Oligomeric Aβ in Alzheimer's Disease: Relationship to Plaque and Tangle Pathology, APOE Genotype and Cerebral Amyloid Angiopathy

Abstract

Despite accumulating evidence of a central role for oligomeric amyloid β (Aβ) in the pathogenesis of Alzheimer's Disease (AD), there is scant information on the relationship between the levels and distribution of oligomeric Aβ and those of other neurodegenerative abnormalities in AD. In the present study, we have found oligomeric Aβ to be associated with both diffuse and neuritic plaques (mostly co‐localized with Aβ_1–42) and with cerebrovascular deposits of Aβ in paraffin sections of formalin‐fixed human brain tissue. The amount of oligomeric Aβ that was labeled in the sections correlated with total Aβ plaque load, but not phospho‐tau load, cerebral amyloid angiopathy (CAA) severity or APOE genotype. Although soluble, oligomeric and insoluble Aβ levels were all significantly increased in AD brain homogenates, case‐to‐case variation and overlap between AD and controls were considerable. Over the age‐range studied (43–98 years), the levels of soluble Aβ, oligomeric Aβ₄₂, oligomeric Aβ₄₀ and insoluble Aβ did not vary significantly with age. Oligomeric Aβ_1–42 and insoluble Aβ levels were significantly higher in women. Overall, the level of insoluble Aβ, but neither oligomeric nor soluble Aβ, was associated with Braak stage, CAA severity and APOEε4 frequency, raising questions as to the role of soluble and oligomeric Aβ in the progression of AD.

INTRODUCTION

Alzheimer's disease (AD) is characterized by distinct pathological abnormalities that include amyloid β (Aβ) plaques. For many years, it was assumed that the fibrillar Aβ that forms plaques was responsible for most of the neurodegenerative changes in AD. However, the distribution and severity of neurofibrillary pathology and neuronal loss correlate poorly with the distribution and number of Aβ plaques 13, 15, 50. It is now thought that soluble oligomeric forms of Aβ are the major toxic species in AD.

Levels of oligomeric Aβ were reported to be increased in human brain tissue from AD patients 32, 37, 40, 48, 57, 59 and a robust correlation was demonstrated between oligomeric Aβ and the presence and degree of cognitive decline 37, 59 and memory loss (20). McLean et al (40) found that levels of soluble Aβ correlated directly with neurofibrillary tangle density.

Neurotoxicity of different species of oligomeric Aβ has been demonstrated by various in vivo and in vitro studies. Memory impairment and changes in neuronal form and function preceded deposition of fibrillar Aβ in hAPP transgenic mice 27, 44, 61. A soluble dodecamer of natural Aβ (Aβ*56) in brains of Tg2576 mice was shown to affect memory (34). Aβ trimers fully inhibited long‐term potentiation in Swiss Webster mice; dimeric and tetrameric species caused incomplete inhibition (56). In cortical neuronal cultures high molecular weight oligomeric species, protofibrils, produced a rapid increase in excitatory post‐synaptic potentials, action potentials and membrane depolarizations, and subsequent cell death (23). In Neuro‐2A neuroblastoma cells, oligomeric Aβ inhibited neuronal viability 10‐fold more than fibrils and approximately 40‐fold more than non‐aggregated, monomeric peptide (14). More recently, Aβ dimers that had been extracted from AD brains were shown to be particularly synaptotoxic (54).

Despite the accumulating evidence of a central role for oligomeric Aβ in the pathogenesis of AD, there is scant information on the relationship between the levels and distribution of oligomeric Aβ and those of other neurodegenerative abnormalities in AD. Our aims in the present study were to examine the quantitative and topographic relationships of oligomeric Aβ to plaque‐ and vessel‐associated Aβ (total Aβ, Aβ₄₀ and Aβ₄₂) and neurofibrillary pathology in histological sections, and the quantitative relationships between oligomeric Aβ, total soluble and total insoluble Aβ in brain tissue homogenates, from AD and age‐matched control brains.

METHODS

Study cohort

We used brain tissue from 90 cases of neuropathologically confirmed AD and 43 neuropathologically normal controls without AD, on whom we had complete data for age at death, gender, post‐mortem delay and disease duration. The cases were from the South West Dementia Brain Bank, University of Bristol. The study had approval from Frenchay Local Research Ethics Committee. The brains had been divided midsagittally at autopsy, the left half sliced and frozen at −80°C and the right half fixed in formalin for paraffin histology. A summary of the demographic and neuropathological characteristics of the study subjects is presented in Table 1. The AD cases ranged from 54 to 98 years in age [mean = 80.0, standard deviation (SD) = 9.1] and comprised 56 females and 34 males. A diagnosis of AD had been made in patients with a clinical presentation of dementia and a neuropathological diagnosis of probable or definite AD according to criteria of the Consortium to Establish a Registry for Alzheimer's Disease (43). The post‐mortem delays were between 4 and 99 h (mean = 45.0, SD = 24.2). Cases with concomitant CNS pathology such as Lewy bodies were excluded from the study. The controls comprised 16 females and 26 males, and were of similar ages to the AD cases, from 43 to 95 years (mean = 76.9, SD = 10.3), had not shown clinical evidence of dementia during the weeks or months before death and had no or only sparse neuritic plaques. The post‐mortem delays were between 3 and 106 h (mean = 41.3, SD = 28.3). The control and AD cases were selected to encompass a spectrum of disease severity across all Braak tangle stages (7) (Table 1).

Table 1.

Summary of demographic and neuropathological characteristics of study subjects. Abbreviation: SD = standard deviation.

	Alzheimer's Disease	Control
Age, years (SD)	80 (9.1)	76.9 (10.3)
Gender, % ♀	62.2%	37.2%
Post‐mortem delay, h (SD)	45.0 (24.2)	41.3 (28.3)
Brain weight, g (SD)	1143.9 (142.7)	1305.9 (142.9)
Aβ plaque load, % (SD)	2.5% (2.0)	0.03% (0.09)
Braak tangle stage
0	0.0%	15.0%
I	0.0%	15.0%
II	0.0%	50.0%
III	6.7%	20.0%
IV	14.4%	0.0%
V	38.9%	0.0%
VI	40.0%	0.0%
APOE genotype frequency, %
ε2/ε3	2.2%	12.2%
ε3/ε3	31.1%	73.2%
ε2/ε4	2.2%	0.0%
ε3/ε4	48.9%	12.2%
ε4/ε4	15.6%	2.4%

Parenchymal Aβ plaque load and cerebral amyloid angiopathy (CAA) severity had previously been determined (11). DNA from the brains studied here had been genotyped for APOE isoform by a polymerase chain reaction method based on that of Wenham and colleagues (60), as reported elsewhere 11, 36.

Immunoperoxidase staining

Sections 7 µm in thickness were cut from large paraffin blocks of frontal lobe in the coronal plane of the head of the caudate nucleus from 65 AD and 37 control brains and collected onto 3‐amino‐propyl‐triethoxy‐silane‐coated slides. After pre‐treatment with formic acid for 20 minutes and subsequent blocking in horse serum solution, all sections were immunostained overnight at room temperature with monoclonal mouse anti‐oligomeric Aβ antibody (1:1000, clone 7A1a, New England Rare Reagents, Gorham, ME, USA), the specificity of which for oligomeric Aβ we had confirmed previously (57). Bound antibody was visualized by subsequent incubation with biotinylated Universal Antibody (Vectastain Universal Elite, Vector Laboratories, Burlingame, CA, USA) and avidin‐biotin horseradish peroxidase followed by reaction with 0.01% H₂O₂ and diaminobenzidine.

To measure the amount of neurofibrillary pathology, we used mouse monoclonal antibody to Ser202 phospho‐tau (1:3000; clone AT8, Autogen Bioclear, Calne, UK). No pre‐treatment was required.

Combined immunofluorescence

To investigate the relationship between oligomeric Aβ and total Aβ, double immunofluorescent labeling was performed on sections of frontal lobe from nine brains (four controls and five AD). We used monoclonal mouse anti‐oligomeric Aβ antibody 7A1a, with overnight incubation, and affinity‐purified polyclonal rabbit anti‐pan‐Aβ antibody (1:100; Zymed Laboratories, South San Francisco, CA, USA) with a 1 h incubation. Bound antibodies to oligomeric Aβ and pan Aβ were visualized using Texas Red‐tagged anti‐mouse IgG and fluorescein isothiocyanate (FITC)‐tagged anti‐rabbit IgG (both fluorochrome‐labeled antibodies from Vector Laboratories), respectively (1:100 in PBS/20% horse serum and 1:100 in PBS/20% goat serum, respectively). To investigate the relationship between oligomeric Aβ and total Aβ_1–42, sections of temporal lobe from six brains (three controls and three AD) were immunolabeled with 7A1a as above, and affinity purified polyclonal rabbit anti‐Aβ_1–42 antibody (1:100, Millipore, Billerica, MA, USA) with a 1 h incubation. Bound 7A1a antibody was visualized as described, and Aβ_1–42 with FITC‐tagged anti‐rabbit IgG (1:100 in PBS/20% goat serum).

Triple immunofluorescent labeling was used to look at the relative distributions of oligomeric Aβ and plaque‐ and vessel‐associated Aβ₄₂ and Aβ₄₀ in frontal lobe sections from 6 AD brains. Following the TSA system protocol (PerkinElmer LAS, Hemel Hempstead, UK), affinity purified polyclonal rabbit anti‐Aβ₄₂ antibody (1:2500, Millipore) was applied to the sections and successfully visualized with streptavidin‐FITC (SA‐FITC). Sections were then immunolabeled for oligomeric Aβ and Aβ₄₀ using the double immunofluorescence protocol. Sections were incubated with 7A1a antibody and visualized as described. Anti‐Aβ₄₀ was added to the sections for 1 h (1:100, Alpha Diagnostic International, San Antonio, TX, USA) and visualized using aminomethylcoumarin (AMCA)‐tagged anti‐rabbit IgG (Vector Laboratories) (1:100 in PBS/20% horse serum).

Double and triple immunofluorescent images were captured by a JVC video camera with appropriate filters, and merged using Kromascan (Kinetic Imaging, Belfast, UK) software.

Quantitative image analysis

The parenchymal oligomeric Aβ load and phosporylated tau load were measured by the same protocol that we had previously used to measure parenchymal Aβ load (11). The field fraction (percentage area) immunopositive for 7A1a or AT8 within the brain parenchyma was measured with the help of Histometrix software (Kinetic Imaging) driving a Leica DM microscope with a motorized stage. The software was used for random selection of 10 × 10 objective fields within three areas of neocortex, the percentage area immunolabeled for oligomeric Aβ or phospho‐tau determined for each of these fields and the mean value calculated.

Oligomeric Aβ ELISA and post‐mortem stability

Oligomeric Aβ_1–40 and Aβ_1–42 levels in the subjects included in this study were reported previously (57): briefly, a sandwich enzyme‐linked immunosorbent assay (ELISA) was designed in which either rabbit polyclonal pan‐Aβ₄₂ or rabbit polyclonal Aβ₄₀ was used as the capture antibody (the same antibodies used for immunofluorescence, as described above) and 7A1a as the detection to specifically detect oligomeric Aβ_1–40 and Aβ_1–42 species.

To assess the stability of oligomeric Aβ within the brain post‐mortem, we used the approach of Fraser et al (18) to simulate post‐mortem delay of up to 3 days at 4°C or room temperature. Samples of frontal cortex from two AD brains with short post‐mortem delays (4–5 h) were used. The samples were subdivided and stored for periods of up to 3 days at room temperature (t = 0 h, 6 h, 12 h, 18 h, 24 h, 48 h, 72 h) or at 4°C (t = 24 h, 48 h, 72 h), after which the oligomeric Aβ_1–42 levels were determined by the ELISA method described above. Duplicate samples were analyzed for each time point and temperature.

Soluble/insoluble Aβ ELISAs

Homogenates were prepared from fresh frozen human brain tissue (frontal neocortex, Brodmann area 6) by the method of Hemming et al (24) in 5 volumes (wt : vol) of TBS extraction buffer [140 mM NaCl, 3 mM KCl, 25 mM Tris (pH 7.4), containing 1% Nonidet P‐40 (NP40)], 5 mM EDTA, 2 mM 1,10‐phenanthroline and protease inhibitors PMSF (10 µM) and aprotinin (1 µg/mL) (all reagents from Sigma Aldrich, Dorset, UK). Tissue (200 mg), allowed to thaw to 4°C, was homogenized for 30 s in a Precellys 24 automated tissue homogenizer (Stretton Scientific, Stretton, UK) with 2.3 mm silica beads (Biospec, Thistle Scientific, Glasgow, UK). Resultant tissue homogenates were spun at 20 817 g for 15 minutes at 4°C and the supernatant (soluble fraction) was stored at −80°C until used. The pellet was homogenized in 6.25 M guanidine HCl in 50 mM Tris (pH 8.0), incubated for 4 h at 25°C and spun at 20 817 g for 20 minutes at 4°C. The resultant supernatant was saved as the guanidine‐HCl soluble fraction and stored at −80°C until used.

Sandwich ELISAs for quantifying human Aβ_1–42 used monoclonal anti‐Aβ (4G8 clone, raised against amino acids 18–22; Millipore) as the capture antibody and biotinylated anti‐human β‐amyloid monoclonal (10H3 clone) (Thermo Fisher Scientific, Northumberland, UK) as the detection antibody. High‐binding Costar 96‐well plates (R&D Systems Europe, Abingdon, UK) were coated with 1 µg/well capture antibody in PBS and incubated overnight at room temperature. After washing five times with 0.05% tween‐20 (Sigma Aldrich), plates were blocked with 300 µL protein‐free™ PBS blocking buffer (Thermo Fisher Scientific) for 3 h at room temperature. After a further five washes, recombinant human Aβ_1–42 (Sigma Aldrich) or brain homogenate samples, prepared in PBS, were incubated for 5 h at room temperature with rocking. The plates were again washed and biotinylated detection antibody 0.1 µg/well was added overnight at 4°C. Following further washing, streptavidin‐HRP (1:100) (R&D Systems Europe) was added to each well for 1 h at room temperature before washing and incubation with substrate solution (TMB) (R&D Systems Europe) for 30 minutes in the dark. The reaction was stopped with 2 N sulfuric acid (R&D Systems Europe) and the optical density for each well read at 450 nm in a FLUOstar plate reader (BMG Labtech, Aylesbury, UK). Total Aβ levels were interpolated from a standard curve generated from serial dilutions of recombinant human Aβ_1–42 (Sigma Aldrich). Each measurement was repeated on two separate occasions and the average calculated.

Statistical analysis

Pearson's test was used to analyze the relationship of oligomeric Aβ load to total Aβ load and phospho‐tau load, and Spearman's test to examine the relationship of oligomeric Aβ load to CAA severity. Mean values for oligomeric Aβ load in AD brains with zero, one or two APOEε4 alleles were compared by one‐way analysis of variance (ANOVA). Inspection of the ELISA measurements of oligomeric Aβ_1–40, oligomeric Aβ_1–42 and total insoluble Aβ showed that the distributions were skewed to the right; the data were normalized by logarithmic transformation, and independent‐samples t‐tests and one‐way ANOVA with Bonferroni's post hoc testing were used to explore the relationships between Aβ concentrations and other disease parameters. Both Pearson's and Spearman's tests were used to analyze the correlations between the levels of oligomeric, soluble and insoluble Aβ. Statistical analyses were performed with the help of Statistical Package for Social Science software (version 12.0.1 for Windows, SPSS Inc., Chicago, IL, USA) and GraphPad Prism 5 (Graphpad Software, Inc., La Jolla, CA, USA). Values of P < 0.05 were considered to be statistically significant.

RESULTS

Immunohistochemical analyses

Distribution of oligomeric Aβ within the brain

In AD, oligomeric Aβ was detected throughout the frontal neocortex (Figure 1A) and in the walls of many cortical and leptomeningeal arterioles and some venules. Much of the oligomeric Aβ was present in diffuse plaques. However, oligomeric Aβ was also detected in mature neuritic and cored plaques (Figure 1B and C). In many sections, oligomeric Aβ was present in the walls of cortical capillaries and formed plaques in the adjacent parenchyma (Figure 1D).

Figure 1.

Immunoperoxidase staining showed abundant oligomeric Aβ in AD compared with control brains. A. In Alzheimer's disease (AD), oligomeric Aβ was detected throughout the neocortex and in the walls of many cortical and leptomeningeal arterioles and some venules. Much of the oligomeric Aβ was present in diffuse plaques (B) but neuritic and cored plaques were also labeled. C. Oligomeric Aβ was also present in the walls of cortical capillaries (D) and formed plaques in the adjacent parenchyma. E,F. Very little oligomeric Aβ was detected immunohistochemically in control brains, and that mainly as diffuse plaques.

Very little oligomeric Aβ was detected immunohistochemically in control brains; what was detected was mainly in the form of diffuse plaques (Figure 1E,F).

Relationship between oligomeric and total Aβ load, tau load, CAA severity and APOE

There was a significant positive correlation between oligomeric and total parenchymal Aβ load in the AD brains (P = 0.002, r² = 0.143; Figure 2A). Although total parenchymal Aβ load also correlated significantly with tau load (P = 0.02, r² = 0.093; Figure 2B), the relationship between oligomeric Aβ load and tau load was not quite significant (P = 0.08; Figure 2C). Oligomeric Aβ load did not show any correlation with CAA severity and did not vary significantly with APOE genotype.

Figure 2.

In paraffin sections, oligomeric Aβ load correlated with total Aβ load but not phospho‐tau load, and was not related to cerebral amyloid angiopathy severity or APOE. A. The area fraction of frontal cortex immunopositive for oligomeric Aβ (ie, oligomeric Aβ load) in 65 cases of Alzheimer's Disease showed a highly significant positive correlation with total Aβ load (P = 0.002, r² = 0.143). B. Total Aβ load correlated with phospho‐tau load (P = 0.02, r² = 0.093). C. Oligomeric Aβ load tended to increase with phospho‐tau load but the correlation did not reach significance (P = 0.08). Each point in the scatterplots represents a single case. The best‐fit linear regression is shown by a solid line and the 95% confidence interval by interrupted lines.

Relative distributions of oligomeric Aβ, Aβ_1–40 and Aβ_1–42

On immunofluorescent staining of paraffin sections, labeling of oligomeric Aβ was much more abundant in AD cases than controls. Relative to total Aβ distribution, oligomeric Aβ was present in most (but not all) diffuse plaques, and formed a corona around the cores of mature plaques (Figure 3A,B). Most plaques that were immunopositive for oligomeric Aβ also contained Aβ_1–42 (Figure 3C). However, some oligomeric Aβ‐positive plaques did not contain demonstrable Aβ_1–42 (Figure 3D). Further investigation with triple immunofluorescence revealed that in those plaques where oligomeric Aβ did not co‐localize with Aβ_1–42, it co‐localized with Aβ_1–40 (Figure 4A). Most vascular accumulations of Aβ in the meninges and cerebral parenchyma contained oligomeric Aβ as well as Aβ_1–40 (Figure 4B,C). In general, the vascular deposits contained smaller amounts of Aβ_1–42 than either oligomeric Aβ or Aβ_1–40.

Figure 3.

Relative distribution of oligomeric Aβ, Aβ and Aβ_1–42 assessed by double immunofluorescence. A,B. Labeling with a pan‐Aβ antibody (green signal) and antibody to oligomeric Aβ (red) showed oligomeric Aβ to be present in most plaques, and to form a corona (arrow in B) around the cores of some mature plaques. C. Most plaques that were immunopositive for oligomeric Aβ (red) also contained Aβ_1–42 (green). D. However, some plaques that were labeled for oligomeric Aβ contained only scanty Aβ_1–42 (arrows).

Figure 4.

Relative distribution of oligomeric Aβ, Aβ_1–40 and Aβ_1–42 assessed by triple immunofluorescence. A. Combined labeling indicated that plaques positive for oligomeric Aβ (red) but largely negative for Aβ_1–42 (green), were positive for Aβ_1–40 (blue) (arrows). Note that some of the larger Aβ_1–42 deposits are negative for oligomeric Aβ. B,C. In vessel walls, oligomeric Aβ (red) was present in association with most deposits of Aβ_1–40 (blue); in general, the vessels contained smaller amounts of Aβ_1–42 (green).

Aβ ELISAs

These results are summarized in Table 2.

Table 2.

Summary of ELISA‐based analyses. Abbreviations: AD = Alzheimer's Disease; CAA = cerebral amyloid angiopathy; ELISA = enzyme‐linked immunosorbent assay; NS = not significant.

	Control vs. AD	Age	Gender			Braak*	CAA†	APOEε4 frequency	Disease duration	Post‐mortem delay
			Control vs. AD	Control only	AD only
Oligomeric Aβ42	P = 0.041 (↑ AD)	NS	P = 0.002 (↑♀)	NS	P < 0.0001 (↑♀)	NS‡	NS	NS††	P = 0.046 (↑)	NS
Oligomeric Aβ40	NS	NS	NS	NS	NS	NS	NS	NS	P = 0.03 (↑)	NS
Soluble Aβ	P = 0.002 (↑ AD)	NS	NS	NS	NS	NS	NS	NS	P = 0.04 (↑)	NS
Insoluble Aβ	P < 0.0001 (↑ AD)	NS	P = 0.01 (↑♀)	P = 0.008 (↑♀)	NS	P < 0.0001§, ¶ (↑)	P < 0.0001** (↑)	P < 0.0001‡‡, §§ (↑)	P = 0.02 (↑)	NS

^* Braak stages are grouped into three categories: 0–II, III–IV, V–VI.

^† CAA severity is divided into two categories: absent to mild, moderate to severe.

^‡ Females with Braak tangle V–VI pathology had significantly higher oligomeric Aβ42 concentration than did corresponding males (P = 0.01).

^§ Post hoc Bonferroni analysis showed significantly increased insoluble Aβ in stages III–IV (P = 0.0001) and V‐VI (P < 0.0001) compared with group 0–II, and significantly increased insoluble Aβ in stages.

^¶ Females with Braak tangle stage III–IV pathology had significantly higher levels of insoluble Aβ than did corresponding males (P = 0.03).

^** A significant increase in insoluble Aβ was found in cases with moderate to severe CAA compared with cases with absent or mild CAA.

^†† Significantly higher levels of oligomeric Aβ42 were found in females with no APOEε4 alleles (P = 0.03) or one APOEε4 allele (P = 0.02) than did corresponding males.

^‡‡ Post hoc Bonferroni analysis revealed that insoluble Aβ levels were significantly higher in individuals with one (P = 0.001) or two (P < 0.001) copies of the APOEε4 alleles, but no significant differences were observed between individuals with either one or two copies of APOEε4.

^§§ Significantly higher levels of insoluble Aβ were found in females with no APOEε4 alleles than did corresponding males (P = 0.005).

Aβ levels in fresh frozen human brain tissue

We previously reported that although case‐to‐case variation was considerable, oligomeric Aβ_1–42 levels were significantly higher in AD cases than controls (P = 0.041, independent samples t‐test) but found no significant difference in oligomeric Aβ_1–40 levels (57). Here, significantly increased levels of total soluble and total insoluble Aβ were also found in the AD brains compared with the controls (P = 0.002 and P < 0.0001 respectively) (Figure 5A,B). No significant correlations were identified between levels of soluble, oligomeric and insoluble Aβ within the AD or control cohorts. Simulated post‐mortem delay of up to 72 h at room temperature or at 4°C did not have a significant effect on the concentration of oligomeric Aβ. We also found that the levels of oligomeric Aβ_1–42 as determined by ELISA did not correlate with the immunohistochemical measurements of oligomeric Aβ load (data not shown).

Figure 5.

Increased Aβ levels in homogenates of frontal cortex in Alzheimer's Disease (AD). Significantly increased levels of (A) total soluble (P = 0.002) and (B) total insoluble Aβ (P < 0.0001), as measured by sandwich ELISA, were found in homogenates of frontal neocortex from AD brains compared with controls, in keeping with previous analysis demonstrating increased oligomeric Aβ_1–42 levels within the same cohort of AD brains compared with controls. Each point represents the mean value in a single case. The horizontal bars indicate the mean and standard deviation.

Age and gender

No significant differences in levels of soluble, oligomeric Aβ or insoluble Aβ were observed with age within the AD or control cohorts.

Levels of insoluble Aβ (P = 0.01) and oligomeric Aβ_1–42 (P = 0.0002) (Figure 6A,B) were significantly higher in females when AD and control cases were analyzed together. This analysis was repeated for AD and control cases separately: in controls, females had significantly increased levels of insoluble Aβ only (P = 0.008) (Figure 6C), whereas in AD, females had significantly increased levels of oligomeric Aβ_1–42 only (P < 0.0001) (Figure 6D).

Figure 6.

Higher levels of insoluble Aβ and oligomeric Aβ_1–42 in females. Levels of (A) insoluble Aβ (P = 0.01) and (B) oligomeric Aβ_1–42 (P = 0.0002) were significantly higher in females when Alzheimer's Disease (AD) and control cases were analyzed together. C. On separate analysis of controls, females had significantly increased levels of insoluble Aβ only (P = 0.008). D. On separate analysis of AD cases, females had significantly increased levels of oligomeric Aβ_1–42 only (P < 0.0001). Each point represents the mean value in a single case. The horizontal bars indicate the mean and standard deviation.

In our study, 60% of females had Braak tangle stage V‐VI pathology, compared with 45% of males, and these females had a significantly higher oligomeric Aβ_1–42 concentration (P = 0.01, Figure 7A). Within Braak stages III‐IV, although proportions of females (20%) and males (26.7%) were similar, significantly higher levels of insoluble Aβ were found in the females (P = 0.03, Figure 7B). No significant differences in Aβ levels were found between males (28.3%) and females (20%) in Braak stages 0‐II.

Figure 7.

Levels of different forms of Aβ in relation to Braak tangle stage. A. In brains with Braak stage V–VI pathology (60% of females and 45% of males), the level of oligomeric Aβ_1–42 was significantly higher in females (P = 0.01); levels of total insoluble Aβ did not differ between males and females. B. In brains with Braak stage III–IV pathology (20% of females and 26.7% of males), a significantly higher level of insoluble Aβ was found in females (P = 0.03). Here, levels of oligomeric Aβ_1–42 did not differ between males and females. No significant differences in Aβ levels were found between males (28.3%) and females (20%) in Braak stages 0–II. Each point represents the mean value in a single case. The horizontal bars indicate the mean and standard deviation.

Sixty‐one percent of females carried at least one APOEε4 allele compared with 39% of males (Table 3). Significantly higher levels of oligomeric Aβ_1–42 (P = 0.03, Figure 8A) and insoluble Aβ (P = 0.005, Figure 8B) were found in females with no APOEε4 alleles compared with corresponding males. Compared with males carrying one APOEε4 allele, females had significantly higher levels of oligomeric Aβ_1–42 (P = 0.02, Figure 8C). No significant differences in Aβ levels were observed between genders for APOEε4 homozygotes.

Table 3.

Gender proportions for APOEε4 alleles frequency.

		Females (%)	Males (%)
# APOEε4	0	40	61
	1	47	34
	2	14	5

Figure 8.

The levels of different forms of Aβ in females were associated with APOE ε4 allele frequency. Levels of (A) oligomeric Aβ_1–42 (P = 0.03) and (B) insoluble Aβ (P = 0.005) were significantly higher in females than males with no APOEε4 alleles. C. Compared with males heterozygous for APOEε4 allele, heterozygous females had significantly higher levels of oligomeric Aβ_1–42 (P = 0.02). No significant differences in Aβ levels were observed between genders for APOEε4 homozygotes. Each point represents the mean value in a single case. The horizontal bars indicate the mean and standard deviation.

Braak stage, CAA, APOE and post‐mortem delay

Braak stage, CAA severity and APOEε4 frequency were significantly associated with the level of insoluble but not soluble or oligomeric Aβ. Comparison of insoluble Aβ level in Braak tangle stages 0–II, III–IV and V–VI yielded a P‐value of <0.0001 (one‐way ANOVA). Post hoc Bonferroni analysis showed significantly increased insoluble Aβ in stages III–IV (P = 0.0001) and V–VI (P < 0.0001) compared with group 0–II, and significantly increased insoluble Aβ in stages V–VI compared with group III–IV (P = 0.007) (Figure 9A).

Figure 9.

Braak stage, cerebral amyloid angiopathy (CAA) severity and APOE ε4 frequency were significantly associated with the level of insoluble but not soluble or oligomeric Aβ. A. Comparison of insoluble Aβ level in Braak tangle stages 0–II, III–IV and V–VI yielded a P‐value of <0.0001 (one‐way ANOVA). Post hoc Bonferroni analysis showed significantly increased insoluble Aβ in stages III–IV (P = 0.0001) and V–VI (P < 0.0001) compared with group 0–II, and significantly increased insoluble Aβ in stages V–VI compared with group III–IV (P = 0.007). B. A significant increase in insoluble Aβ was found in cases with moderate to severe CAA compared with cases with absent or mild CAA (P < 0.0001). C. Insoluble Aβ levels significantly increased with number of APOEε4 alleles (P < 0.0001, one‐way ANOVA). Post hoc Bonferroni analysis revealed that Aβ levels were significantly higher in individuals with one (P = 0.001) or two (P < 0.001) than with no copies of the APOEε4 allele, but no significant differences were observed between individuals with one and two copies.

A significant increase in insoluble Aβ was found in cases with moderate to severe CAA compared with cases with absent or mild CAA (P < 0.0001) (Figure 9B). Insoluble Aβ levels increased significantly with number of APOEε4 alleles (P < 0.0001, one‐way ANOVA) (Figure 9C). Post hoc Bonferroni analysis revealed that Aβ levels were significantly higher in individuals with one (P = 0.001) or two (P < 0.001) copies of the APOEε4 allele, but no significant differences were observed between individuals with either one or two copies of APOEε4.

Soluble Aβ, oligomeric Aβ_1–40, oligomeric Aβ_1–42 and insoluble Aβ levels were positively correlated with disease duration (P = 0.04, r² = 0.07; P = 0.03, r² = 0.28; P = 0.046, r² = 0.05; and P = 0.02, r² = 0.09, respectively) (data not shown).

As might be expected from the data obtained after simulated post‐mortem delay, there was no significant correlation between actual post‐mortem delay and the level of oligomeric Aβ in AD or control brains. Nor did the levels of total soluble or insoluble Aβ correlate with post‐mortem delay.

DISCUSSION

Oligomeric Aβ is considered to be a major cause of neurodegeneration in AD. In the present study, we have shown that in paraffin sections of formalin‐fixed human brain tissue, oligomeric Aβ is associated with both diffuse and neuritic plaques (mostly co‐localized with Aβ_1–42 but in some plaques with Aβ_1–40 and not Aβ_1–42) and with cerebrovascular deposits of Aβ, and that the amount of oligomeric Aβ that can be labeled in the sections correlates with the total Aβ plaque load. We did not find the same correlation between oligomeric and total insoluble (or total soluble) Aβ in brain tissue homogenates, where oligomeric Aβ level was found to be higher in females than males in AD and to correlate with disease duration but not to vary significantly with Braak tangle stage, CAA severity, APOE genotype or post‐mortem delay.

The labeling of a corona of oligomeric Aβ around the cores of mature plaques was previously observed in APP/PS1 transgenic mice (31) in a study that used a different oligomer‐specific antibody (NAB61) (33). The authors suggested that plaques may act as a local reservoir of soluble oligomeric Aβ(31). An alternative explanation is suggested by the work of Carare et al (9). These authors reported that 3 kDa dextran that had been injected into the hippocampus in a triple‐transgenic mouse model of AD co‐localized with plaques rather than draining along perivascular spaces as the tracer did in normal control mice. Plaque deposition and CAA may interfere with the drainage of solutes, including soluble oligomeric forms of Aβ, and lead to their accumulation in association with plaques.

We have found that soluble, oligomeric and insoluble Aβ levels are all significantly increased in AD compared with age‐matched controls. It is possible that the levels of insoluble Aβ reported here underestimate the total insoluble Aβ within the brain, as guanidine hydrochloride extraction, while sufficient for studies with transgenic mice, may not extract all of the highly insoluble Aβ from human brain tissue. In the present series, case‐to‐case variation was considerable. Similar case‐to‐case variation in Aβ levels, as measured by ELISA, was reported by others (62). By examining the effects of simulated post‐mortem delay, we have been able to exclude this as a significant contributor to the case‐to‐case variability in the level of oligomeric Aβ. It is clear that soluble oligomeric species of Aβ are produced in non‐demented age‐matched controls, implying that Aβ oligomers are not disease‐specific. Kuo et al demonstrated that both AD and control brains contained a continuous distribution of Aβ species from monomer up to oligomers in excess of 100 kDa, with the major contribution coming from low‐n oligomers ranging from dimers to octamers (32). Soluble Aβ was detected from brains in age‐matched controls that had neuropathological hallmarks of AD but showed no synapse loss or symptoms of dementia37, 59. Wang et al reported that soluble Aβ₄₀ and Aβ₄₂ constituted the largest fractions of total Aβ in the normal brain but the smallest in AD (59). The marked variation between different control brains or AD brains may reflect differences in production, aggregation and disaggregation as well as in proteolytic degradation or clearance of Aβ from the brain. The antibody to oligomeric Aβ that we used in this study detects all soluble, non‐monomeric, aggregates of Aβ(57) and does not discriminate between relative levels of specific Aβ multimers.

Although our immunohistochemical analysis revealed a significant correlation between oligomeric Aβ load and total Aβ plaque load in the frontal neocortex in AD, we did not find a significant relationship between the levels of oligomeric and insoluble (or soluble) Aβ by ELISA of brain tissue homogenates. Indeed, no correlation was found between oligomeric Aβ load as measured immunohistochemically and levels of oligomeric Aβ_1–42 as measured by ELISA. It should be noted that the method used here for measuring levels of total and oligomeric Aβ load in histological sections, although widely used, is not optimal as it indicates the area occupied by Aβ but does not take account of the intensity of labeling. A recent ELISA‐based study of Aβ species in plasma and brain tissue found a close association between levels of oligomeric and insoluble Aβ₄₂ in brain tissue (62), in agreement with our immunohistochemical analysis but not our ELISAs. However, Xia et al (62) measured Aβ levels in brain tissue from only 9 AD and 7 control brains and the considerable variation in Aβ levels reported here, as well as by Xia et al suggest caution in extrapolating data from small series to larger populations. The difference between the results obtained immunohistochemically and by ELISA may be due in part to the techniques used to preserve the integrity of the brain tissue for immunohistochemical analysis: some of the soluble oligomeric Aβ may be lost during tissue fixation and processing, leaving only what has been cross‐linked by formalin to insoluble plaques and vascular deposits of Aβ. Fixation is also well known to influence retention of antigenicity on immunohistochemistry; for example, fixation of cynomolgus monkey brain fixed in neutral buffered formalin (as in the present study) caused loss of Aβ immunoreactivity in cortical neurons whereas labeling was preserved in sections of brain fixed with paraformaldehyde (30). This discrepancy between immunohistochemical and ELISA results raises the question of what the various assays are actually measuring. More work is required to determine how and where conformation‐dependent antibodies bind to Aβ, and to understand the effects of tissue fixation and paraffin processing on antigen retention, conformation and antibody binding.

The most significant non‐genetic risk factor for AD is age. Overall, we did not observe significant differences in the levels of soluble Aβ, oligomeric Aβ₄₂, oligomeric Aβ₄₀ or insoluble Aβ over the age‐range studied (43–98 years). In keeping with our results, no age‐related alteration in oligomeric Aβ level was observed in plasma (62). Multiple studies have shown increasing levels of various forms of Aβ with age in mouse models of AD 19, 28, 45. However, of several human studies of Aβ plaque burden in relation to age, some have shown no correlation or even a decline 6, 51, 58.

Evidence is mixed on the relationship between gender and prevalence of AD. Several studies have shown that females have a higher risk of developing AD 1, 29, 52, 53 but for reasons that are still unclear, this influence of gender has not been found in all series 2, 35, 55. Our study found significantly higher levels of oligomeric Aβ_1–42 and insoluble Aβ in brain tissue from women. The significantly elevated levels of oligomeric Aβ_1–42 in females with end‐stage (Braak stage V–VI) AD could reflect the contribution of oligomers to the development of tangle pathology, and if aggregation of Aβ is more likely in females, this might help to explain their increased prevalence of AD. An influence of gender on Aβ accumulation and aggregation has been observed in transgenic mice. Callahan et al (8) showed that female Tg2576 APP mice had greater plaque load at 15 and 19 months and ELISA‐quantified Aβ_1–40 (but not Aβ_1–42) level at 15 months (the level was not measured at other ages). In Tg2576 and 3xTg‐AD mice, at the plaque‐free stage, Aβ level did not vary between genders. However, in plaque‐bearing mice, levels of soluble and insoluble Aβ_1–40 and Aβ_1–42 were greater in female mice, which also had more plaque pathology than age‐matched males (25). The authors found that the female mice had higher β‐secretase activity in early life and less efficient Aβ degradation (their assay specifically measured neprilysin) as they aged.

In our study, 61% of females carried at least one APOEε4 allele compared with 39% of males. We previously demonstrated a significant independent relationship between possession of ε4 and reduced expression of neprilysin (41). This and other allele‐specific effects APOE on Aβ degradation and clearance 5, 16, 42 may have contributed to the elevated levels of oligomeric and insoluble Aβ in females in the present study. The influence of APOE on risk of AD may vary between men and women. Payami et al (47) reported that the risk of developing AD was as high in ε4 heterozygous females as in those homozygous ε4 whereas in male heterozygotes the risk was significantly less than in homozygotes.

Overall, we found that the level of insoluble Aβ, but neither oligomeric nor soluble Aβ, was associated with Braak stage, CAA severity and APOEε4 frequency. The fact that Braak stage correlates better with the level of insoluble than neither oligomeric nor soluble Aβ may be explained by the marked delay between the accumulation of neurotoxic forms of Aβ, the initiation of neuronal damage and the eventual development of tau pathology. In Down's syndrome, in which Aβ is produced in excess throughout life, diffuse Aβ deposits have been shown to start accumulating in the mid‐teens, whereas neurofibrillary tangles are seen only when the patients reach their fourth decade, when cored plaques are also observed 38, 39. Although oligomeric Aβ may interfere with synaptic function and induce early symptoms of AD (12), the accumulation of phospho‐tau and development of neurofibrillary changes occur only after the formation of Aβ plaques. In CAA, Aβ_1–40 is the predominant form of Aβ in the vasculature. Oligomeric Aβ₄₀ was not significantly elevated in AD in our study, so the lack of association between levels of oligomeric Aβ and severity of CAA was not surprising. Further investigation is needed to characterize the physical conformations and species of soluble as well as insoluble Aβ in the perivascular compartment. These are probably influenced by APOE, which influences Aβ fibrillogenesis 3, 4, 10, 17, 26 and is strongly associated with severity of CAA 11, 21, 22, 46, 49.