gamma-Secretase complexes containing N- and C-terminal fragments of different presenilin origin retain normal gamma-secretase activity

The gamma-secretase complex processes substrate proteins within membranes and consists of four proteins: presenilin (PS), nicastrin, Aph-1 and Pen-2. PS harbours the enzymatic activity of the complex, and there are two mammalian PS homologues: PS1 and PS2. PS undergoes endoproteolysis, generating the N- and C-terminal fragments, NTF and CTF, which represent the active species of PS. To characterize the functional similarity between complexes of various PS composition, we analysed PS1, PS2, and chimeric PS composed of the NTF from PS1 and CTF from PS2, or vice versa, in assembly and function of the gamma-secretase complex. Chimeric PSs, like PS1 and PS2, undergo normal endoproteolysis when introduced into cells devoid of endogenous PS. Furthermore, PS2 CTF can, at least partially, restore processing in a truncated PS1, which cannot undergo endoproteolysis. All PS forms enable maturation of nicastrin and cleave full length Notch receptors, indicating that both PS1 and PS2 are present at the cell surface. Finally, when co-introduced as separate molecules, NTF and CTF of different PS origin reconstitute gamma-secretase activity. In conclusion, these data show that endoproteolysis, NTF-CTF interactions, and the assembly and activity of gamma-secretase complexes are very conserved between PS1 and PS2.     

Conserved residues within the putative active site of gamma-secretase differentially influence enzyme activity and inhibitor binding

Gamma-secretase performs the final processing step in the generation of amyloid-beta (Abeta) peptides, which are believed to be causative for Alzheimer's disease. Presenilins (PS) are required for gamma-secretase activity and the presence of two essential intramembranous aspartates (D257 and D385) has implicated this region as the putative catalytic centre of an aspartyl protease. The presence of several key hydrogen-bonding residues around the active site of classical aspartyl proteases led us to investigate the role of both the critical aspartates and two nearby conserved hydrogen bond donors in PS1. Generation of cell lines stably overexpressing the D257E, D385E, Y256F and Y389F engineered mutations has enabled us to determine their role in enzyme catalysis and binding of a transition state analogue gamma-secretase inhibitor. Here we report that replacement of either tyrosine residue alters gamma-secretase cleavage specificity, resulting in an increase in the production of the more pathogenic Abeta42 peptide in both cells and membranous enzyme preparations, without affecting inhibitor binding. In contrast, replacement of either of the aspartate residues precludes inhibitor binding in addition to inactivation of the enzyme. Together, these data further incriminate the region around the intramembranous aspartates as the active site of the enzyme, targeted by transition state analogue inhibitors, and highlight the roles of individual residues.     

Length and overall sequence of the PEN-2 C-terminal domain determines its function in the stabilization of presenilin fragments

Gamma-secretase is an aspartyl protease complex that catalyzes the intramembrane cleavage of a subset of type I transmembrane proteins including the beta-amyloid precursor protein (APP) implicated in Alzheimer's disease. Presenilin (PS), nicastrin (NCT), anterior pharynx defective (APH-1) and presenilin enhancer-2 (PEN-2) constitute the active gamma-secretase complex. PEN-2, the smallest subunit, is required for triggering PS endoproteolysis. Stabilization of the resultant N- and C-terminal fragments, which carry the catalytically active site aspartates, but not endoproteolysis itself, requires the C-terminal domain of PEN-2. To functionally dissect the C-terminal domain we created C-terminal deletion mutants and mutagenized several evolutionary highly conserved residues. The PEN-2 mutants were then probed for functional complementation of a PEN-2 knockdown, which displays deficient PS1 endoproteolysis and impaired NCT maturation. Progressive truncation of the C-terminus caused increasing loss of function. This was also observed for an internal deletion mutant as well as for C-terminally tagged PEN-2 with a twofold elongated C-terminal domain. Interestingly, only simultaneous, but not individual substitution of the highly conserved D90, F94, P97 and G99 residues with alanine interfered with PEN-2 function. All loss of function mutants identified allowed PS1 endoproteolysis, but failed to stably associate with the resultant PS1 fragments, which like the PEN-2 loss of function mutants underwent proteasomal degradation. However, complex formation of the PEN-2 mutants with PS1 fragments could be recovered when proteasomal degradation was blocked. Taken together, our data suggest that the PS-subunit stabilizing function of PEN-2 depends on length and overall sequence of its C-terminal domain.     

Co-expressed presenilin 1 NTF and CTF form functional gamma-secretase complexes in cells devoid of full-length protein

The enzyme gamma-secretase catalyzes the intramembrane proteolytic cleavage that generates the amyloid beta-peptide from the beta-amyloid precursor protein. The presenilin (PS) protein is one of the four integral membrane protein components of the mature gamma-secretase complex. The PS protein is itself subjected to endoproteolytic processing, generating stable N- and C-terminal fragment (NTF and CTF, respectively) heterodimers. Here we demonstrate that coexpression of PS1 NTF and CTF functionally mimics expression of the full-length PS1 protein and restores gamma-secretase activity in PS-deficient mammalian cells. The coexpressed fragments re-associate with each other inside the cell, where they also interact with nicastrin, another gamma-secretase complex component. Analysis of gamma-secretase activity following the expression of mutant forms of NTF and CTF, under conditions bypassing endoproteolysis, indicated that the putatively catalytic Asp257 and Asp385 residues have a direct effect on gamma-secretase activity. Moreover, we demonstrate that expression of the wild-type CTF rescues endoproteolytic cleavage of C-terminally truncated PS1 molecules that are otherwise uncleaved and inactive. Recovery of cleavage is critically dependent on the integrity of Asp385. Taken together, our findings indicate that ectopically expressed NTF and CTF restore functional gamma-secretase complexes and that the presence of full-length PS1 is not a requirement for proper complex assembly.     

A domain at the C-terminus of PS1 is required for presenilinase and gamma-secretase activities

The structural requirements for presenilin (PS) to produce active presenilinase and gamma-secretase enzymes are poorly understood. Here we investigate the role the cytoplasmic C-terminal region of PS1 plays in PS1 activity. Deletion or addition of residues at the PS C-terminus has been reported to inhibit presenilinase endoproteolysis of PS and alter gamma-secretase activity. In this study, we use a sensitive assay in PS1/2KO MEFs to define a domain at the extreme C-terminus of PS1 that is essential for both presenilinase and gamma-secretase activities. Progressive deletion of the C-terminus demonstrated that removal of nine residues produces a PS1 molecule (458ST) that lacks both presenilinase processing and gamma-secretase cleavage of Notch and APP substrates. In contrast, removal of four or five residues had no effect (462ST, 463ST), while intermediate truncations partially inhibited PS1 activity. The 458ST mutant was unable to replace endogenous wtPS1 in HEK293 cells. Although 458ST was able to form a gamma-secretase complex, this complex was not matured, illustrated by mutant PS1 instability, lack of endoproteolysis, and little production of mature Nicastrin. These data indicate that the C-terminal end of PS1 is essential for Nicastrin trafficking and modification as well as the replacement of endogenous PS1 by PS1 transgenes.     

Two domains within the first putative transmembrane domain of presenilin 1 differentially influence presenilinase and gamma-secretase activity

Presenilins (PS) are thought to contain the active site for presenilinase endoproteolysis of PS and gamma-secretase cleavage of substrates. The structural requirements for PS incorporation into the gamma-secretase enzyme complex, complex stability and maturation, and appropriate presenilinase and gamma-secretase activity are poorly understood. We used rescue assays to identify sequences in transmembrane domain one (TM1) of PS1 required to support presenilinase and gamma-secretase activities. Swap mutations identified an N-terminal TM1 domain that is important for gamma-secretase activity only and a C-terminal TM1 domain that is essential for both presenilinase and gamma-secretase activities. Exchange of residues 95-98 of PS1 (sw95-98) completely abolishes both activities while the familial Alzheimer's disease mutation V96F significantly inhibits both activities. Reversion of residue 96 back to valine in the sw95-98 mutant rescues PS function, identifying V96 as the critical residue in this region. The TM1 mutants do not bind to an aspartyl protease transition state analog gamma-secretase inhibitor, indicating a conformational change induced by the mutations that abrogates catalytic activity. TM1 mutant PS1 molecules retain the ability to interact with gamma-secretase substrates and gamma-secretase complex members, although Nicastrin stability is decreased by the presence of these mutants. gamma-Secretase complexes that contain V96F mutant PS1 molecules display a partial loss of function for gamma-secretase that alters the ratio of amyloid-beta peptide species produced, leading to the amyloid-beta peptide aggregation that causes familial Alzheimer's disease.     

Zebrafish lacking Alzheimer presenilin enhancer 2 (Pen-2) demonstrate excessive p53-dependent apoptosis and neuronal loss

Gamma-secretase cleavage, mediated by a complex of presenilin, presenilin enhancer (Pen-2), nicastrin, and Aph-1, is the final proteolytic step in generating amyloid beta protein found in brains of Alzheimer's disease patients and Notch intracellular domain critical for proper neuronal development. Here, we employ the zebrafish model to study the role of Pen-2 in neuronal survival. We found that (i) knockdown of Pen-2 using antisense morpholino led to a reduction of islet-1 positive neurons, (ii) Notch signaling was reduced in embryos lacking Pen-2 or other gamma-secretase components, (iii) neuronal loss in Pen-2 knockdown embryos is not as a result of a lack of neuronal precursor cells or cell proliferation, (iv) absence of Pen-2 caused massive apoptosis in the whole animal, which could be suppressed by simultaneous knockdown of the tumor suppressor p53, (v) loss of islet-1 or acetylated tubulin positive neurons in Pen-2 knockdown embryos could be partially rescued by knockdown of p53. Our results demonstrate that knockdown of Pen-2 directly induces a p53-dependent apoptotic pathway that contributes to neuronal loss and suggest that Pen-2 plays an important role in promoting neuronal cell survival and protecting from apoptosis in vivo.     

Immature nicastrin stabilizes APH-1 independent of PEN-2 and presenilin: identification of nicastrin mutants that selectively interact with APH-1

Gamma-secretase is a high molecular mass aspartyl protease complex composed of presenilin (PS1 or PS2), nicastrin (Nct), anterior pharynx-defective-1 (APH-1) and presenilin enhancer-2 (PEN-2). The complex mediates the intramembraneous proteolysis of beta-secretase cleaved beta-amyloid precursor protein (APP) leading to the secretion of the Alzheimer's disease-associated amyloid beta-peptide (Abeta). In order to dissect functionally important domains of Nct required for gamma-secretase complex assembly, maturation, and activity we mutated evolutionary conserved amino acids. The mutant Nct variants were expressed in a cellular background with significantly reduced endogenous Nct. Mutant Nct was functionally investigated by its ability to restore PS, APH-1 and PEN-2 expression as well as by monitoring the accumulation of the APP C-terminal fragments, the immediate substrates of gamma-secretase. We identified three independent mutations within the ectodomain of Nct, which rescued expression of APH-1 but not of PEN-2 or PS and thus failed to restore gamma-secretase activity. Interestingly, these immature Nct variants selectively bound to APH-1, suggesting a stable Nct/APH-1 interaction independent of PS and PEN-2. Consistent with this finding, expression of APH-1 remained largely unaffected in the PS double knock-out and immature Nct co-immunoprecipitated with APH-1 in the absence of PS and PEN-2. Taken together, our findings suggest that immature Nct can stably interact with APH-1 to form a potential scaffold for binding of PS and PEN-2. Moreover, binding of the latter two complex partners critically depends on the integrity of the Nct ectodomain.     

A presenilin-independent aspartyl protease prefers the gamma-42 site cleavage

beta-Amyloid peptides (Abeta40 and Abeta42) are the major constituents of amyloid plaques, which are one of the hallmarks of Alzheimer's disease (AD). The Abeta is derived from sequential cleavages of amyloid precursor protein (APP) by beta- and gamma-secretases. gamma-Secretase consists of at least four proteins where presenilins (PS1 and PS2 or PS) are the catalytic subunit involved in the gamma-site cleavage of APP. Secretion of both Abeta40 and Abeta42 is significantly reduced in PS1 knock-out cells and completely abolished in cells deficient for both PS1 and PS2. Consequently, both the PS proteins play essential roles in the production of the secretory of Abeta from cells. Recent studies in primary neurons, however, suggest that PSs are not required for intracellular Abeta42 accumulation; thus the intracellular Abeta42 appears to be generated in a PS-independent manner. Here we present the first biochemical evidence indicating that Abeta, especially Abeta42, can be generated in the absence of PS based on an in vitrogamma-secretase assay employing membranes prepared from PS-deficient Blastocyst-derived (BD) cells. This PS-independent gamma-secretase (PSIG) activity is sensitive to the changes in pH and displays an optimal activity at pH 6.0. Pepstatin A is a potent inhibitor for this proteolytic activity with IC50 of 1.2 nm and 0.4 nm for Abeta40 and Abeta42 generation, respectively. These results indicate that these PS-independent gamma-site cleavages are mediated by an aspartyl protease. More importantly, the PSIG activity displays a distinct preference in mediating the 42-site cleavage over the 40-site cleavage, thereby generating Abeta42 as the predominant product.     

Syntaxin 5 interacts specifically with presenilin holoproteins and affects processing of betaAPP in neuronal cells

The specific roles of syntaxin 5 (Syx 5) in the interaction with presenilin (PS) and the accumulation of beta-amyloid precursor protein (betaAPP), as well as the secretion of beta-amyloid peptide (Abeta peptide) were examined in NG108-15 cells. Syx 5, which localizes from the endoplasmic reticulum (ER) to the Golgi, bound to PS holoproteins, while the other Syxs studied did not. Among familial Alzheimer's disease (FAD)-linked PS mutants, PS1deltaE9, which lacks the endoproteolytic cleavage site, showed markedly decreased binding to Syx 5. The interaction domains in Syx 5 were mapped to the transmembrane region and to the cytoplasmic region containing the alpha-helical domains, which are distinct from the H3 (SNARE motif). Among all of the Syxs examined, only overexpression of Syx 5 resulted in the accumulation of betaAPP in the ER to cis-Golgi compartment, an attenuation of the amount of the C-terminal fragment (APP-CTF) of betaAPP, and a reduction in the secretion of Abeta peptides. Furthermore, co-expression of Syx 5 with C99 resulted in an increase in APP-CTF and suppressed Abeta secretion. Taken together, these results indicate that Syx 5 may play a specific role in the modulation of processing and/or trafficking of FAD-related proteins in neuronal cells by interaction with PS holoproteins in the early secretory compartment of neuronal cells.     

The same gamma-secretase accounts for the multiple intramembrane cleavages of APP

It has been hypothesized that different C-terminus of beta-amyloid peptide (Abeta) may be generated by different gamma-secretase activities. Recently, we have identified a new zeta-cleavage site at Abeta46, leading to an important finding that the C-terminus of Abeta is produced by a series of sequential cleavages. This finding prompted us to examine the effects of the known gamma-secretase inhibitors on different steps of the gamma-secretase-mediated sequential cleavages and specifically their effects on the formation and turnover of the intermediate Abeta(46). Our results demonstrate that some of the known inhibitors, such as L-685,458 and III-31C as well as inhibitors IV and V, inhibit the formation of secreted Abeta(40/42) by inhibiting the formation of the intermediate Abeta(46). However, most of the other inhibitors show no inhibitory effect on the formation of the intermediate Abeta(46), but rather inhibit the turnover of Abeta(46), resulting in its accumulation. In addition, the non-steroidal anti-inflammatory drugs (NSAIDs) ibuprofen and sulindac sulfide have no effect on the formation and turnover of Abeta(46), but rather modulate the ratio of secreted Abeta at a step after the formation of Abeta(40) and Abeta(42). Thus, our data strongly suggest that the multi-sequential intramembrane cleavages of amyloid precursor protein C (APP) are likely catalyzed by the same gamma-secretase.     

Presenilin 2 familial Alzheimer's disease mutations result in partial loss of function and dramatic changes in Abeta 42/40 ratios

Gene knockout studies in mice suggest that presenilin 1 (PS1) is the major gamma-secretase and that it contributes disproportionately to amyloid beta (Abeta) peptide generation from beta-amyloid precursor protein (APP), whereas PS2 plays a more minor role. Based on this and other observations we hypothesized that familial Alzheimer's disease (FAD) mutations in PS2 would have a dramatic effect on function in order to have an observable effect on Abeta levels in the presence of normal PS1 alleles. Only four of the eight reported FAD mutations in PS2 have altered function in vitro suggesting that the other variants represent rare polymorphisms rather than disease-causing mutations. In support of our hypothesis, the four verified PS2 FAD mutations cause substantial changes in the Abeta 42/40 ratio, comparable with PS1 mutations that cause very-early-onset FAD. Most of the PS2 mutations also cause a significant decrease in Abeta 40, APP C-terminal fragment (CTF)gamma and Notch intracellular domain (NICD) production suggesting that they are partial loss of function mutations. PS2 M239V, its PS1 homolog M233V, and other FAD mutations within transmembrane (TM) 5 of PS1 differentially affect CTFgamma and NICD production suggesting that TM5 of PS are important for gamma-secretase cleavage of APP but not Notch.     

Non-steroidal anti-inflammatory drugs (NSAIDs) in Alzheimer's disease: old and new mechanisms of action

Alzheimer's disease (AD) is characterized by cerebral deposits of beta-amyloid (A beta) peptides and neurofibrillary tangles (NFT) which are surrounded by inflammatory cells. Epidemiological studies have shown that prolonged use of non-steroidal anti-inflammatory drugs (NSAIDs) reduces the risk of developing AD and delays the onset of the disease. It has been postulated that some NSAIDs target pathological hallmarks of AD by interacting with several pathways, including the inhibition of cyclooxygenases (COX) and activation of the peroxisome proliferator-activated receptor gamma. A variety of experimental studies indicate that a subset of NSAIDs such as ibuprofen, flurbiprofen, indomethacin and sulindac also possess A beta-lowering properties in both AD transgenic mice and cell cultures of peripheral, glial and neuronal origin. While COX inhibition occurs at low concentrations in vitro (nM-low microm range), the A beta-lowering activity is observed at high concentrations (< or = 50 microm). Nonetheless, studies with flurbiprofen or ibuprofen in AD transgenic mice show that the effects on A beta levels or deposition are attained at plasma levels similar to those achieved in humans at therapeutic dosage. Still, it remains to be assessed whether adequate concentrations are reached in the brain. This is a crucial aspect that will allow defining the dose-window and the length of treatment in future clinical trials. Here, we will discuss how the combination of anti-amyloidogenic and anti-inflammatory activities of certain NSAIDs may produce a profile potentially relevant to their clinical use as disease-modifying agents for the treatment of AD.     

Presenilin function and gamma-secretase activity

Alzheimer's disease (AD) is the most common form of dementia and is characterized pathologically by the accumulation of beta-amyloid (Abeta) plaques and neurofibrillary tangles in the brain. Genetic studies of AD first highlighted the importance of the presenilins (PS). Subsequent functional studies have demonstrated that PS form the catalytic subunit of the gamma-secretase complex that produces the Abeta peptide, confirming the central role of PS in AD biology. Here, we review the studies that have characterized PS function in the gamma-secretase complex in Caenorhabditis elegans, mice and in in vitro cell culture systems, including studies of PS structure, PS interactions with substrates and other gamma-secretase complex members, and the evidence supporting the hypothesis that PS are aspartyl proteases that are active in intramembranous proteolysis. A thorough knowledge of the mechanism of PS cleavage in the context of the gamma-secretase complex will further our understanding of the molecular mechanisms that cause AD, and may allow the development of therapeutics that can alter Abeta production and modify the risk for AD.     

Amyloid precursor protein and Notch intracellular domains are generated after transport of their precursors to the cell surface

Alzheimer's disease is characterized by brain deposition of extracellular amyloid beta-peptide (Abeta)-containing plaques. The cellular site of gamma-secretase activity, which releases Abeta and the corresponding amyloid precursor protein intracellular domain (AICD), remains controversial. Proposed cleavage sites range from the endoplasmic reticulum (ER), the Golgi apparatus, and the cell surface to endosomal compartments. We now used C99-green fluorescent protein (GFP), a fluorescent reporter substrate for gamma-secretase activity and monitored AICD production in living cells. C99-GFP is efficiently cleaved by gamma-secretase, and AICD-GFP is released into the cytosol. Inhibiting gamma-secretase results in accumulation of C99-GFP in early endosomes. By blocking selective transport steps along the secretory pathway, we demonstrate that gamma-secretase does not cleave its substrates in the ER, the Golgi/trans-Golgi network, or in secretory vesicles. In contrast, inhibition of endocytosis did not inhibit cleavage of C99-GFP. Similar results were obtained for another gamma-secretase substrate, NotchDeltaE. Our results suggest that intracellular domains are generated by gamma-secretase at the plasma membrane and/or early endosomes.     

The presenilin C-terminus is required for ER-retention, nicastrin-binding and gamma-secretase activity

gamma-Secretase is an intramembrane cleaving protease involved in Alzheimer's disease. gamma-Secretase occurs as a high molecular weight complex composed of presenilin (PS1/2), nicastrin (NCT), anterior pharynx-defective phenotype 1 and PS enhancer 2. Little is known about the cellular mechanisms of gamma-secretase assembly. Here we demonstrate that the cytoplasmic tail of PS1 fulfills several functions required for complex formation, retention of unincorporated PS1 and gamma-secretase activity. The very C-terminus interacts with the transmembrane domain of NCT and may penetrate into the membrane. Deletion of the last amino acid is sufficient to completely block gamma-secretase assembly and release of PS1 from the endoplasmic reticulum (ER). This suggests that unincorporated PS1 is actively retained within the ER. We identified a hydrophobic stretch of amino acids within the cytoplasmic tail of PS1 distinct from the NCT-binding site, which is required to retain unincorporated PS1 within the ER. Deletion of the retention signal results in the release of PS1 from the ER and the assembly of a nonfunctional gamma-secretase complex, suggesting that at least a part of the retention motif may also be required for the function of PS1.     

SEL-10 interacts with presenilin 1, facilitates its ubiquitination,and alters A-beta peptide production

Mutations in the human presenilin genes (PS1 or PS2) have been linked to autosomal dominant, early onset Alzheimer's disease (AD). Presenilins, probably as an essential part of gamma-secretase, modulate gamma-cleavage of the amyloid protein precursor (APP) to the amyloid beta-peptide (Abeta). Mutations in sel-12, a Caenorhabditis elegans presenilin homologue, cause a defect in egg laying that can be suppressed by loss of function mutations in a second gene, SEL-10. SEL-10 protein is a homologue of yeast Cdc4, a member of the SCF (Skp1-Cdc53/CUL1-F-box protein) E2-E3 ubiquitin ligase family. In this study, we show that human SEL-10 interacts with PS1 and enhances PS1 ubiquitination, thus altering cellular levels of unprocessed PS1 and its N- and C-terminal fragments. Co-transfection of sel-10 and APP cDNAs in HEK293 cells leads to an alteration in the metabolism of APP and to an increase in the production of amyloid beta-peptide, the principal component of amyloid plaque in Alzheimer's disease.     

A non-amyloidogenic function of BACE-2 in the secretory pathway

beta-Site amyloid precursor protein cleavage enzyme (BACE)-1 and BACE-2 are members of a novel family of membrane-bound aspartyl proteases. While BACE-1 is known to cleave beta-amyloid precursor protein (betaAPP) at the beta-secretase site and to be required for the generation of amyloid beta-peptide (Abeta), the role of its homologue BACE-2 in amyloidogenesis is less clear. We now demonstrate that BACE-1 and BACE-2 have distinct specificities in cleavage of betaAPP in cultured cells. Radiosequencing of the membrane-bound C-terminal cleavage product revealed that BACE-2 cleaves betaAPP in the middle of the Abeta domain between phenylalanines 19 and 20, resulting in increased secretion of APPs-alpha- and p3-like products and reduced production of Abeta species. This cleavage can occur in the Golgi and later secretory compartments. We also demonstrate that BACE-1-mediated cleavage of betaAPP at Asp1 of the Abeta domain can occur as early as in the endoplasmic reticulum, while cleavage at Glu11 occurs in later compartments. These data indicate that the distinct specificities of BACE-1 and BACE-2 in their cleavage of betaAPP differentially affect the generation of Abeta.     

Analysis of transmembrane domain mutants is consistent with sequential cleavage of Notch by gamma-secretase

gamma-Secretase is a lipid-embedded, intramembrane-cleaving aspartyl protease that cleaves its substrates twice within their transmembrane domains (TMD): once near the cytosolic leaflet (at S3/epsilon) and again in the middle of the TMD (at S4/gamma). To address whether this unusual process occurs in two independent or interdependent steps, we investigated how mutations at the S3/epsilon site in Notch1-based substrates impact proteolysis. We demonstrate that such mutations greatly inhibit not only gamma-secretase-mediated cleavage at S3 but also at S4, independent of their impact on NICD stability. These results, together with our previous observations, suggest that hydrolysis at the center of the Notch transmembrane domain (S4/gamma) is dependent on the S3/epsilon cleavage. Notch (and perhaps all gamma-secretase substrates) may be cleaved by sequential proteolysis starting at S3.