Soluble amyloid‐β dimers are resistant to amyloid‐β prion conversion in vivo suggesting antiprion properties

For Alzheimer's disease (AD), one neuropathological hallmark is the accumulation of insoluble amyloid-beta (Aβ) peptide, a protease-processed fragment of amyloid precursor protein (APP), into extracellular amyloid plaques in the brain. Pathological conformations and multimerisation of Aβ are assumed to be a critical initial step in Aβ fibrilisation culminating in the massive deposition of extracellular Aβ plaques. Such plaques may arise de novo from spontaneous seeds or, in addition, after the disintegration of existing fibrils into smaller ones, spread through providing seeds and thus accelerating Aβ-related neuropathology, termed Aβ prions1 in analogy to protein-based replication of macromolecular fibrils as seen classical prion replication, for example, in Creutzfeldt Jakob disease.2 Animal experiments established that intracerebral injection of transgenic APP23 mice, a mouse model reflecting Aβ pathology of AD, with brain extracts from AD patients containing Aβ seeds accelerated the formation of Aβ plaques leading to astrogliosis.1 In the last 20 years, extensive studies have established the existence of different Aβ prion strains3 and structurally different Aβ filaments in vivo,4 with potential relevance to understanding propagation in human AD.5 Because Aβ plaques are the consequence of decade-long aggregation of Aβ oligomer precursors, a common assumption is that Aβ oligomers accelerate Aβ plaque pathology by providing abundant, “ready-made” precursors. The tgDimer mouse is a transgenic mouse that expresses human APP with the Swedish mutation and an artificial mutation (S679C), replacing a serine at position 8 of Aβ with a cysteine, resulting in the generation of exclusive, stable and neurotoxic Aβ-S8C dimers.6 The Aβ-S8C dimer was designed to mimic wild-type Aβ dimers by covalently crosslinking Aβ monomers for investigating Aβ dimer effects independent of other Aβ species.7 In fact, long-lived and stable Aβ dimers have also been demonstrated to build from wild-type monomers and to be abundant in vivo.8 Even though Aβ-S8C dimers maintain the ability to associate with Aβ plaques,6, 9 tgDimer mice do not develop spontaneous Aβ plaques during their lifetime but display early cognitive deficits resembling early AD symptoms.6 We have previously demonstrated that Aβ dimers from the tgDimer mouse inhibit Aβ seeded nucleation but not Aβ plaque growth in a genetic experiment when tgDimer mice were crossed to tgCRND8 mice, a model for AD developing Aβ plaques at as early as 3 months of age.9, 10 Furthermore, in the cell-free in vitro thioflavin T assay, Aβ-S8C dimers inhibited seeded nucleation of wild type Aβ in a dose-dependent manner.9 In order to investigate whether, on top of the above findings, Aβ dimers could be a substrate of or modulate seeded nucleation, that is, prion replication of Aβ fibrils, we inoculated tgDimer mice with insoluble Aβ seeds, backed by various control conditions. We crossed the tgDimer mice with GFAP-luc mice, which express luciferase when GFAP-dependent astrogliosis is turned on after Aβ plaques emerge, to enable in vivo longitudinal monitoring of Aβ-plaque-associated astrogliosis.11 These double transgenic tgDimer/GFPAP-luc mice were inoculated with brain homogenates from plaque-bearing tgCRND8 mouse10 crossed to tgDimer mice, in order to account for any potential prion strain resistance (Figure S1).6 Inoculated tgDimer/GFAP-luc mice did not develop astrogliosis or cognitive deficits during their lifetime compared with the negative controls (Figures 1A,B and S3). A parallel-inoculated positive control, tgAPP23/GFAP-luc mice, expectedly, developed astrogliosis from around 9 months on, accelerated by the inoculum (Figure 1A) and clearly 4 months before the emergence of spontaneous Aβ plaques (Figure 1A), similarly to what has been reported several times before.6, 12 Cognitive deficits in these mice, measured by automated reward-related learning that has the advantage of minimal test interference by stressful animal handling,13 started from month 18 on exclusively in tgAPP23/GFAP-luc mice (Figure 1B). The later time point of this cognitive deficit in reward-related learning in tgAPP23 mice, contrasting to the much earlier deficit in hippocampus-dependent spatial learning,14 can be attributed to the more distributed anatomical network involved (including ventromedial prefrontal regions, anterior cingulate cortex, amygdala and ventral striatum) that seems more resilient to neuropathological lesions. No differences in general activity, measured as total distance moved and the total number of entries, in the reward-related learning task were observed in the mice (Figure S2). Histological analysis for Aβ confirmed that Aβ fibril-inoculated tgDimer mice do not form Aβ plaques even if the right template is provided, compared with tgAPP23/GFAP-luc mice, which displayed Aβ plaques throughout the brain (Figure S3). Similarly, no astrogliosis was observed in the brains of Aβ-inoculated tgDimer mice, whereas the brains of tgAPP23 mice showed strong astrogliosis (Figure S3). The lack of induction of the GFAP promoter and the absence of astrogliosis (Figure S3) indicate that the presence of insoluble Aβ is necessary for astrogliosis. These results are in accordance with our previous findings that even though Aβ-S8C dimers are fundamentally able to associate with insoluble, wild-type Aβ both in vivo and in vitro,6, 9 they delay wild-type Aβ aggregation and decrease seeded nucleation in vivo in a genetic cross with tgCRND8 mice.9 Our findings here indicate that Aβ-S8C dimers also resist Aβ prion propagation by seeded nucleation in vivo. The discovery that Aβ would propagate in a prion-like manner in vivo was, at the time, a surprise, but it is now clear that protein-templated conformational replication is far more common than initially perceived. In addition, the mainstream assumption is that Aβ aggregation proceeds mainly in one direction towards the successive building up of oligomers from monomers to fibrils.15 The yeast prion systems have been exemplary in investigating molecular mechanisms of prion biology and showed that an intricate system of both homologous and heterologous factors regulate prion spreading, the inhibiting ones termed antiprions.16 Along this line, our findings indicate that Aβ prion propagation is similarly regulated by homologous antiprions and thus is a process that is actively regulated by homologous and possible heterologous factors, rather than being a passive process spreading through the brain once initiated. From our work, we suggest that certain Aβ oligomers could have a so far underappreciated roles as antiprions for slowing or preventing Aβ propagation and downstream astrogliosis. The end stage of AD where abundant Aβ plaques populate the brain may therefore also be conceived as a complete breakdown of antiprion systems unable to contain the inherent prion propensity of Aβ. E.v.G. performed experiments and data analysis and wrote the original draft. A.M.S. was a major contributor to the experimental set-up and revised the paper. S.S. provided support on behavioural experiments and analysis. B.K. and M.L. assisted with behavioural set-up and data analysis. C.K. designed the project and supervised and revised the paper. All authors contributed to and approved the final manuscript. This research was funded by a grant from Deutsche Forschungsgemeinschaft (KO 1679/10-1) to C.K. Open Access funding enabled and organized by Projekt DEAL. None. Animal experiments were performed in accordance with the German Animal Protection Law and were authorised by local authorities (LANUV NRW, Germany). The peer review history for this article is available at https://publons.com/publon/10.1111/nan.12895. All original data will be made available upon reasonable request. Figure S1. 10% brain extracts from 5 months old tgCRND8/tgDimer mice in sterile phosphate-buffered saline (PBS) were centrifuged at 3000 x g. 5 μl of the supernatant and 2.5 ng of synthetic Aβ42 was separated on a 16.5% Tris-Tricine gel and transferred to a nitrocellulose membrane. Immunoblotting for Aβ (antibody 4G8, 1:500) revealed the presence of Aβ monomers and dimers in tgCRND8/tgDimer brain extracts. Bands visualised around 15 kDa and 25 kDa are APP-CTF signals [1]. Figure S2. Total distance moved and total number of entries to assess the general activity of the mice during reward-related learning task. No significant differences were observed in the total distance moved and total entries made during the task performed at 6, 13 and 18 months of age. Figure S3. Representative immunohistochemical images of the cortex of 18 months old mice stained with anti-Aβ (biotinylated IC-16) and GFAP antibody. Aβ plaques were only observed throughout the sections in tgAPP23/GFAP-luc mice, and not in the other genotypes Cortices of inoculated tgDimer and Wildtype mice showed little astrogliosis, compared to Aβ inoculated tgAPP23/GFAP-luc mice, which showed strong astrogliosis in the cortex. Scale bar = 200 μm. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.