Angiotensin-converting Enzyme Degrades Alzheimer Amyloid β-Peptide (Aβ); Retards Aβ Aggregation, Deposition, Fibril Formation; and Inhibits Cytotoxicity*

We have demonstrated that the angiotensin-converting enzyme (ACE) genotype is associated with Alzheimer's disease (AD) in the Japanese population (1). To determine why ACE affects susceptibility to AD, we examined the effect of purified ACE on aggregation of the amyloid β-peptide (Aβ)in vitro. Surprisingly, ACE was found to significantly inhibit Aβ aggregation in a dose response manner. The inhibition of aggregation was specifically blocked by preincubation of ACE with an ACE inhibitor, lisinopril. ACE was confirmed to retard Aβ fibril formation with electron microscopy. ACE inhibited Aβ deposits on a synthaloid plate, which was used to monitor Aβ deposition on autopsied brain tissue. ACE also significantly inhibited Aβ cytotoxicity on PC12 h. The most striking fact was that ACE degraded Aβ by cleaving Aβ-(1–40) at the site Asp7-Ser8. This was proven with reverse-phase HPLC, amino acid sequence analysis, and MALDI-TOF/MS. Compared with Aβ-(1–40), aggregation and cytotoxic effects of the degradation products Aβ-(1–7) and Aβ-(8–40) peptides were reduced or virtually absent. These findings led to the hypothesis that ACE may affect susceptibility to AD by degrading Aβ and preventing the accumulation of amyloid plaques in vivo.

Progressive cerebral dysfunction in Alzheimer's disease (AD) 1 is accompanied by innumerable extracellular amyloid deposits in the form of senile plaque and microvascular amyloid. Amyloid protein is derived from the integral membrane polypeptide, ␤-amyloid precursor protein (␤APP). The released 39 -43 residue amyloid ␤-peptide (A␤) may subsequently undergo aggregation to form amyloid fibrils under the influence of various amyloid-associated factors (2). The aggregation and deposition of A␤ has been linked to the toxic effects causing cell damage in AD. Because A␤ is present in both normal and AD subjects, an answer to the question of why A␤ accumulates in AD but not in the normal brain may lead to a possible cure for AD.
Angiotensin-converting enzyme (ACE; dipeptidyl carboxypeptidase, EC 3.4.15.1) is a membrane-bound ectoenzyme. It catalyzes the conversion of angiotensin I (AngI) to angiotensin II (AngII), which plays an important role in blood pressure and body fluid and sodium homeostasis (3). The cloning of the ACE gene revealed a 287-bp insertion (I)/deletion (D) polymorphism in intron 16. The serum ACE activity of the ACE DD genotype was twice as high as that of the ACE II genotype (4). The ACE genotype is considered to be associated with hypertension, coronary artery disease, left ventricular hypertrophy, myocardial infarction, and diabetic nephropathy (5)(6)(7). In particular, the ACE DD genotype is considered to be a risk factor for vascular diseases.
We have compared the distribution of an I/D polymorphism of the gene coding for ACE in 133 Japanese sporadic AD patients and 257 control subjects (1). The association between AD and ACE genotypes or alleles was found to be significant. The frequency of the ACE II genotype was 1.4ϫ higher in AD than in controls, whereas that of ACE DD genotypes was only 0.4ϫ as high. Moreover, the altered distribution of ACE alleles with AD patients appears to be independent of ApoE (1). The association between AD and ACE genotypes was even more significant in the Japanese population than in the British population (8). Although several reports published recently elucidate the association between ACE genotype and AD (9 -11), the mechanism of how ACE influences susceptibility to AD remains unclear. Here, we provide the first evidence that ACE significantly inhibits the aggregation, deposition, and cytotoxicity of A␤ in vitro by degrading A␤-  at the site Asp 7 -Ser 8 .

EXPERIMENTAL PROCEDURES
Preparation of ACE and Immunoblotting-Somatic ACE was purified from human seminal plasma by using lisinopril-coupled Sepharose as described (12). Immunoblotting was done with anti-somatic ACE antibodies as described (12).
ACE Activity Assay-Enzymatic activity of ACE was determined with the ACE color kit (Fujirebio, Japan) in which p-hydroxyhippuryl-L-histidy-L-leucine was used as substrate (13). ACE activity was monitored by absorbance at 505 nm.
Inhibition of ACE Activity by Lisinopril-Lisinopril was added to fixed amounts of PBS-diluted seminal plasma to the final concentrations described in the legend to Fig. 2. After incubation for 15 min at room temperature, ACE activities were determined.
Aggregation Studies-Synthetic A␤ (1-40, 1-7, 8 -40) (Peptide Institute, Osaka, Japan) was dissolved first in dimethyl sulfoxide (Me 2 SO) and then in PBS to form the stock solution (1 mM A␤ containing 25% Me 2 SO). The stock solution was diluted 10-fold with PBS and incubated with or without ACE at 37°C for 4 days. Aggregation of A␤ was measured by adding 10 l of A␤ solution into 0.5 ml of thioflavine T (ThT) solution (final concentration: 3 M in 50 mM sodium phosphate buffer, pH 6.0) and measuring the fluorescence intensity ( ex at 450 nm, em at 482 nm). A␤ Deposition Assay-Various concentrations of ACE were incubated with 10 nCi of 125 I-A␤ (Amersham Biosciences) in 100 l of TE buffer (50 mM Tris, pH 7.5, containing 0.1% BSA) at 37°C for 3 h. Then the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
resulting solution was incubated in a Synthaloid Drug Screening Plate (Quality Controlled Biochemicals Inc.). The deposited A␤ was detected as radioactive signals according to the manufacturer's instructions.
Electron Microscopy-100 M A␤ solutions containing 2.5% Me 2 SO and preincubated with or without ACE and lisinopril (as prepared in the aggregation studies) were examined. The fibril-formed peptide in the solutions was adsorbed onto 200-mesh Formvar-coated copper grids and negative-stained with 2% uranyl acetate. The fibrils were observed with an electron microscope at 80 kV.
Reverse Phase HPLC-Fifty microliters of the reaction mixture was injected onto a TSK gel ODS120T column (0.64 ϫ 25 cm, particle size 5 m) and eluted at 1 ml/min with a linear gradient of 0ϳ80% acetonitrile, over a period of 50 min. The peaks monitored at 210 nm were collected.
Amino Acid Sequence-Microsequencing was performed automatically by a gas-liquid sequencer (Shimadzu, model PSQ1). Phenylthiohydantoin (PTH)-derivatives were identified by Shimadzu LC system PTH-1 on a Wakopak WS-PTH column (0.64 ϫ 25 cm, particle size: 5 m) with isocratic elution of the PTH-derivative mobile phase. The data were analyzed by a chromatopak CR4A data processor (Shimadzu).

Matrix-assisted Laser Desorption/Ionization Time-of-Flight/Mass Spectrometry (MALDI-TOF/MS)-
The HPLC eluates were dissolved in 50% acetonitrile, 0.1% trifluoroacetic acid. Aliquots of 0.5 l were applied onto the MALDI target and allowed to air dry. All mass spectra were recorded with a Voyager-DE PRO mass spectrometer (Applied Biosystems, Japan) operated in the linear or reflection mode. MALDI-MS spectra were calibrated using several peaks as external standards. Obtained spectra were analyzed using the sequest algorithm with public data bases.

RESULTS
Preparation of ACE-Somatic ACE is present in serum and seminal plasma. We measured the ACE activity in human seminal plasma (844.84 Ϯ 344.27 units/liter (n ϭ 139)). 2 In contrast, normal human serum ACE activity has been reported to be 7.60 Ϯ 2.01 units/liter (n ϭ 173) (14). Because the activity in seminal plasma is over 100ϫ higher than that in serum, we purified ACE from seminal plasma using the ACE inhibitor, lisinopril, as an affinity ligand. The purity of ACE eluted from the lisinopril-coupled Sepharose column was confirmed by electrophoregram. Purified ACE showed a single band with a molecular mass of 180 kDa using Coomassie Blue staining, and this band was strongly recognized by the anti-somatic ACE monoclonal antibody in immunoblotting (Fig. 1). The purified ACE had an activity of about 20 unit/mg of protein that could be inhibited by lisinopril at a final concentration ranging from 10 to 0.01 M in a dose response manner (Fig. 2).
ACE Inhibited A␤ Aggregation-Synthetic A␤ in aqueous buffer tends to self-aggregate (15,16), and only self-aggregated A␤ exerts cytotoxicity. We detected A␤ aggregation quantitatively using fluorescence of ThT, a reagent that associates rapidly with aggregated A␤ but not with monomeric or dimeric A␤, giving rise to a new excitation absorption at 450 nm (17). As shown in Fig. 3, 100 M A␤ solution aggregated remarkably after incubating at 37°C for 4 days. When A␤ solution was incubated with ACE, the aggregation was significantly inhibited, and the inhibition was dose-dependent. A concentration of 240 milliunits/100 l ACE reduced A␤ aggregation to about 20% of the control (p ϭ 1.7 ϫ 10 Ϫ5 versus PBS). The presence of 2.5% Me 2 SO in the solution did not affect A␤ aggregation (data not shown).
To elucidate whether the inhibition was specific, a final concentration of 10 M of lisinopril, which could inhibit about 98% of ACE activity (Fig. 2), was added to the ACE solution 15 min before incubation with A␤. As shown in Fig. 3, pretreat-2 M. Kamata and J. Hu, unpublished data. ment with the ACE inhibitor blocked 99% of the inhibitory effect of ACE (p ϭ 0.9 versus PBS), suggesting this inhibitory effect was based on the active site of the enzyme. As a negative control, BSA at double the concentration of ACE was added to the A␤ solution, but no significant alteration was observed (p ϭ 0.5 versus PBS).
ACE Inhibited A␤ Deposition on Synthaloid Plate-Because the process of in vitro A␤ deposition at physiological concentrations onto plaques in AD brain preparations is cumbersome, Esler et al. (18) prepared a synthaloid (synthetic template) for A␤ deposition by immobilizing fibrillar A␤ in a polymer matrix. It was demonstrated that radiolabeled A␤ deposited onto synthaloid similar to plaques in AD brain (the natural template). We used synthaloid to predict the inhibition by ACE of A␤ deposition in autopsied AD brain preparations. The amount of 125 I-A␤ deposited onto the plate was 3200 cpm. Addition of 240 milliunits of ACE reduced the deposition to 1648 cpm, approximately half of the control (Fig. 4). ACE remarkably inhibited the 125 I-A␤ deposition in a dose-dependent manner.
ACE Inhibited A␤ Fibril Formation-Fibril formation of A␤ was investigated by electron microscopy (Fig. 5). Abundant amyloid fibrils were found in the incubated A␤ solution (100 M; containing 2.5% Me 2 SO). In contrast, very few fibrils were observed in the A␤ solution incubated with 240 milliunits of ACE.
ACE Inhibited A␤ Cytotoxicity-Compared to other rat and human cell types of neuronal origin, the rat pheochromocytoma PC12 h cell was found to be the most sensitive to A␤ (19). A novel cell proliferation and cytotoxicity assay method using a tetrazolium salt that produces a water-soluble formazan dye was reported. The new method was able to measure the dehydrogenase enzymes in living cells in a more convenient and sensitive way than the most currently utilized 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) methods (20,21). We therefore used the PC12 h cell and the new cytotoxicity assay to evaluate the inhibitory effect of ACE on A␤ cytotoxicity.
As shown in Fig. 6, incubation of 10 M aggregated A␤ with PC12 h cells for 3 days caused 53% cell death, while preincubation with 240 milliunits of ACE increased cell survival to 80 Ϯ 5.2% (p ϭ 0.02 versus PBS). Similarly, 160 and 60 milliunits of ACE resulted in cell survival to 65 Ϯ 0.7% (p ϭ 0.0001 versus PBS) and 60 Ϯ 0.2% (p ϭ 0.008 versus PBS), respectively. The effect of ACE on PC12 h cell survival was dose-dependent and was significant versus controls. Preincubation of double the amount of BSA, instead of ACE, did not affect cell survival (50 Ϯ 1%). On the other hand, ACE treated with 10 M lisinopril significantly blocked the inhibitory effect of ACE on A␤ cytotoxicity, resulting in the reduction of 57 Ϯ 2% (p ϭ 0.005 versus ACE (240 milliunits)). These data suggest that the inhibitory effect on A␤ cytotoxicity was a specific effect by ACE. and cytotoxicity, we tried to determine if any degradation occurred during the incubation of A␤-(1-40) with ACE and discovered a new degraded fragment using an HPLC chromatogram (Fig. 7). The degraded fragment was eluted at a more hydrophobic region compared with A␤-(1-40) (Fig. 7, A and B). Amino acid sequence analysis showed that the first ten residues of the degraded fragment (Fig. 7A, peak b) was SGYEVH-HQKL, which corresponded to A␤- (8 -17). The elution time of the degraded fragment coincided with that of the synthetic A␤-(8 -40) peptide (Fig. 7, A and C).
To confirm that the degraded fragment is A␤-(8 -40), we examined the molecular weight of the synthetic A␤- (8 -40) and the degraded fragment using MALDI-TOF/MS spectroscopy. Our results for A␤- (8 -40) show 3457.78 and for the degraded peptide, 3457.98 (data not shown). These data suggest that ACE cleaved the Asp 7 -Ser 8 linkage of A␤-(1-40) during the incubation of A␤ with ACE.

DISCUSSION
Schachter et al. (22) reported that ACE polymorphism was associated with human longevity and that the ACE DD genotype was surprisingly increased in centenarians. Therefore, Kehoe et al. (8) hypothesized that the D allele might protect against the development of AD and actually confirmed the hypothesis in British populations. We have reported that the ACE genotype is associated with AD in Japanese population even more significantly than that in the British population reported by Kehoe et al. (1). Recently, a gender-specific association of the ACE genotype with AD in the female clinic population was reported (10). It was also reported that ApoE and ACE genotypes might be independent risk factors for late-onset AD in both Russian and North American populations (11). Although several clinical, epidemiological, and pathological observations suggested that vascular risk factors might be associated with cognitive performances of AD, the mechanism by which the ACE genotypes influenced susceptibility to AD was unknown.
The recent studies on the renin-angiotensin system (RAS) of the mammalian brain may explain the association between ACE and AD in a certain sense. Besides the classical RAS, a local RAS in the brain may play a critical role in the central nervous system. It has been reported that angiotensin in astrocytes is required for the functional maintenance of the blood brain barrier (23), which is impaired in AD (24). Central RAS prevents neuronal cells from apoptosis not only by AngII but also by AngIV, an AngII metabolite (25). Both AngII and AngIV excite hippocampal neuronal activity (26) and regulate cerebral blood flow (27). Colocalization of ACE and AngI receptor in the substantia nigra, the caudate nucleus, and putamen of human and rat suggests central RAS may be important in modulating central dopamine release. In Parkinson's disease, there is a marked reduction of ACE receptors associated with the nigrostriatal dopaminergic neuron loss, and ACE inhibitor modifies the clinical features of Parkinson's disease (28). The striking distribution of AngIV receptors in cholinergic neurons, motor, and sensory nuclei of the brain suggest that AngIV plays an important role in the facilitation of learning and memory (29 -31). These studies demonstrate that angiotensin is essential not only to the circulatory system, but also to the central nervous system. Although these results are helpful in understanding the relationship between AD and ACE, they do not provide direct evidence.
AD is a heterogeneous disorder with a variety of molecular pathologies converging predominantly on abnormal amyloid deposition particularly in the brain. A␤ aggregation into senile plaques is an important pathological hallmark of AD. We hypothesize that ACE may affect A␤ aggregation and deposition in the brain. We have substantiated the hypothesis and elucidate here that ACE inhibits A␤ aggregation, deposition, fibril formation, and cytotoxicity in vitro. These results provide the first evidence of direct involvement of ACE with AD susceptibility.
Several lines of evidence have shown that amorphous, largely nonfilamentous deposits of A␤ (so called "diffuse" or preamyloid plaques) precede the development of fibrillar amyloid, dystrophic neurites, neurofibrillary tangles, and other cytopathological changes in Down's syndrome and AD. In the AD brain, diffuse plaques composed mostly of amorphous A␤ are inert, whereas compact plaques composed of A␤ fibrils are associated with neurodegenerative changes (32,33). In vitro experiments also reveal that the neurotoxicity of A␤ is associated with their ability to form stable aggregates in aqueous solution (34 -36). The aggregation of A␤ only is not sufficient to exert neurotoxicity effect, but further amyloid fibril formation is required (37). Aggregation of A␤ is template-independent initial nidus formation, and deposition of A␤ is template-dependent subsequent to plaque growth. These are considered fundamentally distinct biochemical processes in AD (38). Taken together, these findings suggest that aggregation, deposition, and fibril formation are the necessary processes for A␤ to achieve and strengthen a neurotoxic state. Just in these critical processes, ACE plays an important role in decrease of A␤ neurotoxicity, suggesting the possible cause of ACE genotype in affecting susceptibility to AD.
Three types of proteases, which are designated ␣-, ␤-, and ␥-secretases, cleave APP. Processing by ␣-secretase cleaves within the A␤ sequence whereas ␤and ␥-secretase cleaves on the N-and C-terminal ends of the A␤ region, respectively, releasing A␤ (39). ␥-Secretase cleaves at several adjacent sites to yield A␤ species containing 39 -43 amino acid residues. Because ␣-secretase destroys the A␤ sequence, it is generally thought that ␣-secretase pathway mitigates amyloid formation, although this has not yet been demonstrated unequivocally (40). In addition, the C-terminally truncated form of APP released by ␣-secretase may have trophic actions (41), which could antagonize the neurotoxic effects of aggregated A␤ (42). ACE acts like the ␣-secretase in degrading A␤ and thus preventing aggregation.
Several A␤-degrading enzymes were studied because of their potential usage in AD treatment. BACE can cleave full-length APP at Asp 1 of the A␤ sequence and also at Glu 11 (42). Recently, Chesneau et al. (43) reported that insulin-degrading enzyme (IDE) is sufficient to degrade A␤ and that its degradation products do not promote oligomerization of the intact A␤ peptide (43). Iwata et al. (44) reported that endopeptidase 24.11 (neprylisin) is involved selectively in the catabolism of A␤-  in rat brain parenchyma. Yu et al. (45) reported that midkine formed complexes with A␤-(1-40) and protected PC12 h from A␤-induced cytotoxicity.
In the present study, we report that ACE is a new A␤degrading enzyme that cleaves the A␤ sequence at Asp 7 -Ser 8 . The cleavage site is different from the site that converts AngI to AngII, or other A␤-cleaving sites reported as yet. Although the actual meaning of a real A␤-degrading function of ACE in vivo remains to be further studied, our data strongly lead to the hypothesis that ACE may affect susceptibility to AD by degrading A␤ and preventing the accumulation of amyloid plaques in the brains of AD patients.