c-Abl null mice were bred from c-Abl^loxp/c-Abl^loxp and Nestin-Cre⁺ mice, kindly donated by Dr. AJ Koleske (Yale School of Medicine, New Haven, CT, USA). Genotyping was performed using a polymerase chain reaction (PCR)-based screening (Bradley and Koleske, 2009). Primers: Abl1-floxA: 5′-GATGTCTCTACAGGGTTTAAGATTAAGAGCA-3′; and Abl1-floxB: 5′-AGTTAACACACCTCCAGAGTGAGTGCCCT-3′; Cre: B: 5′-GCAATTTCGGCTATACGTAACAGGG-3′; and A: 5′-GCAAGAACCTGATGGACATGTTCAG-3′.

All protocols were approved and followed local guidance documents generated by the ad hoc Chilean committee (CONICYT), and were approved by the Bioethics and Care of Laboratory Animals Committee of the Pontificia Universidad Católica de Chile (Protocol #150721002). We followed the recommendations of the Guide for Care and Use of Laboratory Animals from US Public Health Service.

Hippocampi from c-Abl knockout (c-Abl^loxp/c-Abl^loxp; Nestin-Cre⁺; c-Abl-KO) and their WT siblings (c-Abl^floxo/floxo; WT) mice embryos at day 18 (E18) were dissected, and primary hippocampal cultures were prepared as previously described (Kaech and Banker, 2006). This conditional c-Abl-KO mice model does not present the c-Abl protein in the brain, unlike their WT littermates, although it is present in other tissues (see Supplementary Figure S1). Pregnant mice were anesthetized with CO₂ and subsequently euthanized by cervical dislocation. Cultures were maintained at 37°C in 5% CO₂ with neurobasal growth medium (Invitrogen, Carlsbad, CA, USA), supplemented with B27, 2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen, Carlsbad, CA, USA). On the next day, cultured neurons were treated with 1 μM AraC to prevent glial cell proliferation. Hippocampal neurons were treated with AβOs at 5 μM final concentration for 5 h.

Human synthetic Aβ_1–42 peptide (Genemed Biotechnologies Inc, San Francisco, CA, USA) was suspended in 1, 1, 1, 3, 3, 3 hexafluoro-2-propanol 0.5 mg/ml (Sigma-Aldrich, St. Louis, MO, USA). Peptide samples were vortexed to obtain a homogenous solution, aliquoted into microfuge tubes and lyophilized. The Aβ_1–42 peptide aliquots were resuspended in nanopure water at 200 μM concentration and vortexed briefly. Aggregation was allowed to proceed for 12 h at 4°C following the protocols by Arimon et al. (2005) and Sokolov et al. (2006). To form fluorescent AβOs (AβOs-FITC), synthetic Aβ_1–42 coupled to FITC (Bachem, Torrance, CA, USA) was used. Gel electrophoresis was performed at 4°C in Tris-tricine gels (4% stacking, 10% spacer and 16% resolutive gel) at 50 V to 80 V. Aβ_1–42 species were transferred onto nitrocellulose membrane (0.22 μm pore) for 1 h and 350 mA. Blocking was performed in TBS-3% non-fat milk, and incubated with the primary antibody WO2 raised against Amyloid-β-peptide (MABN10, Millipore, Burlington, MA, USA 1:1,000; Supplementary Figure S2).

Hippocampal neurons from WT and c-Abl-KO embryos (E18) were seeded onto poly-L-lysine-coated coverslips in 24-well culture plates at a density of 10⁴ cells per well. For transfection of pcDNA 3.0 GFP-plasmids, we used the Magnetofection^TM technology with the reagent Neuromag (OZ Biosciences, NM50200) in 18 DIV neurons. After 24 h, these neurons were treated with 5 μM AβOs for 5 h. For dendritic spine quantification, we analyzed GFP-expressing neurons and the co-localization with PSD95 or TRITC-phalloidin (ECM Biosciences, Versailles, KY, USA) to label actin cytoskeleton and examine spine morphology. Coverslips were mounted with mounting medium (DAKO) and then observed using an Olympus IX81 LSM Fluoview or a Nikon Eclipse C2 Ti-E confocal microscope. Images were processed with ImageJ (NIH). Antibodies used for immunofluorescence were anti-Piccolo [epitope 44aII antibody (Cases-Langhoff et al., 1996; Gundelfinger et al., 2016) produced by Viviana I. Torres and Craig C. Garner]; anti-PSD95 (75–028) from NeuroMab, Davis, CA, USA. Dendrite and spine morphology classification was performed according to the method described by Tyler and Pozzo-Miller (2003).

Hippocampal neurons from WT and c-Abl-KO embryos were plated at a density of 10⁶ cells/cm². At different DIV, they were washed and lysed in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris, 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 0.5% deoxycholate, 1% NP-40, and 0.1% SDS) supplemented with protease inhibitors. Cell lysates were centrifuged at 14,000 rpm for 15 min at 4°C. Protein quantification was performed using the Pierce^® BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). The fractions were subjected to SDS-PAGE and transferred onto nitrocellulose membranes (Thermo Fisher Scientific, Waltham, MA, USA). The antibodies used were: anti-βIII-tubulin (AA10 sc80016, Santa Cruz Biotechnology, Dallas, TX, USA), anti-c-Abl (A5844, Sigma-Aldrich, St. Louis, MO, USA); anti-PSD95 (75–028) and anti-SAP102 (75–058), from NeuroMAb Antibodies Inc.

Hippocampal neurons from WT and c-Abl-KO embryos were seeded onto poly-L-lysine-coated coverslips in 24-well culture plates at a density of 5 × 10⁴ cells per well and treated with 5 μM AβOs for 5 h. Cells were fixated and immunostained with active caspase-3 (AB3623, Millipore, Burlington, MA, USA) and Hoechst (33342, Thermo Fisher Scientific, Waltham, MA, USA), to visualize apoptotic nuclei.

Total RNA from 10 DIV WT and c-Abl-KO hippocampal neurons was extracted using TRIzol (Life Technologies, Carlsbad, CA, USA, 15596). Neurons were treated with 5 μM AβOs for 1, 3, 5 and 12 h. Fibrillary forms of the Aβ-peptide (Aβf) were used as control at 5 μM. Total RNA was reverse-transcribed into cDNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Primers: c-Abl Forward: 5′ AGCATCACTAAAGGCGAGAA 3′; c-Abl Reverse: 5′ CACCCTCCCTTCATACCG 3′; GAPDH Forward: 5′ GGGTGTGAACCACGAGAAATA 3′; GAPDH Reverse: 5′ CTGTGGTCATGAGCCCTTC 3′.

We use ImageJ for dendritic spine quantification every 10 μm of each secondary dendrite. We performed three independent experiments (mice cultures) with a total of four to five embryos per condition, and two to five dendrites per neuron analyzed for a total of 13–17 neurons per condition. The total number of spines is indicated in each figure. For spine profile analyses, we used quantifications for each type (Mushroom, Stubby, Thin, Filopodium and Branched) and normalized with the total number of dendritic spines per dendrite analyzed per neuron. All data are presented as Mean ± Mean Standard Error (SEM). Mean, SEM and SE values and the number of experiments are indicated in each figure. Spine quantification statistical analyses were performed using Mann–Whitney unpaired t-test for dendritic spine type analyses and two-way ANOVA, followed by Tukey’s post hoc multiple comparison test using GraphPad Prism 6 software for genotype/treatment comparisons. Pearson’s and Mander’s colocalization were performed using the ImageJ plugin COLOC2. The significance level was P < 0.05 for all treatments.

First, in order to examine the role of c-Abl in the synapse, we analyzed its protein levels over time. During culture maturation, the post-synaptic scaffold protein SAP102 and the post-synaptic density protein-95 (PSD95), continuously increase their protein levels. Meanwhile, c-Abl displays the highest protein levels during neurite extension at 4 DIV and at full culture maturation at 20 DIV (Figure 1A). At 7 DIV c-Abl is found mainly located at the soma, while at 15 DIV c-Abl is broadly distributed, not only in the soma, but it also has a punctate shape in all neuronal processes (Figure 1B). As shown by Pearson’s correlation (Figure 1C), c-Abl localizes in the post-synaptic compartment, where it co-localizes with PSD95.

To investigate whether c-Abl could be regulating dendritic spine density, we studied spine morphology in cultured c-Abl knock-out (c-Abl-KO) hippocampal neurons transfected with GFP expression plasmids at 18 DIV to label whole single neurons (Figure 2A). One day later, we counted their dendritic spine population in secondary branches. Interestingly, we found that c-Abl deficiency increases dendritic spine density. c-Abl-KO neurons showed an enriched spine density with 4.22 ± 0.25 spines/10 μm dendrite (Figure 2B), in comparison with WT neurons that showed 3.36 ± 0.20 spines/10 μm dendrite. Therefore, c-Abl-KO neurons display a 25.6% dendritic spine increase in their spine density (Figure 2B).

Exposure to AβOs significantly decreases dendritic spine density in neurons. Using a synthetic Aβ_1–42 human peptide (see “Materials and Methods” section), we prepared an overnight solution of AβOs characterized by the presence of dimers and trimers (Supplementary Figure S2). These soluble species of the peptide have been previously described as the most toxic species of AβOs that bind to the synapse of hippocampal neurons (Tu et al., 2014; Mi et al., 2017). Previously, we described that AβOs induce c-Abl activation participating in the loss of synapses (Vargas et al., 2014). Moreover, the binding of FITC-labeled AβOs to dendrites activates c-Abl (p-c-Abl; Figure 1E, Supplementary Figure S3). Thus, we treated hippocampal neurons with AβOs and observed a quick increase (at 1 h) in c-Abl phosphorylation that remains active 5 h after treatment with AβOs. This increase in active c-Abl was also associated with a later increase in both Abl1 mRNA and c-Abl protein levels from 3 h post-AβOs incubation (Figures 1D–F). Fibrillary forms of the Aβ-peptide (Aβf) also increased c-Abl protein levels by 3 h incubation and were used as a control.

Then, we asked whether c-Abl ablation modulates AβOs-induced synapsis loss. GFP-expressing WT and c-Abl-KO hippocampal neurons were exposed to AβOs and after 5 h treatment, the number of dendritic spines was evaluated. Primary dendrites were the least affected by the AβOs treatment, while tertiary and far-away branches were the most affected, and even disrupted by AβOs. Therefore, we quantified dendritic spines in secondary branches. WT neurons showed reduced spine density after AβOs treatment (3.36 ± 0.20 vs. 2.45 ± 0.15 spines/10 μm dendrite), representing 27% spine loss. As well as WT neurons, c-Abl-KO neurons displayed a significant reduction over spine density (4.22 ± 0.22 vs. 3.19 ± 0.15 spines/10 μm dendrite), representing a 24.3% spine loss (Figure 2B, Supplementary Figure S4C). Thus, AβOs treatment induced a decrease in the number of dendritic spines in both, WT and c-Abl-KO neurons. Since we found only a 3% difference in dendritic spine density between c-Abl-KO and WT neurons treated with AβOs, suggesting that AβOs-driven spine loss seems to be independent of c-Abl.

Then, we evaluated spine morphology by analyzing the dendritic spines shape profile following five major categories: mushroom, stubby, branched, thin and filopodia spines. Mature spines are mushroom, big spines in which the head is wider than the neck. Intermediate states are stubby, smaller and neckless spines. Branched spines are two-headed spines. Finally, immature spines are thin protrusions shorter than 2 μm length; and filopodium, in which protrusions are longer than 2 μm length (Tyler and Pozzo-Miller, 2003).

To distinguish between different types of spine morphology, we measure head, length, and neck of dendritic spines in GFP-expressing neurons (Figures 3A–D), and quantified the number of each spine type in a 10 μm dendrite section. We found a tendency for c-Abl-KO neurons to display more mushroom, branched and filopodia spines than WT neurons per 10 μm dendrite, however non-significant (Figure 3B). This trend towards increasing some spine types correlated with the augmented spine density observed before (Figure 2B). Since both, the neuron genotype and the binding of AβOs alter the number of dendritic spines, we normalized to the total population of dendritic spines within each dendrite analyzed. Then, we found that c-Abl deficiency alters the distribution of the spine population.

When we analyzed the head diameter of dendritic spines, we found that c-Abl-KO spines had wider heads compared to WT (0.46 ± 0.01 μm vs. 0.42 ± 0.01 μm in average head diameter, respectively; Supplementary Figure S4A), which correlates with an increased number of mushroom spines. Interestingly, we also found that c-Abl-KO spines were slender than WT spines, with a total length increase for longer filopodia and thin spines (Figure 3D) and an increment of the overall spine length (WT: 0.79 ± 0.04 and KO: 0.98 ± 0.05 μm spine length; Supplementary Figure S4B), probably due to the increase in thin population.

As expected, when AβOs were added to the medium, the most mature, mushroom spines were significantly affected by AβOs treatment. As the spine profile shows, mushroom decreased 42% by AβOs in WT, but also decreased 37% in c-Abl-KO neurons. Stubby spines were also significantly affected by AβOs treatment with a 7.2% less in WT neurons, while they tend to 6.5% augment in KO neurons (Figure 3C). Interestingly, the histogram of dendritic spines changed mushroom, stubby, thin and filopodia in WT vs. WT-treated neurons, while the overall pattern was similar for KO vs. KO-treated neurons. Interestingly, c-Abl deficiency increased the relative abundance of immature spines and maintained the mushroom and stubby population (Figure 3C). We observed a slight increase in the relative abundance of thin and filopodia spines in c-Abl-KO neurons when they were treated with AβOs (thin: 41% vs. 43.4%, and filopodium: 13.3% vs. 11.6% WT+AβOs vs. KO+AβOs, respectively; Figure 3C). Interestingly, the spine profile of WT-treated neurons displays a significant increase for filopodia spines (54.9% increase) as they are usually found within 10 μm dendrite (WT: 0.04 ± 0.01 vs. WT+ AβOs: 0.07 ± 0.02 spines/10 μm dendrite), in correlation with the lengthening of dendritic spines (WT: 0.79 ± 0.04 μm vs. WT+AβOs: 1.03 ± 0.07 μm length; Supplementary Figure S4B). Finally, branched spines were non-significantly affected by AβOs in both WT and c-Abl-KO neurons. These results suggest that AβOs-induced an overall decrease in the dendritic spine population of WT and c-Abl-KO neurons, but the effect becomes attenuated in the immature spine population of c-Abl-KO neurons. Whereas these immature spines, thin and filopodia together comprise 55% of the total spine population in c-Abl-KO treated neurons, they represent 54% in WT control-treated neurons. In the same line, the abundance of mature mushroom spines represents almost 7% of the overall population for WT, while 8% for c-Abl-KO dendritic spines after AβOs incubation (Figure 3C). The same as controls, spine head diameter of WT and c-Abl-KO neurons exposed to AβOs was very similar (WT+AβOs: 0.40 ± 0.02 μm and KO+AβOs: 0.37 ± 0.01 μm average spine head diameter), and no changes were evident between them. However, spine head diameter was significantly reduced in c-Abl-KO neurons exposed to AβOs compared with basal levels (Supplementary Figure S4A), in agreement with the significant reduction of mushroom spines.

Therefore, in the absence of c-Abl, treatment with AβOs enriches the immature dendritic spine population and preserves mushroom and stubby population maintaining the overall spine density, which suggests that this may be a possible mechanism for synaptic resiliency against AβOs.

In AD mice models and AD patients, AβOs induce synapse loss, decreasing levels of pre- and post-synaptic proteins and reducing the number of synaptic contacts (Sze et al., 1997; Masliah et al., 2001; Reddy et al., 2005; Calabrese et al., 2007). Therefore, we next evaluated the effect of c-Abl deficiency on AβOs synapse alteration. We examined synaptic contacts in c-Abl null hippocampal neurons after AβOs treatment by following PSD95 and Piccolo clustering (post-synaptic and pre-synaptic proteins, respectively).

c-Abl-KO neurons display increased numbers of Piccolo clusters in basal conditions in comparison to WTs. Interestingly, AβOs significantly reduced Piccolo clustering in WT neurons (3.82 ± 0.17 clusters/10 μm dendrite). While the number of Piccolo clusters in c-Abl-KO-treated neurons was very similar to its controls and significantly different from WT-treated neurons. Therefore, indicating maintenance of the pre-synaptic terminal (6.41 ± 0.19 clusters/10 μm dendrite; Figures 4A,B). Opposite results were obtained for PSD95, as c-Abl deficiency did not perturb the basal number of PSD95 synaptic clustering (WT: 5.72 ± 0.18 and KO: 5.27 ± 0.24 clusters/10 μm dendrite; Figures 4A,C). However, we observed a strong reduction in PSD95 clusters after WT neurons were treated with AβOs (WT+AβOs: 3.91 ± 0.14 clusters/10 μm dendrite) whereas, c-Abl-KO neurons treated with AβOs display less reduction in PSD95 clusters (KO+AβOs: 4.38 ± 0.15 clusters/10 μm dendrite; Figures 4A,C). However, we did not find significant differences between WT and KO-treated neurons.

On the other hand, WT neurons showed a significant reduction in PSD95/Piccolo clusters when incubated with AβOs whereas the c-Abl-KO neurons did not show synaptic loss (2.53 ± 0.2 and 4.15 ± 0.27 clusters/10 μm dendrite, respectively; Figure 4D). Colocalization analysis by Mander’s and Pearson’s correlation also showed decreased correlation for WT-treated neurons while c-Abl-KO neurons were non-significantly affected, which means protection of synaptic clustering in correlation with synaptic contact quantification (Supplementary Figures S5A,B). Our results strongly suggest that c-Abl deficiency protects the synapse.

Since AβOs-induced synaptotoxicity is linked to neuronal cell death (Yang et al., 2009), and c-Abl inhibition by Imatinib prevents AβOs-induced apoptosis (Cancino et al., 2008), we asked whether c-Abl could be responsible for apoptosis. Therefore, we quantified apoptotic nuclei using Hoechst staining (Figures 5A,C) and caspase-3 immuno-labeling (Figures 5B,D), in WT and c-Abl-KO neurons after a 5 h treatment with AβOs. As expected, in WT neurons, AβOs induced a significant increase in apoptotic nuclei (18.2 ± 4.8% control vs. 49.7 ± 4.1% AβOs-treated neurons) and in the number of caspase-3 positive cells (30.7 ± 7.2% control vs. 54.7 ± 6.5% AβOs-treated neurons). While c-Abl deficiency significantly decreased AβOs-induced apoptosis followed by apoptotic nuclei (16 ± 3.4% control vs. 31 ± 4.9% AβOs-treated neurons) and active caspase-3 positive cells (12 ± 4.1% control vs. 22.9 ± 7.1% AβOs-treated neurons; Figures 5C,D). Therefore, c-Abl deficiency protects against AβOs-induced cell death.

We also analyzed the effect of AβOs over synaptic protein removal by following PSD95 and Piccolo clusters in apoptotic and non-apoptotic WT and c-Abl-KO neurons. As before, we counted the number of clusters per 10 μm dendrite and found a strong and robust reduction of Piccolo clusters in WT neurons under AβOs, both in apoptotic and non-apoptotic neurons (Figure 5E; non-apoptotic WT: 6.52±0.18 vs. WT+AβOs: 4.36 ± 0.21 clusters/10 μm dendrite). Interestingly, the number of Piccolo clusters in c-Abl-KO neurons were maintained after AβOs treatment. Regardless of whether we examined apoptotic or non-apoptotic c-Abl-KO neurons, there was no significant difference with their controls (non-apoptotic KO: 8.19 ± 0.19 vs. KO+AβOs: 7.1 ± 0.24 clusters/10 μm dendrite). Although apoptotic neurons showed fewer Piccolo clusters than non-apoptotic neurons, after AβOs treatment all of them display a loss of Piccolo. In spite of AβOs treatment, c-Abl-KO neurons showed significantly more clusters than WT neurons in both apoptotic and non-apoptotic cells.

However, the effects of c-Abl deficiency and apoptosis on PSD95 clusters reduction induced by AβOs are less clear (Figure 5F). First, there are fewer PSD95 clusters under basal conditions in apoptotic cells compared to non-apoptotic neurons. Additionally, non-apoptotic c-Abl-KO neurons showed reduced PSD95 cluster numbers in comparison with WT under basal conditions (WT: 6.76 ± 0.17 vs. KO: 5.7 ± 0.33 clusters/10 μm dendrite). However, when treated with AβOs, both WT and c-Abl-KO non-apoptotic neurons showed fewer PSD95 clusters, with no significant differences between them (WT+AβOs: 4.29 ± 0.17 vs. KO+AβOs: 4.66 ± 0.22 clusters/10 μm dendrite). Interestingly, c-Abl-KO apoptotic neurons did not show a further reduction in PSD95 clusters in comparison with WT neurons (apoptotic KO+AβOs: 4.11 ± 0.19 clusters/10 μm dendrite).

Our results suggest that AβOs-induced Piccolo clustering reduction is significantly prevented in c-Abl-KO apoptotic and non-apoptotic neurons in comparison with WTs. On the other hand, PSD95 clusters are strongly affected by the apoptotic state of neurons derived from AβOs.

Finally, our results suggest that c-Abl mediates the signaling pathway triggered by AβOs that induces synaptic elimination and neuronal death, and could, therefore, be a relevant player in the early onset of AD.