Abstract

Alzheimer’s disease is characterized by the deposition of amyloid-β peptide in the brain. N-terminal truncation resulting in the formation of AβN3pE and phosphorylation at serine 8 have been reported to modify aggregation properties of amyloid-β. Biochemically, soluble, dispersible, membrane-associated, and insoluble, plaque-associated amyloid-β aggregates have been distinguished. Soluble and dispersible amyloid-β aggregates are both in mixture with the extracellular or intracellular fluid but dispersible aggregates can be cleared from proteins in solution by ultracentrifugation. To clarify the role of phosphorylated amyloid-β and AβN3pE in soluble, dispersible, membrane-associated, and plaque-associated amyloid-β aggregates in the pathogenesis of Alzheimer’s disease we studied brains from 21 cases with symptomatic Alzheimer’s disease, 33 pathologically preclinical Alzheimer’s disease cases, and 20 control cases. Western blot analysis showed that soluble, dispersible, membrane-associated and plaque-associated amyloid-β aggregates in the earliest preclinical stage of Alzheimer’s disease did not exhibit detectable amounts of AβN3pE and phosphorylated amyloid-β. This stage was referred to as biochemical stage 1 of amyloid-β aggregation and accumulation. In biochemical amyloid-β stage 2, AβN3pE was additionally found whereas phosphorylated amyloid-β was restricted to biochemical amyloid-β stage 3, the last stage of amyloid-β aggregation. Phosphorylated amyloid-β was seen in the dispersible, membrane-associated, and plaque-associated fraction. All cases with symptomatic Alzheimer’s disease in our sample fulfilled biochemical amyloid-β stage 3 criteria, i.e. detection of phosphorylated amyloid-β. Most, but not all, cases with pathologically preclinical Alzheimer’s disease had biochemical amyloid-β stages 1 or 2. Immunohistochemistry confirmed the hierarchical occurrence of amyloid-β, AβN3pE, and phosphorylated amyloid-β in amyloid plaques. Phosphorylated amyloid-β containing plaques were, thereby, seen in all symptomatic cases with Alzheimer’s disease but only in a few non-demented control subjects. The biochemical amyloid-β stages correlated with the expansion of amyloid-β plaque deposition and with that of neurofibrillary tangle pathology. Taken together, we demonstrate that AβN3pE and phosphorylated amyloid-β are not only detectable in plaques, but also in soluble and dispersible amyloid-β aggregates outside of plaques. They occur in a hierarchical sequence that allows the distinction of three stages. In light of our findings, it is tempting to speculate that this hierarchical, biochemical sequence of amyloid-β aggregation and accumulation is related to disease progression and may be relevant for an increasing toxicity of amyloid-β aggregates.

Introduction

Alzheimer’s disease is characterized by the formation of amyloid-β (Aβ) protein aggregates in the brain (Alzheimer, 1907; Masters et al., 1985; Hyman et al., 2012). Various types of Aβ aggregates have been described: soluble Aβ oligomers, dispersible Aβ oligomers, protofibrils and fibrils, membrane-associated (SDS soluble) Aβ aggregates, and solid, plaque-associated (formic acid soluble) Aβ fibrils (Harper et al., 1997; Walsh et al., 1997; Kayed et al., 2003; Lesne et al., 2006; Habicht et al., 2007; Shankar et al., 2008; Rijal Upadhaya et al., 2012a). Recently, we showed that dispersible Aβ oligomers, protofibrils, and fibrils are pathologically relevant forms of Aβ aggregates that cause neurotoxic effects in a concentration-dependent manner as demonstrated in APP transgenic mouse models (Rijal Upadhaya et al., 2012a). Dispersible Aβ aggregates represent diffusible but non-soluble Aβ aggregates that differ from insoluble membrane-associated and plaque-associated Aβ aggregates. In contrast to plaque-associated and membrane-associated Aβ, dispersible Aβ is thought to represent not fully dissolved Aβ aggregates mixed with the extra- or intracellular fluid whereas membrane-associated and plaque-associated Aβ are not mixed with the extracellular fluid and become detectable only after SDS or formic acid treatment. Dispersible Aβ can be separated from soluble Aβ by ultracentrifugation (Rijal Upadhaya et al., 2012a). Currently, it is not clear whether dispersible Aβ aggregates play a role in the human Alzheimer’s disease brain.

Post-translational N-terminal truncation of the first two amino acids of the Aβ peptide and subsequent pyroglutamate formation at the N-terminal glutamate resulting in AβN3pE (Saido et al., 1995) as well as phosphorylation of serine 8 of Aβ (Kumar et al., 2011) have been described. Both AβN3pE and phosphorylated Aβ occur in Alzheimer’s disease plaques (Saido et al., 1995; Kumar et al., 2011). AβN3pE increases the aggregation propensity of Aβ by changing the biophysical properties of Aβ fibrils (Schlenzig et al., 2009). Phosphorylated Aβ, on the other hand, promotes the formation of Aβ oligomers that serve as nucleation sites for fibril formation (Kumar et al., 2011). Here, it is suggested that these modifications of Aβ play a role in Alzheimer’s disease pathogenesis and the development of dementia, but it is not clear at which stage of the disease and in which types of aggregates these modifications may become evident.

Alzheimer’s disease begins many years before the cognitive deficits become evident that allow its clinical diagnosis. The recently revised criteria for the clinical diagnosis of Alzheimer’s disease introduced preclinical Alzheimer’s disease as a diagnosis for non-demented individuals with positive biomarkers for Alzheimer’s disease, e.g. Alzheimer’s disease-like Pittsburg compound B retention in the brain (Dubois et al., 2007; Sperling et al., 2011). Current neuropathological guidelines for the assessment of Alzheimer’s disease pathology recommend the description of the level of Alzheimer’s disease pathology in a given brain regardless of the ante-mortem cognitive status (Hyman et al., 2012). As such, by employing the neuropathological diagnosis of Alzheimer’s disease pathology as a biomarker for Alzheimer’s disease, pathologically preclinical Alzheimer’s disease cases are those that were non-demented before death, but exhibit at least low levels of Alzheimer’s disease pathology at autopsy, whereas symptomatic Alzheimer’s disease cases exhibit significant Alzheimer’s disease pathology and cognitive impairment (Monsell et al., 2013). Non-Alzheimer’s disease cases are those without any Aβ plaques (Hyman et al., 2012). The term ‘pathologically preclinical Alzheimer’s disease’ for non-demented cases with amyloid plaques does not necessarily mean that these cases must convert into symptomatic Alzheimer’s disease. It cannot be excluded that progression of pathology may cease in some of these cases although they fulfil the recommended criteria for low levels of Alzheimer’s disease pathology. Whether the biochemical composition of cortical Aβ aggregates has impact on the course of the disease and the development of dementia is not yet clear. Clinical criteria for preclinical Alzheimer’s disease based on CSF-Aβ and CSF-tau protein levels have been published (Vos et al., 2013) that differ from the neuropathological definition of pathologically preclinical Alzheimer’s disease that is used here, synonymous with the term ‘asymptomatic Alzheimer’s disease’ (Monsell et al., 2013).

To clarify the role of phosphorylation of Aβ and AβN3pE formation for the clinicopathological stage of the disease and for the pattern of Aβ aggregate types, we studied brains from 21 patients with Alzheimer’s disease, 33 with pathologically preclinical Alzheimer’s disease, and 20 non-demented control cases immunohistochemically and biochemically.

Materials and methods

Neuropathology

For morphological analysis, autopsy brains from 21 patients with Alzheimer’s disease, 20 control cases without any Aβ-pathology and 33 cases with pathologically preclinical Alzheimer’s disease were used (Tables 1 and 2). None of the investigated cases had a known familial background for Alzheimer’s disease. After autopsy, brains were fixed in a 4% aqueous solution of formaldehyde. Following fixation the medial temporal lobe and tissue from the occipital cortex containing the primary visual field were embedded in paraffin. Further medial temporal lobe tissue of the cases listed in Table 1 was embedded in polyethylene glycol. Paraffin sections were cut at 12 μm, polyethylene glycol sections at 100 μm. Histopathological diagnosis of Alzheimer’s disease was performed by analysing Gallyas, Campbell-Switzer, anti-abnormal tau-protein (anti-PHF-τ) and anti-Aβ17–24 stained sections of the medial temporal lobe and the occipital cortex (Supplementary Table 1). Braak neurofibrillary tangle staging and the assignment of Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) scores for neuritic plaque density were performed on the basis of the Gallyas-stained and anti-PHF-τ-stained sections (Braak and Braak, 1991; Mirra et al., 1991; Braak et al., 2006; Alafuzoff et al., 2008). The distribution of amyloid plaques in the medial temporal lobe (Aβ–medial temporal lobe phase) had been obtained according to previously published criteria (Thal et al., 2000) and represents the distribution of Aβ plaques in the human brain as a semi-quantitative parameter for the overall severity of Aβ plaque pathology (Thal et al., 2002). Aβ–medial temporal lobe phase, Braak–neurofibrillary tangle stages and CERAD scores for neuritic plaques were used to determine the degree of Alzheimer’s disease pathology according to recently published guidelines (Hyman et al., 2012).

Table 1

Cases used for histological and immunohistochemical analysis

Case number Age Gender Neuropathological diagnosis CDR score Aβ-MTL phase Braak-NFT stage CERAD-plaque score NIA-AA AD degree Biochemical-Aβ stage analogue for plaques 
A1 62 Control Not AD 
A2 62 Control Not AD 
A3 66 Control Not AD 
A4 61 Control Not AD 
A5 69 Control Not AD 
A6 66 Control Not AD 
A7 72 Control Not AD 
A8 66 Control Not AD 
A9 60 Control Not AD 
A10 74 Control Not AD 
A11 71 p-preAD Low 
A12 64 p-preAD Low 
A13 83 p-preAD, brain infarction Low 
A14 72 p-preAD Low 
A15 71 p-preAD, brain infarction Low 
A16 84 p-preAD, brain infarction Intermediate 
A17 87 p-preAD Intermediate 
A18 83 p-preAD Intermediate 
A19 63 p-preAD, brain infarction Intermediate 
A20 85 p-preAD Intermediate 
A21 66 p-preAD n.d. Low 
A22 86 p-preAD, VD 0.5 Low 
A23 88 p-preAD, AGD Low 
A24 78 AD Intermediate 
A25 68 AD High 
A26 82 AD Intermediate 
A27 89 AD Intermediate 
A28 87 AD Intermediate 
A29 83 AD Intermediate 
A30 81 AD Intermediate 
A31 89 AD High 
A32 78 AD High 
A33 83 AD High 
A34 86 AD, AGD High 
Case number Age Gender Neuropathological diagnosis CDR score Aβ-MTL phase Braak-NFT stage CERAD-plaque score NIA-AA AD degree Biochemical-Aβ stage analogue for plaques 
A1 62 Control Not AD 
A2 62 Control Not AD 
A3 66 Control Not AD 
A4 61 Control Not AD 
A5 69 Control Not AD 
A6 66 Control Not AD 
A7 72 Control Not AD 
A8 66 Control Not AD 
A9 60 Control Not AD 
A10 74 Control Not AD 
A11 71 p-preAD Low 
A12 64 p-preAD Low 
A13 83 p-preAD, brain infarction Low 
A14 72 p-preAD Low 
A15 71 p-preAD, brain infarction Low 
A16 84 p-preAD, brain infarction Intermediate 
A17 87 p-preAD Intermediate 
A18 83 p-preAD Intermediate 
A19 63 p-preAD, brain infarction Intermediate 
A20 85 p-preAD Intermediate 
A21 66 p-preAD n.d. Low 
A22 86 p-preAD, VD 0.5 Low 
A23 88 p-preAD, AGD Low 
A24 78 AD Intermediate 
A25 68 AD High 
A26 82 AD Intermediate 
A27 89 AD Intermediate 
A28 87 AD Intermediate 
A29 83 AD Intermediate 
A30 81 AD Intermediate 
A31 89 AD High 
A32 78 AD High 
A33 83 AD High 
A34 86 AD, AGD High 

Age in years. Clinical dementia rating (CDR) scores (Morris, 1993), Aβ-MTL phase (Thal et al., 2000), Braak–neurofibrillary tangle stage (Braak et al., 2006), CERAD score for neuritic plaque density (Mirra et al., 1991), and the degree of Alzheimer’s disease pathology (Hyman et al., 2012) were determined as previously published and recommended. The biochemical Aβ stage was determined as depicted in Fig. 6. M = male; F = female, (control) non-demented control; AD = Alzheimer’s disease; AGD = argyrophilic grain disease; ALS = amyotrophic lateral sclerosis; CBD = corticobasal degeneration; FTLD-TDP = frontotemporal lobar degeneration with TDP43-pathology; MCI (AD) = mild cognitive impairment with predominant Alzheimer’s disease pathology; MTL = medial temporal lobe; n.d. = not done; NIA-AA AD degree = Degree of Alzheimer’s disease pathology (Hyman et al., 2012); NFT = neurofibrillary tangle; NMO = neuromyelitis optica; p-preAD = pathologically diagnosed preclinical Alzheimer’s disease; VD = vascular dementia.

Table 2

Cases used for histological, immunohistochemical and for biochemical analysis from frozen neocortex samples

Case number Age Gender Neuropathological diagnosis CDR Score Aβ-MTL phase Braak-NFT stage CERAD- plaque score NIA-AA – AD degree Biochemical-Aβ stage Biochemical-Aβ stage analogue for plaques 
B1 60 Control Not AD 
B2 35 Limbic encephalitis Not AD 
B3 45 Control Not AD 
B4 58 Control Not AD 
B5 66 Control Not AD 
B6 69 Control Not AD 
B7 71 Control Not AD 
B8 46 Control Not AD 
B9 59 Control n.d. Not AD 
B10 57 Control Not AD 
B11 53 p-preAD Low 
B12 72 p-preAD, NMO Low 
B13 78 p-preAD, VD, CBD Low 
B14 73 p-preAD Low 
B15 72 p-preAD Low 
B16 73 p-preAD Low 
B17 68 p-preAD Low 
B18 64 p-preAD, Brain infarction n.d. Low 
B19 82 p-preAD, metastatic lung carcinoma, microinfarcts n.d. Low 
B20 68 p-preAD Low 
B21 74 p-preAD Low 
B22 67 p-preAD Low 
B23 77 p-preAD, VD Low 
B24 73 p-preAD n.d. Low 
B25 84 p-preAD Low 
B26 77 p-preAD Low 
B27 78 p-preAD Low 
B28 71 p-preAD Low 
B29 71 p-preAD Low 
B30 74 p-preAD Intermediate 
B31 91 AD Intermediate n.d. 
B32 79 AD n.d. Intermediate 
B33 84 AD, AGD, ALS, VD Intermediate 
B34 75 MCI (AD) 0.5 Intermediate 
B35 78 AD Intermediate 
B36 72 AD Intermediate 
B37 83 AD Intermediate 
B38 64 AD n.d. High 
B39 62 AD High 
B40 84 AD High 
Case number Age Gender Neuropathological diagnosis CDR Score Aβ-MTL phase Braak-NFT stage CERAD- plaque score NIA-AA – AD degree Biochemical-Aβ stage Biochemical-Aβ stage analogue for plaques 
B1 60 Control Not AD 
B2 35 Limbic encephalitis Not AD 
B3 45 Control Not AD 
B4 58 Control Not AD 
B5 66 Control Not AD 
B6 69 Control Not AD 
B7 71 Control Not AD 
B8 46 Control Not AD 
B9 59 Control n.d. Not AD 
B10 57 Control Not AD 
B11 53 p-preAD Low 
B12 72 p-preAD, NMO Low 
B13 78 p-preAD, VD, CBD Low 
B14 73 p-preAD Low 
B15 72 p-preAD Low 
B16 73 p-preAD Low 
B17 68 p-preAD Low 
B18 64 p-preAD, Brain infarction n.d. Low 
B19 82 p-preAD, metastatic lung carcinoma, microinfarcts n.d. Low 
B20 68 p-preAD Low 
B21 74 p-preAD Low 
B22 67 p-preAD Low 
B23 77 p-preAD, VD Low 
B24 73 p-preAD n.d. Low 
B25 84 p-preAD Low 
B26 77 p-preAD Low 
B27 78 p-preAD Low 
B28 71 p-preAD Low 
B29 71 p-preAD Low 
B30 74 p-preAD Intermediate 
B31 91 AD Intermediate n.d. 
B32 79 AD n.d. Intermediate 
B33 84 AD, AGD, ALS, VD Intermediate 
B34 75 MCI (AD) 0.5 Intermediate 
B35 78 AD Intermediate 
B36 72 AD Intermediate 
B37 83 AD Intermediate 
B38 64 AD n.d. High 
B39 62 AD High 
B40 84 AD High 

Age in years. Clinical dementia rating (CDR) scores (Morris, 1993), Aβ–medial temporal lobe phase (Thal et al., 2000), Braak–neurofibrillary tangle stage (Braak et al., 2006), CERAD score for neuritic plaque density (Mirra et al., 1991), and the degree of Alzheimer’s disease pathology (Hyman et al., 2012) were determined as previously published and recommended. The biochemical Aβ stage was determined as depicted in Fig. 6. M = male; F = female, (control) non-demented control; AD = Alzheimer’s disease; AGD = argyrophilic grain disease; ALS = amyotrophic lateral sclerosis; CBD = corticobasal degeneration; FTLD-TDP = frontotemporal lobar degeneration with TDP43-pathology; MCI (AD) = mild cognitive impairment with predominant Alzheimer’s disease pathology; MTL = medial temporal lobe; n.d. = not done; NIA-AA AD degree = Degree of Alzheimer's disease pathology (Hyman et al., 2012); NFT = neurofibrillary tangle; NMO = neuromyelitis optica; p-preAD = pathologically diagnosed preclinical Alzheimer’s disease; VD = vascular dementia.

The cases had usually been examined 1 to 4 weeks before death by different clinicians according to standardized protocols. The protocols included the assessment of cognitive function and recorded the ability to care for and dress oneself, eating habits, bladder and bowel continence, speech patterns, writing and reading, short-term and long-term memory, and orientation within the hospital setting. In the event that a Clinical Dementia Rating score could not be obtained because of missing clinical data, this is noted in Table 1. These data were used to retrospectively assess Clinical Dementia Rating scores for each patient (Morris et al., 1989). The diagnosis of symptomatic Alzheimer’s disease including Alzheimer’s disease-related mild cognitive impairment was considered for all individuals with a Clinical Dementia Rating score ≥0.5, which exhibited either an intermediate or high degree of Alzheimer’s disease pathology according to the National Institute of Aging Alzheimer Association (NIA-AA) guidelines for the neuropathological diagnosis of Alzheimer’s disease (Hyman et al., 2012). Controls were defined by the absence of any Aβ plaques. They either had no neurofibrillary tangles or not more than Braak–neurofibrillary tangle stage I. Non-demented cases with Aβ plaques, i.e. having low or intermediate degrees of Alzheimer’s disease pathology were categorized as cases with pathologically preclinical Alzheimer’s disease.

For biochemical analysis we used fresh-frozen occipital and temporal lobe tissue from 10 patients with Alzheimer’s disease, 20 patients with pathologically preclinical Alzheimer’s disease and 10 control subjects (Table 2).

The human brain tissue used in this study originated from the Brain Bank of the Laboratory of Neuropathology at the University of Ulm (Germany). This brain bank collects brain tissue in accordance with German legal regulations. The project was approved by the ethics committee of the University of Ulm.

Immunohistochemistry

Morphological and immunohistochemical analyses were carried out on cases shown in Table 1 and 2 (n = 73; Case B25 was not included because only frozen tissue was available). Paraffin sections from the human medial lobe and the occipital cortex were stained with anti-Aβ17–24, anti-Aβ42, anti-AβN3pE, and anti-phosphorylated Aβ (Kim et al., 1988; Saido et al., 1995; Yamaguchi et al., 1998; Kumar et al., 2011) (Supplementary Table 1). The primary antibodies were detected with biotinylated anti-mouse and anti-rabbit IgG secondary antibodies and visualized with avidin-biotin-complex (ABC-Kit, Vector Laboratories) and diaminobenzidine-HCl (DAB). The sections were counterstained with haematoxylin. Positive and negative controls were performed.

Double-label immunofluorescence was performed to demonstrate colocalization of Aβ with AβN3pE and phosphorylated Aβ in a given plaque. Anti-Aβ17–24 and anti-AβN3pE, anti-Aβ42 (IBL, polycloncal) (Supplementary Table 1) and anti-phosphorylated Aβ (monoclonal) as well as anti-AβN3pE and anti-phosphorylated Aβ (monoclonal) were combined. Polyclonal rabbit antibodies were detected with Cy2 or Cy3-labelled secondary antibodies against rabbit IgG. Likewise, monoclonal mouse antibodies were visualized with Cy2 or Cy3-labelled secondary antibodies against mouse IgG (Dianova).

Quantification of amyloid-β load

Aβ load was determined as the percentage of the area in the temporal neocortex (Brodmann area 36) covered by Aβ plaques detected with anti-Aβ17–24. Morphometry for Aβ load determination was performed using ImageJ image processing and analysis software (National Institutes of Health). For plaque measurements the area of the morphologically identified plaques was interactively delineated with a cursor and then measured by using the ImageJ software package (National Institutes of Health). The areas of all plaques in a given cortical region were added up. The area of the respective cortex areas was likewise measured by interactive delineation with a cursor. Accordingly, the AβN3pE load was determined as the percentage of the temporal neocortex area covered by anti-AβN3pE-positive plaques and the phosphorylated Aβ load by that of anti-phosphorylated Aβ-positive plaques.

Preparation of native human brain lysates

Biochemical analysis was carried out from cases shown in Table 2. Protein extraction from fresh frozen brain (0.04 g) was carried out in 2 ml Tris-buffered saline containing a protease and phosphatase inhibitor-cocktail (Complete and PhosSTOP, Roche). The tissue was homogenized with Micropestle (Eppendorf) before sonication. The homogenate was centrifuged for 30 min at 14 000g at 4°C. The supernatant with the soluble and dispersible fraction not separated from one another was retained. The pellet containing the membrane-associated and the solid plaque-associated fraction was resuspended in 2% SDS (Fig. 1). Ultracentrifugation of the supernatant at 175 000g was used to separate the soluble, i.e. the supernatant after ultracentrifugation, from the dispersible fraction, i.e. the resulting pellet (Fig. 1). The pellet of the dispersible fraction was resuspended in TBS and stored at −80°C until further use. After separation from the soluble and the dispersible fraction, the SDS-resuspended pellet was centrifuged at 14 000g, and the supernatant was kept as membrane-associated SDS fraction (Fig. 1). The pellet was further dissolved in 70% formic acid and the homogenate was lyophilized by centrifuging in the vacuum centrifuge (Vacufuge; Eppendorf) and reconstituted in 100 µl of 2× lithium dodecyl sulphate sample buffer (Invitrogen) before heating at 70°C for 5 min. The resulting sample was considered as plaque-associated, formic acid-soluble fraction (Mc Donald et al., 2010). The total protein amounts of soluble, dispersible, and membrane-associated fractions were determined using BCA Protein Assay (Bio-Rad).

Figure 1

Schematic representation of the biochemical fractionation of brain tissue homogenates into soluble, dispersible, membrane-associated SDS-soluble, and plaque-associated (formic acid-soluble) fraction. The dispersible fraction also contains microsomes. Isolation of dispersible oligomers, protofibrils, and fibrils by immunoprecipitation with oligomer or protofibril/fibril-specific antibodies is necessary as previously shown (Rijal Upadhaya et al., 2012a, b).

Figure 1

Schematic representation of the biochemical fractionation of brain tissue homogenates into soluble, dispersible, membrane-associated SDS-soluble, and plaque-associated (formic acid-soluble) fraction. The dispersible fraction also contains microsomes. Isolation of dispersible oligomers, protofibrils, and fibrils by immunoprecipitation with oligomer or protofibril/fibril-specific antibodies is necessary as previously shown (Rijal Upadhaya et al., 2012a, b).

Immunoprecipitation

For immunoprecipitation, 200 μl of the native soluble and dispersible fractions from the brain lysates were incubated with 1 μl A11 antibodies against non-fibrillar oligomers, B10AP antibody fragments for precipitation of protofibrils and fibrils, anti-AβN3pE or anti-phosphorylated Aβ at 4°C for 4 h as previously described (Rijal Upadhaya et al., 2012a) (Supplementary Table 1). Protein G Microbeads (50 μl; Miltenyi Biotec) were added to the mixture and incubated overnight at 4°C. The mixture was then passed through the μ Columns which separate the microbeads by retaining them in the column, while the rest of the lysate flows through. After one mild washing step with TBS at pH 7.4 the microbead-bound proteins were eluted with 95°C heated lithium dodecyl sulphate sample buffer (Invitrogen). To verify specific precipitation of non-fibrillar oligomers with A11 and protofibrils and fibrils with B10AP and to exclude contamination with membrane-coated microsomes, precipitates were analysed for non-fibrillar oligomeric or protofibrillar/fibrillar protein structure by transmission electron microscopy as previously published (Rijal Upadhaya et al., 2012b).

Western blot analysis

The four fractions (soluble, dispersible, membrane-associated and plaque-associated) as well as immunoprecipitation eluates were analysed by SDS-PAGE and subsequent western blot analysis with anti-Aβ1–17, anti-phosphorylated Aβ and anti-AβN3pE antibodies (Supplementary Table 1). Aβ40 and Aβ42 were detected with C-terminus specific antibodies (Supplementary Table 1) after precipitation of AβN3pE and phosphorylated Aβ to clarify whether these post-translational modifications occur in both Aβ peptides, Aβ40 and Aβ42. Blots were developed with an ECL detection system (Supersignal Pico Western system, ThermoScientific-Pierce) and illuminated in ECL Hyperfilm (GE Healthcare).

Because Aβ aggregates readily dissociate in the presence of SDS-containing buffers into monomers and small oligomers, such as dimers, trimers, or Aβ*56 (Rijal Upadhaya et al., 2012b; Watt et al., 2013), we analysed differences among the monomer bands that indicate changes in the protein levels of precipitated Aβ aggregates densitometrically using ImageJ software (National Institutes of Health). This method allows a semi-quantitative assessment of Aβ as previously described in detail (Rijal Upadhaya et al., 2012a).

Statistical analysis

SPSS-Statistics 19.0 (SPSS) software was used to calculate statistical tests. One-way ANOVA was used to compare densitometric data received from western blot quantification and Aβ loads among cases with Alzheimer’s disease, pathologically preclinical Alzheimer’s disease and control cases. The Games-Howell post hoc test was used to correct for multiple testing. Binary logistic regression analysis controlled for age and gender was used to test whether dementia was associated with Aβ, AβN3pE, and phosphorylated Aβ loads. Partial correlation analysis was performed for Aβ–medial temporal lobe phase, Braak–neurofibrillary tangle stage, CERAD score for neuritic plaques, and the biochemical stages of Aβ aggregation and accumulation (biochemical–Aβ stages) as determined in this study. Likewise, partial correlation analysis controlled for age and gender was also carried out among Aβ–medial temporal lobe phase, Braak–neurofibrillary tangle stage, CERAD score for neuritic plaques, and a modified biochemical-Aβ stage represented by the detection of Aβ, AβN3pE, and phosphorylated Aβ in plaques. Fisher’s exact test with subsequent trend test was performed to clarify whether the biochemical-Aβ stages and biochemical-Aβ stage analogues for plaques increase hierarchically with progression of the clinical stage of Alzheimer’s disease from non-Alzheimer’s disease to pathologically preclinical Alzheimer’s disease and finally to symptomatic Alzheimer’s disease.

Results

Biochemical detection of soluble, dispersible, membrane-associated and plaque-associated amyloid-β

SDS-PAGE and western blot analysis with anti-Aβ1–17 demonstrated Aβ in the soluble, dispersible, membrane-associated and plaque-associated fraction in neocortex homogenates from cases with Alzheimer’s disease and cases with pathologically preclinical Alzheimer’s disease. As previously shown Aβ dimers, trimers etc. represent SDS treatment-related dissociation products of larger Aβ aggregates (Rijal Upadhaya et al., 2012b; Watt et al., 2013). Therefore, we did not consider them for a separate analysis in this study. The semi-quantitative assessment of the monomer band density has been demonstrated previously to correlate with the amount of Aβ aggregates (Rijal Upadhaya et al., 2012a) and was used for the semi-quantitative assessment of Aβ aggregates in a given biochemical fraction or being precipitated with A11 and B10AP. Control cases showed no detectable Aβ (Fig. 2A and Supplementary Fig. 1A). Semi-quantitatively, cases with Alzheimer’s disease exhibited significantly more Aβ-positive material than cases with pathologically preclinical Alzheimer’s disease and non-Alzheimer’s disease control cases in all fractions. Pathologically preclinical Alzheimer’s disease cases showed more Aβ-positive material than non-Alzheimer’s disease controls (Fig. 2A and Supplementary Table 2A).

Figure 2

Biochemical detection of soluble, dispersible, membrane-associated and plaque-associated Aβ. (A) Denaturing SDS-PAGE analysis of soluble, dispersible, membrane-associated (SDS-soluble) and plaque-associated (formic acid-soluble) fractions of human brain homogenates. Aβ was detected with anti-Aβ1–17. Quantification revealed highest levels of soluble, dispersible, membrane-associated, and plaque-associated Aβ in Alzheimer’s disease cases whereas pathologically preclinical Alzheimer’s disease cases (p-preAD) exhibited lower Aβ levels than Alzheimer’s disease cases but higher levels than non-Alzheimer’s disease controls, which lack detectable amounts of Aβ aggregates. (B) Cases with symptomatic Alzheimer’s disease exhibited higher levels of soluble, dispersible, membrane-associated and plaque-associated (formic acid soluble) AβN3pE than pathologically preclinical Alzheimer’s disease and control cases. Significant differences occurred between pathologically preclinical Alzheimer’s disease and control cases only in the dispersible and membrane-associated fraction. Soluble, dispersible and plaque-associated AβN3pE was nearly absent in pathologically preclinical Alzheimer’s disease cases. SDS-soluble membrane-associated and plaque-associated AβN3pE was observed in some pathologically preclinical Alzheimer’s disease cases whereas other pathologically preclinical Alzheimer’s disease cases did not exhibit AβN3pE distinguishing biochemical-Aβ stages 1 and 2. An additional dimer band was visible in the plaque-associated fraction. (C) Phosphorylated Aβ was not found in the soluble fraction. In the dispersible, membrane-associated and plaque-associated fractions phosphorylated Aβ monomer bands (4 kDa) were visible in cases with Alzheimer’s disease exhibiting biochemical-Aβ stage 3 whereas most pathologically preclinical Alzheimer’s disease cases did not exhibit phosphorylated Aβ monomer bands. No significant quantitative differences were observed after western blot analysis. The 8 kDa band stained with anti-phosphorylated Aβ was considered unspecific and not relevant for Alzheimer’s disease because it was seen in non-Alzheimer’s disease controls, pathologically preclinical Alzheimer’s disease and symptomatic Alzheimer’s disease cases in similar intensity. Case numbers according to Supplementary Table 1B are provided. Statistical analysis was performed by ANOVA with Games-Howell post hoc test: *P < 0.05; **P < 0.01; ***P < 0.001 (Alzheimer’s disease, n = 10; pathologically preclinical Alzheimer’s disease, n = 20; control, n = 10; Supplementary Table 2A–C). AD = Alzheimer’s disease; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 2

Biochemical detection of soluble, dispersible, membrane-associated and plaque-associated Aβ. (A) Denaturing SDS-PAGE analysis of soluble, dispersible, membrane-associated (SDS-soluble) and plaque-associated (formic acid-soluble) fractions of human brain homogenates. Aβ was detected with anti-Aβ1–17. Quantification revealed highest levels of soluble, dispersible, membrane-associated, and plaque-associated Aβ in Alzheimer’s disease cases whereas pathologically preclinical Alzheimer’s disease cases (p-preAD) exhibited lower Aβ levels than Alzheimer’s disease cases but higher levels than non-Alzheimer’s disease controls, which lack detectable amounts of Aβ aggregates. (B) Cases with symptomatic Alzheimer’s disease exhibited higher levels of soluble, dispersible, membrane-associated and plaque-associated (formic acid soluble) AβN3pE than pathologically preclinical Alzheimer’s disease and control cases. Significant differences occurred between pathologically preclinical Alzheimer’s disease and control cases only in the dispersible and membrane-associated fraction. Soluble, dispersible and plaque-associated AβN3pE was nearly absent in pathologically preclinical Alzheimer’s disease cases. SDS-soluble membrane-associated and plaque-associated AβN3pE was observed in some pathologically preclinical Alzheimer’s disease cases whereas other pathologically preclinical Alzheimer’s disease cases did not exhibit AβN3pE distinguishing biochemical-Aβ stages 1 and 2. An additional dimer band was visible in the plaque-associated fraction. (C) Phosphorylated Aβ was not found in the soluble fraction. In the dispersible, membrane-associated and plaque-associated fractions phosphorylated Aβ monomer bands (4 kDa) were visible in cases with Alzheimer’s disease exhibiting biochemical-Aβ stage 3 whereas most pathologically preclinical Alzheimer’s disease cases did not exhibit phosphorylated Aβ monomer bands. No significant quantitative differences were observed after western blot analysis. The 8 kDa band stained with anti-phosphorylated Aβ was considered unspecific and not relevant for Alzheimer’s disease because it was seen in non-Alzheimer’s disease controls, pathologically preclinical Alzheimer’s disease and symptomatic Alzheimer’s disease cases in similar intensity. Case numbers according to Supplementary Table 1B are provided. Statistical analysis was performed by ANOVA with Games-Howell post hoc test: *P < 0.05; **P < 0.01; ***P < 0.001 (Alzheimer’s disease, n = 10; pathologically preclinical Alzheimer’s disease, n = 20; control, n = 10; Supplementary Table 2A–C). AD = Alzheimer’s disease; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

N3pE was observed in the soluble, dispersible, membrane-associated and plaque-associated fraction of symptomatic Alzheimer’s disease brain homogenates. Cases with pathologically preclinical Alzheimer’s disease exhibited no or only small amounts of soluble and dispersible AβN3pE. SDS-soluble AβN3pE in the membrane-associated fraction and/or plaque-associated AβN3pE was detected in 13 of 20 cases with pathologically preclinical Alzheimer’s disease by western blotting. Some cases with pathologically preclinical Alzheimer’s disease, thereby, exhibited similar amounts of SDS-soluble AβN3pE as symptomatic cases with Alzheimer’s disease (Fig. 2B and Supplementary Fig. 1B). Semi-quantitative comparison of monomer bands from control, pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases revealed that Alzheimer’s disease cases exhibited significantly more AβN3pE in all four fractions than controls and cases with pathologically preclinical Alzheimer’s disease. No significant differences in the levels of soluble and plaque-associated Aβ were observed between control and pathologically preclinical Alzheimer’s disease cases whereas such differences were seen in the dispersible and membrane-associated fraction (Fig. 2B and Supplementary Table 2A).

Phosphorylated Aβ was found in the dispersible, membrane-associated, and plaque-associated fraction of Alzheimer’s disease brain homogenates. Soluble phosphorylated Aβ was not observed. Cases with pathologically preclinical Alzheimer’s disease did not exhibit detectable levels of phosphorylated Aβ in the membrane-associated and plaque-associated fractions. Only isolated pathologically preclinical Alzheimer’s disease cases showed few phosphorylated Aβ in the dispersible fraction. Phosphorylated Aβ was not detected in control cases. A second ∼8 kDa band was also detected with the phosphorylated Aβ antibody. This band presented with similar intensity in soluble, dispersible, and membrane-associated fractions of Alzheimer’s disease, pathologically preclinical Alzheimer’s disease, and control cases as well as in ischiadic nerve samples (Supplementary Fig. 3). Therefore, we did not interpret this band as a dimer-specific band but as unspecific co-staining without any relevance for Alzheimer’s disease because a similar ∼8 kDa band was not observed in the formic acid-soluble, plaque-associated fraction although dimers were seen (Fig. 2C and Supplementary Fig. 1C). Significant differences in the semi-quantitative assessment of the phosphorylated Aβ monomer bands detected by western blotting were not observed (Fig. 2C and Supplementary Table 2A).

To clarify whether the occurrence of AβN3pE and phosphorylated Aβ in Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases was related to a specific accumulation in Aβ oligomers, protofibrils, and/or fibrils we performed immunoprecipitation and western blotting. Non-fibrillar oligomers were precipitated from soluble and dispersible fractions with A11 antibodies whereas protofibrils and fibrils were precipitated with B10AP antibody fragments. These precipitates contained Aβ aggregates as well as oligomeric, protofibrillar, or fibrillar aggregates composed of other proteins (Rijal Upadhaya et al., 2012b). The highest amounts of oligomeric and protofibrillar/fibrillar Aβ aggregates were found in precipitates of the dispersible fraction of Alzheimer’s disease cases. Cases with pathologically preclinical Alzheimer’s disease had detectable but lower levels of dispersible Aβ oligomers, protofibrils, and fibrils than Alzheimer’s disease cases. Non-Alzheimer’s disease controls did not contain measurable amounts of Aβ. The amounts of soluble Aβ oligomers, protofibrils, and fibrils did not vary significantly between Alzheimer’s disease and pathologically preclinical Alzheimer’s disease but were higher in cases with Alzheimer’s disease and cases with pathologically preclinical Alzheimer’s disease than in controls (Fig. 3A, Supplementary Fig. 2A and Supplementary Table 2B).

Figure 3

(A) Analysis of B10AP immunoprecipitated protofibrils and fibrils and A11 immunoprecipitated non-fibrillar oligomers revealed highest levels of these in the soluble and dispersible fractions of Alzheimer’s disease cases. Pathologically preclinical Alzheimer’s disease cases exhibited fewer Aβ oligomers, protofibrils, and fibrils than Alzheimer’s disease cases but more than non-Alzheimer’s disease controls, which did not show Aβ oligomers, protofibrils or fibrils. (B) In the precipitated protofibrils, fibrils, and oligomers, AβN3pE was found only in the dispersible but not in the soluble fraction. The levels of AβN3pE oligomers, protofibrils and fibrils were not significantly different between Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases but higher than in non-Alzheimer’s disease controls. Only a subset of pathologically preclinical Alzheimer’s disease cases (p-preAD) exhibited AβN3pE indicative of biochemical-Aβ stage 2 whereas other pathologically preclinical Alzheimer’s disease cases with anti-Aβ1–17 positive Aβ aggregates did not exhibit anti-AβN3pE-positive material representing biochemical-Aβ stage 1. (C) Dispersible phosphorylated Aβ oligomers, protofibrils, and fibrils were nearly restricted to Alzheimer’s disease cases whereas non-Alzheimer’s disease controls and pathologically preclinical Alzheimer’s disease exhibited nearly negligible amounts. Phosphorylated Aβ in patients with Alzeimer’s disease represented the third stage of the biochemical development of Aβ aggregates throughout the pathogenesis of Alzheimer’s disease (biochemical-Aβ stage 3). The 8 kDa band stained with anti-phosphorylated Aβ was considered unspecific and not relevant for Alzheimer’s disease because it was seen in non-Alzheimer’s disease controls, pathologically preclinical Alzheimer’s disease and symptomatic Alzheimer’s disease cases in similar intensity. No phosphorylated Aβ containing oligomers, protofibrils and fibrils were found in the soluble fraction. Case numbers according to Supplementary Table 1B are provided. Statistical analysis was performed by ANOVA with Games-Howell post hoc test: *P < 0.05; **P < 0.01; ***P < 0.001 (Alzheimer’s disease, n = 10; pathologically preclinical Alzheimer’s disease, n = 20; control, n = 10; Supplementary Table 2A–C). AD = Alzheimer’s disease; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 3

(A) Analysis of B10AP immunoprecipitated protofibrils and fibrils and A11 immunoprecipitated non-fibrillar oligomers revealed highest levels of these in the soluble and dispersible fractions of Alzheimer’s disease cases. Pathologically preclinical Alzheimer’s disease cases exhibited fewer Aβ oligomers, protofibrils, and fibrils than Alzheimer’s disease cases but more than non-Alzheimer’s disease controls, which did not show Aβ oligomers, protofibrils or fibrils. (B) In the precipitated protofibrils, fibrils, and oligomers, AβN3pE was found only in the dispersible but not in the soluble fraction. The levels of AβN3pE oligomers, protofibrils and fibrils were not significantly different between Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases but higher than in non-Alzheimer’s disease controls. Only a subset of pathologically preclinical Alzheimer’s disease cases (p-preAD) exhibited AβN3pE indicative of biochemical-Aβ stage 2 whereas other pathologically preclinical Alzheimer’s disease cases with anti-Aβ1–17 positive Aβ aggregates did not exhibit anti-AβN3pE-positive material representing biochemical-Aβ stage 1. (C) Dispersible phosphorylated Aβ oligomers, protofibrils, and fibrils were nearly restricted to Alzheimer’s disease cases whereas non-Alzheimer’s disease controls and pathologically preclinical Alzheimer’s disease exhibited nearly negligible amounts. Phosphorylated Aβ in patients with Alzeimer’s disease represented the third stage of the biochemical development of Aβ aggregates throughout the pathogenesis of Alzheimer’s disease (biochemical-Aβ stage 3). The 8 kDa band stained with anti-phosphorylated Aβ was considered unspecific and not relevant for Alzheimer’s disease because it was seen in non-Alzheimer’s disease controls, pathologically preclinical Alzheimer’s disease and symptomatic Alzheimer’s disease cases in similar intensity. No phosphorylated Aβ containing oligomers, protofibrils and fibrils were found in the soluble fraction. Case numbers according to Supplementary Table 1B are provided. Statistical analysis was performed by ANOVA with Games-Howell post hoc test: *P < 0.05; **P < 0.01; ***P < 0.001 (Alzheimer’s disease, n = 10; pathologically preclinical Alzheimer’s disease, n = 20; control, n = 10; Supplementary Table 2A–C). AD = Alzheimer’s disease; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

N3pE was not detected in soluble oligomers, protofibrils, and fibrils precipitated with A11 and B10AP but in dispersible oligomers, protofibrils, and fibrils of Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases. Non-Alzheimer’s disease controls did not display such material. Although dispersible AβN3pE oligomers, protofibrils, and fibrils appeared to occur in higher levels in Alzheimer’s disease neocortex than in pathologically preclinical Alzheimer’s disease these differences were not significant (Fig. 3B, Supplementary Fig. 2B and Supplementary Table 2C).

Dispersible phosphorylated Aβ-containing oligomers, protofibrils, and fibrils were found in higher amounts in Alzheimer’s disease cases compared to non-Alzheimer’s disease controls and pathologically preclinical Alzheimer’s disease cases. The amount of dispersible phosphorylated Aβ oligomers, protofibrils and fibrils did not vary significantly between non-Alzheimer’s disease controls and pathologically preclinical Alzheimer’s disease cases. Only a few pathologically preclinical Alzheimer’s disease cases exhibited small amounts of phosphorylated Aβ-containing protofibrils. Soluble phosphorylated Aβ in precipitated oligomers, protofibrils and fibrils was not observed. An 8-kDa band stained with anti-phosphorylated Aβ was considered unspecific and not relevant for Alzheimer’s disease because it was seen in similar intensity in non-Alzheimer’s disease controls, pathologically preclinical Alzheimer’s disease, symptomatic Alzheimer’s disease cases (Fig. 3C, Supplementary Fig. 2C and Supplementary Table 2C) and in western blots of peripheral nervous tissue of the ischiadic nerve (Supplementary Fig. 3). The fact that it was observed after immunoprecipitation with A11 and B10AP indicates a cross-reaction with components of non-Aβ protein complexes sharing A11 and B10AP conformation specific epitopes.

In summary, human Alzheimer’s disease brains can be distinguished from non-Alzheimer’s disease and pathologically preclinical Alzheimer’s disease brains by increasing amounts of soluble and dispersible Aβ oligomers, protofibrils, and fibrils whereby phosphorylation of Aβ at serine 8 was associated with dispersible Aβ oligomers, protofibrils, and fibrils in the Alzheimer’s disease neocortex. The biochemical composition of Aβ aggregates showed a hierarchical sequence in which Aβ, AβN3pE, and phosphorylated Aβ occurred in dispersible, membrane-associated and plaque-associated Aβ-aggregates. All 10 cases with Alzheimer’s disease and 14 of 20 cases with pathologically preclinical Alzheimer’s disease exhibited biochemically detectable Aβ. Six cases with pathologically preclinical Alzheimer’s disease and the 10 non-Alzheimer’s disease cases did not show biochemically detectable amounts of Aβ (Fig. 2A and 3A). Twelve pathologically preclinical Alzheimer’s disease and all 10 Alzheimer’s disease cases also showed anti-AβN3pE-positive material in the Aβ aggregates, suggesting a second stage in the development of Aβ aggregation. Phosphorylated Aβ was found only in 4 of 20 cases with pathologically preclinical Alzheimer’s disease, but in all 10 Alzheimer’s disease cases studied biochemically in a presumably third stage of this process. These three stages of the biochemical aggregation and accumulation are referred to here as biochemical-Aβ stages 1–3.

Immunoprecipitation of post-translational modified AβN3pE and phosphorylated Aβ with subsequent detection of the Aβ40 and Aβ42 C-terminus with C-terminus specific antibodies revealed that AβN3pE-40, AβN3pE-42, phosphorylated Aβ40, and phosphorylated Aβ42 can be detected in the human Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cortex with stronger signals for the Aβ42-C-terminus peptides (Supplementary Fig. 4A).

Immunohistochemical detection of amyloid-β, AβN3pE and phosphorylated amyloid-β in senile plaques

Immunohistochemical staining of brain tissues from all Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases exhibited Aβ plaques detectable with antibodies raised against Aβ17–24 and Aβ42. All cases with Alzheimer’s disease and 30 of 33 pathologically preclinical Alzheimer’s disease cases also showed immunopositivity for anti-AβN3pE. Eleven of the pathologically preclinical Alzheimer’s disease cases with anti-AβN3pE positive plaques and all Alzheimer’s disease cases also had phosphorylated Aβ positive plaques. This hierarchical sequence of plaque staining with anti-Aβ17–24, anti-AβN3pE, and anti-phosphorylated Aβ was identical with that seen for the biochemical detection of Aβ and its accumulation in the dispersible, membrane-associated and plaque-associated fractions of brain homogenates. This sequence of plaque staining is referred to as biochemical-Aβ stage analogue for plaques. However, 6 of 20 cases with pathologically preclinical Alzheimer’s disease cases with Aβ17–24-positive plaques (Table 2) did not exhibit significant amounts of biochemically detectable Aβ. In two further cases with AβN3pE-positive plaques Aβ was seen biochemically but no AβN3pE. Four of 16 cases with phosphorylated Aβ-positive plaques did not exhibit phosphorylated Aβ in the western blot and immunoprecipitation analysis.

Aβ plaques detected with antibodies raised against Aβ17–24 and Aβ42 were prevalent in all pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases (Fig. 4A–C). Alzheimer’s disease cases exhibited higher Aβ loads than pathologically preclinical Alzheimer’s disease cases. Non-Alzheimer’s disease controls had lower Aβ loads than in Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases (Fig. 5A and Supplementary Table 2D).

Figure 4

(A–C) Aβ plaques detected with anti-Aβ42 were found in both Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases. The biochemical-Aβ stage analogues for plaques were provided. (D–F) AβN3pE was found in pathologically preclinical Alzheimer’s disease cases of biochemical-Aβ stage analogue 2 and in Alzheimer’s disease cases. In the biochemical-Aβ stage analogue 1 case depicted in D no anti-AβN3pE-positive plaques were found. (G–I) Phosphorylated Aβ was absent in biochemical-Aβ stage analogues 1 and 2 pathologically preclinical Alzheimer’s disease cases (G and H) but prevalent in the biochemical-Aβ stage 3 case with Alzheimer’s disease (I). Calibration bar in H corresponds to 400 µm (valid for A–I). A, D and G: Case A15; B, E and H: Case A14; C, F and I: Case A30. AD = Alzheimer’s disease; B-Aβ-stage analog = biochemical-Aβ stage analogue; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 4

(A–C) Aβ plaques detected with anti-Aβ42 were found in both Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases. The biochemical-Aβ stage analogues for plaques were provided. (D–F) AβN3pE was found in pathologically preclinical Alzheimer’s disease cases of biochemical-Aβ stage analogue 2 and in Alzheimer’s disease cases. In the biochemical-Aβ stage analogue 1 case depicted in D no anti-AβN3pE-positive plaques were found. (G–I) Phosphorylated Aβ was absent in biochemical-Aβ stage analogues 1 and 2 pathologically preclinical Alzheimer’s disease cases (G and H) but prevalent in the biochemical-Aβ stage 3 case with Alzheimer’s disease (I). Calibration bar in H corresponds to 400 µm (valid for A–I). A, D and G: Case A15; B, E and H: Case A14; C, F and I: Case A30. AD = Alzheimer’s disease; B-Aβ-stage analog = biochemical-Aβ stage analogue; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 5

Aβ load, AβN3pE load, and phosphorylated Aβ load in Alzheimer’s disease, pathologically preclinical Alzheimer’s disease (p-preAD) and control cases. (A) The Aβ load increased gradually from control to pathologically preclinical Alzheimer’s disease and then to Alzheimer’s disease cases. (B) The AβN3pE load in pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases was higher than in non-Alzheimer’s disease cases. Significant differences in the AβN3pE load between pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases were not observed. (C) Alzheimer’s disease cases had significantly higher phosphorylated Aβ loads compared with control and pathologically preclinical Alzheimer’s disease cases. ANOVA with Games-Howell post hoc test: *P < 0.05; ***P < 0.001 (Supplementary Table 2E). AD = Alzheimer’s disease; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 5

Aβ load, AβN3pE load, and phosphorylated Aβ load in Alzheimer’s disease, pathologically preclinical Alzheimer’s disease (p-preAD) and control cases. (A) The Aβ load increased gradually from control to pathologically preclinical Alzheimer’s disease and then to Alzheimer’s disease cases. (B) The AβN3pE load in pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases was higher than in non-Alzheimer’s disease cases. Significant differences in the AβN3pE load between pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases were not observed. (C) Alzheimer’s disease cases had significantly higher phosphorylated Aβ loads compared with control and pathologically preclinical Alzheimer’s disease cases. ANOVA with Games-Howell post hoc test: *P < 0.05; ***P < 0.001 (Supplementary Table 2E). AD = Alzheimer’s disease; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

N3pE positive plaques were frequently observed in most pathologically preclinical Alzheimer’s disease cases and in all Alzheimer’s disease cases (Fig. 4D–F). All types of plaques exhibited AβN3pE. AβN3pE plaque loads were lower than total Aβ1–40/42 plaque loads. Alzheimer’s disease cases had higher AβN3pE plaque loads than cases with pathologically preclinical Alzheimer’s disease. Cases with pathologically preclinical Alzheimer’s disease exhibited higher AβN3pE plaque loads than non-Alzheimer’s disease controls (Fig. 5A and B and Supplementary Table 2D).

The phosphorylated Aβ plaque loads were lower than the Aβ and AβN3pE plaque loads. However, in Alzheimer’s disease cases the phosphorylated Aβ plaque load was higher than in pathologically preclinical Alzheimer’s disease. Non-Alzheimer’s disease controls exhibited no anti-phosphorylated Aβ-positive plaques whereas some cases with pathologically preclinical Alzheimer’s disease showed few phosphorylated Aβ-positive plaques. The phosphorylated Aβ plaque load in pathologically preclinical Alzheimer’s disease cases was slightly higher than in control cases (Figs 4G, H and 5C and Supplementary Table 2D). Single pathologically preclinical Alzheimer’s disease cases exhibiting high amounts of Aβ17–24 and AβN3pE-positive plaques did not exhibit phosphorylated Aβ within these plaques in consecutive sections (Supplementary Fig. 5).

Logistic regression analysis controlled for age and gender revealed a significant association of the Aβ load, AβN3pE load and the phosphorylated Aβ load with Alzheimer’s disease cases in comparison to cases with pathologically preclinical Alzheimer’s disease and non-Alzheimer’s disease control cases (P < 0.05; detailed statistical analysis see Supplementary Table 2E).

Double-label immunohistochemistry revealed that in Alzheimer’s disease cases most Aβ plaques also exhibit AβN3pE whereas phosphorylated Aβ was usually restricted to a subset of plaques, especially cored plaques (Supplementary Fig. 4B–J).

Correlations between the biochemical stages of amyloid-β aggregation and accumulation with the hallmark lesions of Alzheimer’s disease and its associations with dementia

The biochemical-Aβ stages correlated with the Aβ–medial temporal lobe phase (r = 0.79, P < 0.001), the Braak–neurofibrillary tangle-stage (r = 0.609, P = 0.001), and the CERAD score for neuritic plaques (r = 0.56, P = 0.002) as well as with the overall NIA-AA degree of Alzheimer’s disease pathology (r = 0.683, P < 0.001; detailed statistical analysis is shown in Supplementary Table 2F).

Likewise, the biochemical-Aβ stage analogue for plaques correlated with the Aβ–medial temporal lobe-phase (r = 0.834, P < 0.001), the Braak–neurofibrillary tangle-stage (r = 0.564, P = 0.002), the CERAD score for neuritic plaques (r = 0.429, P = 0.023), the overall NIA-AA degree of Alzheimer’s disease pathology (r = 0.76, P < 0.001; detailed statistical analysis is shown in Supplementary Table 2G) as well as with the biochemical-Aβ stages (r = 0.688, P < 0.001).

Using Fisher’s exact test with a subsequent trend test there was a significant association between the increasing clinical stage of Alzheimer’s disease from non-Alzheimer’s disease to pathologically preclinical Alzheimer’s disease and finally to symptomatic Alzheimer’s disease with the biochemical-Aβ stage and the biochemical-Aβ stage analogue for plaques (P < 0.001; detailed statistical analysis Supplementary Table 2H).

Discussion

The major findings of this study are: (i) the prevalence of AβN3pE and phosphorylated Aβ in dispersible, membrane-associated, and plaque-associated Aβ aggregates showed a hierarchical sequence of three stages, in which these post-translationally modified Aβ species occurred in Aβ aggregates: biochemical-Aβ stage 1 = aggregation of Aβ1–40/42 alone, biochemical-Aβ stage 2 = additional detection of AβN3pE, and biochemical-Aβ stage 3 = aggregation of Aβ1–40/42, AβN3pE–40/42, and phosphorylated Aβ40/42 (Fig. 6); (ii) the phosphorylation of Aβ at serine 8 and its aggregation in dispersible oligomers, protofibrils and fibrils was associated with symptomatic Alzheimer’s disease but not with pathologically preclinical Alzheimer’s disease and controls; (iii) the amounts of soluble and dispersible Aβ oligomers, protofibrils and fibrils increased with the development from non-Alzheimer’s disease to pathologically preclinical Alzheimer’s disease and then to Alzheimer’s disease; and (iv) AβN3pE and phosphorylated Aβ were not detectable in soluble oligomers, protofibrils and fibrils but in dispersible ones.

Figure 6

Associations between the diagnosis of Alzheimer’s disease, pathologically preclinical Alzheimer’s disease and control cases with the biochemical-Aβ stages of the biochemical composition of Aβ aggregates and with the Aβ-medial temporal lobe phases. ‘Non-Alzheimer’s disease’ is by definition the absence of Aβ plaques. Pathologically preclinical Alzheimer’s disease cases and cases with symptomatic Alzheimer’s disease can be distinguished by the distribution of Aβ plaque pathology in the brain as represented in the medial temporal lobe (Thal et al., 2000, 2002, 2013) but also by changes in the biochemical composition of soluble, dispersible, membrane-associated, and plaque-associated Aβ aggregates as represented by the biochemical-Aβ stages. These differences in the biochemical composition of Aβ aggregates between cases with pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases are indicated by the detection of phosphorylated Aβ in symptomatic Alzheimer’s disease cases and by the detection of AβN3pE in the soluble and dispersible fraction. The hierarchical sequence, in which Aβ1–40/42, AβN3pE, and phosphorylated Aβ occurred in the Aβ aggregates in the human brain, thereby, allowed the distinction of three biochemical-Aβ stages: biochemical-Aβ stage 1 was defined by the detection of anti-Aβ-positive Aβ aggregates in the absence of detectable amounts of AβN3pE and phosphorylated Aβ; biochemical-Aβ stage 2 was characterized by additional AβN3pE in the aggregates without detectable phosphorylated Aβ; biochemical-Aβ stage 3 represented Aβ aggregates in the brain exhibiting all three types of Aβ, i.e. Aβ1–40/42, AβN3pE, and phosphorylated Aβ. AD = Alzheimer’s disease; AβMTL phase = Aβ medial temporal lobe phase; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Figure 6

Associations between the diagnosis of Alzheimer’s disease, pathologically preclinical Alzheimer’s disease and control cases with the biochemical-Aβ stages of the biochemical composition of Aβ aggregates and with the Aβ-medial temporal lobe phases. ‘Non-Alzheimer’s disease’ is by definition the absence of Aβ plaques. Pathologically preclinical Alzheimer’s disease cases and cases with symptomatic Alzheimer’s disease can be distinguished by the distribution of Aβ plaque pathology in the brain as represented in the medial temporal lobe (Thal et al., 2000, 2002, 2013) but also by changes in the biochemical composition of soluble, dispersible, membrane-associated, and plaque-associated Aβ aggregates as represented by the biochemical-Aβ stages. These differences in the biochemical composition of Aβ aggregates between cases with pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases are indicated by the detection of phosphorylated Aβ in symptomatic Alzheimer’s disease cases and by the detection of AβN3pE in the soluble and dispersible fraction. The hierarchical sequence, in which Aβ1–40/42, AβN3pE, and phosphorylated Aβ occurred in the Aβ aggregates in the human brain, thereby, allowed the distinction of three biochemical-Aβ stages: biochemical-Aβ stage 1 was defined by the detection of anti-Aβ-positive Aβ aggregates in the absence of detectable amounts of AβN3pE and phosphorylated Aβ; biochemical-Aβ stage 2 was characterized by additional AβN3pE in the aggregates without detectable phosphorylated Aβ; biochemical-Aβ stage 3 represented Aβ aggregates in the brain exhibiting all three types of Aβ, i.e. Aβ1–40/42, AβN3pE, and phosphorylated Aβ. AD = Alzheimer’s disease; AβMTL phase = Aβ medial temporal lobe phase; B-Aβ-stage = biochemical-Aβ stage; pAβ = phosphorylated Aβ; p-preAD = pathologically (diagnosed) Alzheimer’s disease.

Dispersible, membrane-associated, and plaque-associated Aβ aggregates exhibited a hierarchical sequence, in which Aβ1-40/42, AβN3pE, and phosphorylated Aβ occurred in these aggregates. This sequence allowed the distinction of three biochemical stages of Aβ aggregation and accumulation (biochemical-Aβ stages). The first stage was characterized by the detection of Aβ1–40/42 in the absence of detectable amounts of AβN3pE and phosphorylated Aβ. Biochemical-Aβ stage 2 was characterized by the additional occurrence of AβN3pE-positive material in these aggregates in the absence of phosphorylated Aβ. Phosphorylated Aβ in biochemical-Aβ stage 3 was restricted to those cases that already exhibited anti-Aβ and anti-AβN3pE-positive material. AβN3pE–40, AβN3pE–42, phosphorylated Aβ40, and phosphorylated Aβ42 were all found in cases with Alzheimer’s disease with biochemical-Aβ-stage 3. However, AβN3pE-42 and phosphorylated Aβ42 were the predominant forms. This sequence was further confirmed by the finding of a similar hierarchical sequence in the occurrence of Aβ, AβN3pE, and phosphorylated Aβ in senile plaques in controls, cases with pathologically preclinical Alzheimer’s disease, and cases with symptomatic Alzheimer’s disease. Comparison between biochemical detection of Aβ aggregates and immunohistochemistry revealed that the biochemical detection of Aβ aggregates by western blotting was less sensitive than immunostaining for plaques. A possible explanation for this finding is that those cases with initial plaque deposition have only very few plaques that may not be included in the samples taken for biochemical analysis or that the amount of plaque-pathology is too low for detection in brain homogenates. The hierarchical staining pattern of plaques and Aβ aggregates seen in this study can be explained by either a hierarchical occurrence of these Aβ species in the aggregates or by different sensitivities of the antibodies. Arguments in favour of a hierarchical occurrence of Aβ1–40/42, AβN3pE and phosphorylated Aβ are that the antibody sensitivity of anti-AβN3pE and anti-phosphorylated Aβ were quite similar (Saido et al., 1995; Kumar et al., 2013) and did not explain the differences between biochemical-Aβ stages 2 and 3, and that biochemical-Aβ stage 1 cases already exhibited significant anti-Aβ1–17 positive material in the absence of AβN3pE and phosphorylated Aβ signals. Moreover, this sequence was seen in Aβ aggregates in brain homogenates as well as in plaques stained immunohistochemically with these antibodies. A further argument in favour of a hierarchical sequence in which Aβ aggregates accumulate distinct types of Aβ peptides is provided by previous reports showing that Aβ plaques first stain for Aβ42, second for Aβ40 (Iwatsubo et al., 1996; Lemere et al., 1996) followed by AβN3pE (Iwatsubo et al., 1996), then for AβN11pE, and, finally, in very few cases, for Aβ17-40/42 (P3) (Iwatsubo et al., 1996; Thal et al., 2005). As Aβ40 and Aβ42 both occur very early in the development of Aβ plaque pathology and as AβN11pE was seen in plaques of Alzheimer’s disease as well as of pathologically preclinical Alzheimer’s disease cases, we focused our study on Aβ, AβN3pE and phosphorylated Aβ. These three different Aβ peptides exhibited a robust hierarchical sequence that provides a backbone for the determination of other peptides and their relation to the development of Alzheimer’s disease-related Aβ aggregation.

A limitation of autopsy studies is that only a single time point can be analysed for each individual. To minimize this limitation we used the Aβ phase, the Braak–neurofibrillary tangle stage, and the CERAD score for neuritic plaque pathology as widely accepted pathological markers for Alzheimer’s disease progression (Hyman et al., 2012). The biochemical-Aβ stages, thereby, correlated with the phases of Aβ plaque distribution (Thal et al., 2002), the Braak–neurofibrillary tangle stages for neurofibrillary tangle distribution (Braak and Braak, 1991) and with the CERAD score for neuritic plaque pathology (Mirra et al., 1991). This correlation was not simply an effect of ageing because we used partial correlation analysis controlled for age and gender, a statistical method that allows one to calculate the correlation between two parameters independent from age and gender effects. Interestingly, the occurrence and amount of phosphorylated Aβ in biochemical-Aβ stage 3 cases was associated with symptomatic Alzheimer’s disease but not with pathologically preclinical Alzheimer’s disease. As such, it is tempting to speculate that the biochemical composition of Aβ aggregates changes with the progression of Alzheimer’s disease from pathologically preclinical Alzheimer’s disease to Alzheimer’s disease cases. Hence, we assume that cases with Alzheimer’s disease contain more soluble and dispersible Aβ oligomers, protofibrils, and fibrils than pathologically preclinical Alzheimer’s disease and non-Alzheimer’s disease cases and the presence of modified AβN3pE and phosphorylated Aβ peptides may stabilize dispersible oligomeric, protofibrillar and fibrillar Aβ aggregates. An argument in favour of this hypothesis is that both AβN3pE and phosphorylated Aβ have the ability to stabilize Aβ aggregates (Schlenzig et al., 2009; Kumar et al., 2011). An alternative explanation for the increased amounts of AβN3pE and phosphorylated Aβ in Alzheimer’s disease cases in comparison with pathologically preclinical Alzheimer’s disease cases is that both are by-products of an increased production or decreased clearance of Aβ without relevance for the disease and its progression. Accordingly, the accumulation of such by-products would be expected to be more predominant in symptomatic Alzheimer’s disease cases compared with pathologically preclinical Alzheimer’s disease cases, and AβN3pE and phosphorylated Aβ would accumulate in parallel with Aβ plaques detected with anti-Aβ17–24 or anti-Aβ42. However, this was not the case for phosphorylated Aβ. As depicted in Supplementary Fig. 5 isolated pathologically preclinical Alzheimer’s disease cases exhibited very high amounts of Aβ17–24 and AβN3pE-positive plaques, even more than some Alzheimer’s disease cases, but no phosphorylated Aβ. On the other hand, phosphorylated Aβ was seen in all symptomatic Alzheimer’s disease cases, even in those that had fewer plaques than some pathologically preclinical Alzheimer’s disease cases. Another argument against the hypothesis that AβN3pE and phosphorylated Aβ are by-products of Aβ accumulation without specific impact on the disease is that both modified forms of Aβ are more prone to form oligomeric and fibrillar aggregates in vitro than non-modified Aβ (Saido et al., 1995; Schlenzig et al., 2009; Kumar et al., 2011). As such, AβN3pE and phosphorylated Aβ promote the formation of oligomeric, protofibrillar and fibrillar aggregates of Aβ and the biochemical-Aβ stages more likely document the biochemical development of Aβ aggregates in the pathogenesis of Alzheimer’s disease. For all that, it is not yet clear whether AβN3pE and phosphorylated Aβ play a directing role in the pathogenesis of Alzheimer’s disease. At least, they serve as marker proteins for the progression of the disease as shown here.

It is important to note that the biochemical development of Aβ aggregates starts with the aggregation of Aβ in all four fractions received after brain homogenization. Immunoprecipitation with B10AP antibody fragments and A11 revealed that these initial Aβ aggregates already contain Aβ oligomers, protofibrils and fibrils. Given the sequence of events in the biochemical-Aβ stages, it is tempting to speculate that modification of initial Aβ aggregates by adding detectable amounts of AβN3pE and phosphorylated Aβ peptides to these aggregates is a critical event for the development of Alzheimer’s disease. Phosphorylation of Aβ at serine 8 indicating biochemical-Aβ stage 3 rather than the mere presence of AβN3pE, thereby seems to be critical for conversion from pathologically preclinical Alzheimer’s disease to Alzheimer’s disease. Arguments in favour of this hypothesis are: (i) pathologically preclinical Alzheimer’s disease cases do not exhibit significant amounts of phosphorylated Aβ in dispersible oligomers, protofibrils and fibrils but Alzheimer’s disease cases do; (ii) cases with Alzheimer’s disease have significant numbers of phosphorylated Aβ-containing plaques (phosphorylated Aβ plaque load = 1.21%) whereas the phosphorylated Aβ plaque load was in mean <0.28% in pathologically preclinical Alzheimer’s disease cases; and (iii) AβN3pE is already present in significant amounts in plaques (AβN3pE plaque load = 2.25%), dispersible oligomers, protofibrils, fibrils, and in the SDS-soluble membrane-associated fraction in pathologically preclinical Alzheimer’s disease cases and increases quantitatively in Alzheimer’s disease (AβN3pE plaque load = 4.22%) but does not indicate a qualitative change in the composition of Aβ aggregates between Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases because AβN3pE already occurs in biochemical-Aβ stage 2, which is seen in pathologically preclinical Alzheimer’s disease cases, and in biochemical-Aβ stage 3 in Alzheimer’s disease cases. In the event that phosphorylation of Aβ increases its tendency to form dispersible aggregates and, thereby, supports conversion from pathologically preclinical Alzheimer’s disease to Alzheimer’s disease, blocking or modulation of Aβ phosphorylation would be an appropriate mechanism to prevent or delay the conversion from pathologically preclinical Alzheimer’s disease to symptomatic Alzheimer’s disease. An aggregation promoting the role for phosphorylated Aβ has been demonstrated (Kumar et al., 2011). However, it is important to test this potential treatment strategy in an appropriate animal model to exclude the possibility that phosphorylated Aβ is merely a by-product of the disease without therapeutic potential.

Phosphorylation of serine residues by protein kinase A similar to serine 8 of the Aβ peptide (Kumar et al., 2011) is also seen in tau protein (Andorfer and Davies, 2000). Thus, one could assume that Aβ and tau phosphorylation are two results of a common problem: increased phosphorylation of proteins in the Alzheimer’s disease brain. Arguments against this hypothesis are that: (i) dispersible Aβ alone was associated with neurodegeneration in APP transgenic mice with an increased Aβ production (Rijal Upadhaya et al., 2012a); (ii) Aβ was capable of exacerbating tau pathology in tau transgenic mice (Gotz et al., 2001; Lewis et al., 2001) suggesting a causative or at least triggering role for Aβ in Alzheimer’s disease-related neurodegeneration; and (iii) tau phosphorylation occurs early in the pathogenesis of neuronal alterations in Alzheimer’s disease (Braak et al., 2011) as well as in other non-Alzheimer’s disease tauopathies (Dickson et al., 2011), whereas Aβ phosphorylation at serine 8 is a late event mainly restricted to symptomatic Alzheimer’s disease cases, as shown here.

As AβN3pE and phosphorylated Aβ have also been found in APP/PS1 transgenic mice without inducing significant levels of tau pathology the hierarchical accumulation of different forms of Aβ peptides alone may not cause Alzheimer’s disease, but in the presence of mild, pre-existing tau pathology as it is regularly the case in elderly humans (Braak et al., 2011), Aβ aggregates may exacerbate tau pathology as also seen in mouse models for Alzheimer’s disease (Gotz et al., 2001; Lewis et al., 2001; Oddo et al., 2004).

Our finding, that soluble and dispersible Aβ oligomers, protofibrils and fibrils increase from pathologically preclinical Alzheimer’s disease to Alzheimer’s disease cases is in line with the previously reported detection of Aβ oligomers, protofibrils and fibrils in Alzheimer’s disease cases (Kayed et al., 2003; Habicht et al., 2007; Mc Donald et al., 2010). However, our data apparently contradict reports by other authors showing that the Aβ plaque loads did not vary significantly between Alzheimer’s disease and non-demented cases with plaque pathology (Arriagada et al., 1992b) and that increasing cognitive decline in patients with Alzeimer’s disease could not be explained by differences in the Aβ loads (Arriagada et al., 1992a). Of note, in the present study, some pathologically preclinical Alzheimer’s disease cases had higher amyloid plaque loads in the temporal neocortex than some cases with Alzheimer’s disease (Supplementary Fig. 5). This might explain the lack of statistically significant differences in Aβ loads reported in the abovementioned studies. However, when staining Aβ plaques for phosphorylated Aβ we found significant differences between pathologically preclinical Alzheimer’s disease and Alzheimer’s disease cases in the respective phosphorylated Aβ plaque loads indicating that changes in the biochemical composition of the Aβ aggregates occur when pathologically preclinical Alzheimer’s disease cases convert to symptomatic Alzheimer’s disease, i.e. the conversion from biochemical-Aβ stage 2 to biochemical-Aβ stage 3. These qualitative changes were also found biochemically in dispersible Aβ oligomers, protofibrils, and fibrils as well as in the membrane-associated and plaque-associated fractions.

Although it is tempting to assume that the hierarchical sequences of Aβ plaque distribution and that of the biochemical evolution of Alzheimer’s disease-related Aβ aggregates represent a pathogenetic sequence of events it is possible that this sequence can be held at a given point or that Aβ deposition is even reversible until a given point in this sequence. Accordingly, cases classified as pathologically preclinical Alzheimer’s disease (non-demented individuals with Alzheimer’s disease pathology according to current NIA-AA criteria for the neuropathological diagnosis of Alzheimer’s disease (Hyman et al., 2012)) do not necessarily develop symptomatic Alzheimer’s disease.

The non-Alzheimer’s disease control and pathologically preclinical Alzheimer’s disease cases included in this study were identified at autopsy and were not tested for Alzheimer’s disease biomarkers, such as CSF-Aβ and CSF-tau protein or amyloid PET. Therefore, the pathologically preclinical Alzheimer’s disease cases in our study cannot be compared with clinically detectable preclinical Alzheimer’s disease cases according to Vos et al. (2013). However, it will be an important issue for future research to verify the neuropathological and biochemical correlatives in amyloid PET-positive or CSF-biomarker positive non-demented cases and to distinguish them from cases with symptomatic Alzheimer’s disease and non-Alzheimer’s disease.

The missing signals for AβN3pE and phosphorylated Aβ in the soluble oligomers, protofibrils, and fibrils argue in favour of aggregation promoting effects of both post-translational modified Aβ species as previously described in vitro (Schlenzig et al., 2009; Kumar et al., 2011). However, AβN3pE was observed in the soluble fraction of Alzheimer’s disease cases indicating that presumably smaller AβN3pE oligomers are present in the Alzheimer’s disease brain that cannot be precipitated with A11 and B10AP.

In conclusion, we have shown qualitative differences in the composition of Aβ plaques and dispersible Aβ oligomers, protofibrils and fibrils between Alzheimer’s disease and pathologically preclinical Alzheimer’s disease cases that allow the distinction of three biochemical-Aβ stages. Although it appears quite obvious that non-phosphorylated full-length Aβ accumulates before truncated and phosphorylated forms become detectable, their sequence of occurrence was associated with a critical step in the pathogenesis of Alzheimer’s disease: phosphorylated Aβ, indicative for biochemical-Aβ stage 3, was specifically associated with symptomatic Alzheimer’s disease. Thus, phosphorylated Aβ may support further accumulation of Aβ oligomers, protofibrils, and fibrils in the event that pathologically preclinical Alzheimer’s disease converts into Alzheimer’s disease. Phosphorylation of Aβ at serine 8 may be a new therapeutic target to prevent conversion from pathologically preclinical Alzheimer’s disease to Alzheimer’s disease.

Acknowledgements

The authors thank Professor Johannes Attems and Dr. Kelly Del Tredici for his/her helpful comments on this manuscript.

Funding

This study was supported by DFG-grants WA1477/6-2 (J.W.), TH624/4-2, TH624/6-1, Alzheimer Forschung Initiative Grants #10810, #13803 (D.R.T.), #12854 (S.K.), SFB610 and the Landesexzellenz-Netzwerk “Biowissenschaften” (Sachsen-Anhalt) (M.F.).

Conflict of interest

D.R.T. received consultant honorary from Simon-Kucher and Partners (Germany), and GE-Healthcare (UK) and collaborated with Novartis Pharma Basel (Switzerland). C.A.F.v.A. received honoraria from serving on the scientific advisory board of Nutricia GmbH and has received funding for travel and speaker honoraria from Sanofi-Aventis, Novartis, Pfizer, Eisai and Nutricia GmbH, and received research support from Heel GmbH.

Supplementary material

Supplementary material is available at Brain online.

Abbreviations

    Abbreviations
  • amyloid-β

  • CERAD

    Consortium to Establish a Registry for Alzheimer’s Disease; CSF = Cerebrospinal fluid; SDS = Sodium dodecyl sulfate

References

Alafuzoff
I
Arzberger
T
Al-Sarraj
S
Bodi
I
Bogdanovic
N
Braak
H
, et al.  . 
Staging of neurofibrillary pathology in Alzheimer’s disease. A study of the BrainNet Europe Consortium
Brain Pathol
 , 
2008
, vol. 
18
 (pg. 
484
-
96
)
Alzheimer
A
Ueber eine eigenartige Erkrankung der Hirnrinde
Allg Zschr Psych
 , 
1907
, vol. 
64
 (pg. 
146
-
8
)
Andorfer
CA
Davies
P
PKA phosphorylations on tau: developmental studies in the mouse
Dev Neurosci
 , 
2000
, vol. 
22
 (pg. 
303
-
9
)
Arriagada
PV
Growdon
JH
Hedley-Whyte
ET
Hyman
BT
Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer's disease
Neurology
 , 
1992a
, vol. 
42
 (pg. 
631
-
9
)
Arriagada
PV
Marzloff
K
Hyman
BT
Distribution of Alzheimer-type pathologic changes in nondemented elderly individuals matches the pattern in Alzheimer's disease
Neurology
 , 
1992b
, vol. 
42
 (pg. 
1681
-
8
)
Braak
H
Alafuzoff
I
Arzberger
T
Kretzschmar
H
Del Tredici
K
Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry
Acta Neuropathol
 , 
2006
, vol. 
112
 (pg. 
389
-
404
)
Braak
H
Braak
E
Neuropathological stageing of Alzheimer-related changes
Acta Neuropathol
 , 
1991
, vol. 
82
 (pg. 
239
-
59
)
Braak
H
Thal
DR
Ghebremedhin
E
Del Tredici
K
Stages of the pathological process in Alzheimer's disease: age categories 1 year to 100 years
J Neuropathol Exp Neurol
 , 
2011
, vol. 
70
 (pg. 
960
-
9
)
Dickson
DW
Kouri
N
Murray
ME
Josephs
KA
Neuropathology of frontotemporal lobar degeneration-tau (FTLD-tau)
J Mol Neurosci
 , 
2011
, vol. 
45
 (pg. 
384
-
9
)
Dubois
B
Feldman
HH
Jacova
C
Dekosky
ST
Barberger-Gateau
P
Cummings
J
, et al.  . 
Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria
Lancet Neurol
 , 
2007
, vol. 
6
 (pg. 
734
-
46
)
Gotz
J
Chen
F
van Dorpe
J
Nitsch
RM
Formation of neurofibrillary tangles in P301l tau transgenic mice induced by Abeta 42 fibrils
Science
 , 
2001
, vol. 
293
 (pg. 
1491
-
5
)
Habicht
G
Haupt
C
Friedrich
RP
Hortschansky
P
Sachse
C
Meinhardt
J
, et al.  . 
Directed selection of a conformational antibody domain that prevents mature amyloid fibril formation by stabilizing Abeta protofibrils
Proc Natl Acad Sci USA
 , 
2007
, vol. 
104
 (pg. 
19232
-
7
)
Harper
JD
Wong
SS
Lieber
CM
Lansbury
PT
Observation of metastable Abeta amyloid protofibrils by atomic force microscopy
Chem Biol
 , 
1997
, vol. 
4
 (pg. 
119
-
25
)
Hyman
BT
Phelps
CH
Beach
TG
Bigio
EH
Cairns
NJ
Carrillo
MC
, et al.  . 
National Institute on Aging–Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease
Alzheimers Dement
 , 
2012
, vol. 
8
 (pg. 
1
-
13
)
Iwatsubo
T
Saido
TC
Mann
DM
Lee
VM
Trojanowski
JQ
Full-length amyloid-beta (1-42(43)) and amino-terminally modified and truncated amyloid-beta 42(43) deposit in diffuse plaques
Am J Pathol
 , 
1996
, vol. 
149
 (pg. 
1823
-
30
)
Kayed
R
Head
E
Thompson
JL
McIntire
TM
Milton
SC
Cotman
CW
, et al.  . 
Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis
Science
 , 
2003
, vol. 
300
 (pg. 
486
-
9
)
Kim
KS
Miller
DL
Sapienza
VJ
Chen
C-MJ
Bai
C
Grundke-Iqbal
I
, et al.  . 
Production and characterization of monoclonal antibodies reactive to synthetic cerebrovascular amyloid peptide
Neurosci Res Commun
 , 
1988
, vol. 
2
 (pg. 
121
-
30
)
Kumar
S
Rezaei-Ghaleh
N
Terwel
D
Thal
DR
Richard
M
Hoch
M
, et al.  . 
Extracellular phosphorylation of the amyloid beta-peptide promotes formation of toxic aggregates during the pathogenesis of Alzheimer's disease
EMBO J
 , 
2011
, vol. 
30
 (pg. 
2255
-
65
)
Kumar
S
Wirths
O
Theil
S
Gerth
J
Bayer
TA
Walter
J
Early intraneuronal accumulation and increased aggregation of phosphorylated Abeta in a mouse model of Alzheimer's disease
Acta Neuropathol
 , 
2013
, vol. 
125
 (pg. 
699
-
709
)
Lemere
CA
Blusztajn
JK
Yamaguchi
H
Wisniewski
T
Saido
TC
Selkoe
DJ
Sequence of deposition of heterogeneous amyloid beta-peptides and APO E in Down syndrome: implications for initial events in amyloid plaque formation
Neurobiol Dis
 , 
1996
, vol. 
3
 (pg. 
16
-
32
)
Lesne
S
Koh
MT
Kotilinek
L
Kayed
R
Glabe
CG
Yang
A
, et al.  . 
A specific amyloid-beta protein assembly in the brain impairs memory
Nature
 , 
2006
, vol. 
440
 (pg. 
352
-
7
)
Lewis
J
Dickson
DW
Lin
WL
Chisholm
L
Corral
A
Jones
G
, et al.  . 
Enhanced neurofibrillary degeneration in transgenic mice expressing mutant tau and APP
Science
 , 
2001
, vol. 
293
 (pg. 
1487
-
91
)
Masters
CL
Multhaup
G
Simms
G
Pottgiesser
J
Martins
RN
Beyreuther
K
Neuronal origin of a cerebral amyloid: neurofibrillary tangles of Alzheimer's disease contain the same protein as the amyloid of plaque cores and blood vessels
EMBO J
 , 
1985
, vol. 
4
 (pg. 
2757
-
63
)
Mc Donald
JM
Savva
GM
Brayne
C
Welzel
AT
Forster
G
Shankar
GM
, et al.  . 
The presence of sodium dodecyl sulphate-stable Abeta dimers is strongly associated with Alzheimer-type dementia
Brain
 , 
2010
, vol. 
133
 (pg. 
1328
-
41
)
Mirra
SS
Heyman
A
McKeel
D
Sumi
SM
Crain
BJ
Brownlee
LM
, et al.  . 
The Consortium to Establish a Registry for Alzheimer's Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer's disease
Neurology
 , 
1991
, vol. 
41
 (pg. 
479
-
86
)
Monsell
SE
Mock
C
Roe
CM
Ghoshal
N
Morris
JC
Cairns
NJ
, et al.  . 
Comparison of symptomatic and asymptomatic persons with Alzheimer disease neuropathology
Neurology
 , 
2013
, vol. 
80
 (pg. 
2121
-
9
)
Morris
JC
The Clinical Dementia Rating (CDR): current version and scoring rules
Neurology
 , 
1993
, vol. 
43
 (pg. 
2412
-
4
)
Morris
JC
Heyman
A
Mohs
RC
Hughes
JP
van Belle
G
Fillenbaum
G
, et al.  . 
The Consortium to Establish a Registry for Alzheimer's Disease (CERAD). Part I. Clinical and neuropsychological assessment of Alzheimer's disease
Neurology
 , 
1989
, vol. 
39
 (pg. 
1159
-
65
)
Oddo
S
Billings
L
Kesslak
JP
Cribbs
DH
LaFerla
FM
Abeta immunotherapy leads to clearance of early, but not late, hyperphosphorylated tau aggregates via the proteasome
Neuron
 , 
2004
, vol. 
43
 (pg. 
321
-
32
)
Rijal Upadhaya
A
Capetillo-Zarate
E
Kosterin
I
Abramowski
D
Kumar
S
Yamaguchi
H
, et al.  . 
Dispersible amyloid β-protein oligomers, protofibrils, and fibrils represent diffusible but not soluble aggregates: their role in neurodegeneration in amyloid precursor protein (APP) transgenic mice
Neurobiol Aging
 , 
2012a
, vol. 
33
 (pg. 
2641
-
60
)
Rijal Upadhaya
A
Lungrin
I
Yamaguchi
H
Fändrich
M
Thal
DR
High-molecular weight Aβ-oligomers and protofibrils are the predominant Aβ-species in the native soluble protein fraction of the AD brain
J Cell Mol Med
 , 
2012b
, vol. 
16
 (pg. 
287
-
95
)
Saido
TC
Iwatsubo
T
Mann
DM
Shimada
H
Ihara
Y
Kawashima
S
Dominant and differential deposition of distinct beta-amyloid peptide species, A beta N3(pE), in senile plaques
Neuron
 , 
1995
, vol. 
14
 (pg. 
457
-
66
)
Schlenzig
D
Manhart
S
Cinar
Y
Kleinschmidt
M
Hause
G
Willbold
D
, et al.  . 
Pyroglutamate formation influences solubility and amyloidogenicity of amyloid peptides
Biochemistry
 , 
2009
, vol. 
48
 (pg. 
7072
-
8
)
Shankar
GM
Li
S
Mehta
TH
Garcia-Munoz
A
Shepardson
NE
Smith
I
, et al.  . 
Amyloid-beta protein dimers isolated directly from Alzheimer's brains impair synaptic plasticity and memory
Nat Med
 , 
2008
, vol. 
14
 (pg. 
837
-
42
)
Sperling
RA
Aisen
PS
Beckett
LA
Bennett
DA
Craft
S
Fagan
AM
, et al.  . 
Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease
Alzheimers Dement
 , 
2011
, vol. 
7
 (pg. 
280
-
92
)
Thal
DR
Capetillo-Zarate
E
Schultz
C
Rüb
U
Saido
TC
Yamaguchi
H
, et al.  . 
Apolipoprotein E co-localizes with newly formed amyloid beta-protein (Abeta)-deposits lacking immunoreactivity against N-terminal epitopes of Abeta in a genotype-dependent manner
Acta Neuropathol
 , 
2005
, vol. 
110
 (pg. 
459
-
71
)
Thal
DR
Rüb
U
Orantes
M
Braak
H
Phases of Abeta-deposition in the human brain and its relevance for the development of AD
Neurology
 , 
2002
, vol. 
58
 (pg. 
1791
-
800
)
Thal
DR
Rüb
U
Schultz
C
Sassin
I
Ghebremedhin
E
Del Tredici
K
, et al.  . 
Sequence of Abeta-protein deposition in the human medial temporal lobe
J Neuropathol Exp Neurol
 , 
2000
, vol. 
59
 (pg. 
733
-
48
)
Thal
DR
von Arnim
C
Griffin
WS
Yamaguchi
H
Mrak
RE
Attems
J
, et al.  . 
Pathology of clinical and preclinical Alzheimer's disease
Eur Arch Psychiatry Clin Neurosci
 , 
2013
, vol. 
263
 
Suppl 2
(pg. 
S137
-
45
)
Vos
SJ
Xiong
C
Visser
PJ
Jasielec
MS
Hassenstab
J
Grant
EA
, et al.  . 
Preclinical Alzheimer's disease and its outcome: a longitudinal cohort study
Lancet Neurol
 , 
2013
, vol. 
12
 (pg. 
957
-
65
)
Walsh
DM
Lomakin
A
Benedek
GB
Condron
MM
Teplow
DB
Amyloid beta-protein fibrillogenesis. Detection of a protofibrillar intermediate
J Biol Chem
 , 
1997
, vol. 
272
 (pg. 
22364
-
72
)
Watt
AD
Perez
KA
Rembach
A
Sherrat
NA
Hung
LW
Johanssen
T
, et al.  . 
Oligomers, fact or artefact? SDS-PAGE induces dimerization of beta-amyloid in human brain samples
Acta Neuropathol
 , 
2013
, vol. 
125
 (pg. 
549
-
64
)
Yamaguchi
H
Sugihara
S
Ogawa
A
Saido
TC
Ihara
Y
Diffuse plaques associated with astroglial amyloid beta protein, possibly showing a disappearing stage of senile plaques
Acta Neuropathol
 , 
1998
, vol. 
95
 (pg. 
217
-
22
)