Figure 2. Structure of b-secretase, BASE1/2 cleavage sites on APP and disease-causing mutations p-secretase (BACE1) and its homolog, BACE2, exhibit 52% amino acid sequence identity and 68% similarity, and BACE2 cleaves APP and short peptides in a p-secretase-like manner: activity is similarly increased at Asp1 by the Swedish double mutation and prevented by a P1 Met-to-Val mutation (17). However, BACE2 is not expressed well in the brain, suggesting that it may play little, if any role in AD plaque formation. Moreover, unlike BACE1, BACE2 generates little Glu11-Ap (Figure 2) but efficiently performs an additional proteolysis in the middle of the Ap region, resulting in the production of Phe20-Ap and Ala21-Ap, with the implication that BACE2 might limit the production of pathogenic forms of Ap. The biological role of BACE2 is unknown. If BACE2's role is critical, inhibitors selective for BACE1 over this homolog may be needed to minimize toxicity.
A crystal structure of p-secretase bound to a hydroxyethyl transition-state analog inhibitor at 1.9 A resolution shows that the bilobal structure of p-secretase has the conserved general folding found in many other aspartyl proteases (18, 19). The six cysteine residues in the ectodomain form three disulfide bonds. The inhibitor is located in the substrate binding cleft between the amino- and carboxy-terminal lobes, and as expected, the transition-state mimicking hydroxyethyl moiety is coordinated with the two active site aspartates. As with a number of other aspartyl proteases, p-secretase possesses a "flap" that partially covers the cleft, and the backbone of the inhibitor is mostly in an extended conformation. However, p-secretase does display some structural differences, at least compared with pepsin, that may be turned to advantage toward the development of selective inhibitors.
Several reports on BACE1 knockout mice concluded that this enzyme is the major p-secretase in the brain (20-22). The BACE1 knockout mice were healthy, viable, and appeared normal in gross anatomy, tissue histology, hematology, and clinical chemistry. These findings indicate that inhibition of BACE should lead to dramatic decreases in brain Ap levels and that such inhibition may not lead to mechanism-based toxicity. Despite the promising results from the BACE1 knockout studies, this protease likely has other normal substrates, and it remains to be seen whether long-term p-secretase inhibition is tolerable in aging adult humans. Most recently, BACE1 expression and p-secretase activity have been reported to be elevated in the AD brain, further emphasizing the connection between this protease and the disease (23, 24).
B-Secretase inhibitors - p-secretase is an attractive pharmacological target for AD, and the search for potent and selective inhibitors of this enzyme has intensified during the past few years. Despite the availability of X-ray structures of the enzyme, most of the reported p-secretase inhibitors are still substrate-based peptidomimetics. The large active site of p-secretase apparently presents a challenge for the development of nonpeptidic small molecule inhibitors. One exception is the tetralin derivative ±, which inhibits activity of recombinant p-secretase with an IC5o of 0.35 nM (25). Identifying nonpeptidic compounds is critical for obtaining effective clinical agents, because peptide analogs typically do not display good enough pharmacokinetic properties (e.g., the ability to cross the blood-brain barrier) to become drugs.
Reported SARs have involved variations on the original peptide structure based on the sequence EVNLDAEF, which is derived from the APP p-secretase site containing the Swedish P2-P1 double mutation. This sequence was used to design potent first-generation p-secretase inhibitors, such as hydroxyethylene isostere transition-state analogues 2 (Ki = 1.6 nM) and later 3 (Ki = 0.2 nM) (19, 26). A recently disclosed version of 2 is 4, which inhibits recombinant p-secretase with an IC5o of 49 nM (27). Importantly, a recent kinetic study using synthetic peptide libraries and mass spectrometric analysis revealed a peptide EIDLMVLD which was cleaved with a kcat/KM value 14-fold better than EVNLDAEF (28). This finding should aid the development of more potent peptidic inhibitors. At the same time, it suggests that other better substrates for p-secretase may exist. Cross-inhibition with the homologous BACE2 has not been reported (29). Such inhibition may cause unforeseen toxicity. BACE2 knockout mice are being generated to address this issue.
Additionally, continual work on the development of statine-based peptidomimetics selective against BACE has resulted recently in the identification of the cell-permeable 5 (IC5o = 0.12 nM), which is being evaluated in vivo (30).
y-Secretase characterization - This enzyme has been considered central to understanding the molecular basis of AD, because it determines the proportion of the highly aggregation-prone Ap42 peptide. y-Secretase has also been of interest because it somehow hydrolyzes within the middle of the transmembrane region of APP. The enzyme is inhibited by classical transition-state mimicking motifs for aspartyl proteases, suggesting that it falls into this mechanistic category of proteases (31-33). Knocking out PS in mice showed that it mediates y-secretase cleavage of APP (34-36), begging the question: What is the role of this protein in y-secretase activity?
PS is cut into two pieces that remain associated (37, 38). This heterodimer is thought to be the active form of the protein, because it is metabolically stable and its formation is tightly regulated by complexation with other cellular factors (37, 39, 40). PS also contains two conserved transmembrane aspartates, each contributed by one of the PS subunits, that are predicted to lie the same distance within the membrane and to roughly align with the y-secretase cleavage site in APP. Mutation of either asparate prevented the formation of PS subunits and blocked y-secretase processing of APP (41).
These results suggested that PS might be the catalytic component of y-secretase: upon interaction with other, limiting cellular factors, PS undergoes autoproteolysis via the two aspartates, and the two PS subunits remain together, each contributing one aspartate to the active site of y-secretase. More direct evidence that PS is the catalytic component of y-secretase came from affinity labeling studies using transition-state analogue inhibitors: the heterodimeric form of PS was specifically tagged (42, 43). Also, a non-transition state analog inhibitor of y-secretase likewise crosslinked presenilin subunits (44).
PS is not only involved in the proteolytic processing of the transmembrane domain of APP but is also critical for processing of the transmembrane region of the Notch receptor, a signaling molecule crucial for cell-fate determinations during embryogenesis (45). Release of the intracellular domain of Notch, a process mediated by PS, is essential for Notch signaling (46, 47). The responsible PS-dependent protease that cleaves Notch appears to be Identical to the one that cleaves APP (35, 36, 48, 49). Because Notch is also required for cell differentiation during adulthood (e.g., hematopoiesis), concerns have been raised about mechanism-based toxicities (e.g., immunosuppression) arising from y-secretase inhibition (50, 51).
Presenilin Mcastrin Aph-1 Peri-2
Presenilin Mcastrin Aph-1 Peri-2
.. .SN RjGAIIGLM VGGVVIATVIVITLVML|KKK. .. | ^ ft |
Figure 3. Scheme of y-secrelase complex, y-secretase cleavage sites on APP and sites of AD-causing mutations
Although PS appears to be the catalytic component of y-secretase, it does not work alone, requiring complexation with at least three other recently identified membrane proteins: nicastrin, Aph-1, and Pen-2 as shown in Figure 3 (52-54). Expression of all four proteins together results in increased formation of PS subunits and elevated y-secretase activity (55-57). Moreover, these components assemble together and bind to an immobilized y-secretase inhibitor (55, 56). These findings are consistent with the hypothesis that, after complexation with its three partners, PS undergoes autoproteolysis to become the catalytic component of y-secretase, the active site lying at the interface between the two PS subunits. This is a major advance: the complete identification of y-secretase should greatly facilitate our understanding of its mechanism and expedite the search for effective inhibitors. Moreover, from the perspective of basic biochemistry, y-secretase appears to be a founding member of a new class of proteases containing membrane-embedded active sites (58).
v-Secretase inhibitors - Although the biochemistry and mechanism of y-secretase are still not completely understood, many new y-secretase inhibitors have been identified as a result of cell-based screening of both diverse and focused libraries. Aspartyl protease transition-state mimics, such as hydroxyethylene-based 6 and (hydroxyethyl)urea-based 7, which are closely related to previously reported hydroxyethylene-based peptidomimetics, potently inhibit A04o and Ap42 production in cell-based and/or cell-free assays (32, 59, 60). A photoactivatable and biotinylated derivative of 7 has been used to assess the mechanism of action of structurally diverse y-secretase inhibitors, and most them apparently act by binding or allostericaliy affecting the y-secretase active site (61). Epoxide 8 has been shown to affect the active site in a time-dependent manner, suggesting it irreversibly inactivates the enzyme, possibly by reacting with one of the active site aspartates (61, 62). SAR studies have resulted in identification of fenchylamine sulfonamide 9 (ICso = 1 nM), a-hydroxyacid-based 10 (IC5o = 160 nM) and dipeptide ester 11 (IC5o = 130 nM) (63-65).
Compound 12 was reported to inhibit the formation of presenilin heterodimers with an IC5o of 60 nM (66). Recently disclosed highly active y-secretase inhibitors have included benzodiazepine 13 (ICso = 300 pM), dipeptide analogue 14 and second generation analogue benzocaprolactam 15 (44,67,68).
Was this article helpful?