Amyloid β peptide load is correlated with increased β-secretase activity in sporadic Alzheimer's disease patients

Materials and Methods

Patient Information. AD was diagnosed by National Institute of Neurological Disorders and Stroke-Alzheimer's Disease and Related Disorders Association criteria and confirmed by autopsy. Tissue for BACE protein and enzymatic activity assays were longitudinally assessed and obtained from routine brain autopsies of subjects enrolled in the Sun Health Research Institute Brain Donation Program. A board-certified neuropathologist using Consortium to Establish a Registry for Alzheimer's Disease and National Institutes of Health Reagan criteria performed neuropathology evaluation. Postmortem intervals of autopsied brain samples averaged as 2.8 h (range of 2-3.5 h). Both AD subjects and control subjects were Caucasian and were enlisted through the Cleo Roberts Center for Clinical Research and the Brain Bank at Sun Health Research Institute. We selected temporal cortical and hippocampal samples by National Institute of Neurological Disorders and Stroke-Alzheimer's Disease and Related Disorders Association criteria.

Northern Blot Analysis. Poly(A) RNA was isolated from both AD and ND brain tissues by using the Invitrogen fasttrack 2.0 kit. Five micrograms of each mRNA was fractionated on 1.2% agarose gels containing 0.22 M formaldehyde, transferred to nylon membrane, and immobilized by UV cross-linking. Probes were labeled with [³²P]dCTP by using the Prime-It II kit (Stratagene). cDNA was used to generate a probe that encoded amino acid residues 1-287. Hybridization was performed at 68°C in the ExpressHybTM hybridization solution (Clontech) for 1.5 h, and the blots were washed, according to the manufacturer's instructions. Hybridization of GAPDH was used as a control for equal loading. Sizes of the transcripts were estimated by comparison with Millennium RNA markers (Ambion, Austin, TX).

Ribonuclease Protection Assay. The human BACE1 fragment, a 220-bp transgene 5′ coding sequence, was subcloned into the vector pPCRI (Invitrogen). The antisense cRNA probe was synthesized from the linearized template DNA by using T7 RNA polymerase in the presence of [α-³²P]UTP. The ribonuclease protection assay was performed by using a commercially available RPAII kit (Ambion). Three micrograms of total RNA was hybridized with 0.3 ng of the cRNA probe (specific activity, 5 × 10⁷ cpm per μg of RNA) at 48°C for 16 h, and was then digested with a mixture of RNase A and T1. The protected bands were separated on a 6% denaturing acrylamide gel and were detected by autoradiography. A riboprobe that directed at GAPDH mRNA was included in all incubations as an internal control.

Aβ ELISA. For tissue preparations, we used 70% formic acid to extract soluble and insoluble Aβ from AD and ND brains. Specifically, the brain extracts were incubated at 37°C for 5 min, diluted in buffer (20 mM sodium phosphate/2 mM EDTA/400 mM NaCl/0.2% BSA/0.05% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate/0.4% Block Ace/0.05% NaN3, pH 7.0) before loading onto ELISA plates. Aβ levels were determined by ELISA, which was performed as described (14-17). The final concentrations of Aβ were calculated as pmol of Aβ per gram of protein. The capture antibodies were 4G8 (Aβ residues 17-24) for total Aβ_1-x, AB40 (Biosource), recognizing Aβ residues 33-40 for Aβ_x-40 species, and AB42 (Biosource) recognizing Aβ residues 36-42 for Aβ_x-42 species. The reporter antibodies were biotinylated 6E10 (to Aβ residues 1-16) for total Aβ_1-x, 4G8 for Aβ_x-40, or AβX-42 species.

Western Blot and Immunoprecipitation Analysis. AD and ND brain samples were homogenized individually in five volumes of homogenizing buffer containing 10 mM Hepes, Tris·HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 10% glycerol, 0.1 mM sodium vanadate, plus 0.5% Triton X-100. The tissue membrane sample was collected (at 25,000 × g for 30 min) and dissolved in the homogenizing buffer. For each sample, 100 μg of protein was separated on a 4-12% SDS Tricine gel and transferred to a nitrocellulose membrane for 2 h with TBS containing 0.1% Tween 20. BACE levels were initially determined by immunoblotting with polyclonal antibodies. SECB1 and SECB2 antisera were raised against GST fusion proteins containing 15 or 16 amino acids from the two different N-terminal regions of human BACE. BACE was detected with various antibodies, such as SECB1, SECB2, anti-BACE, and anti-BACE1 (6, 18). APP and C99 were immunoprecipitated with 4G8 or C8 (gift from D. Selkoe, Harvard Medical School, Boston) at a 1:200 or 1:500 dilution for 12 h at 4°C, and the Western blot was probed with 6E10 antibody, followed by incubation with an horseradish peroxidase-conjugated secondary antibody and processed by using ECL detection (Amersham Pharmacia Biotech). The procedures used for immunoprecipitation and immunoblotting have been described (6, 11).

BACE ELISAs. The temporal cortex or hippocampal homogenates were added to polyclonal anti-BACE1 N-terminal antibody (SECB1)-coated plates at a 1:1,000 dilution, and were incubated at 25°C for 2 h to permit BACE capture. To bind captured BACE, 100 μl of a 0.3 μg/ml anti-BACE C terminus antibody was added. Biotinylated anti-rabbit IgG, avidin-conjugated horseradish peroxidase, and o-phenylenediamine dihydrochloride were used to detect binding. Specificity of the ELISA protocols was verified by serial dilutions of purified BACE standards or BACE-transfected cell lysate. To verify the BACE protein measurement, SECB1 was also used as a capture antibody, and the antibody for BACE C terminus (1:1000) was used as a detection antibody. Similar results were obtained.

BACE Enzymatic Activity Assays. Activity assays for BACE were performed by using synthetic peptide substrates containing either the APPwt BACE site (MCA-Glu-Val-Lys-Met-Asp-Ala-Glu-Phe-(Lys-DNP)-OH) or APPsw, where Lys-Met are substituted by NL, and in the MV mutant, and M is substituted by V. Substrates were used at 50 mM and reactions were performed in 50 mM [2-(N-morpholino)ethanesulfonic acid] with 50 mM acetic acid (pH5.5). Enzymatic crude extracts and fluorescent labeled peptides were incubated for various times at 37°C. The reaction mixtures were quenched and absorbed at 383 nm by a fluorescent plate reader.

Searching for BACE Mutations. Sequencing. The entire ORF of the BACE gene was recovered in two overlapping RT-PCR products. cDNAs were generated from total RNA of brain tissue (TRIzol) by using the firs-strand cDNA synthesis kit (GIBCO). As Nicolaou et al. (18) reported, we used forward primer 1 (5′-CCACCAGCACCACCAGACTT-3′) and reverse primer 1 (5′-CTCCTTCAGTCCATTTTCAGAT-3′) to amplify the first 93-bp fragment that included nucleotides 366-1271 of BACE and used the second set forward primer 2 (5′-CACCTTGCCAGCCTTTTCCTT-3′) and backward primer 2 (5′-ATCCGGCGGGAGTGGTATTAT-3′) to generate a 913-bp fragment included nucleotides 1213-2105 of BACE. The fragment generated by first set of PCR primers was sequenced with sequencing oligonucleotide primer 1 (5′-CCCCGCAGACGCTCAACAT-3′), sequencing oligonucleotide primer 2 (5′-GCCAGTGCGAGCCCAGAG-3′), and sequencing oligonucleotide primer 3 (5′-CCCAGCTCCCCTTCCCACTT-3′). Consistent with the reported result (19), the fragment generated by the second PCR primers was sequenced with sequencing oligonucleotide primer 4 (5′-ACCGCTGCCGTCCTGAACTC-3′) and sequencing oligonucleotide primer 5 (5′-AACCAGTCCTTCCGCATCACC-3′). Mutations were sought by direct inspection of the fluorescent chromatographs and by using the Factura sequence navigation software (version 2.0).

Linkage analysis. Linkage analysis was performed by using nonparametric and parametric approaches implemented in the computer software package genehunter. A parametric analysis was carried out by assigning a constant low penetrance (0.02) to all unaffected at-risk individuals. The genotype of a silent BACE polymorphism in exon 5 (at nucleotide 1239) was determined by using a PCR restriction-site method. The 528-bp PCR product was amplified from genomic DNA (GenBank accession no. AC020997) with primers 2192 (5′-AAGCAGGAAGATGAAAAGGG-3′) and 2193 (5′-CTACGAAGCAAGGCAGTGAC-3′). The PCR products were digested with 0.3 units of HphI for 4 h at 37°C, and the resulting restriction fragments (C allele, 528 base pairs; G allele, 303 base pairs, and 225 base pairs) were resolved on a 1.5% agarose gel.

Results

Aβ Levels in Sporadic AD Brains. Our recent report (11) has examined the levels of Aβ in brain regions that were used for measurement of BACE protein levels and activities. Although no direct linkage between frontal Aβ levels and BACE activities was identified, there was no separation of which Aβ species were measured, and the sum of Aβ_x-40 and Aβ_x-42 was correlated to BACE activities (20). Because the deposition of Aβ₄₂ is believed to play important roles in plaque formation in AD pathogenesis, we aimed to determine whether there is a specific relationship between BACE protein levels/activities and different Aβ fragments deposited in AD brains; i.e., Aβ_x-40, Aβ_x-42, Aβ_1-40, Aβ_1-42, and Aβ_1-x. To do so, we first developed Aβ ELISAs.

To measure all of the Aβ in AD brains, including soluble and insoluble Aβ with several species, we homogenized the temporal cortex in both AD and ND in 70% formic acid. The assay for Aβ_x-40 and Aβ_x-42, we used the captured antibody 4G8 (to Aβ17-24) and detector antibodies biotinylated AB40 and AB42. The assay for Aβ_1-x, Aβ_1-40, and Aβ_1-42 used the captured antibodies 4G8, AB40, and AB42 and biotinylated detector antibody 6E10. Synthetic Aβ_1-40 and Aβ_1-42 peptides were used as positive controls.

We have found that very little Aβ was detected in ND brains that contain no plaques (Fig. 1A). The 4G8/6E10 ELISA for Aβ_1-x showed a total 2,800 ± 1,200 pmol/g, which is significantly higher than that of ND samples (Fig. 1A). There were very low levels of Aβ_1-x in ND brains without plaques, <27.8 pmol/g of protein. However, the level of Aβ_1-x in AD brains was in a range from 1,500 to 5,000 pmol/g of protein. As 6E10 antibody (to Aβ_1-16) cannot capture Aβ species beginning at or beyond 17, we also measured Aβ_x-40 and Aβ_x-42. We have found that most Aβ in the formic acid extracts from the 10 samples ended at Aβ_42, specifically, the 4G8/AB42 ELISA, which is for Aβ_x-42 assay showed 4,500 ± 1,300 pmol of Aβ per g of protein (Fig. 1B).

We have found that most Aβ detected in formic acid extracted brain tissues were Aβ_x-42 and Aβ_1-x, whereas Aβ_x-40 and Aβ_1-40 only occupied a low percentage of total extracted Aβ. Not surprisingly, most AD cases showed much higher Aβ levels, compared with ND cases. Consistent with previous reports (21, 22), we detected much higher levels of Aβ₄₂ than Aβ₄₀ in these human brains, and we found that most of Aβ species analyzed were Aβ_x-42 (89%). Specifically, we directly measured Aβ in eight AD brains and found that all Aβ in the formic acid extracts from these AD brains, and subsequently more Aβ_1-42 was assayed (four times more) by AB42/6E10 ELISA than Aβ_x-40 assayed by AB40/6E10 ELISA (Fig. 1C).

Compared with little amounts (< 9 pmol/g) of Aβ_1-42 measured in ND brains, as the Fig. 1 demonstrated, Aβ_1-42 in formic acid extracts from AD brains ranged from 2,300 to 3,800 pmol/g. Meanwhile, the level of Aβ_1-40 in AD brains is from 230 to 500 pmol/g of protein, whereas very little Aβ_1-40 in ND brains was detected.

Interestingly, when we calculated the difference between Aβ1-X and the sum of Aβ1-40 and Aβ1-42, we noticed a large quantity of C-terminal-truncated Aβ species that were not detected by Aβ₄₀ and Aβ₄₂ specifically. Because BACE is a key factor in Aβ generation and because the BACE protein levels and activities are now available from the same cases, we have examined a linkage between BACE and Aβ levels.

Characterization of BACE Antibodies. To study a specific BACE expression pattern in the brain, we have generated two antibodies. The two BACE antisera (SECB1 and SECB2) were raised against GST fusion proteins containing 15 or 16 amino acids from the two different N-terminal regions of human BACE (GST-SECB1 and GST-SECB2). The resulting antibodies were extensively characterized by Western blot analysis, using overexpression systems or preabsortion before being used to detect the endogenous BACE expressed in the AD brain (Fig. 7A, which is published as supporting information on the PNAS web site). In extracts from the temporal cortex (Fig. 7A), SECB1 strongly reacted with a 70-kDa protein (Fig. 7A), which is similar in mobility to mature BACE expressed in transfected HEK293 cells that has been previously detected by other antibodies (6, 8, 11). Less protein loading (i.e., 10 μg per lane) showed the same pattern of bands (data not shown). To further demonstrate the specificity of SECB1 for BACE and to additionally characterize the bands observed in Fig. 7, the same Western blot membrane was stripped and was reprobed with an affinity-purified polyclonal antibody, SECB2, raised against a different region of the N-terminal region of BACE (Lower). The SECB2 detected the ≈70-kDa band in HEK293 cells specifically, as well as a 30-kDa band on the same blot (Lower, lane B), confirming that both antibodies recognized full-length BACE. Moreover, blocking peptides and preabsorption in this Western blot abolished BACE-immunoreactive bands from transfected cells or purified BACE protein (Fig. 7A, lanes D-F and G, respectively).

Elevation of BACE Protein Expression in AD Brains. Several studies demonstrated that higher mRNA levels of BACE are observed in the brain, compared with peripheral tissues (6-10). The availability of the BACE-specific antibodies allowed us to measure BACE protein expression levels by using Western blots. Temporal cortical and hippocampal tissue samples obtained at autopsy from (<3 h) AD and ND subjects were immediately frozen and stored in vacuum-sealed plastic bags at -80°C until assay. Both AD and ND brain tissue were homogenized with the average yield of 40 mg of total protein per gram of wet tissue. The yield of total protein did not correlate with age at the time of death, or with the postmortem interval for ND or AD patients (data not shown). Western blot analysis in these samples indicated a higher level of BACE protein expression in the hippocampus and the temporal cortex of AD than that in the same brain regions of ND. The statistical significance (P = 0.02, P < 0.05, ANOVA, Student's t test) was reached when the relative large numbers of cases were extended (n = 16) in each group (Fig. 7 B and C). Similar significant differences (Fig. 7D) were found in Western blots detected with additional antibodies as published (6-11).

To further confirm this finding, a sensitive BACE ELISA was used to test more samples. We have randomly selected 40 cases, consisting of 20 AD and 20 ND cases, including those previously used for Western blots, and these cases contained no mutations in PS1, PS2, or APP genes. These brain homogenates were assayed in triplicate for each experiment and were repeated three times. Twenty AD and 20 ND temporal cortex samples and eight AD and eight ND hippocampus samples were quantitated by the ELISA to measure BACE. The specificity of the ELISA protocol was verified by testing either BACE-transfected cells or purified BACE protein in the same ELISA (Fig. 2). We found that the level of BACE protein was significantly higher (28%) in the temporal cortex and even higher in the hippocampus of AD patients (32%), compared with that of ND by ELISA, but no change was observed in the cerebellum (Fig. 2).

BACE mRNA Expression in AD Brains. First, we performed a Northern blot analysis by using BACE cDNA as a probe to determine the mRNA expression of BACE (Fig. 3A). The BACE cDNA probe hybridizes to all BACE transcripts, including 2.1 and 3 kb. Hybridization of AD and ND brain tissue Northern blots revealed a BACE transcript of 2.5 kb in all tissues examined, which was similar to the expression pattern as reported (6, 7). Likewise, BACE was expressed in both AD and ND brains and two mRNA species of 2.1 and 2.3 kb were found. Control hybridizations with a probe specific for GAPDH revealed that equal amounts of mRNA were loaded. Among the 12 cases of brain samples (six AD and six ND) we analyzed, we observed that BACE mRNA expression in four of six AD cases was obviously higher than that in ND cases. These results indicate that BACE mRNA levels in AD brains may be increased. To further verify BACE mRNA expression levels in AD brains, we have used RNase protection assay and examined two cases that did not show obvious change. A ribonuclease protection assay was performed on brain tissues from AD and ND. A riboprobe that directs to the GAPDH mRNA was included in all incubations as an internal control. As expected, total RNA extracted from brain tissues to express the BACE1 gene protected a single band of 220 nucleotides. The level of BACE1 mRNA in the AD brain was higher than that in ND (Fig. 3B). This finding is consistent with the BACE1 protein level we described above.

Elevated BACE Enzymatic Activity in AD Brains. Because a high BACE protein level was found in AD brains, this finding is reflected in the BACE enzymatic activity in AD brains. We used crude AD brain extracts as a BACE enzymatic substrate and tested whether BACE would cleave the fluorescent-labeled peptide bearing with either the APPwt or APPsw sequence. We found that the fluorescent intensity was significantly increased by 23% (P < 0.05) in the crude extracts from the AD temporal cortex, compared with that of ND (Fig. 4). In a parallel test model for a positive control, as previously reported, we transiently transfected BACE cDNA or vector alone into HEK293 cells stably expressing APPwt. Fig. 4 demonstrates a high-fluorescent intensity in BACE-transfected cells, suggesting that BACE-overexpressing cells have higher BACE activity. To confirm this finding further, we designed APPsw, and incubated the labeled fragment with AD and ND temporal cortex crude extracts. We found that many more APPsw fragments were cleaved by both AD and ND crude extracts than were the APPwt fragments (Fig. 4A). Moreover, a mixture of APPsw and AD crude extracts generated a higher fluorescent fragment product (143%, P < 0.01) than that of the mixture of APPsw and ND extracts (Fig. 4B). Our results suggest a higher BACE enzymatic activity in AD brains. Last, to rule out the possible involvement of caspases in AD brains (23), we incubated enzymatic crude extracts with a caspase inhibitor, Z-VAD (Ome)-CH₂F, and this general caspase inhibitor did not perturb BACE activity (data not shown).

BACE is a membrane-bound aspartyl protease protein that can cleave APP into sAPP and C99 from the C terminus of APP. As both protein level and activity of BACE are increased in AD, thus the enzymatic product, C99, is presumably increased. To visualize the BACE-cleavage product, C99, we performed immunoprecipitation of C99 by using C8, an antibody recognizing the C terminus of APP. As demonstrated in Fig. 8, which is published as supporting information on the PNAS web site, we found more C99 fragments were immunoprecipitated from AD brains than that from ND brains (n = 6). The increased level of C99 in AD brains reached a significant difference when compared with the ND group, whereas no significant changes of APP levels were observed between AD and ND brains (Fig. 8). Our data suggest that BACE activity is higher in the AD brain than that in ND.

APPsw Does Not Lead to BACE Expression Change in Vitro. To examine whether elevation of BACE in the AD brain is a causative factor or a secondary effect, we have used APPsw or mutated PS1-transfected cells to analyze BACE protein and mRNA expression. First, we performed a Northern blot analysis by using BACE cDNA as a probe to determine the mRNA expression of BACE (Fig. 5). The BACE cDNA probe hybridizes to the BACE1 transcript. A 2.3-kb transcript was detected with steady levels, compared with empty vector or APPwt-transfected cells. Control hybridizations with a probe specific for GAPDH revealed that equal amounts of mRNA were loaded (Fig. 5). We then examined BACE protein expression from these cell lysates. These results indicate APPsw did not result in alteration of BACE expression in vitro.

Correlation of BACE Levels and Plaques and Aβ Loads in AD Brains.Plaques containing Aβ are one of the established pathological hallmarks in the AD brain, and it is reported that there is a correlation between Aβ plaques and the degree of dementia (24-26). Because BACE is one of the major enzymes in Aβ production and a high level of BACE was found in AD brains (Fig. 6A), it is logical to examine whether the elevated BACE level is correlated with plaque numbers in AD. By using linear regression analysis, we observed a significant correlation between levels of BACE and plaque numbers in AD brains. A similar correlation was also found between BACE levels and disease history of these AD patients (Fig. 6 A and B). In general, the onset for nonfamiliar AD usually progresses slowly. The data may suggest that a reasonable level of increased BACE expression in the brain, due to whatever reason, may possibly be one of the causal factors for AD.

No Gene Mutations Within BACE Coding Region Found. The pathological APPsw double-mutation KM670/671NL, located near the BACE-cleavage site, causes AD, due to an increased production of Aβ by affecting the BACE-cleavage process (27). Thus, we hypothesize that any mutations in BACE having a similar effect on Aβ processing might explain some cases of autosomal dominant AD. Alternatively, although lack of association of BACE with late-onset AD was demonstrated (28), variations in BACE might contribute to the susceptibility of early-onset AD. Therefore, we performed molecular genetic analyses of BACE in selected AD cases that show high levels of BACE activity and protein levels.

Discussion

Our results in this study are consistent with recent reports (11-13) that BACE protein level is increased in a large number samples of AD by using various assay formats. Our present study has extended our recent study that Aβ production is correlated with BACE expression and activity in AD brains. Moreover, we have also found that BACE mRNA expression is increased in rapidly autopsied brains with AD. Our rapidly autopsied brain samples with <3.5 h postmortem time provide a very reasonable and reliable source for studying of APP processing, including BACE mRNA and protein expression and BACE enzymatic activity in AD brains in vivo. For example, our laboratory can isolate living microglia, astrocytes, and even neurons from those rapidly autopsied brains (29-31). However, if postmortem intervals are longer than 6-7 h, very few living cells can be isolated (data not shown), and RNA and proteins are partially degraded. Thus, using our rapid autopsy approach in this study provides a good sample source and a system in which we search for pathological alteration reflective of living AD brains.

Many factors could contribute to an increased BACE level in brain. For example, recent studies demonstrate that studies on mRNA levels in brains from AD vs. ND do not unequivocally prove a uniform increase of the messengers (data not shown). Whether the stability of mRNA could be the cause for enhanced BACE activity remains to be explored. To examine whether genetic factors play role in BACE function, we have examined DNA sequences from the BACE-coding region from >40 AD patients and we found no mutations in these AD patients (data not shown), which is consistent with recent reports (11-13). We have also examined BACE levels in endogenous BACE expression in either APP mutant or PS1 mutant-transfected cells or cotransfected with BACE cells, and we found little changes in BACE expression (data not shown). Furthermore, studies on transgenic animals recently have demonstrated that there was no BACE change in APP-transgenic mice (32). This study suggests that BACE might not be a secondary effect although direct evidence should be ideally obtained from AD patients with APP or PS mutations.

Supplementary Material