Signaling via the p75 neurotrophin receptor facilitates amyloid-β-induced dendritic spine pathology

Synapse and dendritic spine loss induced by amyloid-β oligomers is one of the main hallmarks of the early phases of Alzheimer’s disease (AD) and is directly correlated with the cognitive decline typical of this pathology. The p75 neurotrophin receptor (p75NTR) binds amyloid-β oligomers in the nM range. While it was shown that µM concentrations of amyloid-β mediate cell death, the role and intracellular signaling of p75NTR for dendritic spine pathology induced by sublethal concentrations of amyloid-β has not been analyzed. We describe here p75NTR as a crucial binding partner in mediating effects of soluble amyloid-β oligomers on dendritic spine density and structure in non-apoptotic hippocampal neurons. Removing or over-expressing p75NTR in neurons rescues or exacerbates the typical loss of dendritic spines and their structural alterations observed upon treatment with nM concentrations of amyloid-β oligomers. Moreover, we show that binding of amyloid-β oligomers to p75NTR activates the RhoA/ROCK signaling cascade resulting in the fast stabilization of the actin spinoskeleton. Our results describe a role for p75NTR and downstream signaling events triggered by binding of amyloid-β oligomers and causing dendritic spine pathology. These observations further our understanding of the molecular mechanisms underlying one of the main early neuropathological hallmarks of AD.


Results
The genetic ablation of the p75 neurotrophin receptor protects hippocampal neurons against Aβ 1-42 mediated spine pathology. It is well established that the long-term treatment of both cell lines and primary neurons with amyloid-β (Aβ  ) oligomers in the µM range results in neuronal degeneration and cell death per apoptosis 30 . These effects have been shown to depend on the specific binding of the Aβ  oligomers to the p75 NTR receptor 14,16 and the activation of downstream signaling. Here we were primarily interested in analyzing the sublethal effects of acute Aβ 1-42 /p75 NTR interactions in the nM range. By this means we analyzed possible intracellular signaling events which might lead to dendritic spine alterations. Dendritic spine pathology is one of the principal symptoms of Aβ 1-42 accumulation in early AD.
Initially, f-eGFP expressing DIV14 primary hippocampal neurons, obtained from C57BL/6 wild type animals were incubated with increasing amounts of Aβ  in the nM range and their impact on dendritic spine density and morphology was analyzed after six hours ( Fig. 1A-D). This time point has been chosen since it has been reported previously to cause significant dendritic alterations upon treatment with Aβ oligomers 31 . While incubations with 10 and 50 nM Aβ 1-42 oligomers resulted in no or only a mild reduction in dendritic spine density when compared to the control condition, application of 100 nM Aβ 1-42 oligomers resulted in a significantly lower dendritic spine density (Fig. 1B). This decrease was even stronger upon application of 500 nM Aβ 1-42 oligomers (Fig. 1B), indicating a dose-dependent effect. The decrease in spine density was associated with structural changes in the remaining dendritic spines. Particularly, a significant increase in dendritic spine length (Fig. 1C) and a significant decrease in the spine head width (Fig. 1D) was observed using the higher concentrations of Aβ  oligomers (100-500 nM). Interestingly, while the application of 50 nM Aβ 1-42 oligomers did not alter dendritic spine structure, 10 nM Aβ 1-42 oligomers caused a small increase in the spine head width (Fig. 1D) that coincided with a decrease in the length (Fig. 1C) of treated spines.
In summary, these data demonstrate that a 6-h treatment of wild type mature primary hippocampal neurons with Aβ 1-42 oligomers causes a reduction in dendritic spine density associated with a switch toward an immature phenotype of the remaining dendritic spines. These observations are in line with previous reports 27 . Interestingly, low concentrations of Aβ 1-42 (10 nM) seem to promote dendritic spine maturation. It should be noted that a rather brief incubation for 15 min of primary hippocampal neurons with amyloid-β oligomers had no effect on both the density and the morphology of dendritic spines ( Supplementary Fig. S2).
Intriguingly, when increasing amounts of Aβ  oligomers were added to f-eGFP expressing mature primary hippocampal neurons obtained from p75 NTRexon-IV knockout mice 29 , the alterations observed in wild type mice were completely prevented (Fig. 1A, E-G, Supplementary Fig. S2). In fact, when compared to the control conditions even at the highest concentration of Aβ 1-42 oligomers tested (i.e. 500 nM) no changes in dendritic spine density (Fig. 1E) as well as in dendritic spine length (Fig. 1F) and head width (Fig. 1G) were detected. Considering that p75 NTR is not the only reported neuronal receptor serving as a ligand for amyloid-β 32 , these results indicate that it must play a major role in triggering Aβ 1-42 -induced alterations in dendritic spine morphology.
Overexpression of p75 ntR leads to higher sensitivity of primary hippocampal neurons to the deleterious effects of Aβ 1-42 on dendritic spines. Next, we evaluated the impact of the neuronal overexpression of full-length variants of p75 NTR on mediating amyloid-β induced dendritic changes. For this purpose, primary hippocampal neurons were either transfected with an expression vector encoding the rat p75 NTR together with f-eGFP to visualize the neurons (Fig. 3A) 9 or with the eGFP-tagged human p75 NTR (Fig. 3B) 33 . As a control, neuronal cultures were transfected with the f-eGFP-expressing plasmid alone. Subsequently, the neuronal cultures were treated on DIV14 with increasing concentrations of Aβ 1-42 oligomers for 6 h and the spine density and morphology of individual neurons were evaluated. Since the sole over-expression of p75 NTR causes alterations in dendritic spine morphology 9 , the results obtained from amyloid-β treated dendrites of each experimental group were normalized to the corresponding control samples without amyloid-β stimulation ( Fig. 3C www.nature.com/scientificreports/ Similar to the results described above in Fig. 1, the neurons that expressed f-eGFP alone and were treated with Aβ 1-42 oligomers showed a dose-dependent decrease in dendritic spine density when compared to vehicle controls (Fig. 3C). Likewise, we observed an increase in spine length (Fig. 3D) as well as a decrease in spine head width (Fig. 3E) in those neurons that had received amyloid-β oligomers. Remarkably, the overexpression of both human and rodent p75 NTR isoforms led to a significantly stronger decrease in the density of dendritic spines after both 100 and 500 nM amyloid-β treatment when compared to f-eGFP-expressing control neurons, which had been treated with identical amounts of amyloid-β oligomers (Fig. 3C). The application of Aβ 1-42 oligomers, however, had no comparable effects on the length (Fig. 3D) nor on the head width (Fig. 3E) of the remaining dendritic spines of p75 NTR overexpressing neurons. The ablation of p75 NTR rescues acute amyloid-β induced changes in dendritic spine morphology. Primary hippocampal neurons from wild type (A, B-D) and p75 NTR ko (A, E-G) mice were incubated with vehicle (PBS; 0) or with 10, 50, 100, or 500 nM Aβ 1-42 oligomers for 6 h. (A) shows representative images as maximum intensity projections of dendrites of f-GFP + expressing neurons, which were co-stained with MAP2 to visualize the entire dendritic network. An evaluation of the spine density (B, E), spine length (C, F), and spine head width (D, G) is shown in the graphs where dendritic segments with a length of ≥ 100 µm were used for analysis for wild type and p75 NTR ko neurons. Columns represent mean values + SEM and dots mark data points of individual neurons. The number of evaluated cells, obtained from three independent sets of experiments, is given in each column of the graphs. Statistical significance was tested using a one-way ANOVA with Sidak posttest. ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001. Scale bar: 20 µm. www.nature.com/scientificreports/ The localization of amyloid-β at dendritic spines is dependent on the levels of p75 ntR . Our data suggest that p75 NTR is an important membrane component mediating the effects of Aβ 1-42 oligomers on dendritic spines. Hence, we sought to evaluate in more detail the direct interactions of amyloid-β with dendritic spines depending on the p75 NTR expression levels. Initially, we compared the colocalization of amyloidβ-immunoreactive puncta with dendritic spine heads in f-eGFP expressing wild type and p75 NTR knockout primary hippocampal neurons after a short-term incubation of 15 min with increasing concentrations of Aβ 1-42 oligomers (Fig. 4A,B). We observed that both in wild type and p75 NTR ko neurons the number of amyloidβ-immunoreactive puncta associated with dendritic spines increased in a concentration-dependent manner.
In order to analyze more long-term effects of the interaction of Aβ 1-42 oligomers with dendritic spines, we extended the Aβ 1-42 incubation times to 6 h ( Fig. 4C). At this time point, the differences in the relative amounts of spine-associated amyloid-β-immunoreactive puncta between wild type and p75 NTR ko neurons were less pronounced and did not reach any longer statistical significance (27.7 ± 2.3%, n = 28 vs. 21.2 ± 1.5%, n = 33, respectively, for treatment with 500 nM Aβ 1-42 ; mean ± SEM, p = 0.08, Fig. 4C). The overexpression of rodent or human p75 NTR in neurons, however, significantly augmented the amount of amyloid-β-immunoreactive puncta localized at dendritic spines ( Fig. 4C) when compared to wild type. Interestingly, the overexpression of the human p75 NTR resulted in a higher colocalization than the rodent one (Fig. 4C). This data emphasizes that p75 NTR serves as an important receptor for amyloid-β binding, in particular for the initial phases of amyloid-β/receptor interactions. However, it also confirms the notion that other amyloid-β receptors are located at dendritic spines of hippocampal neurons.
Mechanistic insight: amyloid-β induced activation of RhoA requires presence of p75 ntR . We documented amyloid-β-induced alterations of dendritic spines of primary neurons that could be rescued by the removal of p75 NTR . Moreover, we show that Aβ 1-42 is localized at dendritic spines and that the levels of neuronal p75 NTR expression influence strongly the association of amyloid-β with spines. Therefore, we next searched for a potential downstream mechanism explaining the amyloid-β/p75 NTR mediated changes in spine morphology. Since it has been reported that p75 NTR is capable to activate RhoA 8 at a fast time scale 34 and RhoA signaling is implicated in remodeling the actin cytoskeleton, including the modulation of spine dynamics 35,36 within minutes 37 , we hypothesized that Aβ 1-42 /p75 NTR interactions trigger the activation of the RhoA/ROCK pathway, ultimately leading to changes in the actin composition and dendritic spine architecture.
We first analyzed if amyloid-β is capable to activate RhoA in murine primary hippocampal cultures and whether this cellular event is influenced by the presence of p75 NTR . Hence, we added Aβ 1-42 oligomers for 10 min to primary neurons from wild type and from p75 NTR ko cultures and performed an ELISA-based assay www.nature.com/scientificreports/ to determine the activation state of RhoA molecules, i.e. its GTP-bound form 38 . Intriguingly, we observed a significant activation of RhoA for wild type hippocampal cultures upon treatment with Aβ 1-42 oligomers (Fig. 5). Interestingly, the RhoA activation level in vehicle-treated p75 NTR ko cultures was lower, albeit not significantly (p = 0.0961, Two-way ANOVA with Tukey post-test), than in vehicle-treated wild type neurons (Fig. 5). Importantly, the application of Aβ 1-42 oligomers to p75 NTR -deficient neuronal cultures did not result in any measurable RhoA activation (Fig. 5). This data indicate that Aβ 1-42 oligomers are able to cause the p75 NTR -dependent activation of RhoA in primary hippocampal cultures.
The inhibition of p75 ntR -mediated RhoA/RocK signaling rescues amyloid-induced changes in dendritic spine morphology. Next, we utilized inhibitors known to prevent the activation of p75 NTR downstream signaling to RhoA to test the hypothesis that p75 NTR is able to mediate the effect of Aβ 1-42 oligomers on dendritic spine structure by directly controlling the activation of the RhoA. The TAT-Pep5 fusion peptide has been described to be able to enter the cell and to specifically inhibit p75 NTR -mediated RhoA activation 39 .
In addition, we also employed the Rho-associated coiled-coil containing protein kinase (ROCK) inhibitor Y-27632 40 . Upon a 6-h treatment with 500 nM Aβ 1-42 oligomers, f-eGFP expressing primary hippocampal neurons showed a significant decrease in dendritic spine density (Fig. 6A,B) associated with significant morphological changes in spine length (Fig. 6A,C) and spine head width (Fig. 6A,D), similar to the results described above (see Fig. 1). Interestingly, a pretreatment of the neuronal cultures either with the TAT-Pep5 peptide, inhibiting the p75 NTR -mediated RhoA activation, or with the pan-ROCK inhibitor Y-27632 both protected hippocampal neurons from the Aβ 1-42 oligomer-induced alterations in dendritic spine number and structure (Fig. 6A-D). Specifically, preventing the p75 NTR mediated activation of RhoA or inhibiting ROCK blocked significant changes www.nature.com/scientificreports/ in either dendritic spine density (Fig. 6A,B), spine elongation (Fig. 6A,C), or spine head diameter (Fig. 6A,D) when compared to the corresponding controls.
Involvement of p75 ntR amyloid-β induced changes in the actin filament. Above we analyzed the contribution of the RhoA/ROCK pathway in transducing the interaction of Aβ 1-42 oligomers with the p75 NTR receptor. Finally, we examined whether these p75 NTR mediated signaling events influenced the actin cytoskeletal network in dendritic spines. First, we used phalloidin staining to compare the levels of F-actin within dendritic spines from individual wild type and p75 NTR knockout animals (Fig. 7A,B). Under basal conditions, only low levels of F-actin could be identified in the dendritic spines of both genotypes (Fig. 7A,B).
Next, we sought to assess the effect of the application of Aβ 1-42 oligomers on the F-actin levels within spines by incubating f-eGFP expressing primary hippocampal neurons with 500 nM Aβ 1-42 for 10 min. We observed gives a representative example of the distribution of amyloid-β immunoreactive puncta (red) at a dendritic segment of f-eGFP-transfected neurons (green), which either were obtained from wild type (left) or p75 NTR ko (middle) mice, or were co-transfected with an expression vector encoding rodent p75 NTR (right). Arrows indicate dendritic spines colocalized with amyloid-β-immunoreactive puncta. The red puncta that did not colocalize with dendritic f-eGFP labeled spines stem from out-of-focus amyloid-β oligomers binding to neuronal processes of non-transfected neurons. Treatment of the primary hippocampal cultures (DIV 14) with Aβ 1-42 oligomers was for 15 min. The graphs illustrate the quantification of the proportion of amyloid-β-positive dendritic spines of neurons, which received vehicle or increasing amounts of Aβ 1-42 oligomers for 15 min (B) or 6 h (C). Hippocampal neurons were wild type (B, C; blue), had knockout for p75 NTR (B, C; green), or overexpressed human (C, red) or rat p75 NTR (C, orange). Each bar shows the mean values + SEM. To compare the means from wild type and p75 NTR knockout neurons in B, One-way ANOVA with Sidak post-test was used. In C, a Two-way ANOVA with Tukey post-test was applied. The numbers at the bottom of each column give the amount of analyzed neurons, which were obtained from three independent experiments. Statistical significances as indicated or **** = p < 0.0001; ns = not significant. Scale bar in (A) is equivalent to 5 µm. www.nature.com/scientificreports/ that this brief treatment caused a significant increase in the F-actin intensity specifically in those dendritic spines colocalized with Aβ 1-42 immunoreactive puncta (Fig. 7C,E). Importantly, the levels of F-actin within dendritic spines were strictly dependent on the presence of Aβ 1-42 since dendritic spines of the same neurons lacking Aβ 1-42 immunoreactive puncta did not show any significant increase in F-actin (Fig. 7C,E). Interestingly, in p75 NTR -deficient neurons the F-actin intensity within amyloid-β + dendritic spines increased only slightly, but not significantly, when compared to the control condition (p = 0.08, Fig. 7D,E). Furthermore, the F-actin intensity in spines of p75 NTR ko neurons was clearly, albeit also not significantly, lower than in wild type cells (p = 0.09; Fig. 7E).
In summary, our data suggest that the interaction of Aβ 1-42 oligomers with the p75 NTR triggers an intracellular RhoA/ROCK signaling cascade that likely leads to alterations in actin polymerization in dendritic spines of primary hippocampal neurons. The ablation of p75 NTR largely prevents the activation of these p75 NTR mediated intracellular signaling events of the RhoA/ROCK pathway and the actin polymerization, thereby protecting p75 NTR ko neurons from the Aβ 1-42 -induced dendritic spine pathology.

Discussion
Understanding the role of oligomeric Aβ 1-42 in inducing the dendritic spine pathology most likely causing the symptoms typical of the initial phases of AD requires to elucidate the signaling mechanisms activated by amyloid-β at synapses. While Aβ has been shown to bind to several membrane receptors, our results provide strong evidence for a crucial role of p75 NTR signaling in the Aβ 1-42 -induced structural alterations at synapses directly. In this study, we show that p75 NTR signaling is required for the dendritic spine loss and structural changes observed in mature primary hippocampal neurons upon application of increasing nM concentrations of Aβ 1-42 oligomers. Moreover, we show that the activation of p75 NTR upon Aβ 1-42 oligomer treatment results in a RhoA-ROCK-mediated stabilization of the actin spinoskeleton followed by spine loss as well as shrinking and elongation of the remaining spines.
While Aβ plaques are the typical hallmark of the advanced phases of AD it is now rather believed that it is the soluble Aβ 1-42 oligomers that, binding with high affinity at synaptic sites, impair the function and structure of synapses at the onset of the disease 41 . Soluble Aβ 1-42 oligomers are found in several AD mouse models and in human patients and their levels correlate with synaptic loss 42 , impairment of synaptic transmission and plasticity 43,44 as well as AD-associated cognitive decline 45,46 . While in vitro higher concentrations of Aβ oligomers, in the µM range, are well known to induce neuronal death by apoptosis 30,47,48 , lower pM to nM Aβ concentrations have been shown to bidirectionally regulate activity-dependent synaptic plasticity 49 . However, only few studies so far addressed the sublethal Aβ effects on the architecture of neurons. Here we show that Aβ 1-42 oligomers accumulate at dendritic spines of wild type hippocampal neurons within minutes and reside there up to hours after starting the treatment. Likewise, synthetic Aβ oligomers were previously shown to specifically accumulate at excitatory synapses in an activity-dependent manner 31,50,51 . Since most of the excitatory synapses in the mammalian brain form contacts onto dendritic spines these previous observations are in good agreement with our results. Moreover, we observed a concentration-dependent decrease in dendritic spine density in primary hippocampal neurons treated with synthetic Aβ 1-42 oligomers. In fact, dendritic spine loss is one of the main hallmarks of AD [52][53][54][55] . In addition, we observed that the remaining dendritic spines show an increase in their length and a decrease in the head width, typical of an immature phenotype. These findings are supported by results of previous studies showing a rapid decrease in the density of both synaptic structures and dendritic spine in response to both naturally secreted and synthetic Aβ peptides 27,31,51,56 and identify a strong and reliable phenotype to be www.nature.com/scientificreports/ used to analyze the downstream signaling involved in this processes. Importantly, our results show that under the experimental conditions used in this study, the number of cleaved caspase-3 and TUNEL-positive neurons in hippocampal cultures is extremely low and especially it is not increased upon a 6 h treatment with 500 nM Aβ 1-42 oligomers. This observation strongly indicates that the dendritic spine phenotype observed here is not the consequence of Aβ-induced apoptosis but rather of a sublethal effect of Aβ oligomers. Several surface receptors have been identified to bind Aβ oligomers at the neuronal surface and trigger a series of intracellular signaling pathways negatively affecting their survival and function 57 . However, the role of several of these receptors is still controversial and their downstream signaling mediating the synaptic effects of Aβ 1-42 oligomers are largely unknown. We show here strong evidence for a role of the p75 NTR in mediating the effects of Aβ 1-42 oligomers on the neuronal architecture in a specific manner, since a complete p75 NTR knockout significantly reduces the accumulation of Aβ 1-42 to dendritic spines and completely rescues spine loss and changes in spine architecture. Conversely, p75 NTR overexpression strengthens both Aβ 1-42 accumulation and spine loss without exacerbating the changes in dendritic spine architecture observed under control conditions. A link between p75 NTR signaling and AD is supported by various observations both in human patients 58,59 and in mouse models 60,61 . Furthermore, p75 NTR has been shown to bind Aβ 12 including its oligomeric form 22 and to be required for the Aβ-induced death of primary hippocampal neurons both in vitro and in vivo 16 . Upon treatment www.nature.com/scientificreports/ with Aβ peptides in the µM range p75 NTR has been shown to mediate the activation of downstream signaling cascades involving the activation of JNK 62 and leading to cell death 12,14,16 . However, whether nM concentrations of Aβ 1-42 oligomers may activate a different intracellular cascade resulting in changes in neuronal and synaptic structure rather than cell death is not known. Supporting this notion is the observation that low concentrations of Aβ as well as sAPPα have been shown to induce neurite outgrowth in cerebellar and cortical neurons in a p75 NTR -dependent manner 22,63 . On the other hand, p75 NTR has been shown to also function as a signal transducer for neurite outgrowth inhibitory molecules 64 and to negatively modulate dendritic spine density and their architecture in healthy adult hippocampal neurons 9 supporting our current observations upon Aβ treatment. However, intracellular signaling linking Aβ interactions with p75 NTR to dendritic spine alterations are still unknown. Here we show that Aβ-dependent p75 NTR signaling results in the fast activation of RhoA, triggering the activation of ROCK and causing an increase in the F-actin levels within individual spines. These events require p75 NTR as they are prevented in p75 NTR knockout neurons as well as in neurons treated with the peptide TAT-Pep5, known to act as a downstream signaling silencer by disrupting the recruitment of Rho-GDI to p75 NTR and thereby inhibiting the activation of RhoA 8 . In this respect, it is noteworthy that recent studies described the involvement of the www.nature.com/scientificreports/ ROCK-LIMK pathway in Aβ-induced dendritic spine degeneration 65,66 . Moreover, Aβ-dependent synaptotoxicity in cortical neurons involves a dysregulation of the actin cytoskeleton characterized by a cofilin driven increase in actin stability 67 . Taken together our observations provide a strong support for a role of p75 NTR in mediating the negative effects of Aβ 1-42 oligomers at dendritic spines and we here specifically describe a downstream signaling pathway that involves RhoA-ROCK activation and regulates the stability of the actin cytoskeleton within spines. Thus, our work contributes to a better understanding of the signaling mechanisms activated upon a treatment with Aβ 1-42 oligomers and resulting in the typical dendritic spine pathology observed in the early phases of AD and opens new avenues for further treatment option at the very early onset of AD.

Materials and methods
Mice. Both wild type (WT) and p75 NTRexonIV knockout (p75 NTR ko) 29 mice used in this study were on the C57Bl/6J genetic background. All procedures concerning animals were performed in accordance with the animal welfare representative of the TU Braunschweig and were approved by the LAVES (Oldenburg, Germany, Az. §4 (02.05) TSchB TU BS).
cell cultures. Primary hippocampal cultures were prepared from WT or p75 NTR KO mice at embryonic day 18 as previously described 68 . Following rapid decapitation, isolated embryo brains were immediately immersed in Gey's Balanced Salt Solution (GBSS) supplemented with glucose and adjusted to pH 7.3. Under a dissection microscope, the hippocampi were isolated in ice-cold GBSS and their dissociation was achieved by incubation in Trypsin-EDTA at 37 °C for 30 min followed by mechanical dissociation until a single cell suspension was obtained. Subsequently, the cells were re-suspended in NB + medium comprising Neurobasal medium (Gibco) supplemented with 2% B27 (vol/vol), 1% N2 and 0.5 mM Glutamax (Gibco) and plated at a density of 7 × 10 4 / cm 2 on coverslips (12 mm) or at 450,000 cells/petri dishes (diameter 3 cm), previously coated with poly-L-lysine (Sigma). The cells were incubated at 37 °C with 99% humidity and 5% CO 2 until usage at DIV14. A medium change was done once a week by exchanging 20% of the medium with fresh NB + medium.
Aβ peptide preparation. Soluble Aβ 1-42 oligomers were prepared as previously described 69  Immunofluorescence. Following a 10 min fixation using cold paraformaldehyde (4% in PBS), the cells were incubated for 1 h at room temperature (RT) in a permeabilizing and blocking buffer containing 0.2% Triton X-100 and 1.5% Normal Goat Serum (NGS, Merck) in PBS. Next, the cells were incubated overnight at 4 °C on a rocker with primary antibodies diluted in the permeabilizing/blocking solution. The following primary antibodies were used: anti-β-Amyloid peptide (either clone BAM-10, Sigma, 1:5,000, or clone 6E10, BioLegend, 1:500); mouse anti-microtubule-associated protein 2 (MAP2, 1:1,000, Sigma); rabbit anti-cleaved caspase-3 (1:500, Millipore). Secondary antibody staining was done in the dark at room temperature for 2 h using the following antibodies: anti-mouse IgGFc γ Subclass-I (1:500, Jackson ImmunoResearch Labs); anti-mouse IgG (H + L) Alexa-fluor 647 (1:500, Jackson ImmunoResearch Labs); anti-rabbit IgG (H + L) Cy3 (1:500, Jackson ImmunoResearch Labs). Finally, the cover slips were washed and mounted onto glass slides using anti-fading Fluoro-Gel embedding medium (Electron Microscopy Sciences). For the visualization of F-actin, some sets of cultures were stained with Alexa-fluor-Phalloidin 647 (ThermoFisher Scientific) diluted 1:40 in PBS following the incubation with the secondary antibody also in the dark for 1 h RT. To minimize actin disruption, the Tritonpermeabilization duration was reduced to 15 min. The subsequent steps remained unchanged. www.nature.com/scientificreports/ In one set of experiments, DNA fragmentation in dying neurons was detected using the Dead End fluorimetric TUNEL system (Promega) according to the protocol provided. Briefly, after incubation with the secondary antibodies and washes in PBS, the coverslips were incubated in equilibration buffer for 10 min followed by addition of incubation buffer containing terminal deoxynucleaotidyl transferase (TdT) and fluorescein-12-dUTP (nucleotide mix). The coverslips were stained for 60 min at 37 °C in a humidified chamber kept in the dark. The reaction was then stopped using SSC buffer and the coverslips were washed in PBS, counterstained with DAPI, and mounted with anti-fading Fluoro-Gel embedding medium (Electron Microscopy Sciences).
RhoA G-LiSA. The G-LISA RhoA activation assay was performed strictly according to the manufacturer's instructions (G-LISA activation assay kit, Colorimetric format, Cytoskeleton Inc.). 9 × 10 5 cells were used for each treatment by pooling cells of two petri dishes (diameter: 3 cm) per treatment. After a 10-min treatment with vehicle or Aβ 1-42 oligomers, cell lysates were prepared rapidly on ice. The lysates were clarified by centrifugation at 10,000×g for 1 min at 4 °C. 30 μl of lysates were used for protein quantification and the remainder was snap-frozen in liquid nitrogen. This entire process was carried out under 10 min and was done sequentially for each test group. Protein quantification was performed by incubating the samples for 1 min at room temperature with the Precision Red Advanced Protein Assay Reagent (Cytoskeleton Inc.) provided by the kit. Absorbance readings were performed with a spectrophotometer at 600 nm. All samples were equilibrated to identical protein concentrations (0.5 mg/ml) and loaded in triplicates onto the plates kept on ice. The samples (in triplicates), the positive RhoA control (duplicate) and the buffer blank (duplicate) were incubated with an anti Rho-antibody and, after multiple washes, with the HRP-conjugated secondary antibody. A color reaction was done by application of freshly prepared HRP A/B detection reagent and the wells were read for absorbance at 490 nm.
Image acquisition and analysis. Hippocampal neuronal cultures stained for MAP2 and cleaved caspase-3 were imaged using a Zeiss Axioplan2 microscope equipped with a 10 × objective (NA 0.3) and Zeiss AxioCam MRm camera. Different excitation wavelengths were used to image MAP2, cleaved caspase-3, and TUNEL in each field of view. Cleaved caspase-3 and TUNEL-positive neurons were counted manually based on their positivity for MAP2 in merged images created using ImageJ software (https ://image j.nih.gov/ij/).
For dendritic spine analysis, f-eGFP-labeled dendrite stretches were imaged using a confocal laser scanning microscope (Olympus, FluoView1000) equipped with a 40 × objective (oil, NA 1.3) with a 5 × digital zoom for a pixel size of 0.107 µm. Z-stack images were acquired with a 0.35 μm step-size and constant laser intensity. Well defined dendritic segments of healthy neurons, identified by a pyramidal cell body and by the absence of irregular membranous protrusions around the soma, were selected for imaging. Images were then deconvolved using AutoQuant (Media Cybernetics) and imported to the ImageJ software for analyzing dendritic spine density and morphology, Aβ colocalization with spines and Phalloidin fluorescence intensity.
Spine density was determined by analyzing z-stacks containing the entire dendritic stretch. The number of spines per unit of length (µm) was calculated by using the segmented line tool of ImageJ to measure dendritic length and the multipoint selection tool (ImageJ) to count spines. Morphometric analysis of dendritic spines was done in the same z-stacks as for the spine density using the segmented line tool of ImageJ to measure spine length (from its base at the dendrite to its tip) and head width (measured at the widest point of the dendritic spine).
A f-eGFP-labelled spine was considered to be colocalized with Aβ-immunoreactive puncta when seen within the same focal plane or one above or below. For analysis of Phalloidin intensity, regions of interest (ROIs) were drawn around spine heads on merged images, colocalized or not with Aβ-immunoreactivity, while being blinded for the phalloidin channel. Only dendritic spines with clearly defined heads were chosen for this analysis. Additional ROIs were drawn on background (no dendrite and staining). Phalloidin intensities for all selected spine heads were averaged and normalized to the averaged background intensity for each image individually. All analysis were performed by an experimenter blinded for the genotype and treatment of each sample.
Data representation and statistical analysis. Data replicates generated from at least three independent experiments were used for statistical analysis. The data were collected and organized in Microsoft Excel. For the generation of graphs and performing the statistical tests, the values were imported into GraphPad Prism 6. Unless otherwise mentioned, all data is reported as mean ± SEM. Student's t-test was used to compare two independent groups. One-way or Two-way ANOVA multiple-comparisons followed by Sidak or Tukey post-hoc tests, respectively, was used for assessing significance between more than two groups. Significance was considered for p value < 0.05; plotting used the notation as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Data availability
All data generated and analyzed during this study are available from the corresponding author on reasonable request.