Aβ42 Overproduction Associated with Structural Changes in the Catalytic Pore of γ-Secretase

γ-Secretase is an atypical aspartyl protease that cleaves amyloid β-precursor protein to generate Aβ peptides that are causative for Alzheimer disease. γ-Secretase is a multimeric membrane protein complex composed of presenilin (PS), nicastrin, Aph-1, and Pen-2. Pen-2 directly binds to transmembrane domain 4 of PS and confers proteolytic activity on γ-secretase, although the mechanism of activation and its role in catalysis remain unknown. Here we show that an addition of amino acid residues to the N terminus of Pen-2 specifically increases the generation of Aβ42, the longer and more aggregable species of Aβ. The effect of the N-terminal elongation of Pen-2 on Aβ42 generation was independent of the amino acid sequences, the expression system and the presenilin species. In vitro γ-secretase assay revealed that Pen-2 directly affects the Aβ42-generating activity of γ-secretase. The elongation of Pen-2 N terminus caused a reduction in the water accessibility of the luminal side of the catalytic pore of PS1 in a similar manner to that caused by an Aβ42-raising γ-secretase modulator, fenofibrate, as determined by substituted cysteine accessibility method. These data suggest a unique mechanism of Aβ42 overproduction associated with structural changes in the catalytic pore of presenilins caused commonly by the N-terminal elongation of Pen-2 and fenofibrate.

secretory pathway (10), remains unknown. However, the loss of Aph-1 or Nct decreased the expression of PS and Pen-2. In contrast, the overexpression of Aph-1 together with Nct stabilizes the ␥-secretase complex, suggesting the role of Aph-1 as a molecular scaffold for this atypical membrane-associated protease (5,11).
Pen-2 is a membrane protein harboring two TMDs with a hairpin-like topology (12). Depletion of Pen-2 caused the loss of the ␥-secretase activity accompanied by the accumulation of PS holoprotein-Nct-Aph-1 trimeric complex (5). The overexpression of Pen-2 together with PS, Nct, and Aph-1 provoked the endoproteolysis of PS to generate N-and C-terminal fragments (NTF and CTF, respectively), concomitantly with the acquisition of the ␥-secretase activity. Mutational analyses revealed that Pen-2 directly binds to TMD4 of PS through its proximal two-thirds of the first TMD (13)(14)(15). Assembled ␥-secretase complex is sorted out from the endoplasmic reticulum and exhibits proteolytic activity. Thus, Pen-2 is the functional activator of the ␥-secretase complex in its biosynthetic pathway. It has been shown that the length and the amino acid sequences of the C terminus of Pen-2 affect the stability of PS fragments after maturation (13, 16 -18). However, the role of Pen-2 in endoproteolytic activity remains unknown. Here we report on an unexpected observation that the length of the N terminus of Pen-2 affects the structure of the active site of ␥-secretase in a way to increase A␤42 generation, in a similar fashion to fenofibrate, an A␤42-raising ␥-secretase modulator.
Analysis of Immunoprecipitated A␤ Peptides-Conditioned media supernatants from S2 cells and mouse embryo fibroblast cells were incubated with 20 l of anti-A␤ antibody BAN50 (provided by Takeda Pharmaceutical Co.) in a rotator at 4°C for 5-18 h. Protein G-agarose (Invitrogen) was added, and rotational incubation was continued for additional 2 h. For immunoblot analysis, precipitated A␤ peptides were separated by 10% Tris/Bicine/urea gel (27,28) and probed by anti-A␤ antibody 82E1. Simultaneously, synthetic A␤ peptides were loaded and used as molecular standards.
(%A␤42) is ϳ10% in almost all types of mammalian cultured naïve cells (29). We previously reported that Drosophila S2 cells harbor endogenous ␥-secretase activity with a property quite similar to that of mammalian cells, %A␤42 being at ϳ15-20% (5,19). Furthermore, the overexpression of Drosophila PS (Psn), nicastrin (dNct), Aph-1 (dAph-1), and Pen-2 (dPen-2) increased the levels of Psn fragments as well as of total ␥-secretase activity in Drosophila S2 cells (Fig. 1A) (5). We measured secreted A␤40 and A␤42 from cells expressing tagged Drosophila ␥-secretase components (i.e. Psn, dNct-V5/His, dAph-1-FLAG, or HA-dPen-2) together with SC100, the latter corresponding to the C-terminal fragment of human APP with a signal peptide. Unexpectedly, S2 cells expressing all the four components secreted significantly increased levels of A␤42 compared with that from cells expressing Psn, dNct-V5/His, and dAph-1-FLAG ("mock" in Fig. 1B). %A␤42 was significantly increased to almost 2.5-fold (11.7% in mock, and 28.5% in HA-dPen-2 transfected cells), although the coding sequences of all the ␥-secretase components were of wild-type (shown as -fold changes in %A␤42 compared with mock in Table 1). In contrast, the C-terminal V5/His-tagged dPen-2 had no effect on %A␤42 (1.2-fold compared with mock, p ϭ 0.339 (n ϭ 3)). To ascertain whether the addition of the tag sequence to the N terminus of dPen-2 caused this effect, we overexpressed dPen-2 without a tag together with Psn, dNct-V5/His, and dAph-1-FLAG and examined the secretion of A␤. In contrast to HA-dPen-2, %A␤42 from cells expressing untagged dPen-2 was almost comparable to that in mocktransfected cells ( Fig. 1B and Table  1). To specifically examine the effects of tagged Pen-2 or other ␥-secretase components by eliminating endogenous components, we took a UTR-targeted RNAi/rescue approach on each component (11), which enables us to analyze the function of exogenous proteins under a null-phenotype of the gene of interest in Drosophila S2 cells. The levels of Psn fragments and of the ␥-secretase activity were decreased by the UTR-targeted RNAi, suggesting that dsRNAs against the chosen UTR sequences (5Ј-UTR for Psn and dNct, 3Ј-UTR for dAph-1 and dPen-2, respectively) did suppress the gene expression. The overexpression of exogenous components (i.e. Psn, dNct-V5/ His, dAph-1-FLAG, and HA-dPen-2) in respective knockdown cells restored both the expression of Psn fragment and A␤ generation (Fig. 2, A and B). Intriguingly, however, only Pen-2-rescued cells showed a statistically significant overproduction of A␤42 ( Fig. 2B and Table 2). These data suggest that the reconstitution of the Drosophila ␥-secretase complex harboring dPen-2 tagged with HA at the N terminus caused an increase in the A␤42 generation in Drosophila S2 cells.
distal (residues 10 -16), portion of the hydrophilic N-terminal region of hPEN-2 is dispensable for the acquisition of the ␥-secretase activity, although the levels of secreted A␤42 were not documented (13). To determine whether the HA tag effect on A␤42 overproduction is dependent on the length and/or integrity of the N terminus of dPen-2, we generated a series of N-terminal length-mutants of dPen-2: dPen-2/⌬2-10 lacking residues 2-10 of dPen-2, HA-dPen-2/⌬2-10, a HA-tagged version of dPen-2/⌬2-10, and dPen-2/10HA11 with a HA tag sequence inserted between residues 10 and 11 (Table 3). Because the HA sequence consists of nine amino acid residues, HA-dPen-2/⌬2-10 and dPen-2/⌬10HA11 have N-terminal amino acid lengths equal to those of untagged dPen-2 and HA-dPen-2, respectively. These mutants increased Psn fragments as well as total A␤ secretion in the presence of Psn, dNct-V5/ His, and dAph-1-FLAG (data not shown), suggesting that the proximal region of the N terminus of dPen-2 is dispensable for the ␥-secretase activity, in agreement with the result in hPEN-2 (13). However, %A␤42 in cells expressing dPen-2/⌬2-10 or HA-dPen-2/⌬2-10 was similar to that in cells expressing untagged dPen-2 (Table 3). Thus, the presence of an HA tag sequence at the N terminus of dPen-2 is not sufficient to increase A␤42 production. In contrast, dPen-2/10HA11 significantly increased %A␤42 in a similar manner to those with a series of the N-terminally tagged dPen-2 (Table 3). Finally, we examined the effects of untagged or HA-tagged dPen-2/⌬2-17/rep, which harbor several substitutions of amino acid residues (MDISKAPNPRKLELCRK to MELCRGPQPKRVDISKR) within the entire length of the N-terminal region of dPen-2. This mutant carries a totally different amino acid sequence at the N terminus with similar characteristics in terms of bulkiness and charges other than the proline residues. HA-dPen-2/ ⌬2-17/rep significantly increased the generation of A␤42, whereas dPen-2/⌬2-17/rep did not affect %A␤42 at a statisti-FIGURE 2. Only tagged dPen-2 affected the A␤42 generation in UTR-targeted RNAi and rescue experiment. A, immunoblot analysis of S2 cells transiently transfected with cDNAs encoding wild-type ␥-secretase components (denoted as "Rescue") together with dsRNAs against GFP or UTR region of the target genes (denote as "dsRNA"). Used antibodies were indicated below the panels. B, effect of the tagged components on the secreted A␤ levels (n ϭ 3, mean Ϯ S.E.). Transfected cDNAs and dsRNAs are indicated under the graph. White and black bars denote A␤40 and A␤42, respectively.  cally significant level (Table 3). Collectively, these data strongly suggest that the A␤42 overproduction closely correlates with the extension of the N-terminal length of dPen-2, regardless of the amino acid sequences of the N-terminal luminal region of dPen-2, as well as of the types of the tag sequence added.
N-terminal Tags of Drosophila Pen-2-modulated ␥but Not ⑀-Cleavage-Several lines of evidence indicate that the ␥-secretase complex executes "dual" cleavages, namely, ␥and ⑀-cleavages at middle and proximal positions to cytoplasm, respectively, within the transmembrane domain (2,30). Almost all FAD mutations in PSEN genes affect the ␥-cleavage to increase %A␤42, whereas the effect of FAD mutations on ⑀-cleavage remain controversial. However, the ␥-secretase-mediated cleavage of Notch at the site corresponding to the ⑀-cleavage is required for several biological functions (1-3). Thus, the modulation of the ␥-cleavage site (i.e. specific decrease in A␤42 generation) by small molecule compounds without affecting the ⑀-cleavage is one of the plausible therapeutic strategies for AD (1). To monitor the effect of the tagged dPen-2 on the ⑀-cleavage, a reporter S2 cell line stably expressing SC100 fused to gal4 within the cytoplasmic domain (SC100gal4) and EGFP, together with UAS-controlled, destabilized luciferase plasmid, was established. Upon the ⑀-cleavage of SC100gal4 by the ␥-secretase, the intracellular domain of SC100gal4 enters the nucleus to activate the transcription of the luciferase. Luciferase activity and fluorescence light units of EGFP were dependent on the cell number (data not shown). Treatment with DAPT, a potent ␥-secretase inhibitor, decreased the luciferase activity, whereas the fluorescence light units were unaffected (Fig. 3A). Thus, using this reporter cell line, we can assess the degree of the ⑀-cleavage of SC100gal4 by the relative luciferase activity standardized with the fluorescence light units. When we overexpressed the untagged or HA-tagged dPen-2 together with Psn, dNct-V5/His, and dAph-1-FLAG in the SC100gal4 reporter cell line, the relative luciferase activity of the cells expressing HA-dPen-2 was almost similar to that with untagged dPen-2 (Fig. 3B), suggesting that the tagged dPen-2 affected only the ␥-cleavage site.
Several lines of evidence show an inverse correlation between A␤38and A␤42-generating the ␥-secretase activities in mammalian cells. For example, a subset of non-steroidal anti-inflammatory drugs known as A␤42-lowering ␥-secretase modulators (GSMs) increase A␤38 production (28,31), whereas fenofibrate, which acts as an A␤42-raising GSM, decreases an A␤38 generation (supplemental Fig. S1) (28,32). To ascertain whether the ␥-cleavage site was modulated by the N-terminally tagged dPen-2 in Drosophila cells in a similar manner to that by the A␤42-raising GSM, we examined the effect of the tagged dPen-2 on A␤ secretion from S2 cells expressing SC100 by an immunoprecipitation followed by an immunoblot analysis using Tris/Bicine/urea SDS-PAGE (Fig.  4) (27, 28). Unexpectedly, S2 cells showed robust A␤38 secretion at an almost comparable level to that of A␤40 at a steady state. However, the overexpression of HA-dPen-2 decreased the level of A␤38 concomitantly with an increase in A␤42. Collectively, these data suggest that an N-terminally tagged dPen-2 reciprocally modulates the A␤38and A␤42-generating activities but has little effect on the ⑀-cleavage, in Drosophila cells, that was similar to the effect of A␤42-raising GSMs in mammalian cells.
N-terminally Tagged Human PEN-2 Directly Affected the Structure and the Activity of the ␥-Secretase Complex-We have reported that hPEN-2 can compensate for the loss of function of dPen-2 in RNAi-treated Drosophila S2 cells (5). To test the effect of the tag sequence at the N terminus of hPEN-2, we measured the levels of A␤ secreted from cells expressing hPEN-2 or HA-hPEN-2 together with Psn, dNct-V5/His, and dAph-1-FLAG. %A␤42 of cells expressing HA-hPEN-2 was dramatically increased, whereas the overexpression of untagged hPEN-2 only slightly increased it (Table 4). To test this effect in a mammalian cell expression system, recombinant retrovirus encoding untagged or N-terminally FLAG-tagged hPEN-2 was co-infected with retrovirus encoding human PS1, human NCT (hNCT), and human APH-1b (hAPH-1b) into fibroblast cells  obtained from Psen1/Psen2 double knock-out mice (DKO cells) stably expressing APP Swedish mutant (20). Generation of A␤42 from cells expressing FLAG-hPEN-2 was dramatically augmented at 1.9-fold compared with that from untagged hPEN-2-expressing cells (Table 5). In contrast, consistent with the result of dPen-2/⌬2-10, the deletion of 2-10 of hPEN-2 (hPEN-2/⌬2-10) did not affect the A␤42 generation (supplemental Table S1). These data strongly suggest that the effect of the N-terminal tag of hPEN-2 on A␤42 production is observed beyond species and independent of the expression system.
To gain insights into the molecular mechanism whereby the tagged hPEN-2 increases %A␤42, we directly assessed the enzymatic activity of reconstituted human ␥-secretase complex using a baculovirus/Sf9 cell system. We have reported that the reconstituted ␥-secretase complex containing Hx-hPEN-2 showed high %A␤42 in de novo generated A␤ (6). However, the reconstituted enzyme containing untagged hPEN-2 exhibited lower A␤42 production ( Fig. 5A and Table 6), whereas the expression levels of components as well as the endoproteolytic processing of PS1 were unchanged (data not shown). Then, we examined a Michaelis-Menten plot of the reconstituted ␥-secretase activities for A␤40 and A␤42 generation (␥40-and ␥42-secretase activities, respectively, Fig. 5, B and C). ␥40-Secretase activity showed similar profiles in the reconstituted ␥-secretase complex containing untagged or tagged hPEN-2. However, the ␥-secretase complex reconstituted with the tagged hPEN-2 exhibited a significantly higher ␥42-secretase activity at almost 2-fold in V max value (11.5 pM/h for hPEN-2, 24.0 pM/h for Hx-hPEN-2). Notably, K m values of these reconstituted enzymes were almost similar (1.56 M for hPEN-2 and 1.67 M for Hx-hPEN-2), suggesting that the binding affinity of substrate to enzyme was unaffected. Thus, it seems quite likely that the tagged hPEN-2 directly and specifically increased the ␥42-secretase activity.
Recently, using a substituted cysteine accessibility method, we reported that PS forms a catalytic pore embedded within the lipid bilayer (8). Using this method, we can gain insights into the structure of PS1 by the water accessibility of substituted cysteines in cysteine-less PS1. In particular, single-Cys PS1
*p Ͻ 0.05 by Student's t test. **p Ͻ 0.005 by Student's t test.

TABLE 6
Relative -fold change in %A␤42 of enzymatic activity of reconstituted ␥-secretase complex containing hPEN-2 together with PS1, hNCT, and hAPH-1aL Transmembrane domains and His/Xpress tag were depicted as hatched and black boxes, respectively (n ϭ 3, mean Ϯ S.E.). **p Ͻ 0.005 by Student's t test. mutants harboring A246C or L250C were labeled by methanethiosulfonate ethylammonium (MTSEA)-biotin, a membraneimpermeable thiol-directed reagent, from the extracellular side. These labels were sensitive to a transition-state analogue inhibitor, L-685,458. These residues were hypothesized to function as subsites that directly bind to a substrate and are involved in the proteolysis. Moreover, it was suggested that GSMs allosterically inhibit the binding of L-685,458 to the ␥-secretase and directly affect the structure of PS1 (25,(33)(34)(35). To examine whether the structural change in the catalytic pore was correlated to the overproduction of A␤42, the water accessibility of A246C and L250C from extracellular side was analyzed in DKO cells expressing untagged or N-terminally FLAGtagged hPEN-2 together with single-Cys mutant PS1, hNCT, and hAPH-1b (Fig. 6) (32). Water accessibility of I114C, predicted to be located far from the catalytic pore, was unaffected. Surprisingly, the biotinylation of A246C and L250C was reduced upon expression of FLAG-hPEN-2, suggesting that the N-terminal tag of hPEN-2 decreased the water accessibility of the residues at the luminal side of the catalytic pore. Notably, this effect was more prominent in L250C, whereas the biotinylation of A246C was slightly, but constantly, reduced. Then, we examined the effect of fenofibrate, which modulates the ␥-secretase in a way to enhance A␤42 production and decrease A␤38 without any effect on A␤40 generation (supplemental Fig. S1 and Table S1), on the labeling of these residues (Fig. 6). Both the biotinylation of A246C and L250C, but not I114C, was decreased. Similar to the effect by the N-terminally tagged hPEN-2, the labeling of L250C was more strongly reduced. Taken together, our results suggest that the N-terminally tagged PEN-2 increases the ␥42-secretase activity through a direct effect on the catalytic pore in a similar manner to those by the A␤42-raising GSM.

DISCUSSION
In this study, we found that the N-terminal extension of Pen-2 with various tag sequences directly affected the enzymatic property of the ␥-secretase complex in a way that increases A␤42 production both in insect and mammalian cell systems. Notably, the N-terminal tagged Pen-2 caused a decrease in the water accessibility of the luminal side of the catalytic pore in a similar manner to that caused by an A␤42raising GSM, fenofibrate. Thus, our results suggest that the structure of the catalytic pore determines the ␥-cleavage position of substrates and that Pen-2, especially its N-terminal portion, is involved in the ␥-secretase cleavage through structural regulation of the catalytic pore.
The N-terminal tag of Pen-2 increased A␤42 generation regardless of the expression system and the tag sequence, whereas the HA tag in dPen-2 was most potent. The precise molecular function of Pen-2 still remains unknown. However, our in vitro reconstitution data clearly suggest a direct effect of Pen-2 on the enzymatic activity, especially modulation of the cleavage position. Pen-2 exhibits a hairpin-like topology with both the N and C termini being oriented to the luminal side (36,37). Kim and Sisodia (13) reported that the proximal part of the N terminus (residues 10 -16) of hPEN-2 is required for its function, although preservation of the exact amino acid sequence of this region is unnecessary. In fact, the replacement of most of the amino acid residues at the N terminus as well as the distal part (i.e. dPen-2/⌬2-17/rep and HA-dPen-2/⌬2-10) retained the activity of dPen-2, suggesting that the electrostatic composition of this proximal region determines its function related to the assembly and the activation of the ␥-secretase complex. Because recent structural analyses of the ␥-secretase complex still remain at low resolution, the location and stoichiometry of each component are unknown (22,38). However, recently, we and others showed that hPEN-2 binds to PS1 TMD4 (14,15). Thus, the N terminus of Pen-2 might be located close to the luminal side of the catalytic pore, the latter being formed by TMD6 and -7 of PS1. Consistent with this model, the N-terminal tag of hPEN-2 decreased the water accessibility of the luminal side of the catalytic pore. Therefore, it is tempting to speculate that the N terminus of Pen-2 might behave like a "lid" of the pore, or directly affect the structure of the pore in such a way to regulate the cleavage positions of substrates. However, the length of the tag sequences did not correlate with the degree in the A␤42 overproduction, because 9 and 40 amino acid lengths of FLAG and His/Xpress tags, respectively, yielded comparable increases in %A␤42. In addition, the HA tag showed the most potent effect, although the number of amino acid residues of HA, myc, and FLAG tags are almost similar. The HA2 tag caused an almost comparable increase with the HA-tagged dPen-2, suggesting that the critical sequence at a certain position in the elongated N terminus of dPen-2, but not the length or the net charge of the tag, determines a degree of the increase in A␤42 generation. Notably, the biotinylation of L250C was more prominently affected than that of A246C, the latter being located close to the extracellular side. Thus, we prefer a model in which a tag sequence at the N terminus of Pen-2 renders the catalytic pore around the midportion nar- rower and less accessible to labeling by MTSEA-biotin. It has been shown that antibodies to the N-terminal region of dPen-2 or hPEN-2 failed to pull down the active ␥-secretase complex in CHAPSO-or CHAPS-solubilized lysates, whereas free Pen-2 polypeptides were efficiently immunoprecipitated from Triton X-100-solubilized lysate, in which the ␥-secretase complex is dissociated (Refs. 37 and 39 and data not shown). Moreover, little biotinylation of hPEN-2 by MTSEA-biotin in the active ␥-secretase complex was observed, despite hPEN-2 carrying one luminal cysteine residue closely located to TMD1. 5 In contrast, the overexpressed hPEN-2 that was not bound to PS1 was labeled by MTSEA-biotin. 6 Thus, it is highly likely that the N-terminal region of Pen-2 is buried and hidden within the active ␥-secretase complex. Taken together, the N terminus of Pen-2 might function as a molecular chaperone that contributes to the structural arrangement of TMDs as well as of the catalytic pore. The tags, that contain hydrophilic and charged residues, might change an electrostatic configuration of the N terminus of Pen-2 and modify the structure of the catalytic pore in a way to increase the A␤42 generation.
Recently much attention is focused on the molecular mechanism of the A␤42-lowering GSMs as the plausible therapeutics without Notch-related adverse effects for AD (1). Several GSMs, including a subset of non-steroidal anti-inflammatory drugs, directly regulate the proteolytic cleavages at A␤38 and A␤42 positions, without affecting ⑀-cleavage activity in a cellbased assay (25,31,32). Although the precise molecular mechanism whereby GSMs modulate the ␥-secretase activity remains unknown, enzymatic analyses indicated that the A␤42lowering GSMs modulate the binding site for transition-state analogue inhibitors in a noncompetitive, allosteric manner (25,35). Moreover, FLIM assay showed that the A␤42-lowering GSMs increase the proximity of PS1 NTF and CTF (33), whereas FAD-linked PS mutations that augment A␤42 generation decrease the distance (34). Collectively, these results suggest that the structural changes in the catalytic site formed by PS1 NTF and CTF affect the enzymatic property regarding cleavage positions. Here we found that an A␤42-raising GSM and the N-terminally tagged Pen-2 caused a similar effect on the A␤38/42-generating activity and the water accessibility of the PS1 TMD6, thus constituting the luminal side of the catalytic pore. We previously reported that the biotinylation of L250C was inhibited by L-685,458 and DAPT, potent, non-selective ␥-secretase inhibitors, whereas the labeling of A246C was affected only by L-685,458 (8). Thus, it may be reasonable to speculate that the A␤42-raising GSM directly targets the luminal side of the catalytic pore, as do the transition-state analogue inhibitors. However, it was reported that non-selective ␥-secretase inhibitors (i.e. DAPT and transition-state analogue) had no effect on the proximity of PS1 NTF and CTF in the FLIM analysis (33,40), suggesting that the molecular mechanisms whereby these conventional ␥-secretase inhibitors and GSMs regulate the ␥-secretase activity are different. Another possibility is that GSMs allosterically affect the structure of the catalytic pore through the binding to other regions in PS1 or to different molecular targets, including Pen-2. The observation that the A␤42-raising GSM and the tagged Pen-2 had a similar effect on the biotinylation of the luminal part of the catalytic pore prompts us to speculate that the compound targets the position where the N-terminal region of Pen-2 is located. Further attempts to clarify the mode of binding as well as the molecular targets of GSM would resolve this issue, in the same way that we and others have identified the functional domains in the ␥-secretase components using small compounds harboring photolabile moieties as a molecular probe (26,41,42).
In summary, we have identified a novel mode of the ␥-secretase to increase A␤42, in which the extension of the N-terminal region of Pen-2 affects the regulation of the ␥-secretase cleavage position, in a similar manner to the effect of the A␤42raising GSM. Nevertheless, whatever the precise underlying mechanism may be, our findings are pharmacologically relevant and could have major therapeutic implications, because it is strongly suggested that the N terminus of Pen-2 and/or the structural changes occurring in the catalytic pore are one of the major determinants for the ␥-secretase cleavage positions. Further investigations using a chemical biological approach should provide us with clues to the development of the A␤42-lowering GSMs for AD therapeutics.