Evidence against a role of P-glycoprotein in the clearance of the Alzheimer’s disease Aβ_1–42 peptides

Abstract

It has been proposed that the amyloid-β peptides (Aβ) cause the neuronal degeneration in the Alzheimer’s disease brain. An imbalance between peptide production at the neuronal level and their elimination across the blood–brain–barrier (BBB) results in peptide accumulation inside the brain. The identification and functional characterization of the transport systems in the BBB with the capacity to transport Aβ is crucial for the understanding of Aβ peptide traffic from the brain to the blood. In this context, it has been suggested that the P-glycoprotein (P-gp), expressed in endothelial cells of the BBB, plays a role in the elimination of Aβ. However, there is little, if any, experimental evidence to support this; therefore, the aim of this investigation was to determine whether P-gp is capable of transporting Aβ peptides. Our results show that ATPase activity measured in plasma membrane vesicles of K562 cells overexpressing P-gp is not increased by the presence of Aβ₄₂, suggesting that Aβ₄₂ is not a P-gp substrate. Similarly, P-gp of pirarubicin was unaffected by Aβ₄₂. Moreover, the overexpression of P-gp does not protect cells against Aβ₄₂ toxicity. Taken together, our results support the conclusion that Aβ₄₂ is not transported by P-gp.

Introduction

Alzheimer’s disease (AD) is the most important age-related neurodegenerative disorder characterized by a gradual loss of episodic memory (Niedowicz et al 2011). It is estimated that 60 % of dementia cases are due to AD, and more than 35 million people are currently living with the disease worldwide (Niedowicz et al 2011; Wimo et al 2013) . It is expected that the number of persons that suffers AD will increase to more than 66 million by 2030 (Wimo et al 2013). The two pathological hallmarks of AD are neurofibrillary tangles and senile plaques. These plaques are proteinaceous extracellular deposits of amorphous aggregates of amyloid-β peptides (Aβ) (Cvetković-Dožić et al 2001; Selkoe 2001; Walsh and Selkoe 2004; Perl 2010). Aβ is generated in neurons by proteolytic cleavage of a transmembrane precursor protein (APP) by β-secretase and γ-secretase. As a result of this cleavage, species of 39–43 amino acids are generated and released from the membrane. The 40-residue Aβ (1–40) peptide represents the most abundant Aβ isoform inside the brain (80–90 %), whereas the 42-residue amyloid-β peptide (Aβ₄₂) is less abundant. The isoform Aβ₄₂ is believed to play a central role in AD pathogenesis (Selkoe 2001; Götz et al 2011; Lazarov and Demars 2012).

It has been shown that the Aβ peptides are neurotoxic. In the AD brain, an imbalance between the production of Aβ peptides by the neurons and their elimination through the blood–brain-barrier (BBB) may produce an increase in the levels of the peptides in the brain. This idea is the central core of what it is known as “the amyloid cascade hypothesis,” which postulates that the deposition of the toxic amyloid-β peptides in the brain is a central event in AD pathology (Hardy and Higgins 1992; Haass and Selkoe 2007; Mucke 2009). The mechanisms responsible for extruding Aβ from the brain to the blood to maintain Aβ at normal concentrations remain unknown.

P-Glycoprotein (P-gp) is expressed on the luminal side of endothelial cells of the BBB and has been proposed as a potential Aβ₄₂ transporter (Bendayan et al 2006). Lam et al. (2001) reported enhanced P-glycoprotein ATPase activity in vesicles derived from the multidrug-resistant cell line CH^RB30 in the presence of Aβ(1–40) and Aβ(1–42). In addition, Cirrito et al. (2005) studied the accumulation of Aβ in the brains of mdr1a/b^−/− double knockout (P-gp null) mice, suggesting that P-gp could prevent the risk of developing AD. Similar results observed in intact mice brain capillaries conducted by Hartz et al. (2010). Conflicting data, however, have been reported by others indicating that Aβ₄₂ is not a P-gp substrate. For example, Ito et al. (2006) described in rat BBB a LRP1-mediated transport of Aβ₄₂. In this study they found that the role of P-gp was insignificant and concluded that the transport of Aβ₄₂ could be done by other non-identified protein. In a BBB in vitro model made of endothelial cells hCMEC/D3, Nazer et al. (2008) found that the overexpression of either LRP1 or P-gp alone had no effect in the transport of Aβ₄₂.

We have been studying the P-gp and its role in the multidrug resistance (MDR) phenomenon in tumor cells (Salerno et al 2002; Darghal et al 2006; Saengkhae et al 2007). P-gp is an ubiquitous ABC family transporter with a broad range of substrates. In view of the molecular structures of the numerous known P-gp substrates, Aβ is an unlikely substrate of P-gp.

In this context, the aim of the present work was to study the ATPase and the transport activity of the P-gp in the presence of Aβ₄₂, in order to verify whether the “interaction” between these two proteins actually modifies the functioning of this ABC transporter. For this purpose, we studied: (i) the ATPase activity of P-gp in the presence of Aβ₄₂, in plasma membrane vesicles; (ii) the transport of pirarubicin, a well-known P-gp substrate, in the presence of Aβ₄₂ in intact cells overexpressing P-gp; and (iii) the toxicity of Aβ₄₂ in the same cell lines.

Materials and methods

Drugs and chemicals

Doxorubicin (DOX) (Teva) and pirarubicin (PIRA) were solubilized in water. Concentrations were determined by diluting stock solutions approximately 10⁻⁵ M with ε₄₈₀ = 11,500 M⁻¹ cm⁻¹. Stock solutions were prepared just before use. Triton X-100 was dissolved in water. HEPES buffer solution contains 20 mM HEPES, 132 mM NaCl, 3.5 mM KCl, 1 mM CaCl₂, and 0.5 mM MgCl₂ at pH 7.3, with or without 5 mM glucose. Phosphate-buffered saline (PBS) buffer solution contains 4.3 mM Na₂HPO₄, 137 mM NaCl, 2.7 mM KCl, and 1.4 mM KH₂PO₄ at pH 7.3. Oubain, H₂SO₄, ammonium heptamolybdate, ascorbic acid, and potassium antimmonium (III) tartrate were solubilized in H₂O.

The sodium orthovanadate solutions (100 mM Na₃VO₄) were prepared following indications of Goodno (1982). Before use, the solutions of Na₃VO₄ were boiled until translucent (~5 min). All reagents were of the highest quality available and were purchased from Sigma. Deionized double-distilled water was used throughout the experiments.

Electronic spectroscopy

Electronic spectra were recorded on a Varian Cary 300 spectrometer at 37 °C.

Fluorescence spectroscopy

All fluorescence measurements were performed using a Perkin Elmer LS50B spectrofluorometer.

Preparation of the Aβ₄₂ peptide solutions

Aβ₄₂ peptide (EZBiolab Laboratories) was initially solubilized, at a final concentration of 1 mM, in 1,1,1,3,3,3 hexafluoro-2-propanol or hexafluoruroisopropanol (HFIP) (Fluka). Monomers were prepared as described by Ryan et al. (2010) with some modifications. Briefly, stock solution in HFIP was aliquoted in Eppendorf tubes (20 μL/tube) and dried under N₂ gas atmosphere. The Aβ₄₂ dried films were stored at −20 °C until processing. Aliquots were resuspended at a final concentration of 5 mM in DMSO (Sigma), sonicated using a bath sonicator for 10 min, and diluted to 100 μM with a PBS buffer + 0.05 % sodium dodecyl sulfate (SDS) (Sigma). To study the aggregation kinetics of Aβ₄₂, the stock solutions were diluted to 20 μM with PBS buffer (pH 7.3) and aggregation process started with incubation at 37 °C in a water bath for 72 h.

Aggregation kinetics

In order to follow the kinetics of aggregation of Aβ₄₂, aliquots of the solution of Aβ₄₂ (30 μL) were taken at different times of incubation, and the process was followed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) in 12 % Tris–glycine gels. Silver staining method (Rabilloud et al 1988; Chevallet et al 2006) was used to detect the peptide in gels.

Cell lines and cultures

The K562 cell line is a highly undifferentiated erythroleukemia originally derived from a patient with chronic myelogenous leukemia (Lozzio and Lozzio 1975). K562 leukemia cells and the P-glycoprotein overexpressing K562/ADR cells were cultured in RPMI 1640 medium with GlutaMAX^TMI (GIBCO) medium supplemented with 10 % fetal calf serum (GIBCO) at 37 °C in a humidified incubator with 5 % CO₂. The overexpressing P-gp cell line, K562/ADR, was cultured with 400 nM doxorubicin, until 4 weeks before experiments. K562 cultures were initiated at a density of 10⁵ cells/mL grew exponentially to about 10⁶ cells/mL in 3 days. In order to have enough cells in the exponential phase for assay, cultures were initiated at 5 × 10⁵ cells/mL and allowed to grow for 24 h until use. All cultured cells were counted with a Coulter counter before use.

Plasma membrane vesicles preparation

Vesicle membranes were prepared as a described by Garrigos et al. (1993): K562/ADR cells was centrifuged (1000×g, 15 min, 4 °C). The cells were then resuspended in an hypotonic buffer (10 mM Tris–HCl, 10 mM KCl, 2 mM MgCl₂, 1 mM DTT, 1 mM EGTA, and 5 mM NaN₃ pH 7.8) at a final concentration of 2 × 10⁷ cells/mL and sonicated during 10 s (three times in ice) (VIBRACELL, Sonics and Materials Inc., Dantbury Connecticut, USA) at 20 % of maximal power. Lysed cells were centrifugated (1400×g, 15 min, 4 °C), and supernatant were placed in a tube with a layer of 46 % p/v sucrose in hypotonic buffer, recentrifugated (7000×g, 20 min, 4 °C). Layer at the sucrose interphase was removed, twice diluted with hypotonic buffer and sedimented (135,000×g, 15 min, 4 °C). The pellets containing membrane vesicles were resuspended in hypotonic buffer containing 100 mM NaCl. Membranes were frozen at −20 °C until use.

Protein determination

The protein content of K562/ADR plasma membrane was determined with the Bio-Rad Quick Start^TM Bradford kit, using bovine serum albumin (BSA) as standard.

Measurement of ATPase activity

Membrane suspensions of K562/ADR cells (30 μg proteins) were incubated in a buffer solution containing 50 mM Tris, 2 mM EGTA, 2 mM DTT, 50 mM KCl, 0.5 mM Ouabaine, and 5 mM NaN₃ (pH = 6.8). ATPase reaction was started by the addition of 1 mM ATP, and reaction mixtures were incubated for 1 h at 37 °C in a water bath under stirring. Aliquots of 0.1 mL reaction mixtures were mixed with 0.1 mL of 5 % SDS in order to stop ATP hydrolysis. Inorganic phosphate levels in samples were measured by a modification of method from Bartolommei et al. (2013). Briefly, the SDS containing samples were supplemented with 0.8 mL of a buffer containing 125 mM H₂SO₄, 0.5 mM ammonium heptamolybdate, 10 mM ascorbic acid, and 40 μM potassium antimmonium (III) tartrate, and incubated for 1 h at room temperature under stirring. Absorbance was recorded at 880 nm. Controls of reactions were included in all assays, specifically verapamil (a P-gp substrate) and sodium ortovanadate (a classic inhibitor of ATPase activity).

Determination of the P-gp-mediated efflux of pirarubicin in the presence or absence of Aβ₄₂

K562/ADR cells (10⁶ cells/mL, 2 mL per cuvette) were incubated for 30 min in HEPES buffer with sodium azide, but without glucose (energy-deprived cells). The cells remained viable throughout the experiment, as checked with Trypan blue (not shown). After addition of pirarubicin (PIRA), the decrease in the signal was monitored until steady state was reached. Since the pH of the medium was chosen to equal the intracellular pH, at steady state, the extracellular-free drug concentration (C_e) was equal to the cytosolic-free drug concentration (C_i). Before the addition of glucose, the cells were treated or not with different concentrations of Aβ₄₂ (concentrations, 0.5, 1, and 2 μM) during 5 min. Then, glucose was added, which led to the restoration of control ATP levels within 2 min and to an increase in the fluorescence signal due to the efflux of anthracycline. ATP-dependent PIRA efflux was determined from the slope of the tangent of the curve F = f(t), where F is the fluorescence intensity at the time of addition of glucose. Since under these conditions C_i = C_e, the passive influx and efflux were equal at the moment of addition of glucose, the net initial efflux represents the P-gp-mediated active efflux only (Mankhetkorn et al 1996; Marbeuf-Gueye et al 1999).

Toxicity assays

The different cell lines were incubated with different concentrations of Aβ₄₂ peptide (1, 2, and 4 μM). The assay was carried out in 96-well plates. Briefly, 10⁵ cells/well were incubated for 72 h at 37 °C and 5 % CO₂ in presence or absence of Aβ₄₂ peptide. Toxicity was evaluated using Trypan blue dye. In all experiments, the cells were grown in serum free culture medium in order to prevent the binding of Aβ to albumin (Biere et al 1996).

Results

Aβ₄₂ aggregation kinetics

The goal of this study was to test the hypothesis that P-gp is a transporter for Aβ₄₂. A major impediment to the study of Aβ₄₂ is its low aqueous solubility and rapid aggregation. In order to ensure consistent responses induced by Aβ₄₂, we followed the dissolving protocol proposed by Ryan et al. (2010). Using this protocol, we were able to obtain suspension of Aβ₄₂ consistently made of monomeric forms. The first step of the Aβ aggregation process is the formation of low molecular weight species. Conformational changes occur with time, giving rise to the formation of higher molecular weight species responsible of the peptide toxicity. As shown in Fig. 1, the Aβ₄₂ aggregation process was visualized by SDS-PAGE electrophoresis of a solution of Aβ₄₂ (20 μM), incubated during 72 h in PBS buffer at 37 °C. Under these conditions, the aggregation from monomeric to oligomeric aggregates occurred within 72 h of incubation. The electrophoresis pattern showed bands localized around 4.5 kDa and around 16 kDa that we defined as monomers and tetramers of Aβ₄₂, respectively. These forms are seen from the start of the process. The highest aggregates larger than 66 kDa that could be observed in the figure, which we defined as the oligomeric forms, are seen after 48 h of incubation. In order to ensure mostly monomeric species of Aβ in the experiments that follow, we always used freshly prepared PBS buffer solutions of Aβ. Higher insoluble species resided in the stacker segment of the SDS-PAGE gel, not migrating into the resolving gel.

Fig. 1.

Aβ₄₂ aggregation by SDS-PAGE electrophoresis. Aβ₄₂ solutions, at a concentration of 20 μM and at 37 °C, were incubated for up 72 h in PBS buffer. The electrophoresis pattern showed monomers and tetramers localized around 4.5 and 16 kDa, during the first 4 h of incubation. Highest oligomeric aggregates, larger than 66 kDa, are seen after 48 h of incubation. The figure shows that Aβ₄₂ aggregation from monomeric to oligomeric aggregates occurred within 72 h of incubation

Effect of Aβ₄₂ peptide on the P-gp ATPase activity

The P-gp-mediated transport of molecules is an energy-dependent process wherein the energy is provided by hydrolysis of ATP. In order to study possible interactions between Aβ₄₂ and P-gp in native plasma membranes, we measured the P-gp ATPase activity from membrane vesicles of K562/ADR cells in the presence of Aβ₄₂. For this purpose, the inorganic phosphate (Pi) released during ATP hydrolysis was measured. In order to isolate P-gp ATPase activity from other activities in native plasma membranes, the solutions used in this investigation contained specific ATPase blockers (see solutions in “Materials and methods”). A necessary criterion to consider a molecule as a substrate for the P-gp is that the ATPase activity of P-gp has to be stimulated by the molecule to levels higher than the basal activity level (Sharom 2008).

We measured the effect of Aβ₄₂ on the ATPase activity of P-gp at Aβ₄₂ concentrations from 0 to 10 μM, and the results are shown in Fig. 2a. The various concentrations of Aβ₄₂ were obtained by diluting stock solutions of Aβ₄₂ with PBS. The data obtained with the mixture containing Aβ₄₂ peptide were corrected by subtracting the corresponding data obtained with the solvent mixture alone. As can be noted, data points basically laid over the dotted line that indicates the level of the ATPase basal activity of P-pg showing clearly, that Aβ₄₂ did not modify the ATPase activity of P-gp.

Fig. 2.

P-pg ATPase activity measured in plasma membrane vesicles of K562/ADR cells. Membrane vesicles were exposed to various concentrations of Aβ₄₂ (a), verapamil (b), and vanadate (c). The activity is plotted as the percentage of the ATPase basal activity. Values represent the means ± SD from at least four independent experiments. Reaction medium and Pi determination are described at “Measurement of ATPase activity”. The data obtained in the presence of Aβ₄₂ peptide were corrected by subtracting the corresponding data obtained with the solvent mixture employed for the preparation of Aβ₄₂ solutions alone. The resulting points basically lay over the dotted line that indicates the level of the ATPase basal activity of P-pg. Aβ₄₂ actually did not affect the ATPase activity (a). A stimulatory effect of ATPase activity was observed at 5 μM verapamil (b). For all concentrations, vanadate (c) shows an inhibitory effect of ATPase activity. Dotted lines indicate ATPase basal activity level. Where error bars are not visible, they are contained within the symbols

We also studied the P-gp ATPase activity from membrane vesicles of K562/ADR cells in the presence of several concentrations of verapamil. Verapamil is a known substrate that alters P-gp ATPase activity (Spoelstra et al 1994; Sharom et al 1995b). Verapamil was dissolved in small volumes of ethanol solutions to concentrations up to 10 μM. The same volumes of ethanol solution were tested on the membrane vesicles and showed no effect on the basal ATPase activity of the P-gp (not shown). As it is shown in Fig. 2b, low concentrations of verapamil produced moderate increases above the basal level of the P-gp ATPase activity. A maximum increase was observed at 5 μM. As expected, verapamil concentrations beyond this level were inhibitory. This particular behavior of verapamil on the P-gp ATPase activity has been previously reported (Sharom et al 1995b).

Likewise, we performed another set of measurement of the P-gp ATPase activity from membrane vesicles of K562/ADR cells, in the presence of sodium orthovanadate (0–100 μM). Sodium orthovanadate has been shown to be an inhibitor of P-gp ATPase activity (Sharom et al 1995a). Figure 2c shows the result of this experiment. As described before, all concentrations of sodium orthovanadate tested reduced the P-gp ATPase activity to levels below the basal level. These results are expected from a compound that is known to inhibit the P-gp ATPase activity and confirm the adequacy of the methodology that we followed to evaluate the effect of Aβ₄₂ peptide on the P-gp ATPase activity.

P-gp-mediated efflux of pirarubicin in K562/ADR cells

To further study the possible participation of the P-gp in the cellular transport of Aβ₄₂ we chose a method that has previously been shown to be suitable to study the transport of anthracyclines mediated by P-pg in intact cells (Frezard and Garnier-Suillerot 1990, 1991; Pereira et al 1994; Borrel et al 1994a,b; Mankhetkorn et al 1996). This method allows to accurately measuring the free cytosolic concentration of anthracyclines under steady-state conditions, as well as the initial rates of uptake and the kinetics of anthracyclines active efflux. Thus, we investigated the effect of Aβ₄₂ on the P-gp-mediated ATP-dependent PIRA efflux. Following incubation in a glucose-free HEPES buffer containing sodium azide to deplete internal ATP, cells were incubated for 30 min in media containing 1 μM of the fluorescent probe PIRA to allow the internalization of PIRA, until reaching a steady intracellular concentration. The internalization process can be followed as a decrease in the fluorescence of the external medium and is shown in the initial phase of the records in Figs. 3 and 4. When PIRA reached a steady internal concentration, glucose was added to the medium in order to initiate the synthesis of ATP by the cells, and hence, the active PIRA efflux by the P-pg. This efflux process can be seen in Fig. 3a (control). The rate of the increase in the fluorescence in the external medium is a measurement of the extent of the PIRA efflux mediated by the P-pg. To test the effect of an inhibitor of the P-gp transport activity, a similar experiment was performed in the presence of verapamil, and the results are shown in Fig. 3b. A change in the slope indicates that verapamil affects the P-gp-mediated efflux of PIRA. Comparing the slope of the fluorescence change recorded in Fig. 3a, corresponding to control conditions, verapamil remarkably reduced the slope of the PIRA efflux.

Fig. 3.

P-gp-mediated efflux of pirarubicin (PIRA) in K562/ADR cells. Effect of verapamil. P-gp-mediated efflux of PIRA was studied in K562/ADR cells in control conditions (a) and exposed to verapamil (b). The incorporation of PIRA 1 μM in K562/ADR cells followed after ATP depletion with sodium azide, and the active efflux rate was determined after restoration of the ATP synthesis by the addition of glucose. The rate of the PIRA efflux was assessed as the slope of the fluorescent change indicated with a straight line in the figure. Comparing the slopes of the fluorescence change recorded in a corresponding to control conditions, with the one in the presence of verapamil in b, verapamil remarkably reduced the slope of the P-gp-mediated PIRA efflux

Fig. 4.

P-gp-mediated efflux of pirarubicin (PIRA) in K562/ADR cells. Effect of Aβ₄₂. P-gp-mediated efflux of pirarubicin was studied in K562/ADR cells exposed to Aβ₄₂ (a). The incorporation of PIRA 1 μM in K562/ADR cells followed after ATP depletion with sodium azide, and the active efflux rate was determined after restoration of ATP synthesis by the addition of glucose. The rate of the PIRA efflux was assessed by the slope of the fluorescent change indicated with a straight line in the figure. The ratios of slopes of the fluorescence changes measured in control and in various concentrations of Aβ₄₂, $V_{\mathrm{control}}/V_{{\mathrm{A}\mathrm{\beta }}_{42}}$, are plotted in b. The ratios calculated were virtually unchanged for different concentrations of Aβ₄₂ and could be significantly fitted by a horizontal straight line. Means ± SD, n = 4

To evaluate the effect of Aβ₄₂ on the P-gp-mediated ATP-dependent PIRA efflux, we measured the slope of the fluorescence increase in the presence of Aβ₄₂. Before the addition of glucose, cells were exposed for 5 min with increasing concentrations up to 2 μM of Aβ₄₂. Figure 4a shows the results obtained when cells were exposed to 1 μM of Aβ₄₂. Clearly, the rate of PIRA efflux in the presence of Aβ₄₂ is not much different to the one observed in control. The data plotted in Fig. 4b show that the slope ratios comparing control with the various concentrations of Aβ₄₂, $V_{\mathrm{control}}/V_{{\mathrm{A}\mathrm{\beta }}_{42}}$, remain virtually unchanged. The values of the ratios measured at different concentrations of Aβ₄₂, and in control, were well fit by a horizontal straight line. This finding indicates that Aβ₄₂ does not modify the efflux of PIRA by P-gp in K652/ADR cells.

Aβ₄₂ cytotoxicity remain unchanged in cells overexpressing P-gp

To evaluate the participation of P-gp in the transport of the Aβ peptides across the plasma membrane, we evaluated the toxicity of Aβ₄₂ in the K562 cell line and in the same type of cells that overexpress P-gp, K562/ADR cells. Cells were incubated in media containing Aβ₄₂, and the viability of the cells was measured after 72 h using Trypan blue dye, as described in “Materials and methods.” Figure 5 displays the percentage of dead K562 cells (empty bars) and K562/ADR cells (filled bars) measured from cultures containing increasing concentrations of Aβ₄₂. The results showed that after 72 h. Aβ₄₂ displayed a remarkable cytotoxicity on both cells types. Under these conditions, Aβ₄₂ at a concentration of 4 μM had already produced almost total mortality of the two cell types. Furthermore, the magnitude of the mortality induced by each concentration of Aβ₄₂ was very similar and independent of P-pg overexpression. This result strongly suggests that overexpression of P-gp does not offer any protection against Aβ₄₂ cytotoxicity.

Fig. 5.

Toxicity of Aβ₄₂ peptide in K562 and P-pg overexpressing K562/ADR cells. The toxicity of Aβ₄₂ peptide was evaluated by measuring the viability of cells after exposure to Aβ₄₂ for 72 h. Cell viability was assessed using the Trypan blue dye method. The K562 (empty bars) and K562/ADR cells (filled bars) were incubated with different concentrations of peptide during 72 h in the media without serum (means ± SD, n = 3). No difference was observed in the percentage of death in both types of cells exposed to various concentrations of Aβ₄₂

Discussion

Aβ peptides could be removed from the brain extracellular fluids (BEF) by different clearance mechanisms not completely understood. It is now accepted that, when Aβ peptides concentration in the BEF increases, either by overproduction by the neuronal tissue, or by defective clearance, Aβ will form assemblies ranging from oligomers to protofibrils, fibrils, and amyloid plaques. Indeed, the accumulation of Aβ peptides in senile plaques in the brain is a major pathological hallmark of AD.

Main clearance of Aβ peptides from the brain extracellular fluids is thought to occur at the level of BBB, which controls the exchanges of a variety of substances between blood and brain. Therefore, identification and functional characterization of receptors and transporters presents at the cerebral endothelium is important for understanding Aβ traffic. It is known that Aβ peptide influx to the epithelial cells is mediated by the receptor of advanced glycosylation products (RAGE) and by the low-density lipoprotein receptor-related protein (LRP1) (Shibata et al 2000; Deane et al 2004, 2008; Ito et al 2006; Nazer et al 2008; Fujiyoshi et al 2011). Overexpression of RAGE and a low expression of LRP1 may be related with Aβ peptide accumulation in the BEF and consequently with Alzheimer’s disease (Weiss et al 2009).

Several studies have suggested that amyloid peptide could be a P-gp substrate, and thus, this protein could participate to its clearance from brain. Lam et al. (2001) reported that human synthetic Aβ₄₀ and Aβ₄₂ can bind the purified hamster mdr1 (P-glycoprotein multidrug transporter), increase the ATPase activity of mdr1 in plasma membrane vesicles derived from the multidrug ovary hamster cell line CH^RB30, and be transported across mdr1 membrane vesicles. Cirrito et al. (2005) reported that, in Pgp-null mice, [¹²⁵I] Aβ₄₀ and [¹²⁵I] Aβ₄₂ microinjected into the CNS, clear at half the rate that they have in wild-type mice; therefore, they proposed that P-gp is involved in the transport of Aβ across the BBB in vivo and that ablation of P-gp at the BBB enhances Aβ deposition into the brain. However, it should be noted that, in this situation, LRP1 expression in the brain capillaries of P-gp knockout mice was also suppressed by 51 % compared with wild-type mice (Cirrito et al 2005). This suppression of LRP1 expression could also attenuate the elimination of Aβ40, and therefore, unexpected gene modulation and consequent alteration of systemic Aβ clearance in the knockout mice could not be ruled out (Ito et al 2006). Kuhnke and et al. (Kuhnke et al 2007), using MDR1-transfected LLC cells growing in a polarized cell layer, showed that when FITC-Aβ₄₀ and FITC-Aβ₄₂ were delivered from the basolateral to the apical compartment in the presence of an ABCB1 inhibitor, cyclosporine A, both peptides transport was significantly decreased. These results suggested impaired clearance of Aβ peptides by ABCB1 inhibition and lead them to conclude that ABCB1 played a key role in the clearance of Aβ₄₀ and Aβ₄₂ from the brain. Hartz et al. (2010) studied the accumulation of fluorescein-hAβ₄₂ in capillary lumens of isolated Tg2576 transgenic mouse brain capillaries. In this AD animal model, the brain capillary P-gp expression and transport activity were substantially reduced compared to wild-type control mice. These authors also showed that the restoration of P-gp expression and transport activity in brain capillaries significantly reduced brain Aβ levels compared to control mice. They remarked that, although brain accumulation of Aβ is not the only major contributor to cognitive impairment in AD, reducing Aβ accumulation in the transgenic hAPP (Tg2576) mouse model does delay pathology (Iadecola 2004; Blennow et al 2006; Karlnoski et al 2009). Most recently, reports from Kaddoumi’s lab (Abuznait et al 2011; Qosa et al 2012) have proposed that ABCB1 plays a major role in Aβ transport.

On the other hand, other investigators have questioned the proposed role of P-gp in the transport of Aβ peptides. Evidence against the proposal of P-gp-mediated transport of Aβ has been reported by Ito et al. (2006). This group emphasized that LRP1 partially affects the cerebral elimination of hAβ40 and argue that P-gp does not play a significant role in the brain-to-blood efflux transport of hAβ40 at BBB in rats. They suggested that molecules, yet to be unidentified, may be involved in the cerebral Aβ40 uptake process in rats. In support to this idea, Nazer et al. (2008) reported that, in Madin Darby Canine Kidney (MDCK) cells, stably overexpressing P-gp did not exhibit any increased transport of [¹²⁵I] Aβ across their monolayers when compared with MDCK cells transfected with empty vector and identically cultured. They concluded that it was possible that P-gp functions in conjunction with another transporter or requires an as-yet-unidentified cofactor for Aβ transcytosis such that its inhibition or deletion in vivo inhibited Aβ clearance and increased cortical Aβ levels, but P-gp upregulation failed to promote Aβ efflux in their cellular model.

In this study, our purpose was to clarify if amyloid-β peptide is or not a substrate of P-glycoprotein. To that aim, we studied the interaction of human synthetic Aβ₄₂ with P-gp in intact cells overexpressing P-gp and in P-gp containing membrane vesicles.

The ability of P-gp to transport substrates depends on ATP hydrolysis, and the transport of substrates by P-gp often stimulates ATPase activity (Lam et al 2001). Our results showed that Aβ₄₂ monomers, dissolved in PBS solution, were indeed able to stimulate the ATPase activity. However, the same results were obtained when same volumes of the solvent of Aβ₄₂ alone were used. In addition, Aβ₄₂ did not modify PIRA transport by P-gp in K562/ADR cells. Moreover, the extent of mortality induced on K562/ADR cells overexpressing P-gp compared to Aβ₄₂ toxicity on K562 cells was similar. Taken together, the experimental results described in this investigation strongly support the conclusion that Aβ₄₂ is not a substrate for P-gp and that Aβ₄₂ and P-gp do not interact.

The discrepancy between our results and those proposed by others could be due to the use of difference cell lines. This possibility was already suggested by Abuznait and Kaddoumi (2012) and also by others (Ohtsuki et al 2010), who have proposed that the mechanism of transcellular transport of Aβ depends on the cell line. To this possibility, we like to add that the source and the preparation of the Aβ solutions could also affect the results. Aβ is prone to form a complex variety of aggregates, and the aggregation pattern of synthetic peptide depends on the dissolving protocol followed to prepare the Aβ solutions. Therefore, Aβ preparation could introduce a lot of variability in the results, and most of the studies concerning the conceivable P-gp-mediated transport of Aβ do not provide precise description of the preparation of Aβ solutions, neither the state of aggregation of the peptide used (i.e., monomers, dimers, etc.). Furthermore, binding of Aβ with fluorescein or fluorescein isothiocyanate (FITC) could also modify the peptide behavior in solution and in membranes.

Our results also can be analyzed from the point of view of the molecular interaction of the substrate required for the transport process by P-gp. Various works suggest that P-gp substrates are usually organic molecules with sizes ranging from <0.2 kDa to almost 1.9 kDa. Additionally, the substrates of P-gp should be electrically neutral or positively charged (Spoelstra et al 1992; Sharom et al 1998; Schinkel and Jonker 2003; Fromm 2004; Sharom 2008; Eckford and Sharom 2009; Hill et al 2013). Aβ₄₂ peptide is a hydrophobic molecule of about 4.5 kDa and is negatively charged at physiological pH. Therefore, from this criterion, Aβ peptide does not meet the characteristics of molecular weight and charge required for P-gp substrates. Sharom et al. (1996) studied the interactions of a series of synthetic hydrophobic peptides with P-gp. They found that the maximal interaction corresponds to small peptides made of three residues, and the interaction decreases with increasing the peptide length. Interestingly, no cytosolic peptides occurring naturally in cells have yet been described as P-gp substrates in the literature.

In order to be transported by P-gp, Aβ should be located in the substrate side of the protein. All models proposed for the function of the P-gp describe that substrate binding site of P-gp is localized in the cytoplasmic leaflet of the membrane bilayer (Higgins and Gottesman 1992; Varma et al 2003). Following these models and hypothesizing that Aβ could be a P-gp substrate, the transport of this peptide should first involve its passage through the membrane in order to be localized in the cytoplasm. It is very unlikely that the Aβ peptides will cross the cytoplasmic membrane, like the known cell penetrating peptides (CCPs), to localize in the cytoplasmic leaflet of the membrane. Moreover, the Aβ peptides are highly hydrophobic and form aggregates in an aqueous medium. Therefore, if found in the cytoplasm, Aβ would likely aggregate, making it difficult to be transported by P-gp.

The size of the active site of P-gp is ranked between 5 and 50 Å (Loo and Clarke 2001; Sauna et al 2004). The size of Aβ aggregates could be in the range of 50–100 Å, and the size of one single alpha helix of Aβ₄₂ is about 63 Å (Roychaudhuri et al 2009; Morgado and Fändrich 2011).

As it was mentioned above, to be transported by P-gp external Aβ needs to cross the membrane to the cytosol to reach the substrate binding site of P-gp. So far, there is not a defined answer to the question of how external Aβ can insert and cross the plasma membrane, either in its monomeric or the aggregated forms. To the extent that the crossing through plasma membrane of the cell penetrating peptides (CCPs) has been largely studied, one could try to consider Aβ peptides like CPPs. A thoughtful analysis shows that this consideration is groundless. CCPs are defined as peptides with a length between 5 and 30 amino acids, able to pass through the plasma membrane by mechanisms dependent or independent of energy, and producing low cellular toxicity (Lindgren et al 2000; Lundberg and Langel 2003; Mussbach et al 2011; Milletti 2012; Bechara and Sagan 2013). The Aβ₄₂ peptide, on the other hand, has a 42 amino acids sequence and is highly toxic to cells through interaction with membranes (Verdier et al 2004). Furthermore, most CCPs contain positive charges (Schmidt et al 2010), whereas Aβ₄₂ is negatively charged at physiologic pH. Although hydrophobic peptides are able to fit into the membranes, most probably they must remain attached to the membranes reducing the efficiency of internalization (Walrant et al 2012). It has been generally presumed that peptides insertion in the membrane could modify the stability of the membranes and the functioning of other membrane proteins, such as membrane enzymes and transporter. Under our experimental conditions, no modification of the transport activity of P-gp was observed when cells were exposed to different concentrations of Aβ₄₂.

In summary, the results obtained here show that, under our experimental conditions, the transport activity of P-gp was not modified by the presence of Aβ₄₂. In addition, P-gp does not offer any protection to cells from the damage induced by Aβ. All these data strongly support the idea that Aβ₄₂ is not transported by P-gp.