Direct interaction of beta-amyloid with Na,K-ATPase as a putative regulator of the enzyme function

By maintaining the Na+ and K+ transmembrane gradient mammalian Na,K-ATPase acts as a key regulator of neuronal electrotonic properties. Na,K-ATPase has an important role in synaptic transmission and memory formation. Accumulation of beta-amyloid (Aβ) at the early stages of Alzheimer’s disease is accompanied by reduction of Na,K-ATPase functional activity. The molecular mechanism behind this phenomenon is not known. Here we show that the monomeric Aβ(1-42) forms a tight (Kd of 3 μM), enthalpy-driven equimolar complex with α1β1 Na,K-ATPase. The complex formation results in dose-dependent inhibition of the enzyme hydrolytic activity. The binding site of Aβ(1-42) is localized in the “gap” between the alpha- and beta-subunits of Na,K-ATPase, disrupting the enzyme functionality by preventing the subunits from shifting towards each other. Interaction of Na,K-ATPase with exogenous Aβ(1-42) leads to a pronounced decrease of the enzyme transport and hydrolytic activity and Src-kinase activation in neuroblastoma cells SH-SY5Y. This interaction allows regulation of Na,K-ATPase activity by short-term increase of the Aβ(1-42) level. However prolonged increase of Aβ(1-42) level under pathological conditions could lead to chronical inhibition of Na,K-ATPase and disruption of neuronal function. Taken together, our data suggest the role of beta-amyloid as a novel physiological regulator of Na,K-ATPase.

factor in the facilitating neuronal growth, cellular survival, the modulation of synaptic function and defense against oxidative stress 7 . The physiological concentrations of Aβ benefit learning and memory processes 8 , however its precise function has not been elucidated until now. Experiments on neuronal cell cultures indicated that at submicromolar concentrations Aβ affects cell differentiation, while at micromolar concentrations Aβ has a cytotoxic effect 9 . Recently it has been shown that amilospheroids, the neurotoxic 10-15-nm spherical Aβ oligomers derived from AD-patients, target neuronal Na,K-ATPase and inhibit its hydrolytic activity 10 . However the amilospheroids are not formed at the early stages of AD, and cannot be the cause of Na,K-ATPase inhibition at these stages of disease. We have hypothesize that the monomeric Aβ can directly interact with Na,K-ATPase and regulate its activity.
In this study using isothermal titration calorimetry we have shown that Aβ  and Na,K-ATPase form equimolar complex with the dissociation constant (K d ) 3 μ M. We have found that the Na,K-ATPase:Aβ  complex formation leads to the dose-dependent loss of hydrolytic activity of the enzyme in solution and in neuroblastoma cells SH-SY5Y. Treatment of these cells with Aβ (1-42) also leads to a decrease of Na,K-ATPase transport activity and Src-kinase activation in the first 30 min of incubation. Using fluorescently-labeled Aβ  we further show that amyloid peptide does not penetrate into cells within this time period. Hence, Aβ  directly interacts with the extracellular part of Na,K-ATPase and modulates its function. We have built a model of the Aβ (1-42):Na,K-ATPase complex, according to which binding of Aβ  occurs in the "gap" between the alpha-and beta-subunits, disrupting the function of Na,K-ATPase.

Results
Aβ(1-42) interacts with Na,K-ATPase and inhibits its hydrolytic activity in solution. We used purified preparation of Na,K-ATPase from duck salt glands with high Na,K-ATPase-specific activity which is difficult to achieve when purifying Na,K-ATPase from other tissues 11 . The protein from duck is a homolog of the Na,K-ATPase α 1β 1 isozyme from human tissues. A typical set of isothermal titration calorimetry (ITC) data for Aβ  binding to Na,K-ATPase at 25 °C is shown in Fig. 1A. Titration curve was well fitted to a model with one Aβ  binding site per Na,K-ATPase molecule. Complex formation between Na,K-ATPase and Aβ  was characterized by the following parameters: dissociation constant (K d ) 3 ± 1 μ M, enthalpy variation (Δ H) − 13.2 ± 2.0 kcal/mole and the entropy contribution to binding (TΔ S) − 5.8 ± 1.2 kcal/mole. These thermodynamic parameters indicate that the binding is enthalpy-driven and is largely caused by electrostatic interactions 12 . The shape of ITC curve allowed to obtain the stoichiometry (N) of Aβ  binding to Na,K-ATPase equal to 1.1, which demonstrates that Aβ  is in the monomeric state.
Aβ  binding to Na,K-ATPase leads to the tremendous reduction of the enzyme hydrolytic activity in the dose-dependent manner (Fig. 1B). At 40 μ M Aβ (1-42) almost fully inhibits the ability of Na,K-ATPase to hydrolyze ATP.
The Aβ (1-42) concentration resulting in the enzyme losing half of its hydrolytic activity (IC50) is 24 ± 1 μM. The difference between K d value obtained by ITC and the IC 50 value can be attributed to the presence of adenine nucleotides in the medium during hydrolytic activity measurements. Similar difference between the K d and IC 50 values is observed for Na,K-ATPase when it interacts with various cardiotonic steroids 13 . Dilution of the enzyme solution containing 40 μ M of Aβ (1-42) down to 10 μ M leads to recovery of the Na,K-ATPase activity (Supplementary Figure S1) suggesting that the enzyme inhibition is reversible.
Aβ(1-42) inhibits hydrolytic and transport activity of Na,K-ATPase in neuroblastoma cells. The inhibitory action of Aβ (1-42) on Na,K-ATPase was verified using neuroblastoma cells SH-SY5Y. The cells were incubated in the presence of Aβ (1-42) for 30 min. Then cells were lysed and hydrolytic activity of Na,K-ATPase in the lysates was measured ( Fig. 2A). At concentration of Aβ (1-42) 40 μ M inhibition of the enzyme activity reached 95% and the IC 50 value is equal 4 ± 2 μ M which correlates well with the data on the inhibition of the purified Na,K-ATPase preparation (Fig. 1B).
Transport activity of Na-K-ATPase was evaluated in intact cells using 86 Rb as a radioactive tracer for K + . Pre-treatment by 10 μ M Aβ (1-42) with or without 100 μ M ouabain was followed by the addition of 86 Rb + to the cells (see Methods). As shown in Fig. 2B, total K + influx in the cells treated with Aβ (1-42) was suppressed by 36% compared to the control values. Whereas active K + influx, caused by Na,K-ATPase activity, was suppressed by 65%, the passive ouabain insensitive component (passive K + influx) was unaltered by Aβ . Thus, introduction of Aβ  to the extracellular medium resulted in the inhibition of Na,K-ATPase transport activity in the cell. been built using the shark Na,K-ATPase crystal structure (PDB code 2zxe) and Aβ (1-42) modeled structure. In all 9 top models of the complex generated by docking 15 , Aβ (1-42) was localized within the same binding site between the extracellular parts of the alpha-and beta-subunits of Na,K-ATPase (Supplementary Figure S6). The best score model is shown in Fig. 4. Pro125, Glu909, and Arg979 of the alpha-subunit and Glu88, Ser101, Glu273 and Arg292 of the beta-subunit form hydrogen and ionic bonds with Aβ (1-42) (Supplementary Figure S7). Electrostatic interactions in the Aβ (1-42)-Na,K-ATPase interface are in agreement with the enthalpy changes

Activation of Src-kinase by Aβ(1-42).
It is known that Na,K-ATPase interacts with Src-kinase to form a receptor complex 16 . To reveal the effect of Aβ (1-42) on receptor function of Na,K-ATPase, we have incubated SH-SY5Y cells with 10 μ M of Aβ (1-42) for 30 min and measured the level of phosphorylation of the Src-kinase residue Tyr416. Autophosphorylation of Tyr416, which leads to activation of Src-kinase, was described for Src-kinase dissociation from the complex with Na,K-ATPase when binding ouabain 16 . We have found that incubation with Aβ (1-42) leads to an over 30% increase of phosphorylation level of the Src-kinase residue Tyr416 (Fig. 5). Thus binding of Aβ (1-42) to Na, K-ATPase, by analogy with ouabain leads to the activation of Src-kinase.

Discussion
Decrease of Na,K-ATPase activity is one of the early markers of AD 2,17-20 . Na,K-ATPase plays a key role in maintaining ion balance and resting potential and hence is important for the regulation of neuronal excitability, and synaptic transmission. It is also involved in memory formation, possibly due to the effect on synaptic plasticity of neurons and long-term potentiation 3,5 . There are numerous data indicating that the decrease in activity of Na,K-ATPase is not a consequence of neurodegenerative processes but, on the contrary, precedes them and causes the disruption of neuronal activity, deregulation of cellular ion homeostasis, calcium overload and neuronal death 2,3,5,17,21,22 . On the other hand, there exists evidence that Aβ is required for the regulation of synaptic transmission 7 . We hypothesized that the physiological role of Aβ may involve regulation of the receptor and transport functions of Na,K-ATPase.
We have shown that Aβ  formed an equimolar complex with Na,K-ATPase that was unable to cleave ATP and transport ions (Figs 1 and 2). Inhibition of the Na,K-ATPase function in intact cells by Aβ (1-42) occurred within 30 minutes (Fig. 2B), whereas penetration of the Aβ (1-42) into the cells could be detected much later in time (Fig. 3). This observation suggested that the inhibitory effect is mediated by binding of the Aβ  to the extracellular domain of the enzyme. In line with these data, the model of the Aβ (1-42):Na,K-ATPase complex (Fig. 4) implied binding of the peptide to the extracellular part of Na,K-ATPase, in the "gap" between the alpha-and beta-subunits. According to our model, binding Aβ (1-42) hinders the movement of the alpha-and beta-subunit towards each other during the catalytic cycle 23 , disrupting the Na,K-ATPase function. Hence Aβ is a potential physiological endogenic regulator of Na,K-ATPase. Observed acute effects cannot be directly translated to the pathogenesis of Alzheimer's disease associated with chronic effects of the Aβ .
The cardiotonic steroids (CTS) are known as endogenous regulators of Na,K-ATPase, which bind to the extracellular enzyme domains affecting its activity and receptor function 24,25 . K d value for the Aβ (1-42) complex with Na,K-ATPase (3 μ M) is close to the K d for complexes of Na,K-ATPase with such CTS as ouabain (0.8 μ M) and marinobufagenin (1.5 μ M) at the same conditions 26 . CTS and Aβ are present in the blood plasma in similar concentrations, within several nanomoles 27 . Taking into account that binding of CTS to Na,K-ATPase triggers activation of Src-kinase 28 , it is reasonable to suggest that binding of Aβ to Na,K-ATPase may also cause the activation of the Src signaling pathway. Indeed, we have shown that incubation of neuroblastoma cells with Ab, leading to decrease of Na,K-ATPase activity, causes the activation of Src-kinase (Fig. 5). Previously, it was shown that treatment of neuronal cells by Aβ  leads to phosphorylation of a number of proteins within 10 minutes, and that this phosphorylation was blocked by addition of the Src family tyrosine kinase inhibitor 29 . Thus, it is tempting to suggest that physiological function of CTS and Aβ as signaling entities, as well as toxicity of abnormally high doses of both messengers, may share certain molecular mechanisms.
Endogenous CTS level varies in different pathologies, including ischemia 30 and hypertension 24 , leading to the modulation of Na,K-ATPase activity and launching several signaling cascades in cells. There is only limited data on the causes of changes in the physiological levels of Aβ . It has been shown that the level of Aβ rises during stress due to the increase of APP processing 31 . It also rises upon the changes in hormonal status, particularly in the level of luteinizing hormone and testosterone 32 , and as a result of injuries 33 . Short-term reduction in Na,K-ATPase activity may allow cells to maintain the ATP pool and promote their viability 11 . Thus, short-term increase in the level of Aβ in the extracellular space and subsequent binding of Aβ to Na,K-ATPase would allow to regulate the activity of neuronal Na,K-ATPase under pathological conditions. Binding of Aβ would induce a number of signaling cascades increasing the adaptive capacity of the cells. As we have shown Aβ  in the 10 μ M concentration decreases transport activity in the intact cells by 65% (Fig. 2), but has no toxic effect on cells in short incubation periods ( Supplementary Figures S3 and S4) though it causes Src-kinase activation (Fig. 5). Even at the Aβ (1-42) concentration 40 μ M, which completely inactivates Na,K-ATPase, the fraction of dead neuroblastoma cells in the population did not increase, although a disruption of ionic homeostasis and osmoregulation was observed ( Supplementary Figures S2-S4). Exposure to Aβ (1-42) for longer periods of time led to the loss of viability of SH-SY5Y cells (Supplementary Figure S5), and, as we showed earlier, of the human differentiated neuronal cells 9 and neuroblastoma cells SK-N-SH 34 . Hence, disruption of the Aβ homeostasis leading to a prolonged increase in Aβ local concentrations, e.g. in the synaptic cleft, will lead to the inhibition of Na,K-ATPase, imbalance of the Na + and Ca 2+ levels and the loss of neuronal excitability. Local concentrations of the peptide can vary widely and must reach high values before the aggregates and β -amyloid plaques will form. It was shown that at 1 nM concentration of Aβ  in the extracellular fluid, its accumulation was observed in endosomes/vesicles of SH-SY5Y cells after 24 h incubation 35 . At the concentration of 25 nM in extracellular fluid, the Aβ (1-42) concentration in vesicles constituted over 2.5 μ M after 24 h incubation, i.e. a similar concentration was also reached at the cell surface 35 . These data imply that impaired homeostasis of β -amyloid will lead to its accumulation at cell surface creating high local concentrations of the peptide in the area of Na,K-ATPase localization even at the nanomolar concentrations of Aβ  in extracellular fluid. This is consistent with the hypothesis that disruption of the electrotonic properties of neurons observed in AD is a direct consequence of the long-term inhibition of Na,K-ATPase at early stages of the disease. Additionally, beta-subunit of the enzyme is involved in intercellular adhesion 36 , and Aβ binding to the extracellular part of Na,K-ATPase may lead to disruption of these contacts and to neuronal degeneration.
To summarize, we have shown that the monomeric Aβ (1-42) forms equimolar complex with the Na,K-ATPase, leading to acute inhibition of the enzyme transport and hydrolytic functions. Formation of Ab-ATPase complexes is therefore of high physiological importance. For example, short-term increase of Aβ (1-42) level under stress conditions 31-33 will lead to reversible inhibition of Na,K-ATPase which could prevent  37 . To do this, cold hexafluoroisopropanol (Fluka) was added to dry Aβ (1-42) to a concentration of 1 mM and incubated for 60 min at room temperature. Then this solution was put on ice for 10 min and aliquoted into non-siliconized microcentrifuge tubes (0.56 mg peptide per tube). Peptide in the tubes was dried under vacuum using Eppendorf Concentrator 5301. Dried peptide was stored at − 80 °C. 2.5 mM peptide stock solution was prepared by adding 20 μ L of 100% anhydrous DMSO (Sigma-Aldrich) to 0.22 mg peptide and incubating for 1 h at room temperature. For use in the experiments, the peptide was diluted to the required concentration with buffer solution. Equivalent amount of DMSO was added to the control samples in all experiments. Only freshly prepared peptide solutions were used for all experiments. By dynamic light scattering (DLS) and turbidity methods it was shown that there were no aggregates and higher molecular weight oligomers in Aβ (1-42) preparation (40 μ M) after 1 hour following preparation of the water solution. To determine the turbidity the optical density at 405 nm of freshly prepared solutions of Aβ (1-42) was recorded on Jasco V-560 spectrophotometer within one hour. The absorbance of the solution does not change, which allows to make a conclusion about the absence of particles in solution with a size of 1-100 nm. DLS method allows measuring the particle size from 0.6 to 10 nm. The DLS measurements were carried out on a Zetasizer Nano ZS apparatus (Malvern Instruments Ltd.). According to our data the freshly prepared solution of amyloid (40 μ M) does not contain particles in this size range.
Cell culture. Undifferentiated human neuroblastoma SH-SY5Y cells were cultured in DMEM/F-12 media, containing 20% fetal bovine serum (FBS), 0.2 mg/ml Penicillin-Streptomycin, and 0.01 mM sodium pyruvate on uncoated petri dishes at 37 °C in humid atmosphere with 5% CO 2 . Notably, non-differentiated SH-SY5Y cells are a ubiquitous model frequently used to study mechanism of the Аβ (1-42) effects on cells 35,40 . Amount of Na,K-ATPase at the SH-SY5Y cell surface was close to the amount present at the surface of primary neuronal granular cells, as confirmed be similar values of Na,K-ATPase transport activity in these cells (active K + influx 0.40 mM/(mg × h) for SH-SY5Y cells, Figure 2B, versus 0.35 mM/(mg × h) for primary neurons 41 ). Cell lysates containing Na,K-ATPase were prepared by repeated freezing-thawing cycles 41 . Hydrolytic and transport Na,K-ATPase activity measurements. Hydrolytic activity of Na,K-ATPase in the purified preparation and in cell lysates was measured as ouabain-sensitive (1 mM) ATP cleavage in the medium containing 130 mM NaCl, 20 mM KCl, 3 mM MgCl 2 , 3 mM ATP, and 30 mM imidazole, pH 7.4, 37 °C as described in [41][42][43] .
Transport activity of the enzyme was measured in a separate set of experiments in intact cells using 86 Rb as a radioactive tracer for K + . Ouabain-sensitive unidirectional K + ( 86 Rb) influx was measured on the Petri dishes (diameter 3 cm) containing Tyrode solution. To distinguish between the active, Na,K-ATPase-mediated, and passive K + influx, ouabain at the final concentration of 100 μ M was added to one-half of the samples 15 min before the addition of the radioactive tracer. Aβ (1-42) was added 10 min before introduction of the radioactive tracer. Flux measurements were started by adding 86 RbCl (~0.5 μ Ci/ml cell suspension; Perkin-Elmer). After 10-60 min of incubation with 86 Rb, flux was stopped by immediate dilution with 10 ml ice-cold washing medium (100 mM Mg(NO 3 ) 2 and 10 mM imidazole, pH 7.4, 4 °C). After additional washing from external 86 Rb cells were lysed in 5% TCA. Radioactivity of cells and incubation medium was measured using a Tri-Carb 1600 TR liquid scintillation counter (Packard) in water phase (Cherenkov effect). Accumulation of 86 Rb + by the neuroblastoma cells was linear within at least 30 min incubation with the tracer. Uptake over this time interval was used for the calculations of the unidirectional K + influx. Unidirectional fluxes (J) were calculated using the following equation: J = (A c /A m [K + ] e )/tm protein , where A c and A m are radioactivity of cells in 1 ml suspension and 1 ml medium, respectively; m protein is the amount of protein (mg/ml cell suspension) corrected for the amount of viable cells in suspension, [K + ] e is K + concentration in the incubation medium, and t is the equilibration time with the tracer.

Isothermal titration calorimetry (ITC). The thermodynamic parameters of Aβ (1-42) binding to
Na,K-ATPase were measured using a MicroCal iTC200 instrument, as described elsewhere 44,45 . Experiments were carried out at 25  Aβ(1-42) labeling and monitoring its penetration into cells.  was labeled by NHS495 dye (amine-reactive derivative of fluorescein, E ex /E em = 488/535 нм). Six microliters of Aβ (1-42) (2.5 mM) mixed with 15 μ l of NHS495 (1 mM) were incubated one hour at room temperature. To remove the unreacted "free" dye the volume of reaction mixture was adjusted to 500 μ L by phosphate buffer, pH 7.4 and then passed through desalting column PD MidiTrap G-25 (GE Healthcare). Labeled Aβ  was added to the cell suspension at a concentration of 10 μ M. Penetration of Aβ (1-42) into cells was evaluated by flow cytometry on GALLIOS flow cytometer (Beckman Coulter). The distinction between internalized and surface-bound fluorescein-Aβ (1-42) was performed using trypan blue (1.2 mg/ml) for quenching surface-bound fluorescence as described in 14 . The fluorescence of living cells was assessed after 10, 30, 60, 120, 180 min and 19 h with or without trypan blue.
Flow cytometry. Analysis of the cells was performed on GALLIOS flow cytometer (BeckmanCoulter). In our analysis we have excluded cell debris lying outside the gate R1 (Supplementary Figure S2). The size of the cells was assessed by forward side scatter (FS). Cells with damaged membrane were determined according to staining with propidium iodide (PI) (Sigma) (Ex/Em = 535/617 nm). Mitochondrial membrane potential in intact cells (Ψ ) was detected by MitoProbe DilC1(5) (Ex/Em = 638/658 nm) (Invitrogen), according to Mironova et al. 46 . The Ca 2+ level in intact cells was estimated by staining with fluo-4 (Ex/Em 494/516 nm) (Molecular Probes), according to Mitkevich et al. 47 . Each value is the mean of at least three independent experiments with triplicate samples ± SD.
Immunoblot analysis. Cells were incubated with 10 μ M Aβ (1-42) for 30 min and then lysed in the RIPA buffer (25 mM tris-HCl, pH 7.6, 150 mМ NaCl, 1% Nonidet-P40, 0.1% SDS, 1% sodium deoxycholate) containing 1 μ M of PMSF with stirring at 4 °C for 1 h. The probes were then centrifuged at 13000 g for 10 min and the supernatant was collected. Proteins of cell lysates were separated on SDS-PAGE and transferred to a PVDF membrane. After membrane blocking in 5% nonfat milk in PBST, the detection of phospho (Tyr 416) Src and total Src was carried out by incubating the membrane in the solution of appropriate rabbit polyclonal antibodies (both from Cell Signaling Technology) in PBST. The level of β -actin was also estimated using mouse monoclonal anti-β -actin antibody (Ambion). Visualization of the proteins was performed by the appropriate horseradish Scientific RepoRts | 6:27738 | DOI: 10.1038/srep27738 peroxidase-conjugated secondary antibodies provided by the enhanced chemiluminescence SuperSignal ™ West Femto Maximum Sensitivity Substrate kit (ThermoScientific). Chemiluminescence was detected using Bio-Rad ChemiDoc MP instrument. Densitometric analysis was performed with Image Lab program (Bio-Rad) and the results were expressed as ratio of phospho-Src to total Src band intensity (phospho-Src/Src). Statistical analysis. The comparison of data groups was performed using Student's t-test; p < 0.05 was considered significant. ITC, hydrolytic activity and flow cytometry data are presented as means of at least three independent experiments ± SD. Data of K + ( 86 Rb) influx are presented as means of four independent experiments ± SE.
Modeling of the structure of Aβ(1-42):Na,K-ATPase complex. Model of the Aβ (1-42) peptide was constructed using as templates the Aβ (1-16) structures 1ze7 and 4f37 (PDB ids) and the ab initio model of Aβ  built using the server Bhageerath 48 . The resulting model of Aβ  was minimized in the AMBER99 force field with the MOE program version 2013.08. Modeling of the Aβ (1-42):Na,K-ATPase complex was performed using the structure of Na,K-ATPase from shark glands 2zxe (PDB id) solved at 2.4 Å resolution 49 , and the modeled structure of Aβ . Docking has been carried out with VinaAutoDock program 15 , and the docking was constrained to cover only the extracellular part of the protein.
Determination of Cell index. Analysis was performed on xCELLigence real time cell analyser (RTCA) (ACEA Biosciences) measures focal adhesion of living cells in real-time 50 . Cells were seeded onto custom RTCA E-plates (ACEA Biosciences) coated with high-density gold arrays for measuring electrical impedance. The xCELLigence biosensor measures cellular adhesion, which is converted to Cell index (unit less) by the xCELLigence software (version 1.2.1). Cells were seeded into E16 well plates and allowed to recover until the cells had attained a stable Cell index. Treatment of cells on the plates with Aβ (1-42) was performed in serum free DMEM. After 4 h of treatment inactivated FBS (10%) was added into plates.