Correlation Between Elevated Levels of Amyloid β-Peptide in the Brain and Cognitive Decline FREE

Deposition of the amyloid β-peptide (Aβ) into plaques in the brain parenchyma and cerebral blood vessel walls is one of the distinguishing neuropathological features of Alzheimer disease (AD).¹ The Aβ peptide is generated from the Alzheimer amyloid precursor protein (APP) by 2 consecutive cleavage events: one proteolytic activity, β-secretase, generates the N-terminus of Aβ while another, γ-secretase, generates the C-terminus.²

There is some heterogeneity in the primary structure of Aβ peptides found in media from cultured cells and in cerebrospinal fluid. However, most Aβ peptides appear to terminate at 1 of 2 residues: Val40 or Ala42. The longer species, Aβx-42, is more prone to aggregation than the shorter Aβx-40,³ and immunohistochemical studies suggest that Aβx-42 is the initially deposited amyloid peptide in plaques in AD and Down syndrome.^4- 5 Moreover, mutations in the Alzheimer APP and presenilin genes, associated with rare, early-onset familial forms of AD, result in a relative increase in levels of Aβx-42.⁶ Collectively, these findings suggest a role for Aβx-42 in initiating the events that ultimately lead to accretion of amyloid in AD.

The accumulation of neurofibrillary tangles (NFTs) in neurons is a second distinguishing feature of AD. Neurofibrillary tangles are composed of tau, a protein involved in microtubule formation.⁷ By the use of conformation- and phosphorylation-sensitive antibodies, it has been demonstrated that the tau in NFTs is abnormally folded and phosphorylated. Changes in the conformation and phosphorylation of tau are early events in neurofibrillary lesions and result in a loss of the microtubule-binding properties of tau.⁷

The connection between cognitive impairment and observed neuropathology associated with AD is a key issue in defining the molecular mechanisms responsible for functional loss. The role of different pathogenic molecules, such as Aβ and tau, in the disease process has been the subject of controversy,⁸ possibly stemming from the fact that it is difficult to compare studies using different immunohistochemical protocols, brain regions, subject inclusion criteria, and methods of analysis. Additionally, the typical brain available for postmortem examination rarely derives from patients with early disease, yet only through the examination of such cases is it possible to elucidate the very earliest pathological events. The study of brains with early- and middle-stage disease also facilitates correlative analyses seeking relationships between pathological markers and severity of illness. Such analyses are markedly impaired by the use of brains predominantly from patients with end-stage disease.

We investigated the levels of total Aβx-40 and Aβx-42 in brain tissue from a cohort of 79 subjects who were either nondemented or had questionable, mild, moderate, or severe dementia, as assessed by the Clinical Dementia Rating (CDR) scale. To this end, we devised 2 highly sensitive enzyme-linked immunosorbent assays (ELISAs) specific for Aβ peptides terminating at Val40 and Ala42. These ELISAs were then used to quantify total Aβx-40 and Aβx-42 extracted from 5 distinct brain regions, allowing the comparison of Aβ levels across CDR scores.

The evaluation and selection of the subject cohort have been described in detail elsewhere.⁹ Briefly, the sample cohort was selected from a group of 278 consecutively autopsied subjects (between 1986 and 1997) who had been residents of the Jewish Home and Hospital (JHH) in Manhattan and the Bronx, NY. All subjects had a history of independent functioning in the community prior to entry into the JHH. Admission to the JHH was due to a need for ongoing nursing care resulting from cognitive impairment and/or, in the case of nondemented subjects, a physical condition such as incomplete recovery following a fracture. A multistep approach was applied to the assignment of CDR scores^10- 11 based on cognitive and functional status during the last 6 months of life. Scores were determined at a consensus meeting of a group of senior clinicians and were obtained independently of any postmortem data. The antemortem clinical information that contributed to the CDR score included information from caregivers, family members, and the clinical medical record, including an annual mental status examination. The assignment of CDR scores to these cases has been described previously.^9,12- 13 These subjects are part of a prospective study of AD being performed in the JHH and the Mount Sinai Alzheimer Disease Research Center in New York. Five groups of subjects were formed, consisting of cases falling into CDR score categories of 0.0, 0.5, 1.0, 2.0, and 4.0 or 5.0 (the latter 2 were grouped as 5.0). The demographic data for the final selection of subjects are shown in Table 1. As described previously,⁹ all patients with neuropathological lesions other than those of AD were excluded (eg, lesions of Pick disease, diffuse Lewy body disease, Parkinson disease, stroke, multi-infarct dementia, and severe cerebrovascular disease).

At autopsy, the left hemisphere was sectioned into 0.5- to 0.8-mm coronal slabs, flash frozen in liquid nitrogen–cooled isopentane, and stored at –80°C until used. Prior to assay, specific tissue slabs were warmed to –20°C and the regions of interest were dissected. The regions studied here corresponded to the middle frontal gyrus (Brodmann area [BA] 8), superior temporal gyrus (BA 22), entorhinal cortex (BA 36 and 28, designated BA 36), inferior parietal lobule (BA 7), and primary visual cortex (BA 17). For each region, the entire gyrus represented within a coronal slab (approximately 1-2 g) was dissected taking care not to include more than 1 mm of white matter within each block. The dissected material was crushed and homogenized to a powder using a liquid nitrogen–cooled mortar and pestle according to procedures described previously.¹⁴

Frozen powdered tissue homogenates (20-25 mg) were lightly thawed and homogenized by sonication in 300 µL of phosphate-buffered saline solution (PBS) containing 0.05% sodium dodecyl sulfate and a protease inhibitor cocktail (PIC Complete, Boehringer-Mannheim, Indianapolis, Ind; supplemented with 1 µmol/L of pepstatin). The homogenates were centrifuged at 100,000g for 60 minutes at 4°C. The pellets were delipidated with a slightly modified chloroform-methanol protocol¹⁵ before extraction with 150 µL of 70% formic acid and 100 mmol/L of betaine (FA, sonication followed by mixing via vortex for 30 minutes). The extracts were then subjected to centrifugation at 100,000g for 60 minutes at 4°C. The clear Aβ-containing FA supernatants were aspirated and saved for analyses. In the initial exploratory analyses we quantified the amounts of Aβ in the supernatant from the PBS extract, in the supernatant from the FA extract, and in 1,1,1,3,3,3-hexafluoro-2-propanol (HFP) extracts of the FA pellet. These studies revealed that more than 90% of all Aβ species were found in the supernatant of the FA extract and that the relative proportion of each Aβ species was not significantly different in any extract. We also noted that inclusion of the chloroform-methanol delipidation step in the extraction procedure resulted in substantially improved Aβx-40 and Aβx-42 signals. We therefore chose to analyze the delipidated primary FA extract.

Aliquots (50 µL) of each FA extract were neutralized in 1 mol/L of Tris hydrochloride and 100 mmol/L of betaine at pH 10.3 (850 µL), and further diluted 1:50 in ELISA capture and wash (ECW) buffer (PBS, 5 mmol/L of EDTA, 2 mmol/L of betaine, 0.1% bovine serum albumin, 0.05% Tween 20, 0.2% CHAPS [3-{(3-cholamidopropyl)dimethylammonio}-1-propanesulfonic acid], and 0.05% NaN₃[sodium azide])–containing protease inhibitor cocktail. Particulate matter was removed by centrifugation at 2500g for 1 minute before ELISA analysis.

Peptides corresponding to the 6 C-terminal residues of Aβx-40 and Aβx-42, and containing N-terminal cysteine residues, were synthesized (Research Genetics, Huntsville, Ala), coupled to keyhole limpet hemocyanin, and injected into rabbits (Cocalico Biologicals Inc, Reamstown, Pa). Blood obtained from the rabbits was assayed for titer and cross-reactivity by spot blotting and, more extensively, by Western blotting and peptide ELISAs. The antisera selected from this primary screening were affinity-purified on Sulfo-Link resins (Pierce Chemical Company, Rockford, Ill) derivatized with either the Aβx-40 or the Aβx-42 octomers containing cysteine. After another round of cross-reactivity and affinity testing using the above methods, 1 antibody for each peptide (denoted αAβC40 and αAβC42, respectively) was chosen and used throughout the study (Figure 1). To ensure that these antibodies did not cross-react with full-length APP, they were used for immunoblot analysis of lysates from APP-transfected Chinese hamster ovary cells. No cross-reactivity with full-length APP reactivity was observed with either αAβC40 or αAβC42 (data not shown).

Microtiter plates (Falcon Probind, Becton Dickinson, Franklin Lakes, NJ) were coated with 2 µg/mL of monoclonal antibody 4G8 (Senetek, Napa, Calif) in 50 mmol/L of carbonate-bicarbonate buffer, pH 9.6 (Sigma-Aldrich, St Louis, Mo), at 37°C for 18 hours. The antibody 4G8 recognizes an epitope between residues 17 and 20 of Aβ, enabling detection not only of Aβ beginning at residue 1, but also of different N-terminally truncated Aβ species that are found abundantly in the brain.¹⁵ After washing the plates with ECW buffer, unoccupied binding sites were blocked by incubating the plates with 1% casein in ECW at 37°C for 4 hours. Synthetic Aβ1-40 and Aβ1-42 (US Peptides, Fullerton, Calif), used as standard peptides in the ELISAs, were stored in aliquots at –80°C (100 µg/mL in HFP). The amount of aliquoted peptide was corroborated by amino acid analysis.

Samples and standards were applied in quadruplicate and incubated for 48 hours at 4°C. After the capture phase, the plates were washed twice with ECW buffer and incubated with the appropriate Aβ C-terminal–specific antibody (2-5 µg/mL in ECW buffer) for 18 hours at 22°C. The plates were washed, reporter antibody was added (alkaline phosphatase–conjugated antirabbit IgG, γ-chain specific; Southern Biotechnologies Associates, Birmingham, Ala) at 1 µg/mL in ECW buffer, and the plates were incubated for 2 hours at 22°C. The plates were developed using an alkaline phosphatase substrate yielding a fluorescent product (AttoPhos, Roche Pharmaceuticals, Nutley, NJ) and analyzed with a 96-well fluorescence reader (CytoFluor, Millipore, Bedford, Mass).¹⁶ The detection limit for synthetic Aβx-40 and Aβx-42 was estimated to be 10 pg/mL (approximately 2 pmol/L). All samples were analyzed in the linear range of the ELISA.

Staining of tissue sections using the conformation-specific anti-tau antibody MC1 was performed essentially as described previously.¹⁷ Sections (10 µm) were cut from paraffin-embedded tissue blocks, deparaffinized with xylene, and rehydrated using 100%, 95%, and 80% ethanol. Endogenous peroxidase activity was blocked by incubation in 3% hydrogen peroxide in methanol for 15 minutes. Nonspecific antibody binding was prevented by inclusion of 5% nonfat dried milk in both primary and secondary antibodies. MC1 was used at a dilution of 1:25, and diaminobenzidine (0.3 mg/mL in 0.01% hydrogen peroxide) was used for color development. Sections were counterstained with toluidine blue O.

Aβ1-40, Aβ1-42, and Aβ1-43 peptides (50 ng each) were spotted directly onto polyvinylidene difluoride membranes or electrophoresed (500 ng each) on a 10% to 20% Tris-tricine gel and transferred to nitrocellulose. Membranes were probed with affinity-purified antibodies (1 µg/mL) in PBS containing 0.05% Tween 20 and 5% nonfat dried milk. The blots were developed using horseradish peroxidase-conjugated antirabbit antibodies and the SuperSignal detection system (Pierce).

The Aβ results were analyzed by a mixed design analysis of variance with CDR score as the between-group fixed factor and brain region as the within-group variable. Interaction terms between groups and Aβ levels in different brain areas were examined and were found to be significant in the analysis of Aβx-42. However, our specific hypotheses related more to group effects than to specific group-by-region interactions; therefore, discussion of these latter interactions was omitted. No interaction effects were observed that would alter the conclusions drawn from examination of the main effect of CDR scores. Comparisons between factors and groups for the effects of CDR scores and brain regions were analyzed using the Tukey Honestly Significant Difference test.¹⁸ Correlations were calculated using Pearson product moment correlations. For correlations between CDR scores and Aβ levels in which one variable (CDR) is a scored (ordinal) value, we also carried out Spearman rank correlations. Differences were considered significant at a P value of <.05, after correction for multiple testing using the Bonferroni method. Because of the heterogeneity of variance between groups for the Aβ measurements, the validity of the above-mentioned analyses was assessed by transforming the Aβ values to rank order. Analyses of variance performed on these rank-order transformed Aβ values yielded identical results with respect to significant effects.

Brain tissue Aβx-40 and Aβx-42 were analyzed from a cohort of subjects classified on the basis of CDR scores as either cognitively intact (CDR, 0.0 [n = 16]), questionably demented (CDR, 0.5 [n = 11]), mildly demented (CDR, 1.0 [n = 22]), or moderately impaired (CDR, 2.0 [n = 15]). As a reference group for severe dementia, an additional 15 subjects with a CDR score of 4 or 5 (grouped as CDR 5.0 [n = 15]) were included. Subjects with a CDR score of 4 or 5 represent the extreme of the continuum of dementia and are similar to the cases of Alzheimer disease most often reported in autopsy series. The primary focus of the current study was early dementia (CDR score 0-2), so data were analyzed first without and then with the advanced (CDR score 4-5) cases.

Basic demographic data for the study subjects are presented in Table 1. The age and intervals from death to autopsy (postmortem intervals) did not differ significantly (F_4,76 = 1.4, P>.25 and F_4,76 = 0.63, P>.64, respectively) between the groups. Although there were more women than men in the study cohort as a whole, the proportion of men and women in different CDR groups did not differ significantly (χ²₄ = 0.80, P = .82).

The results from the ELISA screening are shown in Table 2. The average levels of Aβx-42 were higher than those of Aβx-40 in the subjects without dementia. This finding is consistent with previous studies showing a preponderance of Aβx-42–immunoreactive diffuse plaques in subjects without dementia⁴ and with previous studies of total Aβ in the brain.¹⁹ Levels of both Aβx-40 and Aβx-42 were higher in the frontal cortex (BA 8) than in the other regions in the CDR 0.0 group. The average levels of Aβx-40 and Aβx-42 were higher in the CDR 0.5 group compared with the CDR 0.0 group in the frontal, entorhinal, parietal, and visual cortices (Table 2). Since both peptides were elevated in a similar fashion, these results do not support a specific role for Aβx-42 early in the disease process.

The average levels of both Aβx-40 and Aβx-42 were generally higher with increasing severity of disease. The levels of Aβx-42 remained higher than Aβx-40 throughout the progression of disease in all 5 cortical regions examined. Statistical analysis revealed that both Aβx-40 and Aβx-42 correlated with the degree of dementia when including either cases with CDR scores of 0 to 2 or 0 to 5 (Table 3). There was evidence for a strong correlation between Aβ levels and CDR score in all cortical regions except the primary visual cortex (BA 17), suggesting a global accumulation of Aβ during the course of disease. The levels of Aβx-40 and Aβx-42 also were correlated highly and significantly with each other (r range for the 5 regions, 0.42-0.87; P<.05). Moreover, comparing the levels of Aβx-40 and Aβx-42 with the density of neuritic plaques in all of the cerebrocortical regions studied⁹ revealed significant correlation, especially when all CDR categories were included in the analysis (Table 4). Finally, study subjects were also categorized according to established neuropathological and clinical criteria for definite AD,²⁰ and levels of both peptides were robustly increased in definite AD cases compared with elderly controls without dementia (Table 5).

We also investigated AD-associated tau protein pathology in the frontal cortex using antibody MC1, a monoclonal antibody recognizing a conformation-dependent epitope in tau.²¹ MC1 reactivity is a very early marker for AD-related changes in tau and can precede abnormal phosphorylation of tau. We focused on the frontal cortex because this region is involved only late in disease progression and, unlike the transentorhinal and entorhinal cortices, neurofibrillary lesions are not typically found in the frontal cortex of aged, cognitively normal individuals.⁷ MC1 was used to categorize the subject cohort, depending on the tau-stained structures, into 3 groups: those with no MC1 staining, those with staining of neurites only, and those with staining of neurites, neuritic plaques, and NFTs (Figure 2). The categorization was done in a blinded fashion. As seen in Table 6, levels of both Aβx-40 and Aβx-42 were significantly increased as a function of MC1 staining (F_2,59 >8.5, P<.001). The levels of Aβx-42 were significantly (P = .03) higher in the group with only neuritic MC1 staining than in the group negative for MC1 staining, and were highest (P<.001) in the group with fully developed tau pathology (Figure 2 and Table 6). The mean level of Aβx-40 was higher in the group with neuritic MC1 staining; however, the increase reached statistical significance only in the group with fully developed tau pathology (P<.001). It appears then, at least in the frontal cortex, that increases in both Aβ peptides, and in particular Aβx-42, precede the formation of the other main lesion in AD, the NFT, and these increases occur before clinical AD criteria are met.

We present data on total brain Aβ peptides with intact or truncated amino termini and ending at either amino acid 40 or 42 from a cohort of subjects representing a continuum of AD-related cognitive decline. The study subjects fulfilled a number of important parameters: short postmortem intervals, high degree of age matching, and reasonably short periods between cognitive assessment and death. In addition, controls were derived from exactly the same population as the cases.

Levels of both Aβx-40 and Aβx-42 were found to increase very early in the disease process (Table 2), and in the frontal cortex this increase occurred in the absence of significant neurofibrillary pathology (Table 5). Although causality of disease is not proven in a correlative study of the type presented here, the results argue in favor of a central role for Aβ early in the disease process. More than half (55%) of the study subjects without detectable plaques (as assessed by immunostaining) had readily detectable levels of Aβx-42 and 40% of the study subjects without detectable plaques had readily detectable levels of Aβx-40. These findings indicate that the Aβ peptides measured not only reflect classic amyloid deposits, but also soluble Aβ and perhaps Aβ aggregates that are not resolved at the level of light microscopy. Ongoing studies are aimed at elucidating the conformational states and tissue localization of Aβx-40 and Aβx-42 in "preamyloid" cases.

Neurofibrillary lesions can develop in the absence of Aβ deposition into amyloid plaques.⁷ This observation is inconsistent with NFTs developing as a consequence of a neurotoxic effect of extracellular amyloid deposits. However, given the presence of soluble Aβ even in the absence of amyloid plaques, it is possible that the NFTs develop as a consequence of soluble Aβ, a possibility supported by the evidence that, at least in the frontal cortex, Aβ levels increase before AD-associated neuritic changes are detectable. Our studies of early neuritic changes and total Aβ were confined to the frontal cortex. Using the frontal cortex, where the presence of neuritic pathology is diagnostic for AD (in contrast to the hippocampus and entorhinal cortex where NFTs were found in all subjects in this cohort¹²), allowed us to stage cases based on the presence of AD-associated neuritic pathology. We cannot say whether the progression we observed in the frontal cortex is similar to any other brain areas. Finally, NFTs occur in a variety of diseases without Aβ deposition. Thus, alterations in Aβ are not necessarily the only cause of tau pathology.

Aβx-40 and Aβx-42 correlate with the degree of dementia, the levels of both peptides increasing systematically with severity of cognitive decline (Table 3). This result might be considered to contradict some earlier results comparing cognitive impairment and neuritic plaques.^22- 23 However, more recent studies support a correlation between neuritic plaques and cognitive impairment.^9,24- 26 In addition, by using rigorous extraction techniques and quantitative ELISA, we were able to measure total Aβ, not just Aβ in plaques. Studies using immunocytochemistry are not likely to detect soluble Aβ or Aβ in small preamyloid aggregates as we did. In addition, we focused on a rigidly characterized and relatively large sample cohort, in which the majority of subjects (n = 49) were either not demented or questionably or mildly demented. This approach affords the greatest insight into early neurochemical changes in dementia. Although confirmatory studies are needed on subjects of different ethnic origins and age ranges, using either the same or different methods, our results support the fundamental role of Aβ in the dementia process.

The dynamics of Aβ deposition among different brain regions during disease progression also were investigated. The entorhinal cortex is one of the first neuroanatomical structures pathologically affected in AD.²⁵ In support of this, the levels of Aβx-42 were highest in this region, but they also were elevated in the frontal cortex in the CDR 0.5 and 1.0 groups (Table 2). In addition, the levels of Aβx-42 in the primary visual cortex in the severely demented cases were lower than in other regions. This is consistent with immunohistochemical studies showing that occipital regions usually contain less Aβ pathology.²⁷

Aβ has been shown to lower the neuronal threshold to a variety of cellular insults, among them excitatory amino acids, glucose deprivation, and oxidative stress.²⁸ Gliosis and other inflammatory responses, which can be induced with Aβ in vitro, are also key features of the pathological spectrum of AD.^29- 30 Our results suggest that levels of both Aβx-40 and Aβx-42 are elevated very early in the disease process, perhaps causing the onset of the pathological cascade that propagates the disease and ultimately leads to late-onset AD. From a possible treatment perspective, attempts to prevent or slow the course of AD should focus on inhibiting the cellular production of Aβ and/or slowing the pathogenic assembly of Aβ into aggregates.