Difluoro ketone peptidomimetics suggest a large S1 pocket for Alzheimer's gamma-secretase: implications for inhibitor design.

The final step in the generation of the amyloid-beta protein (Abeta), implicated in the etiology of Alzheimer's disease, is proteolysis within the transmembrane region of the amyloid precursor protein (APP) by gamma-secretase. Although considered an important target for therapeutic design, gamma-secretase has been neither well-characterized nor definitively identified. Previous studies in our laboratory using substrate-based difluoro ketone and difluoro alcohol transition-state analogue inhibitors suggest that gamma-secretase is an aspartyl protease with loose sequence specificity. To further characterize the active site of gamma-secretase, we prepared a series of difluoro ketone peptide analogues with varying steric bulkiness in the P1 position and tested the ability of these compounds to inhibit Abeta production in APP-transfected cells. Incorporation of bulky, aliphatic P1 side chains, such as sec-butyl or cyclohexylmethyl, led to increased gamma-secretase inhibitory potency, suggesting a large S1 pocket to accommodate these substituents and providing further evidence for loose sequence specificity. The cyclohexylmethyl P1 substituent allowed N-terminal truncation to a low-molecular-weight compound (<600 Da) that effectively blocked Abeta production (IC(50) approximately 5 microM). This finding suggests that optimal S1 binding may allow the development of potent inhibitors with ideal pharmaceutical properties. Moreover, a difluoro alcohol analogue with a cyclohexylmethyl P1 substituent was equipotent with its difluoro ketone counterpart, providing strong evidence that gamma-secretase is an aspartyl protease. All new analogues inhibited total Abeta and Abeta(42) production with the same rank order of potency and increased Abeta(42) production at low concentrations, providing further evidence for distinct gamma-secretases that are nevertheless closely similar with respect to active site topology and mechanism.

Transition-state analogue inhibitors of gamma-secretase bind directly to presenilin-1.

The beta-amyloid precursor protein (beta-APP), which is involved in the pathogenesis of Alzheimer's disease, and the Notch receptor, which is responsible for critical signalling events during development, both undergo unusual proteolysis within their transmembrane domains by unknown gamma-secretases. Here we show that an affinity reagent designed to interact with the active site of gamma-secretase binds directly and specifically to heterodimeric forms of presenilins, polytopic proteins that are mutated in hereditary Alzheimer's and are known mediators of gamma-secretase cleavage of both beta-APP and Notch. These results provide evidence that heterodimeric presenilins contain the active site of gamma-secretase, and validate presenilins as principal targets for the design of drugs to treat and prevent Alzheimer's disease.

Toward the characterization and identification of gamma-secretases using transition-state analogue inhibitors.

The amyloid-beta protein (A beta), strongly implicated in the etiology of Alzheimer's disease (AD), is formed from the amyloid-beta precursor protein (APP) through sequential proteolysis by beta- and gamma-secretases. Cleavage by gamma-secretase takes place within the middle of the single transmembrane region of APP and results primarily in 40- and 42-amino acid A beta C-terminal variants, A beta 40 and A beta 42. The latter form of A beta is highly fibrillogenic, is invariably elevated in autosomal-dominant forms of AD, and is the major A beta component found presymptomatically in cerebral deposits. Thus, blocking production of A beta in general and A beta 42 in particular is considered an important therapeutic goal. We have developed transition-state analogue inhibitors of gamma-secretase as molecular probes for characterizing the active site of this enzyme, as pharmacological tools for understanding its role in biology, and as affinity labels toward its definitive identification. Specifically, we found that: (1) difluoro ketone and difluoro alcohol peptidomimetics are effective inhibitors of gamma-secretase activity in APP-transfected cells, strongly suggesting an aspartyl protease mechanism; (2) gamma-secretases that form A beta 40 and A beta 42 are pharmacologically distinct but are nevertheless closely similar; (3) large hydrophobic P1 substituents increase the inhibitory potency of these peptidomimetics, suggesting a large complementary S1 pocket for gamma-secretases; (4) A beta 42 production is increased several fold over control by these gamma-secretase inhibitors after replacement with inhibitor-free media; (5) a bromoacetamide derivative of one of these analogues continues to inhibit total A beta and A beta 42 production hours after replacement with compound-free media and should help identify the target(s) of these protease transition-state mimics.

Two transmembrane aspartates in presenilin-1 required for presenilin endoproteolysis and gamma-secretase activity.

Accumulation of the amyloid-beta protein (Abeta) in the cerebral cortex is an early and invariant event in the pathogenesis of Alzheimer's disease. The final step in the generation of Abeta from the beta-amyloid precursor protein is an apparently intramembranous proteolysis by the elusive gamma-secretase(s). The most common cause of familial Alzheimer's disease is mutation of the genes encoding presenilins 1 and 2, which alters gamma-secretase activity to increase the production of the highly amyloidogenic Abeta42 isoform. Moreover, deletion of presenilin-1 in mice greatly reduces gamma-secretase activity, indicating that presenilin-1 mediates most of this proteolytic event. Here we report that mutation of either of two conserved transmembrane (TM) aspartate residues in presenilin-1, Asp 257 (in TM6) and Asp 385 (in TM7), substantially reduces Abeta production and increases the amounts of the carboxy-terminal fragments of beta-amyloid precursor protein that are the substrates of gamma-secretase. We observed these effects in three different cell lines as well as in cell-free microsomes. Either of the Asp --> Ala mutations also prevented the normal endoproteolysis of presenilin-1 in the TM6 --> TM7 cytoplasmic loop. In a functional presenilin-1 variant (carrying a deletion in exon 9) that is associated with familial Alzheimer's disease and which does not require this cleavage, the Asp 385 --> Ala mutation still inhibited gamma-secretase activity. Our results indicate that the two transmembrane aspartate residues are critical for both presenilin-1 endoproteolysis and gamma-secretase activity, and suggest that presenilin 1 is either a unique diaspartyl cofactor for gamma-secretase or is itself gamma-secretase, an autoactivated intramembranous aspartyl protease.

Additive effects of PS1 and APP mutations on secretion of the 42-residue amyloid beta-protein.

Humans harboring missense mutations in the presenilin 1 (PS1) gene undergo progressive cerebral deposition of the 42-residue amyloid beta-protein (A beta 42) at an early age and develop severe Alzheimer's disease. A beta 42 is selectively elevated in the conditioned media of cells expressing mutant but not wild-type PS1, indicating that presenilin mutations alter APP processing. Here we analyze the effects of various PS1 mutant constructs on the cellular production of A beta 42. A construct expressing only the PS1 N-terminal endoproteolytic fragment with the mutation Y115H causes no significant increase in A beta 42, whereas a full-length PS1 construct with the same mutation does. This result suggests that the pathogenic effect of mutant presenilins is produced by the full-length molecule even though only a minor proportion of total PS1 occurs as holoprotein in tissues and cell lines. We demonstrate that the effects of two different PS1 mutations are additive when engineered into the same PS1 molecule. Therefore, two mutations alter gamma-secretase processing of APP more than one and the PS1 mutations described to date do not cause the maximum possible PS1-mediated rise in A beta 42. When a PS1 mutation was expressed in cells carrying the APPV717I mutation, A beta 42 rose dramatically to become the predominant secreted A beta species, an observation of interest for transgenic modeling of AD. Our results are consistent with the hypothesis that presenilin is a major regulator of the proteolytic processing of APP by gamma-secretases.

Evidence that the 42- and 40-amino acid forms of amyloid beta protein are generated from the beta-amyloid precursor protein by different protease activities.

Cerebral deposition of the amyloid beta protein (A beta) is an early and invariant feature of Alzheimer disease (AD). Whereas the 40-amino acid form of A beta (A beta 40) accounts for approximately 90% of all A beta normally released from cells, it appears to contribute only to later phases of the pathology. In contrast, the longer more amyloidogenic 42-residue form (A beta 42), accounting for only approximately 10% of secreted A beta, is deposited in the earliest phase of AD and remains the major constituent of most amyloid plaques throughout the disease. Moreover, its levels have been shown to be increased in all known forms of early-onset familial AD. Thus, inhibition of A beta 42 production is a prime therapeutic goal. The same protease, gamma-secretase, is assumed to generate the C termini of both A beta 40 and A beta 42. Herein, we analyze the effect of the compound MDL 28170, previously suggested to inhibit gamma-secretase, on beta-amyloid precursor protein processing. By immunoprecipitating conditioned medium of different cell lines with various A beta 40- and A beta 42-specific antibodies, we demonstrate a much stronger inhibition of the gamma-secretase cleavage at residue 40 than of that at residue 42. These data suggest that different proteases generate the A beta 40 and A beta 42 C termini. Further, they raise the possibility of identifying compounds that do not interfere with general beta-amyloid precursor protein metabolism, including A beta 40 production, but specifically block the generation of the pathogenic A beta 42 peptide.

Inhibition of amyloid beta-protein production in neural cells by the serine protease inhibitor AEBSF.

Cerebral deposition of amyloid beta protein (A beta) is an early and critical feature of Alzheimer's disease. A beta production requires the proteolytic release of A beta from the beta-amyloid precursor protein (beta APP). Thus, inhibition of A beta release is a prime therapeutic goal. Here, we show that the broad spectrum, irreversible serine protease inhibitor, AEBSF, inhibits the constitutive production of A beta in five different human cell lines, both neural and nonneural. AEBSF also stabilizes full-length beta APP and enhances alpha-secretion, as shown by an increase in the proteolytic derivative, alpha-APPS. Further, we demonstrate that the inhibitory effect of AEBSF is specific for A beta proteins starting at Aspartate 1, suggesting that AEBSF directly inhibits beta-secretase, the Methionine-Aspartate (Met-Asp)-cleaving enzyme. These results indicate that specific inhibition of this A beta-generating protease is possible in living human neural cells and provide information about the characteristics of this as yet unidentified enzyme.

GTP gamma S loading of endothelial cells stimulates phospholipase C and uncouples ATP receptors.

The effects of GTP gamma S, a stable GTP analogue that can activate guanine nucleotide-binding proteins, on phospholipase C activation/calcium mobilization were studied in intact cultured bovine aortic endothelial cells (BAEC). Phosphoinositide metabolism and cytosolic free Ca2+ concentration [( Ca2+]i; fura-2 fluorescence) were studied after the cells were transiently permeabilized, loaded with different guanine nucleotides, and then allowed to reseal and recover. Intracellular GTP gamma S stimulated a dose-dependent [median effective concentration (EC50) 2.5 microM] decrease in basal [3H]phosphoinositide content. Phosphatidylinositol 4,5-bisphosphate, phosphatidylinositol 4-bisphosphate, and phosphatidylinositol levels decreased to 57 +/- 9, 63 +/- 8, and 74 +/- 8% control levels, respectively, in BAEC loaded with approximately 85 microM GTP gamma S. Basal inositol trisphosphate (IP3) and [Ca2+]i were increased in GTP gamma S-loaded BAEC when compared with sham-loaded BAEC. In control BAEC, the purinergic receptor agonist ATP (100 microM) induced rapid increases in [Ca2+]i and IP3. However, BAEC that had been intracellularly loaded with GTP gamma S [median inhibitory constant (IC50) 1 microM] or 5'-guanylyl-imidodiphosphate exhibited decreased calcium mobilization in response to ATP. Ionomycin (calcium ionophore)-releasable pools of calcium were similar in sham- and GTP gamma S-loaded cells, suggesting that total intracellular calcium had not been depleted by the permeabilization protocol. The diminished calcium mobilization in response to ATP was associated with decreases in ATP-stimulated PIP2 hydrolysis and IP3 formation. In addition, GTP gamma S loading did not increase basal levels of cyclic AMP. Intracellular GDP beta S, GDP, or GTP did not inhibit ATP-stimulated increases in [Ca2+]i or IP3.(ABSTRACT TRUNCATED AT 250 WORDS)