The polymorphic β‐amyloid lesions present in individuals with Alzheimer's disease are collectively known as cerebral β‐amyloidosis. Amyloid precursor protein (APP) transgenic mouse models similarly develop β‐amyloid depositions that differ in morphology, binding of amyloid conformation‐sensitive dyes, and Aβ40/Aβ42 peptide ratio. To determine the nature of such β‐amyloid morphotypes, β‐amyloid‐containing brain extracts from either aged APP23 brains or aged APPPS1 brains were intracerebrally injected into the hippocampus of young APP23 or APPPS1 transgenic mice. APPPS1 brain extract injected into young APP23 mice induced β‐amyloid deposition with the morphological, conformational, and Aβ40/Aβ42 ratio characteristics of β‐amyloid deposits in aged APPPS1 mice, whereas APP23 brain extract injected into young APP23 mice induced β‐amyloid deposits with the characteristics of β‐amyloid deposits in aged APP23 mice. Injecting the two extracts into the APPPS1 host revealed a similar difference between the induced β‐amyloid deposits, although less prominent, and the induced deposits were similar to the β‐amyloid deposits found in aged APPPS1 hosts. These results indicate that the molecular composition and conformation of aggregated Aβ in APP transgenic mice can be maintained by seeded conversion.
The aggregation and deposition of the β‐amyloid peptide (Aβ) in brain is considered an early and predictive lesion of Alzheimer's disease (AD) [1–3]. Aβ of various lengths are generated by cleavage of amyloid precursor protein (APP) by β‐secretase and γ‐secretase [4, 5]. The most prominent Aβ species are full‐length or N‐truncated Aβx‐40 (Aβ40) and Aβx‐42 (Aβ42) with Aβ42 being much more prone to aggregation and more neurotoxic compared with Aβ40 [4, 5].
There are multiple lines of evidence for polymorphic Aβ aggregation [6, 7]. Aβ deposits in brain can differ in morphology and in biochemical composition within and among individuals with AD, between normal aging and AD, and among APP transgenic mouse models [6–13]. Such Aβ morphotypes might be governed by host factors, such as Aβ posttranslational modifications, Aβ length variant generation or Aβ amino‐acid substitutions as found in familial AD. However, we have previously shown the induction of different Aβ morphotypes in a given transgenic mouse line after intracerebral application of aggregated Aβ‐containing brain extracts from two different sources . Similarly, in vitro, distinct structural variants of synthetic Aβ fibrils can be grown and are self‐propagating under seeded growth conditions [15, 16].
The aim of the present study was to determine whether the induction of different Aβ morphotypes in genetically defined recipient mice reflect variations in the molecular composition and/or conformation of the aggregated Aβ in the donor brain extracts. This would suggest that the prion conformation strain concept [17–20] might also apply to cerebral β‐amyloidosis and further supports the concept of prion‐like templated misfolding of Aβ.
Results and discussion
Aβ morphotypes in APP23 and APPPS1 mice
APP23 and APPPS1 transgenic mice develop age‐dependent depositions of Aβ in the brain [21, 22]. While amyloid deposits in APP23 mice are characterized by fairly large Aβ deposits consisting of congophilic amyloid cores with diffuse penumbras, as well as diffuse Aβ deposition, APPPS1 mice develop small, compact and highly congophilic Aβ deposits [21, 22]. These mouse strain‐specific plaque morphotypes can be found throughout the neocortex and hippocampus (Fig 1A–D).
Conformational differences of variant Aβ morphotypes can be studied by anionic luminescent conjugated polythiophene (LCP) including trimeric polythiophene acetic acid (tPTAA) . LCPs are flexible amyloid‐binding dyes whose spectral properties depend on the amyloid conformation. An LCP‐based histo‐optical imaging technique has previously been used to discriminate various prion strains, types of systemic amyloids and heterogeneous Aβ deposits [12, 24, 25]. When tPTAA was applied to tissue sections of amyloid‐bearing APP23 and APPPS1 mice, fluorescence images revealed a shift towards bright yellow–greenish colours for the amyloid plaques in APP23 mice, whereas the amyloid plaques in APPPS1 mice showed a reddish appearance (Fig 1E,F). Subsequent spectral analysis revealed tPTAA emission spectra in APP23 mice with a maximum intensity at ∼575 nm and a shoulder at shorter wavelengths (around 535 nm), with more red‐shifted spectra also present (Fig 1G). In contrast, plaques in the APPPS1 mice showed a narrower tPTAA spectral distribution with a maximum intensity at ∼590 nm and a shoulder at longer wavelengths (around 630 nm) (Fig 1G). Quantitative results revealed a significant difference in the 534/631 nm emission ratio between the amyloid deposits in APP23 and APPPS1 mice (Fig 1H; see also supplementary Fig S1 online). The spectral difference was also present, albeit less distinct, when amyloid fibrils were isolated from whole brains of APP23 and APPPS1 mice (supplementary Fig S2 online).
APP23 mice express Swedish‐mutated human APP, with Aβ40 generation exceeding that of Aβ42. In contrast, APPPS1 mice harbour mutated presenilin (PS) 1 in addition to Swedish‐mutated APP, and thus generate more Aβ42 than Aβ40 albeit lower total amounts of total Aβ compared with APP23 mice [21, 22]. To test whether this difference in Aβ length variant generation is also reflected in the amyloid plaques, laser dissection of amyloid plaques with subsequent Aβ‐immunoblotting was performed. Results revealed an Aβ40/42 ratio of 4.7/1 for aged APP23 mice, and a ratio of 0.3/1 for APPPS1 mice (Fig 1I–K)) consistent with enzyme‐linked immunosorbent assays from brain homogenates [14, 21, 22].
Propagation of Aβ morphotypes by seeding
Intracerebral injection of minute amounts of brain extract from β‐amyloid‐laden aged APP23 or APPPS1 mice induces β‐amyloidosis in young pre‐depositing APP23 and APPPS1 mice . Here we have replicated these findings and show that β‐amyloid‐containing APP23 extract injected into the hippocampus (dentate gyrus) of young, pre‐depositing, 4–6‐month‐old APP23 mice induces β‐amyloid deposits surrounded by diffuse, filamentous Aβ immunoreactivity 3 months after injections. In contrast, β‐amyloid‐containing APPPS1 extract injected in the same host induces punctate, coarse and compact Aβ‐plaques 3 months after injections (Fig 2A–C). While the APP23 extract induced Aβ deposits throughout the subgranular and molecular layers of the dentate gyrus, the induced Aβ deposits by the APPPS1 extract were largely confined to the subgranular layer and polymorphic region of the dentate gyrus (Fig 2B,C). Quantification revealed that total Aβ deposition induced by the APP23 extract was more than double compared with the APPPS1 extract (Fig 2B,C). In contrast, when only the compact Aβ deposition was quantified, there was no clear difference between the two extracts (Fig 2B,C).
The reverse experiment, that is, APP23 and APPPS1 extracts injected into young, pre‐depositing 1.5–3‐month‐old APPPS1 mice and analysed 1.5–3 months later, revealed again a more diffuse and filamentous pattern of Aβ‐deposition for the APP23 extract and prominent Aβ‐deposition in the subgranular cell layer for the APPPS1 extract. However, this morphological difference between the extracts was not as obvious as in the APP23 host and also the induced amount and pattern of total or compact Aβ immunoreactivity appeared similar between the two extracts (Fig 2D,E).
APP23 and APPPS1 mice analysed 1 week after the injection of the β‐amyloid‐containing brain extracts did not show amyloid deposition demonstrating that the Aβ deposits did not simply represent the injected Aβ‐containing material (n=8; 2 mice/group; results not shown) [14, 26].
tPTAA spectral emission of seeded Aβ deposits
To study conformational differences of the induced Aβ morphotypes, tPTAA staining with spectral analysis was applied (Fig 2F–I). Strikingly, while the APP23 extract injected in APP23 mice induced Aβ deposits with a yellow–greenish appearance, the APPPS1 extract injected in APP23 mice induced Aβ deposits with a reddish shifted appearance (Fig 2F,G). Vice versa, while the Aβ deposits in APP23 extract‐seeded APPPS1 mice revealed a spectral shift towards a yellow–greenish colour, the APPPS1 extract‐seeded APPPS1 mice revealed a more reddish pattern, although this difference appeared again less striking compared with the APP23 host (Fig 2H,I). Quantitative spectral analysis of the 534/631 nm emission ratio confirmed this qualitative histological impression (Fig 3A). When compared with the endogenous Aβ deposits of the hosts (Fig 1H), the 534/631 nm emission ratios in APP23‐seeded APP23 mice and APPPS1‐seeded APPPS1 mice were reminiscent of the emission ratios in APP23 and APPPS1 mice, respectively.
Aβ40/42 ratio of seeded Aβ deposits
To test whether the ratio of Aβ40/42 might underlie the different morphology and LCP spectral analysis of the induced Aβ deposits, laser dissection of induced Aβ deposition with subsequent Aβ‐immunoblotting was performed (Fig 3B). Intriguingly, there was a significant difference between APP23‐seeded and APPPS1‐seeded Aβ deposits in the APP23 host. While APP23‐seeded Aβ deposits showed a mean Aβ40/42 ratio of 3.7/1, APPPS1‐seeded APP23 mice revealed a mean Aβ40/42 ratio of 1.5/1. (Note that in ‘unseeded’ APP23 mice, the ratio was 4.7/1 and in APPPS1 mice 0.3/1, see Fig 1K). This finding implies the existence of a selection mechanism, whereby the seeded Aβ deposits incorporate mainly the host‐generated Aβ isoform that resembles the composition of the seed. This seeded imprinting of the Aβ40/42 ratio in the APP23 host paralleled the morphological characteristics of the induced Aβ deposits and LCP staining.
In the APPPS1 host and in contrast to the LCP staining, immunoblot analysis of the induced Aβ deposition failed to reach significance between the two extracts (Aβ40/42 ratio of 0.35/1 for APP23 extract versus 0.25/1 for APPPS1 extract; Fig 3B) and the ratios of both APP23‐ and APPPS1‐seeded Aβ deposits were similar to ratio in the ‘unseeded’ APPPS1 host. This observation might reflect that seeded nucleation in the APPPS1 host was obscured by endogenous spontaneous nucleation events and/or cross‐seeding of the abundant and highly aggregate‐prone Aβ42 generated by the APPPS1 host [6, 22].
‘Strain‐like’ Aβ morphotypes
Prion strains might differ in amino‐acid sequence of the prion protein or are based on polymorphic assembled states of the same prion protein molecule [17–20]. While the latter has at least been shown for Aβ in vitro [15, 16], the results of the present study suggest that, for β‐amyloidosis in brain, C‐terminal chain‐length variants might contribute to the nature of the Aβ conformers and can be sustained by seeded induction (‘strains’). In apparent contrast, previous in vitro studies have reported that assemblies of Aβ protofibrils with different Aβ40/42 ratios have similar molecular structures [27, 28]. However, in vivo and within a cellular environment Aβ aggregation is likely different . While the present study only focused on Aβ40 and Aβ42, it should be noted that there are many additional Aβ isoforms with various amino‐ and carboxy terminal lengths. In addition, there are Aβ amino‐acid sequence differences associated with familial forms of cerebral β‐amyloidosis [5, 29].
The physiological relevance of Aβ ‘strains’ and whether they can be linked to phenotypic variability of AD and/or cerebral β‐amyloidosis, as reported for prion strains in prionoses [18–20], remains to be shown. However, in vitro, the neurotoxicity of Aβ fibrils is dependent on their molecular and structural composition, including their Aβ40/42 ratio [15, 1997, 2003]. In vivo, it has been shown that subtle changes in the Aβ40/42 ratio have profound effects on neurotoxicity . Finally, the Aβ40/42 ratio correlates with disease onset in familial AD [34, 35].
Cerebral β‐amyloidosis is likely initiated by stochastic Aβ seed formation, with subsequent propagation and spreading [6, 36]. The finding that variants of Aβ seeds might govern the type (and possibly toxicity) of Aβ aggregates might at least partly explain the heterogeneous morphology, pathogenicity and progression of Aβ lesions and associated pathologies in AD. The identification of factors that influence the conformational characteristics of the initial seed might thus have therapeutic implications.
Detailed methods can be found in the Supplementary Information Online.
Mice and brain extracts. APP23 and APPPS1 transgenic mice develop first amyloid deposits at 6–8 months (APP23) and 2–4 months (APPPS1) of age. In both strains, Aβ deposits develop first in neocortex and later in the hippocampus [21, 22]. Brain extracts were prepared from 27–28‐month‐old APP23 and 16–22‐month‐old APPPS1 mice (10–20 ng Aβ/μl for both extracts). Injections were done into the hippocampus/dentate gyrus (AP −2.5 mm, L+/−2.0 mm, DV −1.8 mm) .
Immunohistochemistry and stereology. Sections were stained with Congo red and polyclonal antibody to Aβ . Induced Aβ deposition was quantified on a set of every 12th systematically sampled coronal section throughout the dentate gyrus . Total Aβ‐load was determined by calculating the areal fraction (percentage) occupied by Aβ‐staining; the compact plaque load by calculating the areal fraction occupied by Congo red.
tPTAA staining and emission spectra. Spectra were collected from tPTAA‐stained sections  with a LSM 780 or 510 META (Carl Zeiss, Jena, Germany) with an argon 488 nm laser, and a Leica DM6000 B fluorescence microscope (Leica, Wetzlar, Germany) equipped with a SpectraCube (optical head) module (Applied Spectral Imaging, Migdal Ha‐Emek, Israel). SpectraView 3.0 EXPO software (Applied Spectral Imaging) was used for images and selection of spectral regions. Spectra were collected from eight spots within 15–20 plaques/animal. Twisted separated LCP molecules emit light at 530–540 nm, whereas planar, stacked (aggregated) LCP multimers emit light at 630–650 nm. The ratio of the intensity of the emitted light at 534 and 631 nm was used as a parameter for spectral distinction of different plaques .
Immunoblotting of Laser‐Dissected Tissue. Laser‐microdissected (MicroBeam, P.A.L.M., Bernried, Germany) patches of the dentate gyrus (see Fig 1J) were cut from sections adjacent to Aβ‐immunostained sections in which the location of the Aβ induction was confirmed. Bicine‐Tris urea SDS–PAGE immunoblotting was used to separate Aβ40 and Aβ42 using antibody 6E10 specific to human Aβ . Densitometric values of band intensities were analysed to calculate the Aβ40:Aβ42 ratio using ImageJ, version 1.42q (http://rsb.info.nih.gov/ij).
Conflict of Interest
The authors declare that they have no conflict of interest.
Source data for Supplementary Figures
We thank Matthias Staufenbiel (Basel) and Lary Walker (Atlanta) for comments to this manuscript; Ulla Welzel, Sarah Fritschi, Uli Obermüller, Claudia Schäfer (Tübingen), and Sofie Nyström (Linköping) for experimental help; Monika Schütz and Ingo Autenrieth (Tübingen) for help with the laser‐microdissection. Supported by BMBF (ERA‐Net NEURON—MIPROTRAN), European Union FP7 HEALTH (LUPAS), the Swedish Foundation for Strategic Research (SSF), and an ERC starting independent researcher grant (K.P.R.N.).
Author contributions: Y.S.E. and M.J. designed the research; G.H., Y.S.E., F.L. S.A.K., R.N., A.Å., P.H. and K.P.R.N. performed the research; G.H., Y.S.E., A.N., K.P.R.N., P.H. and M.J. analysed data and prepared figures; M.J., K.P.R.N., P.H. wrote manuscript.
- Copyright © 2013 European Molecular Biology Organization