|Home | About | Journals | Submit | Contact Us | Français|
Numerous studies have shown that amyloid-β (Aβ) modulate intracellular metabolic cascades and an intracellular Ca2+ homeostasis and a cell surface NMDA receptor expression alteration in Alzheimer’s disease (AD). However most of these findings have been obtained by using non-physiological Aβ concentrations. The present study deals with the effect of low Aβ concentrations on cellular homeostasis. We used nerve growth factor-differentiated PC12 cells and murine cortical neurons sequentially treated with low chronic monomeric or small oligomeric Aβ concentrations and high acute oligomeric Aβ concentrations to bring out a priming effect of chronic treatment on subsequently high Aβ concentrations-elicited cellular response. Both cell types indeed displayed an enhanced capacity to bind oligomeric Aβ after monomeric or small oligomeric Aβ application. Furthermore, the results show that monomeric Aβ1–42 application to the cells induces an increase of the Ca2+-response and of the membrane expression of the extrasynaptic subunit of the NMDA receptor GluN2B in PC12 cells, while the opposite effects were observed in cultured neurons. This suggests a sequential interaction of Aβ with the cellular plasma membrane involving monomers or small Aβ oligomers which would facilitate the binding of the deleterious high molecular Aβ oligomers. This mechanism would explain the slow progression of AD in the human nervous system and the deep gradient of neuronal death observed around the amyloid plaques in the nervous tissue.
The neuropathological hallmarks of Alzheimer’s disease (AD) are characterized by the presence of large extracellular deposits of fibrillar amyloid-β (Aβ) [1–3] and intraneuronal neurofibrillary tangles [4–6]. However, different observations suggest that fibrillar Aβ may not be the primary toxic assembly state responsible for AD pathogenesis. Several groups have reported that individuals with normal cognitive function have Aβ plaque loads that either meet or exceed the criteria for AD diagnosis . Studies have also revealed a positive correlation between soluble Aβ levels and the severity of dementia [8, 9], further suggesting a key role for soluble oligomers species in AD. In vitro studies have reported that not only soluble oligomers and protofibrils are toxic to neuronal cells , but that they are more toxic than fibrils [11, 12].
Considering the progression of AD, the mode of exposure of nerve cells to amyloid-β (Aβ) peptides seems to be very important. The question remains whether low soluble Aβ concentrations induce a change in nerve cells. The initial events probably involve the interaction of soluble oligomers with neuronal membranes . But AD evolves over decades in human, while most in vitro experiments are performed within minutes or hours. Similarly, in most biochemical or electrophysiological experiments designed to decipher the mode of action of Aβ, the used concentrations are not physiological.
In an attempt to address this question we used PC12 cells, a cell line which displays a neuronal phenotype when cultivated in the presence of nerve growth factor and primary cultures of mouse cortical neurons. We show that these cells have an increased capacity to bind Aβ oligomers after pre-incubation with nanomolar concentrations of soluble Aβ1–42 and that the toxic effect of oligomers at micromolar concentrations is enhanced. Moreover, we also show that cell exposure to high concentrations of Aβ1–42 oligomers rapidly induces changes in the membrane expression of the GluN2B subunit of the NMDA receptor. These observations may partly explain the slow progression of the disease in the human brain.
PC12 cells (rat pheochromocytoma) were cultured in RPMI 1640 medium supplemented with 5% heat-inactivated horse serum, 10% fetal bovine serum and antibiotics (penicillin 100 U/ml, streptomycin 100 U/ml) at 37°C in 5% CO2. PC12 cells were reseeded once a week. Before use, cells were plated in 35mm glass-bottom culture dishes coated with rat tail type I collagen (0.1mg/ml) and Poly-L-ornithine (0.5mg/ml) and NGF (50 μg/l) was added in the culture medium to induce their neuronal differentiation. Cells were used after 5 days in vitro.
Except for rat tail type I collagen and poly-L-ornithine which were supplied by Sigma-Aldrich (Sigma, St. Louis, MO) all products used to cultivate PC12 cells were supplied by Life Technologies® (St Aubin, France).
Primary neuronal cultures were prepared from C57Bl6-OF1 mouse embryos (E16-E18). Pregnant females were from Charles River (L’Abresle, France) or from Chronobiotron-CNRS (Strasbourg, France). Cortices were dissociated in a trypsin-EDTA enzymatic solution (0.05% , Life Technologies) for 2min at 37°C. The dissociation was then mechanically completed in Neurobasal medium supplemented with B27 (2% ), glutamax (1% ), and antibiotics (streptomycine 100 U/mL, penicillin 100 U/mL). After centrifugation (500g, 5min), the pellet was resuspended in Neurobasal medium and seeded on glass-bottom culture dishes (Corning, Avon, France). The cultures were kept at 37°C under 5 % CO2 atmosphere during 9 to 14 days with a partial medium renewal every 2 days.
Aβ1–42 (Bachem, Bubendorf, Switzerland) was dissolved in hexafluoroisopropanol (HFIP; Sigma-Aldrich), distributed in aliquots, dried (HFIP film) and stored at −80°C as previously described [14, 15]. The day before the experiment the peptide film was dissolved (1mM) in sterile dimethylsulfoxide (DMSO). The solution was then diluted with phosphate buffered saline (PBS, pH 7.4) at a final nominal concentration of 100 μM and aged overnight at 4°C. The preparation was centrifuged for 15min at 14000× g to remove insoluble aggregates (Protofibrils and fibrils) and the supernatant containing soluble Aβ1–42 oligomers was transferred to new tubes and stored at 4°C. The “monomer” solutions were used immediately after dissolution in DMSO.
We measured Aβ1–42 concentrations in the stock solution with a classical method (Bradford). However in the experimental conditions of cell pre-treatment, the Aβ1–42 solution was applied within the culture medium, which introduces measurement bias. We thus used an ELISA kit specifically developed for Aβ1–42 (BetaMark-x42, Eurogentec, San Diego, CA USA). In this case, the actual final concentration measured was between 2 and 4 nM for a nominal concentration of 10 nM. For this reason, the nominal concentration is systematically quoted, knowing that the final concentration is probably less.
The Aβ1–42 and the Aβ42–1 synthetic peptides (Bachem) were first dissolved in DMSO (Sigma) and then diluted in the working media. According to experimental conditions, two types of peptidic treatments have been applied on the cells. For “chronic” treatment, cells were maintained at 37°C in 5% CO2 and the treatment consisted of an overnight application of peptide Aβ1–42 or Aβ42–1 (10 nM nominal). For “acute” application, Aβ1–42 or the reverse peptide Aβ42–1 were diluted in phosphate buffered saline (PBS) for immunochemistry or Hepes buffer (D–Glucose 5.5mM, MgCl2 1mM, CaCl2 2mM, NaCl 130mM, KCl 5.4mM, Hepes 10mM, pH 7.4) buffer for calcium imaging and then applied to the cells at a final concentration of 1 μM. To quantify the binding of Aβ1–42 oligomers to the cells (see following section), this solution was applied at 200 nM final concentration.
In a first double immunolabeling experiments (Figs. 2 and and3),3), cells were acutely treated ([Aβ1–42 or the reverse Aβ42–1]=1 μM diluted in PBS, for 10min). Cells were subsequently rinsed to eliminate unbound Aβ and fixed with 4% paraformaldehyde in PBS for 10min. Cells were then incubated in a blocking solution (3% BSA in PBS) for 1h at room temperature. Fixed and non-permeabilized PC12 cells were first incubated with Alexa Fluor 488-conjugated cholera toxin subunit B (1:500, Life Technologies) for 30min at room temperature, then washed and incubated overnight with a primary antibody directed against every forms of Aβ (6E10, mouse monoclonal, 1:1000, Covance Inc., Rueil-Malmaison, France). Cells were washed 3×10min with the blocking solution and incubated with the appropriate secondary Dylight 549-conjugated Goat anti-mouse IgG (1:1000, Jackson ImmnunoResearch Laboratories Inc., West Grove, PA, USA). Glass slides were mounted in Mowiol and observed with a Nikon DIAPHOT–TMD inverted fluorescence microscope and images were acquired with a digital camera (DXM1200, Nikon) and the Metamorph software (Molecular Devices).
In experiments aimed at quantifying the binding of Aβ1–42 oligomers (Figs. 6 and and7)7) the cells have undergone a chronic treatment ([Aβ1–42 or the reverse Aβ42–1]=10 nM diluted in culture media overnight) and/or an acute treatment with 200 nM Aβ1–42 oligomer solution or with 200 nM Aβ42–1 in order to avoid micrograph light saturation. A rabbit polyclonal antibody A11 directed against Aβ oligomers (Chemicon International, Temecula, CA; diluted 1:1500 in blocking solution) was applied to the cells overnight at 4°C. Cells were washed three times for 10min with the blocking solution and incubated with the appropriate secondary alexa 546-conjugated antibody (1:2000, Life Technologies) for 2h at room temperature, washed 3×10min in PBS. In this case fixed primary cultured neurons were also incubated with an anti MAP2 primary antibody (chicken polyclonal, 1:5000, Thermo Scientific) and with an anti-GFAP antibody (rabbit polyclonal, 1:5000, Dako, Denmark) and revealed with secondary antibodies (Dylight.649 goat anti chicken IgY, 1:1000 and Dylight 488 anti rabbit IgG, 1:1000; Jackson ImmnunoResearch Laboratories Inc.).
For GluN2B quantification experiments, cells were incubated with an antibody directed against the GluN2B subunit (Rabbit polyclonal, 1:500, Abcam, Bristol, UK). The immunoreactivity was then revealed by incubating the preparation with Dylight 488-conjugated Goat anti-rabbit IgG antibodies (1:1000, Jackson ImmnunoResearch Laboratories Inc.) for 2h at room temperature, washed 3×10min in PBS and mounted in Mowiol.
In some experiments, fixed neurons were also incubated with an anti-Thy-1.2 primary antibody (rat monoclonal, 1:1000; Pharmingen, San Diego, USA) and revealed with Dylight 549-conjugated Goat anti-rat IgG (1:1000, Jackson ImmnunoResearchLaboratories Inc.).
For quantitative imaging, the micrograph recordwas performed on a Carl Zeiss LSM 510 confocal microscope (Munich, Germany). Images were subsequently processed with Image J software (NIH) for quantification and the Pearson coefficient was calculated to assess for the co-localization of the markers.
Fura-2 AM dissolved in DMSO (100 μM) and Pluronic F-127 (16 μM dissolved in DMSO, Life Technologies) were mixed (1/1 v) and added to cells platted on glass–bottom culture dishes in a Hepes buffer for 20min at 37°C (1 μM Fura-2 final concentration). Cells were washed 3 times and again incubated in Hepes buffer for 20min at 37°C to ensure the complete probe hydrolysis. Cells were maintained in Hepes buffer during image acquisition. Data were acquired with an Axiovert 135 microscope (Carl Zeiss), an IMAGE-CDD digital camera (Hamamatsu photonics) and the MetaFluor software (Molecular Devices). Fluorescence was excited by a xenon lamp equipped with two alternating filters allowing a fluorescence excitation at 350nm for Ca2+-bound Fura-2 and at 380nm for unbound Fura-2. Emitted fluorescence intensities (510nm) were measured from regions of interest centered on individual cells. A pair of images was captured every five seconds. Normalized ratio were calculated by dividing all ratio between fluorescence intensities emitted when cells were excited at 350nm and when cells were excited at 380nm by the mean of this ratio during baseline time (F 350/F 380)/mean (F 350/F 380). Variations of these normalized ratio reflect [Ca2+]i changes.
Cells were pretreated with Aβ1–42 or reverse Aβ42–1 (10 nM) overnight when required. Aβ1–42 (1 μM final concentration), reverse Aβ42–1 (1 μM) were dissolved in Hepes buffer and added to the extracellular medium about 3min after starting the recording. When required, N-methyl-D-aspartate receptor (NMDAR) antagonists were added to the extracellular medium just before starting theacquisition.
To characterize the proportions of Aβ monomers and oligomers in the solutions applied on cells, chronic and acute solutions were prepared by diluting Aβ1–42 in RPMI medium (50 nM) or in PBS (1 μM). Samples were diluted in a reducing sample buffer (TrisHCl 63mM, glycerol 30% , SDS 2% , Bromophenol blue 0.00025% , pH 6.8) and electrophoretically separated by SDS-PAGE using 4–20% Tris-Glycine polyacrylamide gels and Novex Tris Glycin running buffer (Life Technologies) at 125mV for 1h 40min. The gels were transferred to nitrocellulose membranes (1h 15min at 125mA) in transfer buffer (Tris 25mM, Glycine 192mM, SDS 0.02% , ethanol 20% ). The membranes were blocked in 0.5% nonfat dry milk in PBS-Tween20 0.1% and incubated for 15min with an monoclonal anti-Aβ 6E10 primary antibody (Covance Inc.) and an HRP-conjugated Goat anti-mouse IgG secondary antibody (Jackson) using the SNAPi.d system (Millipore, Molsheim, France). Final detection was performed with chemioluminescence (Supersignal west femto sensitivity substrate, Thermo Scientific) using CL-Xposure films (Thermo Scientific).
Values are expressed as mean ± standard deviation. The statistical significance was assessed by one way ANOVA and subsequent Bonferroni test analysis.
Each peptide preparation applied to the cells was evaluated for the presence of soluble oligomers by SDS-PAGE and immunoblotting. Gels were processed for western blotting using 6E10 monoclonal antibody, which recognizes every form of Aβ. Western blots showed that preparations at low Aβ concentrations (nanomolar range) contained a mixture of monomers and higher molecular weight oligomers. In preparations containing high Aβ concentrations (micromolar range) most of the peptide was oligomeric (Fig. 1).
To confirm that Aβ1–42 interacts with the plasma membrane of either differentiated PC12 cells or cultured neurons in our experimental conditions, multi-labelling immunochemistry experiments were performed on cells previously incubated for 10min in the presence of 200nM Aβ. Aβ binding on PC12 cells was revealed by using the 6E10 monoclonal antibody and a secondary Dylight-549-conjugated antibody associated with Alexa Fluor 488-conjugated cholera toxin which binds to lipid microdomains . As shown in Fig. 2 Aβ1–42 clearly labels the membrane with a partial but consistent colocalization with cholera toxin. The Pearson coefficient value was 0.226 ± 0.032 significantly higher than the value obtained in control conditions (0.034 ± 0.171 p< 0.05) with a reversed Aβ42–1 peptide.
In cultured neurons, we used an anti-MAP2 primary antibody which is a general marker for differentiated neurons  and an anti-Thy-1.2 primary antibody which labels the neuronal plasma membrane . In MAP2-positive cells, Fig. 3 shows that Aβ1–42 partially co-localizes with Thy-1.2 (Pearson coefficient 0.192 ± 0.64). As a control we also used the inactive reverse peptide Aβ42–1(Pearson coefficient 0.022 ± 0.019). In this case no label was observed on the cultured neurons similarly to cells preincubated with the vehicle solution (Fig. 3). This observation means that Aβ1–42 directly interacts at least partially with the plasma membrane of the cells in each studied model.
Calcium imaging experiments showed that an acute application of Aβ1–42 oligomers induces a rapid rise of intracellular calcium concentration in PC12 cells (Fig. 4A). This calcium response was dose-dependent. When a low concentration of Aβ1–42 (0.1 μM) was applied, the calcium response developed slowly and the amplitude was reduced but detectable; for Aβ1–42 concentrations greater than or equal to 0.5 μM, the calcium response was fast and large with a time-to-peak within 2min. No response was observed following application of the reverse peptide Aβ42–1 suggesting a specific effect of Aβ1–42 (not shown). The response was rapidly desensitizing according to a two exponential kinetics (Fig. 4).
The effects of homogeneous monomeric and oligomeric preparations of soluble Aβ1–42 were examined in fura-2-AM-loaded PC12 cells. Figure 4b illustrates the increase of intracellular free Ca2+. Application of monomers and low molecular weight oligomers at a final concentration of 1 μM evoked no detectable change in Ca2+ concentration whereas application of the same concentration of high molecular weight oligomers evoked a rapid increase of [Ca2+]i. This oligomeric Aβ-induced calcium response correlates with adhesion of Aβ1–42 oligomers on the membrane surface of PC12 cells as revealed by immunocytochemistry with an antibody specific for the oligomeric form of Aβ42 (A11) (see Fig. 6B).
A similar Ca2+-response was observed in cultured neurons after Aβ1–42 oligomers application (Fig. 4C): a sharp peak and a rapid subsequent decrease of the intracellular [Ca2+]. As observed for PC12 cells monomeric Aβ1–42 applied to neurons did not evoke any Ca2+ signal (Fig. 4C) in cultured neurons.
Since it has been shown that Aβ1–42 toxicity involves NMDA receptors (for a review, see ), we first checked whether our models were responsive to NMDA. Either in PC12 cells or in neurons NMDA (100 μM) application induced a progressive increase of the intracellular [Ca2+] without an initial peak in Ca2+ concentration and without desensitization along the duration of the recording (around 20min); indeed this response was completely inhibited by a previous incubation of the cells in the presence of D-APV a non-selective antagonist of NMDA receptors (notillustrated).
Most of the PC12 cells were not responding to Aβ1–42 application in the presence of 50 μM APV (Table 1). In the remaining responding cells (about 5% of the total cell number), the [Ca2+]i increase induced by Aβ1–42 was almost completely inhibited (94 ± 1.5% ) (Table 1).
Aβ1–42 -evoked Ca2+ response following D-APV application was also inhibited in neurons: we observed a partial but consistent inhibition and a decreased number of responding cells (Table 1).
To mimic physiopathological conditions, we incubated PC12 cells overnight with nanomolar concentrations of Aβ1–42 or Aβ42–1.
Acute subsequent application of Aβ1–42 peptide (1 μM; 10min) on PC12 cells triggered a rapid rise in intracellular calcium concentration ([Ca2+]i) (Fig. 5A) and this response was potentiated when the cells were preincubated with the Aβ1–42 peptide (10 nM) overnight or at least 16h (chronic treatment, Fig. 5A). The peak amplitude of the [Ca2+]i response was significantly increased (704 ± 822 nM, n=51, in preincubated cells versus 295 ± 129 nM, n=45, in non preincubated cells, p< 0.01) although this value varied from cell to cell. This [Ca2+]i cellular response was specific for Aβ1–42 application since neither acute application of the reverse peptide Aβ42–1 at μmolar concentration nor chronic treatment with the reverse peptide Aβ42–1 had a significant effect on the cells when applied (Fig. 5A). Moreover, chronic treatment of PC12 cells with Aβ1–42 peptide followed by acute application of the reverse peptide Aβ42–1 did not induce a significant response as compared to control (Fig. 5A). This lack of effect of the reverse peptide Aβ42–1 was also confirmed by the weak percentage of responding cells (Fig. 5B). It should be pointed out that the overnight preincubation of the PC12 cells with Aβ1–42 did not significantly alter the number of responsive cells (Fig. 5B). The preincubation alone did not modified the resting [Ca2+]i (53.6 ± 10.8 nM in control cells versus 53.4 ± 8.9 nM in preincubated cells).
We performed this experiment with increasing concentrations of acutely applied Aβ1–42 after a chronic incubation of the cells with Aβ1–42. This resulted in a dose-dependent increase of the calcium signal, which was not significantly different from the dose dependence observed in cells not chronically incubated with Aβ1–42 (not illustrated).
Similar experiments were performed on cultured neurons. In this case, acute application of Aβ1–42 (1 μM) following an overnight incubation of the neurons in the presence of 10 nM Aβ1–42 resulted in a decrease of the cellular calcium response as compared to the effect of a single acute application of the peptide (Fig. 5C). The specificity of this effect was checked using the reverse peptide Aβ42–1 and no effect was detected with the reverse peptide either on Ca2+ concentration (Fig. 5C) or on the number of responding cells (Fig. 5D). Furthermore as previously observed for PC12 cells no significant change was observed in the basal [Ca2+]i in neurons incubated overnight with 10 nM Aβ1–42 and then acutely treated with vehicle (Fig. 5C).
Two questions arose: what was the tertiary form of the peptide bound to the cells and responsible for the priming effect induced on the intracellular calcium response and was it similar in both cellular models? To answer these points, we performed immunocytochemical experiments: PC12 cells or neurons were incubated with 200 μM oligomeric Aβ1–42 with or without preincubation with 10 nM Aβ1–42, fixed and then revealed with an antibody specifically directed against Aβ oligomers.
In PC12 cells, no significant increase of the fluorescent signal was observed after overnight preincubation with 10 nM Aβ1–42 (Fig. 6C) or after acute application of 200 μM oligomeric Aβ1–42 (Fig. 6B). When overnight preincubation was associated to a subsequent acute Aβ1–42 application, a clear signal increase was observed (Fig. 6D, G) suggesting that most bound Aβ1–42 was under the oligomeric form. The quantitative analysis of fluorescence intensities correlates with the priming effect observed on the Aβ-induced Ca2+ response: the fluorescence intensity in cells chronically incubated with Aβ1–42 was not significantly different from the fluorescence measured in control cells, while the effect of previous chronic incubation on a subsequent acute Aβ1–42 application was highly significant (p< 0.001, Fig. 6G).
In neurons, the pattern of Aβ1–42 binding according to the incubation conditions was similar to that observed in PC12 cells: Fig. 7 shows a significant increase of Aβ1–42 immunodetection following preincubation with 10 nM Aβ1–42 and subsequent application of 200 nM oligomeric Aβ1–42.
Hence the observed opposite effect of Aβ1–42 preincubation combined with subsequent application of high concentration on the induced Ca2+-response in PC12 cells and in cultured neurons was not due to a difference in binding properties of Aβ1–42 oligomers following the preincubation step.
Previous studies have reported an involvement of the GluN2B subunit of NMDA receptor in the cellular response induced by Aβ1–42 application [17–21]. Since we observed that the Ca2+ response induced by Aβ1–42 in PC12 cells was almost completely mediated by NMDA receptors, we checked whether the amyloid peptide was co-localized with NMDA receptors on the cellular membrane. For this purpose we performed double immunolabelling experiments on fixed but non permeabilized PC12 cells. We used an antibody directed against the GluN2B subunit of the NMDA receptor and another antibody (6E10) directed against Aβ1–42. As illustrated in Fig. 8, we observed a partial co-localization of the membrane-bound Aβ1–42 with the GluN2B subunit. This co-localization occurred in the presence of 10 nM monomeric Aβ1–42 as well as 1 μM oligomeric Aβ1–42 (Fig. 8).
We also observed that the fluorescence intensity reflecting GluN2B immunoreactivity was enhanced in cells treated with Aβ (Fig. 8, lower graph). Indeed, a quantification of the fluorescence intensity revealed no significant increase of GluN2B expression after chronic exposure of the PC12 cells to 10 nM monomeric Aβ1–42. However GluN2B immunoreactivity was significantly increased when the cells were incubated with 1 μM oligomeric Aβ1–42. In this case, a trend of priming effect after chronic Aβ1–42 application was observed although not significant. It should be pointed out that the cells appeared heterogeneous in the label immuno-intensity for GluN2B (Fig. 8, arrows). This probably reflects a partial differentiation of some cells, and it may also explain the lack of significance of priming effect of the cells pretreatment.
A similar experiment was performed on cultured cortical neurons. In this case, due to the heterogeneity of the cellular types present in the culture it was necessary to characterize neurons with a membranous neuronal marker. We used double immunolabelling of the non permeabilized cells with antibodies directed against the GluN2B subunit and Thy1.2 respectively. Several cellular types were present in the cell culture; among these cells, some displaying a neuronal morphology were expressing GluN2B but not Thy1.2 (Fig. 9 arrows). Immunolabelling on these cells was not quantified. On Thy1.2 positive cells we observed a significant decrease of GluN2B expression on the membrane after chronic application of 10 nM monomeric Aβ1–42 (Fig. 9). Indeed this decrease was amplified when the cells were also subjected to acute 1 μM oligomeric Aβ1–42 for 10min.
Growing evidence points to a disruption of intra-cellular Ca2+ homeostasis in AD and other neurodegenerative diseases (for a review, see ), and intracellular Ca2+ levels are known to trigger metabolic cascades that ultimately lead to synaptic dysfunction and cell death [22, 23]. It has been established that this deleterious effect is mainly induced by the soluble forms of the amyloid peptides [8, 9]. However considering the pathological and physiological concentrations of soluble Aβ encountered in the nervous tissue raises the question of such low concentrations effects on the cellular homeostasis.
In the present report, we show that micromolar concentrations of Aβ1–42 induce a transient increase of intracellular [Ca2+] in PC12 cells expressing a neuronal phenotype as well as in cultured cortical neurons. Although the concentration used is high compared to physiological concentrations, this is in agreement with numerous previous studies performed either on primary cultures of nerve cells or on cell lines [24–26]).
Additionally the pharmacological characterization of the observed Ca2+ cellular response corroborates previous observations: we show that this response is preferentially induced by Aβ1–42 oligomers in agreement with other published observations [27–29]. Prefibrillar amyloid aggregates have been shown to elevate cytosolic Ca2+ in neurons or neuronal cell lines ([30, 31]), and this phenomenon has been proposed to result from Ca2+ influx across the plasma membrane either through cation-selective channels formed by Aβ1–42 itself or through a general disruption of lipid integrity [12, 31, 32]. Another still controversial hypothesis is based on the effect of Aβ1–42 itself on voltage operated calcium channels although the peptide has been described to inhibit L-type channels at micromolar concentrations  and P/Q type channels at nanomolar concentrations . We cannot exclude these hypotheses but they appear unlike since we observed that the Aβ1–42-induced Ca2+ response was almost completely inhibited by APV, an antagonist of the NMDA receptor, and Ifenprodyl, an antagonist specific for GluN2B subunit (not illustrated), in both cellular models in agreement a previous observation .
This observation needs some comments: Aβ1–42 has been described to directly activate NMDA receptors in transfected cells expressing GluN1, GluN2A, and GluN2B subunits . However one should notice the rapid desensitization of the response induced in PC12 cells and in neuronal cultures (see Figs. 4 and and6).6). By contrast [Ca2+]i is increasing without an initial peak and without desensitization upon continuous NMDA application to PC12 cells or cultured neurons (not illustrated). A more precise observation shows that the decrease rate fits with a bi-exponential curve. This means that NMDA channel activation is not the only component involved but that a second component is also involved in the observed response. The nature of this process is not yet identified. In a recent study, Kessels et al. suggest a metabotropic activity of NMDA receptor induced by Aβ1–42 application . Another possibility would be a massive involvement of intracellular Ca2+ stores triggered by NMDA receptors as already observed [38–40] and a delayed subsequent enzymatic activity of the endoplasmic reticulum calcium pump, which regulates Ca2+ homeostasis and would contribute to the rapid decrease of the response. This point has to be further elucidated.
Overnight exposure of PC12 cells to 10 nM Aβ1–42 did not modify the basal [Ca2+]i in agreement with previously published data . Moreover we did not observe a significant oligomeric Aβ1–42 accumulation on the cellular membranes after chronic exposure of the cells as revealed by immunocytochemistry. Thus we were not able to show any noticeable change, which would be elicited by preincubating the cells with low concentrations of Aβ1–42.
There are some clues in the literature indicating that Aβ1–42 is acting differently depending on its cellular location or on its concentration: Puzzo et al.  showed on hippocampal slices that pmolar Aβ1–42 application was able to increase long term potentiation while μmolar Aβ1–42 application inhibited it. Acute application of Aβ1–42 was described to increase the K +-induced Ca2+ response while chronic application would induce apoptosis . Additionally different species (monomeric or oligomeric forms) of Aβ1–42 were described to interact with cellular membranes of PC12 cells . As revealed by fluorescence measurement, these different species were not exchangeable in solution  and their amount binding to the membranes depended on the peptide conformation according to their size: low concentration would favor monomeric or small oligomeric forms while high peptide concentration triggers the binding of large soluble oligomers  in agreement with our results. In our study, however, we show a priming effect of low concentrations of monomeric Aβ1–42 on the subsequent binding of large oligomers, which suggests a functional coupling between these two interaction modes of the peptide with the cellular membrane. Nag et al.  suggested that monomers or small oligomers bound to the membrane would favor further aggregation of the peptide to form large oligomers at the cellular surface. From our results we cannot exclude such a mechanism. More recently Bateman and Chakrabartty  showed that monomeric Aβ1–42 stays at the surface of differentiated PC12 cells while large oligomers are rapidly internalized, suggesting different properties of these two Aβ1–42 forms. By introducing a sequential aspect our results corroborate and further characterize the Aβ1–42 binding observed previously [43–45]. Furthermore we also demonstrate that this mode of Aβ1–42 interaction is also relevant for cultured cortical neurons.
The functional consequences of the priming effect on Aβ1–42 binding are not even in cultured primary neurons or in PC12 cells. Indeed the amyloid peptide has been suggested to interact with extra synaptic NMDA receptors involving GluN2B subunits [36, 46]. In mouse neurons, the primed binding of Aβ1–42 oligomers induced a decrease in the Ca2+-response and in the membrane expression of GluN2B. This has been often described in the recent literature (for a review see ). In PC12 cells, however, the chronic exposure to low Aβ1–42 concentrations induced a significant increase of the Ca2+ response induced by a subsequent acute application of 1 μM Aβ1–42 and an increased expression of GluN2B on the plasma membranes of the cells. Such a positive priming effect of low Aβ1–42 concentrations was also observed in PC12 cells by Innocent et al. . In this case the K +-induced Ca2+ response was increased. Given the rapid effect observed (within 5min) and the fact that it was revealed in non permeabilized cells, this phenomenon should not be assigned to protein synthesis but rather to a translocation of GluN2B from the cytoplasm or from endogenous vesicles toward the plasma membrane. Is this discrepancy with cultured mouse neurons reflecting species differences between rat PC12 cells and murine cortical neurons? Recently Liu et al.  also reported that Aβ42 induced an increase of GluN1 and GluN2B subunits in rat hippocampal neurons in culture, which would support this hypothesis. However numerous previous data suggested a decrease in GluN2B after Aβ1–42 oligomers application to rat primary neuronal cultures or to organotypic slices (for a review, see ). Several pathways regulating GluN2B including ERK/MAPK, ApoER2/reelin, STEP61, casein kinase, or calpain activity have been described to be involved in the toxicity of Aβ1–42 oligomers . The PC12 cell line derives from a rat tumor and is subject to instability. Whether these regulating pathways are functional in this cell line remains to be clarified.
Hence the present study describes a functional link between two forms of Aβ1–42 interacting on the plasma membrane of PC12 cells and of cultured cortical neurons. Considering these observations we would suggest that a low physiologic Aβ1–42 concentration induces monomers (or small oligomers) binding to cellular plasma membranes, which in turn favors large oligomers to bind when the peptide concentration increases. Coming back to the physiopathology of the disease, this could explain the deep gradient of degenerated neurons around amyloid plaques in the brain tissue and the slow progression of the disease. A continuous secretion of amyloid peptides does not notably hinder neuronal function but prepare the bad end of the neurons. Degeneration maximally occurs in close vicinity to senile plaques. These plaques can be considered as Aβ1–42 reservoirs generating high concentration of soluble amyloid peptide oligomers. Due to the size of the oligomers, the diffusion of the peptide through the brain tissue is limited and therefore the concentration gradient should be high around the amyloid plaques.
This hypothesis does not explain the spreading of the disease throughout the brain during the course of the pathology. It has been suggested similarities between prion diseases and AD involving intracellular transport of misfolded particles [49, 50] before extracellular secretion. According to this hypothesis, monomers or small oligomers of misfolded proteins could serve as seeds for larger pathogenic particles. Our present observation completes this suggestion: soluble misfolded Aβ1–42 could be not only seeds for larger oligomers, protofibrils, and ultimately plaques, but also anchors for soluble large oligomers on cellular membranes.
Additionally Aβ1–42 is observed not only in CNS but also in peripheral tissues including vessels and blood in a number of pathologies. As an example Aβ1–42 is secreted by platelets in AD. It should be of interest to determine whether this priming effect on Aβ1–42 oligomers binding capacity applies to a broad variety of peripheral cell types and could serve as a biomarker for the disease.
This work was supported by CNRS, by a special grant for Alzheimer’s disease research from the Fondation pour la Recherche Médicale and received a financial support from the SATT Conectus. We are grateful to Odile Proscksh for her technical support.
Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/14-2529r2).