APP mutations in the Aβ coding region are associated with abundant cerebral deposition of Aβ38

Abstract

Aβ is the main component of amyloid deposits in Alzheimer disease (AD) and its aggregation into oligomers, protofibrils and fibrils is considered a seminal event in the pathogenesis of AD. Aβ with C-terminus at residue 42 is the most abundant species in parenchymal deposits, whereas Aβ with C-terminus at residue 40 predominates in the amyloid of the walls of large vessels. Aβ peptides with other C-termini have not yet been thoroughly investigated. We analysed Aβ38 in the brains of patients with Aβ deposition linked to sporadic and familial AD, hereditary cerebral haemorrhage with amyloidosis, or Down syndrome. Immunohistochemistry, confocal microscopy, immunoelectron microscopy, immunoprecipitation and the electrophoresis separation of low molecular weight aggregates revealed that Aβ38 accumulates consistently in the brains of patients carrying APP mutations in the Aβ coding region, but was not detected in the patients with APP mutations outside the Aβ domain, in the patients with presenilin mutations or in subjects with Down syndrome. In the patients with sporadic AD, Aβ38 was absent in the senile plaques, but it was detected only in the vessel walls of a small subset of patients with severe cerebral amyloid angiopathy. Our results suggest that APP mutations in the Aβ coding region favour Aβ38 accumulation in the brain and that the molecular mechanisms of Aβ deposition in these patients may be different from those active in patients with familial AD associated with other genetic defects and sporadic AD.

Introduction

Neurofibrillary tangles (NFT) and senile plaques (SP) are the neuropathological hallmarks of Alzheimer disease (AD). NFT are due to the intracellular accumulation of hyper-phosphorylated tau [19], and SP originate from the extracellular deposition of amyloid-beta (Aβ), a family of 3–4.5 kDa peptides that come from the proteolytic processing of the amyloid precursor protein (APP) by β- and γ-secretases [40]. It is thought that Aβ aggregation into oligomers, protofibrils and fibrils is crucial in the development of AD [41]. Aβ40 and Aβ42 are the main products of the β- and γ-secretase cleavage of APP and are consistently detected in amyloid deposits throughout AD brains. Elevated levels of Aβ42 and Aβ40 are associated with several familiar forms of AD (FAD) [3, 20].

Aβ42 is more hydrophobic and tends to oligomerise and aggregate more rapidly and to a greater extent than Aβ40 [21, 52], which has a less negative effect on neuronal viability [11]. On the other hand, Aβ40 is the main component of the amyloid deposits in the vessel walls (cerebral amyloid angiopathy, CAA) associated with AD [45]. Reduced Aβ42 levels in cerebrospinal fluid (CSF) are considered to be an important biochemical marker for a clinical diagnosis of AD [6, 13], although a definite diagnosis still requires neuropathological assessment [16].

The amyloidogenetic properties of Aβ40 and Aβ42 have been extensively studied. However, although Aβ species with other C-termini have not been fully investigated, it is not possible to exclude their involvement in the pathogenesis of AD, either by supporting Aβ40 and Aβ42, or by triggering independent pathogenic events. For example, it has recently been discovered that Aβ43 plays a role in amyloidogenesis, and its investigation is providing important insights into the molecular pathogenesis of AD [39, 54].

Aβ38 has been detected in the CSF, plasma and the medium of cell cultures, and its levels seem to be inversely related to those of Aβ42 [26, 53, 55]. Aβ38 has also been detected in aqueous extracts of AD brains [45], but its cerebral distribution has not been systematically studied by immunohistochemistry.

We analysed Aβ38 in the cerebral cortex of patients with sporadic and familial forms of Aβ amyloidosis. Our results point to a relation between Aβ38 deposition in the brain and APP mutations in the Aβ coding region.

Materials and methods

Patients

We examined the brains of ten FAD patients or hereditary cerebral haemorrhage with amyloidosis (HCHWA) associated with APP mutations, six patients with FAD associated with Presenilin1 (PSEN1) or Presenilin2 (PSEN2) mutations, four adults with Down syndrome (DS), 12 patients with sporadic AD (sAD), and three non-demented controls without cerebral Aβ deposits (Table 1).

Table 1.

Age at death and disease duration of the patients studied and assessment of Aβ deposition and Aβ38 immunoreactivity

Mutations (disease)	Aβ deposition		Aβ38 immunoreactivity		Age at death (years)	Disease duration (years)
APP mutations (FAD/HCHWA)	CAA	SP	CAA	SP
APP A673V (FAD recessive)	+++	+++	+++	+++	46	10
APP E693Q (HCHWA)	+++	+	+++	+	71a	25a
APP E693K (HCHWA)	+++	+	+++	+	60a	16a
APP E693G (FAD)	+++	+++	+++	+++	62	6
APP L705V (HCHWA)	+++	−	+++	−	63a	1a
APP A713T (FAD)	+++	+ +	+++	−	56a	4a
APP V717F (FAD) (2 pts)	+	+++	−	−	45, 49	7, 9
APP V717L (FAD) (2 pts)	+	+++	−	−	44, 51	6, 13
Down syndrome
Patient 1	−	+	−	−	31	/
Patient 2	−	+++	−	−	39	/
Patient 3	−	+++	−	−	46	/
Patient 4	+	+++	−	−	55	/
PSEN1 mutations (FAD)
PSEN1 P117A	+ +	+++	−	−	43	6
PSEN1 I143V	−	+++	−	−	75	11
PSEN1 M146L	+	+++	−	−	43	4
PSEN1 S169L	+ +	+++	−	−	40	11
PSEN2 mutations (FAD)
PSEN2 A85V	+	+++	−	−	82	22
PSEN2 M239V	+	+++	−	−	63	11
Sporadic AD (Braak stage VI) (10 pts)	− to ++	+++	−	−	71–83	6–13
Sporadic AD with severe CAA (2 pts)	+++	+++	+	−	72a, 82	3a, 8
Nondemented controls without Aβ (3 subjects)	−	−	−	−	25, 51, 72	/

CAA cerebral amyloid angiopathy, SP senile plaques

^aOnset and cause of death: cerebral haemorrhage

The APP mutations were A673V (FAD, recessive) [12, 17], E693Q (HCHWA-Dutch type) [24], E693K (HCHWA-Italian I) [7], E693G (FAD, Arctic) [2], L705V (HCHWA-Italian II) [34], A713T (FAD, with severe CAA)[37], V717F (FAD) [32], and V717L (FAD) [33].

The PSEN1 mutations were P117A (unpublished), I143V [15], M146L [5], S169L [47], and the PSEN2 mutations were A85V [36] and M239V [27].

Frozen samples were available from the frontal cortex of one sAD, one sAD with severe CAA (sAD-CAA) and the patients associated with A673V, E693K, E693Q and A713T APP mutations.

Reagents, peptides and antibodies

Unless otherwise stated, all of the chemicals came from Sigma-Aldrich (Munich, Germany) and were of the highest purity available.

The experiments with Aβ synthetic peptides were performed using certified high quality peptides of human Aβ1–38, Aβ1–40, Aβ1–42 and Aβ1–40-Gln22 (successively indicated as Aβ38, Aβ40, Aβ42, Aβ40-Gln22) purchased from Bachem (Bubendorf, Switzerland) and each peptide belonged to a unique lot of production (for Aβ preparations, see Online Resource).

The anti-Aβ antibodies utilized in this work were: a monoclonal antibody reactive to amino acid residues 17–24 of Aβ (clone 4G8, Covance, Emeryville, CA, USA) that is widely used in routine diagnostic procedures [1], monoclonal antibodies reactive to the C-terminus of Aβ and that specifically recognise Aβ40 (clone 11A50-B10, Covance) or Aβ42 (clone 12F4, Covance), the monoclonal antibodies (host species: rabbit) specific for the Aβ isoforms that end at the 38th amino acid (clone BA1–13 and clone 7–14-4, Covance), and a polyclonal anti-Aβ40 antibody (Biosource, USA).

Immunohistochemistry

The routine neuropathological evaluations were based on sections from formalin-fixed, paraffin-embedded blocks that were stained with heamatoxylin–eosin, cresyl violet for Nissl substance, Heidenhain–Woelcke for myelin, and thioflavine S for amyloid. Bodian silver impregnation and immunostaining for phosphorylated tau (clone AT8, 1:300) were used to detect neurofibrillary changes.

Immunohistochemistry (IHC) was performed using the following Aβ antibodies (overnight incubation at room temperature): 4G8 (1:2,000), 11A50-B10 (1:200), 12F4 (1:250), BA1–13 (1:200).

Before Aβ immunostaining, the sections were pre-treated with formic acid (80 %, 45 min). The immunoreactions were visualised by the EnVision Plus/horseradish peroxidase system for rabbit or mouse immunoglobulins, using 3–3′-diaminobenzidine as the chromogen (DakoCytomation, Glostrup, Denmark).

A quantitative analysis of the total Aβ burden (immunolabeling with the 4G8 antibody) and of the deposition of the different Aβ species (Aβ38, Aβ40, Aβ42) was conducted in the different AD brains. The immunostained sections were processed on a computer-assisted image analysis system, consisting of a light microscope (Nikon Eclipse E800, Nikon, Langen, Germany) equipped with a digital camera (Nikon DXM 1200) connected to a personal computer running NIS Elements software (Nikon). The quantitative analysis followed a procedure as previously described [9].

Immunohistofluorescence and confocal microscopy

Double immunohistofluorescence (IHF) methods were used to compare the localisation of Aβ38 with that of total Aβ and Aβ40.

In the first set of experiments, the sections were incubated overnight at 4 °C with a mixture of Aβ38 (1:100) and 4G8 (1:1,000); in the second set, with a mixture of anti-Aβ38 (BA1–13, 1:100) and anti-Aβ40 (11A50-B10, 1:100). They were then incubated for 1 h with a mixture of Alexa Fluor 488 goat anti-rabbit and Alexa Fluor 555 goat anti-mouse (1:100; Life Technologies, Grand Island, NY, USA), at room temperature, in the dark. Cross-sectional information and z-stacks of representative Aβ deposits were obtained and analysed using a confocal laser scanning microscope (Nikon D-Eclipse C-1, Nikon).

Immunoelectronmicroscopy

Electron microscopy (EM) was carried out on formalin-fixed specimens of frontal cortex taken from the patients with the E693K, A673V and A713T APP mutations. The specimens were cut into small blocks, washed in buffer (8 % glucose in 0.4 M phosphate buffer, pH 7.4) for 72 h, and then fixed in 2.0 % EM grade glutaraldehyde [Microscopy Sciences (EMS), Fort Washington, PA, USA] in 0.05 M phosphate buffer, pH 7.4, dehydrated in graded acetone, and embedded in epoxy resin (Spurr) (EMS). Semi-thin sections (1 lm thick) were stained with Toluidine Blue (Carlo Erba Reagents, Arese, Italy) and the areas of interest were selected for immunoelectron microscopy. Post-embedding immunolabelling was performed on non-osmicated sections using anti-Ab38 (BA1–13, 1:10) and 4G8 antibodies (1:100), and the polyclonal Aβ40 antibody (1:100; Biosource, USA), following a previously described protocol [7, 17, 37].

Briefly, 80 nm sections were placed on 200-mesh formvar-coated carbon nickel grids (EMS), and etched in 1 % sodium periodate for 60 min. Antigen expression was enhanced by means of 40 % formic acid for 10 min, and the residual aldehyde groups were quenched with 0.05 M glycine (Sigma) in PBS, pH 4.0, for 20 min. The primary monoclonal antibody was diluted in incubation buffer (Aurion-BSA AURION Immuno Gold Reagents & Accessories, Wageningen, The Netherlands) and applied to the grids overnight at 4 °C. After rinsing in Aurion-BSA, the grids exposed to anti-Aβ38 and anti-Aβ40 antibodies were incubated with goat anti-rabbit (GAR), and the grids exposed to 4G8 antibody were incubated with goat anti-mouse (GAM) secondary antibody conjugated to 10-nm gold particles (1:10; 3 h, room temperature; GAM-10 and GAR-10 Aurion). The sections were fixed in 1 % glutaraldehyde, post-fixed in vapours of 1 % osmium tetroxide (EMS), counterstained with uranyl acetate and lead citrate, and viewed through a transmission electron microscope (EM109, Zeiss, Germany).

Immunoprecipitation/Western blot (WB) analysis

Frozen samples of frontal cortex were homogenised in freshly prepared, ice-cold Tris–HCl 20 mM, pH 7.4 (7:1, Tris–HCl volume/brain wet weight), containing a cocktail of protease inhibitors (Complete Mini, Protease Inhibitor Cocktails, Roche, Mannheim, Germany).

A unique protocol of Aβ immunoprecipitation and WB analysis was used to detect Aβ38, Aβ40 and Aβ42 monomers and low molecular weight (LMW) aggregates in parallel. Briefly, the samples were immunoprecipitated with the pan anti-Aβ polyclonal antibody AW7 [28] and protein G mag Sepharose (GE-Healthcare, Uppsala, Sweden). The Aβ species were liberated by heating the samples at 100 °C for 8 min in 2× lithium dodecyl sulphate (LDS, analogue of sodium dodecyl sulphate, SDS) sample buffer. The total amount of immunoprecipitates was electrophoresed on 4–12 % Bis–Tris gels (Life Technologies, USA) and underwent two colour WB analysis. For the simultaneous detection of Aβ38 and Aβ40 specific bands on a PVDF membrane, a mixture of 11A50-B10 (antibody produced in mouse, 1:1,000) and 7–14-4 (antibody produced in rabbit, 1:1,000) was used as primary antibodies of the WB protocol. Subsequently, a mixture of IRDye 680 conjugated anti-mouse and IRDye 800 conjugated anti-rabbit (LI-COR Biosciences, Lincoln, Nebraska, USA) was employed as the secondary antibodies, in accordance with the instructions provided by LI-COR Biosciences (http://biosupport.licor.com/docs/Western_Blot_Analysis_11488.pdf). Aβ40 bands were detected at 680 nm (red signal) and the Aβ38 bands at 800 nm (green signal), by means of a Li-COR Odyssey near-infrared imaging system.

The same protocol was used for the simultaneous detection of the bands related to monomers and LMW aggregates of Aβ42 and Aβ38, employing a mixture of 12F4 (antibody produced in mouse, 1:1,000) and 7–14-4 (antibody produced in rabbit, 1:1,000) as primary antibodies of the WB protocol. Aβ42 bands were detected at 680 nm (red signal) and the Aβ38 bands at 800 nm (green signal).

The specificity of the immunoprecipitation protocol was verified through a control containing the same amount of homogenate but PBS in place of the AW7 antibody. The specificity for Aβ38 of LMW bands detected in green (emission at 800 nm) in the immunoprecipitates was confirmed loading synthetic Aβ38 as an internal control in each WB. Analogously, synthetic Aβ40 or Aβ42 was used to verify the specificity for Aβ40 or Aβ42 of LMW bands detected in red (emission at 680 nm).

Results

Aβ38 immunohistochemistry in the cerebral cortex of patients with Aβ amyloidosis

Control experiments with synthetic Aβ peptides (dot blot, absorption of BA1–13 and WB) confirmed the specificity of anti-Aβ38 antibodies (see Online Resource).

FAD with the recessive A673V APP mutation

The A673V APP mutation, corresponding to position 2 of Aβ, is unique in terms of its recessive pattern of inheritance and distinctive neuropathological features [12, 17], particularly for the high content of Aβ40-positive deposits [17], and this prompted us to extend the study of this mutation to other Aβ species.

The SP and CAA showed substantial Aβ38 immunoreactivity (Fig. 1a) in the cerebral cortex and also in the cerebellum, where considerable amyloid deposition represents another peculiarity of this mutation. The amyloid deposits were mainly perivascular, and the pattern of Aβ38 immunolabelling closely corresponded to that elicited by thioflavine S amyloid staining. However, the immunoreactivity to Aβ38 was uneven, with intensely positive amyloid deposits in close proximity to negative deposits with very similar morphological characteristics and severity (Fig. 2a–c).

Fig. 1.

Aβ38 immunoreactivity is present in CAA and SP of FAD and HCHWA patients with intra-Aβ APP mutations. In FAD, Aβ38 immunoreactivity is consistent in CAA, perivascular amyloid and SP in the cerebral cortex of the patient with the A673V APP mutation (a), anti-Aβ38 decorates the CAA and SP of the patient with the E693G APP mutation (b), whereas no immunolabeling is present in patients with mutations at codon 717 of APP (V717F in c). In HCHWA, anti-Aβ38 labels CAA and diffuse plaques of patient with the E693Q APP mutation (d), CAA in the arterioles and capillaries of the patient with E693K APP mutation (e) and only the amyloid-laden walls of the leptomeningeal arteries of the patient with L705V APP mutation (f). Aβ38 immunolabeling is absent in most of the patients with sporadic AD (g), and is slight to moderate and confined to a few vessels with CAA in a small subset of cases (h, i). Scale bar in a = 50 μm (a, b, c, d and h have the same magnification); scale bar in e = 100 μm (e, f, g and i have the same magnification)

Fig. 2.

Aβ38 labels a subset of amyloid-laden vessels and parenchymal Aβ deposits. Comparison of adjacent sections immunostained with anti-Aβ38 (a, d, g), treated with thioflavine S (b, e, h) or immunostained with an antibody to total Aβ (c, f, i: 4G8). In A673V FAD, Aβ38 unevenly labels amyloid-laden vessels, perivascular deposits and SP, often adjacent to unlabelled structures in the same area (a–c). In E693Q HCHWA, Aβ38 labels CAA and parenchymal Aβ deposits; and its distribution roughly corresponds to that of thioflavine S fluorescence (d–f). In sAD-CAA, very few amyloid-laden vessels are positive, mainly leptomeningeal arteries (g–i). Scale bar in a = 200 μm (a–c, g–i have the same magnification); scale bar in d = 50 μm (d–f have the same magnification) The diagram in panel j shows the percentage area (values of y axis) of immunoreactivity for total Aβ (4G8) and for Aβ forms with C-termini at residue 42, 40 and 38 in the frontal cortex of the different patients examined

Familial Aβ amyloidosis with APP mutations in the Aβ coding region (intra-Ab APP mutations)

To verify whether Aβ38 deposition distinguishes the A673V APP recessive mutation, we examined the brains of patients carrying other APP or PSEN1 and PSEN2 mutations (Table 1).

The first group included patients with APP mutations in the Aβ coding region: E693G, E693Q and E693K (position 22 of Aβ), L705V (position 34) and A713T (position 42).

The E693G APP mutation (also known as the Arctic mutation) is associated with AD phenotype. Antibodies to Aβ38 revealed CAA and the ring-shaped amyloid plaques typical of this mutation [2], consisting of a crown of delicate amyloid bundles surrounding an area of finely granular or patchy Aβ deposits (Fig. 1b). In the centre of some plaques, small Aβ38 immunoreactive cores or capillaries were observed (Fig. 1b).

The E693Q APP mutation (also known as the Dutch mutation) is associated with HCHWA. Aβ38 immunostaining was consistently present in CAA and involved parenchymal amyloid deposits in most cortical areas (Figs. 1d, 2d–f).

The E693K APP mutation (also known as the Italian 1 mutation) is associated with HCHWA phenotype. In this patient, we detected abundant Aβ38 deposition in the walls of parenchymal and leptomeningeal vessels (Fig. 1e), but few diffuse plaques and pre-amyloid deposits were decorated by Aβ38.

In the HCHWA patient with the L705V APP mutation, no parenchymal deposits were observed and Aβ38 immunoreactivity was related to CAA of the leptomeningeal and, less extensively, to cortical vessels (Fig. 1f).

A similar vascular pattern of Aβ38 deposition was observed in the patient with the A713T APP mutation (AD phenotype), although it involved both leptomeningeal and parenchymal vessels. The SP were intensely immunoreactive for 4G8 antibody, but negative for Aβ38. It is worth noting that the A713T APP mutation lies inside the Aβ42 sequence and adjacent to the γ-secretase cleavage site, but outside the Aβ38 and Aβ40 region.

Aβ amyloidosis associated with other APP genetic defects or mutations in PSEN and PSEN2

We analysed four patients with V717F and V717L APP mutations, located outside the Aβ region and between the γ- and ε-secretase cleavage sites, and four patients with DS and abundant Aβ cerebral deposits (Table 1). Surprisingly, none of them showed Aβ38 immunopositivity (Fig. 1c).

Aβ38 immunostaining was also negative in seven patients with PSEN1 and PSEN2 mutations, thus indicating that mutations in these genes may also not induce consistent Aβ38 accumulation in the brain.

Sporadic AD

To assess the specificity of these results, we extended the study to a series of patients with sAD and different levels of CAA severity (Table 1). Most showed no anti-Aβ38 immunodecoration (Fig. 1g). Very slight, focal immunoreactivity was detected in only two subjects: the sAD-CAA patients (Fig. 1h, i). Aβ38 immunolabelling was confined to CAA, and was more substantial in the leptomeningeal than the parenchymal vessels; the SP were negative (Fig. 2g–i).

The results did not change with modification of the concentration and time of the formic acid incubation. Absence of formic acid pretreatment strongly reduced the immunoreactivity for Aβ38 in positive cases and did not change the absence of immunostaining in negative cases.

Comparison between the deposition of Aβ38 and other Aβ species

In different AD patients, Aβ38 deposits were quantitatively compared to the total Aβ percentage area of immunoreactivity (as detected by 4G8) and to those of the most abundant Aβ peptides in the brain: Aβ40 and Aβ42 (Fig. 2j). Aβ38 immunoreactivity was significant in patients with A673V, E693G and E693Q APP mutations, where the percentage area of immunoreactivity for total Aβ, Aβ42 and Aβ40 was similar, and in patients with E693K and A713T APP mutations, where Aβ42 was scarce. Aβ38 immunoreactivity was absent or very slight in patients with APP V717F and with sAD, also those exhibiting significant Aβ40 deposition (Fig. 2j).

The immunohistochemical distribution of Aβ38 and total Aβ (4G8) was compared also using double-immunofluorescence confocal microscopy (Fig. 3).

Fig. 3.

Relationships of Aβ38 deposits with total Aβ and Aβ40. CAA in the cerebral cortex of the patient with the A713T APP mutation, analysed using confocal microscopy and z-scanning (a–c). Most of the total Aβ immunoreactive area (a, 4G8, red) is also labelled by anti-Aβ38 (b, BA1–13, green), which is widely, but not uniformly distributed in the vessel wall. Comparison of the Aβ40 (d, g, j, red) (clone: 11A50-B10) and Aβ38 (e, h, k, green) (clone: BA1–13) immunoreactivity in SP of the patient with the A673V APP mutation (d–f), in vascular deposits of the patient with the A713T APP mutation (g–i) and in vascular deposits of the sAD-CAA patient (j–l). Aβ38 does not completely overlap the Aβ40 deposition. In particular, some Aβ40-positive SP in the patient with A673V APP mutation (d–f) and perivascular deposits in the patient with A713T APP mutation (g–i) do not have any immunoreactivity for Aβ38. In sAD-CAA, Aβ38 is located in a small area of the vessel wall (k), which is completely decorated by anti-Aβ40 (j). The images on the left (red) and in the middle (green) are merged on the right (yellow) (c, f, i, l). These images of 7 μm brain tissue sections are obtained by z-scanning a number of planes sufficient to visualise the whole surface of SP or vessel. Scale bar in a = 50 μm. All of the other pictures have the same magnification

In the case of the A713T APP mutation, Aβ38 was distributed throughout the vascular Aβ deposits revealed by 4G8, but was completely absent in the SP (Fig. 3a–c).

As the prevalently vascular distribution of Aβ38 is similar to that of Aβ40, we assessed Aβ38 and Aβ40 immunoreactivity in the same amyloid deposits (Fig. 3). In the case of the A673V APP mutation, a subset of Aβ40-positive SP did not show any immunoreactivity for Aβ38, but the two species largely overlapped when Aβ40 and Aβ38 co-existed in the same lesions (Fig. 3d–f). Conversely, in the CAA of the patient with the A713T APP mutation, the labelling of the two peptides did not overlap completely, as Aβ38 was confined to the vascular wall, whereas Aβ40 extended to the surrounding parenchyma (Fig. 3g–i).

In the sAD-CAA patient, the specific signal of Aβ38 was much less intense than that of Aβ40 (Fig. 3j–l).

Electromicroscopic characterisation of Aβ38-positive deposits

Frontal cortex sections of FAD patients with the A673V, A713T and E693K APP mutations were used to investigate the ultrastructure of Aβ38 deposits. Immunoelectron microscopy showed that the material decorated by the anti-Aβ38 antibody consisted of long, straight 8–10 nm fibrils that were morphologically recognisable as amyloid (Fig. 4). In the three patients, Aβ38 immunogold labelling was substantial in CAA, whereas it was also present in parenchymal amyloid only in the patient with the A673V APP mutation. Paralleling the scenario described at optical level, only a fraction of the amyloid deposits was labelled by anti-Aβ38, with no difference in vessel structure or in the characteristics of the amyloid deposits between the Aβ38-positive and Aβ38-negative lesions. Whereas other anti-Aβ antibodies also decorated non-fibrillar material (Fig. 4h), anti-Aβ38 immunogold labelling was only present on fibrils (Fig. 4c, f, i).

Fig. 4.

Electron microscopy of Aβ deposits labelled by 4G8, anti-Aβ40 and anti-Aβ38 antibodies. Comparison of the 4G8 (a, d, g), Aβ40 (b, e, h) and Aβ38 immunoreactivity (c, f, i) in vessels walls of the patients with the A673V APP (a–c), E693K APP (d–f) and A713T APP mutations (g–i). Aβ38 labels long, straight, unbranched fibrils with a diameter of 8–10 nm. The pattern of Aβ38 immunolabelling does not completely overlap the immunodecoration elicited by 4G8 and anti-Aβ40 antibodies. In particular, some non-fibrillar aggregates (arrows) are labelled by anti-Aβ40, but not by anti-Aβ38. Scale bar in a = 100 nm. All the other pictures have the same magnification

Low molecular weight Aβ38 species in the cerebral cortex of patients with Aβ amyloidosis

Although AD is histopathologically distinguishable from other dementias by its association with abundant and widespread extraneuronal Aβ deposition, low molecular weight (LMW)-aggregated forms (Aβ oligomers and protofibrils) have become the focus of the most studies on AD pathogenesis. The aim of this last part of our work was to verify whether there is a relationship between the Aβ38 deposition revealed by immunohistochemistry and the detection of Aβ38 species of LMW.

Homogenates of the subjects with the A673V, A713T APP mutations and one sAD-CAA patient were investigated in parallel. The immunoprecipitation with the pan anti-Aβ polyclonal antibody AW7 and the separation of Aβ species by the two colour WB protocol revealed that the anti-Ab38 antibody (clone 7–14-4) detects a set of bands distinguishable from monomers and aggregates of Aβ40 and Aβ42 (Fig. 5c, f) and characterised by a lower molecular weight than Aβ40 and Aβ42 bands.

Fig. 5.

LMW species of Aβ38 are detectable in the brain of patients with intra-Aβ mutations. Electrophoresis separation of frontal cortex homogenates of a sAD-CAA patient (lanes 1 in all panels), FAD patients with A713T (lanes 2 in all panels), with A673V APP (lanes 3 in all panels) mutations, immunoprecipited with the pan anti-Aβ polyclonal antibody AW7. Monomers and aggregates of Aβ38 and Aβ40 are detected on the same membrane by two colour WB protocol (a–c). Aβ38 bands are identified as green signal at 800 nm (a green signal converted to grey), Aβ40 bands as red signal at 680 nm (b: red signal converted to grey). Using the same protocol, monomers and aggregates of Aβ38 and Aβ42 are detected on the same membrane (d–f). Aβ38 bands are identified as green signal at 800 nm (d: green signal converted to grey) and Aβ42 bands as red signal at 680 nm (e: red signal converted to grey). Panel c is the part of the membrane, shown in a and b, related to LMW species, in the original two colours. Panel f is the part of the membrane, shown in d and e, related to LMW species, in the original two colours. Aβ38 monomer and dimers (grey bands in a, d, green bands in c, f) are detectable only in FAD patients with A713T (lane 2 a, c, d, f) and A673V APP (lane 3 a, c, d, f) mutations. A trimeric band is also present in these two patients but with a lower intensity. In sAD-CAA (lane 1 a, c, d, f) Aβ38 is not detected and the only signal detected at 800 nm is related to the immunoglobulin used to immunoprecipitate Aβ (lane 6 a, c). Aβ40 monomer, dimers and trimers (grey bands in b, red bands in c) are clearly separated both in the patients with A713T (lane 2 b, c) and A673V APP mutation (lane 3 b, c) and in sAD-CAA (lane 1 b, c). Aβ42 (grey bands in e, red bands in f) is mainly present as monomer and dimer in A673V APP (lane 3 e, f) and as dimer in sAD-CAA (lane 1 e, f). No clear Aβ42 bands are detectable in the homogenates of the patient carrying the A713T APP mutation (lane 2 e, f). For each membrane, synthetic Aβ38 has been used as control of the specificity of the green signal for Aβ38 species (lane 5 a–f), synthetic Aβ40 has been used as control of the specificity of red signal for Aβ40 species (lane 4 a–c) and synthetic Aβ42 has been used as control for the specificity of the red signal for Aβ42 species (lane 4 d–f). White bands indicate saturated signal

In particular in A673V and A713T APP, Aβ38 monomer and dimers were always clearly detected, while they were undetectable in the homogenates of the sAD-CAA patient (Fig. 5a, c and d, f). Differently, monomer, dimer and trimer of Aβ40 were clearly separated both in the patients with mutations in the Aβ coding region (A673V and A713T APP mutations) and in sAD-CAA (Fig. 5b, c). Aβ42 was mainly present as monomer and dimer in A673V APP and as dimer in sAD-CAA. No clear Aβ42 bands were identified in the homogenates of the patient associated with A713T APP mutation, but the position of this mutation at 42nd residue of Aβ may have influenced the detection of Aβ42 monomer, dimer and trimers by this protocol (Fig. 5e, f).

Since monomers and dimers of Aβ38 were visualised only in the homogenates of the patients carrying the A673V and A713T APP mutations and they were undetectable in sAD-CAA, who had shown a slight positivity for Aβ38 in IHC, we extended the analysis to other two patients with intra-Ab mutations (E693Q and E693K APP) and another sAD. Also in this case we detected Aβ38 monomer and dimers only in the two patients with APP mutations.

Discussion

Although the most common form of AD is sporadic and has a late-onset, the rare early-onset forms genetically determined by APP, PSEN1 and PSEN2 mutations have provided the background for developing animal models and gaining insights into the basic pathogenetic mechanisms of the disease [14]. Aβ42 and Aβ40 production and fibrillogenesis have been studied in several of these mutations in comparison with sAD. Moreover, amino acid substitutions in APP may produce a variety of effects on APP processing and Aβ aggregation [4]. The identification of a convergence of phenotype in a specific group of FAD patients could represent an important step in the understanding of the processes of Aβ production and accumulation in the brain.

Our findings show that Aβ38 selectively accumulates in FAD patients with intra-Aβ APP mutations and that Aβ38 deposits are predominantly located in the vascular and perivascular compartments.

Examining a relevant number of patients harbouring several different PSEN1, PSEN2 and APP mutations increased the possibility of identifying Aβ38-positive patients.

Our results indicate that Aβ38 deposition is consistently present in pathologic conditions related to APP mutations in the Aβ coding region, while is absent in those associated with other APP genetic defects or modifications of γ-secretase activity due to presenilin mutations.

PSEN1 is a central component of γ-secretase, and its mutations lead to altered APP cleavage and a higher Aβ42/Aβ40 ratio [43]. The undetectability of Aβ38 deposition in the brains of patients with PSEN1 mutations parallels the very low number or absence of Aβ40-immunoreactive amyloid deposits in most of the FAD patients with mutations of this gene [15, 23]. On the other hand, it is known that some PSEN1 mutations not only increase Aβ42 levels, but also those of Aβ43, Aβ45 and peptides even longer than Aβ46 [39, 44, 50], thus suggesting a carboxy-terminal shift of γ-secretase cleavage.

Individuals with DS carry an extra copy of chromosome 21 containing APP, and develop AD neuropathology in the third/fourth decade of the life [18, 35]. Recent studies have demonstrated that an extra copy of APP is also present in cases of FAD [38]. Aβ38 was absent in the subjects with DS, thus indicating that not all of the genetic/chromosomal abnormalities affecting APP lead to Aβ38 deposition. Accordingly, the patients with mutations at codon 717 of APP (i.e. outside the Aβ region) were also Aβ38 negative by immunohistochemistry.

Conversely, Aβ38 was detected in the patient with the A713T APP mutation, corresponding to residue 42 of Aβ (i.e. adjacent to the c-secretase cleavage site). The substantial presence of Aβ38 and Aβ40 in the amyloid deposits of this patient [37] is intriguing, and suggests that even an intra-Aβ mutation lying outside Aβ38 can have an impact on the deposition of this peptide in the brain.

Our observations reinforce the view that intra-Aβ mutations are associated with high levels of Aβ isoforms shorter than Aβ42, including Aβ38. Accordingly, artificial APP mutations in the GxxxG motif between glycine 33 and 37 of Aβ are associated with an increase in the levels of Aβ fragments with C-termini at residues 34, 37 or 38 [30], and mutations in lysine 28 of Aβ shift the primary site of γ-secretase cleavage from Aβ position 40 to 33 [22].

It is noteworthy, in this regard, that the patients who show abundant Aβ38-positive deposits also display substantial Aβ40 deposition. This suggests that Aβ38 co-deposits with Aβ40 and the phenomenon are favored by the presence of intra-Aβ APP mutations, since the sporadic cases with abundant Aβ40 have scanty or absent Aβ38 immunoreactivity and undetectable Aβ38 monomer and dimers.

Like Aβ40, Aβ38 is predominantly localised in vessel walls. Confocal microscopy documented their similar spatial distribution, with the more extensive deposition of Aβ40 and limited areas of exclusively Aβ38 immunolabelling. Although immunoelectron microscopy showed that Aβ38 accumulates in vessels and SP as amyloid fibrils, Aβ38-immunoreactive lesions were frequently unevenly distributed, thus indicating that this peptide accumulates only in selected amyloid deposits.

As cognitive impairment correlates better with soluble Aβ levels than with plaque density in AD brains [25, 48], a revised version of the amyloid cascade hypothesis considers oligomers the most detrimental forms of Aβ for neuronal and synaptic function. It has been discovered that LMW Aβ aggregates (especially Aβ oligomers with relative molecular masses of 8 and 12 kDa) are highly neurotoxic [51], and dimers are the smallest synaptotoxic species [42]. Monomer and SDS-stable dimer are significantly elevated in brain homogenates of AD cases compared to non-demented controls [29] and they are newly emerging as relevant Aβ forms since the earliest phases of the disease [28]. Most of the works in the literature discuss the presence of Aβ oligomers in the brain of AD patients without emphasis on their C-terminus. Our study, focused on Aβ38 in comparison to Aβ40 and Aβ42 LMW species, indicated that A673V, E693Q, E693K and A713T APP mutations favour the aggregation of Aβ38 that produces SDS-stable dimers. Differently, Aβ40 LMW species are detectable, both in FAD with intra-Ab mutations and in sAD-CAA.

The detected Aβ38 dimers may be produced independently or derived from higher and less stable Aβ38 assemblies and broken down to dimers in SDS. In any case, our findings suggest that a very stable bond involves Aβ38 monomers in the process of aggregation. It might be also possible that Aβ38 takes part in hetero-dimers and trimers with Aβ40 and 42, but their definite detection is limited by biochemical methods so far available.

In brief, immunohistochemistry, confocal and immunoelectron microscopy, Aβ38 immunoprecipitation and the electrophoretic separation of LMW species led to the same main result: Aβ38 consistently accumulates the brains of all the analysed FAD patients with intra-Aβ APP mutations, whereas Aβ38 immunoreactivity was present only in a subset of sAD patients with severe CAA.

Overall, these data suggest that specific APP mutations might quantitatively enhance Aβ38 production. This hypothesis is in agreement with recent findings of qualitative changes in the Aβ profiles with increased levels of Aβ38 in the medium of cells expressing some APP mutations, but not in the medium of cells expressing PSEN mutations [8, 31].

Even if our case series is wide considering the rarity of pathologic conditions associated with APP mutations, a limit of our study is that we could examine only one or two cases for each APP mutation. We, therefore, cannot draw conclusion about the variability of Aβ38 deposition within patients carrying the same APP mutation.

Our conclusions are in line with those coming from a study of patients with another intra-Aβ APP mutation, D694N APP (D23N Aβ, Iowa), which documented the presence of Aβ38 in the brain parenchyma and vessels, together with other N- and C-terminally truncated species [49]. We can, therefore, link a group of APP mutations to the accumulation of a specific Aβ peptide, even if the pattern that leads to Aβ38 production is still unclear. It remains to be determined whether this peptide originates from a specific c-secretase cleavage [10] or the proteolysis of longer Aβ peptides [46].

Our parallel analysis of a wide range of AD patients, diverse for genetic and neuropathological features, revealed relationships between Aβ38 accumulation in the brain, mutated Aβ and the vascular compartment, thus drawing attention to Aβ38, a peptide so far not extensively analysed. The results suggest that the molecular mechanisms of Aβ deposition in FAD patients with intra-Aβ APP mutations may be different from those observed in patients with FAD due to other genetic defects or sAD, thus opening up new perspectives for understanding the pathogenesis of AD.