γ-Secretase Catalyzes Sequential Cleavages of the AβPP Transmembrane Domain

Abstract

The biogenesis of the amyloid-β peptide (Aβ) is a central issue in Alzheimer's disease (AD) research. Aβ is produced by β- and γ-secretases from the amyloid-β protein precursor (AβPP). These proteases are targets for the development of therapeutic compounds to downregulate Aβ production. γ-secretase has received more attention 1) because it generates the C-terminus of Aβ, which is important in the pathogenesis of AD because the longer Aβ species are more amyloidogenic, and 2) because it cleaves AβPP within its transmembrane domain. In the understanding the mechanism of γ-secretase cleavage, three major cleavage sites have been identified, namely, γ-cleavage site at Aβ_40/42, ζ-cleavage site at Aβ₄₆, and ε-cleavage site at Aβ₄₉. Moreover, the novel finding that some of the known γ-secretase inhibitors inhibit the formation of secreted Aβ₄₀ and Aβ₄₂, but cause an intracellular accumulation of long Aβ₄₆, provided information extremely important for the development of strategies aimed at the design of γ-secretase inhibitors to prevent and treat AD. In addition, it has been established that the C-terminus of Aβ is generated by a series of sequential cleavages: first, ε-cleavage, followed by ζ-cleavage and finally by γ-cleavage, commencing from the membrane boundary to the middle of the AβPP membrane domain.

INTRODUCTION

Alzheimer's disease (AD) is the most common type of dementia in the elderly population. AD is clinically characterized by progressive dementia, including loss of memory and the ability to learn and think. Pathologically, it is characterized by degeneration of neurons and accumulation of abnormal protein structures in the brain. This abnormal accumulation includes intracellular deposition of neurofibrillary tangles (NFT) that are composed of the microtubule-associated protein tau, and extracellular deposits of various types of amyloid-β (Aβ) in senile plaques. Aβ is a mix of peptides of 39 to 43 amino acids in length and is proteolytically produced from a large amyloid-β protein precursor (AβPP). Several lines of evidence support the hypothesis that progressive accumulation of Aβ is an early and critical event in the pathogenesis of AD [1]. Studies have revealed that the accumulation of Aβ initiates a series of downstream neurotoxic events, including synaptic failure [2] and hyperphosphorylation of tau, which results in neuronal dysfunction and death [3]. The strongest support for this “amyloid cascade” hypothesis comes from observations of molecular genetic studies. First, patients with Down syndrome, who possess an extra copy of chromosome 21 on which the AβPP gene is located, invariably develop AD [4]. Second, familial forms of AD (FAD) are linked to several missense mutations in the AβPP gene that result in increased production of total Aβ or of Aβ₄₂, which is the more amyloidogenic form [5–7]. Furthermore, FAD-linked mutations in the other two AD genes, presenilin 1 (PS1) [8] and presenilin 2 (PS2) [9,10], also cause abnormal production of Aβ, mostly resulting in an increase in the ratio of Aβ₄₂ to Aβ₄₀. Third, genetically engineered mice that carry the human FAD genes develop amyloid plaques as well as symptoms that mimic those of human disease [11]. The relationship of NFT and senile plaques with clinical disease has long been debated. There is an emerging appreciation that these changes themselves may not have a pathogenic or etiological role in AD, but are more likely by-products of a preceding, primary pathobiologic process that results in their accumulation [12,13]. However, evidence now suggests that the soluble forms of Aβ oligomers, rather than the plaques, are indeed the principal “toxic intermediates” of synapse loss and neuronal injury [13, 14]. Thus, elucidating how Aβ oligomers are generated is essential for understanding AD and for developing strategies to prevent and treat this disease. For instance, such tactics might include development of small molecules that inhibit one or another secretase involved in AβPP processing and Aβ formation over long periods of time.

SECRETASES INVOLVED IN Aβ PP PROCESSING AND Aβ FORMATION

Aβ is proteolytically derived from AβPP, which is a ubiquitously expressed, type I transmembrane protein [15–20]. AβPP can be processed via two alternative pathways, the amyloidogenic pathway and the non-amyloidogenic pathway (Fig. 1) [21]. In the nonamyloidogenic pathway, AβPP is cleaved within the Aβ sequence by α-secretase, resulting in the release of a soluble ectodomain, sAβPPα, and a membrane-anchored C-terminal fragment, CTFα. Since α-cleavage occurs within the Aβ sequence, it precludes the formation of full-length Aβ. Alternatively, in the amyloidogenic pathway, AβPP is cleaved at the N-terminal of the Aβ sequence by β-secretase to produce a soluble sAβPPβ and a membrane-anchored C-terminal fragment, CTFβ. Both CTFα and CTFβ are subsequently cleaved within their transmembrane domain by γ-secretase to produce the short peptide p3 from CTFα, and the full-length Aβ from CTFβ,inaddition to the release of the AβPP intracellular domain (AICD) from both CTFα and CTFβ [22]. The Aβ peptides generated in the amyloidogenic pathway vary in length from 38 to 43 amino acids, as detected in conditioned medium and brains of transgenic mice [23,24], with Aβ₄₀ as the predominant secreted Aβ species. Although the longer forms of Aβ, mostly Aβ₄₂, are the minor species, it has been shown that they are more hydrophobic and particularly prone to aggregate and form amyloid fibrils [25,26]. In addition, longer forms of Aβ are believed to play a pivotal role in the nucleation of amyloid plaques [27]. In contrast, p3 peptide produced from CTFα in the non-amyloidogenic pathway is not involved in amyloidogenesis [3]. AICD produced in both pathways is reported to bind to different target proteins and may be involved in various cellular events, including apoptosis, neuronal growth and regulation of gene expression [28]. Thus, these proteases involved in the processing of AβPP are the therapeutic targets for developing compounds that down-regulate Aβ production either by augmenting the non-amyloidogenic processing or by inhibiting the amyloidogenic processing of AβPP.

Fig. 1.

Secretase-mediated AβPP processing pathways. The amyloid-β protein precursor (AβPP) is a type I transmembrane protein with a large extracellular domain, a single transmembrane, and a short intracellular domain. In the predominant non-amyloid pathway, AβPP is first cleaved by α-secretase and results in the release of the large N-terminal soluble ectodomain (sAβPPα) and the generation of the 83 amino acid-long membrane anchored C-terminal fragment (CTFα). In the amyloidogenic pathway, AβPP is first cleaved by β-secretase, resulting in the release of sAβPPβ and the generation of the membrane-anchored CTFβ of 99 amino acids. Both CTFα and CTFβ undergo further cleavage by γ-secretase within the membrane domain leading to the release of the intracellular domain (AICD) and the generation of a p3 fragment from CTFα and Aβ from CTFβ. The gray area represents the membrane.

α-secretase

Sequence analysis of the soluble secreted human brain AβPP [29] and secreted AβPP from various types of cells [30–32] indicates that α-cleavage mainly occurs after Lys-16 (numbered according to Aβ sequence), but multiple minor cleavages around this site have been observed. While the enzyme accountable for the α-cleavage has not been finally identified, studies suggest that α-cleavage is likely mediated by members of the ADAM (a disintegrin and metalloproteinase) family of proteases (for review see [33]).

β-secretase

Since Aβ is produced by β-secretase and γ-secretase, these two enzymes have attracted more attention. Almost simultaneously, several laboratories, using different approaches, identified β-secretase as a type-I transmembrane aspartyl protease characterized by the presence of two D^T /sG^T active site motifs, which is a conserved signature sequence of aspartic proteases. This enzyme was referred to as BACE (β-site AβPP cleaving enzyme) or BACE1, Asp2, and memapsin 2 [34–38]. In contrast to α-secretase, β-secretase cleaves its substrate with loose, but primary sequence specificity [39–41]. Like other known aspartic proteases such as cathepsin D and E, napsin A, pepsin, and renin, BACE prefers a leucine residue at position P1 [40], and this may explain the fact that Swedish mutant AβPP, in which the residues Lys-Met at the P2-P1 positions were substituted by Asn-Leu, serves as a better substrate for β-secretase cleavage. In addition, the facts that: 1) BACE1 is expressed at a higher level in the brain in comparison with other tissues [34,42]; 2) BACE1 knockout animals do not produce detectable levels of Aβ; and 3) BACE1 knockout mice are healthy and develop normally suggest that inhibition of β-secretase activity is a potentially promising therapeutic approach for AD (for review see [43, 44]).

γ-secretase

The other important secretase involved in Aβ formation is γ-secretase, which cleaves AβPP at the C-terminal end of the Aβ sequence to produce the C-termini of the Aβ peptide (Fig. 1). γ-secretase attracts more attention not only because its cleavage determines the ratio of Aβ₄₂/Aβ₄₀, which is a key factor in the amyloidogenesis of AD implying an important role for γ-secretase as a therapeutic target, but also because it occurs within the transmembrane domain of AβPP. This proteolysis within the hydrophobic environment of a membrane has attracted general biological interest. The current knowledge about the γ-secretase-catalyzed intramembrane processing of AβPP is the focus of this review. In search for genetic factors of AD, in addition to AβPP, two other related genes, PS1 and PS2, were identified to be associated with early-onset, auto-somal dominant FAD [8–10]. Shortly after their identification, it was found that the missense mutations in these two genes are associated with increased levels of the long and more aggregation-prone Aβ₄₂, suggesting that these mutations cause disease by altering AβPP processing, especially the formation of the C-termini of Aβ [45,46]. This notion was strongly supported by the finding that FAD-linked mutations in both PS1 and PS2 increase the production of Aβ₄₂ in transfected cells and transgenic mice [47–50]. Subsequently, the discovery that genetic knockout of both PS1 and PS2 resulted in total inactivation of γ-secretase activity suggests that presenilin may be functional γ-secretase [51–53]. The notion that presenilins bear the γ-secretase active site was strongly supported by the observations that mutation of either of the two conserved aspartate residues in PS1 (D257 and D385) or PS2 (D263 and D366) substantially reduced Aβ production and caused a concomitant accumulation of CTFβ [54,55]. The notion that presenilins bear the γ-secretase active site was further supported by the facts that γ-secretase activity was inhibited by aspartyl protease substrate-based peptidomimetic inhibitors [56,57] and, specifically, that aspartyl protease transition-state analog bound directly to PS1 and PS2 [58,59].

Presenilins are multiple transmembrane proteins and have been suggested to span the membrane six, seven, eight, and even nine times [60–62]. After synthesis, the 43 ∼ 47 KDa presenilin holoprotein is subsequently endoproteolytically cleaved within its large cytoplasmic loop between the sixth and seventh transmembrane domains to generate a 27 ∼ 28 KDa N-terminal and a 16 ∼ 18 KDa C-terminal fragment (NTF, CTF) [63–65]. The finding that NTF and CTF of presenilin remained physically associated in a high molecular complex [66–69], suggesting that other components that interact with PS1 may play a functional role in γ-secretase. In an effort to identify the putative components required for γ-secretase activity, nicastrin was first discovered by biochemical isolation [70]. Nicastrin is a 130 kDa type I transmembrane protein with a large ∼ highly glycosylated ectodomain. RNA interference (RNAi)-mediated knockdown experiments demonstrated that nicastrin was essential for γ-secretase-mediated cleavage of both AβPP and Notch proteins [70]. Subsequently, genetic screening for Notch signaling modulators in C. elegans led to the identification of two other novel genes, Aph-1 and Pen-2, which encode proteins of a 25 kDa with seven, and a 10 kDa with two, transmembrane domains, respectively [71,72]. As with nicastrin, knockdown of Aph-1 and Pen-2 by RNAi strongly reduced γ-secretase activity, indicating that these two are also essential components of γ-secretase [71–73]. Now, it is widely accepted that the active γ-secretase complex consists of at least four polypeptides – presenilin, nicastrin, Aph-1, and Pen-2 – with presenilin functioning as the catalytic subunit (Fig. 2, for review see [74]).

Fig. 2.

γ-secretase complex. The γ-secretase complex is composed of four integral membrane proteins: (brown) presenilin, composed of nine transmembrane domains, (dark blue) nicastrin, composed of a single transmembrane, (orange-red) Aph-1, composed of seven trans-membrane domains, and (light blue) Pen-2, composed of two trans-membrane domains. The substrate CTFβ (yellow) is also present. During maturation, presenilin undergoes endoproteolysis, and the resulting N-terminal fragment and C-terminal fragment remain associated. The two conserved aspartates (D) in the adjacent 6^th trans-membrane domain and 7^th transmembrane domain are essential for γ-secretase activity. Mature nicastrin is a highly glycosylated at its N-terminal ectodomain.

IDENTIFICATION OF THE ε-CLEAVAGE SITE

The α-secretase-catalyzed cleavage between the 16^th (Lys¹⁶) and 17^th (Leu¹⁷) amino acids of the Aβ sequence is referred to as the α-cleavage site; the β-secretase-catalyzed cleavage site at the first residue (Asp¹) of the Aβ sequence is referred to as the β-cleavage site. Since the majority of the C-termini of Aβ species generated by γ-secretase end at Aβ₄₀ or Aβ₄₂, the cleavage sites at these positions are commonly referred as to the γ-cleavage sites (Fig. 3). In addition, sequence analysis of the C-terminal fragments of AβPP produced in primary hippocampal neurons revealed that a minor species starts at position 12 in amino acids N-terminal to the β-cleavage site, and this fourth cleavage site was tentatively named the δ-cleavage site [32]. CTFα and CTFβ produced by α- and β-cleavage undergo further γ-secretase mediated cleavage, resulting in the formation of a common AICD, the full-length Aβ from CTFβ, and a p3 fragment from CTFα. Since most Aβ and p3 fragments end at Aβ₄₀ or Aβ₄₂, one may assume that AICD would start at either Aβ₄₁ or Aβ₄₃. That is why AICD time-after-time was referred to as CTFγ, probably based on the idea that it was produced by γ-cleavage. However, follow-up studies surprisingly revealed that most of the AICDs start from Aβ₅₀, with some of them starting from Aβ₄₉ [75–78]. Thus, these findings led to the identification of a new cleavage site, which is now known as the ε-cleavage site at Aβ49 [75–78]. Interestingly, this ε-cleavage site is found to be equivalent to the S3 site of Notch 1 [79]. In addition, a new intramembrane cleavage site, the S4 site in Notch 1, was reported in a recent study, and it is assumed to be homologous to γ-cleavage site in AβPP [80]. These findings further suggest that AβPP and Notch 1 undergo similar intramembrane cleavages, namely, γ (S4) and ε (S3) cleavages [80].

Fig. 3.

γ-secretase-mediated major intramembrane cleavages of AβPP. Now it is clear that γ-secretase catalyzes three major cleavages within the membrane domain of AβPP, namely the γ-cleavage at Aβ_40/42, the ε-cleavage at Aβ₄₉, and the ζ-cleavage at Aβ₄₆. Studies also strongly suggest that cleavage at Aβ₄₃ is also a necessary intermediate step. It is notable that the two well-characterized AD-linked mutations, the Swedish mutation and the AβPP717 mutation, happen to occur at two major cleavage sites vital for the generation of the N-terminus and the C-termini of the Aβ peptides, suggesting the possibility that these mutations cause abnormal Aβ formation by influencing these major cleavages.

IDENTIFICATION OF THE ζ-CLEAVAGE SITE

Since the ε-cleavage site at Aβ₄₉ is 7−9 residues distal to the γ-cleavage site at Aβ_40/42, the identification of the ε-cleavage site raised questions as to the relationship between the γ- and ε-cleavages and whether ε-cleavage is an essential step for generating the C-termini of Aβ [81]. There are several possibilities: one possibility is that CTFβ is firstly cleaved at the γ-cleavage site at Aβ_40/42 and that the resulting CTFγ (CTF41−99 and CTF43−99) undergoes further and rapid degradation by a particular aminopeptidase and/or endopeptidase or undergoes a second γ-secretase-dependent cleavage at the ε-site to produce CTFε (CTF50−99, the AICD mentioned above); the second possibility is that the ε-cleavage at Aβ₄₉ occurs first, followed by γ-cleavage at Aβ_40/42 to produce CTFε and Aβ_40/42; or thirdly, it is also possible that the ε-cleavage and γ-cleavage are two totally unrelated events occurring in parallel or even possibly involved in two separate proteolytic pathways [75–78]. Given the fact that the ε-cleavage is closer to the membrane boundary and is thus easily accessed by water molecules required for the hydrolysis of the peptide bond, the second possibility that ε-cleavage precedes γ-cleavage is more plausible. However, either one of these possibilities need to be experimentally tested. One of the obstacles to addressing these questions is that neither the long Aβ peptides, which end at the ε-cleavage site, nor the C-terminal fragment, which starts with an N-terminal generated by γ-cleavage, has ever been detected. There are several considerable possibilities that may account for the inability to detect these intermediates. First, these intermediates might have been rapidly processed or degraded into short species, Aβ_40/42 and/or CTFε, which are the final products of γ-secretase processing and were usually detected in most of the previous studies. Second, thus far, most of the attention has been concentrated on the secreted forms of Aβ, and less attention has been paid to intracellular or membrane-bound Aβ; however, it is very possible that the trace amount of the intermediate long Aβ species may have been overlooked; and third, as discussed above, it may be that ε-cleavage and γ-cleavage are separate events, and the intermediates do not exist at all. Since Aβ species are short and hydrophobic, it is difficult to distinguish them from each other using the conventional SDS-PAGE (sodium dodecyl sufatepolyacrylamide gel electrophoresis) under normal conditions. To solve this issue, researchers have developed a new SDS-PAGE system, the urea-SDS-PAGE system, which has been shown to be able to successfully separate Aβ₄₂ from Aβ₄₀ [82,83]. By using this system in combination with mass spectrometric analysis, two groups reported the identification of a long form of Aβ, Aβ₄₆, in cultured cells expressing AβPP and in brain tissues of a transgenic mouse carrying the human AβPP gene [84,85]. Since the production of this Aβ₄₆ is inhibited by transition state analog γ-secretase inhibitor and dominant negative presenilin, it is believed that this Aβ₄₆ is produced by γ-secretase activity [84]. Thus, this finding led to the discovery of a new γ-secretase-mediated cleavage site, which was designated as the ζ-cleavage site at Aβ₄₆ between the γ-cleavage site at Aβ_40/42 and ε-cleavage site at Aβ₄₉ [84].

Aβ₄₆ IS A PRECURSOR OF Aβ_40/42

The identification of the new ζ-cleavage site at Aβ₄₆ confirmed that the Aβ species with a C-terminus end at Aβ₄₆, detected in human skeletal muscles in a previous study [86], is not a result of random degradation of AβPP, but a product of γ-secretase activity. These findings strongly support the notion that, as with γ-cleavage and ε-cleavage, ζ-cleavage is a normal physiological event during secretase-mediated AβPP processing. Now, the questions are, what is the relationship of this new ζ-cleavage with the other two known γ-secretase-mediated cleavages, namely γ-cleavage and ε-cleavage, and does this new ζ-cleavage at Aβ₄₆ play a role in the formation of secreted Aβ_40/42? In this regard, it is notable that the newly identified Aβ₄₆ was found to be tightly associated with γ-secretase complex [87]. Thus, it is fair and reasonable to speculate that this Aβ₄₆ may undergo further cleavage at Aβ_40/42, the γ-cleavage site, to produce secreted Aβ_40/42. This speculation is strongly supported by a very interesting finding that several well characterized γ-secretase inhibitors, such as DAPT, DAPM and compound E, which are known as non-transition state inhibitors, inhibited the formation of secreted Aβ_40/42 and caused a concomitant accumulation of intracellular membrane-bound Aβ₄₆ in a dose dependent manner, suggesting a possible precursor-product relationship between Aβ₄₆ and Aβ_40/42 [84]. However, the observation that ζ-cleavage can be differentially inhibited by γ-secretase inhibitors known as transition state analogs, such as L-685,458 and WPE-III-31C (31C), but less affected by non-transition state inhibitors, such as DAPT, DAPM and compound E, complicated this matter [84]. One may even question whether γ-cleavage and ζ-cleavage are catalyzed by different enzymes, which are both dependent on PS1 or a PS1-like protein, but have different sensitivities to transition state and non-transition state inhibitors. Thus, the roles of ε- and ζ-cleavages in the formation of secreted Aβ and the relationship among these three cleavages, namely ε-, ζ-, and γ-cleavages, still remain elusive. To challenge these issues, one group, by employing a differential inhibition strategy, conducted a series of experiments to: 1) explore the relationship between Aβ₄₆ and Aβ_40/42; 2) determine the possible presence of Aβ₄₉ generated by ε-cleavage and its relationship with Aβ₄₆ and Aβ_40/42; 3) elucidate the relationship between the three major cleavages (ε-, ζ-, and γ-cleavages) within the transmembrane domain of AβPP; and 4) establish the roles of ε- and ζ-cleavages in the formation of secreted Aβ_40/42 [87,88]. To determine whether Aβ₄₆ can be processed into Aβ_40/42, the investigators first cultured the cells in the presence of non-transition state inhibitor DAPM to accumulate intracellular membrane-bound Aβ₄₆ for a certain period, and then the DAPM was replaced with transition state analog L-685,458, which blocks new formation of Aβ₄₆ from CTFβ, in a fresh medium. After two hours incubation, when the fresh medium was examined for the presence of possible secreted Aβ_40/42 during a time course of the subsequent culture, surprisingly, along with a decrease of Aβ₄₆, an increase in secreted Aβ_40/42 was observed. Under the same conditions, namely in the presence of transition state analog L-685,458, no Aβ_40/42 was detected in cells without pre-accumulated Aβ₄₆. These results indicate that, in the presence of L-685,458, the secreted Aβ_40/42 detected in the fresh medium during the subsequent culture should be produced solely from the pre-accumulated Aβ₄₆, but not from CTFβ, which remains unchanged since the transition state analog L-685,458 blocked it from being further processed by γ-secretase. These results have been further confirmed using both living cells and cell-free systems and by metabolic labeling approaches [87,88]. Thus, these observations clearly established the precursor-product relationship between Aβ₄₆ and Aβ_40/42 [87,88].

SECRETED Aβ IS PRODUCED BY γ-SECRETASE-MEDIATED SEQUENTIAL CLEAVAGES: FIRST THE ε-CLEAVAGE, FOLLOWED BY A RAPID ζ-CLEAVAGE, AND THEN BY A γ-CLEAVAGE

In these studies, one notable finding is that transition state analog L-685,458 showed no effect on the turnover of Aβ₄₆ in both living cell and cell-free systems, indicating that L-685,458 does not directly inhibit γ-cleavage at Aβ_40/42 [87]. On the other hand, L-685,458 showed a strong inhibitory effect on the formation of Aβ₄₆ by ζ-cleavage and the formation of CTFε by ε-cleavage. These findings strongly suggest that this inhibitor inhibits the formation of secreted Aβ₄₂ by a mechanism other than direct inhibition of γ-cleavage at Aβ_40/42, but indirectly, by inhibiting the formation of Aβ₄₆ by ζ-cleavage. Therefore, the fact that the inhibition of upstream ζ-cleavage by L-685,458 completely prevents the downstream γ-cleavage from taking place strongly supports an important notion that γ-cleavage not only occurs secondarily, but also is dependent on ζ-cleavage occurring first [87]. The next question is, what is the relationship of ε-cleavage to ζ- and γ-cleavages? To answer this question, the key is to determine the presence of Aβ₄₉ produced by ε-cleavage. In an effort to determine the long-sought Aβ₄₉, the researchers performed immunoprecipitation of cell lysates using an Aβ-specific antibody and successfully detected a modest level of Aβ₄₉ in cells cultured in the absence of any inhibitors [87]. The fact that Aβ₄₉, which contains the ζ-cleavage site at Aβ₄₆, is detectable in living cells in the absence of any inhibitors indicates that ε-cleavage occurs prior to ζ-cleavage; otherwise the ε-cleavage product Aβ₄₉ should not have had a chance of being formed. In addition, in the same study, it was observed that Aβ₄₉ can be processed into Aβ_40/42 in the absence of inhibitors and processed into Aβ₄₆ in the presence of non-transition state inhibitors, such as DAPT, DAPM, and compound E. These findings indicate that Aβ₄₉ is a direct precursor of Aβ₄₆. However, in the presence of transition state analogs, such as L-685,458 and 31C, neither Aβ_40/42 nor Aβ₄₆ was detected, indicating that these transition state analogs inhibit ζ-cleavage. Thus, as illustrated in Fig. 4, these observations strongly suggest that after β- or α-cleavage of AβPP, the resulting CTFβ and CTFα first undergo ε-cleavage, followed by a sequential but rapid ζ-cleavage and then a γ-cleavage, commencing at the site closest to the membrane boundary and proceeding towards the site within the middle of the transmembrane domain of AβPP. More importantly, the fact that γ-cleavage not only occurs last but also is dependent on ε- and ζ-cleavages occurring prior to it, established that the ε-cleavage and ζ-cleavage are obligatory steps, and especially, that the ε-cleavage is the initial and rate-limiting step of the sequential ε, ζ-, and γ-cleavages involved in the generation of the C-termini of secreted Aβ.

Fig. 4.

Schematic illustration of γ-secretase-mediated sequential cleavages of AβPP. After β-secretase cleavage, the resulting C-terminal fragment of AβPP (CTFβ ) undergoes a series of sequential cleavages catalyzed by γ-secretase to produce Aβ. As a major path, CTFβ is first scissored by ε-cleavage at Aβ₄₉, followed by ζ-cleavage at Aβ₄₆ to produce the major intermediate Aβ₄₆. Aβ ₄₆ is mainly processed at Aβ₄₃ and the resulting Aβ₄₃ is further processed into Aβ₄₀, which can be further processed into Aβ₃₇, but is principally released from the γ-secretase complex and becomes the major form of secreted Aβ species. Alternatively, Aβ₄₆ can also be processed at Aβ₄₂ at a low efficiency, and the resulting Aβ₄₂ can be either further processed into Aβ₃₈₍₃₉₎ or released at a low rate.

THE SEQUENTIAL CLEAVAGE MODEL PROVIDES NEW INSIGHTS INTO THE MECHANISM OF γ-SECRETASE-MEDIATED A βPP PROCESSING

All proteases are also known as hydrolases because they catalyze the hydrolysis of a peptide bond; therefore, water molecules are required for the cleavage of the peptide bond. According to the catalytic mechanism of the aspartyl protease, in order to hydrolyze the peptide bond of the substrate, one of the two aspartate residues in the enzyme active site, disposed on opposite faces of the peptide bond to be cleaved, needs to first act as a general base to activate the water molecule. The activated water molecule then attacks and breaks the peptide bond, in cooperation with the second aspartate, which acts as general acid to protonate the departing amine product. Thus, one of the most mysterious questions in γ-secretase-catalyzed intramembrane processing of AβPP is how the water molecules, which are required for the peptide bond hydrolysis, enter the γ-secretase active site, which is localized within the hydrophobic environment of the lipid bilayer membrane. In this regard, the sequential cleavage model may provide an answer to this question. As shown in Fig. 5, the ε-cleavage site is close to the membrane boundary and is easily accessed by water molecules in the cytosol. The initial ε-cleavage may not only release the CTFε (AICD), but may also create a path for the water molecule to have access to the next cleavage sites, namely the ζ-cleavage site and then the γ-cleavage site. Accordingly, without removal of the C-terminal fragment, water molecules may not be able to access the γ-cleavage site, resulting in the prevention of γ-cleavage from taking place. Thus, the blockage of water access may account, at least in part, for the fact that γ-cleavage depends on ε- and ζ-cleavages occurring first.

Fig. 5.

Schematic illustration of γ-secretase-mediated AβPP processing and Aβ formation. Using electronic microscopy and single-particle image analysis on the purified active γ-secretase complex, a 3D structure of this enzyme has been recently proposed [98]. Based on this 3D structure, the γ-secretase complex contains a cylindrical interior chamber, in which the catalytic site is located, with two channels facing the two sides of membrane. It has been assumed that the two channels would allow the entry of water molecules, their access to the γ-secretase-active site, and the release of the cleavage products. Based on this electronic microscopic 3D structure, the γ-secretase-mediated sequential cleavages and Aβ formation are dynamically illustrated. Since the ε-cleavage site is close to the membrane boundary and is easily accessed by water molecules in the cytosol, ε-cleavage occurs first (step 1). The initial ε-cleavage may not only release CTFε (AICD), but may also create a path for the water molecules to have access to the next cleavage site, the ζ-cleavage site, and trigger ζ-cleavage to occur. The result is the release of an additional three amino acids (red dot, step 2), thus, allowing the water molecules further access to the γ-cleavage site and allowing the final cleavage step (step 3) to occur, leading to the formation of secreted Aβ_40/42. Then the γ-secretase complex may be disassembled and subjected to recycling or disposal.

The second question is whether these cleavages, namely ε-, ζ-, and γ-cleavages, are catalyzed by the same γ-secretase or different enzymes. The finding that ε-, ζ-, and γ-cleavages can be differentially inhibited by different inhibitors, i.e. ε-/ζ-cleavages are inhibited by transition state analogs and γ-cleavage is inhibited by non-transition state inhibitors, suggests several possibilities. First, these cleavages may be catalyzed by two enzymes, one of them sensitive to transition state analogs and the other sensitive to non-transition state inhibitors. Second, these cleavages may be catalyzed by one enzyme with two inhibitor binding sites, one for the transition state analogs, such as L-685,458, and the other for the non-transition state inhibitors, such as compound E, as suggested by a recent inhibitor-binding kinetic study [89]. In regard to the question of whether there is only one enzyme or multi-enzymes involved in the sequence of ε-, ζ-, and γ-cleavages, the precursor and product relationship between the long forms of Aβ species and the short forms of Aβ species, the sequential relationship of these cleavages, and specifically the finding that γ-cleavage is dependent on ε- and ζ-cleavages occurring first, strongly suggest that γ-, ζ-, and ε-cleavages are catalyzed by a single enzyme. If it is not the same enzyme, it is hard to imagine how, after ε- and ζ-cleavages, the intermediate Aβ₄₆ is transferred from the enzyme that produced it to another enzyme for further processing into Aβ_40/42. In fact, the single enzyme model is further strongly supported by the finding reported by the same study that the intermediate Aβ₄₆ is tightly associated with PS1, the catalytic subunit of the γ-secretase complex, indicating that the same PS1-dependent enzyme executes the whole series of the ε-, ζ-, and γ-cleavage [87]. The one enzyme model is also supported by the fact that both groups of inhibitors have been shown to bind to presenilins [58, 90,91] and the finding that equimolar amounts of CTFε and Aβ are produced from the cleavage of CTFβ by γ-secretase [92].

The third question is, how does a single enzyme sequentially cleave a peptide at multiple sites over a sequence of ten amino acids? In this regard, several studies have reported the identification of intermediate Aβ₄₃ ([84,85,88], see also Fig. 3). This Aβ₄₃ was even detectable in the conditioned medium when the cells were cultured in the presence of transition state analog 31C, probably because 31C may cause a conformational change in the γ-secretase complex, leading to a looser association of Aβ₄₃ with the γ-secretase complex and resulting in the release of this intermediate [88]. The detection of the intermediates, Aβ₄₉, Aβ₄₆, and Aβ₄₃, revealed an interesting pattern of γ-secretase processing of CTFβ, i.e., γ-secretase cleaves CTFβ every three residues to produce the major secreted Aβ species, Aβ₄₀ [85,87,88]. How does the one enzyme model account for the sequential cleavage of every three residues? Evidence suggests that the AβPP transmembrane domain is in an α-helical conformation [56,93]. Based on this α-helical model, as shown in Fig. 6A, in their native state, the peptide bonds between residues Aβ₄₆ and Aβ₄₇ and residues Aβ₄₃ and Aβ₄₄ are aligned to the center of the putative enzyme targeting side, this may account for the fact that the major intermediate Aβ species are detected as Aβ₄₆ and Aβ₄₃ [88]. According to this model, the peptide bonds between residues Aβ₄₁ and Aβ₄₂ and between Aβ₄₈ and Aβ₄₉ are on the side that is precisely opposite to the side on which the peptide bonds between residues Aβ₄₆ and A_β47 and residues Aβ₄₃ and Aβ₄₄ are aligned. This may account for the fact that cleavages between these residues hardly occur; as a result, no Aβ₄₁ and only a very low level of Aβ₄₈ was reported by a recent study [85]. After each cleavage, the α-helical structure may be slightly loosened, unfolded and stretched. This slight unfolding may cause a counterclockwise turn of the α-helix and make the next cleavable peptide bond move toward the enzyme-attacking site; this may partially explain why the majority of the secreted Aβ species is Aβ₄₀ rather than Aβ₄₂. In addition, the stretch of the slightly unfolded α-helix may also help the next cleavable peptide bond to move and reach the enzyme active site. It has been hypothesized that transition state analogs and non-transition state inhibitors bind to different sites, i.e., the transition state analogs bind to the catalytic site and the non-transition state inhibitors bind to a remote site [89]. Based on this hypothesis, it is plausible that the transition state analogs may inhibit the upstream cleavages, ε-cleavage and ζ-cleavage, by directly binding to the catalytic site. As a result, the downstream γ-cleavage is also prevented from taking place. On the other hand, the non-transition state inhibitors may bind to a remote site of the enzyme, inducing conformational changes in the enzyme and preventing the γ-cleavage site from having access to the catalytic site of the enzyme, resulting in the preferential inhibition of γ-cleavage with less effect on the ε- and ζ-cleavages.

Fig. 6.

α-helical wheel arrangement of amino acids 37 to 52 of CTFβ, a view from the cytosol side. The C-terminal residue of each detectable major Aβ species produced by γ-secretase-mediated cleavages is shown in gray with a bold outline. Note, under normal conditions (Fig. 6A), the peptide bond between residues 46 and 47 and the peptide bond between residues 43 and 44 are aligned to the center of the lower half of the α-helical wheel, which may be the enzyme attacking site. Hydrolyses of these peptide bonds produce the major intermediate Aβ₄₆ and Aβ₄₃. It is noted that after each cleavage, the α-helical structure may be slightly loosened, unfolded and stretched, resulting in a counterclockwise turn of the α-helix and this makes the peptide bond between residues 40 and 41 move toward the enzyme-attacking site; this may partially explain why the majority of the secreted Aβ species is Aβ₄₀ rather than Aβ₄₂. In contrast, under some circumstances, such as environmental or genetic abnormalities, specifically, in the presence of AD-linked mutations in presenilin or in AβPP, the relative position between the γ-secretase active site and its substrate may be altered. As shown in Fig. 6B, if these changes in the position of CTFβ relative to the γ-secretase active site can be mimicked by rotating the CTFβ α-helical wheel clockwise 60°, the peptide bond between Aβ₄₂ and Aβ₄₃ will move toward the center of the lower half wheel and become much more susceptible to γ-secretase activity, resulting in an increase in the formation of Aβ₄₂.

However, the one catalytic site model fails to explain the fact that Aβ₄₆ is still processed to Aβ_40/42 by γ-cleavage in the presence of the transition state analog, which is assumed to bind to the catalytic site [89]. Therefore, the other possibility is likely that these sequential cleavages may be catalyzed by an enzyme that has two catalytic sites, one catalyzing the ε-cleavage at Aβ₄₉ and the ζ-cleavage at Aβ₄₆, both of which are specifically inhibited by transition state analogs, and the other one catalyzing the cleavages at Aβ₄₃, Aβ_40/42(γ), which are inhibited by non-transition state inhibitors [87]. Regardless of whether there is one or two catalytic sites, according to the one enzyme model, at high concentrations the non-transition state inhibitors, which preferentially inhibit γ-cleavage, may also inhibit ε- and ζ-cleavages by an allosteric mechanism, i.e., these compounds may induce conformational changes into the γ-secretase complex, resulting in partial inhibition of ε- and ζ-cleavages. The other possible explanation for the fact that non-transition state inhibitors also cause accumulation of CTFβ is that in the presence of non-transition state inhibitors, which inhibit the turnover of Aβ₄₆ into Aβ_40/42, the accumulated intermediate Aβ₄₆ occupies the binding site of the γ-secretase complex and prevents the further binding of CTFβ to the γ-secretase complex, resulting in accumulation of unprocessed CTFβ.

The fourth question is, how does the one enzyme model account for the production of both secreted Aβ₄₀ and Aβ₄₂? Whether Aβ₄₀ and Aβ₄₂ are produced by the same enzyme or by different enzymes has been debated for a long time, and the answer still remains elusive. Based on the observations that the calpain inhibitor had different effects on the production of Aβ₄₀ and Aβ₄₂, it has been hypothesized that the different C-termini of Aβ are generated by different protease activities [94,95]. A recent study also hypothesized that Aβ₄₀ is generated by an Aβ₄₀-secretase that acts from one side of the α-helical wheel and cleaves AβPP every three residues (i.e., Aβ₄₉ to Aβ₄₆ then to Aβ₄₃ and finally to Aβ₄₀) and that Aβ₄₂ is generated by an Aβ₄₂-secretase acting from the other side of the α-helical wheel and cleaving AβPP every three residues (i.e., Aβ₄₈ to Aβ₄₅ and then to Aβ₄₂) [85]. However, as it has been pointed out, there is insufficient evidence to support this hypothesis of the coordinated cleavages for Aβ₄₂, Aβ₄₅, and Aβ₄₈ [85]. Moreover, other studies have shown that both Aβ₄₀ and Aβ₄₂ are generated from the same intermediate Aβ₄₆ [87,88]. In these studies, it was hypothesized that Aβ₄₆ generated by ζ-cleavage is mostly cleaved at Aβ₄₃, which is further processed at Aβ₄₀, resulting in the formation of major secreted Aβ species Aβ₄₀. However, Aβ₄₆ may also be cleaved with low efficiency at Aβ₄₂ (Fig. 4). The Aβ₄₂ produced from Aβ₄₆ can be either released into the medium or undergo further processing to produce shorter Aβ species such as Aβ₃₈₍₃₉₎ [88]. It should be pointed out that, any genetic or environmental changes could cause a conformational change between the enzyme and substrate, resulting in a shift of the cleavage from Aβ₄₃ to Aβ₄₂. For example, as shown in Fig. 6B, if this relative conformational change between enzyme and substrate can be mimicked by rotating the α-helical wheel clockwise to a certain degree, the peptide bond between residues Aβ₄₃ and Aβ₄₂ would move toward the center of the enzyme-attacking site and become more susceptible to the γ-secretase cleavage. This may explains why certain mutations in AβPP and presenilin could cause increased production of Aβ₄₂ with a concomitant decrease in Aβ₄₀. However, as shown in Fig. 6B, when the peptide bond between Aβ₄₂ and Aβ₄₃ is aligned to the center of the enzyme-attacking site, the peptide bond between Aβ₄₈ and Aβ₄₉ is still not in the favorable position. Therefore, the cleavage at Aβ₄₈ may occur under some circumstance, but may not be necessarily required for the production of Aβ₄₂.

THE PATHOLOGICAL AND PHARMACOLOGICAL SIGNIFICANCE OF THE IDENTIFICATION OF THE NEW ζ-CLEAVAGE SITE AT Aβ₄₆

First, according to the amyloid hypothesis that the long Aβ is more amyloidogenic and more pathogenic, these long Aβ species, Aβ₄₃,Aβ₄₆ and Aβ₄₉ may play an important role in disease development. Therefore, the discovery of these long Aβ species may provide new targets for use in diagnosis and treatment of the disease. Second, the fact that Aβ₄₆ can be detected in the absence of inhibitor and accumulate to a high level in cells cultured in the presence of non-transition state inhibitors strongly indicates that the newly identified ζ-cleavage at Aβ₄₆ is another major γ-secretase-mediated cleavage site, besides the previously identified γ-cleavage at Aβ_40/42 and ε-cleavage at Aβ₄₉. It is also notable that the newly identified ζ-cleavage site at Aβ₄₆ is incidentally the AD-linked AβPP717 mutation (also known as the London mutation) site [96,97]. Given the fact that the other AD-linked Swedish mutation site is at the β-cleavage site this finding revealed an interesting fact that the two AD-linked mutations, the Swedish mutation and the AβPP717 mutation, actually both occur at cleavage sites that are vital for the generation of the Aβ peptide. Thus, the identification that the ζ-cleavage site at Aβ₄₆ happens to be the AβPP717 mutation site may open a new avenue for studying the mechanism by which this AD-linked mutation causes increased production of the more amyloidogenic Aβ₄₂.

The identification of ζ-cleavage is also pharmacologically significant. Prior to the identification of the new ζ-cleavage site and its product (the Aβ₄₆), all γ-secretase inhibitors, including both the transition state analogs and non-transition state inhibitors, were characterized by their inhibitory effects on the turnover of the initial substrate CTFβ and the formation of the final product, the secreted Aβ_40/42. This is why both transition state and non-transition state inhibitors were thought to similarly inhibit γ-cleavage. The identification of Aβ₄₆ produced by ζ-cleavage makes it possible for the first time to determine the specificity of the γ-secretaseinhibitors. Studies examining the effects of the known inhibitors on the formation and turnover of the newly identified Aβ₄₆ led to a novel finding that some of the compounds, such as DAPT, DAPM, and compound E, which were previously known to inhibit the formation of secreted Aβ₄₀ and Aβ₄₂, caused an intracellular accumulation of an even longer Aβ species, Aβ₄₆ [84,87]. According to the amyloid theory, the long Aβ is more pathogenic. Therefore, these inhibitors, which prevent the formation of secreted Aβ_40/42, but cause accumulation of the long Aβ₄₆, are rather pharmacological poisons than potential therapeutic compounds. Thus, these findings provide information extremely important for the strategies to prevent and treat AD by the design and use of γ-secretase inhibitors. Thus, the identification of the novel ζ-cleavage at Aβ₄₆ and the finding the Aβ peptide is produced by a series of sequential cleavages not only contribute to a better understanding of the molecular mechanism by which Aβ is generated, but could also contribute to the development of treatments and methods of AD prevention.