commentary

NATURE CELL BIOLOGY VOL 3 OCTOBER 2001 http://cellbio.nature.com E221

Presenilins and the intramembraneproteolysis of proteins: facts and fiction

Bart De Strooper* and Wim Annaert

Missense mutations in the genes coding for presenilin 1 and presenilin 2 cause familial Alzheimer’s disease — aprogressive neurodegenerative disorder of the central nervous system. Loss-of-function mutations of these genesin Drosophila, Caenorhabditis elegans and mice cause severe lethal phenotypes, which implicates the presenilinsgenetically in the Notch signalling pathway. The hypothesis that presenilins are aspartyl proteases that cleave theamyloid precursor protein and Notch can explain the phenotypes. Direct evidence for this hypothesis is, however,difficult to obtain. Moreover, presenilin 1 is a multifunctional protein, as exemplified by its role in the Wnt/β-catenin signalling pathway.

Some years ago, the many diverse viewson the pathogenesis of Alzheimer’sdisease were compared to the descrip-

tions of “blindfolded investigators describ-ing an elephant”1. Whereas someresearchers stressed the importance of thetrunk, others described the presence of thetail, and yet others focused on the enor-mous pillars, which turned out to be thelegs of the animal. The same story could beused now to describe our current knowl-edge of presenilins (PSs). But although thefull picture remains unclear, we should rec-ognize that impressive progress has beenmade since 1995 when mutations in prese-nilin 1 (PS1) and presenilin 2 (PS2) werediscovered to be major causes of familialAlzheimer’s disease2–4 (FAD).

PS1 and its homologue PS2 are serpen-tine proteins with a relative molecular mass(Mr) of about 50,000 (50K) that cross themembrane of the endoplasmic reticulum(ER) several times (Fig. 1). The mostfavoured topological model is one witheight transmembrane domains5,6, althoughfurther detailed structural analysis is need-ed to confirm this. An unknown ‘presenili-nase’ cuts the presenilins (PSs) in a ninthhydrophobic region, which precedes thelarge hydrophilic loop between transmem-brane domains 6 and 7, to give rise to a~30K amino-terminal fragment (NTF) anda ~20K carboxy-terminal fragment (CTF)that stay together in a heterodimeric com-plex7–10. When cell extracts are fractionatedin glycerol gradients, the NTF and CTF runwith apparent molecular weights(150K–250K) that are significantly higherthan expected, indicating that they formoligomers or are associated with other pro-teins to form a larger PS complex9,11,12.

Interestingly, high expression of PS intransfected cells in culture results mainly inincreased levels of the uncleaved PS holo-protein. This holoprotein is not incorporat-ed into the larger complexes and is rapidlydegraded by the proteasome10,13. The stabi-lization of PS apparently depends critically

on its association with additional cellularcomponents that are only available in limit-ed amounts7,12. The incorporation in thiscomplex seems also to be rate limiting forthe ‘presenilinase processing’ of PS.

One component of the PS complex hasbeen identified recently: Nicastrin is a typeI integral membrane protein that bindsquantitatively to PS14. It is involved in

GLP-1 signalling (GLP-1 is a Caenorhabditiselegans homologue of Notch), and muta-tions in a conserved D-Y-I-G-S sequencecause increased generation of the amyloidβ-peptide (Aβ), suggesting that Nicastrin isindeed part of the functional γ-secretasecomplex. β-catenin is also associated withthe PS1 complex14,15, but this interactionseems to have no functional consequences

CHOCHO

KPI

N

C

Furin

TACE

BACE

ADAM10/TACE

KMDA

QKLV

C

VVIA C

GVL

RQRR

EAVK

APP

Notch

PS

NC

TLVM

Figure 1 PS regulated intramembrane proteolysis of Notch and APP. Notch, PS and APP are depictedon the left, taking into account their relative molecular masses. The circled region is enlarged on theright to detail the different proteolytic processes that activate APP and Notch. APP is cleaved in itsectodomain by BACE58 at the β-secretase site (K-M-D-A) or by ADAM10 (ref. 69) or ADAM17 (alsoknown as TACE)70 at the α-secretase site (Q-K-L-V). Removal of the ectodomain is required before PS-dependent γ-secretase cleavage in the transmembrane region of APP can occur71. Notch is cleavedat the S1 site (R-Q-R-R) by furin72 during transit to the cell surface. The two fragments remain associ-ated in a non-covalent fashion and form the functional Notch heterodimeric receptor. After binding toDelta or Jagged, Notch becomes activated by a second cleavage by ADAM17 (TACE) at the S2site73,74 (E-A-V-K). A third cleavage at the S3 site (C-G-V-L) occurs in the transmembrane region,which releases the Notch intracellular domain. The S3 cleavage is PS dependent and can be inhibitedby the same protease inhibitors that inhibit γ-secretase cleavage of APP49. Recent data indicate thatAPP is cleaved at several positions, including one similar to the Notch cleavage site (T-L-V-M)30,32.

© 2001 Macmillan Magazines Ltd

commentary

NATURE CELL BIOLOGY VOL 3 OCTOBER 2001 http://cellbio.nature.comE222

with respect to γ-secretase activity16,17. Manyother proteins can bind to PS as demonstrat-ed by various approaches; however, furtherwork is needed to clarify the physiologicalsignificance of these interactions.

The biological function of the PSs hasbeen investigated in C. elegans, Drosophilamelanogaster and Mus musculus.Deficiencies in the PS genes cause severelethal phenotypes similar to those observedwhen vital elements in the Notch signallingpathway are inactivated18. The Notch path-way controls several complex cell fate deci-sions during embryogenesis and adulthood,which in general dictates that the receivingcells remain in the more undifferentiated(precursor) state while neighbouring cellsdifferentiate towards a mature phenotype.

PS1 in particular is also involved in theWnt/β-catenin signalling pathway17. Thisaspect of PS function is revealed in vivo

when the lethal, Notch-deficient phenotypeof PS1–/– mice19,20 is rescued by restoringexpression of PS1 by using the human Thy-1 promotor21. The resulting mice are viablebut develop epidermal hyperplasia andmalignant skin tumours. Because the Thy-1promotor cannot restore PS1 expression inskin keratinocytes, this malignant pheno-type is probably caused by the cell-specificloss of function of PS1. Further experi-ments in PS1–/– fibroblasts17 or in PS1–/– ker-atinocytes21 have shown the accumulationof phosphorylated β-catenin in the cyto-plasm, and a concomitant activation of thecyclin D1 and other target genes of theWnt/β-catenin signalling pathway. Theseobservations can explain the cell prolifera-tion and tumours that occur in thesemice17,21. PS1 is apparently needed for therapid degradation of phosphorylated β-catenin by the proteasome, but how this

actually works remains largely a matter ofspeculation22.

Other in vitro studies have implicated PSin regulation of apoptosis23, the unfoldedprotein stress response24,25 and capacitiveCa2+ entry26,27. Again, further in vivo studiesare needed to elucidate the physiologicalconsequences of these observations. Forexample, Yu et al.28 did not detect anincrease in apoptosis in the cerebral cortexof conditional PS1 knockout mice. In addi-tion, the role of PS in the unfolded proteinresponse has been questioned29.

To solve this type of controversy, it willbe important to investigate more rigorouslythe exact molecular mechanism by whichPSs operate in these processes. For instance,it has been suggested that PSs control theproteolytic processing of Ire1 — a type Iintegral membrane protein kinase thatsenses the accumulation of misfolded pro-tein in the ER and transduces a signal to thenucleus to upregulate genes involved in theunfolded protein response. Surprisingly,this interesting possibility was only investi-gated by indirect, immunocytochemicalmeans25. A direct comparison of the prote-olysis of Ire1 in PS-deficient and wild-typecells would have yielded much more con-clusive data.

In any event, PS is a major constituent ofthe proteolytic activity responsible for theintramembraneous cleavage of the amyloidprecursor protein (APP), which releases theAβ peptides in the extracellular space.Recent evidence shows that an additionalPS-dependent cleavage releases the APPintracellular domain (AICD) in the cyto-plasm, similar to the PS-dependent S3-cleavage releasing the Notch intracellulardomain (see below)30–32 (Fig 1). The Aβpeptides are the main components of thecharacteristic amyloid plaques in the brainof Alzheimer’s disease patients. This prote-olytic activity has been called γ-secretase,and is apparently relaxed in its specificityfor its target amino-acid residue sequence33.Consequently, a spectrum of Aβ peptides isgenerated containing 39–43 amino-acidresidues.

Studies in fibroblasts derived frompatients with familiar Alzheimer’s disease(FAD) have shown that all clinical muta-tions of PS increase the relative proportionof secreted peptides ending at residue 42(Aβ42) compared with those ending atresidue 40 (Aβ40)34. Aβ42 is considered aspathologically relevant because of its strongtendency to aggregate. But PS1 seems to beneeded for generating the other Aβ pep-tides as well, as shown in neurons derivedfrom PS1-deficient mice35,36. Moreover, allAβ production ceases completely when, inaddition to PS1, the PS2 homologue is inac-tivated in embryonic stem cells37,38.

Checler and collaborators39 have recent-ly challenged this conclusion. They foundnormal levels of Aβ peptides in PS-deficient

X

C

N

CN

C

N

COPII

?

SREBP

SCAP

+ Sterols

- Sterols

ER

N

to nucleus

C

N

CN

CN

C

N

CN

C

N

CN N

C

NC

N

C

S2P

Golgi

N

N

C

S1PC

N

C

N

C

N

C

N

X

XC

Figure 2 Regulated intramembrane proteolysis and vesicular trafficking of SREBP/SCAP. The best-studied example of regulated intramembrane proteolysis is the SREBP/SCAP system involved in theregulation of sterol biosynthesis62. We use this example as a template to lay out a series of questionsthat have not been answered as yet for the APP/PS system (see Fig. 3). The sterol regulatory-ele-ment-binding protein (SREBP) is a hairpin-like membrane protein residing in the endoplasmic reticu-lum (ER) with its N and C termini facing the cytosol. The N-terminal portion has transcriptional activi-ty and needs to be liberated by two sequential cleavages before it can travel to the nucleus. TheSREBP cleavage-activating protein (SCAP) comes into the picture here, as it complexes and exportsSREBP out of the ER to the Golgi, where the responsible proteases reside. This transport is tightlyregulated by the cellular sterol content61, which is sensed by the N-terminal domain of SCAP. Sterolsprobably cause the SREBP/SCAP complex to bind to a (unidentified) ER-retention protein (the orangestructure labelled ‘X’ in this figure)75. Sterol deprivation releases this binding, allowing the complex toleave the ER, probably by means of COPII vesicles76. In the Golgi, the site 1 protease (S1P), a mem-brane-bound subtilisin-like protease63, cleaves SREBP in its luminal loop domain. The site 2 protease(S2P) then cleaves the first transmembrane helix, releasing the N-terminal transcription factordomain77. S2P is extremely hydrophobic and most probably a zinc metalloprotease64. The N-terminalSREBP fragment is translocated to the nucleus, but the subsequent fates of the C-terminal SREBPfragment and SCAP in the Golgi remain unknown (?). Most likely, SCAP travels back to the ER by ret-rograde transport. Whether the C-terminal SREBP fragment remains associated with SCAP needs tobe clarified.

© 2001 Macmillan Magazines Ltd

commentary

NATURE CELL BIOLOGY VOL 3 OCTOBER 2001 http://cellbio.nature.com E223

fibroblasts, which disagrees with all otherpublished reports35,37,38,40–43. Earlier worksuggested that γ-secretase might be anaspartyl type protease44, drawing attentionto aspartate residues 257 and 385 in trans-membrane domains 6 and 7 of PS1, respec-tively45. These aspartates (Asp 257 andAsp 385) are highly conserved in evolutionand even found in the plant homologues ofPS. Moreover, some similarity with a puta-tive catalytic site in the bacterial type 4prepilin proteases has been noticed46.Substitution of the aspartates by otheramino-acid residues results in a proteolyti-cal inactive form of PS45. When thesemutants are expressed in cells at relativelyhigh levels, a ‘replacement’ of the endoge-nous PS1 and PS2 is observed.

Most likely, these mutants compete withwild-type PSs for the afore-mentionedunknown limiting factors that regulatetheir incorporation into the putative ‘func-tional’ complexes. In any event, the finalresult is the strong inhibition of γ-secretasecleavage of APP. A logical hypothesis is,then, that the two aspartates constitute thecatalytic site of γ-secretase45. But alternativeexplanations are possible as mutations ofcharged amino acids in the hydrophobicenvironment of the membrane can causeimportant conformational changes. In linewith this possibility, Yu et al.12 have shownthat PSs containing such aspartate muta-tions are incorporated in lower Mr com-plexes than the 250K, putative ‘active’ com-plexes mentioned above. The strongestargument in favour of the hypothesis thatPSs contain the active γ-secretase site is thecrosslinking of γ-secretase inhibitors toPSs47,48. This argument is critically depend-ent on the specificity of the binding of theinhibitor to the protease, and the assump-tion that the inhibitors are ‘transition-state’specific.

As the working mechanism of γ-secre-tase is not known, the assumption of tran-sition-state specificity is hypothetical.Furthermore, it probably cannot be exclud-ed that the inhibitors also tagged other pro-teins, especially in the higher and lower Mr

range, in a specific way. Nevertheless, it willbe extremely difficult to provide a moreconclusive proof for the ‘PS is γ-secretase’hypothesis. This will require in vitroreconstitution of the γ-secretase activity,which probably depends on the presenceof the other unknown components of thePS complex and the right membranousenvironment. Indeed, it might be easier toconceive experiments that disprove thehypothesis, putting the burden of proof inthe camp of the opponents.

As we mentioned above, PSs are alsoinvolved in the proteolytic processing (S3cleavage) of the Notch integral membranedomain (Fig. 1), which releases the Notchintracellular domain (NICD)49,50. Thisdomain hooks-up with a member of the

CSL (CBF1/Su(H)/Lag1) family of DNA-binding proteins and then acts as a tran-scription factor in the nucleus. The S3proteolytic cleavage of Notch is notobserved in PS-deficient cells, providing amolecular explanation for the predomi-nant Notch phenotype in PS-deficientanimals49,50. The parallels in processingbetween APP and Notch have led to thehypothesis51 that the APP intracellulardomain (AICD, a highly unstable frag-ment52) might be involved, like the NICD, intranscriptional regulation. Two recent piecesof indirect evidence support this hypothesis:the AICD is associated with nuclear frac-tions52; and it can form a transcriptionalactive complex with Fe65 and the histone

acetyltransferase Tip60, which drives expres-sion of a synthetic reporter gene53.

The fact that several inhibitors of APPprocessing by γ-secretase also inhibit Notchprocessing49 and that both Notch and APPare cleaved at similar positions in theirtransmembrane domains30,32 further sug-gest that similar proteolytic activities cleaveAPP and Notch, agreeing, in principle, withthe ‘PS is γ-secretase’ hypothesis. When theeffects of the afore-mentioned Asp 257mutation in PS on Notch and APP process-ing are compared, however, some discrep-ancies become apparent.

Whereas the mutation completely abol-ishes Notch cleavage, it maintains genera-tion of Aβ (although C-terminal fragments

C

COPII? (2)

? (1)

ER Golgi

? (6)

? (7)

PS

APP

Cholesterol?(3)

Plasma membrane/endosomes

Secretion

ADAM1O/TACE

N

CC

N

CN

NC

A βP3

? (5)

? (4)

CC

NC

N

C

N

CC

N

C

N

NC

BACEN

CC

N

NC

NC

C N

N

C

C N

NC

NC

N

C

Figure 3 APP trafficking and proteolytic processing. Despite a lot of work, we still do not understandwell the link between γ-secretase processing of APP and vesicular trafficking, as indicated in this fig-ure by the question marks. Presenilin is localized mainly in the ER but is also detected in the inter-mediate and cis-Golgi compartments, indicating that PS can indeed leave the ER40. Whether, andunder which conditions, PS can travel up to the cell surface has not been established. Also contro-versial is whether PS leaves the ER in complex with APP or not, although several groups have founddirect binding between PS and (immature, unglycosylated) APP78,79. We do not know whether bothproteins leave the ER in a complex (1) or in what type of vesicles they are transported (2). Do thePSs have a role like that of SCAP in the transport of APP and/or γ-secretase, or are they proteasessuch as S2P? Or can they perform both functions together? In and beyond the Golgi, APP becomes asubstrate for ADAM10, ADAM17 and BACE58,69 (Fig. 1). This cleavage is analogous to the S1P cleav-age of SREBP. Interestingly, the cholesterol content of the cell can influence the relative preponder-ance of α- or β-secretase cleavage. In contrast to the SREBP/SCAP system, this is most probably anindirect effect (3)80. At this point, the APP/PS system probably diverges completely from theSREBP/SCAP system. A fundamental issue is how the products generated by α- or β-secretase —that is, the APP-C83 and APP-C99 fragments — become cleaved by γ-secretase (4). At least a post-ER compartment is involved in this process, as simple retention of APP-C99 in the ER (where PSresides) is not sufficient to obtain Aβ66,81,82. As explained in the text, additional cofactors of the PS/γ-secretase complex clearly need to be identified (5). If PS is the catalytic subunit for γ-secretase andcan travel to the cell surface, as suggested by some observations67, this raises the issue of whetherPS can also travel back (6). Finally, probably one of the most important questions concerns the func-tion of the APP cytoplasmic C-terminal fragment (7). This fragment is highly instable but distributesinto the nucleus52, and seems to be involved in regulating gene transcription53.

© 2001 Macmillan Magazines Ltd

commentary

NATURE CELL BIOLOGY VOL 3 OCTOBER 2001 http://cellbio.nature.comE224

of APP accumulate54,55). If Asp 257 and Asp385 do constitute the catalytic site of PS, aspostulated by the hypothesis, then muta-tions of these residues should abolish APPand Notch processing to the same extent.There is clearly a contradiction betweenthis report54 and the original report ofWolfe et al.45, who found that Asp 257mutations abolished generation of Aβ. Onepossible concern with experiments usingdominant-negative forms of PS is thereplacement phenomenon mentionedabove. It is very difficult to control whetherall endogenously expressed wild-type PS iscompeted away by the transfected PS con-struct. Even small amounts of residualwild-type PS activity might cause impor-tant variations between experiments.Experiments using PS1 and PS2 double-negative cell lines37,38 will allow this issue tobe definitively settled.

The ‘PS is γ-secretase’ hypothesis, or atleast the concept that the same proteasecleaves both APP and Notch, has also beenchallenged by Petit et al.56. These authorsused a new class of α-chymotrypsininhibitors that decreased Aβ productionbut did not affect Notch processing. From acertain point of view, this result is interest-ing because it shows that compounds canbe generated that inhibit the production ofAβ without the detrimental side-effect ofshutting-down Notch signalling.

Nevertheless, it is difficult to rule out thepossibility that these new drugs have only anindirect impact on the γ-secretase proteolyticcleavage of APP, especially as relatively highconcentrations (100 µM) had to be used toobserve the effect on Aβ production. Indeed,in the past some misleading results have beenobtained with serine protease inhibitors suchas 4-(2-aminoethyl)-benzenesulfonyl fluo-ride. This compound was shown to interferewith Aβ production and was suggested to bean inhibitor of β-secretase57, but it has nowbeen established that β-secretase is anaspartyl protease58.

There is, however, also some other evi-dence to suggest that the proteases thatcleave Notch and APP are not completelyidentical. For instance, a single amino-acidsubstitution in the Notch sequence (Val1,744 to glycine) is sufficient to inhibitNotch processing59, whereas mutations inthe APP sequence usually cause small shiftsin the preferred cleavage site but not astrong inhibition of γ-secretase process-ing32,33,60. Therefore, the possibility that dif-ferent γ-secretase-like enzymes, acting inconcert with PS, are needed to cleave APPand Notch can clearly also explain the avail-able data. Definitive proof will require fur-ther experiments identifying these catalyticsubunits.

Similar problems are faced in a relatedfield of research investigating the regulat-ed intramembrane proteolysis of thesterol regulatory-element-binding protein

(SREBP)61,62. This transcription factor con-tains two transmembrane domains andresides in the ER in a hairpin conformation(Fig. 2). On cholesterol depletion, SREBPcleavage-activating protein (SCAP) escortsSREBP to the Golgi apparatus where SREBPbecomes cleaved by a subtilisin-like ‘site 1’protease (S1P)63. The remaining membrane-bound N-terminal SREBP fragmentbecomes then a substrate for the ‘site 2’ pro-tease (S2P) (Fig. 2)64.

S2P contains several hydrophobicdomains and a H-E-x-x-H sequence remi-niscent of a metalloprotease motif. Itcleaves the transmembrane domain of theSREBP fragment and releases eventually theN-terminal cytoplasmic tail of SREBP,which regulates the transcription of genesinvolved in lipid biosynthesis. Similar to PS,only indirect evidence is available that S2Pis a protease64. Recent data have shown thatATF6, which is a protein involved in theregulation of the unfolded proteinresponse, is also a substrate for S1P and S2P,consistent with the hypothesis that S2P isindeed a protease65. Interestingly, SCAP isnot involved in the processing of ATF6.

Finally, a brief comment on what wehave called previously the “spatial para-dox”51 — the observation that the subcellu-lar localization of PS does not completelyoverlap with the subcellular compartmentsin which γ-secretase processing occurs66.PS-dependent proteolysis of the integralmembrane domains of APP and Notch isbelieved to proceed mainly at the plasmamembrane or in the endocytic compart-ments, whereas the main pool of PS is asso-ciated with the ER (Fig. 3; and for a discus-sion see ref. 51).

Ray et al. 67 suggested a possible solutionto this paradox by demonstrating that smallamounts of PS1 can travel in complex withNotch to the cell surface. This could theo-retically also hold true for APP. But if thelimited amount of PS1 compared with itssubstrates is taken into account, as dis-cussed by Thinakaran et al. 68, it becomesvery difficult to understand why strongoverexpression of APP does not apparentlysaturate this system and lead to inefficientγ-secretase processing. Moreover, if Ray etal.67 are right, then it should be possible torelocate PS quantitatively from the ER tothe cell surface by simple overexpression ofNotch or APP in cell cultures. This has notbeen demonstrated so far. Finally, even ifone assumes that very little PS in the Golgiand at the cell surface is acting as the γ-sec-retase, then the issue of the function of theprincipal ‘inactive’ pool of PS in the ERneeds to be addressed40,66.

In conclusion, although the field hasmade considerable progress, much morework is needed before we will get the real pic-ture of the ‘presenilin elephant’. Further char-acterization of the PS-containing 150–250Kcomplexes9,12 is a priority in this respect.

Importantly, PSs are multifunctional pro-teins, and their involvement in γ-secretaseprocessing is likely to be only one aspect oftheir molecular function. For example, therole of PS1 in β-catenin turnover can bedissociated from its role in γ-secretase activ-ity by deleting the loop domain in PS1.Such a mutant does not bind or regulate theturnover of β-catenin, but maintains γ-sec-retase processing of APP and Notch17. It isclear that as yet we have only part of theinformation available, and definitive con-clusions concerning the nature of the ‘beast’should therefore be postponed until wehave more and better data clarifying theissues raised here.Center for Human Genetics, Neuronal Cell BiologyLaboratory, The K.U. Leuven and FlandersInteruniversity Institute for Biotechnology,Herestraat 49 3000 Leuven, Belgiume-mail: AD@med.kuleuven.ac.be

1. Wurtman, R. J. Alzheimer’s disease. Sci. Am. 252, 62–66 (1985).

2. Sherrington, R. et al. Cloning of a gene bearing missense muta-

tions in early-onset familial Alzheimer’s disease. Nature 375,

754–760 (1995).

3. Levy-Lahad, E. et al. Candidate gene for the chromosome 1

familial Alzheimer’s disease locus. Science 269, 973–977 (1995).

4. Rogaev, E. I. 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 376, 775–778 (1995).

5. Doan, A. et al. Protein topology of presenilin 1. Neuron 17,

1023–1030 (1996).

6. Li, X. & Greenwald, I. Additional evidence for an eight-trans-

membrane-domain topology for Caenorhabditis elegans and

human presenilins. Proc. Natl Acad. Sci. USA 95, 7109–7114

(1998).

7. Thinakaran, G. et al. Evidence that levels of presenilins (PS1

and PS2) are coordinately regulated by competition for limit-

ing cellular factors. J. Biol. Chem. 272, 28415–28422 (1997).

8. Thinakaran et al. Endoproteolysis of presenilin 1 and accumu-

lation of processed derivatives in vivo. Neuron 17, 181–190

(1996).

9. Capell, A. et al. 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. 273,

3205–3211 (1998).

10. Kim, T. W. et al. Endoproteolytic cleavage and proteasomal

degradation of presenilin 2 in transfected cells. J. Biol. Chem.

272, 11006–11010 (1997).

11. Yu, G. et al. The presenilin 1 protein is a component of a high

molecular weight intracellular complex that contains β-

catenin. J. Biol. Chem. 273, 16470–16475 (1998).

12. Yu, G. et al. Mutation of conserved aspartates affects matura-

tion of both aspartate mutant and endogenous presenilin 1 and

presenilin 2 complexes. J. Biol. Chem. 275, 27348–27353

(2000).

13. Steiner, H. et al. Expression of Alzheimer’s disease-associated

presenilin-1 is controlled by proteolytic degradation and com-

plex formation. J. Biol. Chem. 273, 32322–32331 (1998).

14. Yu, G. et al. Nicastrin modulates presenilin-mediated

notch/glp-1 signal transduction and βAPP processing. Nature

407, 48–54 (2000).

15. Zhou, J. et al. Presenilin 1 interaction in the brain with a novel

member of the Armadillo family. NeuroReport 8, 1489–1494

(1997).

16. Saura, C. A. et al. The nonconserved hydrophilic loop domain

of presenilin (PS) is not required for PS endoproteolysis or

enhanced Aβ42 production mediated by familial early onset

Alzheimer’s disease-linked PS variants. J. Biol. Chem. 275,

17136–17142 (2000).

17. Soriano, S. et al. Presenilin 1 negatively regulates β-catenin/T

cell factor/lymphoid enhancer factor-1 signaling independently

of β-amyloid precursor protein and Notch processing. J. Cell

Biol. 152, 785–794 (2001).

18. Selkoe, D. J. Notch and presenilins in vertebrates and inverte-

brates: implications for neuronal development and degenera-

tion. Curr. Opin. Neurobiol. 10, 50–57 (2000).

© 2001 Macmillan Magazines Ltd

commentary

NATURE CELL BIOLOGY VOL 3 OCTOBER 2001 http://cellbio.nature.com E225

19. Wong, P. C. et al. Presenilin 1 is required for Notch1 and DII1

expression in the paraxial mesoderm. Nature 387, 288–292

(1997).

20. Shen, J. et al. Skeletal and CNS defects in Presenilin-1-deficient

mice. Cell 89, 629–639 (1997).

21. Xia, X. et al. Loss of presenilin 1 is associated with enhanced β-

catenin signaling and skin tumorigenesis. Proc. Natl Acad. Sci.

(in the press). [AUTHOR: update?]

22. De Strooper, B. & Annaert, W. Where Notch and Wnt signaling

meet. The presenilin hub. J. Cell Biol. 152, F17–F20 (2001).

23. Vito, P., Lacana, E. & D’Adamio, L. Interfering with apoptosis:

Ca-binding protein Alg-2 and Alzheimer’s disease gene Alg-3.

Science 271, 521–525 (1996).

24. Katayama, T. et al. Presenilin-1 mutations downregulate the

signalling pathway of the unfolded-protein response. Nature

Cell Biol. 1, 479–485 (1999).

25. Niwa, M., Sidrauski, C., Kaufman, R. J. & Walter, P. A role for

presenilin-1 in nuclear accumulation of Ire1 fragments and

induction of the mammalian unfolded protein response. Cell

99, 691–702 (1999).

26. Leissring, M. A. et al. Capacitative calcium entry deficits and

elevated luminal calcium content in mutant presenilin-1

knockin mice. J. Cell Biol. 149, 793–798 (2000).

27. Yoo, A. S. et al. Presenilin-mediated modulation of capacitative

calcium entry. Neuron 27, 561–572 (2000).

28. Yu, H. et al. APP processing and synaptic plasticity in the adult

cerebral cortex of presenilin-1 conditional knock out mice.

Neuron (in the press).

29. Sato, N. et al. Upregulation of BiP and CHOP by the unfolded-

protein response is independent of presenilin expression.

Nature Cell Biol. 2, 863–870 (2000).

30. Gu, Y. et al. Distinct intramembrane cleavage of the β-amyloid

precursor protein family resembling γ-secretase-like cleavage of

Notch. J. Biol. Chem. 10.1074/jbc.C100357200 (2001).

31. Pinnix, I. et al. A novel γ-secretase assay based on detection of

the putative C-terminal fragment-γ of amyloid β protein pre-

cursor. J. Biol. Chem. 276, 481–487 (2001).

32. Sastre, M. et al. Presenilin-dependent γ-secretase processing of

β-Amyloid precursor protein at a site corresponding to the S3

cleavage of Notch. EMBO Rep. (in the press).

33. Lichtenthaler, S. F. et al. Mechanism of the cleavage specificity

of Alzheimer’s disease γ-secretase identified by phenylalanine-

scanning mutagenesis of the transmembrane domain of the

amyloid precursor protein. Proc. Natl Acad. Sci. USA 96,

3053–3058 (1999).

34. Scheuner, D. et al. Secreted amyloid β-protein similar to that in

the senile plaques of Alzheimer’s disease is increased in vivo by

the presenilin 1 and 2 and APP mutations linked to familial

Alzheimer’s disease. Nature Med. 2, 864–870 (1996).

35. De Strooper, B. et al. Deficiency of presenilin-1 inhibits the

normal cleavage of amyloid precursor protein. Nature 391,

387–390 (1998).

36. Naruse, S. et al. Effects of PS1 deficiency on membrane protein

trafficking in neurons. Neuron 21, 1213–1221 (1998).

37. Zhang, Z. et al. Presenilins are required for γ-secretase cleavage

of β-APP and transmembrane cleavage of Notch-1. Nature Cell

Biol. 2, 463–465 (2000).

38. Herreman, A. et al. Total inactivation of γ-secretase activity in

presenilin-deficient embryonic stem cells. Nature Cell Biol. 2,

461–462 (2000).

39. Armogida, M. et al. Endogeneous β-amyloid production in

presenilin-deficient embryonic mice fibroblasts. Nature Cell

Biol. (in the press).

40. Annaert, W. G. et al. Presenilin 1 controls γ-secretase process-

ing of amyloid precursor protein in pre-golgi compartments of

hippocampal neurons. J. Cell Biol. 147, 277–294 (1999).

41. Li, Y. M. et al. presenilin 1 is linked with γ-secretase activity in

the detergent solubilized state. Proc. Natl Acad. Sci. USA 97,

6138–6143 (2000).

42. Seiffert, D. et al. Presenilin-1 and -2 are molecular targets for γ-

secretase inhibitors. J. Biol. Chem. 275, 34086–34091 (2000).

43. Palacino, J. J. et al. Regulation of amyloid precursor protein

processing by presenilin 1 (PS1) and PS2 in PS1 knockout cells.

J. Biol. Chem. 275, 215–222 (2000).

44. Wolfe, M. S. et al. Peptidomimetic probes and molecular mod-

eling suggest that Alzheimer’s γ-secretase is an intramembrane-

cleaving aspartyl protease. Biochemistry 38, 4720–4727 (1999).

45. Wolfe, M. S. et al. Two transmembrane aspartates in presenilin-

1 required for presenelin endoproteolysis and γ-secretase activ-

ity. Nature 398, 513–517 (1999).

46. Steiner, H. et al. Glycine 384 is required for presenilin-1 func-

tion and is conserved in bacterial polytopic aspartyl proteases.

Nature Cell Biol. 2, 848–851 (2000).

47. Esler, W. P. et al. Transition-state analogue inhibitors of γ-sec-

retase bind directly to presenilin-1. Nature Cell Biol. 2, 428–434

(2000).

48. Li, Y. M. et al. Photoactivated γ-secretase inhibitors directed to

the active site covalently label presenilin 1. Nature 405,

689–694 (2000).

49. De Strooper, B. et al. A presenilin-1-dependent γ-secretase-like

protease mediates release of Notch intracellular domain.

Nature 398, 518–522 (1999).

50. Struhl, G. & Greenwald, I. Presenilin is required for activity

and nuclear access of Notch in Drosophila. Nature 398,

522–525 (1999).

51. Annaert, W. & De Strooper, B. Presenilins: molecular switches

between proteolysis and signal transduction. Trends Neurosci.

22, 439–443 (1999).

52. Cupers, P., Orlans, I., Craessaerts, K., Annaert, W. & De

Strooper, B. The amyloid precursor protein (APP)-cytoplasmic

fragment generated by γ-secretase is rapidly degraded but dis-

tributes partially in a nuclear fraction of neurones in culture. J.

Neurochem. 78, 1168–1178

53. Cao, X. & Sudhof, T. C. A transcriptively active complex of APP

with Fe65 and histone acetyltransferase Tip60. Science 293,

115–120 (2001).

54. Capell, A. et al. Presenilin-1 differentially facilitates endoprote-

olysis of the β-amyloid precursor protein and Notch. Nature

Cell Biol. 2, 205–211 (2000).

55. Morihara, T. et al. Absence of endoproteolysis but no effects on

amyloid β production by alternative splicing forms of prese-

nilin-1, which lack exon 8 and replace D257A. Brain Res. Mol.

Brain Res. 85, 85–90 (2000).

56. Petit, A. et al. New protease inhibitors prevent γ-secretase-

mediated production of Aβ40/42 without affecting Notch

cleavage. Nature Cell Biol. 3, 507–511 (2001).

57. Citron, M. et al. Inhibition of amyloid β-protein production in

neural cells by the serine protease inhibitor AEBSF. Neuron 17,

171–179 (1996).

58. Vassar, R. et al. β-secretase cleavage of Alzheimer’s amyloid

precursor protein by the transmembrane aspartic protease

BACE. Science 286, 735–741 (1999).

59. Huppert, S. S. et al. Embryonic lethality in mice homozygous

for a processing-deficient allele of Notch1. Nature 405,

966–970 (2000).

60. Murphy, M. P. et al. Presenilin 1 regulates pharmacologically

distinct γ-secretase activities. Implications for the role of prese-

nilin in γ-secretase cleavage. J. Biol. Chem. 275, 26277–26284

(2000).

61. Brown, M. S. & Goldstein, J. L. The SREBP pathway: regulation

of cholesterol metabolism by proteolysis of a membrane-

bound transcription factor. Cell 89, 331–340 (1997).

62. Brown, M. S., Ye, J., Rawson, R. B. & Goldstein, J. L. Regulated

intramembrane proteolysis: a control mechanism conserved

from bacteria to humans. Cell 100, 391–398 (2000).

63. Sakai, J. et al. Molecular identification of the sterol-regulated

luminal protease that cleaves SREBPs and controls lipid com-

position of animal cells. Mol. Cell 2, 505–514 (1998).

64. Rawson, R. B. et al. Complementation cloning of S2P, a gene

encoding a putative metalloprotease required for intramem-

brane cleavage of SREBPs. Mol. Cell 1, 47–57 (1997).

65. Ye, J. et al. ER stress induces cleavage of membrane-bound

ATF6 by the same proteases that process SREBPs. Mol. Cell 6,

1355–1364 (2000).

66. Cupers, P. et al. The discrepancy between presenilin subcellular

localization and γ-secretase processing of amyloid precursor

protein. J Cell Biol 154, 731–740 (2001).

67. Ray, W. J. et al. Cell surface presenilin-1 participates in the γ-

secretase-like proteolysis of notch. J. Biol. Chem. 274,

36801–36807 (1999).

68. Thinakaran, G. et al. Stable association of presenilin derivatives

and absence of presenilin interactions with APP. Neurobiol. Dis.

4, 438–453 (1998).

69. Lammich, S. et al. Constitutive and regulated α-secretase cleav-

age of Alzheimer’s amyloid precursor protein by a disintegrin

metalloprotease. Proc. Natl Acad. Sci. USA 96, 3922–3927

(1999).

70. Buxbaum, J. D. et al. Evidence that tumor necrosis factor αconverting enzyme is involved in regulated α-secretase cleavage

of the Alzheimer amyloid protein precursor. J. Biol. Chem. 273,

27765–27767 (1998).

71. Struhl, G. & Adachi, A. Requirements for presenilin-dependent

cleavage of notch and other transmembrane proteins. Mol. Cell

6, 625–636 (2000).

72. Logeat, F. et al. The Notch1 receptor is cleaved constitutively by

a furin-like convertase. Proc. Natl Acad. Sci. USA 95, 8108–8112

(1998).

73. Brou, C. et al. A novel proteolytic cleavage involved in Notch

signaling: the role of the disintegrin-metalloprotease TACE.

Mol. Cell 5, 207–216 (2000).

74. Mumm, J. S. et al. A ligand-induced extracellular cleavage regu-

lates γ-secretase-like proteolytic activation of Notch1. Mol. Cell

5, 197–206 (2000).

75. Yang, T., Goldstein, J. L. & Brown, M. S. Overexpression of

membrane domain of SCAP prevents sterols from inhibiting

SCAP.SREBP exit from endoplasmic reticulum. J. Biol. Chem.

275, 29881–29886 (2000).

76. Nohturfft, A., Yabe, D., Goldstein, J. L., Brown, M. S. &

Espenshade, P. J. Regulated step in cholesterol feedback local-

ized to budding of SCAP from ER membranes. Cell 102,

315–323 (2000).

77. Duncan, E. A., Dave, U. P., Sakai, J., Goldstein, J. L. & Brown,

M. S. Second-site cleavage in sterol regulatory element-binding

protein occurs at transmembrane junction as determined by

cysteine panning. J. Biol. Chem. 273, 17801–17809 (1998).

78. Xia, W. et al. Presenilin complexes with the C-terminal fragments

of amyloid precursor protein at the sites of amyloid β-protein

generation. Proc. Natl Acad. Sci. USA 97, 9299–9304 (2000).

79. Weidemann, A. et al. Formation of stable complexes between

two Alzheimer’s disease gene products: presenilin-2 and β-

amyloid precursor protein. Nature Med. 3, 328–332 (1997).

80. Fassbender, K. et al. Simvastatin strongly reduces levels of

Alzheimer’s disease β-amyloid peptides Aβ42 and Aβ40 in vitro

and in vivo. Proc. Natl Acad. Sci. USA 98, 5856–5861 (2001).

81. Maltese, W. A. et al. Retention of the Alzheimer’s amyloid pre-

cursor protein fragment C99 in the endoplasmic reticulum

prevents formation of amyloid β-peptide. J. Biol. Chem. 276,

20267–20279 (2001).

82. Iwata, H., Tomita, T., Maruyama, K. & Iwatsubo, T. Subcellular

compartment and molecular subdomain of β-amyloid precur-

sor protein relevant to the Aβ42-promoting effects of

Alzheimer mutant presenilin 2. J. Biol. Chem. 276,

21678–21685 (2001).

ACKNOWLEDGEMENTS

We thank the many researchers who provided information before

publication and all members of our laboratory for their intellectu-

al input. We acknowledge the financial support of the Bayer

Research Network, the Flanders Interuniversity Institute for

Biotechnology, The Human Frontier of Science Program, the

National Fund for Scientific Research–Flanders and the Katholieke

Universiteit Leuven.

Correspondence should be addressed to B.D.S

© 2001 Macmillan Magazines Ltd