) [32]. In contrast, for the globular protein -tubulin (fig. 2a) the probability of folding lies mostly between 0.0 and 0.5, indicating ordered secondary structure with the excep-

tion of the C-terminal 30 residues which are known to be disordered and exposed to the exterior [33, 34]. A second approach to the prediction of unfolded proteins is based on the ratio of the mean net charge and the mean hydrophobicity [15]. To show more detailed infor-

Conformations of Tau

Neurodegenerative Dis 2006;3:197–206

0

Foldability

0,4 0.4 0,2 0.2 00,0 -0,2 -0.2 -0,4 -0.4

100

200

300

400

α-tubulin Pondr

a

htau40 Pondr

0,4 0.4 0,2 0.2 00,0 -0,2 -0.2 -0,4 -0.4

b

0,4 0.4 0,2 0.2 0 0,0 -0,2 -0.2 -0,4 -0.4

α-tubulin FoldIndex

0,4 0.4 0,2 0.2 0 0,0 -0,2 -0.2 -0,4 -0.4

c

htau40 FoldIndex 0

100

200

300

d 400

Residue htau40 1

I1 I2

P1 P2

1

2 3 4

441

Fig. 2. Algorithms predict tau to be mostly unfolded. The primary sequences of htau40 (b, d) and -tubulin (a, c) were analyzed

by algorithms which were designed to predict unfolded sequences and to detect the probability of proteins to fold in a compact fashion (foldability). a, b Prediction by PONDR-VL-XT (window size = 9 amino acid residues) with the highest predictive value for foldability of 0.5 and a high likelihood for disordered conformation at –0.5 [64]. Note the low degree of foldability for tau except for the repeat domain and the C-terminal tail. c, d Analysis by the algorithm of FoldIndex [27] is shown for tau (d) and -tubulin (c) with a gliding window of 20 residues. The algorithm predicts a disordered conformation for the whole sequence of tau except for the sequence from the middle of repeat 2 to the end of repeat 3 and around the C-terminus (note that this is contrary to experimental observation). d A scheme of the primary structure of the longest isoform htau40 is shown.

199

mation on different parts of a protein, Prilusky et al. [27] developed an algorithm which performs a calculation based on the prediction of Uversky with a gliding window of 20 amino acids over the whole protein sequence with a stepsize of 5 (http://bioportal.weizmann.ac.il/fldbin/ findex). The scoring is normalized so that the highest likelihood for a compact fold is 0.5, whereas for disorder it is –0.5. The equation determining the FoldIndex I is shown in formula 1: I = 2.785 !H1 – !R 1 – 1.151

where !H1 represents the mean hydrophobicity according to the Kyte/Doolittle scale [35] and !R 1 is the mean net charge as the absolute value of the difference between positive and negative charges at pH 7 divided by the number of residues. In figure 2d, the black line represents the prediction for htau40 (2N4R), showing only negative values except for a short region between the end of the second and the middle of the third repeat and for the C-terminal tail. This prediction is in good agreement with spectroscopic data of tau, especially recent studies by NMR [9], which have revealed that the repeats of tau contain some residual -structure at the beginning of each repeat. Among the four repeats, the tendency to form structure was most prominent in the third repeat. Apart from some short stretches within the repeats, there is some predicted tendency to form -helix near the C-terminus. By comparison, the prediction for human -tubulin is shown in figure 2c. As in the PONDR prediction, nearly the whole sequence shows a significant probability to be folded with the exception of the C-terminal tail. Overall the two algorithms, although based on different principles, predict the majority of the tau sequence to be disordered, consistent with experimental evidence. The predictions show a tendency for some degree of folding within the repeats and at the C-terminus, and the PONDR algorithm also predicts some folding near the N-terminus. However, these features are not backed up by experimental data so far. There are two more pieces of prediction software available to predict disordered sequences. The software GLOBPLOT [36] (http://globplot.embl.de/) defines intrinsically disordered sequences as the absence of ordered structure. The software DisEMBL [37] (http://dis.embl.de/) uses three different approaches: Firstly, all sequences from known protein structures that are neither helical nor -structure are treated as loops. Secondly, from this group a subgroup was defined by using the temperature factors (B-factors) of C atoms, and thirdly these sequences were crosschecked with a database of sequences 200

Neurodegenerative Dis 2006;3:197–206

that did not show up in X-ray structures, assuming that these must be random coil and thereby are not suitable for crystallization. The software package also includes the ‘tango’ algorithm which provides a prediction of amyloid formation of random coil sequences [38] (http://dis. embl.de). This algorithm calculates the entropic costs of a sequence becoming fully buried, and takes into account the amount of hydrogen bonds and electrostatic interactions with regard to the pI of the protein and the buffer system used. Recently, the crystallization of amyloidogenic hexapeptides [39] led to an algorithm which maps sequences against these hexapeptides with respect to the three-dimensional structure and the likeliness to adopt a similar conformation [40] (http://rosettadesign.med. unc.edu/). Predictions of disorder are generally limited to local aspects of folding. No software is available to predict how different domains of a long and unfolded protein chain might interfere. CD or Fourier transform infrared (FTIR) experiments on disordered proteins would barely detect the presence of higher order states of folding. In these cases one has to rely on spectroscopies which report on the vicinity of residues labeled with spectroscopic markers, such as FRET for fluorescent labels, or spin labels used in EPR and NMR. As an example, FRET results show that the N-terminal and C-terminal domains of tau in solution are folded into the vicinity of the repeat domain [41; see also below].

Global Features of Unfolded Proteins

Because of the sharp protein-solvent boundary, compact proteins have a well-defined radius of gyration which can be determined by small-angle X-ray scattering. For globular proteins RG =

3 RS , 5

where RG is the radius of gyration and RS the Stokes radius. The Stokes radii of tau isoforms and different domains comprising the repeats were determined by size exclusion gel chromatography and plotted versus the molecular weight (fig. 3). The results can be compared with the values for a set of well-folded proteins, molten globules and premolten globules [26]. The tau isoforms measured in phosphate-buffered saline (fig. 3a) revealed radii between 48–54 Å. The shorter constructs K19 and K18 consisting of three or four repeats showed Stokes radii of 25 and 28 Å, respectively. The high ratios of Stokes radii von Bergen/Barghorn/Jeganathan/ Mandelkow/Mandelkow

ed ur t na tau de

Stokes radius (Å)

100

ul e l e lob lobu g g n lte ten ule mo mol glob e pr

50

25

10

100 a

100

10

MW (kDa)

BSA

MWM

Molecular weight (kDa) hTau40

66 45 35

100

500

0 1,000

500 100

0 1,000

25 b

Protein:trypsin ratio Fig. 3. Stokes radius and sensitivity to proteolysis prove tau’s disorder. a Stokes radii of a set of standard proteins for well-folded proteins (black circles), molten globules (dark gray circles), premolten globules (light gray circles) and chemically denatured proteins (white circles). Tau isoforms and constructs dissolved in phosphate buffered saline are shown as filled black squares, the same proteins dissolved in 2 M GdHCl are shown in open squares. In PBS, the isoforms and constructs of tau fit to the reported Stokes radius: molecular weight ratio [15]. Chemical denaturation increases the Stokes radius, but only slightly, supporting the hypothesis that native tau is already mostly unfolded. b Sensitivity of tau towards proteolysis. Note that htau40 is readily degraded at a protein:trypsin ratio of 1,000: 1, whereas bovine serum albumin remains unaffected even at much higher trypsin levels (protein:trypsin ratio = 100:1).

to their molecular weight classifies all tau isoforms and the shorter constructs as mostly natively unfolded proteins. Nevertheless, after denaturation with 2 M GdHCl the Stokes radii increased even further. The shift is greatest in the case of the shorter constructs K19 and K18 indicating some residual structure which was affected by GdHCl. The lack of structure makes unfolded protein highly sensitive for proteolytic cleavage. This is illustrated in figure 3b where BSA (a globular protein) which is nearly resistant to low concentrations of trypsin is compared with tau protein which is rapidly degraded at the same low trypsin concentrations. Because of the extended structure of natively unfolded proteins they are expected to expose most of their sequence to the solvent. This can be tested by tryptophan fluorescence, which is sensitive to the solvent. Tryptophan in water shows an emission maximum around 352 nm, whereas the maximum from tryptophan buried within the protein shows a blue shift to 310–340 nm [42]. Tau contains no tryptophan in the wild-type sequence and therefore lends itself to a tryptophan scanning analysis where tryptophans are introduced at different places in the sequence [43]. In all positions, tryptophan exhibits an emission maximum at about 352 nm indicating solvent exposure. However, there is a significant blue shift to 340 nm upon polymerization into PHFs, indicating a more hydrophobic environment. The solvent accessibility can also be tested by fluorescence quenching experiments, which can be used to confirm and extend the information provided by the emission maximum (such as relative degree of solvent exposure). Similarly, in the case of the polymerized protein the solvent accessibility determined by quenching experiments underscores the interpretations derived from the emission maximum. Sites in the sequence with a stronger blue shift after polymerization exhibit also a decreased solvent accessibility [44]. This allows one to generate an accessibility map of the protein in the polymerized state, showing a minimum of accessibility in repeats R2 and R3, moderate accessibility in R1 and R4, and high accessibility elsewhere. This is in excellent agreement with the notion that R2 and R3 lie in the core of PHFs.

Long-Range Structure in Tau

Most spectroscopic techniques such as tryptophan fluorescence, CD and FTIR detect the local conformation and therefore long-range interactions like those between domains of a protein cannot be analyzed. In the case of Conformations of Tau

Neurodegenerative Dis 2006;3:197–206

201

R-R 1.4 1

htau40 - 310W 322DANS I1

I2

441

P1 P2 1 2 3 4

E = 0.67 R = 19.5

0.7

Fluorescence intensity (AU)

0.0

htau40 - 310W 435DANS

R-C

1.4 1

I1

I2

P1 P2 1 2 3

E = 0.73 R = 18.4

0.7

0.0

htau40 - 18W 291DANS

R-N

1.4 1

I1

I2

1.4 1

441

P1 P2 1 2 3 4

E = 0.08 R = 33.5

0.7

0.0

441

4

htau40 - 432W 17DANS

N-C I1

I2

441

P1 P2 1 2 3 4

E = 0.59 R = 20.8

0.7

0.0 350

400

450

500

Wavelength (nm) Fig. 4. FRET reveals residual structure within the repeats and interactions between the C-terminal region and the repeats or the N-terminal region. The curves illustrate FRET experiments with the donor tryptophan (black circle) and the acceptor IAEDANS (dark gray star) at different places within the longest tau isoform htau40. The emission is shown of donor alone (solid line) and in the presence of the acceptor (dashed line). FRET is revealed by a decrease of the tryptophan emission in the presence of the acceptor (arrows). From top to bottom: FRET within the repeats (R-R), between repeats and C-terminus (R-C), between repeats and the N-terminus (R-N) and between the two termini (N-C).

be drawn for proteins on the basis of small angle scattering when the radius of gyration differs from that expected for a ‘random coil’ expectation [45]. Arguments for long-range interactions have also been derived from the reaction of tau with conformation-dependent antibodies which detect nonlinear epitopes. They argue that there is some interaction between different domains of tau, e.g. between the N-terminal domain and the repeats, or between the C-terminal domain and the repeats [46–48]. To detect long-range interactions we followed a FRET-based approach, creating single tryptophan and cysteine mutants in tau. The cysteines were labeled with IAEDANS, which serves as an acceptor for the emission of the tryptophan fluorescence (see fig. 4) [49]. The FRET efficiency allows the calculation of distances between fluorescence donor and acceptor. It was found that the distances between a tryptophan at the position 310 and the IAEDANS located at either 291 or 322 (fig. 4) are shorter than expected on the basis of a random coil model, indicating residual structure that leads to more compact folding (from 310 to 291, observed: 2.16 nm, theoretical: 3.65 nm; from 310 to 322, observed: 1.95 nm, theoretical: 2.9 nm). More significantly, this study revealed a hairpin-like folding of tau in solution which brings the N- and C-terminus into the vicinity of the central repeat domain. Based on the same FRET approach, we analyzed several distances between tryptophan residues placed near the C-terminus and IAEDANS moieties located in the repeat region of tau and vice versa. Interestingly, these FRET pairs showed high FRET efficiencies and consequently distances of 2.2–2.5 nm which is much shorter than predicted by a random coil model (8.7–9.9 nm). A less pronounced but still unexpectedly close vicinity was observed between the N-terminus and the repeat region [49]. The data are summarized in a model (fig. 6) showing the C-terminus coming near both the repeat region and the N-terminus. The interactions within tau are labile and tend to disappear at 1 M concentrations of GdHCl, as judged by the decrease of FRET efficiency. Under these conditions the two termini also approach each other, showing distances of between 2.0–2.4 nm, which is much less than the expected distance in a pure random coil model of 15–17 nm.

(Un)folding and Aggregation

tau protein, the differential effects of denaturants (GdHCl) on the hydrodynamic radii of full length isoforms and constructs point to the fact that some higher-order folding exists in the protein. A similar global conclusion can 202

Neurodegenerative Dis 2006;3:197–206

Among the amyloidogenic proteins and peptides which are linked to diseases one can distinguish two classes regarding their structure and their mechanism of von Bergen/Barghorn/Jeganathan/ Mandelkow/Mandelkow

5e+3 (deg cm2 dmol-1)

Mean res. ellip.

K19

0 -5e+3 PHF

-1e+4 -2e+4

soluble

-2e+4

a 190 200 210 220 230 240 250 260

Wavelength (nm) Rel. absorbance

1.0

K19 soluble

PHF

0.5 C

0.0

b

1,700 1,680 1,660 1,640 1,620 1,600

Wave number (cm-1)

Filament axis

aggregation. The first class comprises those proteins which are normally well-folded into compact structures and need this folding in order to fulfill their physiological function. Examples are transthyretin and prion protein. For these proteins it is necessary to undergo a partial unfolding before they can adopt the amyloidogenic conformation. During this process, their secondary structure is changed in comparison with the physiological state, and therefore this transition is aptly considered as ‘misfolding’. This category includes well-folded proteins whose unfolding in experimental conditions has been shown to lead to amyloid aggregation, e.g. lysozyme and others [50]. The other group of amyloidogenic proteins contains unfolded domains as part of their native state, which then fold more easily into an amyloid structure during pathological aggregation. Examples for this group are the A peptide [51], -synuclein [52] and tau [53, 54]. These proteins aggregate by formation of extended -sheets but in contrast to the first group they do not contain a welldefined conformation to start with. Thus, building up -structure is comparable to the ab initio process of protein folding. Among the partially unfolded amyloidogenic proteins, tau exhibits the longest unfolded sequence (the longest isoform contains 441 amino acids and no significant secondary structure content in the soluble state). Tau is therefore an extreme example of a natively unfolded protein which undergoes a transition from largely random coil conformation to partial -structure. Generally, in amyloidogenic proteins or peptides the increase in structure is associated with aggregation into fibers which contain a cross- structure, where the -strands are aligned perpendicular to fiber axis [for reviews, see 55– 57]. For PHFs, a diffraction pattern is shown in figure 5c, displaying the typical cross- features.

c

Fig. 5. CD, FTIR and X-ray scattering show -structure in PHFs. a Typical CD spectra of soluble (solid line) and aggregated K19-

protein (dashed line). The soluble tau exhibits a minimum around 200 nm indicating a mostly random coil structure, whereas the polymerized sample shows a shift of the minimum to higher wavelengths pointing to an increased content of -structure. b FTIR spectra of soluble (solid line) and aggregated K19 (dashed

line). Note the shift of the maximum from 1,645 cm–1 to about 1,630 cm–1 from the soluble to the polymerized sample, indicating a conformational change from random coil to -structure. c Partially oriented filaments made out of K18K280 and examined by X-ray fiber diffraction. The meridional reflections at 0.47 nm and equatorial reflections at 1.07 nm are characteristic of cross- structure.

Conformations of Tau

Neurodegenerative Dis 2006;3:197–206

203

R2

R1

R3

22. .33 n nm m

2.0 nm

R4 C

N

has been studied. For most amyloid structures, certain compounds have been reported which are capable of dissolving these fibers. In the case of tau, we found that compounds from the class of the anthraquinones are able to inhibit and reverse PHF assembly [60], and comparable results were reported by other groups [61–63]. One would expect that these compounds have to weaken the interactions and thereby the internal structure of the PHFs, which might be dominated by the -sheet interactions between the hexapeptide motifs within the repeats. However, for none of the compounds has the mode of action been clarified so far.

Conclusions

Fig. 6. Model of higher-order structure and long-range interactions in tau. The figure summarizes the higher-order structure within tau protein deduced by FRET. The repeats are color coded (R1 = blue, R2 = green, R3 = red, R4 = yellow). The C-terminus (shown as yellow tubular structure) comes close to the repeats and also close to the N-terminal domain (magenta structure).

Spectroscopic methods such as CD or FTIR allow a rough estimate of the content of secondary structure within aggregated fibers such as PHFs. For example, the percentage of -sheet in tau fibers was estimated to account for about one third of the repeat domain. This indicates that a relatively small number of amino acids (30– 40) is involved in the intra- and intermolecular interactions within the PHFs, while the rest of the protein likely contributes to the fuzzy coat of the filaments [11]. The limited number of interactions might be the reason for the low stability of PHFs in the presence of GdHCl. The half maximal concentration of 1.1 M of GdHCl for dissolution of the tau fibers is astonishingly small, compared with well-folded globular proteins which usually require half maximal concentration of GdHCl for denaturation of 2.75 M [58]. There are little data available on the stability of other amyloid fibers in the presence of GdHCl, but at least protofibrils of insulin appear to have a similar high sensitivity to GdHCl and dissolve at a half maximal concentration near 1 M [59]. In other studies, the stability of amyloids in the presence of low molecular weight compounds 204

Neurodegenerative Dis 2006;3:197–206

The conformation of tau can by described as follows: 1 Tau belongs to the emerging class of natively unfolded proteins, characterized by properties such as the hydrodynamic radius, solvent accessibility, CD, FTIR spectroscopy and sensitivity towards proteolytic digestion. 2 The unfolded state of tau is caused mainly by the low content of hydrophobic amino acids. 3 Superimposed on the mostly unfolded chain, there is evidence for higher-order residual folding, e.g. a compaction of the repeat domain (relative to a ‘random coil’), a partially helical C-terminal tail, or the hairpin folding between the N-terminus, C-terminus, and repeat domain. 4 During pathological aggregation, tau undergoes a conformational change in which some sequences in the repeat domain form -structure, whereas most of the rest of the protein stays unfolded. The remarkable stability of the unfolded state of tau raises the question why it polymerizes at all in AD and related frontotemporal neurodegeneration. To answer this question, a better understanding of the parameters determining the unfolded state will be needed. This might provide tools to protect the natively unfolded state and thus to prevent pathological aggregation.

von Bergen/Barghorn/Jeganathan/ Mandelkow/Mandelkow

References 1 Drubin DG, Kirschner MW: Tau protein function in living cells. J Cell Biol 1986; 103: 2739–2746. 2 Garcia ML, Cleveland DW: Going new places using an old MAP: tau, microtubules and human neurodegenerative disease. Curr Opin Cell Biol 2001; 13:41–48. 3 Goedert M, Spillantini MG: Tau mutations in frontotemporal dementia FTDP-17 and their relevance for Alzheimer’s disease. Biochim Biophys Acta 2000; 1502:110–121. 4 Andreadis A, Brown WM, Kosik KS: Structure and novel exons of the human tau gene. Biochemistry 1992;31:10626–10633. 5 Andreadis A: Tau gene alternative splicing: expression patterns, regulation and modulation of function in normal brain and neurodegenerative diseases. Biochim Biophys Acta 2005;1739:91–103. 6 Goedert M, Spillantini MG, Jakes R, Rutherford D, Crowther RA: Multiple isoforms of human microtubule-associated protein tau: sequences and localization in neurofibrillary tangles of Alzheimer’s disease. Neuron 1989;3:519–526. 7 Friedhoff P, Mandelkow E: Tau protein; in Kreis T, Vale R (eds): Guidebooks to the Cytoskeletal and Motor Proteins and to the Extracellular Matrix and Adhesion Proteins. Oxford, Oxford University Press, 1998. 8 Santarella RA, Skiniotis G, Goldie KN, Tittmann P, Gross H, Mandelkow EM, Mandelkow E, Hoenger A: Surface-decoration of microtubules by human tau. J Mol Biol 2004; 339:539–553. 9 Mukrasch MD, Biernat J, von Bergen M, Griesinger C, Mandelkow E, Zweckstetter M: Sites of TAU important for aggregation populate beta-structure and bind to microtubules and polyanions. J Biol Chem 2005; 280:24978–24986. 10 von Bergen M, Friedhoff P, Biernat J, Heberle J, Mandelkow EM, Mandelkow E: Assembly of tau protein into Alzheimer paired helical filaments depends on a local sequence motif ((306)VQIVYK(311)) forming beta structure. Proc Natl Acad Sci USA 2000; 97: 5129–5134. 11 Wischik CM, Novak M, Edwards PC, Klug A, Tichelaar W, Crowther RA: Structural characterization of the core of the paired helical filament of Alzheimer disease. Proc Natl Acad Sci USA 1988;85:4884–4888. 12 von Bergen M, Barghorn S, Müller SA, Pickhardt M, Biernat J, Mandelkow EM, Davies P, Aebi U, Mandelkow E: The core of Tau paired helical filaments studied by STEM and limited proteolysis. Biochemistry 2006; 45:6446–6457. 13 Wright PE, Dyson HJ: Intrinsically unstructured proteins: re-assessing the protein structure- function paradigm. J Mol Biol 1999;293:321–331.

Conformations of Tau

14 Tompa P: Intrinsically unstructured proteins. Trends Biochem Sci 2002;27:527–533. 15 Uversky VN, Gillespie JR, Fink AL: Why are ‘natively unfolded’ proteins unstructured under physiologic conditions? Proteins 2000;41:415–427. 16 Denning DP, Uversky V, Patel SS, Fink AL, Rexach M: The Saccharomyces cerevisiae nucleoporin Nup2p is a natively unfolded protein. J Biol Chem 2002; 277:33447–33455. 17 Murachi T, Tanaka K, Hatanaka M, Murakami T: Intracellular Ca2+-dependent protease (calpain) and its high-molecular-weight endogenous inhibitor (calpastatin). Adv Enzyme Regul 1980;19:407–424. 18 Uemori T, Shimojo T, Asada K, Asano T, Kimizuka F, Kato I, Maki M, Hatanaka M, Murachi T, Hanzawa H, et al: Characterization of a functional domain of human calpastatin. Biochem Biophys Res Commun 1990;166:1485–1493. 19 Permyakov SE, Millett IS, Doniach S, Permyakov EA, Uversky VN: Natively unfolded C-terminal domain of caldesmon remains substantially unstructured after the effective binding to calmodulin. Proteins 2003; 53: 855–862. 20 Weingarten MD, Lockwood AH, Hwo SY, Kirschner MW: A protein factor essential for microtubule assembly. Proc Natl Acad Sci USA 1975;72:1858–1862. 21 Cleveland DW, Hwo SY, Kirschner MW: Physical and chemical properties of purified tau factor and the role of tau in microtubule assembly. J Mol Biol 1977; 116:227–247. 22 Lee G, Cowan N, Kirschner M: The primary structure and heterogeneity of tau protein from mouse brain. Science 1988; 239: 285– 288. 23 Goedert M, Wischik CM, Crowther RA, Walker JE, Klug A: Cloning and sequencing of the cDNA encoding a core protein of the paired helical filament of Alzheimer disease: identification as the microtubule-associated protein tau. Proc Natl Acad Sci USA 1988;85: 4051–4055. 24 Romero P, Obradovic Z, Li X, Garner EC, Brown CJ, Dunker AK: Sequence complexity of disordered protein. Proteins 2001; 42: 38– 48. 25 Li X, Romero P, Rani M, Dunker AK, Obradovic Z: Predicting protein disorder for N-, C-, and internal regions. Genome Inform Ser Workshop Genome Inform 1999;10:30–40. 26 Uversky VN: Natively unfolded proteins: a point where biology waits for physics. Protein Sci 2002;11:739–756. 27 Prilusky J, Felder CE, Zeev-Ben-Mordehai T, Rydberg E, Man O, Beckmann JS, Silman I, Sussman JL: FoldIndex: a simple tool to predict whether a given protein sequence is intrinsically unfolded. Bioinformatics 2005; 21:3435–3438.

28 Romero, Obradovic, Dunker K: Sequence data analysis for long disordered regions prediction in the calcineurin family. Genome Inform Ser Workshop Genome Inform 1997; 8:110–124. 29 Schweers O, Schonbrunn-Hanebeck E, Marx A, Mandelkow E: Structural studies of tau protein and Alzheimer paired helical filaments show no evidence for beta-structure. J Biol Chem 1994; 269:24290–24297. 30 Barghorn S, Davies P, Mandelkow E: Tau paired helical filaments from Alzheimer’s disease brain and assembled in vitro are based on beta-structure in the core domain. Biochemistry 2004;43:1694–1703. 31 Abraha A, Ghoshal N, Gamblin TC, Cryns V, Berry RW, Kuret J, Binder LI: C-terminal inhibition of tau assembly in vitro and in Alzheimer’s disease. J Cell Sci 2000; 113: 3737– 3745. 32 Esposito G, Viglino P, Novak M, Cattaneo A: The solution structure of the C-terminal segment of tau protein. J Pept Sci 2000; 6: 550– 559. 33 Chau MF, Radeke MJ, de Ines C, Barasoain I, Kohlstaedt LA, Feinstein SC: The microtubule-associated protein tau cross-links to two distinct sites on each alpha and beta tubulin monomer via separate domains. Biochemistry 1998;37:17692–17703. 34 Downing KH, Nogales E: Crystallographic structure of tubulin: implications for dynamics and drug binding. Cell Struct Funct 1999;24:269–275. 35 Kyte J, Doolittle RF: A simple method for displaying the hydropathic character of a protein. J Mol Biol 1982;157:105–132. 36 Linding R, Russell RB, Neduva V, Gibson TJ: GlobPlot: exploring protein sequences for globularity and disorder. Nucleic Acids Res 2003;31:3701–3708. 37 Linding R, Jensen LJ, Diella F, Bork P, Gibson TJ, Russell RB: Protein disorder prediction: implications for structural proteomics. Structure 2003;11:1453–1459. 38 Fernandez-Escamilla AM, Rousseau F, Schymkowitz J, Serrano L: Prediction of sequence-dependent and mutational effects on the aggregation of peptides and proteins. Nat Biotechnol 2004;22:1302–1306. 39 Nelson R, Sawaya MR, Balbirnie M, Madsen AO, Riekel C, Grothe R, Eisenberg D: Structure of the cross-beta spine of amyloid-like fibrils. Nature 2005;435:773–778. 40 Thompson MJ, Sievers SA, Karanicolas J, Ivanova MI, Baker D, Eisenberg D: The 3D profile method for identifying fibril-forming segments of proteins. Proc Natl Acad Sci USA 2006;103:4074–4078. 41 Jeganathan S, von Bergen M, Brutlach H, Steinhoff HJ, Mandelkow E: Global hairpin folding of tau in solution. Biochemistry 2006;45:2283–2293.

Neurodegenerative Dis 2006;3:197–206

205

42 Hannemann F, Bera AK, Fischer B, Lisurek M, Teuchner K, Bernhardt R: Unfolding and conformational studies on bovine adrenodoxin probed by engineered intrinsic tryptophan fluorescence. Biochemistry 2002; 41: 11008–11016. 43 Li L, von Bergen M, Mandelkow EM, Mandelkow E: Structure, stability, and aggregation of paired helical filaments from tau protein and FTDP-17 mutants probed by tryptophan scanning mutagenesis. J Biol Chem 2002;277:41390–41400. 44 Eftink MR: Fluorescence techniques for studying protein structure. Methods Biochem Anal 1991;35:127–205. 45 Kohn JE, Millett IS, Jacob J, Zagrovic B, Dillon TM, Cingel N, Dothager RS, Seifert S, Thiyagarajan P, Sosnick TR, Hasan MZ, Pande VS, Ruczinski I, Doniach S, Plaxco KW: Random-coil behavior and the dimensions of chemically unfolded proteins. Proc Natl Acad Sci USA 2004;101:12491–12496. 46 Jicha GA, Bowser R, Kazam IG, Davies P: Alz-50 and MC-1, a new monoclonal antibody raised to paired helical filaments, recognize conformational epitopes on recombinant tau. J Neurosci Res 1997;48:128–132. 47 Skrabana R, Kontsek P, Mederlyova A, Iqbal K, Novak M: Folding of Alzheimer’s core PHF subunit revealed by monoclonal antibody 423. FEBS Lett 2004;568:178–182. 48 Ghoshal N, Garcia-Sierra F, Fu Y, Beckett LA, Mufson EJ, Kuret J, Berry RW, Binder LI: Tau-66: evidence for a novel tau conformation in Alzheimer’s disease. J Neurochem 2001;77:1372–1385.

206

49 Jeganathan S, von Bergen M, Brutlach H, Steinhoff HJ, Mandelkow E: Global hairpin folding of tau in solution. Biochemistry 2006;45:2283–2293. 50 Dobson CM: Protein folding and misfolding. Nature 2003;426:884–890. 51 Selkoe DJ: Folding proteins in fatal ways. Nature 2003;426:900–904. 52 Caughey B, Lansbury PT Jr: Protofibrils, pores, fibrils, and neurodegeneration: separating the responsible protein aggregates from the innocent bystanders. Annu Rev Neurosci 2003;26:267–298. 53 Mandelkow E, Song YH, Schweers O, Marx A, Mandelkow EM: On the structure of microtubules, tau, and paired helical filaments. Neurobiol Aging 1995;16:347–354. 54 von Bergen M, Barghorn S, Biernat J, Mandelkow EM, Mandelkow E: Tau aggregation is driven by a transition from random coil to beta sheet structure. Biochim Biophys Acta 2005;1739:158–166. 55 Kirschner DA, Teplow DB, Damas AM: Twist and sheet: variations on the theme of amyloid. J Struct Biol 2000; 130:87. 56 Sunde M, Serpell LC, Bartlam M, Fraser PE, Pepys MB, Blake CC: Common core structure of amyloid fibrils by synchrotron X-ray diffraction. J Mol Biol 1997;273:729–739. 57 Makin OS, Serpell LC: Structures for amyloid fibrils. FEBS J 2005;272:5950–5961.

Neurodegenerative Dis 2006;3:197–206

58 Sanchez SA, Brunet JE, Jameson DM, Lagos R, Monasterio O: Tubulin equilibrium unfolding followed by time-resolved fluorescence and fluorescence correlation spectroscopy. Protein Sci 2004;13:81–88. 59 Murali J, Jayakumar R: Spectroscopic studies on native and protofibrillar insulin. J Struct Biol 2005;150:180–189. 60 Pickhardt M, Gazova Z, von Bergen M, Khlistunova I, Wang Y, Hascher A, Mandelkow EM, Biernat J, Mandelkow E: Anthraquinones inhibit tau aggregation and dissolve Alzheimer’s paired helical filaments in vitro and in cells. J Biol Chem 2005; 280: 3628–3635. 61 Necula M, Chirita CN, Kuret J: Cyanine dye N744 inhibits tau fibrillization by blocking filament extension: Implications for the treatment of tauopathic neurodegenerative diseases. Biochemistry 2005; 44: 10227– 10237. 62 Taniguchi S, Suzuki N, Masuda M, Hisanaga S, Iwatsubo T, Goedert M, Hasegawa M: Inhibition of heparin-induced tau filament formation by phenothiazines, polyphenols, and porphyrins. J Biol Chem 2005;280:7614– 7623. 63 Wischik CM, Edwards PC, Lai RY, Roth M, Harrington CR: Selective inhibition of Alzheimer disease-like tau aggregation by phenothiazines. Proc Natl Acad Sci USA 1996; 93:11213–11218. 64 Romero P, Obradovic Z, Dunker AK: Natively disordered proteins: functions and predictions. Appl Bioinformatics 2004; 3:105–113.

von Bergen/Barghorn/Jeganathan/ Mandelkow/Mandelkow

Original Paper

Diseases

Neurodegenerative Dis 2006;3:207–217 DOI: 10.1159/000095258

Signaling from MARK to Tau: Regulation, Cytoskeletal Crosstalk, and Pathological Phosphorylation T. Timm D. Matenia X.-Y. Li B. Griesshaber E.-M. Mandelkow Max Planck Unit for Structural Molecular Biology, Hamburg, Germany

Key Words Actin Microtubules p21-activated kinase Phosphorylation of tau

Abstract The hyperphosphorylation of tau is an early step in the degeneration of neurons in Alzheimer’s disease and other tauopathies. Of particular importance is the phosphorylation of tau in the repeat domain which detaches tau from microtubules. This makes microtubules dynamic for their role in differentiation and neurite outgrowth, and it controls the level of tau on the microtubule surface which keeps the tracks clear for axonal transport. However, the detachment of tau from microtubules can also initiate the reactions that lead to pathological aggregation into neurofibrillary tangles. Phosphorylation of tau in the repeat domain is achieved by the kinase MARK/Par-1, a member of the calcium/calmodulin-dependent protein kinase group of kinases. In this report, we focus on the modes of MARK regulation. MARK contains several domains which offer multiple ways of regulation by posttranslational modification (e.g. phosphorylation), interactions with scaffolding proteins and subcellular targeting (e.g. 14-3-3), and interactions with other proteins. We consider in particular the interactions between MARK and other kinases, notably MARKK/TAO-1 and PAK5. MARKK (a member of the Ste20 family of kinases) activates MARK by phosphorylating it at a critical threonine residue within the activation loop. Activated MARK in turn phosphorylates tau, causes its detachment from microtubules and renders them

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0207$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

labile. PAK5 inactivates MARK, not by phosphorylation, but by binding to the catalytic domain. PAK5 contributes to microtubule stability by preventing the MARK-induced phosphorylation of tau; conversely, PAK5 contributes to actin dynamics, presumably through the activation of cofilin, an F-actin severing protein. Thus, MARK and its regulators MARKK and PAK5 appear to mediate the crosstalk between the actin and microtubule cytoskeleton in an antagonistic fashion. Copyright © 2006 S. Karger AG, Basel

Introduction

The function of tau in neurons is to stabilize microtubules and to ensure axonal transport along microtubules. In degenerating neurons, tau is hyperphosphorylated, detaches from microtubules, and aggregates into pathological filaments (paired helical and straight filaments). The detachment from microtubules is achieved most efficiently by phosphorylating the KXGS motifs in the microtubule-binding domain of tau. Enhanced phosphorylation at these sites occurs early in Alzheimer’s disease (AD) [1]. A search for the responsible kinase led to the identification of microtubule-associated protein (MAP)/ microtubule affinity-regulating kinase (MARK) kinases, a subfamily within the calcium/calmodulin-dependent protein kinase group of kinases [2]. These kinases are related to the partitioning defective mutant 1 (Par-1) kinases in Caenorhabditis elegans and Drosophila melano-

E.-M. Mandelkow Max Planck Unit for Structural Molecular Biology Notkestrasse 85 DE–22607 Hamburg (Germany) Tel. +49 40 8998 2810, E-Mail [email protected]

gaster which are involved in the determination of embryonic polarity [for reviews, see 3, 4], and indeed the activity of MARK/Par-1 kinases is important for neuronal polarity as well [5]. MARK kinases consist of an N-terminal catalytic domain, followed by the ubiquitin-associated (UBA) domain, spacer, and tail domains (kinase-associated, KA, domain). The size of MARK and its multidomain composition suggests that regulation could take place on several levels [6]. Indeed, our recent elucidation of the X-ray structure of several domains of MARK2 [7] suggests at least four possibilities: phosphorylation of the activation loop in the catalytic domain, binding of regulatory proteins to the ‘common docking’ (CD) domain, regulation by ubiquitin on the UBA domain, and dimerization. Further regulatory options, derived from studies of Drosophila Par-1, include the interaction with 14-3-3 (alias Par-5, a scaffolding protein) [8], phosphorylation in the spacer domain by atypical protein kinase C (aPKC) and subsequent binding of 14-3-3 [9, 10], and the interaction between the N- and C-terminal tails [11]. To explore these possibilities, we embarked on a program to identify regulatory partners of MARK. Here, we focus on two interaction partners: (1) the activation of MARK by the upstream kinase MARKK which phosphorylates MARK in the regulatory loop of the catalytic domain, (2) the inhibition of MARK by p21-activated kinase 5 (PAK5) and the ensuing effects on the actin/microtubule cytoskeleton. MARKK belongs to the Ste20 family within the STE group of kinases which can be divided into the PAKs (with an N-terminal p21-binding domain and a C-terminal kinase domain) and the germinal center kinases (GCKs; Nterminal kinase domain, no p21-binding domain). About 31 Ste20-related kinases are known (2 PAK, 8 GCK subfamilies, [12]). The kinases have various effects, including the regulation of apoptosis and the rearrangement of the cytoskeleton, and many play a role in the activation of MAP kinases. Regulation of Ste20 kinases is achieved through the Ras family of GTPases, additional protein kinases, adapter proteins coupled to cytokine receptors, and via dimerization or association with inhibitors combined with autophosphorylation after stimulation [13]. MARKK/ thousand-and-one-aminoacid kinase 1 (TAO-1) is part of the kinase subfamily GCK-VIII, together with the Prostate-derived ste20-like kinase (PSK, alias TAO-2) [14] and c-Jun activating kinase (JNK)-inhibiting kinase (JIK, alias TAO-3) [15]. The sequence of MARKK and its two relatives in humans [12] is unusually long, suggesting several functions besides the kinase activity. The active complex from the brain has a mass of 330 kDa which would 208

Neurodegenerative Dis 2006;3:207–217

be compatible with a complex of 2–3 MARKK molecules, or with a complex between MARKK, scaffolding proteins, and other cofactors. MARKK contains predicted amphipathic helices in the spacer and tail domains which would favor protein-protein interactions. The kinase PSK/TAO-2 has been reported to be a regulator of the stress-activated MAP kinase pathway [16], but this is not the case for MARKK/TAO-1, although it has the ability to phosphorylate mitogen-activated protein kinase kinase 3 (MKK3) and MKK6, the activators of p38 stress-activated kinases in vitro [17]. Endogenous MKK3 can be copurified with transfected TAO-1 and TAO-2 from Sf9 cells [14, 17]. This involves the binding of the substrate-binding domain of TAO to the N-terminal header domain of its substrate MKK3 [14]. At present, the relationship between MAP kinase signaling and MARK signaling is unclear, but it is interesting to note that cellular stress can lead to the phosphorylation of KXGS motifs in tau through the activation of MARK [18]. Furthermore, the transcripts of MARK1 and MKK3 are upregulated after differentiation of PC12 cells [19, and our results]. Thus far, upstream effectors or scaffolding proteins for MARKK are unknown. The PAKs are members of Rac/Cdc42-associated Ser/ Thr protein kinases, characterized by a conserved aminoterminal p21-binding domain and a carboxyl-terminal kinase domain. Six human PAKs, which can be classified into two distinct subfamilies, have been identified. Group I PAKs (PAK 1–3) differ significantly in their structural organization and regulation from the more recently discovered group II PAKs (PAK 4–6) [20 ; reviewed in 21–24]. They regulate death and survival signaling, cell-cycle progression, and furthermore they seem to play a key role in coordinating the dynamics of the actin and microtubule cytoskeletons. PAK5 contains 719 residues, binds Cdc42 and occurs mainly in the brain [25, 26]. It contains a p21binding domain (PBD) around residues 9–30, and an inhibitory KI motif around residues 120–133 [27]. The kinase domain extends from about 453 to 700. Ser602 in the activation loop must be phosphorylated for activity. PAK5 can induce filopodia, neurite outgrowth and dendritic spines [25, 28]. It has a mainly cytosolic distribution where it can activate the JNK kinase pathway [25, 26]. Materials and Methods The experimental procedures have been described in previous studies [5, 29–31]. Briefly, plasmids encoding MARK2, PAK5, and active or inactive mutants were generated by standard cloning techniques [5, 31]. Yeast two-hybrid screening and assays were

Timm/Matenia/Li/Griesshaber/ Mandelkow

performed according to the manufacturer’s instructions (Clontech, Yeast Protocols Handbook). Cell culture and transfection were performed with HEK293, Chinese hamster ovary cells (CHO), LAN5 and Sf9 cells following standard protocols. The preparations of kinases and the kinase activity assays have been described in previous studies [2, 29, 31].

Results and Discussion

The initial motivation for our investigations stems from the involvement of tau protein in AD. The abnormal aggregation of this protein correlates well with the clinical stages of AD, Braak stages 1–6 [32], and an analogous correlation is found in other brain diseases with tauopathy, notably frontotemporal dementia and parkinsonism linked to chromosome 17 [for review, see 33]. The major changes of tau in AD include not only abnormal aggregation, but also abnormal hyperphosphorylation and detachment from microtubules. Microtubules are the physiological partners of tau in neuronal axons. Microtubules are necessary for axonal stability, growth cone advance, and as tracks for axonal transport of vesicles and organelles, mediated by motor proteins. In principle, the physiological regulation of the tau-microtubule interaction (by phosphorylation) can affect microtubule dynamics and microtubule-based traffic. While bound tau stabilizes microtubules, detached tau allows them to become dynamic and disassemble. The dynamics is necessary for restructuring the cytoskeleton during changes of cell shape; for example, microtubules must be dynamic, at least temporarily, for neurite outgrowth and differentiation. On the other hand, bound tau not only stabilizes microtubules but also has the potential of inhibiting motor proteins because tau can occlude binding sites on the microtubule surface. In this situation, the removal of excess tau from microtubules can clear the way for axonal traffic. In both contexts, the MARKK-MARK cascade is operational [5, 29, 30, 34]. If it becomes overactive, the detached cytosolic tau protein is free for other interactions, including those that lead to abnormal tau aggregates (paired helical filaments) which in turn also obstruct the cell interior.

Fig. 1. Tau, MARK target sites, and effects of phosphorylation. a The bar diagram illustrates the domains of tau (2N4R isoform,

the largest in human CNS, 441 residues). The inserts near the N-terminus (N1, N2) and repeat R2 may be absent due to alternative splicing, creating the 6 main isoforms in the human CNS. The repeat domain (R1–R4, containing 3 or 4 repeats) and the flanking regions constitute the microtubule (MT)-binding domain. Each repeat contains a KXGS motif (serines 262, 293, 324, 356), which is a target site of MARK kinases. This type of phosphorylation efficiently detaches tau (or other related MAPs) from microtubules, which results in microtubule destabilization. b Diagram of signaling pathway from MARKK and MARK through tau to microtubules, resulting in microtubule breakdown, tau detachment and abnormal aggregation. The microtubules serve as tracks for the transport of vesicles and are stabilized by tau protein (a). Upon phosphorylation of the tau protein by the MARKK/MARK cascade, tau detaches from the microtubule resulting in its disassembly (b). The unbound tau protein then tends to aggregate into paired helical filaments of AD (c).

MARK Since the interaction between tau and microtubules is regulated by phosphorylation, a great deal of research in the field has been devoted to the identification of kinases that phosphorylate tau. This problem is complicated by the fact that tau is a natively unfolded protein with a large number of phosphorylatable residues (mostly Ser or Thr)

which can be targets of many kinases [for reviews, see 35, 36]. It is therefore difficult to judge which combination of kinases and phosphorylation sites is responsible for the critical changes in tau activity in neurons. We have sought to simplify the issue by concentrating on phosphorylation sites and kinases that have the most pronounced effect on the tau-microtubule interaction. A search for the important phosphorylation sites revealed the ‘KXGS’ motifs in the repeat domain of tau, containing serines 262, 293, 324, and 356 (fig. 1a). In fact, phosphorylation

Signaling from MARK to Tau

Neurodegenerative Dis 2006;3:207–217

209

of Ser262 alone already accounts for a large decrease in the affinity [37]. This finding prompted a search for the responsible kinases which resulted in the identification of the MARK family of kinases, MARK1–4 [2]. Members of this kinase are ubiquitous but enriched in the brain. Activation or expression of the kinase in cells indeed leads to the phosphorylation of tau at the KXGS motifs, detachment from microtubules, and microtubule breakdown (fig. 1b, 2b) [5]. This type of phosphorylation can be detected by specific phosphorylation-dependent antibodies such as 12E8 [38]. The biochemical analysis of phosphotau from AD brain shows that the KXGS motifs are indeed among the detectable phosphorylation sites [39], and the histological analysis shows that MARK is present in neurofibrillary tangles, and that the MARK-type phosphorylation of tau occurs early in the disease process [1, 40]. The identification of MARK as a tau kinase raised the question of its regulation. Judging by its kinase domain, MARK is part of the calcium/calmodulin-dependent protein kinase group of kinases in the human kinome [12]. The various MARK isoforms and splice variants appear to have multiple functions; in C. elegans and D. melanogaster its homologues (termed Par-1) [3, 41] are reponsible for embryonic polarity, and the establishment and maintenance of polarity may be one of their functions in humans as well [42]. As shown in figure 2, MARK is a large multidomain protein which suggests several modes of regulation. The initial identification of MARK2 revealed the presence of phosphorylated residues in the ‘activation loop’ of the kinase catalytic domain. This is a common feature of many kinases and strongly suggests that there must be upstream activating kinases [for reviews, see 6, 43]. MARKK We therefore embarked on a search for kinases that would phosphorylate the activation loop of MARK2 (as a representative of all MARKs since their catalytic domains are highly homologous, 198%). This search turned out to be elusive because the activating kinase is embedded in a 330-kDa protein complex that is easily disrupted during purification. However, the kinase (termed MARKK for MARK-activating kinase) turned out to be a member of the GCK-VIII subfamily of the Ste20 family within the STE group of kinases [32] (fig. 3). This, too, is a large multidomain kinase, suggestive of multiple pathways of regulation. It is highly related to the kinase TAO-1 which is involved in the MKK3/6-p38-signaling cascade [17, 44]. Other members of this subfamily include 210

Neurodegenerative Dis 2006;3:207–217

PSK/TAO2 and JIK which have similar kinase domains and spacers (99 and 97% homology) but differ in their other domains [14, 15, 45]. An intriguing aspect was the possibility of dual regulation at the activation loop of MARK. With most kinases, activation by phosphorylation requires only one phosphorylated residue, but MARK2 peptides isolated from brain were phosphorylated at two sites, Thr208 and Ser212. These positions are reminiscent of the two sites required for the activity of the MAP kinase family. A further search showed that this analogy does not hold for MARK. In this case, only one site in the activation loop (Thr208) is phosphorylated by MARKK and leads to activation. The other site must remain unphosphorylated for activity but may possibly be phosphorylated by some as yet unknown inhibitory kinase. Mutations of Ser212 into Ala or Glu both lead to an inactive kinase. An important question about the proposed MARKKMARK-tau cascade was whether it is operational in cells, and whether it is specific. This was demonstrated by testing neuronal and non-neuronal cell models with constitutively active or inactive variants of the kinases. Overexpression of MARKK in CHOwt cells leads to microtubule breakdown and shrinkage of the cells (fig. 3b 1–3). This phenotype is similar to that of overexpressed MARK itself (fig. 2b) and can be suppressed by stabilization of the microtubules by taxol (fig. 3b 4–6) or overexpression of tau. However, when MARKK and MARK are expressed together, the stabilizing effect of the additional tau can be overcome (fig. 3c). The results showed that indeed the activation of MARKK leads to the activation of MARK and hence to the phosphorylation of tau at the KXGS motifs. But the KXGS motifs in tau could not be phosphorylated when MARK was inactivated (even when MARKK was active), and consequently no destabilization of microtubules was observed [5, 29]. The function of the cascade is important for differentiation and neurite outgrowth because this requires dynamic microtubules. Conversely, if one inactivates the cascade, for example by transfecting with a kinase-dead mutant of MARK, by changing the KXGS into KXGA motifs in tau, or by silencing MARKK with siRNA, then differentiation is no longer possible. This is demonstrated in figure 3d where PC12 cells differentiate readily upon nerve growth factor (NGF) treatment; however, when MARKK is knocked down by siRNA the cells can no longer differentiate [5, 29]. The above results corroborate a kinase cascade pathway that leads to the phosphorylation of tau in the repeat domain, but this is not sufficient to explain why the cascade might be overactive in AD. In particular, we note Timm/Matenia/Li/Griesshaber/ Mandelkow

Fig. 2. Diagram of MARK/PAR-1 kinases and their effects in cells. a The human genome contains 4 MARK genes located on chro-

mosomes 1, 11, 14, and 19. Additional variants are generated by alternative splicing [50, 51]. Sequence analysis reveals 5 basic domains: header, catalytic, UBA, spacer, and tail (also known as KA domain). The kinase is closely related to Par-1 kinases which play a role in the development of embryonic polarity in C. elegans and D. melanogaster [3]. All MARKs can be phosphorylated by MARKK at a conserved threonine in the catalytic domain (corresponding to T208 in MARK2), leading to activation [29]. The

catalytic domain can bind to PAK5, leading to inhibition [31]. Domains known by structural analysis: catalytic and UBA domain by X-ray crystallography [7], KA domain by NMR spectroscopy [52]. b CHO cells expressing the different MARK isoforms MARK1, MARK2, MARK3, and MARK4. The cells were transiently transfected with each of the four MARK isoforms (arrows), leading to the phosphorylation of endogenous MAPs (MAP4) and breakdown of the microtubule network (shown in green, visible best in 8). The severity of this effect is dependent on the expression level of the MARKs.

Fig. 4. PAK5. a Diagram of PAK5. PAK5 is a member of the Ste20 family of kinases and belongs to the subfamily of PAKs, group II (comprising PAK4–6). It contains the domains header, P = PBD (including a variant of the CRIB motif – but note that this is a matter of debate in the case of PAK5), auto-inhibitory domain (AID), spacer, and catalytic domain. The autophosphorylation site at S602 in the activation loop is crucial for activity [25, 27]. b Colocalization of YFP-PAK5wt (green) with transfected CFP-MARK2 (red). 1–3 Cotransfection of PAK5wt and active MARK2(T208E) shows colocalization of both kinases on vesicles and a diffuse background of MARK2(T208E).

Signaling from MARK to Tau

Neurodegenerative Dis 2006;3:207–217

211

Fig. 3. MARKK/TAO-1. a Diagram of MARKK/TAO-1. MARKK is a member of the GCK-VIII subfamily of the Ste20 kinases. Related members of this subfamily include TAO-1, TAO-2, and JIK [14, 15, 17, 45]. These kinases contain an N-terminal header domain, catalytic domain, substrate-binding domain (SBD), spacer domain, and tail domain. Extended coiledcoil sequences are predicted in the spacer and tail domains (S430-A630, K730F900). The substrate-binding domain of the related PSK/TAO-2 binds to the N-terminal header domain of its substrate MKK3, the kinase upstream of p38 MAP kinase. However, it is unclear whether an analogous interaction occurs with MARKK and MARK since the corresponding domains are poorly conserved. The MARKK gene is located on chromosome 17. MARKK is also a multidomain kinase containing the domains header, catalytic, substrate binding, spacer and tail. The 134-kDa protein was identified as the catalytically active component of a 330-kDa complex that activates MARK by phosphorylating the catalytic loop. MARKK itself can be activated by autophosphorylation in the catalytic loop (residue S181 in the MARKK diagram). b Effect of MARKK on CHO cells. 1–6 CHO cell transfected with YFP-MARKK. 2 The transfected cell (arrow) lost its microtubule network, rounded up and appears smaller. When a cell is transfected with YFP-MARKK (4, arrow) and at the same time microtubules are stabilized by 10 M taxol (5), then the breakdown of microtubules by MARKK is prevented. c CHO cells stably transfected with tau (htau40) and cotransfected with YFP-MARKK (1, arrow) and CFP-MARK (2) or staining of microtubules with the antibody YL1/2 (3). Note that in spite of the increased level of tau (which would normally stabilize microtubules), the activation of the MARKK-MARK cascade leads to the destruction of microtubules in the cotransfected cell, similar to b 1–3. d A1–B2 PC12 cells differentiated with NGF (48 h, 100 ng/ml). A1 Mock-transfected cell, stained for endogenous MARKK. A2 Phase image. B1 PC12 cells transfected with RNAi against MARKK exposed to NGF as above. B2 Phase image. MARKK is largely suppressed and the cells cannot differentiate. C1–3 PC12 cells differentiated with NGF (72 h, 100 ng/ml) and immunostained for MARKK (1, antibody TAO-1), MARK (2 , antibody SA2118), and merged image (3). Note that endogenous MARKK and endogenous MARK colocalize, notably beneath the plasma membrane and at growth cones.

212

Neurodegenerative Dis 2006;3:207–217

Timm/Matenia/Li/Griesshaber/ Mandelkow

that MARKK itself must also be activated by phosphorylation in the activation loop, indicating that there must be further upstream kinases. In addition, it is not known which physiological or pathological signals trigger the cascade. The MARK-type phosphorylation of tau is elevated in fetal tau, and indeed MARK is elevated in fetal brain tissue [2, 18, 38]. A possible rationale is that at this stage, prior to massive neuronal outgrowth, tau must be present but only loosely bound to microtubules, ready to stabilize microtubules once the neurites grow out. In this function, the phosphorylation of tau would be in concert with the splicing pattern which keeps the smallest tau isoform (0N3R) predominant before the stage of neurite outgrowth because this isoform has only three repeats and shows the weakest binding and stabilization of microtubules [46]. Conversely, microtubules in adult brain would be stabilized by the higher molecular weight splice isoforms of 4-repeat tau, and by shutting down the activity of MARK. PAK5 The activation cascade described above concerns phosphorylation in the strict sense; however, considering the multidomain structure of the interacting components we take the possibility of other types of regulation of MARK into account which would affect the phosphorylation of tau. Thus, yeast two-hybrid analysis was used to search for interaction partners of MARK. We took the full sequence of MARK2 and screened a human fetal cDNA library. Inactivating mutations (T208A/S212A) were introduced because this was known to stabilize the interaction with partners. The search revealed a number of positive clones coding for isoforms of the 14-3-3 protein, and clones coding for PAK5. This kinase was particularly interesting since it is a member of the mammalian PAK II subfamily and is predominantly expressed in the brain [25, 26]. As in the previous cases, PAK5 is a large multidomain kinase (fig. 4). Its catalytic domain is located not near the N-terminus but in the C-terminus and is preceded by a GTPase-binding domain (PBD). A further interaction assay with different truncation mutants of MARK2 and PAK5 showed that the interaction between MARK2 and PAK5 was based on the catalytic domains. This initially suggested some regulation by phosphorylation. However, surprisingly, PAK5 interacted with MARK2 independently of its state of phosphorylation. The subsequent analysis, using a tau-derived peptide to assess the activity of MARK2, showed that the binding of PAK5 alone is sufficient to inhibit MARK2.

Signaling from MARK to Tau

Cotransfection of PAK5 and MARK2 into CHO cells followed by analysis with fluorescence microscopy or immunofluorescence showed that MARK2 and PAK5 colocalize and have a vesicular distribution. To identify the nature of the vesicles carrying PAK5, we labeled the cells with different markers. Partial colocalization of transfected PAK5 and MARK2 was observed with vesicles carrying adaptor protein (AP) complexes of the AP1/2 family. More specifically, in a subcellular fractionation experiment the vesicles carrying MARK2 and PAK5 were identified as part of the trans-Golgi network. PAK5 and MARK2 occurred in the same fractions as 1/2- and adaptin, the subunits of the AP1 complex. Immunofluorescence experiments using CHO cells show that active or inactive PAK5 eliminates the effect of MARK2 on the cytoskeleton (i.e. the destabilization of microtubules; fig. 5). Figure 5b shows that if constitutively active PAK5 (S573N/S602E) is coexpressed with constitutively active MARK2(T208E) the microtubule network is protected (fig. 5b 1–3) while actin stress fibers and focal adhesions are dissolved, which correlates with the emergence of filopodia (fig. 5b 4–6). A similar effect of MARK2 inhibition and microtubule preservation is obtained by coexpressing inactive PAK5 (S602M/T606M) with active MARK2(T208E). However, in this case the actin stress fibers are not rendered dynamic, and consequently focal adhesions are preserved and filopodia do not evolve (data not shown). This experiment shows that active PAK5(S573N/S602E) has two independent effects on the cytoskeleton. First, it stabilizes microtubules by binding and inhibiting MARK2, and second, it makes actin dynamic by dissolving stress fibers and focal adhesions and inducing the formation of filopodia. Inactive PAK5(S602M/T606M) only shows the first effect (on microtubules), but the second effect is absent because it would require PAK5 activity. To check the inhibition of MARK2 in cells, the ability of PAK5 to suppress the phosphorylation of tau by MARK2 was tested using htau40 stably transfected CHO cells (fig. 5c). The data clearly confirm that the PAK5 acts as a MARK2 inhibitor. Outlook: Modes of MARK Regulation The kinase MARK is regulated by several different mechanisms (fig. 6). First, the activity of the kinase is increased through phosphorylation by MARKK at the activation loop. This is a common regulatory mechanism for serine/threonine kinases [6, 43]. Furthermore, the activity is modulated by interaction with other proteins. The Ste20-kinase PAK5 can bind to the catalytic domain of MARK, which results in inhibition [31]. MARK can Neurodegenerative Dis 2006;3:207–217

213

5

214

Neurodegenerative Dis 2006;3:207–217

Timm/Matenia/Li/Griesshaber/ Mandelkow

Fig. 5. Crosstalk between MARK and

PAK5 in the regulation of the cytoskeleton. The actin and microtubule cytoskeleton networks in the cell often fulfill complementary functions. Each system is regulated by their set of associated proteins, posttranslational modifications (e.g. by kinases), severing or crosslinking proteins, but there is also a crosstalk which often results in complementary outcomes: When microtubules become unstable, actin fibers become stable (e.g. stress fibers), and vice versa. This can be achieved, for example, by microtubule end-binding proteins which sense the MT dynamics and signal (via GTP-exchange factors or GTPase-activating proteins) to small G-proteins (Rho, Rac) that regulate the actin network [for review, see 53]. A related principle is illustrated here. The activity of MARK tends to make microtubules labile and dynamic which correlates with an increased stability of the actin network. However, MARK can be inhibited by PAK5 (whose normal role is to make the actin network more dynamic), hence microtubules are stabillized. a Effects of PAK5 and MARK2 on the stability of microtubules and actin filament networks. CHO cells transfected with YFP-PAK5 and YFP-MARK2 (green) were cultured for 16 h, fixed, and costained with the YL1/2 antibody for tubulin and fluorescently labeled (TRITC) secondary antibody (MT staining, red). Actin was stained using rhodamine-conjugated phalloidin (red). Transfected cells are indicated by arrows. In cells expressing wild-type YFP-MARK2 (1, 3), microtubules disappear (2 , arrow) and actin stress fibers are stabilized (4). In contrast, constitutively active PAK5 (5–8) stabilizes MT (6), but the actin stress fibers are dissolved (8). b PAK5 inhibits the MARK2 effect on the microtubule and actin networks. CHO cells coexpressing the constitutively active form of PAK5 (1, 4, yellow) and MARK2 (2 , 5, cyan) show a stabilized microtubule network (3, green) and a dynamic actin cytoskeleton discernible by loss of actin stress fibers (6, red). c Cells stably transfected with tau and transiently transfected with CFP-MARK alone (1, 2) and cotransfected with YFPPAK5 and CFP-MARK (3–5). Expression of MARK (1, cyan) leads to phosphorylation of overexpressed tau in the KXGS motifs visible with the 12E8 antibody (2 , red). Coexpression of inactive

PAK5 (3, yellow) and active MARK2 (4, cyan) results in an inhibition of MARK2 as seen by the low level of phospho-KXGS tau (5, red). Transfected cells labeled by arrows. Fig. 6. Modes of regulating MARK. a The diagram summarizes known or plausible modes of MARK regulation which would affect the phosphorylation of tau and hence the stability of microtubules and the aggregation of tau. (a) Activation via phosphorylation by MARKK at the activation loop (T208) [29]. (b) Inhibition by binding of PAK5 to the catalytic domain [31]. (c) Regulation by interaction of the UBA domain with ubiquitin (not proven, but suggested by X-ray structure) [7]. (d) Regulation by interaction of the CD motif with a cofactor, in analogy with MAP kinases where upstream or downstream kinases can be bound [7, 54]. (e) Localization by interaction of the catalytic domain with the AP 14-3-3, in analogy with Drosophila Par-1 [8, 29]. This interaction does not depend on prior phosphorylation of MARK. (f) Localization and probably inhibition by interaction of the spacer domain with 14-3-3, after prior phosphorylation by aPKC which creates a 14-3-3 binding motif on MARK [9, 10]. (g) Interaction between the C-terminal tail and the N-terminal header or catalytic domain (dotted line), creating a folded and inhibited MARK molecule (proposed for the yeast homolog Kin-1) [11]. b Summary of the antagonistic regulation of MARK by MARKK and PAK5 and the effects on the cytoskeleton. MARK is switched on by an upstream kinase, MARKK, and this leads to increased dynamics of microtubules and stabilization of the actin cytoskeleton. However, switching MARK off can be achieved by PAK5, resulting in increased actin dynamics but stabilization of microtubules.

Signaling from MARK to Tau

Neurodegenerative Dis 2006;3:207–217

6

215

also interact with 14-3-3. This can happen in two different modes: 14-3-3 can bind in a nonphosphorylation-dependent manner to the N-terminal half of MARK [8], it can also bind to the spacer domain after this has been phosphorylated by atypical PKC [9, 10]. These interactions do not only regulate MARK spatially by altering its localization but also inhibit the catalytic activity of the enzyme, probably by stabilizing the inhibitory interaction of the KA domain with the N-terminal header or the catalytic domain [11 and unpublished data]. From the structural analysis of MARK [7], one can speculate about regulatory interactions of (yet unknown) proteins with the UBA domain and the CD motif. This motif is known in MAP kinases for multiple interactions with upstream and downstream effectors [54], and the CD motif can be found in all MARK family members [7]. Kinases are often involved in more than one signaling cascade. For example, MARK/Par-1 signaling is involved not only in regulating microtubule dynamics (e.g. in neurite outgrowth) and the Par cell polarity determinants but also in Wnt signaling [47]. Furthermore, the upstream kinase MARKK/TAO-1 activates the p38 stress

pathway by phosphorylating MKK3 and MKK6 [17]. On the other hand, PAK5 is not only involved in remodeling the actin cytoskeleton, but also in inducing the JNK stress pathway [25, 26] and in the apoptotic pathway [48]. It is interesting to note that the two kinases whose interaction we have studied in the context of the cytoskeleton also have a relationship to the cell’s stress response. Consistent with this, the activation of MARK and the phosphorylation of its downstream target tau is elevated by cellular stress [18, 49]. This might explain the increased phosphorylation of tau at early stages of neurodegeneration in AD and frontotemporal dementias [1]. The impact of PAK5 on MARK during neurodegeneration will be an interesting question to pursue. Acknowledgements We thank Anja Thiessen for expert technical assistance, and Cindy Johne, Jian Jiao and Jacek Biernat for helpful discussion. This project was supported by the Deutsche Forschungsgemeinschaft.

References 1 Augustinack JC, Schneider A, Mandelkow EM, Hyman B: Specific tau phosphorylation sites correlate with severity of neuronal cytopathology in Alzheimer’s disease. Acta Neuropathol 2002;103:26–35. 2 Drewes G, Ebneth A, Preuss U, Mandelkow E-M, Mandelkow E: MARK, a novel family of protein kinases that phosphorylate microtubule-associated proteins and trigger microtubule disruption. Cell 1997; 89:297–308. 3 Pellettieri J, Seydoux G: Anterior-posterior polarity in C. elegans and Drosophila – PARallels and differences. Science 2002; 298: 1946–1950. 4 Macara IG: Par proteins: partners in polarization. Curr Biol 2004:14;R160–R162. 5 Biernat J, Wu YZ, Timm T, Zheng-Fischhofer Q, Mandelkow E, Meijer L, Mandelkow EM: Protein kinase MARK/par-1 is required for neurite outgrowth and establishment of neuronal polarity. Mol Biol Cell 2002; 13: 4013– 4028. 6 Huse M, Kuriyan J: The conformational plasticity of protein kinases. Cell 2002; 109: 275–282. 7 Panneerselvam S, Marx A, Mandelkow EM, Mandelkow E: Structure of the catalytic and ubiquitin-associated domains of the protein kinase MARK/Par1. Structure 2006;14:173– 183.

216

8 Benton R, Palacios IM, St. Johnston D: Drosophila 14-3-3/PAR-5 is an essential mediator of PAR-1 function in axis formation. Dev Cell 2002;3:659–671. 9 Hurov J, Watkins J, Piwnica-Worms H: Atypical PKC phosphorylates PAR-1 kinases to regulate localization and activity. Curr Biol 2004;14:736–741. 10 Suzuki A, Hirata M, Kamimura K, Maniwa R, Yamanaka T, Mizuno K, Kishikawa M, Hirose H, Amano Y, Izumi N, Miwa Y, Ohno S: aPKC acts upstream of PAR-1b in both the establishment and maintenance of mammalian epithelial polarity. Curr Biol 2004; 14: 1425–1435. 11 Elbert M, Rossi G, Brennwald P: The yeast Par-1 homologs Kin1 and Kin2 show genetic and physical interactions with components of the exocytic machinery. Mol Biol Cell 2005;16:532–549. 12 Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S: The protein kinase complement of the human genome. Science 2002; 298:1912–1934. 13 Dan I, Watanabe NM, Kusumi A: The Ste20 group kinases as regulators of MAP kinase cascades. Trends Cell Biol 2001; 11:220–230.

Neurodegenerative Dis 2006;3:207–217

14 Chen Z, Hutchison M, Cobb MH: Isolation of the protein kinase TAO2 and identification of its mitogen-activated protein kinase/ extracellular signal-regulated kinase kinase binding domain. J Biol Chem 1999; 274: 28803–28807. 15 Tassi E, Biesova Z, Di Fiore PP, Gutkind JS, Wong WT: Human JIK, a novel member of the STE20 kinase family that inhibits JNK and is negatively regulated by epidermal growth factor. J Biol Chem 1999; 274:33287– 33295. 16 Chen Z, Raman M, Chen L, Lee SF, Gilman AG, Cobb MH: TAO (thousand-and-one amino acid) protein kinases mediate signaling from carbachol to p38 mitogen-activated protein kinase and ternary complex factors. J Biol Chem 2003; 278:22278–22283. 17 Hutchison M, Berman KS, Cobb MH: Isolation of TAO1, a protein kinase that activates MEKs in stress-activated protein kinase cascades. J Biol Chem 1998; 273:28625–28632. 18 Jenkins SM, Johnson GV: Microtubule/ MAP-affinity regulating kinase (MARK) is activated by phenylarsine oxide in situ and phosphorylates tau within its microtubulebinding domain. J Neurochem 2000; 74: 1463–1468.

Timm/Matenia/Li/Griesshaber/ Mandelkow

19 Brown AJ, Hutchings C, Burke JF, Mayne LV: Application of a rapid method (targeted display) for the identification of differentially expressed mRNAs following NGF-induced neuronal differentiation in PC12 cells. Mol Cell Neurosci 1999;13:119–130. 20 Manser E, Leung T, Salihuddin H, Zhao ZS, Lim L: A brain serine/threonine protein kinase activated by Cdc42 and Rac1. Nature 1994;367:40–46. 21 Etienne-Manneville S, Hall A: Rho GTPases in cell biology. Nature 2002;420:629–635. 22 Jaffer ZM, Chernoff J: p21-activated kinases: three more join the Pak. Int J Biochem Cell Biol 2002;34:713–717. 23 Bokoch GM: Biology of the p21-activated kinases. Annu Rev Biochem 2003;72:743–781. 24 Hofmann C, Shepelev M, Chernoff J: The genetics of Pak. J Cell Sci 2004; 117: 4343– 4354. 25 Dan C, Nath N, Liberto M, Minden A: PAK5, a new brain-specific kinase, promotes neurite outgrowth in N1E-115 cells. Mol Cell Biol 2002;22:567–577. 26 Pandey A, Dan I, Kristiansen TZ, Watanabe NM, Voldby J, Kajikawa E, Khosravi-Far R, Blagoev B, Mann M: Cloning and characterization of PAK5, a novel member of mammalian p21-activated kinase-II subfamily that is predominantly expressed in brain. Oncogene 2002;30:3939–3948. 27 Ching YP, Leong VY, Wong CM, Kung HF: Identification of an autoinhibitory domain of p21-activated protein kinase 5. J Biol Chem 2003;278:33621–33624. 28 Bryan B, Kumar V, Stafford LJ, Cai Y, Wu G, Liu M: GEFT, a rho family guanine nucleotide exchange factor, regulates neurite outgrowth and dendritic spine formation. J Biol Chem 2004;279:45824–45832. 29 Timm T, Li XY, Biernat J, Jiao J, Mandelkow E, Vandekerckhove J, Mandelkow EM: MARKK, a Ste20-like kinase, activates the polarity-inducing kinase MARK/PAR-1. EMBO J 2003;22:5090–5101. 30 Mandelkow EM, Thies E, Trinczek B, Biernat J, Mandelkow E: MARK/PAR1 kinase is a regulator of microtubule-dependent transport in axons. J Cell Biol 2004; 167:99–110. 31 Matenia D, Griesshaber B, Li XY, Thiessen A, Johne C, Jiao J, Mandelkow E, Mandelkow EM: PAK5 kinase is an inhibitor of MARK/ Par-1, which leads to stable microtubules and dynamic actin. Mol Biol Cell 2005;16:4410– 4422.

Signaling from MARK to Tau

32 Braak H, Braak E: Demonstration of amyloid deposits and neurofibrillary changes in whole brain sections. Brain Pathol 1991; 1: 213–216. 33 Lee VM, Goedert M, Trojanowski JQ: Neurodegenerative tauopathies. Annu Rev Neurosci 2001;24:1121–1159. 34 Stamer K, Vogel R, Thies E, Mandelkow E, Mandelkow EM: Tau blocks traffic of organelles, neurofilaments, and APP vesicles in neurons and enhances oxidative stress. J Cell Biol 2002;156:1051–1063. 35 Mandelkow EM, Mandelkow E: Tau in Alzheimer’s disease. Trends Cell Biol 1998; 8: 425–427. 36 von Bergen M, Li L, Mandelkow E: Intrinsic fluorescent detection of tau conformation and aggregation. Methods Mol Biol 2005; 299:175–184. 37 Biernat J, Gustke N, Drewes G, Mandelkow EM, Mandelkow E: Phosphorylation of Ser262 strongly reduces binding of tau to microtubules: distinction between PHF-like immunoreactivity and microtubule binding. Neuron 1993;11:153–163. 38 Seubert P, Mawaldewan M, Barbour R, Jakes R, Goedert M, Johnson GVW, Litersky J, Schenk D, Lieberburg I, Trojanowski J, Lee VMY: Detection of phosphorylated Ser(262) in fetal tau, adult tau, and paired helical filament tau. J Biol Chem 1995; 270: 18917– 18922. 39 Morishima-Kawashima M, Hasegawa M, Takio K, Suzuki M, Yoshida H, Titani K, Ihara Y: Proline-directed and non-prolinedirected phosphorylation of PHF-tau. J Biol Chem 1995;270:823–829. 40 Chin JY, Knowles RB, Schneider A, Drewes G, Mandelkow EM, Hyman BT: Microtubule-affinity regulating kinase (MARK) is tightly associated with neurofibrillary tangles in Alzheimer brain: a fluorescence resonance energy transfer study. J Neuropathol Exp Neurol 2000;59:966–971. 41 Kemphues K: PARsing embryonic polarity. Cell 2000;101:345–348. 42 Cohen D, Brennwald PJ, Rodriguez-Boulan E, Müsch A: Mammalian PAR-1 determines epithelial lumen polarity by organizing the microtubule cytoskeleton. J Cell Biol 2004; 164:717–727. 43 Adams JA: Activation loop phosphorylation and catalysis in protein kinases: is there functional evidence for the autoinhibitor model? Biochemistry 2003; 42:601–607. 44 Raman M, Cobb MH: MAP kinase modules: many roads home. Curr Biol 2003;13:R886– R888.

45 Moore TM, Garg R, Johnson C, Coptcoat MJ, Ridley AJ, Morris JD: PSK, a novel STE20like kinase derived from prostatic carcinoma that activates the c-Jun N-terminal kinase mitogen-activated protein kinase pathway and regulates actin cytoskeletal organization. J Biol Chem 2000; 275:4311–4322. 46 Kosik KS, Orecchio LD, Bakalis S, Neve RL: Developmentally regulated expression of specific tau sequences. Neuron 1989; 2: 1389–1397. 47 Sun TQ, Lu B, Feng JJ, Reinhard C, Jan YN, Fantl WJ, Williams LT: PAR-1 is a Dishevelled-associated kinase and a positive regulator of Wnt signalling. Nat Cell Biol 2001;3: 628–636. 48 Cotteret S, Jaffer ZM, Beeser A, Chernoff J: p21-Activated kinase 5 (Pak5) localizes to mitochondria and inhibits apoptosis by phosphorylating BAD. Mol Cell Biol 2003; 23:5526–5539. 49 Schneider A, Laage R, von Ahsen O, Fischer A, Rossner M, Scheek S, Grunewald S, Kuner R, Weber D, Kruger C, Klaussner B, Gotz B, Hiemisch H, Newrzella D, Martin-Villalba A, Bach A, Schwaninger M: Identification of regulated genes during permanent focal cerebral ischaemia: characterization of the protein kinase 9b5/MARKL1/MARK4. J Neurochem 2004;88:1114–1126. 50 Hueso M, Beltran V, Moreso F, Ciriero E, Fulladosa X, Grinyo JM, Seron D, Navarro E: Splicing alterations in human renal allografts: detection of a new splice variant of protein kinase Par1/Emk1 whose expression is associated with an increase of inflammation in protocol biopsies of transplanted patients. Biochim Biophys Acta 2004; 24: 58– 65. 51 Kato T, Satoh S, Okabe H, Kitahara O, Ono K, Kihara C, Tanaka T, Tsunoda T, Yamaoka Y, Nakamura Y, Furukawa Y: Isolation of a novel human gene, MARKL1, homologous to MARK3 and its involvement in hepatocellular carcinogenesis. Neoplasia 2001; 3:4–9. 52 Tochio N, Koshiba S, Inoue M, Kigawa T, Yokoyama S: Solution structure of kinase associated domain 1 of mouse MAP/microtubule affinity-regulating kinase 3. http://www. pdb.org 2003;PDB 1v5s. 53 Small JV, Kaverina I: Microtubules meet substrate adhesions to arrange cell polarity. Curr Opin Cell Biol 2003; 15:40–47. 54 Tanoue T, Nishida E: Molecular recognitions in the MAP kinase cascades. Cell Signal 2003;15:455–462.

Neurodegenerative Dis 2006;3:207–217

217

Original Paper

Diseases

Neurodegenerative Dis 2006;3:218–226 DOI: 10.1159/000095259

Subcellular Trafficking of the Amyloid Precursor Protein Gene Family and Its Pathogenic Role in Alzheimer’s Disease Stefan Kins Nadine Lauther Anita Szodorai Konrad Beyreuther Zentrum für Molekulare Biologie der Universität Heidelberg, Heidelberg, Germany

Key Words Axonal transport Kinesin c-Jun N-terminal kinase-interacting protein Cargo receptor

ronal viability and function. Thus, changes in APP and tau expression may cause perturbed axonal transport and changes in APP processing, contributing to cognitive decline and neurodegeneration in Alzheimer’s disease. Copyright © 2006 S. Karger AG, Basel

Abstract Changes in the intracellular transport of amyloid precursor protein (APP) affect the extent to which APP is exposed to - or -secretase in a common subcellular compartment and therefore directly influence the degree to which APP undergoes the amyloidogenic pathway leading to generation of -amyloid. As the presynaptic regions of neurons are thought to be the main source of -amyloid in the brain, attention has been focused on axonal APP trafficking. APP is transported along axons by a fast, kinesin-dependent anterograde transport mechanism. Despite the wealth of in vivo and in vitro data that have accumulated regarding the connection of APP to kinesin transport, it is not yet clear if APP is coupled to its specific motor protein via an intracellular interaction partner, such as the c-Jun N-terminal kinaseinteracting protein, or by yet another unknown molecular mechanism. The cargo proteins that form a functional complex with APP are also unknown. Due to the long lifespan, and vast extent, of neurons, in particular axons, neurons are highly sensitive to changes in subcellular transport. Recent in vitro and in vivo studies have shown that variations in APP or tau affect mitochondrial and synaptic vesicle transport. Further, it was shown that this axonal dysfunction might lead to impaired synaptic plasticity, which is crucial for neu-

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0218$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Subcellular Trafficking of APP/APLPs

The amyloid precursor protein (APP), similar to other typical type I transmembrane proteins, is cotranslationally translocated during migration across the ER membrane and remains anchored there by a single membrane spanning -helical region. APP is inserted in the membrane such that only a short tail remains in the cytosol, whereas the major N-terminal forms a relatively large extracellular domain. APP is then transported from the ER to the Golgi apparatus, whereby it undergoes different posttranslational covalent modifications, including Nand O-glycosylation, sialylation, and modification with chondroitin sulfate glycosaminoglycan and/or dermatan sulfate glycosaminoglycan [1, 2]. During passage through the different subcompartments of the Golgi apparatus, APP is sorted into secretory vesicles. On its way from the Golgi to the plasma membrane and to endosomes, a portion of APP is processed by different secretases. The extracellular domain of APP can be cleaved by -secretase or, alternatively, by the -secretase BACE 1 (-site APP cleaving enzyme 1). The resulting membrane-retained

Dr. Stefan Kins Zentrum für Molekulare Biologie der Universität Heidelberg (ZMBH) Im Neuenheimer Feld 282 DE–69120 Heidelberg (Germany) Tel. +49 6221 546 847, Fax +49 6221 545 891, E-Mail [email protected]

C-terminal fragments are subsequently processed by cleavage within the transmembrane region by the -secretase complex [3]. - and -cleavage of APP results in the release of amyloid- (A) peptides, while - and -cleavage generate p3 fragments. Besides the Golgi apparatus, the plasma membrane was shown to be one of the two major sites where -secretase cleavage of APP takes place [4, 5]. In contrast, BACE1 is most active at mildly acidic pH [6], making endosomes the most likely location for BACE cleavage of APP. Moreover, BACE 1 was reported to cycle between the cell surface and endosomes [7]. It was also shown that internalization of cell surface APP via endocytosis leads to elevated A generation and subsequent release into the medium [8]. Thus, changes in subcellular APP trafficking affect the time that APP spends together with - or -secretase in a common subcellular compartment and, thereby, the APP processing. In particular, neurons, which have very long processes (the axon of human motoneurons can be up to 1 m in length), are dependent on an appropriate subcellular localization of transcripts and proteins which presumes a highly complex transport system. Due to their morphology and long life span, neurons are highly vulnerable to any perturbation of their transport machinery. There is emerging evidence that impaired axonal transport is causal for different neurodegenerative disorders, including Alzheimer’s disease (AD). APP is anterogradely transported in tubular vesicles along the axons [9–15] with a velocity up to 10 m per second [16, 17]. These transport vesicles differ from axonal vesicles transporting neurotransmitters towards the synapse [18] or from those much slower vesicles, transporting synaptophysin along the axon [17]. The anterograde transport of APP depends on kinesin-1 [2, 9, 17, 19–22], which consists of two kinesin heavy and light chains (KHC, KLC). Motor activity and specificity of cargo binding are localized in the KHC [23], whereas the KLC is thought to regulate the motor activity and bind via its TPR-like motif to vesicular receptor and linker proteins. In mammals, there are three different KHCs: kinesin-1A, -1B and -1C (KIF5A, B, C) [24–28], and three KLCs [29], KLC1, KLC2 and KLC3. Kinesin-1A, kinesin1C and KLC1 are specific to neurons, whereas kinesin-1B and KLC2 are expressed ubiquitously [24, 25]. The KLC3 expression in the brain is very low, but it plays a unique role in the spermatids in testis [30]. Different kinesin-1 tetrameric heterocomplexes consisting of two KHCs and two KLCs may participate in selective transport by using adapter or scaffolding proteins to recognize and bind specific cargoes [31].

It was proposed that APP may play a role as a kinesin cargo receptor [32], directly interacting with the motor protein kinesin-1, connecting it to a certain axonal vesicle class with specific cargo proteins, including the APP secretases, BACE and presinilin (PS) [33]. A production was observed in sciatic nerve extracts, indicating that APP may be processed in axonal transport vesicles [33]. The hypothesis that the APP intracellular domain (AICD) interacts directly with kinesin-1 was based on GST pull-down analyses, showing high affinity binding (KD = 16–18 nM) of GFP-AICD with GST-KLC1 and GSTKLC2 fusion proteins. More detailed GST pull-down experiments revealed that this binding was caused by a nonspecific hydrophobic interaction to the tandem repeats in the carboxy terminus of bacterially expressed KLC1 [34]. Consistent with this, GST pull-down analyses with GSTAICD and recombinant KLC1 expressed in mammalian cells revealed no specific interaction, whereas other known interaction partners of APP, such as Fe65 or Numb, were specifically retained on beads loaded with GST-AICD. Together, these data strongly suggest that KLC1 does not interact directly with the cytoplasmic tail of APP. Nonetheless, KLC1 and KLC2 appear to be associated with different subsets of APP-containing membrane compartments [S.K., pers. obs.; 32]. Thus, an indirect interaction of APP with kinesin via a cytosolic APP interaction partner could be postulated. However, coimmunoprecipitation studies of APP and kinesin-1 from mouse brain lysates with any of a wide range of specific antibodies to different epitopes on KLC, KHC, and APP N- and C-terminus (kindly obtained from G. Multhaup, Berlin, and T. Hartmann, Heidelberg) [34 ; S.K., pers. commun.] revealed, in contrast to previous studies from Kamal et al. [32], no indication for a common complex of APP and kinesin. Importantly, other putative kinesin-cargo receptors, such as JIP3/JSAP1 (Sunday driver, Syd) [35], have been coimmunoprecipitated under identical conditions [34]. Nevertheless, based on the negative results from the in vitro binding studies and coimmunoprecipitation studies, an indirect association of APP to kinesin-1 via an AICD interacting protein cannot be excluded. Different linker molecules are known that connect a specific motor protein to a certain class of vesicles. For example, 1 adaptin was reported to couple KIF13A, a plus end-directed microtubule-dependent motor protein, to the AP-1 adaptor complex and mannnose-6-phosphate receptor [36]. Other linker proteins are scaffolding proteins that are capable of interacting with a set of different interaction partners and which often cluster functional complexes, such as PSD-95 [37], glutamate-receptor-in-

Pathogenic Role of APP Axonal Transport

Neurodegenerative Dis 2006;3:218–226

219

Table 1. Different linker molecules connect specific motor protein to certain class of vesicles

Motorprotein

Motorprotein new nomenclature [28]

Linker

Cargo

References

KHC

Kinesin-1

Syntabulin-

Syntaxin-1, mitochondria

[84, 85]

KIF5

Kinesin-1

GRIP1

AMPA receptor

[38, 40]

KIF5

Kinesin-1

JIP1,2,3 (MAPK8IP1, 2, 3)

JNK signaling cascade (DLK), ApoER2

[34, 35, 45–51]

KIF3A

Kinesin-2

–

Opsin, arrestin

[86]

KIF3, KAP

Kinesin-2

–

Chromosomes, flagellar cargoes

[87–90]

KIF13A

Kinesin-3

AP-1

clathrin, M6PR

[36]

KIF17

Kinesin-3

X11 (mLIN-10, MINT)

mLIN-2 (CASK), mLIN-7 (VELIS/ MALS), NMDAR (NR2B subunit)

[43, 44]

KIF1A

Kinesin-3A

Liprin-, (PPFIA2)

GRIP, AMPAR, RIM, GIT1, PIX, synaptic vesicles

[39, 41, 42, 91]

KIF1B

Kinesin-3B

PSD-95 (SAP90)

PSD-95, synaptic precursor vesicles

[37, 92]

PPFIA2 = Protein-tyrosine phosphatase, receptor-type, F polypeptide-interacting protein 2; PSD-95 = postsynaptic density 95; SAP90 = synapse-associated protein 90; GRIP1 = glutamate receptor-interacting protein 1; AP-1 = adaptor protein 1; mLIN-10 = vertebrate LIN10 homolog; MINT = MUNC18-1-interacting protein; NMDAR = N-methyl-D -aspartate receptor; CASK = calcium/ calmodulin-dependent serine protein kinase; AMPA = -amino-3-hydroxy-5-methyl-4-isoxazolpropionate; GRK = G-protein-coupled receptor kinase; GIT1 = GRK-interacting protein 1; PAK = p21-activated kinase and phospholipase C-interacting protein 1; PIX = PAK-interacting exchange factor; M6PR = mannnose-6-phosphate receptor.

teracting protein 1 [38–42], X11 (LINs, MINT) [43, 44], Syd, a c-Jun N-terminal kinase (JNK)-interacting protein (JIP) 3 homologue in Drosophila or JIP1b and JIP2 [34, 35, 45–51] (table 1). X11 and JIP1b/2 have a phosphotyrosine interaction/ binding domain [52–54] that interacts with the NPTY motif of AICD. Thus, they may couple APP-containing vesicles to a certain signaling pathway by creating a scaffold [52–54] and may link APP to a special type of kinesin transport. Considering that APP is driven with different velocities (2–10 m/s), the different linker proteins may couple APP to different kinesin transport complexes with different velocity characteristics. Each of these adapter proteins could represent the unknown linker protein for a specific subset of vesicles. Based on in vitro analyses it was proposed that JIP-1b might connect APP-bearing axonal vesicles to the kinesin-1 motor mechanism [46, 47, 50]. To test this proposal in vivo, we performed APP/JIP1b coimmunoprecipitation studies and analyzed the subcellular transport of APP in primary neurons treated with specific siRNA to knockdown JIP-1b/2. Under conditions that allowed specific coimmunoprecipitation of complexes containing APP and 220

Neurodegenerative Dis 2006;3:218–226

APLP1 or APLP2 [55], coimmunoprecipitation of APP and JIP-1b was not observed. However, these data do not exclude a role of JIP-1b as a linker of APP to the transport machinery, which might require only a transient, low affinity binding. To address the influence of JIP1b on APP trafficking more directly, we screened different siRNAs directed against JIP1b and found that in mixed cortical primary neurons (DIV11) a knockdown of endogenous JIP-1b to less than 5% of the endogenous level (fig. 1) was observed after siRNA treatment. Immunocytochemical analyses of wild-type neurons with anti-JIP1b antibodies revealed that endogenous JIP1b accumulated at the tips of the neurites (fig. 2a–c). In JIP-1b knockdown neurons, no specific JIP1b immunoreactivity was detected (fig. 2d–e). Surprisingly, costaining of JIP-1b siRNA-treated neurons with anti-APP antibodies revealed that APP was localized not just in the perinuclear region, but also at the tips of neurites, showing that APP can be anterogradely transported in the absence of JIP1b (fig. 2d, f). These data are consistent with the results from a recent study, showing that the majority of APP does not colocalize with JIP-1b and that a knockdown of JIP1b only has an influence on transport of the minor pool of APP phosphorylated at Kins/Lauther/Szodorai/Beyreuther

6h

Fig. 1. JIP-1b knockdown in primary neu-

rons. Mouse primary neurons (DIV11) were transfected with JIP-1b cDNA only or simultaneously treated with 40 n M siRNA against JIP1b. Expression of JIP1b was evaluated 6–24 h after transfection. The cells were lysed and subjected to PAGE and Western blot analyses using antiJIP-1b [1: 500] antibody. Anti--tubulin [1:10,000] antibody was used as a loading control. Efficient knockdown of JIP-1b occurs after 18 h up to 24 h.

JIP1b siRNA-JIP1b kDa 130

+ –

18 h + +

+ –

26 h + +

+ –

+ +

JIP1b

100

-Tubulin

55

threonin 668 [49]. Altogether, it is clear that APP does not bind directly to kinesin and that JIP1b is not the main linker that couples APP to the anterograde transport machinery (fig. 3). However, further analyses are necessary to determine the molecular link between APP and the kinesin motor mechanism. The molecular composition of APP/APLPs-containing vesicles also remains unclear. BACE and PS1 are transported along the axon [56, 57] and, as mentioned before, Kamal et al. [33] proposed that these two APP secretases are specific cargoes of APP-containing vesicles transported by the fast anterograde transport machinery. However, careful studies assessing the trafficking of membrane proteins in the sciatic nerves of mice with heterozygous or homozygous deletions of APP show that the axonal transport kinetics of BACE and PS1 are different from APP and that they are unchanged in APP knockout mice [34]. These data argue that the APP-processing apparatus is transported in TGN vesicles different from those that transport APP, suggesting axonal processing of APP by BACE and the -secretase complex takes place at the plasma membrane or endosomes. However, it is not clear whether -secretase might be cotransported with APP. Other proposed cargo proteins cotransported with APP have not been verified in subsequent studies [33, 34]. In recent studies from our laboratory, we determined that APP, APLP1 and APLP2 not only have very similar structures, but also form homo- and heterotypic complexes, suggesting a close functional relation between the APP gene family members [55]. This view is also supported by genetic analyses of the APP gene family [58, 59] and studies addressing the influence of APLP1 and

APLP2 on APP processing [60]. We could show in extensive coimmunoprecipitation studies that both mature and immature APP/APLPs can interact with similar efficiency [55]. This observation is consistent with studies showing in vitro dimerization of recombinantly expressed soluble APP [61]. However, we found that endogenous heterocomplexes contain exclusively mature APP/ APLPs. Moreover, these high molecular weight species of APP family interacting proteins strongly accumulate in synaptic plasma membrane fractions [55]. As only mature APP is present at the cell surface [62], we favor a model in which endogenous APP/APLP heterointeraction in the brain is limited to the cell surface, which would allow transcellular binding. APP and APLP2 have previously been shown to be transported to presynaptic terminals [1, 10] and growth cones of neurons [63], while APLP1 has been reported to localize to the postsynapse terminals [64]. Interestingly, all APP family members exhibit developmentally increased expression levels correlating with postembryonic synaptogenesis [13, 65, 66], and recent genetic analyses in mouse and Drosophila revealed that the APP gene family members are required for neuromuscular synaptogenesis [67, 68]. Together with our data, these findings indicate that transcellular APP/APLP interaction is part of the regulation of synaptogenesis. In regard to the different subcellular localizations of APP, APLP1 and APLP2 in neurons [1, 64], it is reasonable that APP, APLP1 and APLP2 are transported in different types of vesicles and are differently coupled to specific anterograde transport mechanisms. In principle, each subtype of transport vesicle is thought to mediate the axonal/dendritic transport of a specific set of functional protein complexes [69, 70]. Thus, it will be of

Pathogenic Role of APP Axonal Transport

Neurodegenerative Dis 2006;3:218–226

221

2

a

Postsynaptic sites

b

Axonal transport vesicles

(a ) (b)

1

2

APP

APP ?

(1)

(2)

(c )

(3)

JIP1B/X11

?

(4)

Presynaptic terminal

Kinesin-1

(5)

APP APLP1 APLP2

3

222

Neurodegenerative Dis 2006;3:218–226

Kins/Lauther/Szodorai/Beyreuther

great interest to elucidate the molecular composition of the specific transport vesicle types transporting APP, APLP1 and APLP2. These analyses will also allow conclusions regarding their specific molecular function (fig. 3).

Pathogenic Role of the APP Gene Family in Axonal Transport

In addition to the importance of interactions between APP and its processing machinery during transport, another potential important role of APP in axonal transport was shown in studies of Drosophila wandering third instar larvae with a deletion of the APPL gene (the Drosophila homologue of APP) or overexpressing APP/APPL. Changes in the dose of APP/APPL resulted in increased accumulation of synaptic vesicles in peripheral axons – a phenotype also observed in JIP 1/2 (APLIP1) or JIP3/ JSAP1 (Syd) mutant Drosophila larvae [35, 46, 71]. JIP1 and JIP2 – as well as JIP3, although very different in primary sequence – belong to a family of scaffold factors for

Fig. 2. The anterograde transport of APP is independent of JIP1b.

Mouse primary neurons (DIV5) were treated with JIP1b siRNA (d–f) or nonsense siRNA (a–c) for control. The cells were subjected to immunocytochemical analyses using anti-APP (a, c, d, f ; red) and anti-JIP-1b (b, c, e, f ; green) antibodies 18 h after transfection. Colocalization of endogenous APP (red) and JIP1b (green) was detected in the cell soma and at the tips of neurites, as indicated by the overlap (c, f; yellow). Scale bar: 20 m (insets: 6 m). Fig. 3. Model of the subcellular transport of the APP family members. APP, APLP1 and APLP2 (APP/APLPs) are transported in neurons to the dendritic and axonal compartment. a They are capable of forming homo- and heterocomplexes in a cis- and transcellular fashion (1–5), promoting cell-cell adhesion. It is possible that the transcellular interaction takes place between pre- and postsynaptic membranes in neurons. Alternatively, transcellular interaction involving APP/APLPs may take place between neurons and glia cells. APP/APLPs can be transported in common (b, c) or separate (a) vesicles in the form of monomers, homo- or heterocomplexes, towards the plasma membrane. The different dimer combinations and vesicle compositions may modulate the targeting and functions of APP/APLPs in neurons. b Axonal transport of APP/APLPs is driven by kinesin-1. A part of APP/ APLPs might be coupled via JIP1b to the motor machinery (1); however, recent data indicate that APP/APLPs can be transported by kinesin in absence of JIP1b, suggesting the existence of an additional molecular linker, coupling APP-containing vesicles to the kinesin motor machinery (2).

Pathogenic Role of APP Axonal Transport

the mitogen-activated protein kinase (MAPK) cascades, and accumulate at the leading edges of cells [72–74]. Mutation of APLIP1 caused axonal transport defects of both anterograde and retrograde vesicle transport. Thus, APLIP1 may be an important part of motor-cargo linkage complexes for kinesin-1 and dynein. Alternatively, APLIP1 and its associated JNK signaling proteins may serve as an important signaling module in MAPK pathway for regulating transport by the two opposing mechanisms [46, 75]. APP might function as a regulator of the scaffolding activity of JIP1 in the JNK signaling pathway, as recently shown for the Notch intracellular domain [76]. In a recent paper, Stokin et al. [77] identified axonal defects in APP transgenic mice. They found in fibers of the nucleus basalis of Meynert, which provide the major cholinergic input to the cerebral cortex, that the axonal varicosities exhibited substantial variation in size and morphology. Generally, varicosities correspond to en passant synaptic boutons, are regularly spaced, and have relatively constant diameters. The varicosities in the transgenic mice, in contrast, were often unusually large and irregularly spaced and contained large numbers of mitochondria, sporadic multilamellar bodies, vacuoles, and vesicles. Varicosities of diameters larger than 3 m, termed swellings, were in transgenic mice three times as many as in controls. This is a feature suggestive of axonal transport defects, as found previously in tau transgenic mouse and Drosophila models [78–80] and in dystrophic neuritis embedded in amyloid in AD [80, 81]. Interestingly, a 50% reduction in genetic dosage of KLC1 in these APP transgenic mice caused a significant increase of axonal swellings and A generation preceding the onset of amyloid deposition, suggesting that impaired transport may contribute to early stages of AD [77]. Recently, it was reported that transgenic expression of human tau (0N3R) in Drosophila larval motor neurons causes morphological and functional disruption to the neuromuscular junctions [80]. Despite reduction in size with irregular and abnormal bouton structure and abnormal endo-/exocytosis, tau-expressing neuromuscular junctions retain synaptotagmin expression and can form active zones. Electrophysiological studies showed that following high frequency stimulation (50 Hz), evoked synaptic potentials were significantly decreased [80]. The mechanism underlying the change in evoked synaptic potentials is not clear, but might contribute to a significant reduction in the number of mitochondria in the presynaptic terminals of motor neurons expressing mutant tau. These results suggest that disruption of axoNeurodegenerative Dis 2006;3:218–226

223

nal transport and synaptic transmission may be key components of the pathogenic mechanism that underlie neuronal dysfunction in the early stages of tauopathies. It will be interesting whether changes in the dose of APP/ APPL, which also affect the axonal transport of synaptic vesicles and morphology of synapses in mouse and Drosophila models, may also lead to altered synaptic plasticity, thought to represent the neurophysiological correlate of learning and memory [82, 83]. If true, this would strongly favor the hypothesis that changes in the dose of

APP and tau cause altered neurotransmission, contributing to cognitive decline and neurodegeneration of AD. Acknowledgements We thank Scott DeBoer (University of Chicago, Ill., USA) for critical reading of the manuscript and Marianne Giuffra for assistance in editing. S.K. is grateful for the support provided by the DFG.

References 1 Lyckman AW, Confaloni AM, Thinakaran G, Sisodia SS, Moya KL: Post-translational processing and turnover kinetics of presynaptically targeted amyloid precursor superfamily proteins in the central nervous system. J Biol Chem 1998; 273:11100–11106. 2 Tienari PJ, De Strooper B, Ikonen E, Simons M, Weidemann A, Czech C, Hartmann T, Ida N, Multhaup G, Masters CL, Van Leuven F, Beyreuther K, Dotti CG: The beta-amyloid domain is essential for axonal sorting of amyloid precursor protein. EMBO J 1996; 15: 5218–5229. 3 Reinhard C, Hebert SS, De Strooper B: The amyloid-beta precursor protein: integrating structure with biological function. EMBO J 2005;24:3996–4006. 4 Lammich S, Kojro E, Postina R, Gilbert S, Pfeiffer R, Jasionowski M, Haass C, Fahrenholz F: Constitutive and regulated alphasecretase cleavage of Alzheimer’s amyloid precursor protein by a disintegrin metalloprotease. Proc Natl Acad Sci USA 1999; 96: 3922–3927. 5 Skovronsky DM, Moore DB, Milla ME, Doms RW, Lee VM: Protein kinase C-dependent alpha-secretase competes with betasecretase for cleavage of amyloid-beta precursor protein in the trans-golgi network. J Biol Chem 2000; 275:2568–2575. 6 Vassar R, Bennett BD, Babu-Khan S, Kahn S, Mendiaz EA, Denis P, Teplow DB, Ross S, Amarante P, Loeloff R, Luo Y, Fisher S, Fuller J, Edenson S, Lile J, Jarosinski MA, Biere AL, Curran E, Burgess T, Louis JC, Collins F, Treanor J, Rogers G, Citron M: Beta-secretase cleavage of Alzheimer’s amyloid precursor protein by the transmembrane aspartic protease BACE. Science 1999;286:735–741. 7 Huse JT, Pijak DS, Leslie GJ, Lee VM, Doms RW: Maturation and endosomal targeting of beta-site amyloid precursor protein-cleaving enzyme. The Alzheimer’s disease beta-secretase. J Biol Chem 2000; 275:33729–33737.

224

8 Koo EH, Squazzo SL: Evidence that production and release of amyloid beta-protein involves the endocytic pathway. J Biol Chem 1994;269:17386–17389. 9 Koo EH, Sisodia SS, Archer DR, Martin LJ, Weidemann A, Beyreuther K, Fischer P, Masters CL, Price DL: Precursor of amyloid protein in Alzheimer disease undergoes fast anterograde axonal transport. Proc Natl Acad Sci USA 1990;87:1561–1565. 10 Yamazaki T, Selkoe DJ, Koo EH: Trafficking of cell surface beta-amyloid precursor protein: retrograde and transcytotic transport in cultured neurons. J Cell Biol 1995; 129: 431–442. 11 Sisodia SS, Koo EH, Hoffman PN, Perry G, Price DL: Identification and transport of full-length amyloid precursor proteins in rat peripheral nervous system. J Neurosci 1993; 13:3136–3142. 12 Morin PJ, Abraham CR, Amaratunga A, Johnson RJ, Huber G, Sandell JH, Fine RE: Amyloid precursor protein is synthesized by retinal ganglion cells, rapidly transported to the optic nerve plasma membrane and nerve terminals, and metabolized. J Neurochem 1993;61:464–473. 13 Moya KL, Confaloni AM, Allinquant B: In vivo neuronal synthesis and axonal transport of Kunitz protease inhibitor (KPI)-containing forms of the amyloid precursor protein. J Neurochem 1994;63:1971–1974. 14 Yam PS, Takasago T, Dewar D, Graham DI, McCulloch J: Amyloid precursor protein accumulates in white matter at the margin of a focal ischaemic lesion. Brain Res 1997; 760: 150–157. 15 Stone JR, Singleton RH, Povlishock JT: Antibodies to the C-terminus of the beta-amyloid precursor protein (APP): a site specific marker for the detection of traumatic axonal injury. Brain Res 2000;871:288–302. 16 Stamer K, Vogel R, Thies E, Mandelkow E, Mandelkow EM: Tau blocks traffic of organelles, neurofilaments, and APP vesicles in neurons and enhances oxidative stress. J Cell Biol 2002;156:1051–1063.

Neurodegenerative Dis 2006;3:218–226

17 Kaether C, Skehel P, Dotti CG: Axonal membrane proteins are transported in distinct carriers: a two-color video microscopy study in cultured hippocampal neurons. Mol Biol Cell 2000;11:1213–1224. 18 Ikin AF, Annaert WG, Takei K, De Camilli P, Jahn R, Greengard P, Buxbaum JD: Alzheimer amyloid protein precursor is localized in nerve terminal preparations to Rab5containing vesicular organelles distinct from those implicated in the synaptic vesicle pathway. J Biol Chem 1996; 271: 31783– 31786. 19 Ferreira A, Niclas J, Vale RD, Banker G, Kosik KS: Suppression of kinesin expression in cultured hippocampal neurons using antisense oligonucleotides. J Cell Biol 1992;117: 595–606. 20 Amaratunga A, Morin PJ, Kosik KS, Fine RE: Inhibition of kinesin synthesis and rapid anterograde axonal transport in vivo by an antisense oligonucleotide. J Biol Chem 1993; 268:17427–17430. 21 Buxbaum JD, Thinakaran G, Koliatsos V, O’Callahan J, Slunt HH, Price DL, Sisodia SS: Alzheimer amyloid protein precursor in the rat hippocampus: transport and processing through the perforant path. J Neurosci 1998;18:9629–9637. 22 Simons M, Ikonen E, Tienari PJ, Cid-Arregui A, Monning U, Beyreuther K, Dotti CG: Intracellular routing of human amyloid protein precursor: axonal delivery followed by transport to the dendrites. J Neurosci Res 1995;41:121–128. 23 Vale RD, Fletterick RJ: The design plan of kinesin motors. Annu Rev Cell Dev Biol 1997; 13:745–777. 24 Nakagawa T, Tanaka Y, Matsuoka E, Kondo S, Okada Y, Noda Y, Kanai Y, Hirokawa N: Identification and classification of 16 new kinesin superfamily (KIF) proteins in mouse genome. Proc Natl Acad Sci USA 1997; 94: 9654–9659.

Kins/Lauther/Szodorai/Beyreuther

25 Niclas J, Navone F, Hom-Booher N, Vale RD: Cloning and localization of a conventional kinesin motor expressed exclusively in neurons. Neuron 1994;12:1059–1072. 26 Xia C, Rahman A, Yang Z, Goldstein LS: Chromosomal localization reveals three kinesin heavy chain genes in mouse. Genomics 1998;52:209–213. 27 Kanai Y, Okada Y, Tanaka Y, Harada A, Terada S, Hirokawa N: KIF5C, a novel neuronal kinesin enriched in motor neurons. J Neurosci 2000;20:6374–6384. 28 Lawrence CJ, Dawe RK, Christie KR, Cleveland DW, Dawson SC, Endow SA, Goldstein LS, Goodson HV, Hirokawa N, Howard J, Malmberg RL, McIntosh JR, Miki H, Mitchison TJ, Okada Y, Reddy AS, Saxton WM, Schliwa M, Scholey JM, Vale RD, Walczak CE, Wordeman L: A standardized kinesin nomenclature. J Cell Biol 2004; 167:19–22. 29 Rahman A, Friedman DS, Goldstein LS: Two kinesin light chain genes in mice. Identification and characterization of the encoded proteins. J Biol Chem 1998; 273: 15395– 15403. 30 Junco A, Bhullar B, Tarnasky HA, van der Hoorn FA: Kinesin light-chain KLC3 expression in testis is restricted to spermatids. Biol Reprod 2001; 64:1320–1330. 31 Hirokawa N, Takemura R: Molecular motors and mechanisms of directional transport in neurons. Nat Rev Neurosci 2005;6:201–214. 32 Kamal A, Stokin GB, Yang Z, Xia CH, Goldstein LS: Axonal transport of amyloid precursor protein is mediated by direct binding to the kinesin light chain subunit of kinesinI. Neuron 2000;28:449–459. 33 Kamal A, Almenar-Queralt A, LeBlanc JF, Roberts EA, Goldstein LS: Kinesin-mediated axonal transport of a membrane compartment containing beta-secretase and presenilin-1 requires APP. Nature 2001; 414: 643–648. 34 Lazarov O, Morfini GA, Lee EB, Farah MH, Szodorai A, DeBoer SR, Koliatsos VE, Kins S, Lee VM, Wong PC, Price DL, Brady ST, Sisodia SS: Axonal transport, amyloid precursor protein, kinesin-1, and the processing apparatus: revisited. J Neurosci 2005; 25: 2386–2395. 35 Bowman AB, Kamal A, Ritchings BW, Philp AV, McGrail M, Gindhart JG, Goldstein LS: Kinesin-dependent axonal transport is mediated by the Sunday driver (SYD) protein. Cell 2000;103:583–594. 36 Nakagawa T, Setou M, Seog D, Ogasawara K, Dohmae N, Takio K, Hirokawa N: A novel motor, KIF13A, transports mannose-6phosphate receptor to plasma membrane through direct interaction with AP-1 complex. Cell 2000;103:569–581.

Pathogenic Role of APP Axonal Transport

37 Mok H, Shin H, Kim S, Lee JR, Yoon J, Kim E: Association of the kinesin superfamily motor protein KIF1Balpha with postsynaptic density-95 (PSD-95), synapse-associated protein-97, and synaptic scaffolding molecule PSD-95/discs large/zona occludens-1 proteins. J Neurosci 2002; 22:5253–5258. 38 Hoogenraad CC, Milstein AD, Ethell IM, Henkemeyer M, Sheng M: GRIP1 controls dendrite morphogenesis by regulating EphB receptor trafficking. Nat Neurosci 2005; 8: 906–915. 39 Ko J, Kim S, Valtschanoff JG, Shin H, Lee JR, Sheng M, Premont RT, Weinberg RJ, Kim E: Interaction between liprin-alpha and GIT1 is required for AMPA receptor targeting. J Neurosci 2003;23:1667–1677. 40 Setou M, Seog DH, Tanaka Y, Kanai Y, Takei Y, Kawagishi M, Hirokawa N: Glutamate-receptor-interacting protein GRIP1 directly steers kinesin to dendrites. Nature 2002;417: 83–87. 41 Shin H, Wyszynski M, Huh KH, Valtschanoff JG, Lee JR, Ko J, Streuli M, Weinberg RJ, Sheng M, Kim E: Association of the kinesin motor KIF1A with the multimodular protein liprin-alpha. J Biol Chem 2003; 278: 11393– 11401. 42 Wyszynski M, Kim E, Dunah AW, Passafaro M, Valtschanoff JG, Serra-Pages C, Streuli M, Weinberg RJ, Sheng M: Interaction between GRIP and liprin-alpha/SYD2 is required for AMPA receptor targeting. Neuron 2002;34:39–52. 43 Setou M, Nakagawa T, Seog DH, Hirokawa N: Kinesin superfamily motor protein KIF17 and mLin-10 in NMDA receptor-containing vesicle transport. Science 2000; 288: 1796– 1802. 44 Guillaud L, Setou M, Hirokawa N: KIF17 dynamics and regulation of NR2B trafficking in hippocampal neurons. J Neurosci 2003; 23:131–140. 45 Byrd DT, Kawasaki M, Walcoff M, Hisamoto N, Matsumoto K, Jin Y: UNC-16, a JNK-signaling scaffold protein, regulates vesicle transport in C. elegans. Neuron 2001; 32: 787–800. 46 Horiuchi D, Barkus RV, Pilling AD, Gassman A, Saxton WM: APLIP1, a kinesin binding JIP-1/JNK scaffold protein, influences the axonal transport of both vesicles and mitochondria in Drosophila. Curr Biol 2005; 15: 2137–2141. 47 Inomata H, Nakamura Y, Hayakawa A, Takata H, Suzuki T, Miyazawa K, Kitamura N: A scaffold protein JIP-1b enhances amyloid precursor protein phosphorylation by JNK and its association with kinesin light chain 1. J Biol Chem 2003; 278: 22946– 22955. 48 Matsuda S, Matsuda Y, D’Adamio L: Amyloid beta protein precursor (AbetaPP), but not AbetaPP-like protein 2, is bridged to the kinesin light chain by the scaffold protein JNK-interacting protein 1. J Biol Chem 2003; 278:38601–38606.

49 Muresan Z, Muresan V: Coordinated transport of phosphorylated amyloid-beta precursor protein and c-Jun NH2-terminal kinase-interacting protein-1. J Cell Biol 2005; 171:615–625. 50 Verhey KJ, Meyer D, Deehan R, Blenis J, Schnapp BJ, Rapoport TA, Margolis B: Cargo of kinesin identified as JIP scaffolding proteins and associated signaling molecules. J Cell Biol 2001;152:959–970. 51 Sakamoto R, Byrd DT, Brown HM, Hisamoto N, Matsumoto K, Jin Y: The Caenorhabditis elegans UNC-14 RUN domain protein binds to the kinesin-1 and UNC-16 complex and regulates synaptic vesicle localization. Mol Biol Cell 2005; 16:483–496. 52 Borg JP, Ooi J, Levy E, Margolis B: The phosphotyrosine interaction domains of X11 and FE65 bind to distinct sites on the YENPTY motif of amyloid precursor protein. Mol Cell Biol 1996;16:6229–6241. 53 McLoughlin DM, Miller CC: The intracellular cytoplasmic domain of the Alzheimer’s disease amyloid precursor protein interacts with phosphotyrosine-binding domain proteins in the yeast two-hybrid system. FEBS Lett 1996;397:197–200. 54 Tomita S, Ozaki T, Taru H, Oguchi S, Takeda S, Yagi Y, Sakiyama S, Kirino Y, Suzuki T: Interaction of a neuron-specific protein containing PDZ domains with Alzheimer’s amyloid precursor protein. J Biol Chem 1999; 274:2243–2254. 55 Soba P, Eggert S, Wagner K, Zentgraf H, Siehl K, Kreger S, Lower A, Langer A, Merdes G, Paro R, Masters CL, Muller U, Kins S, Beyreuther K: Homo- and heterodimerization of APP family members promotes intercellular adhesion. EMBO J 2005;24:3624– 3634. 56 Sheng JG, Price DL, Koliatsos VE: The betaamyloid-related proteins presenilin 1 and BACE1 are axonally transported to nerve terminals in the brain. Exp Neurol 2003;184: 1053–1057. 57 Chen XH, Siman R, Iwata A, Meaney DF, Trojanowski JQ, Smith DH: Long-term accumulation of amyloid-beta, beta-secretase, presenilin-1, and caspase-3 in damaged axons following brain trauma. Am J Pathol 2004;165:357–371. 58 Heber S, Herms J, Gajic V, Hainfellner J, Aguzzi A, Rulicke T, von Kretzschmar H, von Koch C, Sisodia S, Tremml P, Lipp HP, Wolfer DP, Muller U: Mice with combined gene knock-outs reveal essential and partially redundant functions of amyloid precursor protein family members. J Neurosci 2000; 20:7951–7963. 59 Herms J, Anliker B, Heber S, Ring S, Fuhrmann M, Kretzschmar H, Sisodia S, Muller U: Cortical dysplasia resembling human type 2 lissencephaly in mice lacking all three APP family members. EMBO J 2004: 23;4106–4115.

Neurodegenerative Dis 2006;3:218–226

225

60 Neumann S, Schobel S, Jager S, Trautwein A, Haass C, Pietrzik CU, Lichtenthaler SF: APLP1 influences endocytosis and proteolytic processing of the amyloid precursor protein. J Biol Chem 2006; 281:7583–7594. 61 Scheuermann S, Hambsch B, Hesse L, Stumm J, Schmidt C, Beher D, Bayer TA, Beyreuther K, Multhaup G: Homodimerization of amyloid precursor protein and its implication in the amyloidogenic pathway of Alzheimer’s disease. J Biol Chem 2001; 276: 33923–33929. 62 De Strooper B, Annaert W: Proteolytic processing and cell biological functions of the amyloid precursor protein. J Cell Sci 2000; 113:1857–1870. 63 Beher D, Elle C, Underwood J, Davis JB, Ward R, Karran E, Masters CL, Beyreuther K, Multhaup G: Proteolytic fragments of Alzheimer’s disease-associated presenilin 1 are present in synaptic organelles and growth cone membranes of rat brain. J Neurochem 1999;72:1564–1573. 64 Kim TW, Wu K, Xu JL, McAuliffe G, Tanzi RE, Wasco W, Black IB: Selective localization of amyloid precursor-like protein 1 in the cerebral cortex postsynaptic density. Brain Res Mol Brain Res 1995;32:36–44. 65 Sandbrink R, Masters CL, Beyreuther K: APP gene family: unique age-associated changes in splicing of Alzheimer’s betaA4amyloid protein precursor. Neurobiol Dis 1994;1:13–24. 66 Lorent K, Overbergh L, Moechars D, De Strooper B, Van Leuven F, Van den Berghe H: Expression in mouse embryos and in adult mouse brain of three members of the amyloid precursor protein family, of the alpha-2-macroglobulin receptor/low density lipoprotein receptor-related protein and of its ligands apolipoprotein E, lipoprotein lipase, alpha-2-macroglobulin and the 40,000 molecular weight receptor-associated protein. Neuroscience 1995;65:1009–1025. 67 Wang P, Yang G, Mosier DR, Chang P, Zaidi T, Gong YD, Zhao NM, Dominguez B, Lee KF, Gan WB, Zheng H: Defective neuromuscular synapses in mice lacking amyloid precursor protein (APP) and APP-like protein 2. J Neurosci 2005;25:1219–1225. 68 Ashley J, Packard M, Ataman B, Budnik V: Fasciclin II signals new synapse formation through amyloid precursor protein and the scaffolding protein dX11/Mint. J Neurosci 2005;25:5943–5955. 69 Hirokawa N: The mechanisms of fast and slow transport in neurons: identification and characterization of the new kinesin superfamily motors. Curr Opin Neurobiol 1997;7:605–614.

226

70 Hirokawa N: Kinesin and dynein superfamily proteins and the mechanism of organelle transport. Science 1998;279:519–526. 71 Hurd DD, Saxton WM: Kinesin mutations cause motor neuron disease phenotypes by disrupting fast axonal transport in Drosophila. Genetics 1996;144:1075–1085. 72 Takino T, Nakada M, Miyamori H, Watanabe Y, Sato T, Gantulga D, Yoshioka K, Yamada KM, Sato H: JSAP1/JIP3 cooperates with focal adhesion kinase to regulate c-Jun N-terminal kinase and cell migration. J Biol Chem 2005;280:37772–37781. 73 Matsuda S, Yasukawa T, Homma Y, Ito Y, Niikura T, Hiraki T, Hirai S, Ohno S, Kita Y, Kawasumi M, Kouyama K, Yamamoto T, Kyriakis JM, Nishimoto I: c-Jun N-terminal kinase (JNK)-interacting protein-1b/isletbrain-1 scaffolds Alzheimer’s amyloid precursor protein with JNK. J Neurosci 2001;21: 6597–6607. 74 Yasuda J, Whitmarsh AJ, Cavanagh J, Sharma M, Davis RJ: The JIP group of mitogenactivated protein kinase scaffold proteins. Mol Cell Biol 1999; 19:7245–7254. 75 Gdalyahu A, Ghosh I, Levy T, Sapir T, Sapoznik S, Fishler Y, Azoulai D, Reiner O: DCX, a new mediator of the JNK pathway. EMBO J 2004;23:823–832. 76 Kim JW, Kim MJ, Kim KJ, Yun HJ, Chae JS, Hwang SG, Chang TS, Park HS, Lee KW, Han PL, Cho SG, Kim TW, Choi EJ: Notch interferes with the scaffold function of JNKinteracting protein 1 to inhibit the JNK signaling pathway. Proc Natl Acad Sci USA 2005;102:14308–14313. 77 Stokin GB, Lillo C, Falzone TL, Brusch RG, Rockenstein E, Mount SL, Raman R, Davies P, Masliah E, Williams DS, Goldstein LS: Axonopathy and transport deficits early in the pathogenesis of Alzheimer’s disease. Science 2005;307:1282–1288. 78 Ishihara T, Hong M, Zhang B, Nakagawa Y, Lee MK, Trojanowski JQ, Lee VM: Age-dependent emergence and progression of a tauopathy in transgenic mice overexpressing the shortest human tau isoform. Neuron 1999;24:751–762. 79 Spittaels K, Van den Haute C, Van Dorpe J, Geerts H, Mercken M, Bruynseels K, Lasrado R, Vandezande K, Laenen I, Boon T, Van Lint J, Vandenheede J, Moechars D, Loos R, Van Leuven F: Glycogen synthase kinase-3beta phosphorylates protein tau and rescues the axonopathy in the central nervous system of human four-repeat tau transgenic mice. J Biol Chem 2000; 275:41340–41349.

Neurodegenerative Dis 2006;3:218–226

80 Chee FC, Mudher A, Cuttle MF, Newman TA, MacKay D, Lovestone S, Shepherd D: Over-expression of tau results in defective synaptic transmission in Drosophila neuromuscular junctions. Neurobiol Dis 2005; 20: 918–928. 81 Terry RD, Gonatas NK, Weiss M: Ultrastructural studies in Alzheimer’s presenile dementia. Am J Pathol 1964; 44:269–297. 82 Hawkins RD, Kandel ER, Siegelbaum SA: Learning to modulate transmitter release: themes and variations in synaptic plasticity. Annu Rev Neurosci 1993;16:625–665. 83 Sigrist SJ, Reiff DF, Thiel PR, Steinert JR, Schuster CM: Experience-dependent strengthening of Drosophila neuromuscular junctions. J Neurosci 2003;23:6546–6556. 84 Su Q, Cai Q, Gerwin C, Smith CL, Sheng ZH: Syntabulin is a microtubule-associated protein implicated in syntaxin transport in neurons. Nat Cell Biol 2004; 6:941–953. 85 Cai Q, Gerwin C, Sheng ZH: Syntabulin-mediated anterograde transport of mitochondria along neuronal processes. J Cell Biol 2005;170:959–969. 86 Marszalek JR, Liu X, Roberts EA, Chui D, Marth JD, Williams DS, Goldstein LS: Genetic evidence for selective transport of opsin and arrestin by kinesin-II in mammalian photoreceptors. Cell 2000;102:175–187. 87 Morris RL, English CN, Lou JE, Dufort FJ, Nordberg J, Terasaki M, Hinkle B: Redistribution of the kinesin-II subunit KAP from cilia to nuclei during the mitotic and ciliogenic cycles in sea urchin embryos. Dev Biol 2004;274:56–69. 88 Mueller J, Perrone CA, Bower R, Cole DG, Porter ME: The FLA3 KAP subunit is required for localization of kinesin-2 to the site of flagellar assembly and processive anterograde intraflagellar transport. Mol Biol Cell 2005;16:1341–1354. 89 Miller MS, Esparza JM, Lippa AM, Lux FG, 3rd, Cole DG, Dutcher SK: Mutant kinesin-2 motor subunits increase chromosome loss. Mol Biol Cell 2005; 16:3810–3820. 90 Haraguchi K, Hayashi T, Jimbo T, Yamamoto T, Akiyama T: Role of the kinesin-2 family protein, KIF3, during mitosis. J Biol Chem 2006;281:4094–4099. 91 Miller KE, DeProto J, Kaufmann N, Patel BN, Duckworth A, Van Vactor D: Direct observation demonstrates that Liprin-alpha is required for trafficking of synaptic vesicles. Curr Biol 2005;15:684–689. 92 Zhao C, Takita J, Tanaka Y, Setou M, Nakagawa T, Takeda S, Yang HW, Terada S, Nakata T, Takei Y, Saito M, Tsuji S, Hayashi Y, Hirokawa N: Charcot-Marie-Tooth disease type 2A caused by mutation in a microtubule motor KIF1Bbeta. Cell 2001;105:587–597.

Kins/Lauther/Szodorai/Beyreuther

Original Paper

Diseases

Neurodegenerative Dis 2006;3:227–232 DOI: 10.1159/000095260

Presenilin Function in Caenorhabditis elegans Agata Smialowska a Ralf Baumeister a, b a

Bio3/Bioinformatics and Molecular Genetics, and b ZBSA/Freiburg Center for Systems Biology, University of Freiburg, Freiburg, Germany

Key Words Presenilin Caenorhabditis elegans Alzheimer’s disease Modifier screen

dependent demethylase of histone H3. Mutations in spr-1, spr-3, spr-4 and spr-5 genes suppress the egg-laying phenotype of sel-12 loss of function mutants by derepressing the expression of the second C. elegans presenilin gene, hop-1. Copyright © 2006 S. Karger AG, Basel

Abstract The genome of the nematode Caenorhabditis elegans contains homologs of several genes associated with familial Alzheimer’s disease in humans. apl-1 encodes a transmembrane protein belonging to the amyloid precursor protein family, sel-12 and hop-1 are the two somatically expressed presenilin genes that resemble PS1 and PS2 on both a structural and a functional level. Mutations in the sel-12-encoded presenilin gene cause defective Notch/lin-12 signaling and result in reduced egg-laying, caused by cell specification and cell attachment defects. spr-1, spr-3, spr-4 and spr-5 were identified as the suppressors of the egg-laying defect of presenilin/sel-12 loss of function mutants in genetic suppressor screens. The corresponding proteins are C. elegans homologs of human REST, CoREST and LSD1, respectively. REST/NSRF (Re1 silencing transcription factor/neural-restrictive silencing factor) is a transcriptional repressor that blocks the expression of neuronal genes in non-neuronal tissues in vertebrates. CoREST is a conserved histone deacetylase and demethylase-containing co-repressor complex possessing a potential chromatin-modifying activity. It is recruited to the promoter via REST-mediated DNA binding. LSD1 is a flavin-

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0227$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Introduction

Alzheimer’s disease (AD) is a multifactorial syndrome. It is now widely accepted that the generation of the 40- to 42-amino-acid-long amyloid- (A) peptide, a proteolytic product derived from the 695- to 770-amino-acidlong amyloid precursor protein (APP), is critical for the onset of both sporadic and familial forms of AD [1]. Three genes have so far been identified to be involved in heritable forms of AD. Mutations in all three genes result in a significantly increased level of A42, the 42-amino-acid variant of A. These mutations occur in the APP gene itself that encodes a class I transmembrane protein, likely involved in neurite outgrowth and cell-cell interactions, and in the two closely related genes encoding the presenilins PS1 and PS2. Mutations in the presenilin genes are the major cause of early onset forms of familial AD (FAD). The presenilin genes encode a family of six- to eighttransmembrane (TM)-domain proteins that are evolu-

Ralf Baumeister, PhD ZBSA/Bioinformatics and Molecular Genetics Schaenzlestrasse 1 DE–79104 Freiburg (Germany) Tel. +49 761 203 2799, Fax +49 761 203 8351, E-Mail [email protected]

Fig. 1. Sex muscle defects in C. elegans presenilin and Notch mutants. a The schematic representation of the C. elegans egg-laying

apparatus. The vulval muscles are innervated by the pair of hermaphrodite-specific neurons (HSN). b The defects in vulval muscle attachment in presenilin sel-12 single mutants or hop-1;sel-12 double mutants resemble the weak Notch loss of function mutants (lin-12(0)). Modified with permission from Eimer et al. [8].

tionary conserved. Our data and those of several other labs provide ample evidence that the function of the presenilins is conserved in Caenorhabditis elegans, Drosophila, Zebrafish, mouse and human [2]. The C. elegans genome contains three genes with significant similarity to human presenilins. One of them, spe-4, is only expressed in spermatids during spermatogenesis. Its function involves the correct transport and formation of intracellular organelles and will be described in detail elsewhere. The two other C. elegans presenilin genes, sel-12 (suppressor/enhancer of lin-12) and hop-1 (homolog of presenilin) have similar functions and can substitute for one another. hop-1;sel-12 double mutants completely inhibit signaling through the two C. elegans Notch receptors glp-1 and lin-12, underscoring their fundamental role for Notch signaling. In agreement with this model, mouse embryos with a targeted deletion of PS1 or PS1 and PS2 die prematurely and display reduced Notch activity [3]. Deletions of the Drosophila presenilin locus also have been shown to result in embryonic lethality that resembles a Notch –/– phenotype. 228

Neurodegenerative Dis 2006;3:227–232

Taken together, these results suggest that presenilins function in the embryonic development of most, if not all, animals and facilitate Notch signaling. Although redundant in their enzymatic activity, sel-12 and hop-1 mutants differ in the phenotype that is the consequence of their deletion. Whereas hop-1 null mutants are superficially inconspicuous, all known mutations in sel-12 have a strong egg-laying defect, and also display additional deficits not obviously linked to egg-laying.

Presenilin Function in C. elegans

Signaling by C. elegans GLP-1/LIN-12 receptors mediates a wide range of developmental cell fate decisions involving Notch function already at the four-cell embryo and in the germline [4, 5]. Notch dysfunction results in defective somatic gonad and vulva development, germline sterility and maternal-effect embryonic lethality [6] (see fig. 2). The egg-laying apparatus in C. elegans requires the coordinated function of multiple tissues including the uterSmialowska /Baumeister

us, the vulva, the egg-laying musculature (sex muscles) and a pair of the hermaphrodite-specific motor neurons that stimulate the muscle contraction (fig. 1). The development of these structures is coordinated by a series of cell-cell interactions involving both the Notch and Ras/ EGF signaling events [7]. The LIN-12/Notch signaling is involved at several steps during vulva formation, and some of them are affected in sel-12 mutants [8, 9]. LIN-12 signaling specifies the fate of the gonadal anchor cell that functions as an organizer cell to initiate vulval cell fate induction [10]. The anchor cell, in turn, induces a group of cells to form the vulva and the vulval-uterine connection. This inductive signaling is perturbed in sel-12 null mutants [8] and weak lin-12 loss of function mutants [9]. However, the penetrance of this cell specification defect is not complete, whereas the egg-laying defect of sel-12 loss of function mutants is. This discrepancy suggested an additional function for sel-12 in the vulva development. We identified an additional role of sel-12 in controlling the sex muscle attachment to hypodermal tissues. These muscle defects are highly penetrant and correlate with the inability of the mutant animals to lay eggs [8]. Although all sixteen uterine and vulva muscle cells required for the functional egg-laying apparatus are generated in sel-12 mutant animals (in contrast to strong lin12/Notch mutants, in which the muscle precursors are turned into different cell types and the muscles are not generated at all), they do not adhere properly to their epidermal target structures (fig. 1). The penetrance of the morphological and attachment defects is strongly enhanced in hop-1;sel-12 double mutants. This defect can be compensated by expressing presenilin genes exclusively in the vulva and uterine muscles, indicating the cell autonomy of this defect [11]. Interestingly, already the weak loss of function mutant sel-12 C60S produced these attachment defects. It is tempting to speculate that neurite outgrowth defects that were previously observed in sel-12 and hop-1;sel-12 mutant animals may also be the consequence of a defective cell attachment [12]. It is currently not known whether a partial inactivation of APP-like genes, for which a role in neuritic growth has been shown, may be involved in these phenotypes.

Suppressors and Enhancers of Presenilin in C. elegans

In contrast to PS1 mutants in mice that result in a recessive postnatal lethality due to respiratory deficits, mutations in both C. elegans presenilin genes are viable. This Presenilin Function in Caenorhabditis elegans

Fig. 2. Different levels of presenilin activity control distinct cellular functions. Developmental decisions of C. elegans are not equally sensitive to the level of LIN-12/Notch signaling (right panel). The genetic screens for suppressors and/or enhancers of a given phenotype led to the discovery of regulatory factors, some of them being the members of the -secretase complex per se (left panel).

allows the design of modifier screens. In particular, the sel-12 loss of function phenotype, a robust egg-laying defect, offered the possibility to identify both enhancers of this defect (with an assumed synthetic lethal phenotype) and suppressors, ameliorating or eliminating the egg-laying defect. These screens proved to be highly effective and revealed additional components of the -secretase complex, in particular alleles of all components of the -secretase complex, including aph-1, aph-2/nicastrin, pen-2 as well as new hop-1 alleles [13]. The relative simplicity of the nematode as a model also facilitated the discovery of the critical role of aph-2/nicastrin in the Notch signaling and thus -secretase activity [14] (fig. 2). The screens for suppressors of PS phenotype, on the other hand, have lead to the discovery of the transcriptional suppressors of the second, less abundant C. elegans presenilin hop-1. Over 25 suppressor alleles, designated spr (suppressor of presenilin) genes, were identified in our lab, and together with experiments conducted elsewhere revealed the identity of five members of a transcriptional repressor complex termed CoREST [11, 15, 16]. The mammalian CoREST complex has been shown to mediate the repression of genes responsible for key neuronal functions [17] and neuronal plasticity [18]. Neurodegenerative Dis 2006;3:227–232

229

In our genetic screens we identified, among others, spr-1, the C. elegans homolog of the human CoREST, spr-3 and spr-4, that most closely resemble REST DNAbinding proteins, and spr-5, encoding a polyamine oxidase-like protein now termed LSD1. At least three additional complementation groups were not yet assigned to [our unpubl. results]. CoREST is a co-repressor protein

Fig. 3. SPR-1–5 encode orthologues of the human CoREST tran-

scriptional repressor complex. The putative chromatin-modifying activities are provided by histone demethylase SPR-5/LSD1 and histone deacetylase (HDAC).

a

able to recruit further components to the complex via the multiple domains implicated in the protein-protein interactions [unpubl. results] (fig. 3 and 4) [19]. The chromatin-modifying component of the complex is potentially provided by the histone deacetylase and the histone demethylase LSD1 [20, 21]. A function of LSD1 was recently identified and suggests that of a flavin-dependent human lysine (K)-specific histone demethylase [20]. SPR-5/LSD1 catalyzes the removal of mono- and dimethyl marks on lysine 4 of histone 3 (H3K4), the chromatin mark associated with the sites of active transcription [22]. The function of this repressor complex has been described; however, the nature of its regulation remains largely unknown [21]. The mutations in spr-1, spr-3, spr-4 and spr-5 that we identified in suppressors of sel-12 result in a loss of function phenotype (fig. 4c). The point mutations map to the putative catalytical domain of SPR-5/LSD1, possibly interfering with its enzymatic activity as an amine oxidase/ histone demethylase [11]. As a consequence of the inactivation of at least some of the aspects of the CoREST complex activity, the subset of genes which are repressed in the wild-type animals is derepressed in the spr-5 mutant background. The fact that the suppression of the sel-12 loss of function phenotype by spr mutants is fully dependent on the presence of intact HOP-1 presenilin suggests

c

Gene

#

spr-1 spr-3 spr-4 spr-5

1 7 7 6

Loss of function alleles by133 by108, by109, by110, by131, by135, by136, by137 by105, by107, by112, by114, by129, by130, by132 by101, by113, by119, by128, by134, by139

ELM2

SANT

GATA

SANT

SPR-1

SWIRM

Amine oxidase

SPR-5

b 36 35 08 30 39 28 19 33 y 1 b y 1 b y 1 b y 1 b y 1 by 1 b y 1 by 1 pe b y ; ; ; ; ; ; ; ; 2 2 2 2 2 2 2 2 2 -t i l d e l - 1 e l - 1 e l - 1 e l - 1 e l - 1 el - 1 e l - 1 e l - 1 e l - 1 s s s s s s s W s s

hop-1 mRNA

Fig. 4. Genetic suppressors of sel-12 defects. Suppressors were identified in a random mutagenesis screen. a The table summarizes the alleles identified. b The Northern blot of L1 larval stages total RNA of the indicated alleles. Mutations in spr-1, spr-3, spr-4, spr-5 strongly derepress transcription of hop-1. c The nonsense mutations in SPR-1 and SPR-5 are localized in the putative protein-protein interaction domains of SPR-1 the putative enzymatic domain of SPR-5.

230

Neurodegenerative Dis 2006;3:227–232

Smialowska /Baumeister

that hop-1 could be one of the targets of SPR/CoRESTmediated transcriptional repression. Indeed, we have identified the promoter region in hop-1 which confers the differential expression of hop-1 in different genetical backgrounds. Under wild-type conditions, hop-1 transcriptional activity is strongly repressed by CoREST/ LSD1, and significant amounts of hop-1 mRNA were mostly seen in later developmental stages (L4 larvae, adults) and in the embryo. This suggests a major role of hop-1 at the early stages of development. Derepression of hop-1 transcription is sufficient to bypass the loss of functional SEL-12 protein in sel-12 null or FAD mutants. Two conclusions can be drawn from this result: (1) Under in vivo conditions, the functions of SEL-12 and HOP-1 are equivalent and they can substitute for one another, as long as the expression levels are adjusted. (2) All the deficits caused by the FAD mutant SEL-12 C60S can be suppressed by the derepression of HOP-1 through mutations in CoREST/LSD1. Given the fact that loss of function mutations in all components of this transcriptional repressor complex are potent suppressors, the most likely explanation for the suppression is that it only requires the disruption of the CoREST/LSD1 complex. If the same regulatory mechanism functions in human tissues, this complex may have the potential to serve as a therapeutic target for an anti-AD drug.

C. elegans as a Model for Presenilin Suppression

We have shown that the interference with the activity of CoREST complex in C. elegans suppresses sel-12 presenilin defects by upregulating hop-1 expression (fig. 4b) [11, 15]. We have proposed the development of an in vivo reporter system to test the CoREST transcriptional repressing activity. Such a system explores the identification of the hop-1 as a target gene for the repressor activity and is based on the hop-1 regulatory sequences [unpubl. results]. It would facilitate further investigation of the nature of the CoREST function as a transcriptional repres-

sor, including analysis of the chromatin changes on a target promoter in vivo. Furthermore, the identification of additional factors involved in the corepressor complex activity and the means of its regulation during development and in disease would be possible.

Future Perspectives

Work of the last years has shown that the assembly of presenilins and the other components of the -secretase complex, as well as the transport and function of this complex to/at the membrane, are tightly regulated. Several other genes that encode proteins with similarity to the presenilins have been identified in the meantime, although it is not known whether they participate in similar protease complexes. One of them, SPE-4, shares both the amino acid similarity and membrane topology of presenilins. In fact, the knowledge on SPE-4 function allowed the first speculation on the role of PS1 once it was cloned in 1995 [23]. However, data obtained in both C. elegans and cell culture indicate that SPE-4 cannot be incorporated into the -secretase complex. However, engineered chimera between human PS1 and worm SPE-4 allowed the identification of protein domains crucial for individual functions [unpubl. data from collaboration between Haass and Baumeister labs]. Given the relative simplicity of the model organism and its successful usage for revealing the -secretase complex components and its activity, it is now tempting to use the nematode for functional analyses of other proteases.

Acknowledgements We would like to thank all past and current members of the Baumeister and Haass lab for the fruitful scientific exchange over the past 6 years of this priority program. Work described here was supported by a grant from DFG SPP1085 (R.B). Additional support was provided by the Fonds der Chemischen Industrie (R.B).

References 1 Mattson MP: Pathways towards and away from Alzheimer’s disease. Nature 2004; 430: 631–639. 2 Brunkan AL, Goate AM: Presenilin function and gamma-secretase activity. J Neurochem 2005;93:769–792.

Presenilin Function in Caenorhabditis elegans

3 Guo Q, Sebastian L, Sopher BL, Miller MW, Ware CB, Martin GM, Mattson MP: Increased vulnerability of hippocampal neurons from presenilin-1 mutant knock-in mice to amyloid beta-peptide toxicity: central roles of superoxide production and caspase activation. J Neurochem 1999;72:1019–1029.

4 Priess JR, Schnabel H, Schnabel R: The glp-1 locus and cellular interactions in early C. elegans embryos. Cell 1987;51:601–611. 5 Austin J, Kimble J: glp-1 is required in the germ line for regulation of the decision between mitosis and meiosis in C. elegans. Cell 1987;51:589–599.

Neurodegenerative Dis 2006;3:227–232

231

6 Westlund B, Parry D, Clover R, Basson M, Johnson CD: Reverse genetic analysis of Caenorhabditis elegans presenilins reveals redundant but unequal roles for sel-12 and hop-1 in Notch-pathway signaling. PNAS 1999;96:2497–2502. 7 Sundaram MV: The love-hate relationship between Ras and Notch. Genes Dev 2005;19: 1825–1839. 8 Eimer S, Donhauser R, Baumeister R: The Caenorhabditis elegans presenilin sel-12 is required for mesodermal patterning and muscle function. Dev Biol 2002; 251: 178– 192. 9 Cinar HN, Sweet KL, Hosemann KE, Earley K, Newman AP: The SEL-12 presenilin mediates induction of the Caenorhabditis elegans uterine pi cell fate. Dev Biol 2001; 237: 173–182. 10 Greenwald IS, Sternberg PW, Robert Horvitz H: The lin-12 locus specifies cell fates in Caenorhabditis elegans. Cell 1983;34:435–444. 11 Eimer S, Lakowski B, Donhauser R, Baumeister R: Loss of spr-5 bypasses the requirement for the C. elegans presenilin sel-12 by derepressing hop-1. EMBO J 2002; 21: 5787– 5796. 12 Wittenburg N, Eimer S, Lakowski B, Rohrig S, Rudolph C, Baumeister R: Presenilin is required for proper morphology and function of neurons in C. elegans. Nature 2000; 406: 306–309.

232

13 Francis R, McGrath G, Zhang J, Ruddy DA, Sym M, Apfeld J, Nicoll M, Maxwell M, Hai B, Ellis MC: aph-1 and pen-2 are required for notch pathway signaling, -secretase cleavage of -APP, and presenilin protein accumulation. Developmental Cell 2002; 3: 85– 97. 14 Goutte C, Hepler W, Mickey KM, Priess JR: aph-2 encodes a novel extracellular protein required for GLP-1-mediated signaling. Development 2000;127:2481–2492. 15 Lakowski B, Eimer S, Gobel C, Bottcher A, Wagler B, Baumeister R: Two suppressors of sel-12 encode C2H2 zinc-finger proteins that regulate presenilin transcription in Caenorhabditis elegans. Development 2003; 130: 2117–2128. 16 Jarriault S, Greenwald I: Suppressors of the egg-laying defective phenotype of sel-12 presenilin mutants implicate the CoREST corepressor complex in LIN-12/Notch signaling in C. elegans. Genes Dev 2002; 16: 2713– 2728. 17 Ballas N, Battaglioli E, Atouf F, Andres ME, Chenoweth J, Anderson ME, Burger C, Moniwa M, Davie JR, Bowers WJ: Regulation of neuronal traits by a novel transcriptional complex. Neuron 2001;31:353–365.

Neurodegenerative Dis 2006;3:227–232

18 Ballas N, Grunseich C, Lu DD, Speh JC, Mandel G: REST and its corepressors mediate plasticity of neuronal gene chromatin throughout neurogenesis. Cell 2005; 121: 645–657. 19 Andres ME, Burger C, Peral-Rubio MJ, Battaglioli E, Anderson ME, Grimes J, Dallman J, Ballas N, Mandel G: CoREST: a functional corepressor required for regulation of neural-specific gene expression. PNAS 1999; 96:9873–9878. 20 Shi Y, Lan F, Matson C, Mulligan P, Whetstine JR, Cole PA, Casero RA, Shi Y: Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 2004; 119:941–953. 21 Shi YJ, Matson C, Lan F, Iwase S, Baba T, Shi Y: Regulation of LSD1 histone demethylase activity by its associated factors. Mol Cell 2005;19:857–864. 22 Bannister AJ, Kouzarides T: Reversing histone methylation. Nature 2005; 436: 1103– 1106. 23 Arduengo PM, Appleberry OK, Chuang P, L’Hernault SW: The presenilin protein family member SPE-4 localizes to an ER/Golgi derived organelle and is required for proper cytoplasmic partitioning during Caenorhabditis elegans spermatogenesis. J Cell Sci 1998; 111:3645–3654.

Smialowska /Baumeister

Original Paper

Diseases

Neurodegenerative Dis 2006;3:233–238 DOI: 10.1159/000095261

Functional Role of the Low-Density Lipoprotein Receptor-Related Protein in Alzheimer’s Disease Elaine Waldron Sebastian Jaeger Claus U. Pietrzik Institute of Physiological Chemistry and Pathobiochemistry, Johannes Gutenberg University Mainz, Mainz, Germany

Key Words Alzheimer’s disease Low-density lipoprotein receptor-related protein Amyloid precursor protein

Abstract Alzheimer’s disease (AD) is the most common age-related neurodegenerative disorder, characterized by neuronal loss, neurofibrillary tangle formation and the extracellular deposition of amyloid- (A) plaques. The amyloid precursor protein (APP) and the enzymes responsible for A generation seem to be the base elements triggering the destructive processes. Initially, the low-density lipoprotein receptor-related protein (LRP) was genetically linked to AD and later it emerged to impact on many fundamental events related to this disease. LRP is not only involved in A clearance but is also the major receptor of several AD-associated ligands, e.g. apolipoprotein E and 2-macroglobulin. APP processing is mediated by LRP on many levels. Enhanced APP internalization through LRP decreases cell surface APP levels and thereby reduces APP shedding. As a consequence of increased APP internalization LRP enhances A secretion. These effects could be attributed to the cytoplasmic tails of LRP and APP. The receptors bind via their NPXY motifs to the two PID domains of FE65 and form a tripartite complex. However, it appears that the second NPVY motif of LRP is the one re-

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0233$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

sponsible for the observed influence over APP metabolism. A more in-depth knowledge of the mechanisms regulating APP cleavage may offer additional targets for therapeutic intervention. Copyright © 2006 S. Karger AG, Basel

The low-density lipoprotein receptor-related protein (LRP) is a member of the low-density lipoprotein (LDL)receptor gene family. It was discovered during a screen of cDNA libraries for proteins that contain the cysteine-rich amino acid type repeats present in the ligand-binding domain of the LDL receptor. In 1988, LRP was cloned and later identified as the 2-macroglobulin (2M) receptor. The other seven family members include megalin, verylow-density lipoprotein receptor, apolipoprotein E receptor 2 (ApoER2/LRP8), sorLA-1/LR11, LRP5, LRP6, and LRP1B. Following its synthesis in the ER as a 600-kDa type I transmembrane glycoprotein, LRP is cleaved in the Golgi compartment by furin producing two subunits, which are 515 kDa and 85 kDa in size. These two subunits remain noncovalently associated during their transport to the cell surface. Via a receptor-recycling pathway, LRP is responsible for the endocytosis of more than 30 ligands. Among the LDLR family members, LRP has the fastest rate of endocytosis. The C-terminal domain of LRP har-

Claus U. Pietrzik Institute of Physiological Chemistry and Pathobiochemistry Johannes Gutenberg University Mainz, Duesbergweg 6 DE–55099 Mainz (Germany) Tel. +49 6131 392 5390, Fax +49 6131 392 6488, E-Mail [email protected]

Fig. 1. Schematic representation of the en-

zymes and cleavages involved in APP and LRP metabolism. BACE catalyzes the initial cleavage of APP, releasing its extracellular domain (APPs) and leaving behind the C-terminal fragment. The same or a similar sheddase catalyzes the cleavage in the LRP extracellular domain. The remaining fragments of both proteins become substrates for -secretase, which generates A and AICD from APP and LICD from LRP.

bors two NPXY motifs, the second of which overlaps with a YXXL domain. The distal NPXYXXL motif in conjunction with the C-terminal LL domain appears to be important in receptor uptake. Early immunoelectron microscopy studies indicated that the majority of LRP in the plasma membrane is localized within coated pits. Furthermore, LRP does not bind LDL and thus is incapable of compensating for the lack of LDL receptor function in LDL receptor-deficient individuals [for reviews, see 1–4]. Hence, the role of LRP in lipoprotein metabolism is much more restricted than that of the LDL receptor, since the latter can fully compensate for the lack of functional LRP in the liver [5]. In addition to its role in endocytosis of a variety of ligands, LRP has been implicated to play a crucial role in cell signaling. Several adaptor proteins involved in signaling cascades can bind to the LRP tail, some of which seem to play a role in neuronal calcium signaling, PDGF signaling [2] or MAPK signaling [6]. This multifunctionality might explain the early embryonic lethality of mouse embryos in which the LRP gene has been disrupted in all cells by conventional gene targeting. Although most of these embryos died during the peri-implantation period, a few were found to survive to approximately embryonic day 10 [7]. Recently, it has become clear that LRP imparts an influence over amyloid precursor protein (APP) processing and thus contributes to AD pathogenesis. While the evidence supporting a role for LRP in AD pathology continues to grow and become more diverse, much remains to be learned. Here, we summarize the evidence implicating LRP as a key participant in AD pathogenesis, which later focuses on the impact of LRP on APP processing. LRP is highly expressed in neurons, where it is predominantly 234

Neurodegenerative Dis 2006;3:233–238

found in neuronal cell bodies and dentritic processes [1]. In neurons, LRP is a major receptor for ApoE/lipoprotein-containing particles, and the 4 allele of ApoE is a strong genetic risk factor for late-onset AD. LRP-mediated brain metabolism of ApoE/lipoprotein-containing particles can also influence the metabolism of cholesterol, which has also been suggested to contribute to the pathogenesis of AD [8]. LRP and several of its ligands (APP, ApoE, 2M, all of which are found within senile plaques) have been genetically associated with AD [9]. The conformational state of ApoE4, namely its capacity for domain interaction with itself, appears to impart neuropathological effects, i.e. increases in A production, potentiation of A-induced lysosomal leakage and apoptosis, and enhanced proteolytic cleavage in neurons [10]. Additionally, its conformational state, which may be brought about through association with lipoproteins, influences its affinity for its receptor. ApoE must be present in large amounts on lipoprotein-containing particles in order to be recognized by LRP [10]. So far, it appears that the different conformations of ApoE alleles induced by its enrichment on lipoprotein particles or its own self-assembly determine the binding abilities by LRP [11]. Furthermore, it has come to knowledge that LRP may also undergo proteolysis by the enzymes responsible for APP cleavage. The protease BACE1, which generates the A N-terminus through the cleavage of APP was shown to be capable of cleaving the extracellular domain of LRP. In addition, the aspartyl protease -secretase, which releases A and the APP intracellular domain (AICD) by cleavage at the C-terminus of the A sequence, is also responsible for the cleavage of LRP releasing its intracellular domain (LICD) (fig. 1) [3]. This cleavage was coined Waldron/Jaeger/Pietrzik

the S3 cleavage or -cleavage event with the products, AICD and LICD, being analogous to the Notch intracellular domain NICD. All of the aforementioned intracellular tails have been suggested to have important signaling roles, particularly in transcription. The last 105 amino acids of the LRP -chain (LPR105) inhibited transcriptional activity of the AICD, FE65, and Tip60 complex in a heterologous signaling system, presumably by interfering with the interaction between AICD and Tip60. It is worth noting that LRP has been shown to mediate clearance of A-2M and A-ApoE complexes in vitro, which led us to propose that LRP may contribute to AD pathogenesis through modulating the clearance and hence levels of A in vivo [12]. Clearance of A by LRP has also been implicated in the transport of A out of the brain parenchyma via the blood-brain barrier [13]. The importance of A clearance via an LRP-mediated pathway is supported by a significant increase in amyloid deposition seen in transgenic APP mice deficient in receptor-associated protein, a well known inhibitor of ligand binding to all receptors of the LDL receptor family. These mice show an 80% reduction in LRP levels due to receptor misfolding and accelerated degradation [3]. Current in vitro and in vivo data, together with previously reported genetic associations of ApoE, 2M, and LRP to AD, strongly implicate a dysfunction in soluble A clearance in AD pathogenesis. A more in-depth understanding of A-clearance via an LRP-mediated pathway might lead to new therapeutic strategies that could ultimately lessen the A-burden seen in AD postmortem brains. We and others have demonstrated LRP’s effectiveness over A production through an impact on APP processing [14, 15]. When LRP binding was blocked by receptorassociated protein in LRP-deficient or LRP wild-type cells stably expressing APP751, A secretion was dramatically reduced in both cases [14]. From this the authors proposed that the association of APP via its KPI domain to LRP at the cell surface modulates APP internalization and hence accounts for the alteration in A secretion. However, no biochemical data supporting this interaction had been provided by these or other investigators. On the other hand, evidence for an interaction between the two receptors through their cytoplasmic domains had come into place as both tails of LRP and APP were reported to interact independently of each other with the cytoplasmic adaptor proteins FE65 and Dab-1 [2]. It was thereafter proposed that the respective cytoplasmic regions may communicate with each other via these adaptor molecules. In agreement with this hypothesis, our Role of LRP in APP Processing

Fig. 2. Schematic representation of the trimeric complex consisting of APP, FE65 and LRP. Binding of the PID domains of FE65 to the NPXY motifs of the two receptors accounts for their assembly. Although FE65 can bind to both NPXY motifs in the C-terminus of LRP, the more distal one mediates the effects on APP processing.

studies with APP751 and APP695 showed the LRP-mediated influence over APP processing to be dependent on the cytoplasmic tails and independent of the KPI domain. While LRP-APP complexes can be detected, APP695 is not known to directly bind to LRP. As FE65 interacts with both LRP and APP independently and colocalizes with both proteins, a scenario was proposed where FE65 bridges LRP and APP, allowing the formation of a trimeric complex (fig. 2). FE65 has two colinear PID-binding domains that bind LRP and APP. Although the concept that FE65 can act as a scaffold linking APP and LRP is very attractive, no biochemical data had been provided. Interestingly, we were able to demonstrate that both LRP NPXY motifs interact with FE65 [16]. Bearing our previous results in mind, which showed the distal NPVY motif to be critical for APP processing, the interaction between the second NPVY motif of LRP and the first PID domain of FE65 is the bona fide complex combination, or at least the more important one responsible for the effects seen on APP processing [15]. Further evidence supporting this hyNeurodegenerative Dis 2006;3:233–238

235

pothesis came from our observation that the deletion around the second NPXY motif resulted in increased APPs and decreased A secretion. This decrease in A can be correlated to a reduction in APP internalization from the cell surface. The impairment in APP internalization is in agreement with the hypothesis that endocytic processing is a major contributor to A release. Similarly, the increase in APP secretion is consistent with previous reports showing that loss of the APP endocytic signal elevates the levels of APPs in medium [17]. These changes in APP metabolism mimic the APP processing phenotype in LRP-deficient cells, indicating that internalization of cell surface APP is altered when the endocytosis signal in the LRP sequence has been changed. However, deletion of mutations around the first NPXY motif had no such influence on APP processing. Hence, we concluded that the LRP-mediated effect on APP processing is brought about by the second NPXY motif, probably by reduced LRP endocytosis and consequently reduced APP internalization. However, in a recently published article by Roebroek et al. [18] the authors have demonstrated that knock-in mutations in the distal NPXY motif do not significantly alter the development and fertility of the mice, whereas mutations in the proximal NPXY motif show dramatic effects on mice development. This finding is most interesting, since it has been proposed that most cellular functions of LRP, as shown for the endocytosis of ligands and APP processing are mediated through the distal NPXY motif. In regard to AD research, this finding might be of benefit, since one could argue that tampering with the distal NPXY binding motif of LRP might influence APP processing, leaving the physiological relevant aspects of LRP untouched. To provide a mechanistic explanation for the LRP-mediated influence over APP processing, one would anticipate a protein-protein interaction between the two proteins to be accountable. Since no direct interaction between the C-terminal intracellular domains of APP and LRP had been shown, we investigated the possibility that FE65 might act as a link molecule between the two proteins. We demonstrated that FE65 does indeed act as a bridge between LRP and APP (fig. 2). Firstly, in pulldown assays, an APP C-terminal fusion protein was able to precipitate cell-associated LRP only when intact FE65 was also expressed but not with an FE65 construct lacking the second PID motif. This indicated that both PID domains of FE65 are required to bridge APP with LRP. Secondly, an LRP C-terminal fusion protein was able to pull down in vitro translated APP695 only when FE65 was also co-translated but not when FE65 was omitted. 236

Neurodegenerative Dis 2006;3:233–238

Thirdly, APP-LRP complexes can be coimmunoprecipitated in HS683 cells, but interestingly the levels of APPLRP immune complexes were decreased by overexpression of either FE65 or mutant FE65 lacking the second PID domain. Together, our findings provided strong evidence that FE65 alone is sufficient to function as a scaffold for APP and LRP [16]. This observation sheds light on the different results which have been published using FE65-overexpressing cells [19]. Since FE65 overexpression in different cell lines caused contradictory results regarding A secretion, endogenous LRP expression levels might explain the findings from these cell lines. In cells expressing high LRP levels, an overexpression of FE65 might lead to the formation of more APP-FE65LRP complexes and therefore to an increased internalization rate; hence one would expect an increase in A secretion. In contrast, a low endogenous LRP expression level in FE65-overexpressing cells would result in a disruption of this tripartite complex, since every LRP and APP molecule would be able to bind one single FE65 molecule. This would result in a virtual LRP knockout phenotype, causing a decrease in A secretion, since FE65 would be incapable of bridging LRP and APP. The latter is in line with a double-transgenic FE65/APP mouse model expressing endogenous LRP, showing decreased A secretion and reduced amyloid plaque formation [19]. Further confirmation of the hypothesis that LRP mediates APP internalization through FE65 came from a collaborative effort with the group of Stefan Lichtenthaler, where we observed that APLP1 overexpression resulted in an increased APP secretion and reduced internalization of surface APP. Since APLP1 harbors the same binding motif for FE65 as APP, APLP1 overexpression results in the seizure of most of the available FE65, such that less is accessible for binding to endogenous APP [20]. Regarding APP, this would account for a virtual LRP knockout phenotype causing a decrease in A secretion. Hence, by offering more binding partners for FE65 one alters the stoichiometry of the LRP-FE65-APP complex, which finally leads to the disruption of the latter, resulting in a virtual LRP knockout phenotype regarding APP processing. In addition, competition with other adaptor proteins may result in a total complex disruption and hence an LRP knockout phenotype. Indeed, X11 overexpression results in decreased A secretion [21]. Although the original explanation of the authors differs from this hypothesis, it would be feasible to assume that X11 competes with FE65 for the NPXY-binding site, resulting in a disruption of the LRP-FE65-APP complex.

Waldron/Jaeger/Pietrzik

Since the initial discovery that Notch and APP are cleaved by -secretase, multiple substrates that undergo -secretase cleavage have been identified, including ErbB4, E- and N-cadherins, CD44, nectin-1, the Notch ligands Delta and Jagged, and LRP [22]. Although all are cleaved by PS1-dependent -secretase activity, it might be expected that the -secretase complex is composed of different subunits depending on the substrate. Recent biochemical evidence points to the existence of an initial substrate-binding site in -secretase distinct from the PS1 substrate binding site. In principle, they might be substrates of different secretase complexes or they might compete for a single docking site on the same secretase complex. For example, APP and PS1 remain in close proximity in the presence of -secretase transition state analogues, measured by FRET. LRP behaves much like APP in terms of its interactions with -secretase: strong FRET between LRP and PS1 can be detected and this close proximity persists after treatment with -secretase active site inhibitors. These data are consistent with the notion of an initial substrate-binding site in or in close proximity to PS1. Additionally, coimmunoprecipitation experiments revealed the binding of LRP only to the autocatalytically processed N-terminal fragment of PS1. Besides reducing A40 and A42, overexpression of LRP-derived constructs reduced the p3 peptide indepen-

dent of APP internalization. However, an inverse relation between the levels of expression of LRP constructs and the levels of A has been observed. Taken together, the latter indicates a competition between LRP and APP for the enzymatic cleavage site [23]. It is clear that LRP influences APP metabolism through the formation of a tripartite complex, and breaking up this complex should reduce APP internalization and hence A production. In vivo confirmation of the LRP impact on APP processing and a deeper understanding of the nature of ApoE4 might make the disruption of the APP-Fe65-LRP complex as well as the binding of ApoE to LRP an interesting target to prevent AD. T validate LRP as a therapeutic target resides in revealing the aftermath of the APP-FE65-APP complex disruption in vivo. Over the last few years, it has become clear that LRP physiology expands beyond the function of lipid uptake. It has become more evident that signal transduction processes play a crucial role in LDL receptor biology. More refined insights into the diverse molecular mechanisms by which LRP exerts its diverse functions come from observations that link LRP to AD. Therefore, its involvement in AD is not yet fully understood; LRP is likely to play an important role as a major neuronal receptor for ApoE and 2M in APP processing and as the potential modulator of A aggregation through A clearance.

References 1 Herz J, Bock HH: Lipoprotein receptors in the nervous system. Annu Rev Biochem 2002;71:405–434. 2 May P, Herz J, Bock HH: Molecular mechanisms of lipoprotein receptor signalling. Cell Mol Life Sci 2005;62:2325–2338. 3 Zerbinatti CV, Bu G: LRP and Alzheimer’s disease. Rev Neurosci 2005;16:123–135. 4 Krieger M, Herz J: Structures and functions of multiligand lipoprotein receptors: macrophage scavenger receptors and LDL receptor-related protein (LRP). Annu Rev Biochem 1994;63:601–637. 5 Rohlmann A, Gotthardt M, Hammer RE, Herz J: Inducible inactivation of hepatic LRP gene by cre-mediated recombination confirms role of LRP in clearance of chylomicron remnants. J Clin Invest 1998; 101: 689– 695. 6 Hu K, Yang J, Tanaka S, Gonias SL, Mars WM, Liu Y: Tissue-type plasminogen activator acts as a cytokine that triggers intracellular signal transduction and induces matrix metalloproteinase-9 gene expression. J Biol Chem 2005;281:2120–2127.

Role of LRP in APP Processing

7 Herz J, Clouthier DE, Hammer RE: LDL receptor-related protein internalizes and degrades uPA-PAI-1 complexes and is essential for embryo implantation. Cell 1992; 71: 411– 421. 8 Burns M, Duff K: Use of in vivo models to study the role of cholesterol in the etiology of Alzheimer’s disease. Neurochem Res 2003; 28:979–986. 9 Hyman BT, Strickland D, Rebeck GW: Role of the low-density lipoprotein receptor-related protein in beta-amyloid metabolism and Alzheimer disease. Arch Neurol 2000; 57:646–650. 10 Mahley RW, Weisgraber KH, Huang Y: Apolipoprotein E4: a causative factor and therapeutic target in neuropathology, including Alzheimer’s disease. Proc Natl Acad Sci USA 2006;103:5644–5651. 11 LaDu MJ, Stine WB Jr, Narita M, Getz GS, Reardon CA, Bu G: Self-assembly of HEK cell-secreted ApoE particles resembles ApoE enrichment of lipoproteins as a ligand for the LDL receptor-related protein. Biochemistry 2006;45:381–390.

12 Kang DE, Pietrzik CU, Baum L, Chevallier N, Merriam DE, Kounnas MZ, Wagner SL, Troncoso JC, Kawas CH, Katzman R, Koo EH: Modulation of amyloid beta-protein clearance and Alzheimer’s disease susceptibility by the LDL receptor-related protein pathway. J Clin Invest 2000;106:1159–1166. 13 Tanzi RE, Bertram L: Twenty years of the Alzheimer’s disease amyloid hypothesis: a genetic perspective. Cell 2005;120:545–555. 14 Ulery PG, Beers J, Mikhailenko I, Tanzi RE, Rebeck GW, Hyman BT, Strickland DK: Modulation of beta-amyloid precursor protein processing by the low density lipoprotein receptor-related protein (LRP). Evidence that LRP contributes to the pathogenesis of Alzheimer’s disease. J Biol Chem 2000;275:7410–7415. 15 Pietrzik CU, Busse T, Merriam DE, Weggen S, Koo EH: The cytoplasmic domain of the LDL receptor-related protein regulates multiple steps in APP processing. EMBO J 2002; 21:5691–5700.

Neurodegenerative Dis 2006;3:233–238

237

16 Pietrzik CU, Yoon IS, Jaeger S, Busse T, Weggen S, Koo EH: FE65 constitutes the functional link between the low-density lipoprotein receptor-related protein and the amyloid precursor protein. J Neurosci 2004;24:4259– 4265. 17 Koo EH, Squazzo SL, Selkoe DJ, Koo CH: Trafficking of cell-surface amyloid beta-protein precursor. I. Secretion, endocytosis and recycling as detected by labeled monoclonal antibody. J Cell Sci 1996;109:991–998. 18 Roebroek AJ, Reekmans S, Lauwers A, Feyaerts N, Smeijers L, Hartmann D: Mutant Lrp1 knock-in mice generated by recombinase-mediated cassette exchange reveal differential importance of the NPXY motifs in the intracellular domain of LRP1 for normal fetal development. Mol Cell Biol 2006; 26: 605–616.

238

19 Santiard-Baron D, Langui D, Delehedde M, Delatour B, Schombert B, Touchet N, Tremp G, Paul MF, Blanchard V, Sergeant N, Delacourte A, Duyckaerts C, Pradier L, Mercken L: Expression of human FE65 in amyloid precursor protein transgenic mice is associated with a reduction in beta-amyloid load. J Neurochem 2005;93:330–338. 20 Neumann S, Schobel S, Jager S, Trautwein A, Haass C, Pietrzik CU, Lichtenthaler SF: Amyloid precursor-like protein 1 influences endocytosis and proteolytic processing of the amyloid precursor protein. J Biol Chem 2005;281:7583–7594.

Neurodegenerative Dis 2006;3:233–238

21 Sastre M, Turner RS, Levy E: X11 interaction with beta-amyloid precursor protein modulates its cellular stabilization and reduces amyloid beta-protein secretion. J Biol Chem 1998;273:22351–22357. 22 Haass C: Take five – BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 23 Lleo A, Waldron E, von Arnim CA, Herl L, Tangredi MM, Peltan ID, Strickland DK, Koo EH, Hyman BT, Pietrzik CU, Berezovska O: Low density lipoprotein receptor-related protein (LRP) interacts with presenilin 1 and is a competitive substrate of the amyloid precursor protein (APP) for gammasecretase. J Biol Chem 2005; 280: 27303– 27309.

Waldron/Jaeger/Pietrzik

Original Paper

Diseases

Neurodegenerative Dis 2006;3:239–246 DOI: 10.1159/000095262

The Functions of Mammalian Amyloid Precursor Protein and Related Amyloid Precursor-Like Proteins Brigitte Anliker a, 1 Ulrike Müller a, b a

Department of Neurochemistry, Max Planck Institute for Brain Research, Frankfurt, Department of Neuroinformatics and Functional Genomics, Institute for Pharmacy and Molecular Biotechnology, University of Heidelberg, Heidelberg, Germany

b

Key Words Alzheimer’s disease -Amyloid precursor protein, in vivo function Amyloid precursor-like protein Knockout mice Functional redundancy

Abstract It is well established that proteolytic processing of the -amyloid precursor protein (APP) generates -amyloid which plays a central role in the pathogenesis of Alzheimer’s disease. In contrast, the physiological role of APP and the question of whether a loss of these functions contributes to Alzheimer’s disease are still unclear. For a long time, the characterization of APP functions was markedly hampered by the high redundancy between APP and the related APP family members amyloid precursor-like proteins 1 and 2. The generation and analyses of combined gene deficiencies for APP and amyloid precursor-like proteins in mice finally marked the beginning of uncovering the in vivo roles of these proteins in mammals. In the current review, we summarize recent insights into the functions of the APP gene family from mice lacking one, two or all three family members. Copyright © 2006 S. Karger AG, Basel 1 Present address: Department of Molecular Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037 (USA).

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0239$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Introduction

The pivotal role of the -amyloid (A) peptide in the pathogenesis of Alzheimer’s disease (AD) put the -amyloid precursor protein (APP) and its proteolytic processing in the focus of intense research. APP is an integral type 1 membrane protein with a large extracellular domain, a single transmembrane region and a short cytoplasmic domain [1]. Processing by several secretases results in diverse extra- and intracellular APP fragments [1]. Cleavage by - or -secretase secretes the soluble ectodomain of APP (APPs- and APPs-) and generates C-terminal stubs (CTF and CTF). Subsequent intramembrane processing of CTFs by -secretase releases A and the APP intracellular domain AICD. -Secretase, in contrast, cleaves APP within the A region, thereby preventing the formation of A. It is widely accepted, that aberrant processing of APP accompanied by increased production and aggregation of A peptides is crucial in AD and thus, the secretases involved in APP processing constitute major therapeutic targets [1]. The in vivo role of APP and its fragments, on the other hand, is not well understood, and therapeutic interference with APP processing is likely to alter physiological APP functions. In addition, it has been proposed that reductions in the level or activity of certain APP fragments may contribute to the cognitive dysfunction associated

Ulrike Müller Department of Neuroinformatics and Functional Genomics Institute for Pharmacy and Molecular Biotechnology University of Heidelberg, INF 364, DE–69120 Heidelberg (Germany) Tel. +49 6221 546 717, Fax +49 6221 545 891, E-Mail [email protected]

with AD [2]. To address these important questions directly, we and others generated mice deficient for APP either alone or in combination with other members of the APP gene family. Here, we focus on recent insights into the biological functions of the APP gene family from these mouse mutants. For a more general overview including functional studies in Caenorhabditis elegans and Drosophila we refer the reader to a recent review [3].

The APP Gene Family

The APP family is an evolutionary conserved gene family comprising APL-1 in C. elegans, APPL in Drosophila, and the amyloid precursor-like proteins, APLP1 and APLP2, in mammals [3]. APP and both APLPs are highly related proteins sharing conserved regions over large portions of the ectodomain and within the intracellular tail [3]. Interestingly, APLP1 and APLP2 lack the A region. Nevertheless, they are similarly processed by -, -, and -secretases as APP [4]. Likewise, APP and APLP2 expression are highly similar showing ubiquitous and largely overlapping expression patterns during embryonic development and in adulthood [5]. APLP1 expression, in contrast, is restricted to the nervous system [5]. The overall extensive similarities between APP and APLPs suggested functional conservation between the family members. Indeed, the generation and analysis of APP/APLP single- and doubleknockout mice revealed partially redundant functions of APP and APLPs. This redundancy considerably complicated the reverse genetic approaches to unravel the in vivo functions of APP.

APP/APLP Loss of Function Studies

Three APP mouse mutants, one carrying a hypomorphic mutation of APP (APP) [6] and two with complete deficiencies of APP [7, 8] have been generated by gene targeting. All mutants were viable, fertile, and revealed comparable phenotypes (table 1). Disruption of APP functions resulted in retarded somatic growth, reduced grip strength, reduced brain weight, reduced size of forebrain commissures, agenesis of the corpus callosum in a 129SvEv genetic background, reactive gliosis, increased sensitivity to kainate-induced seizures, increased copper levels in the cerebral cortex and the liver, impaired long-term potentiation and behavioral deficits 240

Neurodegenerative Dis 2006;3:239–246

such as reduced locomotor activity, reduced exploratory behavior and impaired spatial learning [6–14]. Whether APP also plays a crucial role in kinesin-mediated axonal transport is currently controversial [15, 16]. In vitro analyses of cells derived from APP knockouts revealed a slightly reduced survival rate and impaired neurite outgrowth in APP-deficient hippocampal neurons [17]. In contrast, cortical neurons lacking APP showed normal survival rates and neurite outgrowth [18, 19]. Recently, a role of APP in the cell cycle progression has been suggested since APP deficiency increased the duration of late G2 and mitosis [20]. Although APP-deficient mice are overall slightly retarded in their development and exhibit behavioral deficits, the observed effects were unexpectedly mild in view of the widespread expression and evolutionary conservation of APP. To test the hypothesis that a loss of APP functions is compensated by the related APLP proteins, we subsequently generated APLP-deficient mice which were intercrossed to finally generate a complete set of combined APP/APLPs deficiencies. APLP1-deficient mice revealed a somatic growth deficit as the only abnormality [18], whereas APLP2 knockout mice appeared normal [21]. Combined gene deficiencies for APP/APLP2 or APLP1/APLP2, in contrast, resulted in perinatal lethality (table 1) [18, 21]. Both double mutants were born at the expected Mendelian frequencies, but died shortly after birth with a penetrance of 80–100% for APP/APLP2 and 100% for APLP1/APLP2 double knockouts [18, 21]. The few APP/APLP2-deficient mice surviving into adulthood revealed an enhanced deficit of somatic growth compared to APP single knockouts, and some behavioral abnormalities including difficulty in righting, ataxia, spinning and head tilting [18, 21]. Unlike the lethal phenotypes of APP/APLP2 and APLP1/APLP2 double knockouts, the third combination of dual gene deficiency, APP/APLP1, turned out to be viable, fertile, and without any additional abnormalities as compared to single knockouts (table 1) [18]. An upregulation of the remaining APP and/or APLP proteins was not observed in single and double knockouts at P0 [18]. In older APP-deficient mice, however, both APLP1 and APLP2 protein levels were increased in brain in an age-dependent manner, suggesting compensatory mechanisms relevant for aging [22]. In summary, these findings demonstrate functional redundancy between the APP family members whose physiological roles proved to be essential for early postnatal development. Whereas a loss of single family members is compensated by the remaining APP/APLP proteins resulting in minor phenotypes, dual gene deficienAnliker/Müller

Table 1. Phenotypes of APP/APLP knockout mice Genotype

Viability/fertility

Phenotype

Cellular effects

APP–/–

Viable and fertile [6, 7]

Reduced body weight [6, 7, 11] Reduced grip strength [6, 7, 11] Reduced brain weight [12] Reduced size of forebrain commissures [12] Agenesis of the corpus callosum (129SvEv background) [6, 12] Reactive gliosis [7, 9, 13] Increased sensitivity to kainate-induced seizures [10] Increased copper levels in the cerebral cortex and the liver [14] Increased cholesterol and sphingomyelin levels in brain [37] Impaired long-term potentiation [9, 13] Behavioral deficits (reduced locomotor activity, reduced exploratory behavior, impaired spatial learning) [6, 7, 9, 11]

Reduced survival rate of hippocampal neurons [17] Impaired neurite outgrowth of hippocampal neurons [17] Alteration in cell cycle progression of cortical precursor cells (lengthening of G2/mitosis) [20] Defective Ca2+ mobilization from internal Ca2+ stores in MEFs [32] Increased cholesterol and sphingomyelin levels in brain [37]

APLP1–/–

Viable and fertile [18]

Reduced body weight [18]

–/–

APLP2

Viable and fertile [21]

No apparent phenotype [21] No histological abnormalities [21]

APP–/– APLP1–/–

Viable and fertile [18]

Reduced body weight [18] No histological abnormalities [18]

APP–/– APLP2–/–

Perinatally lethal [18, 21] (80–100% penetrance)

Defective formation of neuromuscular junctions and impaired synaptic transmission [24] Surviving mice revealed a reduction of body weight and behavioural deficits (difficulty in righting, ataxia, spinning and head tilting) [18, 21]

APLP1–/– APLP2–/–

Perinatally lethal [18] (100% penetrance)

APP–/– APLP1–/– APLP2+/–

Perinatally lethal [18, 23] (95–98% penetrance)

APP–/– APLP1–/– APLP2–/–

Perinatally lethal [23] (100% penetrance)

Increased cholesterol and sphingomyelin levels in MEFs [37] Decreased intercellular adhesion on MEFs under Ca2+-free conditions [22] Largely diminished expression and activity of neprilysin in MEFs [31]

No histological abnormalities [18]

Cortical dysplasia resembling human type II lissencephaly (68% penetrance) and other cranial abnormalities [23] Reduced survival rate of CR cells in the embryonic cortex at E18.5 [23]

cies involving APLP2 are lethal. The unique viability of APP/APLP1-deficient mice indicates that APLP2 is the only family member that can compensate for a simultaneous loss of both other APP family members. This physiological key role of APLP2 is further corroborated by the lethality of APP–/–APLP1–/–APLP2+/– mice revealing haploinsufficiency of a single APLP2 allele in the absence of APP and APLP1 [18, 23].

Functions of APP and APLPs

Decreased intercellular adhesion of MEFs under Ca2+-free conditions [22]

Roles of APP/APLP Proteins in the Formation of Neuromuscular Synapses and in Cortical Development

Although the reverse genetic studies clearly revealed an essential role of APP and APLPs during early postnatal development, the precise molecular nature of this function is still largely unknown. In spite of an intense effort to identify the reason for the lethality of APP/ APLP2 and APLP1/APLP2 double knockouts, no major Neurodegenerative Dis 2006;3:239–246

241

Fig. 1. Cortical dysplasia in APP–/–

APLP1–/–APLP2–/– triple-knockout mice. a, b Frontal section (cresyl violet staining) of a triple-knockout mouse (T) at E18.5 exhibiting a prominent protrusion (P) of the right hemispheric cortical plate. b Upon higher magnification (boxed area in a) it becomes apparent that ectopic neurons completely disrupt the subplate (SP) and cortical plate (CP) as cells have migrated into and beyond the marginal zone (MZ). Adapted from Herms et al. [23].

histological abnormalities in the central nervous system or in peripheral organs were observed that could be associated with lethality [18, 21]. Very recently, however, detailed analyses of lethal double and even triple knockouts revealed for the first time distinct in vivo functions of the APP gene family. Immunohistological and ultrastructural analyses of neuromuscular junctions (NMJs) in APP/APLP2 doubleknockout mice uncovered an important role of APP and APLP2 in the development of neuromuscular synapses [24]. The NMJs of APP/APLP2 double-knockout mice were characterized by aberrant presynaptic structures consisting of excessive nerve terminal sprouting and an impaired apposition of presynaptic marker proteins with postsynaptic acetylcholine receptors. Furthermore, neurotransmission was severely impaired as evidenced by a highly reduced frequency of miniature endplate potentials associated with a reduction in synaptic vesicle density [24]. In contrast to these striking abnormalities at neuromuscular synapses, the density, ultrastructure and vesicle number of synapses in the developing brainstem of newborn mutants were comparable to wild-type control mice [18]. It remains to be seen whether death of lethal APP/APLP double knockouts is somehow linked to defective synaptic transmission at NMJs. Interestingly, partially overlapping defects at NMJs have been observed in mice deficient for the acetylcholine-synthesizing enzyme choline acetyltransferase (ChAT) [25, 26]. Although ChAT-deficient mice are perinatally lethal as well, the cause of death may be different in the two mutants, since ChAT–/– mice are most likely dying from absent neuromuscular transmission (e.g. in the diaphragm), are unable to move and exhibit a hunched posture. Newborn APP/ APLP2 and APLP1/APLP2 double knockouts, in contrast, are initially able to breath, suckle, and appear healthy for several hours before they weaken and die [18]. 242

Neurodegenerative Dis 2006;3:239–246

To unravel the full range of physiological APP/APLP functions, all three proteins were finally eliminated simultaneously in mice. Similarly to the lethal double knockouts, triple mutants survive through embryonic development but die shortly after birth [23]. Unlike the double mutants which did not display histological alterations in the brain, 80% of all analyzed triple knockouts showed cranial abnormalities (table 1) [23]. Most frequently, focal cortical dysplasia resembling human type II lissencephaly was observed. Within affected areas, neuronal cells from the cortical plate migrated beyond their normal positions and protruded into the marginal zone and the subarachnoid space (fig. 1). It is noteworthy that ectopic clusters were very focal and found exclusively within dorsal regions of the frontoparietal cortex. Analyses of triple knockouts at different embryonic stages and BrdU incorporation experiments indicated that ectopic neuronal clusters originate around E12–14 and progressively become bigger with increasing age [23]. The cortical cytoarchitecture underneath the neuronal protrusions was completely disrupted and staining patterns for several neuronal marker proteins were highly disorganized [23]. The basement membrane was either fragmented or missing within the protrusion zones. In contrast, reticulin and laminin immunostainings of the basal lamina indicated an unaltered structure of the basement membrane outside the protrusions [23]. Likewise, immunohistological characterization of radial glial cells, which guide migrating neurons from the ventricular zone to their final positions in the cortical layers, revealed disturbed interactions between basal lamina and radial glia exclusively within protrusions. The only consistent alteration along the entire cortex of triple mutants was a reduction in Cajal-Retzius (CR) cells [23]. The total number of CR cells was reduced by about 37% in APP/APLP1/APLP2-deficient mice compared to control mice. CR cells secrete reelin, a protein Anliker/Müller

APP/APLP deficiency in meningeal fibroblasts

APP, APLPs

APP, APLPs

Ligand? ECM component? Integrins

PS1

tin s

Ac

Mena N P T Fe65 Y Tip60

-

W W

nt

me

N mDab1 P T PID2 X11 Y Fe65

fi la

C N

ti n

N P T Y

in

-Secretase

Ac

Fig. 2. Model of putative APP/APLP-mediated mechanisms involved in heterotopia formation. -Secretase/PS1 activity liberates the APP/APLP intracellular domains (AICDs) which form transcriptionally active complexes with the adaptor protein Fe65 and the histone acetyltransferase Tip60. These complexes may regulate expression of genes (e.g. in meningeal fibroblasts or neurons) involved in neuronal adhesion and positioning. Alternatively, membrane-integrated APP/APLP proteins may transduce external signals through intracellular interactions between the NPTY motif and macromolecular signaling complexes. Fe65 may link APP/APLPs to Mena which is involved in actin reorganization. Integrins, which are localized with APP/Fe65/Mena at adhesion complexes, are indirectly associated with the actin cytoskeleton through actin-binding proteins such as -actinin. Immediate proximity to integrins at sites of dynamic actin turnover may allow the APP/Fe65/ Mena complexes to modulate integrinbased cell adhesion. Fe65 competes with the adapter proteins X11/mint1 and mDab1 for binding to the NPTY motif. X11/mint1 may represent another link of APP/APLPs to cell adhesion via interaction with CASK, whereas binding to mDab1, a crucial player in the reelin signaling pathway, may provide a further platform for APP/APLP-mediated signals.

Cell adhesion

Cell motility Actin remodeling

Gene expression

Neuronal adhesion, positioning within cortical layers

that is critical for neuronal migration and accurate positioning of neurons in the cortical layers [27]. A reduction in CR cells may induce local changes of reelin secretion resulting in neuronal overmigration. Intriguingly, very similar neuronal ectopias and loss of CR neurons have been observed in mice lacking presenilin-1 (PS1) [28]. PS1 is an essential component of the -secretase complex involved in APP/APLP processing [1]. In PS1-deficient mice, the loss of CR cells is presumably due to defective trophic support of CR cells by meningeal fibroblasts as leptomeninges became fibrotic [28, 29]. Since meninges do not only express PS1 during cortical neuronal migration but also APP/APLP proteins [5, 28], it is possible that loss of CR cells and ectopia are linked to lack of -secretase-mediated processing of APP and APLPs in these cells. An attractive hypothesis is that APP/APLP intracellular C-terminal fragments generated by -secretase activity, socalled ‘AICDs’, serve as effector molecules possibly via the formation of transcriptionally active complexes with the

nuclear adaptor protein Fe65 and the histone acetyltransferase Tip60 [30]. AICD/Fe65/Tip60 complexes could thus regulate the expression of target genes involved in trophic support of CR cells and/or neuronal positioning (fig. 2). A physiological role of AICDs in gene expression has recently been provided by showing that the intracellular domains of APP/APLPs regulate the expression level of neprilysin, a key A-degrading enzyme [31]. In addition to gene regulation, AICDs could serve a more general signaling function and induce direct cellular effects such as triggering calcium release from internal stores [32]. Alternatively, APP/APLP proteins might influence neuronal migration and positioning by serving as neuronal cell surface receptors. They could terminate radial migration at the marginal zone by regulating cell motility itself, by interfering with neuron-glia adhesion (thereby inducing detachment of neurons from radial glial cells) or finally, by promoting adhesion with ECM components. Although these possibilities are still hypothetical, several

Functions of APP and APLPs

Neurodegenerative Dis 2006;3:239–246

243

interacting partners of the APP/APLP cytoplasmic domain meet the basic requirements for regulating cell adhesion and migration (fig. 2) [3]. Interaction with the adaptor protein Fe65 may link APP/APLPs to Mena, a regulator of the actin cytoskeleton that is localized at sites of dynamic actin turnover [33]. APP and Mena can interact simultaneously with Fe65, and colocalize with Fe65 in lamellipodia [33]. Overexpression studies in non-neuronal cells indicated that APP and Fe65 can indeed regulate cell migration [33]. In addition, APP and Fe65 were found to specifically localize with 1 integrin at adhesion complexes, indicating a putative role in integrin-based adhesion (fig. 2) [33]. The multidomain adaptor protein X11/ mint1 may also link APP/APLPs to cell adhesion via interaction with CASK (fig. 2) [3]. A role of APP/APLPs in cell adhesion is further supported by biochemical interaction of APP proteins with collagen, heparin, laminin, fasciclin, glypican and F-spondin [3]. Cell adhesion could eventually be mediated via trans-dimerization of membrane-integrated APP/APLP proteins, since a recent collaborative study with Soba et al. [22] indicated trans-interactions of APP and APLPs resulting in homo- and heterocomplexes that promote intercellular adhesion. The same NPTY motif of APP/APLPs that mediates binding to Fe65 and X11/mint1 can also bind the adaptor protein mDab1 (fig. 2) [3]. This protein interacts in a similar way with the intracellular NPxY motif of the reelin receptors ApoER2 and VLDLR and has been shown to play a crucial role in the reelin signaling cascade [34]. It is noteworthy that APP/APLP triple knockouts did not show reeler-like lamination defects as observed in mice lacking reelin, ApoER2 and VLDLR, or mDab1 [34]. Although the lacking reeler-like phenotype in triple knockouts suggests that APP/APLPs are not directly involved in the reelin signaling cascade, it is still possible that aberrant mDab1 signaling contributes to the neuronal overmigration observed in the cortices of triple knockouts. Clearly, we need a better understanding of the downstream signaling processes involving AICDs and/or intracellular interaction partners of APP to figure out the mechanism linking APP/ APLP functions to neuronal migration and positioning in the developing cortex. Whatever the precise mechanism of dysplasia formation, whether it is based on lacking AICD-mediated signaling or lacking protein interactions of the C-termini of membrane-integrated APP/APLPs, Fe65 seems to be pivotal in this process, as recently generated double-knockout mice, deficient for Fe65 and the related protein Fe65L1, showed a strikingly similar phenotype of neuronal heterotopias [35]. Finally, there exists also the possibility that secreted fragments of APP/APLPs 244

Neurodegenerative Dis 2006;3:239–246

are involved in cortical development. A prominent role of the secreted ectodomains of APP and APLP2 as regulators of adult neurogenesis in the subventricular zone has recently been described in vitro and in vivo [36]. To fully understand the APP/APLP functions, it will eventually be necessary to identify both ligands and receptors of the APP gene family in vivo.

Conclusions

The generation of combined gene deficiencies for APP and APLP proteins circumvents functional complementation between the APP family members and allowed first insights into the role of these proteins in vivo. Although we are just at the beginning of understanding the roles of APP and APLPs, it is evident that these proteins are involved in various physiological processes during development and in the adult. Whereas membrane-integrated proteins are likely to function as cell surface receptors, the processing of APP/APLPs generates several fragments that could act as ligands for other receptors (APPs-/, A) or regulate gene expression and intracellular signaling (AICDs). Indeed, distinct functions have been proposed for all the proteolytic fragments of APP [3]. Even for the neurotoxic A peptides, physiological functions such as regulation of cholesterol and sphingolipid biosynthesis [37] are emerging. Interestingly, there is also evidence for A as a homeostatic regulator of excitatory neurotransmission [38], consistent with earlier observations of a hypersensitivity to kainate-induced seizures in APP-deficient mice [10]. In view of the numerous functional units of APP and APLPs and the potential diversity of their effects, it may be necessary to segregate the different subdomains of the proteins and study their roles individually at the cellular level as well as in the whole organism.

Acknowledgement We are grateful to Heinrich Betz for continuous encouragement and support. This work was supported by grants from the Deutsche Forschungsgemeinschaft (MU-1457/4-1 to 4-3), and the Bundesministerium für Bildung und Forschung (01GS0469) to U.M. B.A. was supported by a scholarship from the Boehringer Ingelheim Foundation.

Anliker/Müller

References 1 Haass C: Take five – BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 2 Turner PR, O’Connor K, Tate WP, Abraham WC: Roles of amyloid precursor protein and its fragments in regulating neural activity, plasticity and memory. Prog Neurobiol 2003; 70:1–32. 3 Reinhard C, Hebert SS, De Strooper B: The amyloid-beta precursor protein: integrating structure with biological function. EMBO J 2005;24:3996–4006. 4 Eggert S, Paliga K, Soba P, Evin G, Masters CL, Weidemann A, Beyreuther K: The proteolytic processing of the amyloid precursor protein gene family members APLP-1 and APLP-2 involves alpha-, beta-, gamma-, and epsilon-like cleavages: modulation of APLP1 processing by n-glycosylation. J Biol Chem 2004;279:18146–18156. 5 Lorent K, Overbergh L, Moechars D, De Strooper B, Van Leuven F, Van den Berghe H: Expression in mouse embryos and in adult mouse brain of three members of the amyloid precursor protein family, of the alpha-2-macroglobulin receptor/low density lipoprotein receptor-related protein and of its ligands apolipoprotein E, lipoprotein lipase, alpha-2-macroglobulin and the 40,000 molecular weight receptor-associated protein. Neuroscience 1995;65:1009–1025. 6 Muller U, Cristina N, Li ZW, Wolfer DP, Lipp HP, Rulicke T, Brandner S, Aguzzi A, Weissmann C: Behavioral and anatomical deficits in mice homozygous for a modified betaamyloid precursor protein gene. Cell 1994; 79:755–765. 7 Zheng H, Jiang M, Trumbauer ME, Sirinathsinghji DJ, Hopkins R, Smith DW, Heavens RP, Dawson GR, Boyce S, Conner MW, Stevens KA, Slunt HH, Sisoda SS, Chen HY, Van der Ploeg LH: beta-Amyloid precursor protein-deficient mice show reactive gliosis and decreased locomotor activity. Cell 1995; 81:525–531. 8 Li ZW, Stark G, Gotz J, Rulicke T, Gschwind M, Huber G, Muller U, Weissmann C: Generation of mice with a 200-kb amyloid precursor protein gene deletion by Cre recombinase-mediated site-specific recombination in embryonic stem cells. Proc Natl Acad Sci USA 1996;93:6158–6162. 9 Dawson GR, Seabrook GR, Zheng H, Smith DW, Graham S, O’Dowd G, Bowery BJ, Boyce S, Trumbauer ME, Chen HY, Van der Ploeg LH, Sirinathsinghji DJ: Age-related cognitive deficits, impaired long-term potentiation and reduction in synaptic marker density in mice lacking the beta-amyloid precursor protein. Neuroscience 1999; 90:1– 13.

Functions of APP and APLPs

10 Steinbach JP, Muller U, Leist M, Li ZW, Nicotera P, Aguzzi A: Hypersensitivity to seizures in beta-amyloid precursor protein deficient mice. Cell Death Differ 1998;5:858–866. 11 Tremml P, Lipp HP, Muller U, Ricceri L, Wolfer DP: Neurobehavioral development, adult openfield exploration and swimming navigation learning in mice with a modified beta-amyloid precursor protein gene. Behav Brain Res 1998;95:65–76. 12 Magara F, Muller U, Li ZW, Lipp HP, Weissmann C, Stagljar M, Wolfer DP: Genetic background changes the pattern of forebrain commissure defects in transgenic mice underexpressing the beta-amyloid-precursor protein. Proc Natl Acad Sci USA 1999; 96: 4656–4661. 13 Seabrook GR, Smith DW, Bowery BJ, Easter A, Reynolds T, Fitzjohn SM, Morton RA, Zheng H, Dawson GR, Sirinathsinghji DJ, Davies CH, Collingridge GL, Hill RG: Mechanisms contributing to the deficits in hippocampal synaptic plasticity in mice lacking amyloid precursor protein. Neuropharmacology 1999;38:349–359. 14 White AR, Reyes R, Mercer JF, Camakaris J, Zheng H, Bush AI, Multhaup G, Beyreuther K, Masters CL, Cappai R: Copper levels are increased in the cerebral cortex and liver of APP and APLP2 knockout mice. Brain Res 1999;842:439–444. 15 Kamal A, Almenar-Queralt A, LeBlanc JF, Roberts EA, Goldstein LS: Kinesin-mediated axonal transport of a membrane compartment containing beta-secretase and presenilin-1 requires APP. Nature 2001; 414: 643–648. 16 Lazarov O, Morfini GA, Lee EB, Farah MH, Szodorai A, DeBoer SR, Koliatsos VE, Kins S, Lee VM, Wong PC, Price DL, Brady ST, Sisodia SS: Axonal transport, amyloid precursor protein, kinesin-1, and the processing apparatus: revisited. J Neurosci 2005; 25: 2386–2395. 17 Perez RG, Zheng H, Van der Ploeg LH, Koo EH: The beta-amyloid precursor protein of Alzheimer’s disease enhances neuron viability and modulates neuronal polarity. J Neurosci 1997;17:9407–9414. 18 Heber S, Herms J, Gajic V, Hainfellner J, Aguzzi A, Rulicke T, von Kretzschmar H, von Koch C, Sisodia S, Tremml P, Lipp HP, Wolfer DP, Muller U: Mice with combined gene knock-outs reveal essential and partially redundant functions of amyloid precursor protein family members. J Neurosci 2000; 20:7951–7963. 19 Harper SJ, Bilsland JG, Shearman MS, Zheng H, Van der Ploeg L, Sirinathsinghji DJ: Mouse cortical neurones lacking APP show normal neurite outgrowth and survival responses in vitro. Neuroreport 1998; 9: 3053– 3058.

20 Lopez-Sanchez N, Muller U, Frade JM: Lengthening of G2/mitosis in cortical precursors from mice lacking beta-amyloid precursor protein. Neuroscience 2005; 130: 51– 60. 21 von Koch CS, Zheng H, Chen H, Trumbauer M, Thinakaran G, van der Ploeg LH, Price DL, Sisodia SS: Generation of APLP2 KO mice and early postnatal lethality in APLP2/ APP double KO mice. Neurobiol Aging 1997; 18:661–669. 22 Soba P, Eggert S, Wagner K, Zentgraf H, Siehl K, Kreger S, Lower A, Langer A, Merdes G, Paro R, Masters CL, Muller U, Kins S, Beyreuther K: Homo- and heterodimerization of APP family members promotes intercellular adhesion. EMBO J 2005; 24: 3624– 3634. 23 Herms J, Anliker B, Heber S, Ring S, Fuhrmann M, Kretzschmar H, Sisodia S, Muller U: Cortical dysplasia resembling human type 2 lissencephaly in mice lacking all three APP family members. EMBO J 2004; 23:4106–4115. 24 Wang P, Yang G, Mosier DR, Chang P, Zaidi T, Gong YD, Zhao NM, Dominguez B, Lee KF, Gan WB, Zheng H: Defective neuromuscular synapses in mice lacking amyloid precursor protein (APP) and APP-Like protein 2. J Neurosci 2005;25:1219–1225. 25 Brandon EP, Lin W, D’Amour KA, Pizzo DP, Dominguez B, Sugiura Y, Thode S, Ko CP, Thal LJ, Gage FH, Lee KF: Aberrant patterning of neuromuscular synapses in choline acetyltransferase-deficient mice. J Neurosci 2003;23:539–549. 26 Misgeld T, Burgess RW, Lewis RM, Cunningham JM, Lichtman JW, Sanes JR: Roles of neurotransmitter in synapse formation: development of neuromuscular junctions lacking choline acetyltransferase. Neuron 2002;36:635–648. 27 Soriano E, Del Rio JA: The cells of CajalRetzius: still a mystery one century after. Neuron 2005;46:389–394. 28 Hartmann D, De Strooper B, Saftig P: Presenilin-1 deficiency leads to loss of CajalRetzius neurons and cortical dysplasia similar to human type 2 lissencephaly. Curr Biol 1999;9:719–727. 29 Wines-Samuelson M, Handler M, Shen J: Role of presenilin-1 in cortical lamination and survival of Cajal-Retzius neurons. Dev Biol 2005;277:332–346. 30 Cao X, Sudhof TC: A transcriptionally [correction of transcriptively] active complex of APP with Fe65 and histone acetyltransferase Tip60. Science 2001;293:115–120.

Neurodegenerative Dis 2006;3:239–246

245

31 Pardossi-Piquard R, Petit A, Kawarai T, Sunyach C, Alves da Costa C, Vincent B, Ring S, D’Adamio L, Shen J, Muller U, St George Hyslop P, Checler F: Presenilin-dependent transcriptional control of the Abeta-degrading enzyme neprilysin by intracellular domains of betaAPP and APLP. Neuron 2005; 46:541–554. 32 Leissring MA, Murphy MP, Mead TR, Akbari Y, Sugarman MC, Jannatipour M, Anliker B, Muller U, Saftig P, De Strooper B, Wolfe MS, Golde TE, LaFerla FM: A physiologic signaling role for the gamma-secretase-derived intracellular fragment of APP. Proc Natl Acad Sci USA 2002; 99: 4697– 4702.

246

33 Sabo SL, Ikin AF, Buxbaum JD, Greengard P: The Alzheimer amyloid precursor protein (APP) and FE65, an APP-binding protein, regulate cell movement. J Cell Biol 2001;153: 1403–1414. 34 Tissir F, Goffinet AM: Reelin and brain development. Nat Rev Neurosci 2003; 4: 496– 505. 35 Guenette S, Chang Y, Hiesberger T, Richardson JA, Eckman CB, Eckman EA, Hammer RE, Herz J: Essential roles for the FE65 amyloid precursor protein-interacting proteins in brain development. EMBO J 2006; 25: 420–431.

Neurodegenerative Dis 2006;3:239–246

36 Caille I, Allinquant B, Dupont E, Bouillot C, Langer A, Muller U, Prochiantz A: Soluble form of amyloid precursor protein regulates proliferation of progenitors in the adult subventricular zone. Development 2004; 131: 2173–2181. 37 Grimm MO, Grimm HS, Patzold AJ, Zinser EG, Halonen R, Duering M, Tschape JA, De Strooper B, Muller U, Shen J, Hartmann T: Regulation of cholesterol and sphingomyelin metabolism by amyloid-beta and presenilin. Nat Cell Biol 2005; 7:1118–1123. 38 Kamenetz F, Tomita T, Hsieh H, Seabruck G, Borchelt D, Iwatsubo T, Sosodia S, Malinow R: APP processing and synaptic function. Neuron 2003;37:925–937.

Anliker/Müller

Original Paper Neurodegenerative Dis 2006;3:247–254 DOI: 10.1159/000095263

Diseases

Control of Amyloid--Peptide Generation by Subcellular Trafficking of the -Amyloid Precursor Protein and -Secretase Jochen Walter Department of Neurology, University of Bonn, Bonn, Germany

Key Words Alzheimer’s disease -Amyloid precursor protein Glycosphingolipids Subcellular trafficking

Abstract Amyloid- (A) peptides are major components of Alzheimer’s disease (AD)-associated senile plaques and generated by sequential cleavage of the -amyloid precursor protein (APP) by -secretase and -secretase. While -secretase activity is exerted by the aspartic protease BACE1, -secretase consists of a protein complex of at least four essential proteins with the presenilins as the catalytically active components. The understanding of the subcellular trafficking of APP and proteases involved in its proteolytic processing has increased rapidly in the last years. APP as well as the secretases are membrane proteins, and recent work demonstrated that alterations in the lipid composition of cellular membranes could affect the proteolytic processing of APP and A generation. We identified glycosphingolipids as membrane components that modulate the subcellular transport of APP and the generation of A. By cell biological and biochemical methods we also characterized the role of BACE1 and its homologue BACE2 in the proteolytic processing of APP. Here, I summarize and discuss these findings in the context of other studies focused on the function of BACE1 and BACE2 and the role of subcellular trafficking in the proteolytic processing of APP. Copyright © 2006 S. Karger AG, Basel

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0247$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Introduction

Alzheimer’s disease (AD) is neuropathologically characterized by the accumulation of -amyloid plaques in the brain parenchyma [1]. Major components of plaques are amyloid- (A) peptides that derive from the -amyloid precursor protein (APP) by endoproteolytic processing involving sequential cleavages by - and -secretases [1–3] (fig. 1a). The initial cleavage of APP by secretase generates a membrane-bound C-terminal fragment that contains the A domain. This fragment represents a substrate for -secretase, which mediates the apparently intramembraneous cleavage of APP resulting in the liberation of A [4]. In an alternative pathway, APP can be cleaved by secretase in the center of the A domain, thereby precluding the generation of A. Proteolytic processing of APP by -, - and -secretases could occur in distinct subcellular compartments in the secretory and endocytic pathway (fig. 1a). Thus, molecular mechanisms that regulate the subcellular transport of APP and secretases play important roles in the generation of A. Trafficking and Proteolytic Processing of APP

APP is transported in the secretory pathway from the endoplasmic reticulum (ER) to the cell surface, from where it can be internalized into endosomal/lysosomal

Jochen Walter, PhD Department of Neurology, University of Bonn Sigmund-Freud-Strasse 25 DE–53127 Bonn (Germany) Tel. +49 228 9782, Fax +49 228 4387, E-Mail [email protected]

Fig. 1. Subcellular trafficking and proteolytic processing of APP. a The subcellular

trafficking of APP in the secretory and endocytic pathway is shown. The major processing events in the distinct subcellular compartments are indicated. However, the individual secretases could cleave APP to some extent also in other compartments (see text for further details). CTF = C-terminal fragment. b Schematic of APP and amino acid sequence of its cytoplasmic domain. The amino acid sequence motifs that are involved in the subcellular trafficking of APP are underlined and in bold. N = N-terminus; C = C-terminus; A = A domain; CHO = N-glycosylation site; KPI = Kunitz protease inhibitor domain.

compartments [5, 6] (fig. 1a). During transport and at the cell surface, APP is predominantly cleaved by -secretase [7–9]. Interestingly, it has been shown that the ‘Swedish’ double-mutation of APP (595KM596 ] 595NL596 ; numbering for 695 isoform) that causes familial early-onset AD led to increased cleavage by -secretase within secretory vesicles and increased secretion of A [10, 11]. This effect is consistent with the finding that ‘Swedish mutant’ APP is cleaved much more efficiently than wild-type APP by the purified -secretase BACE-1 [12– 14]. Thus, the ‘Swedish mutant’ of APP leads to increased production of A by increasing its -secretory processing in the secretory pathway. In contrast, wildtype APP is cleaved by -secretase preferentially after internalization from the cell surface in endocytic compartments [15, 16] (fig. 1a). 248

'

'

Neurodegenerative Dis 2006;3:247–254

The endocytosis of full-length APP from the plasma membrane is mainly determined by the 681GYENPTY687 motif within its cytoplasmic domain (fig. 1b). Mutation of this motif strongly decreases endocytosis and results in increased levels of APP at the cell surface [17, 18]. The surface-located fraction of APP is predominantly cleaved by -secretase, leading to increased secretion of soluble APP (APPS) and decreased production of A. Several cytoplasmic adaptor proteins, including disabled-1, Fe65, X11, c-JUN N-terminal kinase interacting protein (JIP)-1, interact with the 681GYENPTY687 motif via their phospho-tyrosine binding domains [19–24]. It has also been demonstrated that the interaction of APP with JIP-1 or Fe65 could affect the subcellular trafficking and the proteolytic processing of APP [20, 25, 26 ; for review, see 27]. Walter

The cytoplasmic domain of APP contains another tyrosine-containing trafficking signal that is located close to the transmembrane domain (fig. 1b). However, this motif (653YTSI656) does not strongly affect endocytosis, but has been shown to determine sorting of fulllength APP to the basolateral membrane in polarized MDCK cells [18, 28, 29]. Mutations of this motif led to increased sorting of APP to the apical surface, while wild-type APP is sorted predominantly basolaterally. Since APPS is also sorted predominantly basolatererally, additional unknown sorting information appears to be also present in the ectodomain of APP [28, 29]. The mechanisms described here have been mainly demonstrated in cultured cell lines. The subcellular trafficking of APP in neuronal cells is reviewed by Kins et al. [this issue, pp. 218–226].

Role of Glycosphingolipids in the Subcellular Transport of APP and the Generation of A

Accumulating evidence indicates that the generation of A is also affected by the lipid composition of cellular membranes [30–32]. Research was focused predominantly on the role of cholesterol, and it has been shown that reduction of cholesterol could decrease the generation of A in cultured cells, mice and guinea pigs [30, 33]. In addition, the inhibition of acyl-coenzyme A: cholesterol acyltransferase also led to a strong reduction of A generation in cultured cells and transgenic mice, indicating that cholesterol esters also influence proteolytic processing of APP [34, 35]. Notably, it has also been shown in cultured neurons that a moderate decrease in cholesterol could stimulate A generation, while stronger reductions of cholesterol significantly inhibit A secretion [36]. The molecular mechanisms underlying these effects are not fully understood, but could involve the distribution of APP and secretases in detergent-resistant membrane microdomains. The picture in humans obtained from clinical trials with cholesterol-lowering drugs is less clear. While some studies with statins showed a protective effect against AD, others did not reveal a significant benefit. Since the role of cholesterol in AD has been summarized and discussed extensively in recent reviews [30– 33], I will focus here on the role of other membrane lipids in APP processing. Beside cholesterol, glycosphingolipids (GSLs) have also been involved in the pathogenesis of AD. Several studies showed that A binds to the ganglioside GM1 in vitro and respective complexes could be also isolated Control of Amyloid--Peptide Generation

from brains of both humans and transgenic mice. It was also speculated that these complexes could serve as a seed for A oligomers and aggregates [37, 38]. On the other hand, the injection of GM1 into APP transgenic mice decreased the levels of soluble A and -amyloid plaque load, probably by binding to A in peripheral blood and draining A out of the brain [39]. It has also been shown that levels of several gangliosides are altered in the brains of AD patients [40]. The biosynthesis of GSLs starts with the generation of glucosylceramide from UDP glucose and ceramide by glucosylceramide synthase in the ER and early Golgi compartments. Glucosylceramide represents the precursor of a large variety of GSLs [41], which are modified in the Golgi and then transported in the secretory pathway to the cell surface [30, 42]. Thus, the subcellular trafficking of GSLs in the secretory pathway is very similar to that of APP. The physiological functions of GSLs include the regulation of cell adhesion, cell differentiation, and signal transduction [42–44]. Dysfunction of GSL degradation is associated with several inherited diseases that are characterized by accumulation of GSLs in endosomal/lysosomal compartments [45, 46]. We addressed the role of GSLs in the proteolytic processing of APP in cell culture models. The inhibition of GSL biosynthesis by d-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol significantly decreased the secretion of APPS in several cell types, including human neuronal SH-SY5Y cells [47]. Interestingly, the inhibition of GSL biosynthesis also decreased the secretion of A. By pulse chase experiments, we found that the maturation of APP is significantly decreased by glucosylceramide synthase inhibition [47]. In addition, the expression of APP at the cell surface was also strongly decreased in GSL-depleted cells. Very similar data were also obtained in a genetic model of GSL deficiency. In mouse melanoma GM95 cells with defective GSL biosynthesis, maturation of APP was also strongly decreased as compared to the parental B16 cell line [47]. Moreover, levels of APP were significantly decreased in GSL-depleted cells also indicating an effect on the stability of APP [47]. Together, these data indicated that GSLs are involved in the forward transport of APP along the secretory pathway to the cell surface. In agreement with our data, GSLs have recently been shown to be implicated in the subcellular transport of membrane proteins, and appear to modulate transport of individual proteins, probably at distinct steps in the secretory pathway [42, 48]. Since the inhibition of GSL biosynthesis impairs the forward transport of APP in the secretory pathway and Neurodegenerative Dis 2006;3:247–254

249

Fig. 2. Schematic of BACE1/2 and their cleavage sites within APP. a Model of BACE1 and BACE2. The aspartyl protease active site motifs (DTGS, DSGT) in the lumenal/extracellular domain are indicated by asterisks. The signal peptide (black rectangle) and the prodomain (hatched rectangle) are indicated. The amino acid sequences of the cytoplasmic domains of BACE1 and BACE2 are given in single letter code. The phosphorylation site of BACE1 (S 498) is in bold and the DXXLL in BACE1 motif is underlined. b Amino acid sequence of the A domain with the indicated cleavage sites for BACE1 and BACE2.

reduces the levels of APP at the cell surface, amyloidogenic processing might be decreased by segregation of APP and secretases into distinct subcellular compartments. The detailed molecular mechanisms underlying the role of GSLs in APP processing need further investigation. However, targeting GSLs or enzymes involved in their biosynthesis might be a strategy to decrease A generation in therapeutic approaches against AD. Role of BACE1 and BACE2 in APP Processing

The initial step in A generation is the cleavage of APP by -secretase that was identified in 1999 by several groups as the aspartic protease BACE1 (beta-APP cleaving enzyme), together with a close homolog termed BACE2 [12–14, 49, 50]. Both enzymes contain the two characteristic D(T/S) G(T/S) motifs of aspartyl proteases, which form their catalytic site and share significant sequence homology with other members of the pepsin family of aspartyl proteases. In contrast to all other aspartyl proteases of the pepsin family, BACE1 and BACE2 are type I transmembrane 250

Neurodegenerative Dis 2006;3:247–254

proteins with a large lumenal domain containing the active center, a single transmembrane domain and a small cytoplasmic tail (fig. 2a). Because BACE1 initiates A generation and represents a potential target to lower A levels in the brain, it is important to understand the molecular mechanisms that regulate its cellular metabolism and activity. Despite the high homology between BACE1 and BACE2, a number of biochemical and cell biological as well as in vivo studies with knockout and transgenic mice indicate that BACE1 is the major enzyme involved in neuronal A production [51, 52]. Our studies also indicated significant differences in the specificity of BACE1 and BACE2 in the cleavage of APP (fig. 2b). While BACE1 cleaves APP at the N-terminus of the A domain and at an internal site at Glu11, BACE2 cleaves APP in the middle of the A domain between phenylalanine 19 and 20 (fig. 2b) [53–55]. The BACE2-mediated cleavage results in increased secretion of APPS- and p3like products and reduced production of A species [55]. Thus, the cleavage of APP by BACE2 is similar to secretase rather than to -secretase [55, 56]. The stimulation of BACE2, like that of -secretase, might therefore represent an additional strategy to decrease A generation. Cell biological studies demonstrated that cleavage of APP by BACE1 and BACE2 occurs predominantly in the Golgi and later secretory and endocytic compartments [3, 57]. However, in cells overexpressing BACE1, aberrant cleavage of APP in earlier secretory compartments has also been demonstrated [12, 49, 53, 55, 56, 58]. Notably, a decrease in A levels in conditioned media was observed under these conditions, while a C-terminally truncated variant of A ending at position 34 (A34) was found to be increased [59]. The generation of this species is dependent on -secretase activity, and it was speculated that position 34 represents a novel cleavage site for -secretase [60]. However, we could demonstrate that both BACE1 and BACE2 can cleave A after its -secretase-dependent release from the membrane at position 34, which results in decreased A levels (fig. 2b) [59]. Thus, BACE1 and BACE2 can also mediate the degradation of A. These findings are consistent with the observation that transgenic mice with high BACE1 expression in the brain show decreased A levels and amyloid plaque load [61]. However, the functional relevance of the A-degrading activity of BACE1 and BACE2 in the pathogenesis of AD is not understood and needs further investigation.

Walter

Subcellular Transport of BACE1 and BACE2

The characterization of the subcellular transport of BACE1 and BACE2 revealed that both proteins are transported in the secretory pathway from the ER to the Golgi compartment, where they undergo maturation by complex N-glycosylation [3, 12]. After release of the proteases from the ER and during further transport through the Golgi, prodomains at the N-termini of BACE1 and BACE2 are proteolytically removed. While the propeptide cleavage of BACE1 is catalyzed by furin and/or a furin-like protease to generate the mature enzyme [22–25], that of BACE2 occurs auto-catalytically [62]. It has been shown that removal of the prodomain of BACE1 does not strongly affect its enzymatic activity, as one may expect for a typical zymogen, and rather plays a role in folding and forward transport of BACE1 through the secretory pathway [24, 26]. After complete maturation of the enzymes by glycosylation and propeptide cleavage, BACE1 and BACE2 are further transported from the trans-Golgi network (TGN) to the cell surface [27, 28]. As revealed by pulse-chase experiments, BACE1 is more stable than BACE2 [55]. While BACE2 is degraded at the cell surface or shortly after endocytosis, BACE1 is efficiently endocytosed from the plasma membrane into early endosomal compartments and further transported retrogradely to the TGN (fig. 3). Several amino acid sequence-based trafficking signals have been identified in the cytoplasmic domain of BACE1 (fig. 2a). A dileucine motif has been shown to regulate endocytosis and recycling of BACE1 to the plasma membrane [63]. In addition, we identified a phosphorylation site for casein kinase 1 at Ser498 that regulates the retrograde trafficking of BACE1 between endosomal compartments and the TGN (fig. 2a, 3) [53]. While the wild-type protein and a mutant that mimics phosphorylated BACE1 (S498D) are efficiently retrieved from endosomal compartments to the TGN, a mutant that cannot be phosphorylated (S498A) accumulates in endosomal compartments. The phosphorylation site of BACE1 in the cytoplasmic domain is located within a characteristic binding motif (DXXLL) for the Golgi-localized gamma ear containing (ADP) ribosylation factor binding (GGA) protein family (fig. 2a). GGA proteins are monomeric adaptors that sort specific cargo proteins, like the mannose-6-phosphate receptors, from the TGN to endosomal/lysosomal compartments [64, 65]. By cell biological and biochemical methods, it was demonstrated that GGAs directly interact with BACE1 but not BACE2 [66], which is consistent with the lack of a canonical recognition motif in the cyControl of Amyloid--Peptide Generation

Fig. 3. Subcellular trafficking of BACE1. BACE1 is transported from the ER via the TGN to the cell surface from where it can be reinternalized into endosomal compartments. Phosphorylation of its C-terminal serine residue 498 enhances its interaction with GGA proteins and the retrograde transport to the TGN, while nonphosphorylated BACE1 can recycle to the cell surface. A direct transport route for BACE1 from the TGN to endosomal compartments might also exist (marked by ?).

toplasmic domain of BACE2 (fig. 2a). The phosphorylation of BACE1 increases binding to GGAs, suggesting that these proteins could regulate the phosphorylation state-dependent subcellular transport of BACE1. Indeed, the expression of a dominant-negative variant of GGA1 inhibited retrograde transport of BACE1 to the TGN and led to the accumulation of BACE1 in endosomal compartments, very similar to the inhibition of BACE1 phosphorylation (fig. 3) [67].

Role of GGA Proteins in the Proteolytic Processing of APP

Immunohistochemical analyses revealed that GGA1 is also expressed in neurons of the human brain, where it colocalizes with BACE1. Moreover, the expression of GGA1 was found to be significantly decreased in brain lysates from AD patients as compared to controls [Wahle and Walter, unpubl. obs.]. However, several studies indicated that the secretion of APPS and A was not affected by GGA proteins in cells overexpressing BACE1 [68, 69]. This might be partly due to the aberrant cleavage of APP by BACE1 early in the secretory pathway, while under physiological expression levels, BACE1 cleaves APP mainly in later secretory and endocytic compartments [3, 57]. Neurodegenerative Dis 2006;3:247–254

251

To investigate whether GGA proteins are functionally involved in the BACE1-dependent generation of A, we used a cellular model with endogenous expression of BACE1. The overexpression of the GGA1 dominant-negative variant in human embryonic kidney 293 cells significantly decreased the secretion of APPS. In addition, A secretion was also decreased, demonstrating that GGA proteins are involved in the regulation of the proteolytic processing of APP and A generation [Wahle and Walter, unpubl. obs.]. What mechanisms could contribute to the GGA-dependent proteolytic processing of APP? GGAs are known to mediate forward transport of cargo proteins from the TGN to endosomal/lysosomal compartments [64]. Whether BACE1 is also transported on this route remains to be demonstrated. On the other hand, we could demonstrate that GGA proteins are also involved in the retrograde transport of BACE1 from endosomes to the TGN (fig. 3) [67]. Since BACE1 can cleave APP in both compartments to initiate A generation, the GGA-de-

pendent transport between the TGN and endosomes could likely regulate the amyloidogenic processing of APP by affecting the relative contribution of - and secretory processing of APP. However, GGAs are not only involved in the trafficking of BACE1, but also in that of the mannose-6-phosphate receptors and sorLA [64, 70]. Since both proteins could also affect the proteolytic processing of APP [70, 71], it will be interesting to investigate the role of GGA proteins in the pathogenesis of AD in the future.

Acknowledgements I am grateful to the members of my lab for their excellent contribution to the projects and comments on the manuscript. I also thank Drs. C. Haass and G. Multhaup and their co-workers for close collaboration and stimulating discussion on the described projects. This work has been supported by grants of the Deutsche Forschungsgemeinschaft.

References 1 Selkoe DJ: Alzheimer’s disease: genes, proteins, and therapy. Physiol Rev 2001;81:741– 766. 2 Annaert W, de Strooper B: A cell biological perspective on Alzheimer’s disease. Annu Rev Cell Dev Biol 2002; 18:25–51. 3 Walter J, Kaether C, Steiner H, Haass C: The cell biology of Alzheimer’s disease: uncovering the secrets of secretases. Curr Opin Neurobiol 2001;11:585–590. 4 Steiner H, Haass C: Intramembrane proteolysis by presenilins. Nat Rev Mol Cell Biol 2000;1:217–224. 5 Koo EH, Squazzo SL, Selkoe DJ, Koo CH: Trafficking of cell-surface amyloid beta-protein precursor. I. Secretion, endocytosis and recycling as detected by labeled monoclonal antibody. J Cell Sci 1996;109:991–998. 6 Yamazaki T, Koo EH, Selkoe DJ: Trafficking of cell-surface amyloid beta-protein precursor. II. Endocytosis, recycling and lysosomal targeting detected by immunolocalization. J Cell Sci 1996;109:999–1008. 7 Sisodia SS, Koo EH, Beyreuther K, Unterbeck A, Price DL: Evidence that beta-amyloid protein in Alzheimer’s disease is not derived by normal processing. Science 1990; 248:492–495. 8 Hung AY, Haass C, Nitsch RM, Qiu WQ, Citron M, Wurtman RJ, Growdon JH, Selkoe DJ: Activation of protein kinase C inhibits cellular production of the amyloid beta-protein. J Biol Chem 1993; 268:22959–22962.

252

9 Golde TE, Estus S, Younkin LH, Selkoe DJ, Younkin SG: Processing of the amyloid protein precursor to potentially amyloidogenic derivatives. Science 1992;255:728–730. 10 Haass C, Lemere CA, Capell A, Citron M, Seubert P, Schenk D, Lannfelt L, Selkoe DJ: The Swedish mutation causes early-onset Alzheimer’s disease by beta-secretase cleavage within the secretory pathway. Nat Med 1995;1:1291–1296. 11 Thinakaran G, Teplow DB, Siman R, Greenberg B, Sisodia SS: Metabolism of the ‘Swedish’ amyloid precursor protein variant in neuro2a (N2a) cells. Evidence that cleavage at the ‘beta-secretase’ site occurs in the golgi apparatus. J Biol Chem 1996;271:9390–9397. 12 Vassar R, Bennett BD, Babu-Khan S, Kahn S, Mendiaz EA, Denis P, Teplow DB, Ross S, Amarante P, Loeloff R, Luo Y, Fisher S, Fuller J, Edenson S, Lile J, Jarosinski MA, Biere AL, Curran E, Burgess T, Louis JC, Collins F, Treanor J, Rogers G, Citron M: Beta-secretase cleavage of Alzheimer’s amyloid precursor protein by the transmembrane aspartic protease BACE. Science 1999;286:735–741. 13 Sinha S, Anderson JP, Barbour R, Basi GS, Caccavello R, Davis D, Doan M, Dovey HF, Frigon N, Hong J, Jacobson-Croak K, Jewett N, Keim P, Knops J, Lieberburg I, Power M, Tan H, Tatsuno G, Tung J, Schenk D, Seubert P, Suomensaari SM, Wang S, Walker D, John V, et al: Purification and cloning of amyloid precursor protein beta-secretase from human brain. Nature 1999;402:537–540.

Neurodegenerative Dis 2006;3:247–254

14 Yan R, Bienkowski MJ, Shuck ME, Miao H, Tory MC, Pauley AM, Brashier JR, Stratman NC, Mathews WR, Buhl AE, Carter DB, Tomasselli AG, Parodi LA, Heinrikson RL, Gurney ME: Membrane-anchored aspartyl protease with Alzheimer’s disease betasecretase activity. Nature 1999;402:533–537. 15 Haass C, Koo EH, Mellon A, Hung AY, Selkoe DJ: Targeting of cell-surface beta-amyloid precursor protein to lysosomes: alternative processing into amyloid-bearing fragments. Nature 1992;357:500–503. 16 Koo EH, Squazzo SL: Evidence that production and release of amyloid beta-protein involves the endocytic pathway. J Biol Chem 1994;269:17386–17389. 17 Lai A, Sisodia SS, Trowbridge IS: Characterization of sorting signals in the beta-amyloid precursor protein cytoplasmic domain. J Biol Chem 1995;270:3565–3573. 18 Perez RG, Soriano S, Hayes JD, Ostaszewski B, Xia W, Selkoe DJ, Chen X, Stokin GB, Koo EH: Mutagenesis identifies new signals for beta-amyloid precursor protein endocytosis, turnover, and the generation of secreted fragments, including Abeta42. J Biol Chem 1999;274:18851–18856. 19 Cao X, Sudhof TC: A transcriptionally [correction of transcriptively] active complex of APP with Fe65 and histone acetyltransferase Tip60. Science 2001;293:115–120.

Walter

20 Pietrzik CU, Yoon IS, Jaeger S, Busse T, Weggen S, Koo EH: FE65 constitutes the functional link between the low-density lipoprotein receptor-related protein and the amyloid precursor protein. J Neurosci 2004;24:4259– 4265. 21 Scheinfeld MH, Roncarati R, Vito P, Lopez PA, Abdallah M, D’Adamio L: Jun NH2-terminal kinase (JNK) interacting protein 1 (JIP1) binds the cytoplasmic domain of the Alzheimer’s beta-amyloid precursor protein (APP). J Biol Chem 2002; 277:3767–3775. 22 Biederer T, Cao X, Sudhof TC, Liu X: Regulation of APP-dependent transcription complexes by Mint/X11s: differential functions of Mint isoforms. J Neurosci 2002; 22:7340– 7351. 23 Homayouni R, Rice DS, Sheldon M, Curran T: Disabled-1 binds to the cytoplasmic domain of amyloid precursor-like protein 1. J Neurosci 1999;19:7507–7515. 24 Muresan Z, Muresan V: Coordinated transport of phosphorylated amyloid-beta precursor protein and c-Jun NH2-terminal kinase-interacting protein-1. J Cell Biol 2005; 171:615–625. 25 Guenette SY, Chen J, Ferland A, Haass C, Capell A, Tanzi RE: hFE65L influences amyloid precursor protein maturation and secretion. J Neurochem 1999;73:985–993. 26 Chang Y, Tesco G, Jeong WJ, Lindsley L, Eckman EA, Eckman CB, Tanzi RE, Guenette SY: Generation of the beta-amyloid peptide and the amyloid precursor protein Cterminal fragment gamma are potentiated by FE65L1. J Biol Chem 2003; 278: 51100– 51107. 27 King GD, Scott TR: Adaptor protein interactions: modulators of amyloid precursor protein metabolism and Alzheimer’s disease risk? Exp Neurol 2004;185:208–219. 28 Haass C, Koo EH, Capell A, Teplow DB, Selkoe DJ: Polarized sorting of beta-amyloid precursor protein and its proteolytic products in MDCK cells is regulated by two independent signals. J Cell Biol 1995; 128: 537– 547. 29 de Strooper B, Craessaerts K, Dewachter I, Moechars D, Greenberg B, Van Leuven F, Van den Berghe H: Basolateral secretion of amyloid precursor protein in Madin-Darby canine kidney cells is disturbed by alterations of intracellular pH and by introducing a mutation associated with familial Alzheimer’s disease. J Biol Chem 1995; 270: 4058–4065. 30 Simons K, Ehehalt R: Cholesterol, lipid rafts, and disease. J Clin Invest 2002; 110: 597– 603. 31 Hartmann T: Cholesterol, A beta and Alzheimer’s disease. Trends Neurosci 2001; 24(suppl 11):S45–S48. 32 Puglielli L, Tanzi RE, Kovacs DM: Alzheimer’s disease: the cholesterol connection. Nat Neurosci 2003;6:345–351. 33 Wolozin B: Cholesterol and the biology of Alzheimer’s disease. Neuron 2004;41:7–10.

Control of Amyloid--Peptide Generation

34 Puglielli L, Konopka G, Pack-Chung E, Ingano LA, Berezovska O, Hyman BT, Chang TY, Tanzi RE, Kovacs DM: Acyl-coenzyme A: cholesterol acyltransferase modulates the generation of the amyloid beta-peptide. Nat Cell Biol 2001; 3:905–912. 35 Hutter-Paier B, Huttunen HJ, Puglielli L, Eckman CB, Kim DY, Hofmeister A, Moir RD, Domnitz SB, Frosch MP, Windisch M, Kovacs DM: The ACAT inhibitor CP-113,818 markedly reduces amyloid pathology in a mouse model of Alzheimer’s disease. Neuron 2004;44:227–238. 36 Abad-Rodriguez J, Ledesma MD, Craessaerts K, Perga S, Medina M, Delacourte A, Dingwall C, de Strooper B, Dotti CG: Neuronal membrane cholesterol loss enhances amyloid peptide generation. J Cell Biol 2004; 167:953–960. 37 Yanagisawa K, Odaka A, Suzuki N, Ihara Y: GM1 ganglioside-bound amyloid beta-protein (A beta): a possible form of preamyloid in Alzheimer’s disease. Nat Med 1995; 1: 1062–1066. 38 Hayashi H, Kimura N, Yamaguchi H, Hasegawa K, Yokoseki T, Shibata M, Yamamoto N, Michikawa M, Yoshikawa Y, Terao K, Matsuzaki K, Lemere CA, Selkoe DJ, Naiki H, Yanagisawa K: A seed for Alzheimer amyloid in the brain. J Neurosci 2004; 24: 4894–4902. 39 Matsuoka Y, Saito M, LaFrancois J, Saito M, Gaynor K, Olm V, Wang L, Casey E, Lu Y, Shiratori C, Lemere C, Duff K: Novel therapeutic approach for the treatment of Alzheimer’s disease by peripheral administration of agents with an affinity to beta-amyloid. J Neurosci 2003;23:29–33. 40 Cutler RG, Kelly J, Storie K, Pedersen WA, Tammara A, Hatanpaa K, Troncoso JC, Mattson MP: Involvement of oxidative stress-induced abnormalities in ceramide and cholesterol metabolism in brain aging and Alzheimer’s disease. Proc Natl Acad Sci USA 2004;101:2070–2075. 41 Kolter T, Proia RL, Sandhoff K: Combinatorial ganglioside biosynthesis. J Biol Chem 2002;277:25859–25862. 42 Sprong H, van der Sluijs P, van Meer G: How proteins move lipids and lipids move proteins. Nat Rev Mol Cell Biol 2001; 2: 504– 513. 43 Kolter T, Sandhoff K: Recent advances in the biochemistry of sphingolipidoses. Brain Pathol 1998;8:79–100. 44 Brown DA, London E: Functions of lipid rafts in biological membranes. Annu Rev Cell Dev Biol 1998; 14:111–136. 45 Marks DL, Pagano RE: Endocytosis and sorting of glycosphingolipids in sphingolipid storage disease. Trends Cell Biol 2002; 12: 605–613. 46 Schuette CG, Doering T, Kolter T, Sandhoff K: The glycosphingolipidoses – from disease to basic principles of metabolism. Biol Chem 1999;380:759–766.

47 Tamboli IY, Prager K, Barth E, Heneka M, Sandhoff K, Walter J: Inhibition of glycosphingolipid biosynthesis reduces secretion of the beta-amyloid precursor protein and amyloid beta-peptide. J Biol Chem 2005;280: 28110–28117. 48 Muniz M, Morsomme P, Riezman H: Protein sorting upon exit from the endoplasmic reticulum. Cell 2001;104:313–320. 49 Hussain I, Powell D, Howlett DR, Tew DG, Meek TD, Chapman C, Gloger IS, Murphy KE, Southan CD, Ryan DM, Smith TS, Simmons DL, Walsh FS, Dingwall C, Christie G: Identification of a novel aspartic protease (Asp 2) as beta-secretase. Mol Cell Neurosci 1999;14:419–427. 50 Lin X, Koelsch G, Wu S, Downs D, Dashti A, Tang J: Human aspartic protease memapsin 2 cleaves the beta-secretase site of beta-amyloid precursor protein. Proc Natl Acad Sci USA 2000;97:1456–1460. 51 Luo Y, Bolon B, Kahn S, Bennett BD, BabuKhan S, Denis P, Fan W, Kha H, Zhang J, Gong Y, Martin L, Louis JC, Yan Q, Richards WG, Citron M, Vassar R: Mice deficient in BACE1, the Alzheimer’s beta-secretase, have normal phenotype and abolished beta-amyloid generation. Nat Neurosci 2001; 4: 231– 232. 52 Cai H, Wang Y, McCarthy D, Wen H, Borchelt DR, Price DL, Wong PC: BACE1 is the major beta-secretase for generation of Abeta peptides by neurons. Nat Neurosci 2001;4:233–234. 53 Walter J, Fluhrer R, Hartung B, Willem M, Kaether C, Capell A, Lammich S, Multhaup G, Haass C: Phosphorylation regulates intracellular trafficking of beta-secretase. J Biol Chem 2001;276:14634–14641. 54 Capell A, Meyn L, Fluhrer R, Teplow DB, Walter J, Haass C: Apical sorting of betasecretase limits amyloid beta-peptide production. J Biol Chem 2002; 277:5637–5643. 55 Fluhrer R, Capell A, Westmeyer G, Willem M, Hartung B, Condron MM, Teplow DB, Haass C, Walter J: A non-amyloidogenic function of BACE-2 in the secretory pathway. J Neurochem 2002;81:1011–1020. 56 Yan R, Munzner JB, Shuck ME, Bienkowski MJ: BACE2 functions as an alternative alpha-secretase in cells. J Biol Chem 2001;276: 34019–34027. 57 Vassar R, Citron M: Abeta-generating enzymes: recent advances in beta- and gammasecretase research. Neuron 2000;27:419–422. 58 Capell A, Steiner H, Willem M, Kaiser H, Meyer C, Walter J, Lammich S, Multhaup G, Haass C: Maturation and pro-peptide cleavage of beta-secretase. J Biol Chem 2000; 275: 30849–30854. 59 Fluhrer R, Multhaup G, Schlicksupp A, Okochi M, Takeda M, Lammich S, Willem M, Westmeyer G, Bode W, Walter J, Haass C: Identification of a beta-secretase activity, which truncates amyloid beta-peptide after its presenilin-dependent generation. J Biol Chem 2003;278:5531–5538.

Neurodegenerative Dis 2006;3:247–254

253

60 Vandermeeren M, Geraerts M, Pype S, Dillen L, Van Hove C, Mercken M: The functional gamma-secretase inhibitor prevents production of amyloid beta 1–34 in human and murine cell lines. Neurosci Lett 2001; 315:145–148. 61 Lee EB, Zhang B, Liu K, Greenbaum EA, Doms RW, Trojanowski JQ, Lee VM: BACE overexpression alters the subcellular processing of APP and inhibits Abeta deposition in vivo. J Cell Biol 2005; 168:291–302. 62 Hussain I, Christie G, Schneider K, Moore S, Dingwall C: Prodomain processing of Asp1 (BACE2) is autocatalytic. J Biol Chem 2001; 20:20. 63 Huse JT, Pijak DS, Leslie GJ, Lee VM, Doms RW: Maturation and endosomal targeting of beta-site amyloid precursor protein-cleaving enzyme. The Alzheimer’s disease beta-secretase. J Biol Chem 2000; 275:33729–33737. 64 Bonifacino JS: The GGA proteins: adaptors on the move. Nat Rev Mol Cell Biol 2004; 5: 23–32.

254

65 Robinson MS: Adaptable adaptors for coated vesicles. Trends Cell Biol 2004; 14:167–174. 66 He X, Chang WP, Koelsch G, Tang J: Memapsin 2 (beta-secretase) cytosolic domain binds to the VHS domains of GGA1 and GGA2:implications on the endocytosis mechanism of memapsin 2. FEBS Lett 2002; 524:183–187. 67 Wahle T, Prager K, Raffler N, Haass C, Famulok M, Walter J: GGA proteins regulate retrograde transport of BACE1 from endosomes to the trans-Golgi network. Mol Cell Neurosci 2005;29:453–461. 68 He X, Li F, Chang WP, Tang J: GGA proteins mediate the recycling pathway of memapsin 2 (BACE). J Biol Chem 2005; 280: 11696– 11703. 69 von Arnim CA, Tangredi MM, Peltan ID, Lee BM, Irizarry MC, Kinoshita A, Hyman BT: Demonstration of BACE (beta-secretase) phosphorylation and its interaction with GGA1 in cells by fluorescence-lifetime imaging microscopy. J Cell Sci 2004; 117: 5437– 5445.

Neurodegenerative Dis 2006;3:247–254

70 Andersen OM, Reiche J, Schmidt V, Gotthardt M, Spoelgen R, Behlke J, von Arnim CA, Breiderhoff T, Jansen P, Wu X, Bales KR, Cappai R, Masters CL, Gliemann J, Mufson EJ, Hyman BT, Paul SM, Nykjaer A, Willnow TE: Neuronal sorting protein-related receptor sorLA/LR11 regulates processing of the amyloid precursor protein. Proc Natl Acad Sci USA 2005;102:13461–13466. 71 Mathews PM, Guerra CB, Jiang Y, Grbovic OM, Kao BH, Schmidt SD, Dinakar R, Mercken M, Hille-Rehfeld A, Rohrer J, Mehta P, Cataldo AM, Nixon RA: Alzheimer’s disease-related overexpression of the cation-dependent mannose 6-phosphate receptor increases Abeta secretion: role for altered lysosomal hydrolase distribution in betaamyloidogenesis. J Biol Chem 2002; 277: 5299–5307.

Walter

Original Paper

Diseases

Neurodegenerative Dis 2006;3:255–261 DOI: 10.1159/000095264

-Secretase Activation – An Approach to Alzheimer’s Disease Therapy Falk Fahrenholz Rolf Postina Institute of Biochemistry, University of Mainz, Mainz, Germany

Key Words -Secretase ADAM10 Retinoic acid Cholesterol

Abstract The nonamyloidogenic pathway of processing the amyloid precursor protein (APP) involves the cleavage within the amyloid- peptide sequence, and thus precludes amyloid- formation. The identification of a member of the disintegrin and metalloproteinase family, ADAM10, as an -secretase that prevents plaque formation and hippocampal deficits in vivo gave us the possibility to examine the -secretase as a potential target for the therapy of Alzheimer’s disease. Within the priority program Cellular Mechanisms of Alzheimer’s Disease, we investigated several approaches to stimulate the -secretase pathway. Two protein convertases were found to be responsible for the removal of the prodomain, and for the formation of the mature enzyme with -secretase activity. The cloning and characterization of the human ADAM10 promoter provided the basis to examine ADAM10 gene expression. We found a common upregulation of ADAM10, APP, and APP-like protein 2 during differentiation of neuronal cells by retinoic acid, and increased -secretase cleavage of the two substrates. Other approaches for enhancing secretase activity are the reduction of cellular cholesterol and the stimulation of G protein-coupled neuropeptide receptors. Our results suggest medications and dietary regiments which enhance the nonamyloidogenic pathway of APP processing to be a valuable approach to Alzheimer’s disease therapy. Copyright © 2006 S. Karger AG, Basel

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0255$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Introduction

A main strategy to treat Alzheimer’s disease (AD) is to prevent the formation of amyloid- (A) peptides, and their deposition as senile plaques in the brain. A40 and A42, especially their oligomeric aggregates, are believed to play a central role in AD by causing neurotoxicity and age-related cognitive deficits. Therefore, - and -secretases which generate A by sequential cleavage of the amyloid precursor protein (APP) are obvious and main targets for the development of specific inhibitors. However, the development of such inhibitors of the -secretase BACE or of presenilins, the major components in the secretase complex, that are able to cross the blood-brain barrier, is still a challenge. The nonamyloidogenic pathway of APP processing involves APP cleavage within the A sequence, and thus precludes the A formation. -Secretase cleavage releases the N-terminal ectodomain of APP (APPs), which has neurotrophic and neuroprotective properties [for review, see 1]. Therefore, activation of the nonamyloidogenic pathway provides a logical alternative strategy for treatment of AD. This idea has almost been forgotten until the -secretase was discovered to be a member of the ADAM (a disintegrin and metalloproteinase) family of proteinases. Within the priority program Cellular Mechanisms of Alzheimer’s Disease, we investigated the cellular function of the -secretase, its expression, maturation and activity. As a proof of principle we wanted to answer the question whether activation of the -secretase Falk Fahrenholz Becherweg 30 DE–55099 Mainz (Germany) Tel. +49 6131 392 5833, Fax +49 6131 392 5348, E-Mail [email protected]

APP

Fig. 1. Thioflavine S-stained amyloid plaques in the subiculum from 18-monthold APP transgenic mice and either double-transgenic ADAM10 ! APP, or ADAM10-E384A ! APP mice. The human APP variant expressed in all strains is the V717I mutant. In double-transgenic mice overexpressing the -secretase ADAM10, amyloid plaque formation is nearly prevented. In mice overexpressing the catalytically inactive ADAM10-E384A mutant, amyloid plaque formation is significantly enhanced because of inhibition of endogenous -secretase activity.

Activation of -secretase

ADAM10 × APP

in vivo might be beneficial for the treatment of AD. Finally, we explored different approaches to upregulate -secretase activity. This review will focus mainly on results obtained within the priority program, but other recent approaches to upregulate -secretase activity will be included to complete the evaluation of the -secretase as a potential therapeutic target. Identification of ADAM10 as an -Secretase

The APP belongs to the family of type I membrane glycoproteins and is constitutively expressed in many types of mammalian cells. Its cleavage by -secretases was the first proteolytic pathway of APP to be characterized in detail [2–4]. The trafficking of APP mainly follows the constitutive secretory route which involves cleavage by a putative -secretase in the trans-Golgi network [5, 6] and at the cell surface [2]. The -secretase cleaves within the A sequence and generates a soluble N-terminal APP fragment (APPs) of 105–125 kDa and a C-terminal APP fragment of about 10 kDa (p10). Further processing of p10 by the -secretase yields truncated A fragments of about 3 kDa (p3) that generally are not found in amyloid cores of classical plaques, or in amyloid deposits in the cerebral vasculature. Studies in various cell types confirmed that the major -secretase cleavage site is between lysine-16 and leucine-17 in the A domain [3]. To explore upregulation of the -secretase activity as an approach for AD therapy, it was essential to identify 256

Neurodegenerative Dis 2006;3:255–261

Inhibition of -secretase

ADAM10-E384A × APP

the proteinase responsible for -secretase activity. In 1999 we demonstrated that the disintegrin and metalloproteinase ADAM10 has -secretase activity in vitro and in cultured cells [7]: (a) Purified ADAM10 cleaved APPderived peptides at the -secretase cleavage site. (b) Overexpression of ADAM10 in several cell lines resulted in an increased -secretase activity for the different isoforms of APP. (c) Expression of mutated ADAM10, containing the amino acid substitution E384A in its zinc-binding site, significantly decreased the endogenous -secretase activity. (d) It was possible to stimulate the -secretase activity of ADAM10 with phorbol esters, and to inhibit it by hydroxamic acid-based inhibitors for metalloproteinases. Thus, ADAM10 exhibits many properties of a physiologically relevant -secretase, as expected from various publications since 1992. It has been hypothesized for a long time that upregulation of -secretase activity might preclude the formation of A peptides and their deposition as plaques. To investigate this concept in vivo, we generated transgenic mice which overexpress either ADAM10 or the catalytically inactive ADAM10 mutant. We found that even a moderate neuronal overexpression of ADAM10 in mice transgenic for human APP[V717I] increased the release of the neurotrophic N-terminal APP domain (APPs), reduced the formation of A40 and A42, and prevented their deposition in plaques (fig. 1). Functionally, impaired long-term potentiation and cognitive deficits were alleviated. Expression of the catalytically inactive ADAM10 led to an enhancement of the number and size of amyloid plaques in the brains of double-transgenic mice [8]. Fahrenholz/Postina

In our study, we demonstrate a competition of the -secretase ADAM10 with the -secretase BACE for the substrate APP: overexpression of ADAM10 in young animals inhibited the production of APPs and A, which in older animals led to an almost complete prevention of plaque formation. Although there is evidence that ADAM10 plays a role in the Notch signaling pathway during neuronal development, no indication was found that neuronal ADAM10 overexpression increases Notch signaling in the brains of adult ADAM10 transgenic mice. Neuronal overexpression of ADAM10 had no detrimental effects on ADAM10 single-transgenic mice: these animals exhibited normal behavioral abilities [9]. Gene profiling studies of transgenic mice recently showed that overexpression of ADAM10 does not lead to an increased expression of genes coding for proinflammatory or proapoptotic proteins [unpubl. data]. These results are supportive for the strategy to treat AD by increasing the secretase activity. Regulation of the -Secretase ADAM10 by Its Prodomain and by Proprotein Convertases

These results suggest a new approach to enhance the nonamyloidogenic -secretase pathway: Inhibitors or antibodies with the ability to prevent the association of the prodomain and the catalytic domain of ADAM10 should be able to increase the -secretase activity. In addition, upregulation of the PC activity should promote the nonamyloidogenic -secretase pathway. Tumor necrosis factor--converting enzyme (TACE) or ADAM17 is another member of the ADAM family that shows -secretase activity in cellular systems. It is converted by the same PCs (furin and PC7) to its mature form [15]. Furin and PC7– according to our studies – are also involved in maturation of ADAM10. We found that long-term treatment of several cell lines with phorbol esters decreases the cellular amount of mature TACE. This effect was specific for TACE, as mature ADAM10 was not affected by a similar treatment. Thus, mature forms of TACE and ADAM10 differ in their cellular stability, which may affect their -secretase activity in vivo. Transcriptional Regulation of the Human -Secretase ADAM10 Gene

For some of the ADAM proteinases it has been shown that the catalytic site is maintained inactive via a socalled cysteine switch mechanism performed by their Nterminal prodomain [10, 11]. The essential step for zymogen activation is the proteolytic processing by proprotein convertases at a characteristic motif, which is located between the prodomain and the metalloproteinase domain of the ADAMs. Proprotein convertases (PCs) form a family of calcium-dependent endoproteinases, which comprises seven distinct members [12, 13]. We investigated the role of PCs and of the ADAM10 prodomain in the regulation of the -secretase activity of ADAM10 [14]. Overexpression of the PCs PC7 or furin in cell lines revealed increased ADAM10 maturation resulting in enhanced -secretase-mediated processing of APP. A mutation of the PC recognition sequence in ADAM10 as well as the use of a PC inhibitor and of a furin-deficient cell line confirmed the role of PCs, in particular of PC7, in ADAM10 maturation and activation. It is of note that the prodomain of ADAM10 has a dual function: When coexpressed in trans as a separate polypeptide, the prodomain acted as an internal chaperone and functionally rescued the -secretase activity of a former inactive ADAM10 mutant lacking the prodomain. On the other hand, it inhibited the activity of wildtype ADAM10.

Since enhancing the ADAM10 gene expression appears to be a reasonable approach for the treatment of AD, we functionally analyzed the human ADAM10 gene [16]. Both human and mouse ADAM10 genes comprise 160 kbp and are composed of 16 exons. The 5-flanking regions of human, mouse, and rat ADAM10 genes have a high degree of identity within the first 500 bp upstream of each translation start site. The gene expression of ADAM10 in these species appears to be regulated in a similar way, because multiple putative transcription factor-binding sites are found to be preserved between –508 and –1 bp. We observed the highest ADAM10 promoter activity in neuronal cell lines, and lower activities in kidney and liver cells. This finding is in agreement with reports showing that ADAM10 is expressed ubiquitously [17, 18], but additionally indicates a favored transcription of ADAM10 in the brain. By a combination of several approaches we identified nucleotides –508 to –300 as the core promoters, and found Sp1 and USF elements to modulate its activity. Some of the potential transcription factor-binding sites found in the human ADAM10 gene are also present in the promoters of other genes involved in AD, including the APP and presenilin genes: Sp1 sites have been reported for the APP [19], presenilin-1 [20] and presenilin-2 [21]

-Secretase Activation – An Approach to AD Therapy

Neurodegenerative Dis 2006;3:255–261

257

promoters. In vitro transcription and cotransfection studies showed that USF activates transcription from the human APP promoter [22]. Deletion of Brn-2, SREBP, Oct-1 and CREB1/cJun consensus sequences did not significantly affect ADAM10 promoter activity, suggesting that these elements are not functionally important for constitutive promoter activity. This result does not exclude a role of the corresponding transcription factors in regulating the expression of ADAM10 in response to certain stimuli in other cellular backgrounds. Two putative binding sites for the retinoic acid (RA) receptor are located 302 and 203 nucleotides upstream of the translation initiation site of the ADAM10 gene. Thus, an effect of RA on ADAM10 gene expression is possible. We analyzed the effect of RA on an ADAM10 promoter luciferase reporter gene stably expressed in neuroblastoma cells. After treatment for 4 days, the luciferase activity was significantly increased. Moreover, endogenous ADAM10 mRNA levels were also increased upon RA treatment. To discover the site of RA receptor action, EMSAs were performed. We found region 203 upstream of the translation initiation site to be protein-bound; for site 302, no protein/DNA interaction could be disclosed. By identifying RA as an activator of the -secretase ADAM10 promoter, we now link two cellular pathways which are involved in the pathogenesis of AD: reduced -secretase activity, and impaired retinol metabolism [23].

lines upregulated the expression of APP, APLP2 and ADAM10, thus leading to an increased release of APPs and soluble APLP2. Because TACE was not positively affected by RA, but even degraded in neuroblastoma cells, we again demonstrated a higher stability of ADAM10 compared with TACE, which was also degraded selectively after PMA treatment of cultured cells [15]. Since ADAM10 has been implicated in ectodomain shedding of other substrates, its common upregulation with its substrates APP and APLP2 may result in a preferential cleavage of these two substrates compared to other ADAM10 substrates during cellular differentiation. In late-onset AD, there is genetic, metabolic and dietary evidence for defective retinoid transport and function [23, 25, 26]. In accordance with these findings is the observation that impairment of long-term potentiation induced by experimental vitamin A deficiency in adult mice can be reversed by direct application of RA to hippocampal slices [27]. Because ADAM10 together with its substrates in cell cultures is upregulated via RA, our results suggest that bioactive retinoids in the hippocampus lead to an increased -secretase activity, and to an increased release of the neurotrophic soluble ectodomains of APP and APLP2. These findings suggest that pharmacological targeting of retinoid receptors may increase the expression of the -secretase ADAM10 with beneficial effects on AD pathology. Activation of -Secretase via G Protein-Coupled Receptors

Common Upregulation of ADAM10 and Its Substrates APP and APP-Like Protein 2 by RA

The APP-like protein 2 (APLP2) which has been shown to be essential for the development and survival of mice, is also a substrate for the two proteinases with -secretase activity, ADAM10 and ADAM17 [24]. Overexpression of either ADAM10 or TACE in cultured cells increased the release of neurotrophic soluble APLP2 several fold. The strongest inhibition of APLP2 shedding in neuroblastoma cells was observed with an ADAM10-preferring inhibitor. Transgenic mice with neuron-specific overexpression of ADAM10 showed significantly increased levels of soluble APLP2 and its Cterminal fragments. To elucidate a possible regulatory mechanism of APP and APLP2 shedding in the neuronal context, we examined RA-induced differentiation of neuroblastoma cells. RA treatment of two neuroblastoma cell 258

Neurodegenerative Dis 2006;3:255–261

Stimulation of G protein-coupled receptors and their downstream signal transduction pathways localized in brain areas affected by AD is an approach to activate the nonamyloidogenic pathway. This principle has been validated by in vivo studies with a new class of M1 agonists in AD patients [28], and in a transgenic mouse model [29]. Activators of protein kinase C stimulated the secretase pathway and attenuated symptoms of AD pathology in transgenic mouse models [30]. Considering the beneficial effect of -secretase overexpression, it will certainly be necessary to further explore this approach. Recently, Kojro et al. [31] discovered that the neuropeptide pituitary adenylate cyclase-activating polypeptide (PACAP) promotes the -secretase pathway. PACAP has neurotrophic as well as antiapoptotic properties, and is involved in learning and memory processes. Its specific G protein-coupled receptor PAC1 is expressed in sevFahrenholz/Postina

Fig. 2. Proteinases as targets for therapy of AD. Activation of the nonamyloidogenic APP processing pathway is of particular interest because in addition to preventing generation of neurotoxic A peptides the release of neurotrophic APPs is mediated solely by this pathway. The nonamyloidogenic pathway can be enhanced, for example, by neuropeptides such as PACAP, PKC activators,

statins and retinoids. Generation of neurotoxic A peptides may also be blocked by specific inhibitors for - and -secretases which, in addition, may also shift APP processing into the nonamyloidogenic pathway. Furthermore, degradation of A peptides could be enhanced by activation of neprilysin or insulin-degrading enzyme (IDE).

eral CNS regions including the hippocampal formation. Stimulation of endogenously expressed PAC1 receptors with PACAP in human neuroblastoma cells increased APPs secretion which was completely inhibited by the PAC1 receptor-specific antagonist PACAP(6–38). In cells stably overexpressing functional PAC1 receptors, PACAP-27 and PACAP-38 strongly stimulated -secretase cleavage of APP. This increase in -secretase activity was completely abolished by hydroxamate-based metalloproteinase inhibitors. By using several specific protein kinase inhibitors it was found that the MAP-kinase pathway and the phosphatidylinositol-3-kinase mediate PACAP-induced -secretase activation. These findings identify PACAP peptides as stimulators of the nonamyloidogenic pathway, which might be related to their neuroprotective properties. Since a peptide transport system has been identified, which allows the bi-directional transport of PACAP across the blood-brain barrier [32], PAC1 receptor agonists may therefore be useful for the treatment of AD.

of enhanced -secretase activity. Increased membrane fluidity and impaired internalization of APP were responsible for the effect observed after a reduction in membrane cholesterol with methyl--cyclodextrin. Furthermore, treatment with lovastatin resulted in higher protein levels of the -secretase ADAM10 [33]. It has been reported that simvastatin treatment of a small group of AD patients affected the brain cholesterol metabolism and favored the nonamyloidogenic pathway of APP processing [34]. However, statins turn out not to act simply through -secretase upregulation alone. Due to their multiple modes of action, they exert their antiAD effects also via other mechanisms, such as their antiinflammatory or antioxidant properties.

Low Cholesterol Stimulates the Nonamyloidogenic Pathway

Cellular studies have shown that a reduction in cholesterol channels APP to the nonamyloidogenic processing. Several mechanisms were elucidated to be the basis -Secretase Activation – An Approach to AD Therapy

Concluding Remarks

Despite potential problems emerging from -secretase upregulation, it is encouraging that a variety of currently available medications and endogenous hormones, with only few side effects, have been shown to increase secretase activity at the cellular level and in animal models (fig. 2). The proof of the concept that -secretase upregulation could have beneficial effects has been provided by our transgenic mouse model with moderate overexpression of ADAM10.

Neurodegenerative Dis 2006;3:255–261

259

It has been shown recently that acetylcholine esterase inhibitors, in addition to their classical role, support the nonamyloidogenic pathway. One mechanism which has been suggested to explain their ‘disease-modifying effects’ is the enhanced transport of ADAM10 and APP to the cell surface [35]. Moreover, increased levels of acetylcholine may contribute via G protein-coupled muscarinic receptors to an increase in endogenous -secretase activity. Another noteworthy recent finding is that caloric restriction may promote the nonamyloidogenic APP processing. Caloric intake and metabolic defects in terms of insulin resistance have been linked to AD, but the underlying mechanism is not clear. In vivo studies in APP transgenic mice demonstrated that caloric restriction

shifted APP processing to the nonamyloidogenic pathway, which could potentially be attributed to a statistically significant increase in ADAM10 levels [36]. Further studies with medications and dietary regimens that enhance the nonamyloidogenic pathway of APP processing are thus valuable approaches for AD therapy.

Acknowledgments The authors thank Dr. Christian Haass (Munich), Dr. Ulrike Müller (Heidelberg), Dr. Gerd Multhaup (Berlin) and Dr. Boris Schmidt (Darmstadt) for cooperation in the DFG priority program. This work was supported by the Deutsche Forschungsgemeinschaft within the priority program SPP1085 – Cellular Mechanisms of Alzheimer’s Disease.

References 1 Kojro E, Fahrenholz F: The non-amyloidogenic pathway: structure and function of alpha-secretases. Subcell Biochem 2005; 38: 105–127. 2 Sisodia SS: Beta-amyloid precursor protein cleavage by a membrane-bound protease. Proc Natl Acad Sci USA 1992; 89: 6075– 6079. 3 Esch FS, Keim PS, Beattie EC, et al: Cleavage of amyloid beta peptide during constitutive processing of its precursor. Science 1990; 248:1122–1124. 4 Pasternack JM, Palmert MR, Usiak M, et al: Alzheimer’s disease and control brain contain soluble derivatives of the amyloid protein precursor that end within the beta amyloid protein region. Biochemistry 1992; 31: 10936–10940. 5 Kuentzel SL, Ali SM, Altman RA, Greenberg BD, Raub TJ: The Alzheimer beta-amyloid protein precursor/protease nexin-II is cleaved by secretase in a trans-Golgi secretory compartment in human neuroglioma cells. Biochem J 1993;295:367–378. 6 Sambamurti K, Shioi J, Anderson JP, Pappolla MA, Robakis NK: Evidence for intracellular cleavage of the Alzheimer’s amyloid precursor in PC12 cells. J Neurosci Res 1992; 33:319–329. 7 Lammich S, Kojro E, Postina R, et al: Constitutive and regulated alpha-secretase cleavage of Alzheimer’s amyloid precursor protein by a disintegrin metalloprotease. Proc Natl Acad Sci USA 1999;96:3922–3927. 8 Postina R, Schroeder A, Dewachter I, et al: A disintegrin-metalloproteinase prevents amyloid plaque formation and hippocampal defects in an Alzheimer disease mouse model. J Clin Invest 2004;113:1456–1464.

260

9 Schmitt U, Hiemke C, Fahrenholz F, Schroeder A: Phenotyping of mice overexpressing two different forms of the alpha-secretase ADAM10. Neurosci Meet 2004. 10 Loechel F, Overgaard MT, Oxvig C, Albrechtsen R, Wewer UM: Regulation of human ADAM 12 protease by the prodomain. Evidence for a functional cysteine switch. J Biol Chem 1999; 274:13427–13433. 11 Milla ME, Leesnitzer MA, Moss ML, et al: Specific sequence elements are required for the expression of functional tumor necrosis factor-alpha-converting enzyme (TACE). J Biol Chem 1999; 274:30563–30570. 12 Steiner DF: The proprotein convertases. Curr Opin Chem Biol 1998; 2:31–39. 13 Seidah NG, Chretien M: Proprotein and prohormone convertases: a family of subtilases generating diverse bioactive polypeptides. Brain Res 1999;848:45–62. 14 Anders A, Gilbert S, Garten W, Postina R, Fahrenholz F: Regulation of the alpha-secretase ADAM10 by its prodomain and proprotein convertases. FASEB J 2001; 15: 1837– 1839. 15 Endres K, Anders A, Kojro E, Gilbert S, Fahrenholz F, Postina R: Tumor necrosis factoralpha converting enzyme is processed by proprotein-convertases to its mature form which is degraded upon phorbol ester stimulation. Eur J Biochem 2003;270:2386–2393. 16 Prinzen C, Muller U, Endres K, Fahrenholz F, Postina R: Genomic structure and functional characterization of the human ADAM10 promoter. FASEB J 2005;19:1522– 1524. 17 Howard L, Lu X, Mitchell S, Griffiths S, Glynn P: Molecular cloning of MADM: a catalytically active mammalian disintegrinmetalloprotease expressed in various cell types. Biochem J 1996;317:45–50.

Neurodegenerative Dis 2006;3:255–261

18 Yavari R, Adida C, Bray-Ward P, Brines M, Xu T: Human metalloprotease-disintegrin Kuzbanian regulates sympathoadrenal cell fate in development and neoplasia. Hum Mol Genet 1998;7:1161–1167. 19 Izumi R, Yamada T, Yoshikai S, Sasaki H, Hattori M, Sakaki Y: Positive and negative regulatory elements for the expression of the Alzheimer’s disease amyloid precursor-encoding gene in mouse. Gene 1992; 112: 189– 195. 20 Mitsuda N, Roses AD, Vitek MP: Transcriptional regulation of the mouse presenilin-1 gene. J Biol Chem 1997; 272:23489–23497. 21 Pennypacker KR, Fuldner R, Xu R, et al: Cloning and characterization of the presenilin-2 gene promoter. Brain Res Mol Brain Res 1998;56:57–65. 22 Kovacs DM, Wasco W, Witherby J, et al: The upstream stimulatory factor functionally interacts with the Alzheimer amyloid betaprotein precursor gene. Hum Mol Genet 1995;4:1527–1533. 23 Goodman AB, Pardee AB: Evidence for defective retinoid transport and function in late onset Alzheimer’s disease. Proc Natl Acad Sci USA 2003;100:2901–2905. 24 Endres K, Postina R, Schroeder A, Mueller U, Fahrenholz F: Shedding of the amyloid precursor protein-like protein APLP2 by disintegrin-metalloproteinases. FEBS J 2005;272:5808–5820. 25 Puchades M, Hansson SF, Nilsson CL, Andreasen N, Blennow K, Davidsson P: Proteomic studies of potential cerebrospinal fluid protein markers for Alzheimer’s disease. Brain Res Mol Brain Res 2003;118:140– 146.

Fahrenholz/Postina

26 Rinaldi P, Polidori MC, Metastasio A, et al: Plasma antioxidants are similarly depleted in mild cognitive impairment and in Alzheimer’s disease. Neurobiol Aging 2003; 24: 915–919. 27 Misner DL, Jacobs S, Shimizu Y, et al: Vitamin A deprivation results in reversible loss of hippocampal long-term synaptic plasticity. Proc Natl Acad Sci USA 2001; 98: 11714– 11719. 28 Nitsch RM, Deng M, Tennis M, Schoenfeld D, Growdon JH: The selective muscarinic M1 agonist AF102B decreases levels of total Abeta in cerebrospinal fluid of patients with Alzheimer’s disease. Ann Neurol 2000; 48: 913–918. 29 Caccamo A, Oddo S, Billings LM, Green KN, Martinez-Coria H, Fisher A, LaFerla FM: M1 receptors play a central role in modulating AD-like pathology in transgenic mice. Neuron 2006;49:671–682.

-Secretase Activation – An Approach to AD Therapy

30 Etcheberrigaray R, Tan M, Dewachter I, et al: Therapeutic effects of PKC activators in Alzheimer’s disease transgenic mice. Proc Natl Acad Sci USA 2004;101:11141–11146. 31 Kojro E, Postina R, Buro C, Meiringer C, Gehrig-Burger K, Fahrenholz F: The neuropeptide PACAP promotes alpha-secretase pathway for processing Alzheimer amyloid precursor protein. FASEB J 2006; 20: 512– 514. 32 Dogrukol-Ak D, Tore F, Tuncel N: Passage of VIP/PACAP/secretin family across the blood-brain barrier: therapeutic effects. Curr Pharm Des 2004;10:1325–1340.

33 Kojro E, Gimpl G, Lammich S, Marz W, Fahrenholz F: Low cholesterol stimulates the nonamyloidogenic pathway by its effect on the alpha-secretase ADAM 10. Proc Natl Acad Sci USA 2001;98:5815–5820. 34 Hoglund K, Thelen KM, Syversen S, et al: The effect of simvastatin treatment on the amyloid precursor protein and brain cholesterol metabolism in patients with Alzheimer’s disease. Dement Geriatr Cogn Disord 2005;19:256–265. 35 Zimmermann M, Gardoni F, Marcello E, et al: Acetylcholinesterase inhibitors increase ADAM10 activity by promoting its trafficking in neuroblastoma cell lines. J Neurochem 2004;90:1489–1499. 36 Wang J, Ho L, Qin W, et al: Caloric restriction attenuates beta-amyloid neuropathology in a mouse model of Alzheimer’s disease. FASEB J 2005;19:659–661.

Neurodegenerative Dis 2006;3:255–261

261

Original Paper

Diseases

Neurodegenerative Dis 2006;3:262–269 DOI: 10.1159/000095265

Ectodomain Shedding of the Amyloid Precursor Protein: Cellular Control Mechanisms and Novel Modifiers Stefan F. Lichtenthaler Adolf Butenandt Institute, Ludwig Maximilians University, Munich, Germany

Key Words Amyloid precursor protein Alzheimer’s disease Ectodomain shedding Secretases Endocytosis Endophilin

recent developments in our understanding of the cellular regulation of APP -secretase cleavage. Moreover, it highlights the particular importance of endocytic APP trafficking as a prime modulator of APP shedding. Copyright © 2006 S. Karger AG, Basel

Abstract Proteolytic cleavage in the ectodomain of the amyloid precursor protein (APP) is a key regulatory step in the generation of the Alzheimer’s disease amyloid- (A) peptide and occurs through two different protease activities termed and -secretase. Both proteases compete for APP cleavage, but have opposite effects on A generation. At present, little is known about the cellular pathways that control APP - or -secretase cleavage and thus A generation. To explore the contributory pathways in more detail we have recently employed an expression cloning screen and identified several activators of APP cleavage by - or -secretase. Among them were known activators of APP cleavage, for example protein kinase A, and novel activators, such as endophilin and the APP homolog amyloid precursor-like protein 1 (APLP1). Mechanistic analysis revealed that both endophilin and APLP1 reduce the rate of APP endocytosis and strongly increase APP cleavage by -secretase. This review summarizes the results of the expression cloning screen in the context of

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0262$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Proteolytic Processing of APP

The amyloid precursor protein (APP) is one of a large number of membrane proteins that are proteolytically converted to their soluble counterparts. This process is referred to as ectodomain shedding and is an important way of regulating the biological activity of membrane proteins [reviewed in 1, 2]. The shedding of APP may occur through two different protease activities termed and -secretase, which cleave APP within its ectodomain close to its transmembrane domain [for a review, see 3]. APP cleavage by - or -secretase is a key regulatory process in the generation of the amyloid- (A) peptide, which is assumed to play an essential role in the pathogenesis of Alzheimer’s disease (AD). -Secretase, which is the aspartyl protease BACE1, cleaves APP at the N-terminus of the A peptide domain and thus catalyzes the first step in A peptide generation [4]. Subsequently, the Stefan F. Lichtenthaler Adolf Butenandt Institute, Ludwig Maximilians University Schillerstrasse 44, DE–80336 Munich (Germany) Tel. +49 89 2180 75453, Fax +49 89 2180 75415 E-Mail [email protected]

whereas -secretase cleavage of APP mainly occurs after endocytosis in the endosomes. Third, the interaction of APP with cytoplasmic adaptor proteins alters APP shedding [for a review, see 10], presumably by affecting APP trafficking and access of APP to the secretases. For all three categories the molecular mechanisms underlying the increase in APP shedding are only partly understood. Thus, we have recently employed expression cloning to explore the contributory cellular pathways systematically and obtained several proteins activating APP shedding [11, 12]. The identified proteins fall into the three general categories of APP shedding activators described above. This review summarizes their mechanistic analysis in the context of recent developments in our understanding of the cellular regulation of APP -cleavage. This highlights the particular importance of the endocytic trafficking of APP as a prime modulator of APP shedding. Fig. 1. Ectodomain shedding of APP shedding by - and -secre-

tase. -Secretase cleavage of wild-type APP occurs at or very close to the cell surface, whereas -secretase cleavage mainly takes place after endocytosis in the endosomes.

Expression Cloning Screen for Modifiers of APP Shedding

remaining C-terminal APP fragment is cleaved by secretase within its transmembrane domain at the C-terminus of the A domain, leading to the secretion of the A peptide [5]. In contrast to -secretase, -secretase cleaves within the A sequence, and thereby precludes the generation of the A peptide. -Secretase is a member of the ADAM (a disintegrin and metalloprotease) family of proteases [for a review, see 6]. - and -secretase compete for the ectodomain cleavage of APP [7] (fig. 1), but have opposite effects on A generation. Additionally, - but not -secretase generates a secreted form of APP (APPs), which has neurotrophic and neuroprotective properties [reviewed in 8]. Thus, shifting APP shedding away from - towards -secretase cleavage may be therapeutically beneficial for AD. In order to do so, it is essential to understand the cellular pathways that regulate the activity of both proteases. At present, little is known about the cellular regulation of BACE1. In contrast, APP -secretase cleavage can be regulated in the cell through different mechanisms, which can be broadly grouped into three categories. First, an activation of distinct intracellular signaling mechanisms or a change in the membrane composition increases APP -shedding [for reviews, see 6, 8]. Second, changes in APP endocytosis alter - and -secretase cleavage, because -secretase cleavage occurs at or very close to the plasma membrane [9],

For the expression cloning screen, a reporter cell line was used that allows measurement of APP shedding in a high-throughput format. For this aim, human embryonic kidney 293 cells were used that stably express a fusion protein consisting of alkaline phosphatase fused to the N-terminus of full-length APP [12]. The 293 cells are a well-established cellular model for the analysis of APP shedding and have the additional advantage that they can be transfected with very high efficiency, which was an essential requirement for the screening approach used. In a ‘sib-selection’ or ‘pool-subdivision’ approach, we first used pools of 96 cDNAs from a human brain cDNA library and screened them for activators of APP shedding (fig. 2). Next, to identify the individual cDNA within the pool, which was responsible for activation of APP shedding, the pooled cDNAs were further subdivided and individual cDNAs from that pool were tested for their APP shedding-enhancing activity. With this approach, eight cDNAs were obtained that stimulate the shedding of APP (table 1). They encode protein kinase A (PKA), an N-terminally truncated form of the kinase MEKK2, metabotropic glutamate receptor 3 (mGluR3), endophilins A1 and A3, numblike, an N-terminally truncated form of the palmitoyl-protein thioesterase 1 and the APP homolog amyloid precursor-like protein 1 (APLP1) [11, 12]. cDNAs inhibiting APP secretion were not obtained. Altogether, around 100,000 cDNAs were screened. Considering that cDNA libraries contain many partial cDNAs

Cellular Control of APP Shedding

Neurodegenerative Dis 2006;3:262–269

263

Fig. 2. Expression cloning screen for mod-

ulators of APP shedding. The reporter cell line consists of human embryonic kidney 293 cells, which stably express a fusion protein of alkaline phosphatase (AP) and full-length APP. Cleavage of APP by - or -secretase leads to the secretion of the fusion protein into the conditioned medium, where it can be detected and quantified by measuring the alkaline phosphatase reporter enzyme activity. Cells were plated into 96-well plates and transfected with pools of cDNAs. APP fusion protein secretion was measured in all wells. In wells with altered APP secretion, the corresponding cDNA was identified and mechanistically characterized.

AP

Transfection of cDNA library

AP APP

Measurement of APP secretion

APP

✂ Identification of cDNA Reporter cell line for APP shedding

Table 1. Activators of APP shedding obtained by expression cloning

Protein encoded by cDNA

Full-length or partial cDNA

Protein kinase A, catalytic -subunit MEKK2 (member of the MAPKKK family)

full-length lacking the 5 end of the coding sequence; encoding an N-terminally truncated protein full-length full-length full-length full-length lacking the 5 end of the coding sequence; encoding an N-terminally truncated protein full-length

mGluR3 Endophilin A1 Endophilin A3 Numblike Palmitoyl-protein thioesterase 1 APLP1

and that cDNAs expressed at high levels are overrepresented in the library, we assume that many more cDNA clones would need to be screened to cover all distinct cDNAs found in the library. The identification of PKA is in agreement with previous publications showing that an activation of PKA by forskolin in rat pheochromocytoma PC12 cells [13] and in human embryonic kidney 293 cells [14] increased APP shedding. This validates the screening approach as it shows that physiologically relevant cDNAs can be obtained.

Specificity of Identified cDNAs for APP Shedding

ding of unrelated membrane proteins, such as TNF receptor 2 (TNFR2), P-selectin glycoprotein ligand-1 (PSGL-1) or L-selectin [12]. Like APP, all three proteins are subject to ectodomain shedding by ADAM proteases. APLP1 strongly activated shedding of APP but not of L-selectin (fig. 3), demonstrating that APLP1 does not stimulate the shedding of all ADAM protease substrates. Other proteins, such as PKA and the kinase MEKK2, activated the shedding of L-selectin (fig. 3, shown for MEKK2) or of TNFR2 [12] much more strongly than the shedding of APP, showing that they are not specific activators of APP shedding, but instead may contribute to a general cellular program controlling ectodomain shedding.

Some of the identified cDNAs, such as the endophilins and mGluR3, activated APP shedding in a relatively specific manner, as they had essentially no effect on the shed264

Neurodegenerative Dis 2006;3:262–269

Lichtenthaler

modulate APP -secretase cleavage. The underlying molecular mechanisms are partly understood and have been described in more detail elsewhere [for a review, see 6]. One of the proteins identified in the screen, mGluR3, is a novel activator of APP shedding and belongs to the eight-member family of mGluRs. mGluR3 may be particularly interesting for studying APP shedding, as it was one of the cDNAs showing a specific effect on the shedding of APP. Two members of the mGluR family, mGluR1 and mGluR5, have previously been shown to stimulate the secretion of APP [16]. Both mGluRs activate phospholipase D. In contrast, the identified mGluR3 negatively regulates adenylate cyclase and thus, points to a possible role of this separate pathway in the control of APP shedding.

8

APP 6

Rel. alkaline phosphatase acitvity

4 2 0 14 12

L-selectin

10 8 6 4

Control of APP Shedding by Modulators of General Endocytosis

2 0 Con

APLP1

MEKK2

The APP shedding activators PKA and MEKK2 are part of signal transduction pathways, which have previously been shown to control the amount of APP shedding. The -secretase cleavage of APP can be stimulated by MAP kinase signaling, insulin signaling and signaling through PKA or specific G protein-coupled receptors [for reviews see 6, 8] as well as by calcium [15]. For example, growth factors, such as EGF, or the phorbol ester PMA can activate the MAP kinase cascade and stimulate APP shedding. The molecular and cellular processes activated by these cascades and the mechanisms by which they finally mediate the increase in APP shedding remain largely unknown. Additionally, a diverse group of compounds, such as cholesterol, steroid hormones, nonsteroidal antiinflammatory drugs and cholinesterase inhibitors can

A strong activator of APP shedding identified in the screen was endophilin A3, which belongs to the endophilin family of endocytic and signal transducing proteins [17]. Endophilin consists of a Bin/amphiphysin/Rvs domain, which may be involved in protein dimerization and sensing of membrane curvature, and of a SH3 domain. Endophilin A3 increased APP shedding even stronger than the metalloprotease ADAM10 [11, 12], which is one of the candidate -secretases for APP. Importantly, we found that endophilin A3 specifically increased APP -secretase cleavage and had no significant effect on -secretase cleavage. Mechanistically, endophilin A3 inhibits the rate of APP endocytosis, as measured in a validated anti-APP antibody uptake assay using COS cells cotransfected with APP and either endophilin A3 or GFP as a control [11, 12]. As a result, more APP becomes available at the cell surface for an increased -secretase cleavage. This fits with previous studies showing that a mutant form of APP, which lacks its cytoplasmic domain including its internalization motif, shows a strong reduction in endocytosis resulting in more APP at the cell surface and increased APP shedding [18]. A strong increase in APP shedding, mainly mediated by -secretase, was also observed for a dominant-negative mutant of the endocytic GTPase dynamin, which inhibits endocytosis of many membrane proteins, including APP [19, 20]. Conversely, expression of the small G protein Rab5 in murine L1 cells enhances APP endocytosis, resulting in an increased APP cleavage by -secretase, in increased A generation and in abnormally enlarged endosomal structures [21].

Cellular Control of APP Shedding

Neurodegenerative Dis 2006;3:262–269

Fig. 3. Specificity of the stimulatory effect of APLP1 on the shed-

ding of APP. Kidney 293 cells stably expressing alkaline phosphatase fusion proteins of APP or L-selectin were transiently transfected with control vector (Con), APLP1 or MEKK2. AP activity was measured in the conditioned medium and represents the mean and standard deviation of two to three independent experiments, each one carried out in duplicate. Alkaline phosphatase activity was normalized to the protein concentration in the cell lysate. The data for APP were part of the set of experiments shown in Neumann et al. [11].

Cellular Control of APP Ectodomain Shedding by Signal Transduction Cascades

265

Expression of endophilin A3 not only inhibited the endocytosis of APP but also the endocytosis of fluorescently labeled transferrin [12], revealing that endophilin A3 is a general, negative regulator of endocytosis, potentially similar to mutants of the endocytic GTPase dynamin. Despite its more general role in endocytosis, the endophilin A3 had a strong effect only on APP shedding but no or only a minor effect on the shedding of other membrane proteins, such as TNFR2 or PSGL-1 [12]. This suggests that APP stands out among other shedding substrates in that its shedding is particularly sensitive to changes in the rate of endocytosis. At present, it is unclear why this is so. Potentially, the endocytosis of TNFR2 and PSGL-1 is regulated differently than the endocytosis of APP or may have a different time course.

A APP

LRP

B

C

APP

APP

LRP

APLP1

M FE65

FE65

FE65

APP endocytosis

Fig. 4. Model for the APP-FE65-LRP complex. A In LRP express-

APP Interactors Influence APP Shedding

APP is at the center of a complex protein-protein interaction network involving cytoplasmic adaptor and transmembrane proteins, but the functional role of these interactions is only partly understood. Most of the cytoplasmic interactors seem to compete for the same binding site at or around the conserved GYENPTY motif in the cytoplasmic tail of APP [for recent reviews on APP interactors, see 10, 22]. For example, FE65, X11, and JIP have been shown to bind to this motif and to alter APP shedding, revealing that a change in the interaction of APP with its binding partners is a way to modulate APP shedding. FE65 and X11 have been best studied among the interactors and have opposite effects on APP cleavage. X11 decreases APP shedding, presumably by retaining APP in early compartments of the secretory pathway. In contrast, FE65 stimulates APP shedding. A specific mechanism of how FE65 controls the shedding of APP, but not of unrelated membrane proteins, has been put forward by us and several other groups using different experimental approaches [11, 23–27]. According to this model, FE65 links APP to the LDL receptorrelated protein (LRP; fig. 4A), which is a multifunctional cell surface receptor for proteins involved in lipoprotein metabolism [28]. Formation of the APP-FE65-LRP complex allows efficient endocytosis of APP (fig. 4A). In contrast, disruption of the complex leads to a reduction in APP endocytosis, resulting in an accumulation of APP at the cell surface, where it undergoes increased -secretase cleavage and reduced -secretase cleavage (fig. 4B). This is the case in LRP-deficient cells or upon RNAi-mediated knockdown of FE65 [23, 24, 29]. Like266

Neurodegenerative Dis 2006;3:262–269

ing wild-type cells APP, FE65 and LRP form a complex, allowing efficient APP endocytosis (bold vertical arrow) and resulting in low levels of APP shedding (thin horizontal arrow). Size of proteins is not drawn to scale. B In LRP-deficient cells (LRP–/–), endocytosis of APP is reduced and APP shedding is increased (bold horizontal arrow). C In cells transfected with APLP1, LRP preferentially forms a complex with APLP1, resulting in APP not being complexed to LRP. This results in a state resembling LRP deficiency (B) and an increase in APP shedding (bold horizontal arrow). M = Membrane. Figure adapted from Neumann et al. [11].

wise, overexpression of FE65 results in the disruption of the complex, presumably by leading to APP-FE65 complexes and to separate FE65-LRP complexes. Both kinds of complexes lack the third binding partner and therefore are not functional with regard to APP endocytosis.

APLP1 Modulates APP Shedding

An essential validation of the model described above came from work that we carried out in collaboration with Claus Pietrzik and Christian Haass [11]. We found that the APP-FE65-LRP complex can also be disrupted when proteins are expressed that can functionally replace APP in terms of complex formation with FE65 and LRP, leaving APP without its binding partners (fig. 4C). This happens when the two homologs of APP, APLP1 and APLP2, are expressed. A detailed mechanistic analysis revealed that APLP1 expression reduces APP endocytosis, strongly increases APP -secretase cleavage and reduces APP -secretase cleavage [11]. Moreover, the APLP1 effect on Lichtenthaler

APP shedding is only observed in LRP-expressing cells, but not in LRP knock-out cells, showing that this effect is LRP dependent. Additionally, mutational analysis revealed that the FE65-binding motif in APLP1 needs to be present in order to increase APP shedding. Importantly, proteins that do not bind FE65 did not affect APP shedding, consistent with the APP-FE65-LRP complex being required for APP endocytosis and shedding. Together, these experiments raise the possibility that changes in the expression levels of the APP homologs APLP1 and APLP2 may influence the shedding of APP. In fact, expression levels of APLP1, APLP2 and even of APP are differentially regulated upon physiological and pathophysiological stimuli, such as during embryonic development, neuronal migration and wound repair [30, 31]. These stimuli may in turn alter the amount of APP shedding. Future studies need to show, whether the complex only consists of the three proteins APP-FE65-LRP, or whether it is part of a multi-protein complex. Given that FE65 consists of several protein-protein interaction domains, it is likely that FE65 may link the complex to other proteins. In fact, FE65 colocalizes with APP in actin-rich lamellipodia in neuronal growth cones [32] and may link APP to cellular motility [33].

tate with APP [36, 37]. Again, the function of BRI2 is unknown, but mutant forms of BRI2 have been linked to dementia and cerebellar ataxia in Danish and British kindreds. The name BRI seems to be derived from the British origin [38]. Currently, it is unclear, whether the interaction between APP and both novel proteins, SorLA and BRI2, occurs directly or is mediated by adapter proteins, as it is the case for LRP. Interestingly, expression of both SorLa and BRI2 strongly inhibited APP -secretase cleavage and also A generation. Although the underlying mechanisms remain to be established in detail, both proteins seem to retain APP in early cellular compartments of the secretory pathway, where APP cannot reach the secretases. A surprising additional membrane protein interactor has recently been suggested: APP itself. Soba et al. [39] proposed that APP can dimerize in cis and in trans at the cell surface. It will be interesting to see in future studies, whether the dimerization status of APP can influence APP shedding, potentially by modulating the formation or the endocytosis of the APP-FE65-LRP complex.

Conclusion

Interestingly, a homolog of LRP, LRP1B, may form a similar complex with APP as LRP itself. A recent study showed that LRP1B can also be coimmunoprecipitated with APP [34]. It remains to be established whether this interaction is also mediated by FE65. In contrast to LRP, which is rapidly endocytosed and mediates efficient APP endocytosis, LRP1B is very slowly endocytosed. LRP1B reduced APP endocytosis and again increased APP shedding by -secretase [34]. Presumably, LRP1B forms a complex with APP at the expense of LRP, similar to APLP1, which forms the complex with LRP at the expense of APP. These experiments reinforce the notion that changes in the rate of APP endocytosis determine the amount of APP -secretase cleavage. Besides LRP and LRP1B, two novel transmembrane interactors of APP have recently been described. Sorting protein-related receptor (SorLa) is a type I transmembrane protein of unknown function, which is expressed in neurons and was shown to coimmunoprecipitate with APP [35]. The other protein is BRI2, which is the first type II membrane protein shown to coimmunoprecipi-

Recently, we and others have described several novel modulators of APP trafficking and processing. Although we are only beginning to understand the underlying mechanisms, it becomes more and more clear that not only signaling cascades, changes in membrane composition and interaction of APP with cytoplasmic adaptors (discussed above in the first paragraph) but also changes in APP trafficking, and specifically in the rate of APP endocytosis, can have a major effect on APP processing by - and -secretase. APP endocytosis can be altered by general modulators of endocytosis, such as dynamin, endophilin and Rab5, or by specifically targeting APP endocytosis, such as by altering the amount and the composition of the APP-FE65-LRP complex. Because -secretase cleavage occurs at or very close to the cell surface, whereas -secretase cleavage of wild-type APP mainly occurs in the endosomes, a reduction in APP endocytosis will favor -secretase cleavage, whereas an increase in APP endocytosis will enhance -secretase cleavage and A generation. Interestingly, one of the first pathological changes in AD brain are abnormalities in endosomal morphology [reviewed in 40]. Enlarged endosomal structures are observed long before the onset of the disease and are very similar to the changes seen in cultured cells with experimentally induced increases in endocytosis [21].

Cellular Control of APP Shedding

Neurodegenerative Dis 2006;3:262–269

Modulation of APP Shedding by the Transmembrane Proteins LRP1B and BRI2

267

Together, these findings indicate that alterations in the rate of APP endocytosis may increase A generation not only in cultured cells but also in vivo and may contribute to AD pathogenesis.

Acknowledgements I thank Stephanie Neumann and Susanne Schöbel for critical reading of the manuscript. I am grateful to the Deutsche Forschungsgemeinschaft for financial support through SPP1085/2 (Li862/4-1).

References 1 Blobel CP: ADAMs: key components in EGFR signalling and development. Nat Rev Mol Cell Biol 2005; 6:32–43. 2 Huovila AP, Turner AJ, Pelto-Huikko M, Karkkainen I, Ortiz RM: Shedding light on ADAM metalloproteinases. Trends Biochem Sci 2005;30:413–422. 3 Selkoe DJ, Schenk D: Alzheimer’s disease: molecular understanding predicts amyloidbased therapeutics. Annu Rev Pharmacol Toxicol 2003;43:545–584. 4 Citron M: beta-Secretase inhibition for the treatment of Alzheimer’s disease – promise and challenge. Trends Pharmacol Sci 2004; 25:92–97. 5 Haass C: Take five-BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 6 Allinson TM, Parkin ET, Turner AJ, Hooper NM: ADAMs family members as amyloid precursor protein alpha-secretases. J Neurosci Res 2003;74:342–352. 7 Skovronsky DM, Moore DB, Milla ME, Doms RW, Lee VM: Protein kinase C-dependent alpha-secretase competes with betasecretase for cleavage of amyloid-beta precursor protein in the trans-golgi network. J Biol Chem 2000; 275:2568–2575. 8 Vardy ER, Catto AJ, Hooper NM: Proteolytic mechanisms in amyloid-beta metabolism: therapeutic implications for Alzheimer’s disease. Trends Mol Med 2005;11:464–472. 9 Sisodia SS: Beta-amyloid precursor protein cleavage by a membrane-bound protease. Proc Natl Acad Sci USA 1992; 89: 6075– 6079. 10 Russo C, Venezia V, Repetto E, Nizzari M, Violani E, Carlo P, Schettini G: The amyloid precursor protein and its network of interacting proteins: physiological and pathological implications. Brain Res Brain Res Rev 2005;48:257–264. 11 Neumann S, Schobel S, Jager S, Trautwein A, Haass C, Pietrzik CU, Lichtenthaler SF: Amyloid precursor-like protein 1 influences endocytosis and proteolytic processing of the amyloid precursor protein. J Biol Chem 2006;281:7583–7594. 12 Schobel S, Neumann S, Seed B, Lichtenthaler SF: Expression cloning screen for modifiers of amyloid precursor protein shedding. Int J Dev Neurosci 2006;24:141–148.

268

13 Xu HX, Sweeney D, Greengard P, Gandy S: Metabolism of Alzheimer beta-amyloid precursor protein: Regulation by protein kinase A in intact cells and in a cell-free system. Proc Natl Acad Sci USA 1996; 93: 4081– 4084. 14 Marambaud P, Chevallier N, Ancolio K, Checler F: Post-transcriptional contribution of a cAMP-dependent pathway to the formation of alpha- and beta/gamma-secretasesderived products of beta APP maturation in human cells expressing wild-type and Swedish mutated beta APP. Mol Med 1998;4:715– 723. 15 Camden JM, Schrader AM, Camden RE, Gonzalez FA, Erb L, Seye CI, Weisman GA: P2Y2 nucleotide receptors enhance alphasecretase-dependent amyloid precursor protein processing. J Biol Chem 2005; 280: 18696–18702. 16 Lee RK, Wurtman RJ, Cox AJ, Nitsch RM: Amyloid precursor protein processing is stimulated by metabotropic glutamate receptors. Proc Natl Acad Sci USA 1995; 92: 8083–8087. 17 Reutens AT, Glenn Begley C: Endophilin-1: a multifunctional protein. Int J Biochem Cell Biol 2002;34:1173–1177. 18 Koo EH, Squazzo SL: Evidence that production and release of amyloid beta-protein involves the endocytic pathway. J Biol Chem 1994;269:17386–17389. 19 Chyung JH, Selkoe DJ: Inhibition of receptor mediated endocytosis demonstrates generation of amyloid beta-protein at the cell surface. J Biol Chem 2003; 278:51035–51043. 20 Carey RM, Balcz BA, Lopez-Coviella I, Slack BE: Inhibition of dynamin-dependent endocytosis increases shedding of the amyloid precursor protein ectodomain and reduces generation of amyloid beta protein. BMC Cell Biol 2005; 6:30. 21 Grbovic OM, Mathews PM, Jiang Y, Schmidt SD, Dinakar R, Summers-Terio NB, Ceresa BP, Nixon RA, Cataldo AM: Rab5-stimulated up-regulation of the endocytic pathway increases intracellular beta-cleaved amyloid precursor protein carboxyl-terminal fragment levels and Abeta production. J Biol Chem 2003;278:31261–31268. 22 King GD, Scott Turner R: Adaptor protein interactions: modulators of amyloid precursor protein metabolism and Alzheimer’s disease risk? Exp Neurol 2004;185:208–219.

Neurodegenerative Dis 2006;3:262–269

23 Pietrzik CU, Busse T, Merriam DE, Weggen S, Koo EH: The cytoplasmic domain of the LDL receptor-related protein regulates multiple steps in APP processing. EMBO J 2002; 21:5691–5700. 24 Pietrzik CU, Yoon IS, Jaeger S, Busse T, Weggen S, Koo EH: FE65 constitutes the functional link between the low-density lipoprotein receptor-related protein and the amyloid precursor protein. J Neurosci 2004; 24: 4259–4265. 25 Kinoshita A, Whelan CM, Smith CJ, Mikhailenko I, Rebeck GW, Strickland DK, Hyman BT: Demonstration by fluorescence resonance energy transfer of two sites of interaction between the low-density lipoprotein receptor-related protein and the amyloid precursor protein: role of the intracellular adapter protein Fe65. J Neurosci 2001; 21:8354–8361. 26 Trommsdorff M, Borg JP, Margolis B, Herz J: Interaction of cytosolic adaptor proteins with neuronal apolipoprotein E receptors and the amyloid precursor protein. J Biol Chem 1998;273:33556–33560. 27 Rebeck GW, Moir RD, Mui S, Strickland DK, Tanzi RE, Hyman BT: Association of membrane-bound amyloid precursor protein APP with the apolipoprotein E receptor LRP. Brain Res Mol Brain Res 2001;87:238–245. 28 Herz J, Strickland DK: LRP: a multifunctional scavenger and signaling receptor. J Clin Invest 2001;108:779–784. 29 Ulery PG, Beers J, Mikhailenko I, Tanzi RE, Rebeck GW, Hyman BT, Strickland DK: Modulation of beta-amyloid precursor protein processing by the low density lipoprotein receptor-related protein (LRP). Evidence that LRP contributes to the pathogenesis of Alzheimer’s disease. J Biol Chem 2000;275:7410–7415. 30 Kummer C, Wehner S, Quast T, Werner S, Herzog V: Expression and potential function of beta-amyloid precursor proteins during cutaneous wound repair. Exp Cell Res 2002; 280:222–232. 31 Beckman M, Iverfeldt K: Increased gene expression of beta-amyloid precursor protein and its homologues APLP1 and APLP2 in human neuroblastoma cells in response to retinoic acid. Neurosci Lett 1997;221:73–76.

Lichtenthaler

32 Sabo SL, Ikin AF, Buxbaum JD, Greengard P: The amyloid precursor protein and its regulatory protein, FE65, in growth cones and synapses in vitro and in vivo. J Neurosci 2003;23:5407–5415. 33 Sabo SL, Ikin AF, Buxbaum JD, Greengard P: The Alzheimer amyloid precursor protein (APP) and FE65, an APP-binding protein, regulate cell movement. J Cell Biol 2001;153: 1403–1414. 34 Cam JA, Zerbinatti CV, Knisely JM, Hecimovic S, Li Y, Bu G: The low density lipoprotein receptor-related protein 1B retains betaamyloid precursor protein at the cell surface and reduces amyloid-beta peptide production. J Biol Chem 2004; 279:29639–29646.

Cellular Control of APP Shedding

35 Andersen OM, Reiche J, Schmidt V, Gotthardt M, Spoelgen R, Behlke J, von Arnim CA, Breiderhoff T, Jansen P, Wu X, Bales KR, Cappai R, Masters CL, Gliemann J, Mufson EJ, Hyman BT, Paul SM, Nykjaer A, Willnow TE: Neuronal sorting protein-related receptor sorLA/LR11 regulates processing of the amyloid precursor protein. Proc Natl Acad Sci USA 2005;102:13461–13466. 36 Fotinopoulou A, Tsachaki M, Vlavaki M, Poulopoulos A, Rostagno A, Frangione B, Ghiso J, Efthimiopoulos S: BRI2 interacts with amyloid precursor protein (APP) and regulates amyloid beta (Abeta) production. J Biol Chem 2005; 280:30768–30772. 37 Matsuda S, Giliberto L, Matsuda Y, Davies P, McGowan E, Pickford F, Ghiso J, Frangione B, D’Adamio L: The familial dementia BRI2 gene binds the Alzheimer gene amyloid-beta precursor protein and inhibits amyloid-beta production. J Biol Chem 2005; 280: 28912– 28916.

38 Vidal R, Frangione B, Rostagno A, Mead S, Revesz T, Plant G, Ghiso J: A stop-codon mutation in the BRI gene associated with familial British dementia. Nature 1999; 399: 776– 781. 39 Soba P, Eggert S, Wagner K, Zentgraf H, Siehl K, Kreger S, Lower A, Langer A, Merdes G, Paro R, Masters CL, Muller U, Kins S, Beyreuther K: Homo- and heterodimerization of APP family members promotes intercellular adhesion. EMBO J 2005; 24: 3624– 3634. 40 Nixon RA: Endosome function and dysfunction in Alzheimer’s disease and other neurodegenerative diseases. Neurobiol Aging 2005;26:373–382.

Neurodegenerative Dis 2006;3:262–269

269

Original Paper

Diseases

Neurodegenerative Dis 2006;3:270–274 DOI: 10.1159/000095266

Amyloid Precursor Protein and BACE Function as Oligomers Gerd Multhaup Institute of Chemistry and Biochemistry, Free University of Berlin, Berlin, Germany

Key Words Amyloid precursor protein Homodimerization Oligomerization -Secretase Amyloid A

Abstract Processing of the amyloid precursor protein (APP) by - and -secretases leads to the generation of amyloid- (A) peptides, which are the toxic agents in the pathogenesis of Alzheimer’s disease. The molecular reasons for the sequential A generation by secretase activities have remained unclear. Our studies support an oligomerization-dependent mechanism for the conversion of APP into A. By different lines of evidence, we showed that APP is capable of forming homodimers and tetramers. Oligomerization of APP occurs in a zipper-like mechanism primarily mediated by two highly conserved sites of the ectodomain. We also found that in human brain tissue -secretase (BACE) occurred as a dimer, whereas the soluble ectodomain of truncated BACE exclusively occurred in the monomeric form. A mutational analysis of the active sites supports the idea that BACE might have acquired a specific catalytic activity by oligomerization, which is stabilized through the transmembrane and the cytoplasmic domains. Our results predict that APP homodimers are functionally active within the plasma membrane

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0270$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

and most likely represent substrates for BACE oligomers. Understanding the molecular tasks of homophilic binding of substrates and secretases will allow to find secretase inhibitors which specifically bind to contact sites of dimers and thus inhibit A formation. Copyright © 2006 S. Karger AG, Basel

Although much progress has been made over the past years regarding the involvement of the amyloid precursor protein (APP) in Alzheimer’s disease, the biological function of cellular APP still remains enigmatic. APP is part of a superfamily from which 16 homologous amyloid precursor-like proteins (APLPs) and APP species homologues are derived [1]. APP is found on the cell surface of neurons possessing a neurite outgrowth activity [2]. Its intracellular domain interacts with the cytoskeleton to recruit and activate signaling molecules [3, 4]. Soluble forms of APP also promote neurite outgrowth, cell adhesion, and are neuroprotective [5]. Thus, extra- and intracellular domains of APP are involved in similar biological functions. In initial studies, we identified a conserved motif for collagen binding and APP-APP interactions in the carbohydrate domain of APP [6], which is also called the E2- or

Gerd Multhaup Free University of Berlin, Institute of Chemistry and Biochemistry Thielallee 63 DE–14195 Berlin (Germany) Tel. +49 30 838 555 33, Fax +49 30 838 565 09, E-Mail [email protected]

Fig. 1. Domain structure of membrane-anchored APP-751. HBD-I and II are the hepa-

rin-binding domains at the N-terminus (low affinity binding site) and the C-terminus (high affinity binding site); CuBD is the copper-binding domain, KPI the Kunitz-type protease inhibitor domain and CBD the collagen-binding domain. The A region is part of the ectodomain and the transmembrane sequence (TM). E1, E2 and CT indicate conserved regions of the APP protein family. HBD-II and CBD are within the central APP domain (CAPPD). HDS indicates homodimerization sites 1, 2 and 3.

the central APP domain (CAPPD) [7]. A recent structural analysis of the E2 domain revealed that this domain generates a better groove for its heparin-binding site [8] upon dimerization when both monomers contribute to the positively charged surface patch [7]. Thus, a dynamic alteration of the APP multimerization state could modulate cellcell and cell-matrix interactions by competing with homophilic binding of APP (see below). These sites overlap with the heparin-binding site of the E1 domain and the collagen binding site within the E2 domain (fig. 1). Whereas conflicting results on dimerization of APP domains in solution are described for recombinant APP purified from bacterial systems [7, 9], the glycosylated ectodomain of APP (residues 18–350) secreted by yeast cells preferentially dimerizes in solution [10]. By three lines of evidence (dynamic light scattering, cross-linking and size exclusion chromatography), we demonstrated that APP is capable of forming noncovalent homodimers and tetramers [10]. Moreover, the analysis of posttranslationally modified wild-type and disulfide cross-linked mutant APP from SY5Y cells revealed that APP dimers appear to assemble early in the ER either during or shortly after the synthesis, suggesting that the homodimeric state could be an essential prerequisite for its sorting to the trans-Golgi network and secretory vesicles [10]. Experimental observations with differently tagged APP constructs provided evidence that homodimerization of APP occurs in a zipper-like manner from the Nterminus to the C-terminus, primarily mediated by two highly conserved homodimerization sites (HDS1 and HDS2; fig. 1). The disulfide bridge between Cys98 and Cys105 stabilizes a -hairpin loop with several basic residues that contribute to a positively charged surface [11]. Likewise, they represent the heparin-binding site within

the E1 domain. Synthetic peptides representing APP residues 91–111 with a loop formed by the disulfide bridge between Cys98 and Cys105 were able to inhibit homodimerization of soluble APP [Multhaup, unpubl. obs.; 10]. APP residues 448–465 of the collagen-binding site within the E2 domain were earlier shown by us to be of critical importance for the regulation of homodimerization of holo-APP [6]. According to a hypothesis that we have published 5 years ago [10], dimerization of APP occurs cotranslationally during folding through the two strong separate effector sites HDS1 and HDS2 (fig. 1) and induces an extended third contact site HDS3 between juxtamembrane and intramembrane -helices, thus enabling a spatial proximity of the amyloid- (A) region between two APP molecules [10]. Very recently, we were able to prove this hypothesis when we analyzed the dimerization of the APP transmembrane sequence mediated by glycine residues G29 and G33 of three hitherto unrecognized GxxxG motifs within the APP transmembrane sequence [12, 13; Munter et al., submitted]. We used fluorescence resonance energy transfer (FRET) to examine whether APP dimerization also depends on the GxxxG motif in living cells. FRET efficiencies of wild-type APP revealed that the high affinity contact sites assured APP homodimerization in living cells and that the initial homodimerization of holo-APP remained uninfluenced by the GxxxG motif. This motif promoted TM helix self-interactions of the C-terminal 100 APP residues (CTF) as analyzed in a bacterial test system. Most interestingly, G29 and G33 mutants (A numbering) were unique in their ability to selectively decrease A42 levels. This revealed that A42 production requires dimerization within the membrane through the GxxxG motif, which represents a third con-

APP and BACE

Neurodegenerative Dis 2006;3:270–274

271

tact site of APP (fig. 1). As soon as A is released from APP, the third site is part of a -strand segment that spans A residues 30–35 and is involved in the transformation of the 42-residue A peptide into the amyloid fibrils of Alzheimer’s disease. This process is likely controlled by the GxxxG motif and occurs independently from the helical interaction of transmembrane sequences of two APP molecules [Munter et al., submitted]. In earlier studies, we could show that a biochemically defined assembly of A into A dimers probably represents the initial step in amyloidogenesis [14]. Spectroscopic analyses even revealed that engineered dimeric peptides ending at residue 42 displayed a much more pronounced -structural transition than corresponding monomers rendering it prone to aggregation [14]. Taken together, the APP ectodomain homodimerizes by two different sites (HDS1 and HDS2) and possesses a third contact site (HDS3) within the A region, which determines A42 cleavage that essentially requires dimerization within the membrane through the GxxxG motif [Munter et al., submitted] (fig. 1). In turn, the A homodimer that is possibly released from APP dimers (see below) represents a nidus for plaque formation [14]. The aspartyl protease BACE is a type I transmembrane protease (fig. 2), which is the major -secretase [15]. BACE cleaves APP to release the N-terminal ectodomain referred to as sAPP. After BACE cleavage of APP, the remaining membrane-bound C-terminal APP fragment becomes a substrate for -secretase to produce A40/A42. In addition to APP, other substrates have been described for BACE, such as the APLPs, the voltagegated sodium channel 2 subunit, the low-density lipoprotein receptor-related protein, the sialyltransferase ST6Gal I and the adhesion protein P-selectin glycoprotein ligand-1. We and others have found that A itself complements the substrate profile of BACE [16, 17]. Certainly, a large number of other type I TM proteins known already as -secretase substrates will be discovered as further BACE substrates [15]. Less is known about the subcellular localization of APP processing and substrate-enzyme recognition mechanisms of APP, BACE and the -secretase complex. Naively, increased BACE activity is expected to increase A pathology. Whereas modest BACE overexpression in mice increases steady-state A levels, high BACE expression paradoxically decreased A deposition despite enhanced -cleavage of APP [18]. In addition, BACE overexpression increases -cleavage early in the secretory pathway. Much of these unexpected findings underscore the necessity of a thorough understanding of the molecu272

Neurodegenerative Dis 2006;3:270–274

Fig. 2. Domain structure of soluble membrane-anchored BACE.

SP is the signal peptide, PRO the pro-peptide and D-93 and D-289 represent the active site aspartyl residues. HDS-1 and HDS-2 are the proposed homodimerization sites.

lar architecture and function of BACE as a membraneanchored aspartic acid protease. When we analyzed BACE from human brain tissue, we observed SDS-stable dimers with a polyclonal antibody against the C-terminus of BACE [19]. A detailed analysis revealed that endogenously and ectopically expressed BACE forms oligomers and that membrane attachment was required for dimerization [19, 20]. Since the sequence of the BACE transmembrane domain alone did not affect BACE dimerization, we propose that contact sites within the ectodomain first induce the association of BACE with itself, similar to the mechanism elaborated for dimerization of APP through its ectodomain (see above). The only other known aspartic acid protease, which exists as a homodimer is cathepsin D, which possesses two fully catalytically active monomers [21]. Homodimerization of cathepsin D leads to a slightly changed enzymatic activity that is evident from an increased pH and temperature stability. Likewise, membrane-retained BACE displayed a higher affinity and turnover rate with a synthetic Swedish APP-like substrate compared with soluble BACE lacking the transmembrane and the cytoplasmic domains [20]. Our finding of a tight binding of dimeric BACE and a loose binding of soluble BACE suggested that two N-terminal DTGS active site motifs (fig. 2) could be used instead of the carboxy-terminal active site motif DSGT within the polypeptide chain. Together with residual BACE activity observed in BACED289A mutant transfected cells [19], this supports the idea that BACE dimers might have acquired a specific catalytic activity by oligomerization through the transmembrane and the cytoplasmic domain (fig. 2), which is different from the activity of soluble C-terminally truncated BACE [19]. Thus, the catalytic activity of BACE may be regulated by oligomerization, which is most likely started as early as during the synthesis in the ER and then stabilized Multhaup

by the transmembrane and the cytoplasmic domain of BACE. Membrane-bound BACE localizes to rafts and a small proportion of APP is also present in these domains [22]. APP molecules mainly reside as homodimers in the plasma membrane of living cells [Munter et al., in preparation]. By using a mutationally induced dimerization of different Cys-mutant APP constructs forming disulfide bridges, we could already show that dimeric APP might be an ideal substrate for BACE [Schmechel et al., unpubl. results]. In addition to the homodimerization of BACE, its trafficking through the Golgi, to endosomes and lysosomes and its targeting to rafts represent key mechanisms by which BACE cleavage of APP might be regulated. Posttranslational modifications of the cytoplasmic domain, such as phosphorylation of Ser498 or palmitoylation of cysteines 478, 482 and 485 are suspected to regulate BACE trafficking to late endosomes, the Golgi and to lipid rafts. Alternatively, these cysteine residues coordinate Cu(I), which is highly likely transferred to BACE by the copper chaperone CCS. CCS itself binds to the cytopolasmic domain of BACE [23]. The formation of BACE-Cu complexes may influence trafficking and thus the ability of BACE to process APP [23]. Moreover, we have recently discovered that BACE has at least two binding sites for Cu(II) within the ectodomain [Bethge et al., in preparation]. APP itself possesses a copper-binding domain (CuBD), with histidine residues 147 and 151 being the main residues to coordinate Cu(II) (fig. 1). First, Cu(II) is bound by APP in a planar coordination sphere and upon reduction of Cu(II) to Cu(I) it adopts a tetrahedral arrangement [24, 25]. The CuBD is surface exposed and the close prox-

imity of the CuBD to the HDS1 may thus allow the regulation of homodimerization of APP and/or modulate the substrate-enzyme recognition between APP and BACE. In summary, our results conclusively show that APP homodimers are functionally active within the plasma membrane and most likely represent appropriate substrates of BACE oligomers. Thus, at least four polypeptide chains, i.e., two APP and two BACE molecules, may act together in a stable complex, although the presence of homodimers has not entirely been proven during this initial processing step in the A generating pathway. Many questions regarding the substrate specificity, multiple cleavage sites and substrate-binding sites remain to be clarified. Studying the binding of substrates in secretase complexes is a crucial step in understanding the molecular mechanism of A formation. The proposed active component of the -secretase complex, presenilin (PS), was shown to homodimerize as a PS dimer in the catalytic core of the -secretase with PS transmembrane domains contributing to the interaction [26]. Other questions that remain to be answered include how different complexes of BACE and PS preferentially cleave certain substrates, and most importantly how sequential processing of APP by -secretase and -secretase proteolytic activities is regulated. It appears that different requirements have to be met for cleavages of individual substrates to occur. Further work is needed to better understand the molecular tasks of homophilic binding of substrates and secretases. This may lead to the discovery of specific secretase inhibitors that will specifically attack the homophilic interactions and inhibit A formation.

References 1 Bayer TA, Cappai R, Masters CL, Beyreuther K, Multhaup G: It all sticks together – the APP-related family of proteins and Alzheimer’s disease. Mol Psychiatry 1999; 4: 524– 528. 2 Qiu WQ, Ferreira A, Miller C, Koo EH, Selkoe DJ: Cell-surface beta-amyloid precursor protein stimulates neurite outgrowth of hippocampal neurons in an isoform-dependent manner. J Neurosci 1995;15:2157–2167. 3 Cao X, Sudhof TC: Dissection of amyloidbeta precursor protein-dependent transcriptional transactivation. J Biol Chem 2004; 279:24601–24611.

APP and BACE

4 Hass MR, Yankner BA: A -secretase-independent mechanism of signal transduction by the amyloid precursor protein. J Biol Chem 2005;280:36895–36904. 5 Mattson MP: Cellular actions of beta-amyloid precursor protein and its soluble and fibrillogenic derivatives. Physiol Rev 1997;77: 1081–1132. 6 Beher D, Hesse L, Masters CL, Multhaup G: Regulation of amyloid protein precursor (APP) binding to collagen and mapping of the binding sites on APP and collagen type I. J Biol Chem 1996; 271:1613–1620. 7 Wang Y, Ha Y: The X-ray structure of an antiparallel dimer of the human amyloid precursor protein E2 domain. Mol Cell 2004;15: 343–353.

8 Multhaup G, Bush AI, Pollwein P, Masters CL: Interaction between the zinc (II) and the heparin binding site of the Alzheimer’s disease beta A4 amyloid precursor protein (APP). FEBS Lett 1994;355:151–154. 9 Dulubova I, Ho A, Huryeva I, Sudhof TC, Rizo J: Three-dimensional structure of an independently folded extracellular domain of human amyloid-beta precursor protein. Biochemistry 2004;43:9583–9588. 10 Scheuermann S, Hambsch B, Hesse L, Stumm J, Schmidt C, Beher D, Bayer TA, Beyreuther K, Multhaup G: Homodimerization of amyloid precursor protein and its implication in the amyloidogenic pathway of Alzheimer’s disease. J Biol Chem 2001; 276: 33923–33929.

Neurodegenerative Dis 2006;3:270–274

273

11 Rossjohn J, Cappai R, Feil SC, Henry A, McKinstry WJ, Galatis D, Hesse L, Multhaup G, Beyreuther K, Masters CL, Parker MW: Crystal structure of the N-terminal, growth factor-like domain of Alzheimer amyloid precursor protein. Nat Struct Biol 1999; 6: 327–331. 12 Brosig B, Langosch D: The dimerization motif of the glycophorin A transmembrane segment in membranes: importance of glycine residues. Protein Sci 1998;7:1052–1056. 13 Russ WP, Engelman DM: The GxxxG motif: a framework for transmembrane helix-helix association. J Mol Biol 2000;296:911–919. 14 Schmechel A, Zentgraf H, Scheuermann S, Fritz G, Pipkorn R, Reed J, Beyreuther K, Bayer TA, Multhaup G: Alzheimer beta-amyloid homodimers facilitate A beta fibrillization and the generation of conformational antibodies. J Biol Chem 2003; 278: 35317– 35324. 15 De Strooper B: Aph-1, Pen-2, and nicastrin with presenilin generate an active gammasecretase complex. Neuron 2003;38:9–12. 16 Fluhrer R, Multhaup G, Schlicksupp A, Okochi M, Takeda M, Lammich S, Willem M, Westmeyer G, Bode W, Walter J, Haass C: Identification of a beta-secretase activity, which truncates amyloid beta-peptide after its presenilin-dependent generation. J Biol Chem 2003;278:5531–5538.

274

17 Shi XP, Tugusheva K, Bruce JE, Lucka A, Wu GX, Chen-Dodson E, Price E, Li Y, Xu M, Huang Q, Sardana MK, Hazuda DJ: Betasecretase cleavage at amino acid residue 34 in the amyloid beta peptide is dependent upon gamma-secretase activity. J Biol Chem 2003; 278:21286–21294. 18 Lee EB, Zhang B, Liu K, Greenbaum EA, Doms RW, Trojanowski JQ, Lee VM: BACE overexpression alters the subcellular processing of APP and inhibits Abeta deposition in vivo. J Cell Biol 2005; 168:291–302. 19 Schmechel A, Strauss M, Schlicksupp A, Pipkorn R, Haass C, Bayer TA, Multhaup G: Human BACE forms dimers and colocalizes with APP. J Biol Chem 2004;279:39710–39717. 20 Westmeyer GG, Willem M, Lichtenthaler SF, Lurman G, Multhaup G, Assfalg-Machleidt I, Reiss K, Saftig P, Haass C: Dimerization of beta-site beta-amyloid precursor proteincleaving enzyme. J Biol Chem 2004; 279: 53205–53212. 21 Tsukuba T, Masayuki Y, Hori H, Azuma T, Yamamoto K: Comparison of biochemical properties of natural and recombinant cathepsin E. Adv Exp Med Biol 1995; 362: 331– 334.

Neurodegenerative Dis 2006;3:270–274

22 Ehehalt R, Keller P, Haass C, Thiele C, Simons K: Amyloidogenic processing of the Alzheimer beta-amyloid precursor protein depends on lipid rafts. J Cell Biol 2003; 160: 113–123. 23 Angeletti B, Waldron KJ, Freeman KB, Bawagan H, Hussain I, Miller CC, Lau KF, Tennant ME, Dennison C, Robinson NJ, Dingwall C: BACE1 cytoplasmic domain interacts with the copper chaperone for superoxide dismutase-1 and binds copper. J Biol Chem 2005;280:17930–17937. 24 Barnham KJ, McKinstry WJ, Multhaup G, Galatis D, Morton CJ, Curtain CC, Williamson NA, White AR, Hinds MG, Norton RS, Beyreuther K, Masters CL, Parker MW, Cappai R: Structure of the Alzheimer’s disease amyloid precursor protein copper binding domain. A regulator of neuronal copper homeostasis. J Biol Chem 2003; 278: 17401– 17407. 25 Multhaup G, Schlicksupp A, Hesse L, Beher D, Ruppert T, Masters CL, Beyreuther K: The amyloid precursor protein of Alzheimer’s disease in the reduction of copper(II) to copper(I). Science 1996;271:1406–1409. 26 Cervantes S, Saura CA, Pomares E, Gonzalez-Duarte R, Marfany G: Functional implications of the presenilin dimerization: reconstitution of gamma-secretase activity by assembly of a catalytic site at the dimer interface of two catalytically inactive presenilins. J Biol Chem 2004; 279:36519–36529.

Multhaup

Original Paper

Diseases

Neurodegenerative Dis 2006;3:275–283 DOI: 10.1159/000095267

Assembly, Trafficking and Function of -Secretase Christoph Kaether Christian Haass Harald Steiner1 Laboratory for Alzheimer’s and Parkinson’s Disease Research, Department of Biochemistry, Adolf Butenandt Institute, Ludwig Maximilians University, Munich, Germany

Key Words Alzheimer’s disease Amyloid- peptide -Secretase Presenilin

Abstract -Secretase catalyzes the final cleavage of the -amyloid precursor protein to generate amyloid- peptide, the principal component of amyloid plaques in the brains of patients suffering from Alzheimer’s disease. Here, we review the identification of -secretase as a protease complex and its assembly and trafficking to its site(s) of cellular function. In reconstitution experiments, -secretase was found to be composed of four integral membrane proteins, presenilin (PS), nicastrin (NCT), PEN-2 and APH-1 that are essential and sufficient for -secretase activity. PS, which serves as a catalytic subunit of -secretase, was identified as a prototypic member of novel aspartyl proteases of the GxGD type. In human cells, -secretase could be further defined as a heterogeneous activity consisting of distinct complexes that are composed of PS1 or PS2 and APH-1a or APH-1b homologues together with NCT and PEN-2. Using green fluorescent protein as a reporter we localized PS and -secretase activity at the plasma membrane and endosomes. Investigation of secretase complex assembly in knockdown and knockout cells of the individual subunits allowed us to develop a mod-

1

Co-corresponding author: [email protected]

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0275$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

el of complex assembly in which NCT and APH-1 first stabilize PS before PEN-2 assembles as the last component. Furthermore, we could map domains in PS and PEN-2 that govern assembly and trafficking of the complex. Finally, Rer1 was identified as a PEN-2-binding protein that serves a role as an auxiliary factor for -secretase complex assembly. Copyright © 2006 S. Karger AG, Basel

Introduction

Alzheimer’s disease (AD) is the most common neurodegenerative disorder in industrial countries. Neuropathological hallmarks of AD are the deposition of the 40–42 amino acid amyloid- peptide (A40, A42) as senile plaques and of hyperphosphorylated tau as neurofibrillary tangles. In their accompanying articles, Mandelkow and Mandelkow review in detail the cellular mechanisms causing neurofibrillary tangle formation and its impact on AD. A42 is highly neurotoxic and believed to be the culprit of the disease, which initiates a cascade of pathological events that ultimately lead to dementia [1]. A is generated by proteolytic processing of the -amyloid precursor protein (APP), a type I transmembrane protein [2]. APP is first processed by -secretase, which leaves a 99 amino acid C-terminal fragment (C99) in the membrane. A is subsequently liberated into the extracellular space by -secretase cleavage of C99 within the membrane. This cleavage also releases the

Dr. Christoph Kaether Group ‘Membrane trafficking of proteins involved in Alzheimer’s Disease’ Leibniz Institute for Age Research – Fritz Lipmann Institute Beutenbergstrasse 11, DE–07745 Jena (Germany) Tel. +49 3641 65 6230, Fax +49 3641 65 6040, E-Mail [email protected]

APP intracellular domain (AICD) into the cytosol. A competing nonamyloidogenic processing pathway of APP involves cleavage by -secretase within the A domain, generating an 83 amino acid membrane bound Cterminal fragment (C83). Subsequent cleavage of C83 by -secretase causes the liberation of the nonamyloidogenic peptide p3. -Secretase cleavage thus precludes the formation of A. Fahrenholz and Postina [this issue, pp. 255–261] and Lichtenthaler [this issue, pp. 262–269] describe this pathway in detail. A central question of AD research of nearly two decades has been the elucidation of the identity of the secretases which generate A. In particular -secretase, which is the focus of this review, has been an enigmatic enzyme for a long time [3]. Identification of -Secretase

Ten years ago, a handful of mutations causing earlyonset familial AD (FAD) were discovered and mapped to two novel genes located on chromosomes 1 and 14 [4–7]. The responsible genes, termed presenilin 1 (PS1) and presenilin 2 (PS2) were predicted to encode two homologous 50-kDa polytopic membrane proteins, that according to recent models consist of 9 transmembrane domains (TMDs) [8–10]. Both are endoproteolytically cleaved between TMDs 6 and 7 into an N-terminal and C-terminal fragment (NTF, CTF) [11]. Although PSs lacked obvious functional homology to other proteins when they were first described, the identification of Caenorhabditis elegans SEL-12 as PS homologue [12] strongly implicated PSs in the Notch signaling pathway required for cell differentiation [see also the review article by Smialowska and Baumeister in this issue, pp. 227–232]. Like FADassociated mutations found earlier in the C-terminal end of the APP TMD, the FAD-associated mutations in PS1 and PS2 increased the levels of A42 [13]. The number of mutations found increased rapidly to more than 150 in 2005, including 144 PS1 mutations. PS1 is therefore the major gene responsible for early-onset FAD. The finding that FAD-associated mutants shifted the cleavage specificity of -secretase towards A42 formation suggested that PSs might be modulators of -secretase causing a gain of function. On the other hand, knockout of the mouse PS1 gene caused a severe reduction of total A generation [14]. Furthermore, cells derived from PS1 knockout mice showed an accumulation of the C-terminal fragments of APP, while leaving - and -secretase cleavage unchanged. Taken together, these data indicated that PS is intimately associated with -secretase activity 276

Neurodegenerative Dis 2006;3:275–283

and hinted to the possibility that PSs were either -secretase itself or alternatively an essential co-factor of it. Considerable evidence for the first hypothesis that PS is identical with -secretase was obtained when mutagenesis of either of two aspartate residues of PS1 with an unusual location in TMD6 and 7 caused the same -secretase loss of function phenotype as found for the PS1 gene deletion [15]. Moreover, when these aspartates were mutated, PS accumulated as uncleaved full-length holoprotein, suggesting the possibility that PS autoactivates itself. These findings were consistent with the earlier observation that -secretase is an aspartyl protease [16]. Further support for the hypothesis that PS is identical with secretase was obtained when we observed loss of -secretase function coupled with a deficiency of PS endoproteolysis upon mutation of the corresponding aspartate residues in PS2 and zebrafish PS1, thus demonstrating that the critical PS aspartates were functionally conserved during evolution [17, 18]. Another piece of evidence was the finding that crosslinkable -secretase inhibitors designed to mimic the transition-state of an aspartyl protease mechanism were found to covalently bind the PS NTF and CTF [19, 20]. While these observations were consistent with PS being a candidate aspartyl protease identical with -secretase, the lack of any homology to other aspartyl proteases and the lack of the canonical D(T/S) G(T/S) active motif of aspartyl proteases remained in apparent contrast to the hypothesis. This puzzling issue was resolved when we identified G384 in PS1 as an essential residue of PS function [21]. This residue is part of a highly conserved small GxGD motif that includes the critical aspartate in TMD7 of PS. Moreover, we identified this motif in the bacterial type 4 prepilin peptidase (TFPP) family, polytopic proteases that function as leader peptidases. Similar to PSs, the TFPPs contained two critical aspartate residues, directly adjacent to the TMD boundaries, required for their proteolytic function [22]. The latest piece of evidence for a proteolytic function of PS was the subsequent identification of signal peptide peptidase (SPP) and its homologues, the SPP-like (SPPL) proteases [23]. Like TFPPs and PSs, SPP and SPPLs contain the GxGD active site motif [24]. In addition, all three families contain a short conserved PxL motif at the C-terminus. Despite these conserved regions, no further homologies are found. Interestingly, PS and SPP differ in their orientation of the active sites towards the substrate. While PS cleaves substrates like APP in type I orientation, SPP [23] and SPPLs [Fluhrer et al., submitted] use type II membrane proteins as substrates. Taken together, PS was identified as a founding member of novel polytopic aspartyl Kaether/Haass/Steiner

proteases of the GxGD type [25]. For a detailed description of SPP and SPPLs, we refer the reader to the accompanying review by Haffner and Haass. While these findings strongly suggested that PS might indeed be identical with the long-sought -secretase and provided compelling evidence that PS has to be regarded as a novel aspartyl protease, other findings indicated that it might not fulfill the -secretase function alone. In fact, data by others and us suggested that PS resides in a high molecular weight (HMW) complex [26–28], and indeed it was subsequently shown that -secretase activity was present in an HMW complex as PS-dependent activity [29]. Moreover, overexpression of PS neither led to an increase in the NTF and CTF [30] nor to increased -secretase activity. This finding suggested that PS expression is regulated by the presence of other (limiting) factors, which together with PS assemble into an HMW complex [30] allowing PS endoproteolysis. Consistent with this observation, we found that excess PS holoprotein that fails to become processed into its stable fragments is rapidly degraded by the proteasome [31]. Using an immunoaffinity isolation procedure, the type I membrane glycoprotein nicastrin (NCT) was the first PS-binding partner identified [32]. In addition, screening for Notch pathway components in C. elegans identified two novel candidate PS partner proteins besides NCT, the polytopic membrane proteins PEN-2 and APH-1 [33, 34]. Coimmunoprecipitation studies revealed that PEN-2 and APH-1 are indeed in association with PS and NCT [35, 36]. Strikingly, when we coexpressed these four proteins in baker’s yeast, which does not contain homologues of these proteins and has no endogenous -secretase activity, -secretase activity towards an APP-based substrate was fully reconstituted [37]. -Secretase activity required the coexpression of all four components and was not observed when either one of the four components was lacking. Moreover, reconstitution of -secretase activity was associated with PS endoproteolysis and was found to be dependent on biologically active PS. These experiments demonstrated that -secretase is a complex of four core components that are necessary and sufficient for the activity of the -secretase enzyme. Similar results were obtained when the four components were overexpressed in mammalian cells [38–40]. -Secretase activity was significantly enhanced when all four proteins were coexpressed, suggesting a reconstitution of the enzyme. These findings also demonstrated that NCT, APH-1 and PEN-2 were the elusive limiting factors for PS expression. Like for PSs, two homologues of APH-1 were identified in mammalian cells, APH-1a and APH-1b, with

APH-1a occurring in two splice variants differing in their C-termini [33, 36]. In a coimmunoprecipitation analysis, we found that PS1 and PS2 and the APH-1 homologues/ splice variants are contained in separate -secretase complexes in human cells [35, 41]. Thus, the term -secretase reflects a heterogeneous activity in human cells that consists of several distinct complexes depending on the respective tissue expression of the core components. Similar findings were obtained in rodents [42].

Assembly, Trafficking and Function of -Secretase

Neurodegenerative Dis 2006;3:275–283

Localization of -Secretase and Its Cellular Site(s) of Activity

Shortly after the initial identification of PS1, its intracellular localization was determined to be mostly in the endoplasmic reticulum (ER) [43]. In contrast, presumed sites of -secretase activity ranged from the ER to the Golgi, TGN, secretory vesicles, plasma membrane (PM) and endosomes/lysosomes [for a discussion, see 44]. Subsequent analysis in neurons and other cell types led to the proposal of the ‘spatial paradox’, a term coined by Annaert and De Strooper [45], that referred to the apparent discrepancy in the localization of PS in the ER and the sites of -secretase activity proposed to be in later compartments of the secretory pathway. To clarify the localization of PS, we used green fluorescent protein (GFP) as a reporter for live cell staining. GFP-tagged PS1 (PS1-GFP) was shown to be fully functional in all aspects tested. PS1-GFP replaced endogenous PS1/2, was incorporated into an HMW complex and rescued -secretase activity in PS1/2–/– cells. Using total internal reflection microscopy and cell surface biotinylation we could show that small but significant amounts of PS1-GFP were localized at the PM [46]. Moreover, PS1 bound to NCT could be detected at the PM, showing that it is indeed complex-associated PS, which is at the PM [46]. In living cells expressing PS1-CFP and NCT-YFP, both subunits colocalize at the PM and in endosomes/lysosomes, supporting the idea that fully assembled complexes leave the ER and reach later compartments of the secretory pathway (fig. 1). Others subsequently confirmed these results by the demonstration that all four -secretase complex components are localized in an active form at the PM [47] and in lysosomes [48]. To determine the localization of -secretase activity with a novel approach, we again made use of GFP. A secretase substrate, the APP C-terminal stub (C99), was tagged C-terminally with GFP (C99-GFP) [49]. When stably expressed in cells, this substrate is efficiently cleaved 277

A

PS1 CFP

Nct YFP

merge

PS1 CFP

Nct YFP

Lyso

B

1

2

278

Neurodegenerative Dis 2006;3:275–283

Kaether/Haass/Steiner

by -secretase, resulting in a diffuse cytoplasmic GFP staining. When -secretase is blocked using specific inhibitors, C99-GFP is no longer cleaved and remains membrane associated, highlighting the compartment where it resides. To determine the sites of -secretase activity, biosynthetic transport was inhibited at defined steps along the secretory pathway. The rationale behind that study was that if -secretase cleavage occurs in the compartment where transport was blocked, C99-GFP would be cleaved and the same diffuse GFP fluorescence as in control conditions would be observed. If however -secretase is not active in the compartment where transport is blocked, C99-GFP would remain membrane bound and highlight the respective organelle. Transport of C99-GFP was inhibited at the level of the ER, the Golgi and the TGN. In all cases, membrane-bound C99-GFP accumulated in the respective organelle, indicating that there is no substantial -secretase activity in the ER, the Golgi and the TGN. When exocytosis of post-Golgi vesicles was inhibited, bright fluorescent vesicles accumulated below the PM of C99-GFP-expressing cells, demonstrating that C99-GFP is not cleaved before it reaches the PM. In contrast, when endocytosis was blocked in C99-GFP-expressing cells, GFP staining was weak and diffuse, showing that -secretase cleaves C99-GFP at the PM. Whether -secretase is in addition active in endosomes could not be tested using this system. Taken together, our data show that -secretase activity is localized to the PM and/or endosomes [49].

Fig. 1. Live cell microscopy of HEK293 cells stably expressing PS1-

CFP and NCT-YFP. In addition, these cells express siRNA against endogenous NCT. PS1-CFP and NCT-YFP are fully functional and assemble into -secretase complexes. A Two-color microscopy of living cells shows a high degree of colocalization of PS1CFP and NCT-YFP at the PM (arrowheads) and in vesicular structures. Note that some vesicles appear green or red only in the merged image due to vesicular movement during image acquisition. B Three-color live cell microscopy demonstrates that the vesicular structures seen in A are endosomes/lysosomes, as demonstrated by labeling with lysotracker (arrows). Fig. 2. Assembly of -secretase. (1) Unassembled subunits of the -secretase complex are retained in the ER by specific ER retention signals (red bars). (2) APH-1 and NCT form a first assembly intermediate, which then stabilizes PS holoprotein (3). (4) Finally PEN-2 joins the complex, endoproteolysis of PS and conformational change of NCT take place and the fully assembled complex is exported from the ER through the Golgi via post-Golgi vesicles (PGV, green vesicles) to the PM. Protein-protein interacting domains are depicted in light blue. At the PM and/or in endosomes/ lysosomes (EL, red vesicles) the complex cleaves C99 (green) to release A and AICD.

Assembly, Trafficking and Function of -Secretase

Assembly of the -Secretase Complex

How do the components assemble to build a -secretase complex? Insight into this question was largely obtained from experiments using knockouts and knockdowns of the individual subunits in cultured cells. We found that knockdown of NCT and APH-1 by RNAi was accompanied by a strong reduction of the PS fragment levels [41, 50]. Furthermore, when PEN-2 expression was knocked down, the PS holoprotein accumulated in an unprocessed form [51]. In contrast, the knockout of PS was accompanied with decreased PEN-2 levels, while levels of NCT and APH-1 remained largely unchanged [35, 52]. Interestingly, NCT accumulated in its immature form, suggesting that the complex cannot exit the ER in the absence of PS [50]. Furthermore, we could show that the NCT ectodomain undergoes a conformational change during the assembly process [52]. Taken together, these results and the data from other investigators [36, 38–40, 53, 54] suggested a model for stepwise assembly of secretase complex(es) [41]. First, NCT and APH-1 form an initial stable scaffold for the PS holoprotein, which becomes stabilized by the interaction with the NCT/APH-1 assembly intermediate. Next, association of PEN-2 to this trimeric assembly intermediate triggers the endoproteolytic cleavage of the PS holoprotein (fig. 2). Finally, on the molecular level we and others could identify TMD4 of APH-1 [55, 56], the C-terminus of PEN-2 [51, 57–59], the TMD of NCT [60, 61] and the PS1 C-terminus [62, 63] as essential domains for functional -secretase complex assembly. The NCT TMD and the PS1 C-terminus interact directly with each other [62]. Recent data also suggest direct interactions of the PEN-2 TMD1 with the PS1 TMD4 [64, 65]. Are there specific signals or domains within the secretase subunits, which govern the assembly of the secretase complex? Analogous to ion channels and cell surface receptors, which are frequently composed of several subunits, we and others found that -secretase is assembled in the ER [59, 66]. In the case of ion channels and cell surface receptors, it is known that control mechanisms ensure that only fully assembled complexes leave the ER, while unassembled subunits are retained/retrieved by specific retention/retrieval signals. We hypothesized that similar mechanisms steer the correct assembly and export of -secretase out of the ER. Indeed, using reporter proteins to study cell surface transport, we could identify ER retention/retrieval signals in two -secretase subunits, PS1 and PEN-2. The ER retention/ retrieval signal in PS1 is located in the C-terminus and Neurodegenerative Dis 2006;3:275–283

279

includes the PALP-motif [62]. The ER-retention signal of PS1 is probably embedded in the membrane, as a number of groups recently showed that the hydrophobic part of the PS1 C-terminus spans the membrane. This suggests a topology with nine TMDs for PS [8–10]. In the case of PEN-2, the retention/retrieval signal is located in TMD1 and involves a critical asparagine [Kaether et al., submitted]. The molecular machinery recognizing these signals in mammals is unknown. In yeast, a protein called Rer1p was shown to retrieve unassembled subunits of several complexes to the ER. Retrieval was based on retrieval signals located in transmembrane segments and involved polar or charged amino acids surrounded by hydrophobic amino acids [67]. The human orthologue, Rer1, is a 23-kDa protein with four TMDs that can complement a yeast RER1 gene deletion strain [68]. We could show with reporter protein assays and deglycosylation experiments that human Rer1 is involved in the retention/retrieval of PEN-2. In addition, we showed that the mammalian Rer1 binds directly to unassembled PEN-2. Binding depends on a critical asparagine in the first TMD of PEN-2. Furthermore, overexpression of Rer1 stabilizes PEN-2 and enhances maturation of immature NCT, indicating an enhanced rate of complex formation. These data support the idea that PEN-2 is rate limiting for -secretase complex formation and identify Rer1 as a possible auxiliary factor for -secretase complex assembly [Kaether et al., submitted]. Cellular Function of -Secretase

PSs have been implicated in the Notch signaling pathway, which is required for cell differentiation during development and adulthood, due to the discovery of the C. elegans PS homologue SEL-12 as a key component of this signaling pathway [12]. Consistent with this finding, knockout of PS and the other -secretase subunits in mice causes phenotypes with Notch-like embryonic development deficiency [69]. In addition, we found in collaboration with Baumeister’s group that FAD-associated PS mutations largely fail to rescue the Notch deficiency phenotype of sel-12 mutant worms, whereas PS active site mutations do not rescue at all [17, 21, 70–72]. Notch is a cell surface receptor with type I membrane topology, which is processed in a very similar manner like APP [69]. Following cleavage of the Notch ectodomain at the cell surface, we found that -secretase cleaves the resultant C-terminal Notch membrane fragment to release N, an A-like peptide, into the extracellular space and 280

Neurodegenerative Dis 2006;3:275–283

the Notch intracellular domain (NICD) into the cytosol [73]. The NICD translocates to the nucleus, where it functions as a transcriptional regulator of target genes required for cell differentiation. Thus, enabling Notch signaling is a major function of -secretase. However, besides the two substrates APP and Notch, a rapidly increasing number of other substrates of -secretase have been discovered recently. Among these substrates are CD44 and LRP, both of which were identified in laboratories from the priority program [74, 75]. These substrates have little in common except that they are all type I transmembrane proteins that need to undergo ectodomain shedding that removes the bulk of their extracellular domains to become substrates for the enzyme [76]. While some of the ICDs liberated might have a function in nuclear signal transduction similar to the NICD, a more general function of -secretase may be the removal of membrane stubs of type I membrane proteins after ectodomain shedding. Interestingly, recent data suggest that NCT serves as a -secretase substrate sensor probably by measuring the length of the ectodomains of type I transmembrane proteins [77].

Conclusions

Together with the work in a number of other laboratories worldwide, the -secretase research within the priority program 1085 of the DFG has considerably expanded our knowledge about a long-sought, enigmatic enzyme that was known to be responsible for the final step in the biogenesis of A. We could clarify the identity of -secretase by demonstrating that -secretase is a complex consisting of four subunits that are required for its activity. We further developed an understanding of how the secretase complex assembles and how the complex traffics through the secretory pathway to its functional site(s). Finally, the identification of PS, the catalytic subunit of -secretase, as a prototype of novel aspartyl protease families and the elucidation of signals that govern secretase complex assembly also contributed to the opening of new research fields in cell biology.

Acknowledgments We wish to thank the Deutsche Forschungsgemeinschaft for funding our research. The work of our past and present co-workers is gratefully acknowledged. Due to space restrictions we apologize for not being able to properly cite all relevant work.

Kaether/Haass/Steiner

References 1 Hardy J, Selkoe DJ: The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science 2002;297:353–356. 2 Haass C: Take five-BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 3 Steiner H: Uncovering gamma-secretase. Curr Alzheimer Res 2004;1:175–181. 4 Sherrington R, Rogaev EI, Liang Y, Rogaeva EA, Levesque G, Ikeda M, Chi H, Lin C, Li G, Holman K, et al: Cloning of a gene bearing missense mutations in early-onset familial Alzheimer’s disease. Nature 1995; 375: 754– 760. 5 Rogaev EI, Sherrington R, Rogaeva EA, Levesque G, Ikeda M, Liang Y, Chi H, Lin C, Holman K, Tsuda T, et al: Familial Alzheimer’s disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer’s disease type 3 gene. Nature 1995;376:775–778. 6 Levy-Lahad E, Wasco W, Poorkaj P, Romano DM, Oshima J, Pettingell WH, Yu CE, Jondro PD, Schmidt SD, Wang K, et al: Candidate gene for the chromosome 1 familial Alzheimer’s disease locus. Science 1995; 269: 973–977. 7 Levy-Lahad E, Wijsman EM, Nemens E, Anderson L, Goddard KA, Weber JL, Bird TD, Schellenberg GD: A familial Alzheimer’s disease locus on chromosome 1. Science 1995;269:970–973. 8 Laudon H, Hansson EM, Melen K, Bergman A, Farmery MR, Winblad B, Lendahl U, von Heijne G, Naslund J: A nine-transmembrane domain topology for presenilin 1. J Biol Chem 2005;280:35352–35360. 9 Oh YS, Turner RJ: Evidence that the COOH terminus of human presenilin 1 is located in extracytoplasmic space. Am J Physiol Cell Physiol 2005;289:C576–581. 10 Henricson A, Kall L, Sonnhammer EL: A novel transmembrane topology of presenilin based on reconciling experimental and computational evidence. FEBS J 2005;272: 2727– 2733. 11 Thinakaran G, Borchelt DR, Lee MK, Slunt HH, Spitzer L, Kim G, Ratovitsky T, Davenport F, Nordstedt C, Seeger M, Hardy J, Levey AI, Gandy SE, Jenkins NA, Copeland NG, Price DL, Sisodia SS: Endoproteolysis of presenilin 1 and accumulation of processed derivatives in vivo. Neuron 1996;17:181–190. 12 Levitan D, Greenwald I: Facilitation of lin12-mediated signalling by sel-12, a Caenorhabditis elegans S182 Alzheimer’s disease gene. Nature 1995;377:351–354. 13 Scheuner D, Eckman C, Jensen M, Song X, Citron M, Suzuki N, Bird TD, Hardy J, Hutton M, Kukull W, Larson E, Levy-Lahad E, Viitanen M, Peskind E, Poorkaj P, Schellenberg G, Tanzi R, Wasco W, Lannfelt L, Selkoe

Assembly, Trafficking and Function of -Secretase

14

15

16

17

18

19

20

21

22

D, Younkin S: Secreted amyloid beta-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. Nat Med 1996;2:864–870. 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 amyloid precursor protein. Nature 1998; 391: 387–390. Wolfe MS, Xia W, Ostaszewski BL, Diehl TS, Kimberly WT, Selkoe DJ: Two transmembrane aspartates in presenilin-1 required for presenilin endoproteolysis and gammasecretase activity. Nature 1999;398:513–517. Wolfe MS, Xia W, Moore CL, Leatherwood DD, Ostaszewski B, Rahmati T, Donkor IO, Selkoe DJ: Peptidomimetic probes and molecular modeling suggest that Alzheimer’s gamma-secretase is an intramembranecleaving aspartyl protease. Biochemistry 1999;38:4720–4727. Steiner H, Duff K, Capell A, Romig H, Grim MG, Lincoln S, Hardy J, Yu X, Picciano M, Fechteler K, Citron M, Kopan R, Pesold B, Keck S, Baader M, Tomita T, Iwatsubo T, Baumeister R, Haass C: A loss of function mutation of presenilin-2 interferes with amyloid beta-peptide production and notch signaling. J Biol Chem 1999; 274:28669–28673. Leimer U, Lun K, Romig H, Walter J, Grunberg J, Brand M, Haass C: Zebrafish (Danio rerio) presenilin promotes aberrant amyloid beta- peptide production and requires a critical aspartate residue for its function in amyloidogenesis. Biochemistry 1999; 38: 13602– 13609. Li YM, Xu M, Lai MT, Huang Q, Castro JL, DiMuzio-Mower J, Harrison T, Lellis C, Nadin A, Neduvelli JG, Register RB, Sardana MK, Shearman MS, Smith AL, Shi XP, Yin KC, Shafer JA, Gardell SJ: Photoactivated g-secretase inhibitors directed to the active site covalently label presenilin 1. Nature 2000;405:689–694. Esler WP, Kimberly WT, Ostaszewski BL, Diehl TS, Moore CL, Tsai JY, Rahmati T, Xia W, Selkoe DJ, Wolfe MS: Transition-state analogue inhibitors of g-secretase bind directly to presenilin-1. Nature Cell Biol 2000; 2: 428–433. Steiner H, Kostka M, Romig H, Basset G, Pesold B, Hardy J, Capell A, Meyn L, Grim MG, Baumeister R, Fechteler K, Haass C: Glycine 384 is required for presenilin-1 function and is conserved in polytopic bacterial aspartyl proteases. Nature Cell Biol 2000; 2:848–851. LaPointe CF, Taylor RK: The type 4 prepilin peptidases comprise a novel family of aspartic acid proteases. J Biol Chem 2000; 275: 1502–1510.

23 Weihofen A, Binns K, Lemberg MK, Ashman K, Martoglio B: Identification of signal peptide peptidase, a presenilin-type aspartic protease. Science 2002;296:2215–2218. 24 Krawitz P, Haffner C, Fluhrer R, Steiner H, Schmid B, Haass C: Differential localization and identification of a critical aspartate suggest non-redundant proteolytic functions of the presenilin homologues SPPL2b and SPPL3. J Biol Chem 2005;280:39515–39523. 25 Haass C, Steiner H: Alzheimer disease gamma-secretase: a complex story of GxGD-type presenilin proteases. Trends Cell Biol 2002; 12:556–562. 26 Seeger M, Nordstedt C, Petanceska S, Kovacs DM, Gouras GK, Hahne S, Fraser P, Levesque L, Czernik AJ, George-Hyslop PS, Sisodia SS, Thinakaran G, Tanzi RE, Greengard P, Gandy S: Evidence for phosphorylation and oligomeric assembly of presenilin 1. Proc Natl Acad Sci USA 1997;94:5090–5094. 27 Capell A, Grunberg J, Pesold B, Diehlmann A, Citron M, Nixon R, Beyreuther K, Selkoe DJ, Haass C: The proteolytic fragments of the Alzheimer’s disease-associated presenilin-1 form heterodimers and occur as a 100– 150-kDa molecular mass complex. J Biol Chem 1998;273:3205–3211. 28 Yu G, Chen F, Levesque G, Nishimura M, Zhang DM, Levesque L, Rogaeva E, Xu D, Liang Y, Duthie M, St George-Hyslop PH, Fraser PE: The presenilin 1 protein is a component of a high molecular weight intracellular complex that contains b-catenin. J Biol Chem 1998;273:16470–16475. 29 Li YM, Lai MT, Xu M, Huang Q, DiMuzioMower J, Sardana MK, Shi XP, Yin KC, Shafer JA, Gardell SJ: Presenilin 1 is linked with gamma-secretase activity in the detergent solubilized state. Proc Natl Acad Sci USA 2000;97:6138–6143. 30 Thinakaran G, Harris CL, Ratovitski T, Davenport F, Slunt HH, Price DL, Borchelt DR, Sisodia SS: Evidence that levels of presenilins (PS1 and PS2) are coordinately regulated by competition for limiting cellular factors. J Biol Chem 1997; 272:28415–28422. 31 Steiner H, Capell A, Pesold B, Citron M, Kloetzel PM, Selkoe DJ, Romig H, Mendla K, Haass C: Expression of Alzheimer’s diseaseassociated presenilin-1 is controlled by proteolytic degradation and complex formation. J Biol Chem 1998; 273:32322–32331. 32 Yu G, Nishimura M, Arawaka S, Levitan D, Zhang L, Tandon A, Song YQ, Rogaeva E, Chen F, Kawarai T, Supala A, Levesque L, Yu H, Yang DS, Holmes E, Milman P, Liang Y, Zhang DM, Xu DH, Sato C, Rogaev E, Smith M, Janus C, Zhang Y, Aebersold R, Farrer LS, Sorbi S, Bruni A, Fraser P, St George-Hyslop P: Nicastrin modulates presenilin-mediated notch/glp-1 signal transduction and betaAPP processing. Nature 2000;407:48–54.

Neurodegenerative Dis 2006;3:275–283

281

33 Francis R, McGrath G, Zhang J, Ruddy DA, Sym M, Apfeld J, Nicoll M, Maxwell M, Hai B, Ellis MC, Parks AL, Xu W, Li J, Gurney M, Myers RL, Himes CS, Hiebsch R, Ruble C, Nye JS, Curtis D: Aph-1 and pen-2 are required for Notch pathway signaling, gamma-secretase cleavage of betaAPP, and presenilin protein accumulation. Dev Cell 2002; 3:85–97. 34 Goutte C, Tsunozaki M, Hale VA, Priess JR: APH-1 is a multipass membrane protein essential for the Notch signaling pathway in Caenorhabditis elegans embryos. Proc Natl Acad Sci USA 2002;99:775–779. 35 Steiner H, Winkler E, Edbauer D, Prokop S, Basset G, Yamasaki A, Kostka M, Haass C: PEN-2 is an integral component of the gamma-secretase complex required for coordinated expression of presenilin and nicastrin. J Biol Chem 2002; 277:39062–39065. 36 Lee SF, Shah S, Li H, Yu C, Han W, Yu G: Mammalian APH-1 interacts with presenilin and nicastrin and is required for intramembrane proteolysis of amyloid-beta precursor protein and Notch. J Biol Chem 2002; 277:45013–45019. 37 Edbauer D, Winkler E, Regula JT, Pesold B, Steiner H, Haass C: Reconstitution of gamma-secretase activity. Nat Cell Biol 2003; 5: 486–488. 38 Kim SH, Ikeuchi T, Yu C, Sisodia SS: Regulated hyperaccumulation of presenilin-1 and the ‘gamma-secretase’ complex. Evidence for differential intramembranous processing of transmembrane subatrates. J Biol Chem 2003;278:33992–34002. 39 Kimberly WT, LaVoie MJ, Ostaszewski BL, Ye W, Wolfe MS, Selkoe DJ: Gamma-secretase is a membrane protein complex comprised of presenilin, nicastrin, Aph-1, and Pen-2. Proc Natl Acad Sci USA 2003; 100: 6382–6387. 40 Takasugi N, Tomita T, Hayashi I, Tsuruoka M, Niimura M, Takahashi Y, Thinakaran G, Iwatsubo T: The role of presenilin cofactors in the gamma-secretase complex. Nature 2003;422:438–441. 41 Shirotani K, Edbauer D, Kostka M, Steiner H, Haass C: Immature nicastrin stabilizes APH-1 independent of PEN-2 and presenilin: identification of nicastrin mutants that selectively interact with APH-1. J Neurochem 2004;89:1520–1527. 42 Hebert SS, Serneels L, Dejaegere T, Horre K, Dabrowski M, Baert V, Annaert W, Hartmann D, De Strooper B: Coordinated and widespread expression of gamma-secretase in vivo: evidence for size and molecular heterogeneity. Neurobiol Dis 2004;17:260–272. 43 Kovacs DM, Fausett HJ, Page KJ, Kim TW, Moir RD, Merriam DE, Hollister RD, Hallmark OG, Mancini R, Felsenstein KM, Hyman BT, Tanzi RE, Wasco W: Alzheimer-associated presenilins 1 and 2: neuronal expression in brain and localization to intracellular membranes in mammalian cells. Nat Med 1996;2:224–229.

282

44 Pasternak SH, Callahan JW, Mahuran DJ: The role of the endosomal/lysosomal system in amyloid-beta production and the pathophysiology of Alzheimer’s disease: reexamining the spatial paradox from a lysosomal perspective. J Alzheimers Dis 2004; 6: 53– 65. 45 Annaert W, De Strooper B: Presenilins: molecular switches between proteolysis and signal transduction. Trends Neurosci 1999; 22: 439–443. 46 Kaether C, Lammich S, Edbauer D, Ertl M, Rietdorf J, Capell A, Steiner H, Haass C: Presenilin-1 affects trafficking and processing of betaAPP and is targeted in a complex with nicastrin to the plasma membrane. J Cell Biol 2002;158:551–561. 47 Chyung JH, Raper DM, Selkoe DJ: Gammasecretase exists on the plasma membrane as an intact complex that accepts substrates and effects intramembrane cleavage. J Biol Chem 2005;280:4383–4392. 48 Pasternak SH, Bagshaw RD, Guiral M, Zhang S, Ackerley CA, Pak BJ, Callahan JW, Mahuran DJ: Presenilin-1, nicastrin, amyloid precursor protein, and gamma-secretase activity are co-localized in the lysosomal membrane. J Biol Chem 2003; 278:26687–26694. 49 Kaether C, Schmitt S, Willem M, Haass C: APP and Notch intracellular domains are generated after transport of their precursors to the cell surface. Traffic 2006, in press. 50 Edbauer D, Winkler E, Haass C, Steiner H: Presenilin and nicastrin regulate each other and determine amyloid beta -peptide production via complex formation. Proc Natl Acad Sci USA 2002;99:8666–8671. 51 Prokop S, Shirotani K, Edbauer D, Haass C, Steiner H: Requirement of PEN-2 for stabilization of the presenilin NTF/CTF heterodimer within the gamma-secretase complex. J Biol Chem 2004; 279:23255–23261. 52 Shirotani K, Edbauer D, Capell A, Schmitz J, Steiner H, Haass C: Gamma-secretase activity is associated with a conformational change of nicastrin. J Biol Chem 2003; 278: 16474–16477. 53 Luo WJ, Wang H, Li H, Kim BS, Shah S, Lee HJ, Thinakaran G, Kim TW, Yu G, Xu H: PEN-2 and APH-1 coordinately regulate proteolytic processing of presenilin 1. J Biol Chem 2003;278:7850–7854. 54 Leem JY, Vijayan S, Han P, Cai D, Machura M, Lopes KO, Veselits ML, Xu H, Thinakaran G: Presenilin 1 is required for maturation and cell surface accumulation of nicastrin. J Biol Chem 2002; 277:19236–19240. 55 Edbauer D, Kaether C, Steiner H, Haass C: Co-expression of nicastrin and presenilin rescues a loss of function mutant of APH-1. J Biol Chem 2004; 279:37311–37315. 56 Lee SF, Shah S, Yu C, Wigley WC, Li H, Lim M, Pedersen K, Han W, Thomas P, Lundkvist J, Hao YH, Yu G: A conserved GXXXG motif in APH-1 is critical for assembly and activity of the gamma-secretase complex. J Biol Chem 2004;279:4144–4152.

Neurodegenerative Dis 2006;3:275–283

57 Prokop S, Haass C, Steiner H: Length and overall sequence of the PEN-2 C-terminal domain determines its function in the stabilization of presenilin fragments. J Neurochem 2005;94:57–62. 58 Hasegawa H, Sanjo N, Chen F, Gu YJ, Shier C, Petit A, Kawarai T, Katayama T, Schmidt SD, Mathews PM, Schmitt-Ulms G, Fraser PE, St George-Hyslop P: Both the sequence and length of the C terminus of PEN-2 are critical for intermolecular interactions and function of presenilin complexes. J Biol Chem 2004;279:46455–46463. 59 Kim SH, Yin YI, Li YM, Sisodia SS: Evidence that assembly of an active gamma-secretase complex occurs in the early compartments of the secretory pathway. J Biol Chem 2004;279: 48615–48619. 60 Capell A, Kaether C, Edbauer D, Shirotani K, Merkl S, Steiner H, Haass C: Nicastrin interacts with gamma-secretase complex components via the N-terminal part of its transmembrane domain. J Biol Chem 2003; 278: 52519–52523. 61 Morais VA, Crystal AS, Pijak DS, Carlin D, Costa J, Lee VM, Doms RW: The transmembrane domain region of nicastrin mediates direct interactions with APH-1 and the gamma-secretase complex. J Biol Chem 2003; 278:43284–43291. 62 Kaether C, Capell A, Edbauer D, Winkler E, Novak B, Steiner H, Haass C: The presenilin C-terminus is required for ER-retention, nicastrin-binding and gamma-secretase activity. EMBO J 2004;23:4738–4748. 63 Bergman A, Laudon H, Winblad B, Lundkvist J, Naslund J: The extreme C terminus of presenilin 1 is essential for gamma-secretase complex assembly and activity. J Biol Chem 2004;279:45564–45572. 64 Kim SH, Sisodia SS: Evidence that the ‘NF’ motif in transmembrane domain 4 of presenilin 1 is critical for binding with PEN-2. J Biol Chem 2005;280:41953–41966. 65 Watanabe N, Tomita T, Sato C, Kitamura T, Morohashi Y, Iwatsubo T: Pen-2 is incorporated into the gamma-secretase complex through binding to transmembrane domain 4 of presenilin 1. J Biol Chem 2005; 280: 41967–41975. 66 Capell A, Beher D, Prokop S, Steiner H, Kaether C, Shearman MS, Haass C: Gammasecretase complex assembly within the early secretory pathway. J Biol Chem 2005; 280: 6471–6478. 67 Sato K, Sato M, Nakano A: Rer1p, a retrieval receptor for ER membrane proteins, recognizes transmembrane domains in multiple modes. Mol Biol Cell 2003; 14:3605–3616. 68 Füllekrug J, Boehm J, Rottger S, Nilsson T, Mieskes G, Schmitt HD: Human Rer1 is localized to the Golgi apparatus and complements the deletion of the homologous Rer1 protein of Saccharomyces cerevisiae. Eur J Cell Biol 1997; 74:31–40.

Kaether/Haass/Steiner

69 Brunkan AL, Goate AM: Presenilin function and gamma-secretase activity. J Neurochem 2005;93:769–792. 70 Baumeister R, Leimer U, Zweckbronner I, Jakubek C, Grunberg J, Haass C: Human presenilin-1, but not familial Alzheimer’s disease (FAD) mutants, facilitate Caenorhabditis elegans Notch signalling independently of proteolytic processing. Genes Funct 1997; 1:149–159. 71 Steiner H, Romig H, Grim MG, Philipp U, Pesold B, Citron M, Baumeister R, Haass C: The biological and pathological function of the presenilin-1delta exon 9 mutation is independent of its defect to undergo proteolytic processing. J Biol Chem 1999; 274: 7615– 7618.

Assembly, Trafficking and Function of -Secretase

72 Capell A, Steiner H, Romig H, Keck S, Baader M, Grim MG, Baumeister R, Haass C: Presenilin-1 differentially facilitates endoproteolysis of the beta-amyloid precursor protein and Notch. Nature Cell Biol 2000; 2: 205–211. 73 Okochi M, Steiner H, Fukumori A, Tanii H, Tomita T, Tanaka T, Iwatsubo T, Kudo T, Takeda M, Haass C: Presenilins mediate a dual intramembranous gamma-secretase cleavage of Notch-1. EMBO J 2002;21:5408– 5416. 74 Lammich S, Okochi M, Takeda M, Kaether C, Capell A, Zimmer AK, Edbauer D, Walter J, Steiner H, Haass C: Presenilin-dependent intramembrane proteolysis of CD44 leads to the liberation of its intracellular domain and the secretion of an Abeta – like peptide. J Biol Chem 2002;277:44754–44759.

75 Lleo A, Waldron E, von Arnim CA, Herl L, Tangredi MM, Peltan ID, Strickland DK, Koo EH, Hyman BT, Pietrzik CU, Berezovska O: Low density lipoprotein receptor-related protein (LRP) interacts with presenilin 1 and is a competitive substrate of the amyloid precursor protein (APP) for gammasecretase. J Biol Chem 2005; 280: 27303– 27309. 76 Struhl G, Adachi A: Requirements for presenilin-dependent cleavage of Notch and other transmembrane proteins. Mol Cell 2000; 6: 625–636. 77 Shah S, Lee SF, Tabuchi K, Hao YH, Yu C, LaPlant Q, Ball H, Dann CE, 3rd, Sudhof T, Yu G: Nicastrin functions as a gamma-secretase-substrate receptor. Cell 2005; 122: 435– 447.

Neurodegenerative Dis 2006;3:275–283

283

Original Paper

Diseases

Neurodegenerative Dis 2006;3:284–289 DOI: 10.1159/000095268

Cellular Functions of -Secretase-Related Proteins Christof Haffner Christian Haass Laboratory for Alzheimer’s and Parkinson’s Disease Research, Department of Biochemistry, Adolf-Butenandt-Institute, Ludwig-Maximilians-University, Munich, Germany

Key Words Nodal modulator Presenilin Signal peptide peptidase Nodal signaling pathway Nicastrin-like protein

protein complex different from -secretase. We found that during zebrafish embryogenesis this complex is involved in the patterning of the axial mesendoderm, a process controlled by the Nodal signaling pathway. Copyright © 2006 S. Karger AG, Basel

Abstract Amyloid- peptide (A) is generated by -secretase, a membrane protein complex with an unusual aspartyl protease activity consisting of the four components presenilin, nicastrin, APH-1 and PEN-2. Presenilin is considered the catalytic subunit of this complex since it represents the prototype of the new family of intramembrane-cleaving GxGD-type aspartyl proteases. Recently, five novel members of this family and a nicastrin-like protein were identified. Whereas one of the GxGD-type proteins was shown to be identical with signal peptide peptidase (SPP), the function of the others, now called SPP-like proteins (SPPLs), is not known. We therefore analyzed SPPL2b and SPPL3 and demonstrated that they localize to different subcellular compartments suggesting nonredundant functions. This was supported by different phenotypes obtained in knockdown studies in zebrafish embryos. In addition, these phenotypes could be phenocopied by ectopic expression of putative active site mutants, providing strong evidence for a proteolytic function of SPPL2b and SPPL3. We also identified and characterized the nicastrin-like protein nicalin which, together with the 130kDa protein NOMO (Nodal modulator), forms a membrane

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0284$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Introduction

Intramembrane proteolysis is an only recently identified mechanism underlying various important cellular processes such as cholesterol homeostasis, ER stress, cell fate decisions, signal peptide cleavage, and removal of transmembrane domain (TMD) stubs [1]. An extensively studied example of intramembrane proteolysis is the secretase-mediated cleavage of the -amyloid precursor protein (APP), a critical step in the development of Alzheimer’s disease (AD) [2]. Upon initial ectodomain shedding by - or -secretase, the remaining APP C-terminal fragment is cleaved within its TMD by -secretase, a membrane protein complex of high molecular weight with an unusual aspartyl protease activity. The consecutive cleavage of APP by - and -secretase results in the generation of small peptides, including the 40- to 42amino acid amyloid- peptide (A), which are released into the extracellular space [2]. The neurotoxic, highly amyloidogenic 42-amino acid A variant is believed to Christof Haffner, Laboratory for Alzheimer’s and Parkinson’s Disease Research Department of Biochemistry, Adolf-Butenandt-Institute Ludwig-Maximilians-University, DE–80336 Munich (Germany) Tel. +49 89 218 075 484, Fax +49 89 218 075 415 E-Mail [email protected]

be the pathological agent that initiates AD. The physiological role of the APP processing is not known, but secretase-mediated cleavage of other substrates, most importantly the Notch receptor, is required during development and maintenance of multicellular organisms [3]. Four -secretase complex components have been identified: presenilin, nicastrin, APH-1 and PEN-2 [4]. Their simultaneous expression in yeast, an organism that lacks any endogenous -secretase activity, results in the reconstitution of -secretase complex formation and activity, demonstrating that these four membrane proteins are the core components of the complex [5]. -Secretase activity depends on the presence of two conserved aspartate residues in either presenilin 1 or presenilin 2, two polytopic membrane proteins with partially redundant function [4]. Presenilins constitutively undergo endoproteolysis, leading to the generation of N- and C-terminal fragments which remain bound to each other and apparently constitute the catalytic site of the -secretase. The active site aspartate residues of the presenilins reside within their TMDs 6 and 7, in agreement with an intramembrane cleavage mechanism. The TMD 7 aspartate is located within the highly conserved sequence motif GxGD [6], which is also found in other intramembrane cleaving proteases, the bacterial type 4 prepilin peptidases and signal peptide peptidase (SPP) (see below). These and other data strongly suggest that the presenilins are the proteolytically active components of the -secretase complex. In contrast, the role of the other three subunits is less well understood. The type I glycoprotein nicastrin (see below) and the polytopic membrane protein APH-1 might form a precomplex to which first presenilin holoprotein is added [7]. The subsequent addition of PEN-2 facilitates presenilin endoproteolysis leading to the generation of the active complex. For a detailed description of -secretase assembly and function see the paper by Kaether et al. [this issue, pp. 275–283].

SPP-Like Proteins

SPP and SPP-like proteins (SPPLs) were originally identified as presenilin homologues (PSHs) in a database search [8]. Five PSH genes were described and shown to encode polytopic membrane proteins with a predicted topology similar, but inverted to that of presenilins (fig. 1a). Moreover, their sequences contained two conserved aspartate residues within putative TMDs and a highly conserved PALLYL motif reminiscent of the presenilin PALP sequence, suggesting the existence of a family of unusual Cellular Functions of -Secretase-Related Proteins

aspartyl proteases with the sequence GxGD as a signature motif [9]. This hypothesis was confirmed by the finding that one of the PSHs is identical to SPP [10], a protease cleaving the hydrophobic signal peptides after their removal from newly synthesized secreted or membrane proteins by signal peptidase. Mutation of the aspartate within the GxGD motif of SPP abolished this activity, supporting the idea that SPP is indeed an aspartyl protease. Subsequently, -secretase inhibitors were shown to also block SPP activity, suggesting a common cleavage mechanism of -secretase and SPP [11]. In addition, the liberation of signal peptides by signal peptidase, a step analogous to the ectodomain shedding of -secretase substrates, was shown to be required for intramembrane processing, indicating that SPP and -secretase use a similar substrate recognition mechanism [12]. Thus, the discrimination of SPP and -secretase substrates appears to be primarily based on their membrane orientation: secretase cleaves only type I substrates, SPP accepts only type II substrates, in agreement with the inverted membrane orientation of their active site-forming TMDs [13]. However, SPP differs from presenilin in that it does not undergo endoproteolysis and that it does not need a highmolecular-weight complex for activity. Although SPP dimerization has been shown to occur [13–15], its biological relevance is not clear. One of the physiological functions of signal peptide processing in mammalian cells is the generation of fragments from MHC class I molecules which serve as human lymphocyte antigen E epitopes to report the immunocompetence of antigen-presenting cells [16]. In Caenorhabditis elegans, SPP has been shown to be required for embryonic development and suggested to be involved in the lipoprotein receptor pathway [17]. In contrast, nothing is known about the function of the other four PSHs, now known as SPPLs. Based on their sequence homology to SPP [8] and their membrane topology [13], they might represent intramembrane cleaving aspartyl proteases as well, but no substrates have been identified so far. Two of the four human SPPLs, SPPL2b and SPPL3, are conserved in all vertebrates, including the zebrafish, and were chosen for a biochemical and functional study [15]. Analysis of their subcellular localization in cultured human embryonic kidney HEK293 cells by immunofluorescence microscopy revealed that SPPL3, like SPP, localizes to the endoplasmic reticulum, whereas SPPL2b is predominantly found in endocytic/lysosomal compartments (fig. 1b). This suggested that at least one of the SPPLs might have a function different from SPP. To examine this further, we studied the role of SPP, SPPL2b and SPPL3 Neurodegenerative Dis 2006;3:284–289

285

cytosol NH 2

presenilins Fig. 1. a Presenilins and SPP/SPPLs have opposite membrane topology. Both are believed to contain nine TMDs with the presenilins’ N-termini in the cytosol and the SPP/SPPLs’ N-termini in the lumen (SPPL2a, b, c also contain a signal peptide which is not shown here). In contrast to SPP/SPPLs, presenilins are endoproteolyzed within the large cytoplasmic loop during -secretase complex formation. b SPPL2b localizes to endosomal/lysosomal compartments. Immunofluorescent microscopy of HEK293 cells shows a high degree of colocalization of SPPL2b with LAMP-2, a marker for lysosomes, and only little colocalization with BiP, an endoplasmic reticulum-resident protein.

D D COOH

GxGD cytosol COOH

SPP/SPPLs a

D D NH 2

SPPL2b

LAMP-2

merge

SPPL2b

BiP

merge

b

during zebrafish embryonic development [15]. The corresponding zebrafish genes, spp, sppl2 and sppl3 are highly homologous to their human counterparts and expressed throughout early embryogenesis. Knockdown of each of these genes using antisense oligonucleotides resulted in embryonic lethal phenotypes demonstrating an essential role in zebrafish development. Whereas injection of spp and sppl3-specific oligonucleotides led to a neuronal degeneration phenotype due to enhanced cell 286

Neurodegenerative Dis 2006;3:284–289

death, knockdown of sppl2b resulted in the development of an abnormal caudal vein [15]. Importantly, we obtained identical phenotypes by injecting mRNAs encoding SPP or SPPL proteins harboring a mutation of the aspartate residue within the GxGD motif. This mutation has already been shown to act as dominant-negative mutation in human SPP [10]. We, therefore, concluded that SPPLs most likely represent aspartyl proteases with nonredundant functions. Haffner/Haass

Fig. 2. a Co-expression of ncl1 and nomo

b Nodal

NE X T

receptor P

Presenilin APH-1

Nicastrin PEN-2

Nicalin NOMO Smad2/3 P

Smad4

NICD

ER

Smad4

Smad2/3

NICD

P

Cofactors

Cofactors

Nicalin (Nicastrin-Like Protein)

The type I transmembrane protein nicastrin was originally identified as presenilin-binding factor that affects APP processing [18] as well as Notch signaling in C. elegans [18, 19]. It was subsequently shown to represent an obligate component of the -secretase complex [20] and to be required for the reconstitution of -secretase activity [5]. Its extracytosolic domain contains N-linked carbohydrates which undergo maturation during passage through the secretory pathway [20, 21]. This maturation, together with a conformational change leading to trypsin resistance [22], has been used as an indicator of -secretase complex assembly and trafficking. Within nicastrin’s ectodomain, a 200-amino acid region was found which is predicted to adopt a fold similar to the aminopeptidase (AP) domain. Although this domain does not confer pepCellular Functions of -Secretase-Related Proteins

DSL ligands

Nodal

N o tc h

in zebrafish embryos results in cyclopia. Morphology of a wild-type embryo (left panel) and an embryo co-injected with ncl1 and nomo capped RNAs (right panel) 36 h after fertilization. Note the prominent cyclopia of the injected embryo (arrow). b Roles of nicalin and nicastrin in the Nodal (left panel) and Notch (right panel) signaling pathways, respectively. The Nodal signal is transmitted by kinase receptors which phosphorylate intracellular signal transducers of the Smad family leading to the formation and nuclear translocation of heteromeric Smad complexes. In the nucleus, these complexes associate with cofactors to activate the transcription of target genes. The nicalin/NOMO complex localizes to the endoplasmic reticulum (ER) membrane and antagonizes the Nodal pathway at an undefined step. The Notch receptor is activated by ligands of the Delta/Serrate/LAG-2 (DSL) family and undergoes two proteolytic cleavage events. The second cleavage is mediated by secretase, a membrane protein complex consisting of at least four components including nicastrin. The generated Notch intracellular domain (NICD) translocates to the nucleus and, together with cofactors, regulates the transcription of target genes. NEXT = Notch extracellular truncation; APH-1 = anterior pharynx defective-1; PEN-2 = presenilin enhancer-2.

nucleus

nucleus

tidase activity [23], its integrity is necessary for -secretase activity [18, 22]. We have used a generalized sequence profile constructed from nicastrin ectodomains of various species to search databases for nicastrin-related sequences and identified a novel protein which we termed nicalin (nicastrin-like protein) [24]. The sequence similarity is confined to a region of 180 residues, which roughly corresponds to nicastrin’s AP domain. Northern and Western blot analysis revealed ubiquitous expression of nicalin in human tissues and cell lines, albeit at varying levels. Examination of native, endogenous nicalin from HEK293 cells by Blue-Native polyacrylamid gel electrophoresis revealed its presence in a high-molecular-weight membrane protein complex of 200 kDa [Haffner and Haass, unpubl. data], a size clearly differing from the 500–550 kDa determined for -secretase [20]. Moreover, -secretase components were not detected in nicalin imNeurodegenerative Dis 2006;3:284–289

287

munoprecipitates and vice versa, demonstrating that these complexes are unrelated. Immunoaffinity purification of nicalin resulted in the co-purification of a 130kDa protein, which was identified by mass spectrometry as pM5 [24], a membrane protein of unknown function. Based on the results of our analysis in zebrafish embryos (see below) this protein was termed NOMO (Nodal modulator). Like nicalin, NOMO is highly conserved in higher eukaryotes and expressed ubiquitously. Native, endogenous NOMO has a molecular weight of 200 kDa strongly suggesting its presence in the Nicalin complex. Glycosylation and subcellular localization studies revealed that both proteins have type I topology and localize to the endoplasmic reticulum [Haffner and Haass, unpubl. data]. To identify the biological function of this novel protein complex, we examined the role of nicalin and NOMO in the zebrafish. One NOMO (nomo) and two nicalin orthologs (ncl1, ncl2) were found in the zebrafish genome, of which nomo and ncl1 are expressed during early development. Interfering with their expression levels in embryos failed to produce phenotypes related to Notch signaling deficiencies, indicating distinct functions of secretase and the nicalin/NOMO complex in vivo [24]. In contrast, we found that simultaneous ectopic expression of nomo and ncl1, but not of each factor alone, led to cyclopic, ‘squint-eyed’ embryos (fig. 2a). Blocking expression of nomo with an antisense oligonucleotide resulted in the development of a massively enlarged hatching gland. Both phenotypes can arise through a failure in the proper patterning of embryonic mesendodermal tissue, a process regulated by the TGF factor Nodal [25]. In situ hybridization experiments demonstrated that downregulation of nomo reduced the amount of posterior axial mesendoderm, indicating enhanced Nodal signaling. These data suggested an antagonistic role of NOMO in

the Nodal pathway, and were confirmed by the finding that enhanced inhibition of Nodal signaling by overexpressing the specific Nodal inhibitor lefty was counteracted by blocking nomo expression [24]. We therefore concluded that the nicalin/NOMO complex acts as inhibitor of the Nodal signaling pathway.

Concluding Remarks

We characterized three novel -secretase-related proteins, SPPL2b, SPPL3 and nicalin. SPPL2b and SPPL3 are members of the GxGD family of aspartyl proteases and our data strongly support the hypothesis that they are active proteases [15]. This is confirmed by very recent results from our lab identifying a substrate for SPPL2b [Fluhrer et al., in press]. Moreover, it appears that SPPL2b and SPPL3 fulfill nonredundant functions important for vertebrate development. The characterization of substrates for SPPLs and their mode of action will provide important insights into the mechanisms underlying the process of intramembrane proteolysis. Our analysis of the nicastrin-like protein nicalin and its binding partner NOMO led to the identification of a novel membrane protein complex involved in the Nodal signaling pathway [24]. Thus, nicalin as well as nicastrin are both part of high-molecular-weight protein complexes which are involved in signaling pathways controlling cell fate decisions during embryonic development (fig. 2b). Very recently, it has been suggested that nicastrin functions as a receptor for -secretase substrates and that its AP domain is involved in this process by recognizing the N-termini of processed peptides [26]. Future studies of nicalin and NOMO might help to understand the precise role of nicastrin within the -secretase complex.

References 1 Weihofen A, Martoglio B: Intramembranecleaving proteases: controlled liberation of functional proteins and peptides from membranes. Trends Cell Biol 2003; 13:71–78. 2 Steiner H, Haass C: Intramembrane proteolysis by presenilins. Nat Rev Mol Cell Biol 2000;1:217–224. 3 Fortini ME: Gamma-secretase-mediated proteolysis in cell-surface-receptor signalling. Nat Rev Mol Cell Biol 2002;3:673–684.

288

4 Haass C: Take five – BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 5 Edbauer D, Winkler E, Regula JT, Pesold B, Steiner H, Haass C: Reconstitution of gamma-secretase activity. Nat Cell Biol 2003; 5: 486–488.

Neurodegenerative Dis 2006;3:284–289

6 Steiner H, Kostka M, Romig H, Basset G, Pesold B, Hardy J, Capell A, Meyn L, Grim MG, Baumeister R, Fechteler K, Haass C: Glycine 384 is required for presenilin-1 function and is conserved in polytopic bacterial aspartyl proteases. Nat Cell Biol 2000; 2:848–851. 7 Takasugi N, Tomita T, Hayashi I, Tsuruoka M, Niimura M, Takahashi Y, Thinakaran G, Iwatsubo T: The role of presenilin cofactors in the gamma-secretase complex. Nature 2003;422:438–441.

Haffner/Haass

8 Ponting CP, Hutton M, Nyborg A, Baker M, Jansen K, Golde TE: Identification of a novel family of presenilin homologues. Hum Mol Genet 2002;11:1037–1044. 9 Haass C, Steiner H: Alzheimer disease gamma-secretase: a complex story of GxGD-type presenilin proteases. Trends Cell Biol 2002; 12:556–562. 10 Weihofen A, Binns K, Lemberg MK, Ashman K, Martoglio B: Identification of signal peptide peptidase, a presenilin-type aspartic protease. Science 2002;296:2215–2218. 11 Weihofen A, Lemberg MK, Friedmann E, Rueeger H, Schmitz A, Paganetti P, Rovelli G, Martoglio B: Targeting presenilin-type aspartic protease signal peptide peptidase with gamma-secretase inhibitors. J Biol Chem 2003;278:16528–16533. 12 Lemberg MK, Martoglio B: Requirements for signal peptide peptidase-catalyzed intramembrane proteolysis. Mol Cell 2002; 10: 735–744. 13 Friedmann E, Lemberg MK, Weihofen A, Dev KK, Dengler U, Rovelli G, Martoglio B: Consensus analysis of signal peptide peptidase and homologous human aspartic proteases reveals opposite topology of catalytic domains compared with presenilins. J Biol Chem 2004;279:50790–50798. 14 Nyborg AC, Kornilova AY, Jansen K, Ladd TB, Wolfe MS, Golde TE: Signal peptide peptidase forms a homodimer that is labeled by an active site-directed gamma-secretase inhibitor. J Biol Chem 2004; 279:15153–15160.

Cellular Functions of -Secretase-Related Proteins

15 Krawitz P, Haffner C, Fluhrer R, Steiner H, Schmid B, Haass C: Differential localization and identification of a critical aspartate suggest non-redundant proteolytic functions of the presenilin homologues SPPL2b and SPPL3. J Biol Chem 2005;280:39515–39523. 16 Lemberg MK, Bland FA, Weihofen A, Braud VM, Martoglio B: Intramembrane proteolysis of signal peptides: an essential step in the generation of HLA-E epitopes. J Immunol 2001;167:6441–6446. 17 Grigorenko AP, Moliaka YK, Soto MC, Mello CC, Rogaev EI: The Caenorhabditis elegans IMPAS gene, imp-2, is essential for development and is functionally distinct from related presenilins. Proc Natl Acad Sci USA 2004;101:14955–14960. 18 Yu G, Nishimura M, Arawaka S, Levitan D, Zhang L, Tandon A, Song YQ, Rogaeva E, Chen F, Kawarai T, Supala A, Levesque L, Yu H, Yang DS, Holmes E, Milman P, Liang Y, Zhang DM, Xu DH, Sato C, Rogaev E, Smith M, Janus C, Zhang Y, Aebersold R, Farrer LS, Sorbi S, Bruni A, Fraser P, St George-Hyslop P: Nicastrin modulates presenilin-mediated notch/glp-1 signal transduction and betaAPP processing. Nature 2000; 407:48–54. 19 Goutte C, Hepler W, Mickey KM, Priess JR: aph-2 encodes a novel extracellular protein required for GLP-1-mediated signaling. Development 2000;127:2481–2492.

20 Edbauer D, Winkler E, Haass C, Steiner H: Presenilin and nicastrin regulate each other and determine amyloid beta-peptide production via complex formation. Proc Natl Acad Sci USA 2002;99:8666–8671. 21 Herreman A, Van Gassen G, Bentahir M, Nyabi O, Craessaerts K, Mueller U, Annaert W, De Strooper B: Gamma-secretase activity requires the presenilin-dependent trafficking of nicastrin through the Golgi apparatus but not its complex glycosylation. J Cell Sci 2003;116:1127–1136. 22 Shirotani K, Edbauer D, Capell A, Schmitz J, Steiner H, Haass C: Gamma-secretase activity is associated with a conformational change of nicastrin. J Biol Chem 2003; 278: 16474–16477. 23 Fergani A, Yu G, St George-Hyslop P, Checler F: Wild-type and mutated nicastrins do not display aminopeptidase M- and B-like activities. Biochem Biophys Res Commun 2001;289:678–680. 24 Haffner C, Frauli M, Topp S, Irmler M, Hofmann K, Regula JT, Bally-Cuif L, Haass C: Nicalin and its binding partner Nomo are novel Nodal signaling antagonists. EMBO J 2004;23:3041–3050. 25 Schier AF: Nodal signaling in vertebrate development. Annu Rev Cell Dev Biol 2003;19: 589–621. 26 Shah S, Lee SF, Tabuchi K, Hao YH, Yu C, Laplant Q, Ball H, Dann CE, 3rd, Sudhof T, Yu G: Nicastrin functions as a gamma-secretase-substrate receptor. Cell 2005; 122: 435– 447.

Neurodegenerative Dis 2006;3:284–289

289

Original Paper Neurodegenerative Dis 2006;3:290–297 DOI: 10.1159/000095269

Diseases

Modulators and Inhibitors of - and -Secretases Boris Schmidt Stefanie Baumann Rajeshwar Narlawar Hannes A. Braun Gregor Larbig Clemens Schöpf Institute for Organic Chemistry and Biochemistry, TU Darmstadt, Darmstadt, Germany

Key Words Alzheimer’s disease Secretase Aspartic protease BACE inhibitors Presenilin

Abstract Most gene mutations associated with Alzheimer’s disease point to the metabolism of amyloid precursor protein as a potential cause. The - and -secretases are two executioners of amyloid precursor protein processing resulting in amyloid-. Significant progress has been made in the selective inhibition of both proteases, regardless of structural information for -secretase. Several peptidic and nonpeptidic leads were identified for both targets. Copyright © 2006 S. Karger AG, Basel

which differ in length from 38 to 42 amino acids, are generated from the amyloid precursor protein (APP) by two aspartic proteases: -secretase and -secretase (fig. 1). Both secretases are rather promiscuous, they have multiple substrates and cause several distinctly different cleavages of APP. The membrane localization of both enzymes is crucial for selectivity, as cell-free conditions shift the cleavage pattern or result in additional cleavage sites. Usually 90% of APP is degraded by the benign secretase pathway, and a mere 10% of APP is degraded by the consecutive cleavages of - and -secretases to result in the build-up of extracellular A deposits. Neither the pathological consequences of deposited or soluble A are established beyond doubt, nor does plaque formation adequately correlate to the progress of AD. A definite proof of the A hypothesis is still missing for humans.

Introduction BACE Inhibitors

Alzheimer’s disease (AD) is the most common progressive, irreversible dementia with neither definitely assigned cause nor an available causal therapy. The symptoms of the disease include memory loss, confusion, impaired judgment, personality changes, disorientation, and loss of language skills [1]. A hallmark of AD is the accumulation of extracellular amyloidic plaques in the brain. The -amyloid (A) peptide, which is the major constituent of these amyloid plaques, performs a central role in the neuropathology of AD. The A peptides, © 2006 S. Karger AG, Basel 1660–2854/06/0035–0290$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

Several reviews on BACE inhibition summarize the biology and chemical concepts [2–6]. The majority of potent inhibitors are still peptide-based transition state analogues. Hydroxyethylenes, statins, norstatins, bisstatins, hydroxyethylamines and hydroxyethylureas were employed. The hydroxyethylenes delivered the first highly potent inhibitors. Their subsite specificity was revealed by the cleavage rates of substrate mixtures and selective inhibitors, resulting in the design of the heptapeptides

Boris Schmidt Clemens Schöpf Institute for Organic Chemistry and Biochemistry TU Darmstadt, Petersenstrasse 22 DE–64287 Darmstadt (Germany) Tel. +49 6151 163 075, Fax +49 6151 163 278, E-Mail [email protected]

Amyloid plaques

sAPP sAPP

P3 A40/42

-Secretase

-Secretase C99

APP

C83

Fig. 1. Processing of APP by secretases.

Glu-Val-Asn-(Leu-Ala)-Ala-Glu-Phe (1, OM99-2, Ki = 1.6 nM, fig. 2) and Glu-Leu-Asp-(Leu-Ala)-Val-GluPhe (2, OM00-3, Ki = 0.31 nM) [7]. The two inhibitors allowed co-crystallization with BACE and structure determination at 1.9 Å and 2.1 Å, respectively (PDB: OM99-2, 1FKN; OM00-3, 1M4H) [8]. The hydroxyethylenes are coordinated by four hydrogen bonds to the two catalytic aspartates. Essentially, the structure of the enzyme and the backbone conformations from P3 to P2 are the same, although they differ in side-chain orientation in several subsites. The introduction of the isophthalamide was an important step towards less peptidic compounds 3–6 (fig. 2), which is mandatory to obtain sufficient oral absorption and blood-brain barrier penetration [9]. We adopted this moiety for our BACE inhibitor program utilizing our novel methodology for iodomethanols [10]. This resulted in the stereoselective synthesis of an advanced intermediate 7. Opening of the epoxide with benzylamines furnished a series of hydroxyethylene isosters (fig. 3), which were tested in collaboration with M. Willem, LMU München, and F. Hoffmann-La Roche, Basel. Almost all compounds displayed poor activity; we attribute this to the wrong S-stereochemistry of the hydroxyl group. However, the epoxide 7 turned out to be an irreversible inhibitor of BACE; the irreversibility is apparent from the time dependence of the inhibition. Unfortunately, the compound displays poor activity in cellular assays; this

may be due to rapid degradation by other proteases, which in turn is an indicator of lacking selectivity [10]. We currently revise our synthesis to obtain the necessary diastereomers. Despite all efforts in the development of BACE-1 inhibitors, two major hurdles have hampered progress: blood-brain barrier permeability and oral bioavailability. To overcome these problems, novel nonpeptidic lead structures are of great interest. For several years, there were few structures, usually with an obscure mode of action [11, 12]. However, detailed activities were reported by A. Simon for 12 at the Alzheimer/Parkinson conference in Sorrento, March 2005 (IC50 = 15 nM BACE-1, IC50 = 230 nM BACE-2, IC50 = 7,620 nM cathepsin D, T1/2 2.1 h, clearance 76 ml/min/kg, Vdiss 7.2 l/kg). Furthermore, the compound did not pass the blood-brain barrier in mice. A proof of concept was attempted via intracerebroventricular dosage (7.5 mg/kg/day) for 14 days. A40 was reduced by 47% at the end of the trial [13]. A compound with improved properties was reported recently [14]. Acylated tetronic and tetramic acids have been investigated as aspartic protease inhibitors before [15]. This was due to their similarity to Tipranavir, an active site inhibitor of the HIV-1 aspartic protease. Co-crystallization with the HIV-1 protease and structure determination revealed that the acidic hydroxyl of Tipranavir interacts with the catalytic aspartates [16]. The more compact

Modulators and Inhibitors of - and -Secretases

Neurodegenerative Dis 2006;3:290–297

291

Fig. 2. BACE inhibitors I.

tetronates and tetramic acids may adopt a similar orientation in the active site, placing their substituents into lipophilic pockets. At the time we started the synthetic program on tetronic and tetramic acids, researchers from F. Hoffmann-La Roche identified a broad series of tetronic and tetramic acids with BACE inhibitor activity (13, IC50 = 11 M) [17]. We explored several synthetic strategies in solution and on polymeric supports. A cyclization/ cleavage strategy allowed to introduce diversity and resulted in more than 70 derivatives. The compounds 14–17 292

Neurodegenerative Dis 2006;3:290–297

(fig. 4) were tested in collaboration with F. HoffmannLa Roche, Basel, and M. Willem at the LMU, München. However, the activities were moderate at best. The removal of the acyl substituent or the replacement by a sulfoxide reduced the activity (IC50 1 200 M) in FRET assays on isolated BACE-1 dramatically. The compounds were inactive in a radio ligand displacement assay against an active site-directed inhibitor. Furthermore, they displayed very poor inhibition in cellular assays, which points at a weak allosteric mode of action [18]. Schmidt/Baumann/Narlawar/Braun/ Larbig

Fig. 3. BACE inhibitors II.

-Secretase Inhibitors

Special features of the -secretase complex hinder crystallization and thus crystallographic analysis of the enzyme, which is a major obstacle for structure-based drug design. Furthermore, the information available on inhibitor-binding sites is still limited. Therefore, all selective, nonpeptidic -secretase inhibitors had to be provided by high throughput screening efforts. Peptidic PS1 inhibitors, like Merck’s L-685,458 (18, IC50 = 17 nM) (fig. 5), are potent inhibitors [19]. The all-lipophilic sequence with 3 phenylalanines was somewhat anticipated, as several studies had indicated the lipophilic binding pockets Modulators and Inhibitors of - and -Secretases

(P2, P1, P1, P2, even P4 and P7) in proximity to the cleavage site [20]. It was suggested that compound 18 acts as a direct transition-state analogue of the A 1–40 and 1–42 cleavage sites. Elan’s semipeptidic -secretase inhibitor DAPT (19, IC50 = 20 nM) was developed from an N-dichlorophenylalanine lead. Structure activity relationships studies revealed phenylglycine and difluorophenylacetic acid to be crucial for activity [21]. DAPT has demonstrated robust efficacy in vivo at relatively high doses. Several preclinical studies revealed in vivo toxicity, because DAPT affects the Notch pathway at higher levels (100- to 1,000-fold) [22]. We speculated in 2002 that DAPT may be binding in close contact to the aspartic acNeurodegenerative Dis 2006;3:290–297

293

Fig. 4. Nonpeptidic BACE inhibitors.

ids of the active site and developed a series of acid-labile DAPT analogues. The compounds were intended to result in H+ catalyzed fragmentation and reactive cationic intermediates. Some of the DAPT analogues were indeed potent inhibitors, but none of the compounds displayed the predicted time-dependent inhibition in the assays of C. Haass and H. Steiner, LMU München. We concluded that there is no irreversible inhibition by these pH sensors [23]. The difluorophenylacetyl moiety in DAPT can be replaced by 5-bromopyridin-3-ylacetyl without loss of activity. Methylation of the pyridine results in membrane-blocked DAPT analogues; these were evaluated in the reconstituted -secretase assay (C. Haass, H. Steiner, LMU München) and cellular assays (K. Baumann, M. Brockhaus, F. Hoffmann-La Roche, Basel). All compounds displayed activity in the reconstituted assay, yet at a varying degree. The quaternized, membrane-blocked compounds lacked activity in the cellular assay. The analysis of these compounds is ongoing. In the meantime, we 294

Neurodegenerative Dis 2006;3:290–297

moved on to immobilize DAPT analogue esters on affigel (Biorad) by 3 different linkers of varying length. One of the linkers included a photolabile nitrobenzyl ester to allow a mild cleavage without denaturation of the assembled complex. However, all affinity gels displayed high and unspecific background binding, the isolation of active -secretase by these gels was unsuccessful. The -secretase inhibitors must reduce A production sufficiently to alleviate the cause of AD, but must not totally abolish either its production or the processing of other proteins, which have important roles in neuronal structure and function. Several substrates must be considered in addition to APP and Notch: the Notch ligands Delta and Jagged, apoER2 lipoprotein receptor, the lowdensity lipoprotein receptor-related protein, ErbB4 receptor tyrosine kinase, CD44, p75 neurotrophin and subunits of voltage-gated sodium channels. The important issue is: are there -secretase inhibitors that reduce APP processing without generating an unacceptable side Schmidt/Baumann/Narlawar/Braun/ Larbig

Fig. 5. Inhibitors and modulators of -secretase.

effect? Gleevec (21) inhibits A production but not Notch cleavage [24]. IC50 values for A40, A42 and AICD were recently reported to be 75 M. The generation of NICDFlag was not inhibited, even at 110-fold concentrations. Selected nonsteroidal anti-inflammatory drugs (NSAIDs) reduce A production without affecting alternative cleavages [25, 26]. The mechanisms of different -secretase inhibitors were explored in detail [27, 28]. Most of these bind directly to the active site or alter it through an allosteric interaction. Torrey Pines Pharmaceuticals disclosed a large number of aminothiazol-derivatives (23) with A42/A40-lowering activity at a concentration of about 30 M [29]. Approximately 60 of these

structures were claimed to display modulation of -secretase (activity ! 0.2 M). We decided to synthesize compound 24 and to evaluate it in collaboration with C. Haass and Hoffman-La Roche, Basel. The substance did not display modulation but inhibition: EC50 (A38) = 1.5 M, IC50 (A40) = 1.8 M, IC50 (A42) = 1.6 M. A selective secretase modulator (25) was reported by Merck Sharp & Dohme [30]; the carboxylic acid seems to be relevant for the desired ratio of A38/A40/A42. This modulation is distinctly different from inhibition as the total A load may be unaffected. This was observed for several NSAIDs and is unrelated to COX1 inhibition [25]. Some COX1 inhibitors are suitable candidates to improve their ini-

Modulators and Inhibitors of - and -Secretases

Neurodegenerative Dis 2006;3:290–297

295

Fig. 6. NSAID-derived -secretase modulators.

tially weak activity. The analogy of BMS-299897 to COX1 inhibitors inspired us to explore derivatives of commercial NSAIDs: sulindac, flurbiprofen, ibuprofen, indomethacin, diclofenac, naproxen, carprofen, ketoprofen, and diflunisal. We prepared more than 150 derivatives, initially esters and amides of the parent carboxylic acids and identified several full -secretase inhibitors. The veratryl amides obtained from sulindac and diclofenac caused reduction of A38, A40 and A42 in cellular assays (IC50 10–50 M). However, some amides and esters displayed ‘inverse’ NSAID properties: A38 levels decreased, whereas A40 and A42 levels increased in cellular assays. -Secretase modulation was observed as the third mode of action: A38 increased, whereas A40 and A42 decreased in cellular assays. A structure-activity relationship is already apparent: the carboxylic acid is strictly required, and a lipophilic substituent branching out from the core structure improves the activity 10- to 100-fold. The scaffold can either derive from carprofen or carbazole. The lipophilic branch can be attached by alkylation or by sulfonylation of the aniline. The best cores have been attached to biotinylated, photoreactive linkers to result in 5 biotinylated compounds (just one example is provided in figure 6). However, these compounds turned out to be unsuitable baits for the -secre296

Neurodegenerative Dis 2006;3:290–297

tase complex, although some displayed significant modulation of -secretase activity. They resulted in unspecific binding or no detectable binding at all. A robust bait for the successful pull-down of the binding domain within -secretase is still unknown. Outlook

The availability of peptidic and peptidomimetic inhibitors for - and -secretase inhibitors, both as tool and lead structures, made a huge impact on the research area. But these potent peptidomimetics come with costly price tags: oral availability, cost of goods and blood-brain barrier penetration impose severe obstacles on drug development. Despite the extremely rapid progress in the field, there are no reports of brain penetrating secretase inhibitors in phase II or III (02/2006). The only -secretase inhibitor in phase I has a red flag associated with it. This is due to its impact on the Notch pathway. The selective modulation of -secretase by NSAIDs is pointing in the right direction: allosteric modulation of the active site, which can be identified by additional cleavage sites of presenilin-1.

Schmidt/Baumann/Narlawar/Braun/ Larbig

Collaborations within the German Research Foundation Priority Program 1085 A ligand-based approach to Tau aggregation inhibitors was conducted in collaboration with the participant E. Mandelkow, Hamburg. A manuscript was submitted (30/01/06). -Secretase inhibitors, -secretase inhibitors and modulators, 20S proteasome inhibitors and targeted screening collections were supplied to R. Baumeister, C. Haass, T. Hartmann, U. Müller, G. Multhaup , J. Walter, S. Weggen, M. Willem. This resulted in 6 publications [2, 3, 10, 31–33] and 1 patent application (not yet published).

Acknowledgements The authors thank the DFG (SPP1085 SCHM1012-3-1/2) and the EU (contract LSHM-CT-2003-503330; APOPIS) for support. We thank K. Baumann, M. Brockhaus and C. Czech (all at F. Hoffmann-La Roche, Basel) for biological assays, valuable input and long discussions.

References 1 WHO: The World Health Report 2001 Mental Health: New Understanding, New Hope. Geneva, WHO, 2001. 2 Schmidt B: Aspartic proteases involved in Alzheimer’s disease. Chembiochem 2003; 4: 366–378. 3 Schmidt B, Baumann S, Braun HA, Larbig G: Inhibitors and modulators of - and -secretase. Curr Top Med Chem 2006;6:377–392. 4 Schmidt B, Braun HA, Narlawar R: Drug development and PET-diagnostics for Alzheimer’s disease. Curr Med Chem 2005; 12: 1677–1695. 5 Roggo S: Inhibition of BACE, a promising approach to Alzheimer’s disease therapy. Curr Top Med Chem 2002;2:359–370. 6 John V, Beck JP, Bienkowski MJ, Sinha S, Heinrikson RL: Human -secretase (BACE) and BACE inhibitors. J Med Chem 2003; 46: 4625–4630. 7 Ghosh AK, Hong L, Tang J: -Secretase as a therapeutic target for inhibitor drugs. Curr Med Chem 2002;9:1135–1144. 8 Hong L, Turner RT III, Koelsch G, Shin D, Ghosh AK, Tang J: Crystal structure of memapsin 2 (-secretase) in complex with an inhibitor OM00-3. Biochemistry 2002; 41: 10963–10967. 9 Maillaird M, Hom C, Gailunas A: Preparation of substituted amines to treat Alzheimer’s disease. WO 200202512, 2002. 10 Braun HA, Meusinger R, Schmidt B: 2-Iodoethanols from aldehydes, diiodomethane and isopropylmagnesium chloride. Tet Lett 2005;46:2551–2554. 11 Miyamoto M, Matsui J, Fukumoto H, Tarui N: Preparation of 2-[2-amino- or 2-(N-heterocyclyl)ethyl]-6-(4-biphenylyl-methoxy)tetralin derivatives as -secretase inhibitors. WO 0187293, 2001. 12 Ramakrishna NVS, Kumar EKSV, Kulkarni AS: Screening of natural products for new leads as inhibitors of b-amyloid production: Latifolin from Dalbergia sissoo. Indian J Chem 2001;40B:539–540. 13 Barrow JC, Coburn CA, Nantermet PG: Preparation of phenylamides and pyridylamides as 1-secretase inhibitors. WO2005065195, 2005.

Modulators and Inhibitors of - and -Secretases

14 Stachel SJ, Coburn CA, Steele TG, et al: Conformationally biased P3 amide replacements of beta-secretase inhibitors. Bioorg Med Chem Lett 2006;16:641–644. 15 Yehia NA, Antuch W, Beck B, et al: Novel nonpeptidic inhibitors of HIV-1 protease obtained via a new multicomponent chemistry strategy. Bioorg Med Chem Lett 2004; 14: 3121–3125. 16 Thaisrivongs S, Strohbach JW: Structurebased discovery of tipranavir disodium (PNU-140690E): a potent, orally bioavailable, nonpeptidic HIV protease inhibitor. Biopolymers 1999; 51:51–58. 17 Godel T, Hilpert H, Humm R: Preparation of tetronic and tetramic acids as beta-secretase inhibitors. WO 2005119329, 2005. 18 Larbig G, Schmidt B: Synthesis of tetramic and tetronic acid as beta-secretase inhibitors. J Comb Chem 2006;8:480–490. 19 Shearman MS, Beher D, Clarke EE, et al: L-685,458, an aspartyl protease transition state mimic, is a potent inhibitor of amyloid beta-protein precursor gamma-secretase activity. Biochemistry 2000;39:8698–8704. 20 Lichtenthaler SF, Wang R, Grimm H, Uljon SU, Masters CL, Beyreuther K: Mechanism of the cleavage specificity of Alzheimer’s diseases -secretase identified by phenylalanine-scanning mutagenesis of the transmembrane domain of the amyloid precursor protein. PNAS 1999;96:3053–3058. 21 Dovey HF, John V, Anderson JP, et al: Functional -secretase inhibitors reduce b-amyloid peptide levels in brain. J Neurochem 2001;76:173–181. 22 Geling A, Steiner H, Willem M, Bally-Cuif L, Haass C: A -secretase inhibitor blocks Notch signaling in vivo and causes a severe neurogenic phenotype in zebrafish. EMBO Rep 2002;3:688–694.

23 Larbig G, Zall A, Schmidt B: Inhibitors designed for presenilin 1 utilizing by means of aspartic acid activation. Helv Chim Acta 2004;87:2334–2340. 24 Netzer WJ, Dou F, Cai D, et al: Gleevec inhibits beta-amyloid production but not Notch cleavage. PNAS 2003; 100: 12444– 12449. 25 Weggen S, Eriksen JL, Das P, et al: A subset of NSAIDs lower amyloidogenic A42 independently of cyclooxygenase activity. Nature 2001;414:212–216. 26 Lanz TA, Fici GJ, Merchant KM: Lack of specific amyloid-beta(1–42) suppression by nonsteroidal anti-inflammatory drugs in young, plaque-free Tg2576 mice and in guinea pig neuronal cultures. J Pharmacol Exp Ther 2005;312:399–406. 27 Beher D, Clarke EE, Wrigley JD, et al: Selected non-steroidal anti-inflammatory drugs and their derivatives target gamma-secretase at a novel site. Evidence for an allosteric mechanism. J Biol Chem 2004; 279: 43419– 43426. 28 Kornilova AY, Das C, Wolfe MS: Differential effects of inhibitors on the -secretase complex. J Biol Chem 2003; 278:16470–16473. 29 Cheng S, Comer DD, Mao L, Balow GP, Pleynet D: Aryl compounds and uses in modulating amyloid . WO 2004110350, 2004. 30 Beher D, Bettati M, Checksfield GD: Preparation of tetrahydrocarbazole-1-alkanoic acids for the treatment of Alzheimer’s disease and related conditions. WO 2005013985, 2005. 31 Schmidt B, Ehlert DK, Braun HA: E-1,2-dichlorovinyl ethers as irreversible protease inhibitors. Tet Lett 2004;45:1751–1753. 32 Braun HA, Umbreen S, Groll M, et al: Tripeptide mimetics inhibit the 20 S proteasome by covalent bonding to the active threonines. J Biol Chem 2005; 280:28394–28401. 33 Schmidt B, Siegler A: Aspartic proteases involved in Alzheimer’s disease; in Schmuck C, Wennemers H (eds): Highlights in Bioorganic Chemistry. Weinheim, Wiley-VCH, 2004, pp 262–276.

Neurodegenerative Dis 2006;3:290–297

297

Original Paper

Diseases

Neurodegenerative Dis 2006;3:298–304 DOI: 10.1159/000095270

-Secretase Modulation with A42-Lowering Nonsteroidal Anti-Inflammatory Drugs and Derived Compounds Eva Czirr Sascha Weggen Emmy Noether Research Group, Institute of Physiological Chemistry and Pathobiochemistry, Johannes Gutenberg University Mainz, Mainz, Germany

Key Words Alzheimer’s disease Amyloid- peptide Nonsteroidal anti-inflammatory drugs Ibuprofen -Secretase

Abstract The amyloid- (A) peptides and specifically the highly amyloidogenic isoform A42 appear to be key agents in the pathogenesis of familial and sporadic forms of Alzheimer’s disease (AD). The final step in the generation of A from the amyloid precursor protein is catalyzed by the multiprotein complex -secretase, which constitutes a prime drug target for prevention and therapy of the disease. However, highly potent -secretase inhibitors that block formation of all A peptides have provoked troubling side effects in preclinical animal models of AD. This toxicity can be readily explained by the promiscuous substrate specificity of -secretase and its essential role in the NOTCH signaling pathway. For that reason and because of the crucial role of A42 in the pathogenesis of the disease, selective inhibition of A42 production would seem to be a more promising alternative to complete inhibition of -secretase activity. This theoretical concept has edged much closer to clinical reality with the surprising finding that certain nonsteroidal anti-inflammatory drugs (NSAIDs), including ibuprofen, and derived compounds display preferential A42-lowering activity. In contrast to -secretase inhibitors, these -secretase modulators effectively suppress A42 production while sparing processing of NOTCH and other -secretase substrates. Although

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0298$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

not fully resolved on the molecular level, the mechanism of action of A42-lowering NSAIDs is independent of cyclooxygenase inhibition and most likely involves direct interaction with components of the -secretase complex or its substrates. Current efforts to improve the pharmacological shortcomings of available -secretase modulators will hopefully lead to the development of clinically useful A42-lowering compounds in the near future. Copyright © 2006 S. Karger AG, Basel

Introduction: A42 as a Therapeutic Target in Alzheimer’s Disease

Alzheimer’s disease (AD) is the most common agerelated neurodegenerative disorder. In 2000, the number of prevalent AD cases in the entire European region was 4.7 million, and projections are that this number will surge to about 11.2 million cases in the year 2050 [1], a trend that poses a dramatic challenge to the public health care system. The neuropathology of AD is primarily distinguished by neuronal loss, neurofibrillary tangle formation and the extracellular deposition of amyloid- (A) plaques, and conclusive evidence indicates that aberrant production and deposition of A plays a causal role in the pathogenesis of the disease [2]. A is a proteolytic fragment of the amyloid precursor protein (APP), a ubiquitously expressed type I transmembrane protein. To produce A, Sascha Weggen Institute of Physiological Chemistry and Pathobiochemistry Johannes Gutenberg University Mainz DE–55128 Mainz (Germany) Tel. +49 6131 392 6720, Fax +49 6131 392 6755, E-Mail [email protected]

APP molecules are sequentially cleaved by two aspartyl proteases, first by -secretase (BACE) which generates the amino-terminus (N-terminus) of A, and then by -secretase which cleaves APP approximately in the middle of the transmembrane domain (TMD) to generate the carboxy-terminus (C-terminus) of the A peptide. -Secretase does not display strict sequence specificity and generates A peptides of variable length at the C-terminus, with peptides ending after 40 and 42 amino acids being the most prevalent species [3]. In addition to cleavage in the middle of the TMD (-cleavage), -secretase conducts another cleavage event close to the cytosolic border of the membrane (-cleavage), liberating the APP intracellular domain (AICD). Similar intracellular domains (ICDs) with potential signaling functions are generated from other substrates of -secretase. At least three major observations imply that the longer A42 peptide is the crucial pathogenic species in AD and constitutes an excellent therapeutic target. First, mutations in the APP and presenilin (PS) genes, which are associated with early-onset familial AD (FAD), invariably increase production of A42 in the plasma of mutation carriers, in transfected cells and in transgenic animals. Second, although A40 is predominantly produced, A42 is the species initially deposited in the AD brain. Third, A42 is exceptionally prone to aggregation and precipitation, and aggregated A peptides are toxic to cells in vitro and in vivo. Remarkably, although many APP and PS mutations increase A42 production by only 30–100%, they result in disease onset 20–40 years prior to the development of sporadic late-onset AD. Pharmacological suppression of A42 production may therefore postpone onset of sporadic AD by an equivalent amount of time, and prevent development of AD in most individuals [2].

Considerable advances have been made to target A production for treatment or prevention of AD. Since A peptides are produced by sequential cleavage of APP by - and -secretase, the most straightforward approach for pharmacological intervention appears to be the use of small molecule inhibitors of these proteases [4]. The development of -secretase inhibitors is still in early stages, whereas -secretase inhibitors have already been thoroughly investigated in preclinical AD models. -Secretase is a multiprotein complex that is assembled from at

least four obligatory proteins, PS, nicastrin, anterior pharynx defective-1 (APH-1) and presenilin enhancer-2 (PEN-2), and the PS proteins seem to contain the active site of this enzymatic activity [5]. Highly potent -secretase inhibitors block A production in cultured cells with IC50 values in the picomolar range [4]. However, animal studies with -secretase inhibitors have also uncovered mechanism-based toxicity, which can be largely explained by the essential role of -secretase in proteolytic processing of the NOTCH receptor. Similarly to APP, the NOTCH receptors undergo intramembrane cleavage by -secretase, releasing the Notch intracellular domain (NICD) to regulate transcription of target genes involved in cell fate decisions during embryogenesis but also in mitotic cell populations of adult mammals including lymphocytes and intestinal epithelial cells. Almost all published secretase inhibitors indiscriminately block cleavages within the TMDs of -secretase substrates and prevent A and NICD formation with the same potency. Unsurprisingly then, subchronic -secretase treatment in APPtransgenic mice at doses that effectively reduced A levels in plasma and cerebrospinal fluid further led to severe hematopoietic phenotypes and gastrointestinal toxicity [4]. These findings raise the possibility of prohibitive side effects in human clinical trials and indicate that, at a minimum, clinical dosing of -secretase inhibitors will need to be adjusted to allow residual NOTCH signaling in the periphery. As A42 seems to be the key pathogenic agent that initiates the neurodegenerative cascade, another promising way to prevent or treat AD could be selective targeting of A42 production instead of complete -secretase inhibition. However, until recently it remained unclear whether pharmacological suppression of A42 production with small molecules is feasible. The existence of such molecules was proven by the unexpected discovery that some nonsteroidal anti-inflammatory drugs (NSAIDs), including ibuprofen, selectively lowered A42 production in cell-based assays and in an APP-transgenic mouse model of AD [6]. NSAIDs have been discussed as a potential treatment option for AD for at least two decades. The amyloid pathology in the AD brain is accompanied by an inflammatory response with activation of astrocytes and microglia, complement activation and local upregulation of inflammatory makers [7], and it has been generally assumed that this A-induced chronic inflammation contributes to the neurodegeneration in AD. Strong support for a preventive effect of anti-inflammatory drugs is drawn from epidemiological studies, which have shown that chronic use of NSAIDs is associated with

-Secretase Modulation with NSAIDs

Neurodegenerative Dis 2006;3:298–304

-Secretase Inhibition versus -Secretase Modulation

299

a lower risk of developing AD [8]. Aspirin and NSAIDs achieve their principal therapeutic effects by blocking cyclooxygenase (COX)-mediated synthesis of inflammatory prostaglandins. Consequently, the proposition seems reasonable that NSAIDs are effective in AD through their anti-inflammatory properties. However, the finding that certain NSAIDs selectively lower A42 challenges this idea and provides an attractive alternative explanation for their protective effects in AD [6]. In contrast to secretase inhibitors, these A42-lowering NSAIDs and derived compounds can be termed -secretase modulators as they do not impair processing of NOTCH and other -secretase substrates within a certain range of concentrations. A42-Lowering Compounds: In vitro Studies and Mechanism of Action

The initial study in 2001 reported three NSAIDs with selective A42-lowering activity in a variety of permanent cell lines [6]. Two structurally related compounds, sulindac sulfide and indomethacin, lowered A42 with an IC50 value of 25–50 M whereas the third compound, ibuprofen, lowered A42 with an IC50 of around 250 M (fig. 1). At maximal nontoxic concentrations, 70–80% A42 inhibition was observed without significant reduction of A40 levels. Importantly, this activity was not associated with all NSAIDs and other commonly prescribed NSAIDs like naproxen and aspirin did not change either A42 or A40 levels [6]. A second screen of all FDA-approved NSAIDs identified few other NSAIDs with A42lowering activity, i.e. fenoprofen, flurbiprofen (fig. 1), meclofenamic acid [9]. Subsequently, other laboratories corroborated these results with cell lines of peripheral and neuronal origin [10–14]. Intriguingly, some COX-2specific inhibitors, including celecoxib (fig. 1), but also unrelated compounds such as the peroxisome proliferator-activated receptor- (PPAR) antagonist fenofibrate were shown to selectively increase A42 levels [15, 16]. It was further demonstrated that A42-lowering compounds like indomethacin can be converted into A42raising substances by derivatization of their carboxylic acid function [11, 15]. This revealed that small molecules could modulate A42 production in both directions. Most importantly from a clinical perspective, cellbased studies with A42-lowering compounds have demonstrated that -secretase modulators do not provoke the same molecular side effects as -secretase inhibitors in regard to APP processing and effects on other substrates 300

Neurodegenerative Dis 2006;3:298–304

Fig. 1. Chemical structures of A42-modulating compounds.

of -secretase [6, 10, 12, 16–18]. Assays with APP-transfected cell lines showed that A42-lowering NSAIDs did not change APP expression, turnover, internalization or release of the APP ectodomain, and, in marked contrast to conventional -secretase inhibitors, did not cause accumulation of APP C-terminal fragments [6, 10, 16]. Furthermore, treatment with sulindac sulfide, indomethacin and ibuprofen did not impair -secretase-mediated NOTCH receptor cleavage and NICD formation [6, 12, 16], and additional studies confirmed that generation of the ICDs of APP and the ErbB-4 receptor was likewise not affected [18]. This consistent lack of effect on generation of the ICDs of several -secretase substrates indicates that -secretase modulators and improved compounds following an analogous mechanism of action may avoid the toxicity that was observed with nonselective -secretase inhibitors. Czirr/Weggen

Fig. 2. Molecular mechanism of A42-lowering compounds. The primary pharmacological targets of A42-lowering NSAIDs are COX enzymes, which control the synthesis of inflammatory prostaglandins. Several non-COX pathways are modulated by A42lowering NSAIDs, including the NFB pathway, which regulates genes involved in the inflammatory response, PPARs, which control genes involved in lipid metabolism and LOX, which produce inflammatory leukotrienes. A42-lowering NSAIDs have also been proposed to affect ROCK activity, which mediates cytoskeletal rearrangements. COX and these known non-COX targets have been ruled out as potential mediators of the A42-lowering activity. In contrast, A42-modulating compounds seem to act by direct interaction with the -secretase complex or its substrates. A42-lowering NSAIDs induce a subtle shift in -secretase activity, which manifests as a selective reduction in A42 production and a concomitant increase in the production of shorter A species such as A38.

-Secretase Modulation with NSAIDs

Substantial progress has been made to determine the mechanism of action of A42-lowering compounds (fig. 2). Since COX enzymes are the main pharmacological target of NSAIDs, inhibition of COX and suppression of prostaglandin synthesis seemed to be an obvious mechanism by which NSAIDs could reduce A42 secretion. However, several arguments rule out any involvement of COX in the A42-lowering activity: (1) only few NSAIDs display A42-lowering activity, whereas all NSAIDs by definition inhibit COX [6]; (2) the A42-lowering activity of sulindac sulfide was not impaired in COX-1/-2-deficient cells [6]; (3) NSAID derivatives have been reported that lower A42 but lack COX inhibitory activity [9–11]. Similar arguments dismiss any role for COX in the molecular mechanism of A42-raising compounds [15]. Several COX-independent mechanisms of NSAIDs are well established, including nuclear factor B (NFB) activation, inhibition or activation of lipoxygenases (LOX) and modulation of PPAR signaling [19] (fig. 2). Although not as definitive as for COX itself, available results suggest that a mechanistic involvement of these non-COX targets in the A42-lowering activity is highly unlikely [10, 20]. More recently, Y-27632, an inhibitor of Rho-kinase (ROCK) was shown to reduce A42 levels in vitro and in vivo in an indistinguishable fashion to A42-lowering NSAIDs, suggesting that ROCK inhibition is a critical downstream event of NSAID treatment. Unexpectedly, we have not been able to confirm these findings and found instead that ROCK inhibitors reduced the overall production of A but showed no selectivity for A42 [15, 21]. These data clearly indicate that ROCK is not mechanistically linked to the A42-lowering activity of NSAIDs [21]. In contrast, it now seems likely that A42-lowering compounds act by direct modulation of -secretase activity (fig. 2). A first hint for such a mechanism was provided by mass spectrometry analysis which had shown that the A42-lowering NSAID sulindac sulfide concomitantly increased A38 levels [6]. This suggested a rather subtle shift in the -secretase cleavage pattern with increased production of the shorter A38 species at the expense of the longer A42 species. Since then, the most convincing evidence for direct -secretase modulation was provided by the fact that A42-lowering compounds are active in cell-free -secretase assays [9, 12, 17, 22]. Compounds that exert activity in these assays are presumed to modulate A production either by direct interaction with the -secretase enzyme complex or the substrate APP, and this has been demonstrated for some secretase inhibitors, which were shown to directly bind to PS proteins [4]. A42-lowering NSAIDs inhibited Neurodegenerative Dis 2006;3:298–304

301

A42 in these assays in a noncompetitive manner with respect to substrate, but they were able to displace a transition-state -secretase inhibitor from the active center of the enzyme [17]. A42-raising compounds like celecoxib were also active in -secretase in vitro assays, indicating that they follow a similar but opposite mechanism of action [15]. -Secretase in vitro assays have further been utilized to investigate whether very high concentrations of A42-lowering compounds, which cause toxicity in cell-based assays, would eventually block overall A production and ICD formation from -secretase substrates, and this has added critical insights into the mode of action of these compounds [12, 17, 18]. As exemplified for sulindac sulfide, lower concentrations (20–100 M) of this compound selectively reduced A42 production in cell-free assays in a very similar fashion to what had been observed in cell-based assays [17, 22]. However, at higher concentrations (100 M to 1 mM), A40 and AICD production were also impaired, indicating that selectivity for A42 was lost and sulindac sulfide acted as a nonselective -secretase inhibitor [17]. These results suggested that A42-lowering compounds are not entirely specific for A42 but rather offer a 5- to 10-fold window of modulation where A42 production is selectively reduced without effects on overall A production or ICD generation. Over 150 mutations in the PS proteins, which cause the majority of FAD cases, have been described and they all appear to selectively increase production of A42 [2]. It has been proposed that all these mutations induce similar conformational changes in the -secretase complex, which in turn result in alterations in the pattern of the generated A peptides [3]. One credible explanation for the selective A42-lowering activity of NSAIDs is that these compounds change -secretase conformation in a similar yet opposite way to PS mutations. Studies using fluorescence lifetime imaging to measure the distance between two epitopes within PS1 have produced some evidence that PS mutations and A42-lowering NSAIDs may indeed promote such opposite conformational changes [23, 24]. Observations that certain FAD PS1 mutations change the cellular response to A42-lowering NSAIDs provide further indirect evidence for -secretase modulation by these compounds [22]. In summary, it now appears certain that A42-modulating compounds target the -secretase complex, but the molecular details are far from determined. One major issue is to identify the interaction partner of A42-modulating compounds within the -secretase complex. Although PS seems to be the primary candidate, interaction with any of the components in the -secretase complex 302

Neurodegenerative Dis 2006;3:298–304

or with the substrate APP is plausible. Even more enigmatic remains how the presumed conformational changes induced by A42-modulating compounds lead to the observed shift in cleavage specificity from A42 to A38. One potential explanation is that these compounds induce subtle changes in the presentation of the substrate APP to the catalytic center in the -secretase complex. Preclinical in vivo Studies and Clinical Trials with A42-Lowering Compounds

Several long-term and short-term treatment studies with A42-lowering NSAIDs in APP-transgenic mouse models of AD have been performed to date. Long-term treatment studies have explored whether NSAIDs can prevent or ameliorate amyloid plaque deposition and secondary pathologies mostly in the Tg2576 mouse model of AD. Even before the A42-lowering activity of certain NSAIDs was reported, a groundbreaking study by Lim et al. [25] had demonstrated that chronic treatment with high doses of ibuprofen for 6 months strongly reduced both amyloid pathology and inflammatory responses in Tg2576 mice. Devised as a prevention trial, the intervention was started before substantial amyloid deposition is seen in this mouse model, and the dose of ibuprofen (62.5 mg/kg/day) moderately exceeded the highest approved human dose. In the ibuprofen-treated animals, the total number and area of amyloid plaques was reduced by 50%, and soluble and insoluble A in the brain was reduced by 30–40%. Significant reductions in the number of plaque-associated activated microglia cells and in the levels of proinflammatory markers were also observed [25]. Two subsequent studies have provided important confirmation for the effectiveness of ibuprofen in the Tg2576 mouse model [14, 26]. Indomethacin was also shown to potently reduce amyloid pathology in the same model [27], whereas two NSAIDs without A42-lowering activity, celecoxib and nimesulide, did not cause significant changes [26, 27]. In addition, short-term treatment studies have demonstrated that A42-lowering NSAIDs can acutely lower A42 levels in brain of young, plaque-free APP-transgenic mice. In the initial report [6], 3-month-old Tg2576 mice were orally dosed for 3 days with 50 mg/kg/day of ibuprofen or naproxen. Treatment with ibuprofen resulted in a significant 39% decrease in SDS-soluble A42 without any changes in A40 levels, whereas naproxen had no effect. Further studies with a similar treatment protocol have shown in vivo A42-lowering activity for Czirr/Weggen

additional NSAIDs including sulindac sulfide, indomethacin and flurbiprofen [9]. In summary, both short- and long-term studies with A42-lowering NSAIDs seem to support their efficacy to prevent the development of amyloid pathology in APPtransgenic mice. However, a number of important issues require further investigation. First, clarification of the mechanism of action of A42-lowering NSAIDs in chronic treatment studies is necessary. Since ibuprofen and indomethacin are dual-mechanism drugs with A42-lowering activity and anti-inflammatory properties, it currently seems impossible to decide which of these two activities or if a synergistic mechanism is responsible for the therapeutic results. Second, only very limited data exist to suggest that A42-lowering NSAIDs may further be able to reverse behavioral alterations in the Tg2576 and other AD mouse models [28]. Third, the A42-lowering activity of ibuprofen was only apparent in cultured cells at high micromolar concentrations (250– 500 M) [6], whereas brain concentrations in Tg2576 mice stayed in the low micromolar range (up to 2 M) [9] even with high-dose ibuprofen treatment. Right now, there is no adequate explanation for this striking discrepancy between effective NSAID concentrations in vitro and in the brain. Finally, some groups have not been able to replicate the results from short-term treatment studies with A42-lowering NSAIDs [29, 30]. To resolve these inconsistencies, additional in vivo studies in Tg2576 and other AD mouse models would be desirable. Only few clinical trials with NSAIDs and AD patients have been conducted to date. In the earliest trial with the A42-lowering NSAID indomethacin, some slowing of cognitive decline was observed, but interpretation of this study was complicated by its small size and a high dropout rate [31]. Subsequent trials with NSAIDs lacking A42lowering activity like diclofenac, naproxen, celecoxib and rofecoxib were without significant benefits [32]. These negative results could be related to dose, duration of treatment, time point of intervention or choice of drug, but the available trial data are clearly too limited to exclude any of these possibilities. Serious side effects associated with long-term use of NSAIDs and inhibition of COX restrict the clinical use of NSAIDs, particularly in elderly AD patients, but A42-lowering NSAIDs remain candidate drugs for a treatment approach targeting A42 in AD. Another promising strategy may be the use of specific NSAID enantiomers without COX activity, which exert a better safety profile. A particular interesting compound is the (R)-enantiomer of flurbiprofen, which lacks COX activity but is equally potent in reducing A42 in vitro -Secretase Modulation with NSAIDs

and in vivo as the COX-inhibiting stereoisomer (S)-flurbiprofen [9, 10]. Based on these data, (R)-flurbiprofen is under development for treatment of AD, and a phase II clinical trial with a duration of 12 months has recently been completed [34]. Results showed that the drug was well tolerated, and positive trends were observed with the highest 800-mg twice-daily dose in patients with mild but not with moderate AD. A subgroup of patients with mild disease and high plasma drug levels showed significantly less decline in 2 out of 3 primary outcomes. Recruitment for a phase III clinical study is ongoing.

Future Directions

The A42 peptide is a key molecule in the pathogenesis of AD. The finding that certain NSAIDs selectively inhibit production of A42 in vitro and in vivo proved that pharmacological suppression of A42 generation with small molecules is possible. The clinical use of approved A42-lowering NSAIDs is limited by toxicity associated with inhibition of COX. However, ibuprofen with its relatively benign toxicity profile remains an obvious drug candidate [33]. Major pharmacological shortcomings of current A42-lowering compounds are low potency, low brain permeability, short plasma half-life and potent activity against COX, and efforts are ongoing to improve all of these features. Novel A42-lowering compounds have so far mainly been disclosed in patent applications. However, Peretto et al. [11] recently reported on improved derivatives of flurbiprofen such as 11c (fig. 1), and substantial improvements in potency against A42 and elimination of COX activity were achieved by simple and defined structural substitutions. The (R)-enantiomer of flurbiprofen, which retains A42-lowering activity but lacks COX activity, is currently being evaluated in a phase III clinical trial. A positive outcome could have immediate benefits for AD patients, and would further provide strong support for the use of A42-lowering compounds in treatment or prevention of AD.

Acknowledgement We thank Michael Plenikowski for preparing figure 2. Research by the authors is generously supported by the Emmy Noether program of the Deutsche Forschungsgemeinschaft (WE 2561/1-3).

Neurodegenerative Dis 2006;3:298–304

303

References 1 Wancata J, Musalek M, Alexandrowicz R, Krautgartner M: Number of dementia sufferers in Europe between the years 2000 and 2050. Eur Psychiatry 2003;18:306–313. 2 Golde TE: Alzheimer disease therapy: can the amyloid cascade be halted? J Clin Invest 2003;111:11–18. 3 Selkoe DJ: Alzheimer’s disease: Genes, proteins, and therapy. Physiol Rev 2001;81:741– 766. 4 Beher D, Graham SL: Protease inhibitors as potential disease-modifying therapeutics for Alzheimer’s disease. Expert Opin Investig Drugs 2005;14:1385–1409. 5 Haass C: Take five – BACE and the gammasecretase quartet conduct Alzheimer’s amyloid beta-peptide generation. EMBO J 2004; 23:483–488. 6 Weggen S, Eriksen JL, Das P, Sagi SA, Wang R, Pietrzik CU, Findlay KA, Smith TE, Murphy MP, Bulter T, Kang DE, Marquez-Sterling N, Golde TE, Koo EH: A subset of NSAIDs lower amyloidogenic Abeta42 independently of cyclooxygenase activity. Nature 2001;414:212–216. 7 Akiyama H, Barger S, Barnum S, Bradt B, Bauer J, Cole GM, Cooper NR, Eikelenboom P, Emmerling M, Fiebich BL, Finch CE, Frautschy S, Griffin WS, Hampel H, Hull M, Landreth G, Lue L, Mrak R, Mackenzie IR, McGeer PL, O’Banion MK, Pachter J, Pasinetti G, Plata-Salaman C, Rogers J, Rydel R, Shen Y, Streit W, Strohmeyer R, Tooyoma I, Van Muiswinkel FL, Veerhuis R, Walker D, Webster S, Wegrzyniak B, Wenk G, WyssCoray T: Inflammation and Alzheimer’s disease. Neurobiol Aging 2000;21:383–421. 8 Etminan M, Gill S, Samii A: Effect of nonsteroidal anti-inflammatory drugs on risk of Alzheimer’s disease: systematic review and meta-analysis of observational studies. BMJ 2003;327:128. 9 Eriksen JL, Sagi SA, Smith TE, Weggen S, Das P, McLendon DC, Ozols VV, Jessing KW, Zavitz KH, Koo EH, Golde TE: NSAIDs and enantiomers of flurbiprofen target gamma-secretase and lower Abeta 42 in vivo. J Clin Invest 2003;112:440–449. 10 Morihara T, Chu T, Ubeda O, Beech W, Cole GM: Selective inhibition of Abeta42 production by NSAID R-enantiomers. J Neurochem 2002;83:1009–1012. 11 Peretto I, Radaelli S, Parini C, Zandi M, Raveglia LF, Dondio G, Fontanella L, Misiano P, Bigogno C, Rizzi A, Riccardi B, Biscaioli M, Marchetti S, Puccini P, Catinella S, Rondelli I, Cenacchi V, Bolzoni PT, Caruso P, Villetti G, Facchinetti F, Del Giudice E, Moretto N, Imbimbo BP: Synthesis and biological activity of flurbiprofen analogues as selective inhibitors of beta-amyloid1–42 secretion. J Med Chem 2005;48:5705–5720. 12 Takahashi Y, Hayashi I, Tominari Y, Rikimaru K, Morohashi Y, Kan T, Natsugari H, Fukuyama T, Tomita T, Iwatsubo T: Sulindac

304

13

14

15

16

17

18

19

20

21

22

sulfide is a noncompetitive gamma-secretase inhibitor that preferentially reduces Abeta 42 generation. J Biol Chem 2003; 278: 18664–18670. Zhou Y, Su Y, Li B, Liu F, Ryder JW, Wu X, Gonzalez-DeWhitt PA, Gelfanova V, Hale JE, May PC, Paul SM, Ni B: Nonsteroidal anti-inflammatory drugs can lower amyloidogenic Abeta42 by inhibiting Rho. Science 2003;302:1215–1217. Yan Q, Zhang J, Liu H, Babu-Khan S, Vassar R, Biere AL, Citron M, Landreth G: Anti-inflammatory drug therapy alters beta-amyloid processing and deposition in an animal model of Alzheimer’s disease. J Neurosci 2003;23:7504–7509. Kukar T, Murphy MP, Eriksen JL, Sagi SA, Weggen S, Smith TE, Ladd T, Khan MA, Kache R, Beard J, Dodson M, Merit S, Ozols VV, Anastasiadis PZ, Das P, Fauq A, Koo EH, Golde TE: Diverse compounds mimic Alzheimer disease-causing mutations by augmenting Abeta42 production. Nat Med 2005; 11:545–550. Gasparini L, Rusconi L, Xu H, Del Soldato P, Ongini E: Modulation of beta-amyloid metabolism by non-steroidal anti-inflammatory drugs in neuronal cell cultures. J Neurochem 2004;88:337–348. Beher D, Clarke EE, Wrigley JD, Martin AC, Nadin A, Churcher I, Shearman MS: Selected non-steroidal anti-inflammatory drugs and their derivatives target gamma-secretase at a novel site. Evidence for an allosteric mechanism. J Biol Chem 2004; 279:43419–43426. Weggen S, Eriksen JL, Sagi SA, Pietrzik CU, Golde TE, Koo EH: Abeta42-lowering nonsteroidal anti-inflammatory drugs preserve intramembrane cleavage of the amyloid precursor protein (APP) and ErbB-4 receptor and signaling through the APP intracellular domain. J Biol Chem 2003; 278: 30748– 30754. Tegeder I, Pfeilschifter J, Geisslinger G: Cyclooxygenase-independent actions of cyclooxygenase inhibitors. FASEB J 2001; 15: 2057–2072. Sagi SA, Weggen S, Eriksen J, Golde TE, Koo EH: The non-cyclooxygenase targets of nonsteroidal anti-inflammatory drugs, lipoxygenases, peroxisome proliferator-activated receptor, inhibitor of kappa B kinase, and NF kappa B, do not reduce amyloid beta 42 production. J Biol Chem 2003;278:31825–31830. Leuchtenberger S, Kummer MP, Kukar T, Czirr E, Teusch N, Sagi SA, Berdeaux R, Pietrzik CU, Ladd TB, Golde TE, Koo EH, Weggen S: Inhibitors of Rho-kinase modulate amyloid-beta (Abeta) secretion but lack selectivity for Abeta42. J Neurochem 2006; 96:355–365. Weggen S, Eriksen JL, Sagi SA, Pietrzik CU, Ozols V, Fauq A, Golde TE, Koo EH: Evidence that nonsteroidal anti-inflammatory drugs decrease amyloid beta 42 production

Neurodegenerative Dis 2006;3:298–304

23

24

25

26

27

28

29

30

31

32

33 34

by direct modulation of gamma-secretase activity. J Biol Chem 2003; 278:31831–31837. Berezovska O, Lleo A, Herl LD, Frosch MP, Stern EA, Bacskai BJ, Hyman BT: Familial Alzheimer’s disease presenilin 1 mutations cause alterations in the conformation of presenilin and interactions with amyloid precursor protein. J Neurosci 2005;25:3009–3017. Lleo A, Berezovska O, Herl L, Raju S, Deng A, Bacskai BJ, Frosch MP, Irizarry M, Hyman BT: Nonsteroidal anti-inflammatory drugs lower Abeta42 and change presenilin 1 conformation. Nat Med 2004;10:1065–1066. Lim GP, Yang F, Chu T, Chen P, Beech W, Teter B, Tran T, Ubeda O, Ashe KH, Frautschy SA, Cole GM: Ibuprofen suppresses plaque pathology and inflammation in a mouse model for Alzheimer’s disease. J Neurosci 2000;20:5709–5714. Jantzen PT, Connor KE, DiCarlo G, Wenk GL, Wallace JL, Rojiani AM, Coppola D, Morgan D, Gordon MN: Microglial activation and beta-amyloid deposit reduction caused by a nitric oxide-releasing nonsteroidal anti-inflammatory drug in amyloid precursor protein plus presenilin-1 transgenic mice. J Neurosci 2002;22:2246–2254. Sung S, Yang H, Uryu K, Lee EB, Zhao L, Shineman D, Trojanowski JQ, Lee VM, Pratico D: Modulation of nuclear factor-kappa B activity by indomethacin influences A beta levels but not A beta precursor protein metabolism in a model of Alzheimer’s disease. Am J Pathol 2004; 165:2197–2206. Lim GP, Yang F, Chu T, Gahtan E, Ubeda O, Beech W, Overmier JB, Hsiao-Ashec K, Frautschy SA, Cole GM: Ibuprofen effects on Alzheimer pathology and open field activity in APPsw transgenic mice. Neurobiol Aging 2001;22:983–991. Lanz TA, Fici GJ, Merchant KM: Lack of specific amyloid-beta(1–42) suppression by nonsteroidal anti-inflammatory drugs in young, plaque-free Tg2576 mice and in guinea pig neuronal cultures. J Pharmacol Exp Ther 2005;312:399–406. Gasparini L, Ongini E, Wenk G: Non-steroidal anti-inflammatory drugs (NSAIDs) in Alzheimer’s disease: old and new mechanisms of action. J Neurochem 2004; 91: 521– 536. Rogers J, Kirby LC, Hempelman SR, Berry DL, McGeer PL, Kaszniak AW, Zalinski J, Cofield M, Mansukhani L, Willson P, et al: Clinical trial of indomethacin in Alzheimer’s disease. Neurology 1993;43:1609–1611. Imbimbo BP: The potential role of non-steroidal anti-inflammatory drugs in treating Alzheimer’s disease. Expert Opin Investig Drugs 2004;13:1469–1481. Tabet N, Feldmand H: Ibuprofen for Alzheimer’s disease. Cochrane Database Syst Rev 2003:CD004031. ht t p : //w w w. my r i ad .c om /a l z hei mer s/ phase2_ad.php

Czirr/Weggen

Original Paper Neurodegenerative Dis 2006;3:305–311 DOI: 10.1159/000095271

Diseases

Role of Amyloid Precursor Protein, Amyloid- and -Secretase in Cholesterol Maintenance Tobias Hartmann Center for Molecular Biology, University of Heidelberg, Heidelberg, Germany

Key Words Amyloid precursor protein Amyloid- Cholesterol Sphingomyelin Statins Alzheimer’s disease therapy Secretase

Abstract Lipids play an important part as risk factors for Alzheimer’s disease. This article summarizes the current understanding of the molecular mechanism by which amyloid- (A) peptides regulate cholesterol and sphingomyelin metabolism, and how in return cholesterol and sphingomyelin regulate A peptide production. An understanding of the physiological function of amyloid precursor protein processing and A function is critical for the development of future therapeutic approaches, e.g. statin treatment. Copyright © 2006 S. Karger AG, Basel

Introduction

Ever since the discovery of the Alzheimer’s disease (AD) amyloid precursor protein (APP) [1] and amyloidbeta (A) peptides derived from this protein [2], scientists have been wondering about their physiological func-

© 2006 S. Karger AG, Basel 1660–2854/06/0035–0305$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

tion or whether they possess any nontoxic activity at all. Lipids confer essential biological function to cells; without them life as we know it would not be possible. Controlling the narrow balance of cellular lipid composition is a complex task, and sophisticated but highly complex mechanism evolved to guarantee proper conditions. The best, but by far not fully understood system is that of cholesterol maintenance. Dysfunctional regulation is almost always linked with disease [3]. Today, it appears that APP processing is an indispensable part of this and another closely associated lipid homeostasis systems [4, 5].

APP Processing

Processing of APP follows the simple yet elegant rules of regulated intramembrane proteolysis (RIP) [6]. Unlike what the ambiguous abbreviation may suggest RIP does not put things to (final) peace. Rather it sets things into motion and is known to work at least in one regulatory system as the central leverage. The best established model is that of sterol regulatory element-binding protein (SREBP) processing. As the name indicates, RIP stands in the very center of cholesterol regulation, which has been investigated and clarified in great detail by the

Tobias Hartmann Zentrum für Molekulare Biologie Heidelberg (ZMBH), University of Heidelberg DE–69120 Heidelberg (Germany) Tel. +49 6221 546 844, Fax +49 6221 545 891 E-Mail [email protected]

Fig. 1. A function in cholesterol and SM

homeostasis. High levels of cholesterol increase APP processing and A release. A40 decreases HMGR activity, resulting in reduced cholesterol de novo synthesis. The Brown and Goldstein pathway regulates HMGR activity as well, but works in the opposite direction. Here, decreased cholesterol levels increase HMGR transcription. Statins, like A40, inhibit HMGR activity. It is currently unknown whether the interaction of HMGR and A40 is direct, but known subcellular localizations suggest that the interaction requires an additional factor. In contrast to this, SM regulation involves the direct interaction of A42 with SMases, SM degrading the enzymes. A42 activates SMase, resulting in reduced SM levels. SMases, like HMGR, are the main rate-determining enzymes in their respective pathway. The A activities are concentration dependent and limited to the physiological A concentration range. At pathological A levels, presumably due to peptide aggregation or conformational changes, their regulatory potential decreases and toxic effects working by different mechanisms appear to dominate.

Brown and Goldstein laboratory [7]. In brief, when cellular cholesterol levels in the endoplasmic reticulum are low SREBP will be cleaved successively by two proteases. The site 1 protease cleavage acts first and this scission is an essential precondition for further processing during which a second cut is done by the site 2 protease. This principle is highly analogue to APP processing, where the -cleavage precedes the -secretase cut [8, 9]. As a consequence of SREBP processing a SREBP fragment is released, which acts as intracellular signaling molecule, eventually resulting in increased hydroxymethylglutaryl-coenzyme A reductase (HMGR) levels. This is critical for cholesterol regulation because HMGR is the rate-limiting enzyme in cholesterol biosynthesis. The analogy between SREBP and APP processing is already fascinating and might help to better understand the regulation of A release. But the relationship appears to extend much further, because the mechanistic analogy appears to be associated with an extensive functional relationship as well, and it is at this level where direct links to AD have been observed (the interaction of APP processing with lipid homeostasis is summarized in figure 1). 306

Neurodegenerative Dis 2006;3:305–311

Lipids – Risk Factor for AD

Apo E is not the only lipid-related risk factor for AD [10–12]. Over the last decade, it has been established that also high cholesterol levels increase the risk for AD [13, 14]. Moreover, treatment with high-dose statins, HMGR inhibitors, slowed down the cognitive decline in AD in clinical pilot studies [15, 16]. There is an abundance of excellent reviews on this topic [17–30]. Also, there is increasing insight into the molecular mechanisms involved. Early on, it was noticed that cholesterol changes APP secretion by -secretase [31] in vitro; but possibly of much greater importance is that cerebral A generation in vitro [32, 33] and in vivo is cholesterol dependent [24, 34, 35], and drastically lowered A levels were achieved by treating guinea pigs with simvastatin [36], a clinically used cholesterol-lowering drug. Cholesterol is a highly potent regulator for A generation and clearly it is the VIP among lipids. But it should not be forgotten that cholesterol interacts with other lipids and that some of those, especially the ganglioside GM1, are far more potent upregulators of -secretase activity [37]. Moreover, previous Hartmann

studies have shown that levels of cholesteryl ester are directly correlated to A production. Inhibitors of acylcoenzyme A:cholesterol acyltransferase (ACAT) could decrease A production. When the ACAT inhibitor was tested in APP transgenic mice, both A levels and plaques were reduced. Thus cholesterol is the most prominent, but not the only lipid involved in the regulation of A metabolism. Exactly how cholesterol increases A production is still an unresolved question. It has been hypothesized that cholesterol alters the conformation of the -secretase complex [38], or that trafficking of the protease [39–41] or APP [42] is cholesterol dependent. There is sufficient evidence for -secretase cleavage to be raft dependent and -secretase has been found in its active form to reside in rafts. Therefore a compelling alternative that puts both options together is that rafts, cholesterol-containing lipid microdomains [41, 43–50], play a role in A generation. Interestingly, Notch, another -secretase substrate, is not cleaved by raft-resident -secretase [41]. This may suggest that cholesterol or other raft-determining lipids are involved in substrate specificity for the -secretase complex. Despite all of this evidence for a direct involvement of lipids in AD and A generation, this does not reveal a functional context explaining why the APP processing system responds in such a sensitive manner to cholesterol and sphingolipids. A first clue in this direction might be contained in the above-mentioned similarity between the regulation of cholesterol homeostasis and APP processing. The first experimental indication that this is indeed the case was obtained when it was observed that the localization of -secretase changes as soon as subcellular cholesterol trafficking is impaired [39, 40]. Moreover, this resulted in an increased production of A from -cleaved APP, but not from full-length APP, strongly pointing towards -secretase as a player in cholesterol biology. Does this indicate that A may act as a signaling molecule to correct altered cholesterol levels? A conclusion which might be derived from the analogy to SREBP processing. For quite some time, A aggregates and highly concentrated A solutions, mimicking the pathological situation observed in AD, had already been suspected to interfere with cholesterol trafficking and homeostasis [51–55]. A related explanation might be that -secretase itself is actively involved in cholesterol trafficking or homeostasis. The latter assumption provides a simple and readily explorable approach. Knockout of presenilins (PSs) and hence -secretase activity in embryonic mouse fibroblasts indeed increases cholesterol levels

drastically, supporting the latter assumption [4]. The direction of this response is important, because in the analogue experiment with the site 2 protease cholesterol levels decrease. Assuming that -secretase functions as a protease in cholesterol regulation, this protease would act in the opposite direction of the Brown and Goldstein pathway. Indeed, inhibition of the proteolytic activity of -secretase either in fibroblast cells, human neuroblastoma cells or primary mouse neurons equally increases cholesterol levels, which not only supports the analogy to the Brown and Goldstein pathway for cholesterol regulation, but also shows that -secretase is involved in cholesterol homeostasis. The inhibition of its proteolytic activity further indicates presence of a -secretase substrate downstream of -secretase-mediated cholesterol regulation. While there are many known substrates for -secretase, none of those with already established function would seem to be able to change cellular cholesterol levels in the described way. However, the physiological function of the APP protein family (APPs), despite their discovery two decades ago, has remained enigmatic [56, 57]. Knockout of APP and APLP2 (APLP1 is expressed mainly in the brain) results within mouse embryonic fibroblast cells in the identical lipid phenotype which was previously observed with PSs knockout, suggesting that APPs are downstream targets of -secretase-mediated cholesterol homeostasis [4]. Are additional -secretase substrates involved in this regulatory pathway? This could be answered when APP/APLP2 knockout fibroblast cells (which have increased cholesterol levels) were treated with a -secretase inhibitor (which in wild-type cells increases cholesterol levels). No change in cholesterol level was observed. Therefore in the absence of APPs, no other -secretase substrate is involved in cholesterol regulation in mouse embryonic fibroblast cells, indicating a linear pathway in vitro.

A and Lipid Biology

Neurodegenerative Dis 2006;3:305–311

The Role of APP and PSs in vivo

While APP and -secretase are infamous as neuronal proteins, the inverse assumption that their physiological functions must thus be neuronal grossly overlooks their ubiquitous expression patterns. Moreover, APP is equally ubiquitously processed to A. Indeed, the vast majority of the information available about APP/PS biology has been obtained from nonneuronal cell cultures. Any main function of APP/PS should therefore reflect this expression pattern and should not be specific to neurons. APP processing by -secretase and their consequential func307

tion in cholesterol homeostasis fulfill these criteria and were observed in cell cultures and tissues from various cerebral and noncerebral origins of diverse mammal species, including man. That this function is of physiological relevance is indicated by the role APP and PSs play in vivo. The single APP knockout is perfectly viable, and no major functional changes had been observed in such mice, especially none which would meet the criteria set by the ubiquitous expression pattern. Increased cholesterol levels were observed in brains and various other organs from APP knockout animals [4], indicating that cholesterol homeostasis is a ubiquitous and physiological function of APP. Although overexpression might not be directly comparable to a knockout scenario, especially if this involves mutated APP and PS, altered cholesterol homeostasis was recently observed in transgenic mice [58]. According to the above, PSs knockout should result in a similar lipid phenotype as does APP knockout. PS-1 and PS-2 are qualitatively able to replace each other in APP processing, making a PSs knockout necessary. Such mice are not viable; therefore, only the data from the conditional PS-1/PS-2 knockout are available, which show a significant but moderate cholesterol increase. The reduced phenotype strength appears to be caused in part by the blend of cells with functional and nonfunctional secretase activity. But there is another important factor involved – the APP processing product. Cleavage by /secretase results in several different APP products, including A, p3 and the APP intracellular domain. Incubation of APP knockout cells with the conditioned medium of wild-type cells partly recovers the lower cholesterol levels observed in wild-type cells, whereas incubation with the control medium of APP knockout cells does not, indicating that the active cholesterol-regulating APP processing product is secreted. When these cells were incubated with synthetic A, at a concentration typically found in the cerebrospinal fluid or blood plasma [59] or with APP knockout cell conditioned medium with added synthetic A40, cholesterol homeostasis was restored. Therefore, A has the ability to restore cellular cholesterol level in the absence of APP or -secretase. This function might provide a mechanistic explanation for an earlier observation that neuronal survival in vitro is reduced in the presence of -secretase inhibition, but rescued by the presence of picomolar to nanomolar A40 [60]. Interestingly, in this work the need for A was only found for neuronal cells which do not require fetal calf serum, a major source of lipids, whereas survival of nonneuronal cells, which do require substantial amounts of 308

Neurodegenerative Dis 2006;3:305–311

fetal calf serum, apparently did not profit from A presence. The molecular target for A40 is HMGR. As mentioned before, HMGR is situated at the entry of the cholesterol biosynthesis pathway. It is the rate-limiting enzyme regulated by the Brown and Goldstein pathway and inhibited by statins. The exact mechanism by which A and HMGR interact is unknown and presents a spatial problem, because HMGR is localized to the endoplasmic reticulum and A40 is usually not, suggesting the existence of at least one other intermediate molecule. Statins and A-Mediated Cholesterol Downregulation

The close functional overlap between statins and A may have significant relevance for future AD therapy. When cholesterol levels are high, more A is produced, which then results in a feedback signal reducing cellular cholesterol synthesis. For the brain this could be of special weight, because unlike in peripheral organs which receive some cholesterol from diet, the brain is believed to be almost exclusively dependent on in situ synthesized cholesterol. In principle, the close functional relationship and the identical molecular target of A and statins provide a favorable pharmacological situation. However, APP processing and A functions are not limited to cholesterol regulation. Therefore, further HMGR-related issues may have to be taken into consideration. A or statin-mediated cholesterol de novo synthesis regulation occurs very early in the biosynthetic pathway. Since this pathway is the source of several other essential molecules, especially those derived from the isoprenoid pathway, reduced HMGR activity may result in additional and noticeable effects unrelated to cholesterol. This highlights another twist to the A-lipid theme. Even in the presence of significant statin levels, sufficient HMG-Co A is converted to mevalonate to allow isoprenoid production; however, when this is experimentally prevented intracellular A levels increase. This does not depend on cellular cholesterol levels, but rather on mevalonate levels, a common precursor for isoprenoids and sterols. Under limiting conditions, most mevalonate is converted to isoprenoids and little de novo cholesterol synthesis occurs. Therefore, mild and moderate inhibition of HMGR activity already affects cholesterol and consequentially A production, but has little impact on ispoprenoids. However when, in vitro, HMGR is severely inhibited, things appear to Hartmann

change and the reduced isoprenoid levels become the dominant factor in determining A production. An excellent review on this topic was recently published by Cole and Vassar [61]. The cholesterol, isoprenoid and sphingolipid pathways (see below) as well as some polyunsaturated fatty acids are first examples how tightly APP processing is involved in lipid biology. The mechanistic details differ and what applies for one pathway appears often to be quite the opposite for one of the other pathways. This can be used as an advantage, since it allows to target those pathways specifically. For example, with clinically reasonable statin levels, the isoprenoid production is, if at all, barely touched. For the noncholesterol pathways, little in vivo data are currently available. Whether they will be suitable for therapeutic approaches is a challenging but also very promising question, as they hold the potential to greatly extend the therapeutic treasure box.

man neutral placenta SMase reaches a peak at approximately 1 nM A42 concentration. The following decline appears to be due to the reduction in the concentration of freely available A42, because preaggregated A42 has only residual SMase-stimulating activity. However, at approximately 1,000-fold higher A42 concentration stimulation of SMase reoccurs, this time not due to A-SMase interaction, but due to the production of reactive oxygen by aggregated A [65]. This situation might be further aggravated due to increased ceramide release [5, 66]. Whereas we could observe neither increased apoptosis nor susceptibility to apoptosis-inducing factors in PS or APP knockout cells [Walcak and Hartmann, unpubl. obs.], nor increased ROS release [4], this might drastically differ when massive amyloid deposits are present as this occurs late during AD pathogenesis.

Conclusion A42 and Sphingomyelin

Another role of A which is not directly linked with cholesterol is its function in controlling sphingolipid homeostasis. Cholesterol and sphingolipids, including the glycosphingolipids, are most abundant in the plasma membrane, and together are the main components of lipid microdomains or rafts [43, 62–64]. Because sphingolipid and cholesterol content strongly affects the functional properties of rafts, the ratio of these lipids is under strict control. Like in cholesterol synthesis, the levels of sphingomyelin (SM), the main sphingolipid, can be regulated by SREBP and A. With respect to A, an important difference is that only A40 decreases HMGR activity, whereas at physiological concentrations only A42 is involved in SM regulation [4]. SM levels are mainly controlled by SMase-catalyzed degradation of SM to ceramide. Several different SMases exist and in different subcellular localizations, including the plasma membrane, and acidic SMase can be secreted maintaining enzymatic activity. The interaction between A42 and SMases appears to be direct, because only A42 is needed to activate the turnover of SM to ceramide in the presence of purified SMase. This is supported by the subcellular localizations of A, SM and SMases, all of which are present or exposed to the outer membrane layer. Remarkably, for both A and SMase secreted forms exist. Indeed, it is possible to increase cellular SMase activity in APP or PS knockout cells by adding pM amounts of synthetic A to the cell culture medium. Interestingly, activation of huA and Lipid Biology

Physiological A peptides appear to play an essential role in lipid homeostasis, have the potential to regulate lipid microdomain function, and A40 even reduces cholesterol levels in vivo. However, when A starts to aggregate, it looses these capabilities and becomes an uncontrolled factor which eventually induces neurodegeneration. How A turns from good to evil, how this can be prevented continues to be the main challenge of AD research. The hope is that with a first glimpse at the physiological function of A the challenge is now a little bit easier to address.

Acknowledgement This study was supported by the Deutsche Forschungsgemeinschaft.

Neurodegenerative Dis 2006;3:305–311

309

References 1 Kang J, Lemaire HG, Unterbeck A, Salbaum JM, Masters CL, Grzeschik KH, Multhaup G, Beyreuther K, Muller Hill B: The precursor of Alzheimer’s disease amyloid A4 protein resembles a cell-surface receptor. Nature 1987;325:733–736. 2 Masters CL, Simms G, Weinman NA, Multhaup G, McDonald BL, Beyreuther K: Amyloid plaque core protein in Alzheimer disease and Down syndrome. Proc Natl Acad Sci USA 1985;12,82:4245–4249. 3 Maxfield FR, Tabas I: Role of cholesterol and lipid organization in disease. Nature 2005; 438:612–621. 4 Grimm MO, Grimm HS, Patzold AJ, Zinser EG, Halonen R, Duering M, Tschape JA, De Strooper B, Muller U, Shen J, Hartmann T: Regulation of cholesterol and sphingomyelin metabolism by amyloid-beta and presenilin. Nat Cell Biol 2005; 7:1118–1123. 5 Mattson MP, Cutler RG, Jo DG: Alzheimer peptides perturb lipid-regulating enzymes. Nat Cell Biol 2005;7:1045–1047. 6 Brown MS, Goldstein JL: A proteolytic pathway that controls the cholesterol content of membranes, cells, and blood. Proc Natl Acad Sci USA 1999; 96:11041–11048. 7 Horton JD, Goldstein JL, Brown MS: SREBPs: Activators of the complete program of cholesterol and fatty acid synthesis in the liver. J Clin Invest 2002; 109:1125–1131. 8 Brown MS, Ye J, Rawson RB, Goldstein JL: Regulated intramembrane proteolysis: A control mechanism conserved from bacteria to humans. Cell 2002;100:391–398. 9 Reinhard C, Hebert SS, De Strooper B: The amyloid-beta precursor protein: integrating structure with biological function. EMBO J 2005;24:3996–4006. 10 Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, Pericak-Vance MA: Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 1993;261:921–923. 11 Czech C, Monning V, Tienari PJ, Hartmann T, Masters C, Beyreuther K, Forstl H: Apolipoprotein E-epsilon 4 allele and Alzheimer’s disease. Lancet 1993; 342:1309. 12 Levi O, Lutjohann D, Devir A, von Bergmann K, Hartmann T, Michaelson DM: Regulation of hippocampal cholesterol metabolism by apoE and environmental stimulation. J Neurochem 2005;95:987–997. 13 Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Iivonen S, Mannermaa A, Tuomilehto J, Nissinen A, Soininen H: Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease. Ann Intern Med 2002; 137: 149–155.

310

14 Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Soininen H, Tuomilehto J, Nissinen A: Midlife vascular risk factors and Alzheimer’s disease in later life: Longitudinal, population based study. BMJ 2001;322:1447–1451. 15 Simons M, Schwarzler F, Lutjohann D, von Bergmann K, Beyreuther K, Dichgans J, Wormstall H, Hartmann T, Schulz JB: Treatment with simvastatin in normocholesterolemic patients with Alzheimer’s disease: a 26week randomized, placebo-controlled, double-blind trial. Ann Neurol 2002; 52: 346–350. 16 Sparks DL, Sabbagh MN, Connor DJ, Lopez J, Launer LJ, Browne P, Wasser D, JohnsonTraver S, Lochhead J, Ziolwolski C: Atorvastatin for the treatment of mild to moderate Alzheimer disease: preliminary results. Arch Neurol 2005;62:753–757. 17 Hartmann T: Cholesterol, Abeta and Alzheimer’s disease. TINS 2001;24:45–48. 18 Puglielli L, Tanzi RE, Kovacs DM: Alzheimer’s disease: the cholesterol connection. Nat Neurosci 2003;6:345–351. 19 Dominguez DI, De Strooper B: Novel therapeutic strategies provide the real test for the amyloid hypothesis of Alzheimer’s disease. Trends Pharmacol Sci 2002;23:324–330. 20 Wood WG, Schroeder F, Igbavboa U, Avdulov NA, Chochina SV: Brain membrane cholesterol domains, aging and amyloid beta-peptides. Neurobiol Aging 2002; 23: 685–694. 21 Casserly I, Topol E: Convergence of atherosclerosis and Alzheimer’s disease: inflammation, cholesterol, and misfolded proteins. Lancet 2004;363:1139–1146. 22 Michikawa M: Neurodegenerative disorders and cholesterol. Curr Alzheimer Res 2004;1: 271–275. 23 Hartman T: Cholesterol and Alzheimer’s disease: statins, cholesterol depletion in APP processing and Abeta generation. Subcell Biochem 2005;38:365–380. 24 Burns M, Duff K: Use of in vivo models to study the role of cholesterol in the etiology of Alzheimer’s disease. Neurochem Res 2003; 28:979–986. 25 Shobab LA, Hsiung GY, Feldman HH: Cholesterol in Alzheimer’s disease. Lancet Neurol 2005;4:841–852. 26 Kaether C, Haass C: A lipid boundary separates APP and secretases and limits amyloid beta-peptide generation. J Cell Biol 2004; 167:809–812. 27 Wolozin B: Cholesterol, statins and dementia. Curr Opin Lipidol 2004; 15:667–672. 28 Wolozin B: Cholesterol and the biology of Alzheimer’s disease. Neuron 2004;41:7–10. 29 Rockwood K, Darvesh S: The risk of dementia in relation to statins and other lipid lowering agents. Neurol Res 2003;25:601–604.

Neurodegenerative Dis 2006;3:305–311

30 Marx J: Neuroscience. Preventing Alzheimer’s: a lifelong commitment? Science 2005; 309:864–866. 31 Bodovitz S, Klein WL: Cholesterol modulates alpha-secretase cleavage of amyloid precursor protein. J Biol Chem 1996; 271: 4436–4440. 32 Simons M, Keller P, De Strooper B, Beyreuther K, Dotti CG, Simons K: Cholesterol depletion inhibits the generation of beta-amyloid in hippocampal neurons. Proc Natl Acad Sci USA 1998;95:6460–6464. 33 Mizuno T, Nakata M, Naiki H, Michikawa M, Wang R, Haass C, Yanagisawa K: Cholesterol-dependent generation of a seeding amyloid beta-protein in cell culture. J Biol Chem 1999;274:15110–15114. 34 Sparks DL, Scheff SW, Hunsaker JC 3rd, Liu H, Landers T, Gross DR: Induction of Alzheimer-like beta-amyloid immunoreactivity in the brains of rabbits with dietary cholesterol. Exp Neurol 1994; 126:88–94. 35 Refolo LM, Pappolla MA, Malester B, LaFrancois J, Bryant-Thomas T, Wang R, Tint GS, Sambamurti K, Duff K: Hypercholesterolemia accelerates the Alzheimer’s amyloid pathology in a transgenic mouse model. Neurobiol Dis 2000;7:321–331. 36 Fassbender K, Simons M, Bergmann C, Stroick M, Lutjohann D, Keller P, Runz H, Kuhl S, Bertsch T, von Bergmann K, Hennerici M, Beyreuther K, Hartmann T: Simvastatin strongly reduces levels of Alzheimer’s disease beta-amyloid peptides Abeta 42 and Abeta 40 in vitro and in vivo. Proc Natl Acad Sci USA 2001;98:5856–5861. 37 Zha Q, Ruan Y, Hartmann T, Beyreuther K, Zhang D: GM1 ganglioside regulates the proteolysis of amyloid precursor protein. Mol Psychiatry 2004;9:946–952. 38 Wahrle S, Das P, Nyborg AC, McLendon C, Shoji M, Kawarabayashi T, Younkin LH, Younkin SG, Golde TE: Cholesterol-dependent gamma-secretase activity in buoyant cholesterol-rich membrane microdomains. Neurobiol Dis 2002;9:11–23. 39 Runz H, Rietdorf J, Tomic I, de Bernard M, Beyreuther K, Pepperkok R, Hartmann T: Inhibition of intracellular cholesterol transport alters presenilin localization and amyloid precursor protein processing in neuronal cells. J Neurosci 2002;22:1679–1689. 40 Burns M, Gaynor K, Olm V, Mercken M, LaFrancois J, Wang L, Mathews PM, Noble W, Matsuoka Y, Duff K: Presenilin redistribution associated with aberrant cholesterol transport enhances beta-amyloid production in vivo. J Neurosci 2003;23:5645–5649. 41 Vetrivel KS, Cheng H, Kim SH, Chen Y, Barnes NY, Parent AT, Sisodia SS, Thinakaran G: Spatial segregation of gammasecretase and substrates in distinct membrane domains. J Biol Chem 2005; 280: 25892–25900.

Hartmann

42 Marlow L, Cain M, Pappolla MA, Sambamurti K: Beta-secretase processing of the Alzheimer’s amyloid protein precursor (APP). J Mol Neurosci 2003;20:233–239. 43 Simons K, Ehehalt R: Cholesterol, lipid rafts, and disease. J Clin Invest 2002; 110: 597– 603. 44 Ehehalt R, Keller P, Haass C, Thiele C, Simons K: Amyloidogenic processing of the Alzheimer beta-amyloid precursor protein depends on lipid rafts. J Cell Biol 2003; 160: 113–123. 45 Kalvodova L, Kahya N, Schwille P, Ehehalt R, Verkade P, Drechsel D, Simons K: Lipids as modulators of proteolytic activity of BACE: involvement of cholesterol, glycosphingolipids, and anionic phospholipids in vitro. J Biol Chem 2005;280:36815–36823. 46 Kametaka S, Shibata M, Moroe K, Kanamori S, Ohsawa Y, Waguri S, Sims PJ, Emoto K, Umeda M, Uchiyama Y: Identification of phospholipid scramblase 1 as a novel interacting molecule with beta-secretase (betasite amyloid precursor protein (APP) cleaving enzyme (BACE)). J Biol Chem 2003; 278: 15239–15245. 47 Parkin ET, Turner AJ, Hooper NM: Amyloid precursor protein, although partially detergent-insoluble in mouse cerebral cortex, behaves as an atypical lipid raft protein. Biochem J 1999;344:23–30. 48 Riddell DR, Christie G, Hussain I, Dingwall C: Compartmentalization of beta-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts. Curr Biol 2001;11:1288–1293. 49 Cordy JM, Hussain I, Dingwall C, Hooper NM, Turner AJ: Exclusively targeting betasecretase to lipid rafts by GPI-anchor addition up-regulates beta-site processing of the amyloid precursor protein. Proc Natl Acad Sci USA 2003;100:11735–11740.

A and Lipid Biology

50 Grimm MOW, Tschape JA, Grimm HS, Zinser EG, Hartmann T: Altered membrane fluidity and lipid raft composition in presenilin-deficient cells. Acta Neurol Scan, in press. 51 Liu Y, Peterson DA, Schubert D: Amyloid beta peptide alters intracellular vesicle trafficking and cholesterol homeostasis. Proc Natl Acad Sci USA 1998;95:13266–13271. 52 Koudinov A, Koudinova N: Amyloid beta protein restores hippocampal longterm potentiation: a central role for cholesterol? Neurobiol Lipids 2003;1:8. 53 Igbavboa U, Pidcock JM, Johnson LN, Malo TM, Studniski AE, Yu S, Sun GY, Wood WG: Cholesterol distribution in the Golgi complex of DITNC1 astrocytes is differentially altered by fresh and aged amyloid beta-peptide-(1–42). J Biol Chem 2003; 278: 17150– 17157. 54 Michikawa M, Gong JS, Fan QW, Sawamura N, Yanagisawa K: A novel action of Alzheimer’s amyloid beta-protein (Abeta): oligomeric Abeta promotes lipid release. J Neurosci 2001;21:7226–7235. 55 Gong JS, Sawamura N, Zou K, Sakai J, Yanagisawa K, Michikawa M: Amyloid beta-protein affects cholesterol metabolism in cultured neurons: implications for pivotal role of cholesterol in the amyloid cascade. J Neurosci Res 2002;70: 438–446. 56 Heber S, Herms J, Gajic V, Hainfellner J, Aguzzi A, Rulicke T, von Kretzschmar H, von Koch C, Sisodia S, Tremml P, Lipp HP, Wolfer DP, Muller U: Mice with combined gene knock-outs reveal essential and partially redundant functions of amyloid precursor protein family members. J Neurosci 2000; 20:7951–7963. 57 Herms J, Anliker B, Heber S, Ring S, Fuhrmann M, Kretzschmar H, Sisodia S, Muller U: Cortical dysplasia resembling human type 2 lissencephaly in mice lacking all three APP family members. EMBO J 2002; 23:4106–4115.

58 Wirths O, Thelen K, Breyhan H, Luzon-Toro B, Hoffmann KH, Falkai P, Lutjohann D, Bayer TA: Decreased plasma cholesterol levels during aging in transgenic mouse models of Alzheimer’s disease. Exp Gerontol 2006; 41:220–224. 59 Ida N, Hartmann T, Pantel J, Schröder J, Zerfass R, Förstl H, Sandbrink R, Masters C, Beyreuther K: Analysis of heterogeneous A4 peptides in human cerebrospinal fluid and blood by a newly developed sensitive Western blot assay. J Biol Chem 1996; 271: 22908–22914. 60 Plant LD, Boyle JP, Smith IF, Peers C, Pearson HA: The production of amyloid beta peptide is a critical requirement for the viability of central neurons. J Neurosci 2003;23: 5531–5535. 61 Cole SL, Vassar R: Isoprenoids and Alzheimer’s disease: a complex relationship. Neurobiol Dis 2006;22:209–222. 62 Simons K, Vaz WL: Model systems, lipid rafts, and cell membranes. Annu Rev Biophys Biomol Struct 2004;33:269–295. 63 Simons K, Ikonen E: How cells handle cholesterol. Science 2000;290:1721–1726. 64 Simons K, Ikonen E: Functional rafts in cell membranes. Nature 1997;387:569–572. 65 Jana A, Pahan K: Fibrillar amyloid-beta peptides kill human primary neurons via NADPH oxidase-mediated activation of neutral sphingomyelinase: implications for Alzheimer’s disease. J Biol Chem 2004; 279: 51451–51459. 66 Cutler RG, Kelly J, Storie K, Pedersen WA, Tammara A, Hatanpaa K, Troncoso JC, Mattson MP: Involvement of oxidative stress-induced abnormalities in ceramide and cholesterol metabolism in brain aging and Alzheimer’s disease. Proc Natl Acad Sci USA 2004;101:2070–2075.

Neurodegenerative Dis 2006;3:305–311

311

Author Index Vol. 3, No. 4–5, 2006

Diseases Anliker, B. 239 Barghorn, S. 197 Baumann, S. 290 Baumeister, R. 227 Beyreuther, K. 218 Braun, H.A. 290 Czirr, E. 298 Fahrenholz, F. 255 Griesshaber, B. 207 Haass, C. 193, 275, 284 Haffner, C. 284 Hartmann, T. 305 Jaeger, S. 233

Jeganathan, S. 197 Kaether, C. 275 Kins, S. 218 Larbig, G. 290 Lauther, N. 218 Li, X.-Y. 207 Lichtenthaler, S.F. 262 Mandelkow, E. 197 Mandelkow, E.-M. 197, 207 Matenia, D. 207 Müller, U. 239 Multhaup, G. 193, 270 Narlawar, R. 290

Pietrzik, C.U. 233 Postina, R. 255 Schmidt, B. 290 Smialowska, A. 227 St. George-Hyslop, P. 191 Steiner, H. 275 Szodorai, A. 218 Timm, T. 207 von Bergen, M. 197 Waldron, E. 233 Walter, J. 247 Weggen, S. 298

Subject Index Vol. 3, No. 4–5, 2006

Actin 207 ADAM10 255 Alzheimer’s disease 197, 227, 233, 239, 247, 262, 275, 290, 298 – – therapy 305 Amyloid A 270 – precursor protein 233, 262, 270, 305 – precursor-like protein 239 Amyloid- 305 – peptide 275, 298 -Amyloid precursor protein 247 – – –, in vivo function 239 Aspartic protease 290 Axonal transport 218 BACE inhibitors 290 Caenorhabditis elegans 227 Cargo receptor 218 Cholesterol 255, 305

© 2006 S. Karger AG, Basel Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com

Accessible online at: www.karger.com/ndd

c-Jun N-terminal kinase-interacting protein 218 Ectodomain shedding 262 Endocytosis 262 Endophilin 262 Functional redundancy 239 Glycosphingolipids 247 Homodimerization 270 Ibuprofen 298 Kinesin 218 Knockout mice 239 Low-density lipoprotein receptor-related protein 233 Microtubule-associated proteins 197 Microtubules 207 Modifier screen 227 Nicastrin-like protein 284 Nodal modulator 284 – signaling pathway 284

Nonsteroidal anti-inflammatory drugs 298 Oligomerization 270 p21-activated kinase 207 Paired helical filament 197 Phosphorylation of tau 207 Presenilin 227, 275, 284, 290 Retinoic acid 255 Secretase 290, 305 -Secretase 255 -Secretase 270 -Secretase 275, 298 Secretases 262 Signal peptide peptidase 284 Sphingomyelin 305 Statins 305 Subcellular trafficking 247 Tau protein 197