Amyloid β_1–42 peptide alters the gating of human and mouse α-bungarotoxin-sensitive nicotinic receptors

Abstract

The β-amyloid_1–42 peptide (Aβ_1–42), a major constituent of the Alzheimer's disease amyloid plaque, specifically binds to the neuronal α-bungarotoxin (α-BuTx)-sensitive α7 nicotinic acetylcholine receptor (α7 nAChR). Accordingly, Aβ_1–42 interferes with the function of α7 nAChRs in chick and rodent neurons. To gain insights into the human disease, we studied the action of Aβ_1–42 on human α7 nAChRs expressed in Xenopus oocytes. In voltage-clamped oocytes expressing the wild-type receptor, Aβ_1–42 blocked ACh-evoked currents. The block was non-competitive, required over 100 s to develop and was partially reversible. In oocytes expressing the mutant L248T receptor, Aβ_1–42 activated methyllycaconitine-sensitive currents in a dose-dependent manner. Peptide-evoked unitary events, recorded in outside-out patches, showed single-channel conductances and open duration comparable to ACh-evoked events. Aβ_1–42 had no effect on the currents evoked by glutamate, GABA or glycine in oocytes expressing human or mouse receptors for these transmitters. Muscle nAChRs are also α-BuTx-sensitive and we therefore investigated whether they respond to Aβ_1–42. In human kidney BOSC 23 cells expressing the fetal or adult mouse muscle nAChRs, Aβ_1–42 blocked ACh-evoked whole-cell currents, accelerating their decay. Outside-out single-channel recordings showed that the block was due to a reduced channel open probability and enhanced block upon ACh application. We also report that the inverse peptide Aβ_42–1, but not Aβ_40–1, partially mimicked the effects of the physiological Aβ_1–42 peptide. Possible implications for degenerative neuronal and muscular diseases are discussed.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder whose histological hallmark is the presence of amyloid plaques in the limbic and cerebral cortices (for review, see Selkoe, 1994). Although multiple neural systems are affected, a key feature of the neurodegenerative process is the loss of cholinergic neurons as well as nicotinic acetylcholine receptors (nAChRs) throughout the brain (Guan et al. 2000; Nordberg, 2001). The major constituent of the amyloid plaques is a 42-amino-acid β-amyloid peptide (Aβ_1–42), derived from the proteolytic cleavage of the amyloid precursor protein, which is present in almost all tissues and whose physiological functions are still unknown (Selkoe, 1994, 2001).

Aβ_1–42 has recently been reported to bind specifically and with picomolar affinity to the neuronal nAChR containing the α7 subunit (α7 nAChR) (Wang et al. 2000a,b). The binding affinity of Aβ_1–42 to α7 nAChRs appears to be at least 1000-fold higher than that of the specific blockers α-bungarotoxin (α-BuTx) and methyllycaconitine (MLA) (Wang et al. 2000a), whereas the binding affinity of Aβ_1–42 for α-BuTx-insensitive neuronal nicotinic receptors, i.e. receptors that do not contain the α7 subunit, is much smaller (Wang et al. 2000b). Consistent with these data, Aβ_1–42 functionally blocks the ACh-evoked current responses in rat hippocampal slices (Pettit et al. 2001). In cultured mouse hippocampal neurons and chick ciliary ganglion nerve cells the block appears to be specific for α7 nAChRs, with little, if any, effect on α-BuTx-insensitive nAChRs (Liu et al. 2001). A small, slowly developing block of rat α7 nAChRs expressed in Xenopus oocytes has also been described (Tozaki et al. 2002). At complete variance with all these pieces of evidence, picomolar concentrations of Aβ_1–42 have been reported to activate rat α7 nAChRs expressed in Xenopus oocytes (Dineley et al. 2002), although only upon the very first exposure of the oocyte to the amyloid peptide. No current activation was reported for rat hippocampal neurones exposed to similar Aβ_1–42 concentrations (Liu et al. 2001). A more robust current response has been described for the rat α7 nAChR carrying a point mutation in the pore-forming region (Dineley et al. 2002). The latter observation is in line with the behaviour of several antagonists of chick and human wild-type (WT) α7 nAChRs, which become agonists of the mutant receptors carrying that particular threonine-for-leucine substitution (L247T in chick, L250T in rat and mouse, L248T in human) (Palma et al. 1996, 1998, 1999; Maggi et al. 1999; Fucile et al. 2000, 2002). It is therefore quite likely that Aβ_1–42-induced activation of the mutated nAChR accounts for the Ca²⁺-induced activation of the mitogen-activated protein kinase (MAPK) pathway described in mice heterozygous for the L250T α7 nAChR allele (Dineley et al. 2001). Activation of MAPK is required for contextual and spatial memory formation in mammals (Atkins et al. 1998), which processes are impaired in AD patients. Thus, assessing the ability of the Aβ_1–42 peptide to activate human α7 nAChRs may provide clues to the physiological and/or pathological relevance of the Aβ_1–42-α7 nAChR interaction to AD. Since no functional data is available for human α7 nAChRs, in this paper we investigated the effects of Aβ_1–42 on human WT and L248T α7 nAChRs expressed in Xenopus oocytes.

Additional insights into the physio-pathological importance of the interaction between Aβ_1–42 and nAChRs may come from a different disease, inclusion body myositis (IBM), which represents the most common myopathy after 50 years of age. It is characterised by the presence of plaques, within muscle fibres, where ‘AD characteristic’ proteins, such as Aβ_1–42 and presenilin-1, are accumulated (reviewed in Askanas & Engel, 1998), together with the end-plate nAChR. To date, IBM appears to be the only non-neuronal progressive disease caused by Aβ deposition (Sugarman et al. 2002). Moreover, both the fetal and adult forms of muscle nAChRs (γ- and ε-nAChRs, respectively) share with the α7 nAChR the sensitivity to α-BuTx, and could thus possibly become targets for Aβ_1–42 as well. Indeed, block of the Torpedo nAChR by Aβ_1–42 has been reported (Tozaki et al. 2002). Furthermore, Aβ_1–42 content is elevated in the muscle of AD patients (Kuo et al. 2000b). These considerations prompted us to investigate whether Aβ_1–42 also modulates the functional properties of mammalian α-BuTx-sensitive muscle nAChRs expressed by transient transfection in human kidney BOSC 23 cells.

METHODS

Expression of nAChRs in oocytes and BOSC 23 cells

Recombinant DNA plasmids encoding human WT α7 (gift from Janssen, Belgium) and L248T α7 neuronal nicotinic subunits in the pcDNA3 vector, or the human GluR1 subunit (flip-splice variant) in the pCEP4 expression vector were intranuclearly injected into stage V-VI oocytes (2 ng cDNA in 10 nl buffer). Preparation of oocytes and nuclear injection procedures were as previously detailed (Palma et al. 1996). Oocytes were collected under anaesthesia from frogs that were humanely killed after the final collection. In other experiments, oocytes were injected with membranes extracted from mouse cortex, according to procedures described elsewhere (Miledi et al. 2002). Oocytes were used for electrophysiological determinations 1–4 days after injection. Full length cDNAs in SV-40-based pSM expression vector coding for the α1, β, γ and δ (γ-nAChR) or the α1, β, ε and δ (ε-nAChR) subunits (obtained from Dr J. Patrick, Baylor College of Medicine, Houston, TX, USA; 0.2 µg each per 35-mm dish) were transiently transfected in human kidney BOSC 23 cells (ATCC) using a Ca²⁺-phosphate method, as previously described (Fucile et al. 1996). The cell line BOSC 23 was maintained in culture in Dulbecco's modified Eagle's medium (Euroclone, UK), supplemented with 10 % calf serum (Euroclone). Cells were washed twice 8–12 h after the start of transfection and used for experiments 36–48 h after transfection.

Voltage-clamp recordings and analysis

Membrane currents were recorded in the voltage-clamp mode using two microelectrodes filled with 3 m KCl, at controlled room temperature (20–21 ^oC). The oocytes were placed in a recording chamber (0.1 ml) continuously superfused (12 ml min⁻¹) with oocyte Ringer solution. Throughout the experiments, oocyte membrane potential was maintained at −60 mV, except when otherwise indicated. Multiple ACh applications to the same oocyte were performed with at least 3 min intervals. Drugs, dissolved in oocyte Ringer solution, were applied by superfusion, using electromagnetic valves (BioLogic, France) to achieve solution exchange. Currents were digitised at 50–200 Hz (Digidata 1200 analog-to-digital converter, Axon Instruments, USA) and analysed off-line using pClamp 6.0.2 routines (Axon Instruments), as detailed in Palma et al. (1996). The ACh concentration yielding half-maximal current response (EC₅₀) or inhibition (IC₅₀) and the Hill coefficient (n_H) were obtained as previously reported (Palma et al. 1996).

Patch-clamp recordings in oocytes and BOSC 23 cells

Outside-out patch-clamp recordings were performed on oocytes whose vitelline membrane had been mechanically removed after exposure to a hypertonic solution for 10–20 min, as previously described (Methfessel et al. 1986), using patch pipettes with narrow tips, in order to avoid the occurrence of stretch-activated channels (Methfessel et al. 1986). An Axopatch 200B amplifier (Axon Instruments) was used for recordings. Excised patches were continuously superfused with oocyte Ringer solution (supplemented with ammonia, when appropriate, to the same final concentration as Aβ-containing solutions) or agonist-containing solutions via independent tubes, positioned 50–100 µm from the electrode tip and connected to a gravity-driven fast-exchanging perfusion system (RSC 200, BioLogic). This system was also used in all the experiments with BOSC 23 cells. Unless otherwise indicated, whole-cell and outside-out recordings were performed at a membrane holding potential of −70 mV for BOSC 23 cells and −50 mV for oocytes. Whole-cell currents were digitised at 500 Hz and analysed with pCLAMP programs (pCLAMP 8, Axon Instruments). The time to half-decay (T_0.5), defined as the time taken for the current to decrease from peak to half-peak value, was used to estimate the rate of current decay. Single-channel currents were recorded in the cell-attached or outside-out configuration. Data were sampled at 10 kHz and analysed after Gaussian digital filtering at 2 kHz, using a threshold-crossing method by pCLAMP 6.0.2 routines, as previously detailed (Fucile et al. 1996). Total channel open probability (NP_op) was estimated as the percentage of time spent in the open state, taking into account multiple openings. Once exposed to Aβ_1–42, cells were discarded. Statistical significance was accepted for P < 0.05.

Drugs, chemicals and solutions

Analytical grade reagents were purchased from Sigma (USA), except for methyllycaconitine (MLA, RBI, USA). Amyloid β peptides were obtained from different companies: Aβ_1–42 from Alexis (USA), Bachem (CH) or Sigma; Aβ_42–1 from Bachem; Aβ_40–1 from Sigma. Peptides were dissolved in water (Aβ_40–1), 0.1 % ammonia (Bachem Aβ_1–42 and Aβ_42–1), 100 % DMSO (Alexis Aβ_1–42) or 100 mm acetic acid (Sigma Aβ_1–42) at concentrations ranging from 0.2 to 2 mm and stored in aliquots at −20 ^oC until use. As in other studies (Liu et al. 2001; Pettit et al. 2001), no attempts were made to control the aggregation state of the peptide. However, Aβ peptides were diluted to the final concentration just prior to use, which minimises aggregation. Different lots of Aβ_1–42 from each source were used. Two lots of Bachem Aβ_1–42 were poorly effective on nAChRs, as previously reported for Aβ_1–40 from the same company (Simmons et al. 1994). Oocyte Ringer solution contained (mm): NaCl 82.5, KCl 2.5, CaCl₂ 2.5, MgCl₂ 1, Hepes/NaOH 5 (pH 7.4). The patch pipettes for outside-out recordings in oocytes were filled with a solution containing (mm): CsF 80, EGTA 5, Hepes/CsOH 5; pH 7.4. BOSC 23 cells were bathed in a salt solution composed of (mm): NaCl 140, KCl 2.8, CaCl₂ 2, MgCl₂ 2, Hepes/NaOH 10, glucose 10 (pH 7.3) (plus ammonia 0.0001 % or DMSO 0.05 %, if required). The patch pipettes for recordings in BOSC 23 cells were filled with the above saline for cell-attached recordings, or with an internal solution containing (mm): CsCl 145, BAPTA 5, Hepes/CsOH 10, Mg-ATP 2 (pH 7.3) for whole-cell and outside-out recordings.

RESULTS

Aβ_1–42 blocks WT α7 nAChRs

The main aim of this paper was to investigate the functional modulation of the human α7 nAChR upon exposure to the Aβ_1–42 peptide. The current evoked by ACh (I_ACh) was measured in oocytes expressing WT α7 nAChRs, the best characterised expression system for this receptor. In 13 oocytes tested (three donors, 13/3), Aβ_1–42 at concentrations ranging from 10 pm to 1 µm was unable to elicit current responses (Fig. 1A). Each dose of Aβ_1–42 was applied, in random order, for 2–10 s, followed by a 5 min wash-out. All the oocytes were responsive to ACh (100 µm, the EC₅₀ for this preparation, see below) (e.g. Fig. 1A), which was only applied at the end of the trials with amyloid peptide, to avoid artefacts due to solution contamination. The inhibitory action of Aβ_1–42 was investigated using the same peptide concentration (100 nm) as used by other investigators (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002; Tozaki et al. 2002). I_ACh did not change when Aβ_1–42 was co-applied with ACh (data not shown). However, after 180 s exposures to Aβ_1–42, the amplitude of the current evoked by ACh (100 µm) was markedly reduced, in agreement with previous reports (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002; Tozaki et al. 2002). In the 16 oocytes tested from four donors (16/4), the amplitude of I_ACh was −0.55 ± 0.18 µA, i.e. 51 ± 8 % (mean ± s.e.m.) of the control (Fig. 1B and C). A comparable reduction of I_ACh was observed at test potentials of −100 mV and −60 mV (4/2), indicating that the effect of Aβ_1–42 was voltage independent in this range (data not shown). Aβ_1–42 exerted no effect on current decay, with similar values of T_0.5 measured before and during treatment (0.53 ± 0.11 and 0.57 ± 0.13 s, respectively). The block was poorly reversible, as 35 min after wash-out of Aβ_1–42, I_ACh was 75.7 ± 3 % of control (e.g. Fig. 1B). To test whether the lack of full recovery was due to voltage-dependent interactions between Aβ_1–42 and α7 nAChRs, the oocyte holding potential was stepped to +30 mV for 10 s during Aβ_1–42 wash-out. However, recovery was not accelerated, I_ACh amplitude being 71 % of control 25 min after peptide withdrawal (2/2).

Figure 1. Block of WT α7 nAChRs by Aβ_1–42.

A, Aβ_1–42 at concentrations of 0.01, 0.1 and 10 nm (open bars) fails to elicit current responses in an oocyte sensitive to ACh (100 µm, filled bar). Traces representative of 8 experiments, where Aβ_1–42 concentrations were applied in random order. B, inward currents evoked by 100 µm ACh (filled bars) in an oocyte expressing human WT α7 nAChRs in standard solution (left), after 180 s preincubation with 100 nm of Aβ_1–42 (open bar, middle), and 35 min after wash-out (right). Note the incomplete recovery of I_ACh. C, block of I_ACh by increasing doses of Aβ_1–42. In each of 5 oocytes, currents evoked by ACh (100 µm) plus Aβ_1–42 after 180 s preincubation with the peptide were normalised to the response to ACh alone (-0.67 ± 0.11 µA, 5/1). Best fitting to the Hill equation yielded an IC₅₀ of 90 nm. Inset, histogram representing the effects of 100 nm Aβ_1–42, Aβ_42–1 and Aβ_40–1. I_ACh was measured after 180 s incubation with the peptides and normalised to the control value in each cell (bar labelled ACh). Bars represent mean ± s.e.m. of 5–11 oocytes (3 donors). *Statistically not different from control (Student's t test, P = 0.2). ACh concentration, 100 µm. Note the reduced effect of Aβ_42–1 as compared to Aβ_1–42 and the ineffectiveness of Aβ_40–1. D, Aβ_1–42 (0.4 µm for 180 s) is unable to block currents evoked by AMPA (50 µm plus cyclothiazide 50 µm), kainate (200 µm), GABA (1 mm) or glycine (1 mm), while blocking α7 nAChRs in Xenopus oocytes. Filled bars represent mean ± s.e.m. from 4 oocytes injected with mouse brain membranes. Inward current amplitude was: 0.08–0.68 µA (GABA); 0.01–0.03 µA (glycine); 0.09–0.13 µA (kainate); 0.3–0.44 µA (AMPA). Open bars represent mean ± s.e.m. of 5 oocytes injected with cDNAs encoding human homomeric GluR1 or α7 nAChRs, as indicated. Current ranged from 0.2–0.4 µA (hGluR1, activated by AMPA as above), from 0.2–1.0 µA (α7 nAChR, ACh 100 µm). Holding potential was −80 mV; the protocol of Aβ_1–42 treatment was as in B. *Statistically not different from control (Student's t test, P > 0.23).

The half-inhibitory concentration for Aβ_1–42 was investigated. At concentrations below 5 nm, Aβ_1–42 was not able to block α7 nAChRs, whereas at doses exceeding 100 nm there was a plateau in the inhibitory effect of the peptide, with I_ACh reaching 42 % of control (Fig. 1C). A plateau was also reported in hippocampal neurones (Liu et al. 2001). The apparent IC₅₀ of Aβ_1–42 was 90 nm (Fig. 1C).

To test for the specificity of the Aβ_1–42-induced block of α7 nAChRs, we examined the effects of the peptide on the responses evoked by other neurotransmitters. In a batch of oocytes (5/1) where I_ACh was blocked to 44.6 ± 4.5 % of control by Aβ_1–42 (0.4 µm), the peptide was ineffective on the current evoked by AMPA (50 µm plus cyclothiazide 50 µm) in oocytes (5/2) injected with the human GluR1 subunit cDNA. In oocytes injected with mouse brain membranes (4/1), the responses evoked by AMPA (50 µm plus cyclothiazide 50 µm), kainate (200 µm), GABA (1 mm) or glycine (1 mm) were also unaffected (Fig. 1D). This lack of effect cannot be attributed to the structural differences between human and rodent Aβ_1–42 (3 residues), as human Aβ_1–42 is able to inhibit mouse muscle and neuronal nAChRs (Liu et al. 2001; Pettit et al. 2001; see also below). These data show that Aβ_1–42 specifically inhibits α7 nAChRs.

We next investigated the effects of the widely used, biologically inactive peptide Aβ_40–1 (100 nm). In agreement with former studies (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002), this peptide was ineffective on I_ACh, since after a 180 s exposure to Aβ_40–1 current amplitude was 92 ± 5 % (5/2) of control (Fig. 1C, inset), a reduction that was not statistically significant (Student's t test, P = 0.2). At variance with former studies, we also tested the effects of peptide Aβ_42–1 (100 nm). To our surprise, it reduced I_ACh to 69 ± 3 % (5/2) of control values (Fig. 1C, inset). The block was not enhanced by raising the Aβ_42–1 concentration to 400 nm (data not shown). The most striking difference between Aβ_1–42 and Aβ_42–1 was the good reversibility of the latter. In fact, the I_ACh amplitude fully recovered to control values within 3 min of Aβ_42–1 removal (not shown), suggesting that the actions of Aβ_1–42 and Aβ_42–1 on α7 nAChRs are different.

The nature of the interaction between Aβ_1–42 and α7 nAChRs is controversial, as Aβ_1–42 has been reported to competitively displace α-BuTx binding (Wang et al. 2000a), whereas the inhibition of I_ACh appears to be non-competitive (Liu et al. 2001). We therefore examined how Aβ_1–42 affects the ACh dose-current response relation of human WT α7 nAChRs. In four oocytes (1 donor), during treatment with Aβ_1–42 (100 nm), neither the EC₅₀ nor n_H were significantly modified (Fig. 2A), in spite of the reduction of I_ACh amplitude, suggesting a non-competitive block of α7 nAChRs. In particular, the current evoked by a saturating ACh concentration (2 mm) was blocked to the same extent (51 ± 5 %, 4/1) as the response to 100 µm ACh (Fig. 2A, inset).

Figure 2. Non-competitive nature of Aβ_1–42-induced block of WT α7 nAChRs.

A, ACh dose-response relationships obtained from 4 oocytes (1 donor) in standard solution (○) and after 180 s preincubation with 100 nm Aβ_1–42 (•). I_ACh was normalised to values obtained at 2 mm ACh (-2.07 µA for ○, −1.02 µA for •). Best fitting with the Hill equation yielded: ○, EC₅₀= 108 µm, n_H= 1.48; •, EC₅₀= 112 µm, n_H= 1.20. Inset, typical currents evoked by 2 mm ACh (filled bars) in an oocyte representative of four (2 donors) in standard solution (left) and after a preincubation with 100 nm Aβ_1–42 (right). Note the same percentage of block as in Fig. 1A. B, currents evoked by ACh (100 µm), alone (filled bar, trace labelled C) or plus Aβ_1–42 (800 nm, open bar, trace labelled T). Between trace C and T, oocytes were treated (150 s) with Aβ_1–42 (800 nm), alone (top), together with ACh (1 mm, middle) or with ACh (1 mm) alone (bottom), then washed with normal Ringer (240 s). Note the same percentage of Aβ_1–42 -induced inhibition in the T traces, independent of the presence of ACh during treatment period. All the traces were recorded from one oocyte, representative of three experiments.

Given the slow onset and poor reversibility of Aβ_1–42-induced inhibition of I_ACh, it is possible that the competition at the ACh binding site is obscured by Aβ_1–42 dissociating too slowly to be displaced by ACh during the brief applications eliciting I_ACh. In order to test this hypothesis, we compared the block induced by treating the oocytes (150 s) with Aβ_1–42 alone (800 nm) or with Aβ_1–42 plus ACh (1 mm), so that competition can take place during the onset of the current inhibition. I_ACh (ACh concentration, 100 µm) was measured after a 240 s wash-out, which allowed for full recovery of α7 nAChRs from ACh-induced desensitisation (Fig. 2B, bottom). In the four oocytes tested, I_ACh was reduced to 44 ± 10 % of control when Aβ_1–42 was applied alone and to 46 ± 12 % when Aβ_1–42 was applied in the presence of ACh (Fig. 2B). We also tested whether the application of ACh (1 mm) during Aβ_1–42 wash-out could speed up I_ACh recovery, by accelerating the displacement of the bound peptide. Neither 10 s nor 20 s applications of ACh accelerated the recovery Aβ_1–42-inhibited current (data not shown). All these data taken together strongly support the non-competitive interaction of Aβ_1–42 with α7 nAChRs, and suggest that the mechanism of inhibition may involve the slow transition of nAChRs into a long-lived closed or blocked state.

Our data are probably explained by the reported specific binding of Aβ_1–42 to human α7 nAChRs (Wang et al. 2000a,b, 2002). However, there remains the possibility that the action of Aβ_1–42 is mediated through intracellular effectors ultimately acting on α7 nAChRs. This would be much more unlikely should we be able to demonstrate that Aβ_1–42, like many other antagonists of WT α7 nAChRs, behaves as an agonist of the mutated receptor bearing a threonine-for-leucine exchange in the M2 channel domain. We therefore studied the outcome of the exposure to Aβ_1–42 of oocytes expressing the human L248T α7 nAChR.

Aβ_1–42 is an agonist of the L248T α7 nAChR

Voltage-clamp recordings showed that brief applications (2–10 s) of Aβ_1–42 evoked currents readily blocked by methyllycaconitine (MLA, 0.2 µm) (Fig. 3A). Current amplitude depended on the concentration of Aβ_1–42 (Fig. 3B and C), reaching about half the amplitude of the response elicited by ACh at the saturating concentration of 100 µm (-1.1 µA; see Fucile et al. 2002) with a peptide concentration of 400 nm (Fig. 3C). In all the 15 oocytes tested (5 donors), the currents were sustained during Aβ_1–42 application, with a negligible decay observed only at high peptide concentrations (1 µm, e.g. Fig. 3B), as expected for this non-desensitising nAChR. Multiple Aβ_1–42 applications evoked responses of fairly constant amplitude (data not shown). These findings contrast with the observations made on rat L250T α7 nAChRs (Dineley et al. 2002), where responses desensitise upon multiple or prolonged applications.

Figure 3. Activation of L248T α7 nAChRs by Aβ_1–42.

A, currents activated by Aβ_1–42 (400 nm, filled bars), in an oocyte (representative of 15 oocytes, 4 donors) expressing L248T α7 nAChRs. Note the complete block by 0.2 µm MLA (≈40 s preincubation, open bar). This particular response was slightly smaller than average. B, currents evoked by Aβ_1–42 at the indicated concentrations in two other oocytes. Note the sustained response during agonist application. C, histogram comparing the agonism of Aβ peptides (as indicated), normalised to the response evoked by 100 µm ACh in each oocyte. Each bar represents the mean ± s.e.m. of 6–8 oocytes (4 donors) expressing L248T α7 nAChRs.

The inactive Aβ_40–1 peptide (1 µm) failed to evoke responses in three out of six oocytes tested (2 donors), whereas in the other three oocytes it yielded a current whose amplitude was 12 % of the response elicited by Aβ_1–42 (1 µm) in the same oocytes. Aβ_42–1 was slightly more potent in mimicking the active peptide, eliciting currents with amplitudes which were 22 ± 15 % (6/2) of the responses elicited by Aβ_1–42 (Fig. 3C). However, these data confirm that current activation was largely due to a specific action of Aβ_1–42 on L248T α7 nAChRs, taking into account the high concentrations of peptides used in these experiments.

The action of amyloid peptides on the mutated nAChR was also investigated by performing outside-out patch-clamp recordings in oocytes expressing L248T α7 nAChRs, as determined by preliminary tests of ACh sensitivity.

Spontaneous openings of brief, MLA-sensitive channels at a frequency of 5–50 Hz were observed in all the 19 excised patches examined (17 oocytes from 9 donors), as detailed elsewhere (Fucile et al. 2002). It must be noted that these events differ from the well-characterised stretch-activated channels (Methfessel et al. 1986) both in conductance and kinetics. Application of Aβ_1–42 (1 µm) raised single-channel open probability (NP_op) by about 4-fold above the spontaneous background (9 patches), with MLA completely abolishing channel activity (Fig. 4A). In parallel experiments, ACh (0.1 µm) raised NP_op by about 8-fold (10 patches, data not shown).

Figure 4. Single-channel properties of L248T α7 nAChRs activated by Aβ_1–42.

A, spontaneous and Aβ_1–42-evoked single-channel activity, blocked by MLA (0.2 µm), in an outside-out patch from an oocyte expressing L248T α7 nAChRs. Inset, part of the trace on an expanded time scale, to show three classes of channel conductance in Aβ_1–42-evoked channel openings (γ_L= 39.7 pS, γ_M= 53.2 pS, γ_H= 67 pS). Spontaneous channels in the same patch had matching conductances. Inward currents downwards. B, open time distributions and sample traces for Aβ_1–42-activated channels, recorded from a different patch. Superimposed lines: best fitting exponential curves with time constants (weight): τ_o1= 0.29 ms (28 %), τ_o2= 1.02 ms (56 %), τ_o3= 4.11 ms (16 %), τ_op= 1.86 ms (n = 1689). All recordings were performed at −50 mV. Aβ_1–42 concentration, 1 µm.

Spontaneous and evoked unitary events showed three levels of current amplitude (e.g. Fig. 4A, inset), corresponding to the conductance values given in Table 1. Unitary current (i)-V relations were linear in the potential range tested (-90 to −50 mV, data not shown). More than one class of channel conductance was observed in 16 of the 19 patches examined. In each patch, the same number of conductance levels was observed for spontaneous and evoked channels. For instance, in the nine patches exposed to Aβ_1–42, the three conductance levels were simultaneously observed in five (6 out of 10 for ACh), for both spontaneous and evoked channel (Fig. 4A, inset). Since no transition among the conductance levels was observed, they are likely to represent three independent gating modes of L248T α7 nAChR-channels, rather than conductance substates of a single population. This agrees with data previously described for the chick L247T α7 nAChR (Revah et al. 1991; Palma et al. 1997).

Table 1.

Single-channel properties of L248T α7 nAChR expressed in oocytes

Agonist	Conductance			Open time
	γL (pS)	γM (pS)	γH (pS)	γo1 (ms)	γo2(ms)	γo3(ms)
Spontaneous	41.5 ± 0.7	51.7 ± 1.7	66.2 ± 1.2	0.52 ± 0.09*	2.67 ± 0.66**	—
(19 patches)	(38%)	(37%)	(25%)	(67%)	(33%)
Aβ1–42	43.9 ± 1.6	53.6 ± 0.9	67.4 ± 1.2	0.34 ± 0.03*	1.34 ± 0.14**	6.3 ± 0.8***
(9 patches)	(42%)	(29%)	(27%)	(39%)	(48%)	(13%)
Ach	39.1 ± 1.1	52.8 ± 1.2	64.3 ± 1.8	0.28 ± 0.01*	1.15 ± 0.05**	10.2 ± 2.2***
(10 patches)	(52%)	(34%)	(14%)	(43%)	(42%)	(15%)

Results are given as means ± S.E.M. (weight, %) of the indicated number of patches. Test potential, −50 mV. *, **, *** Statistically not different (one-way ANOVA, P > 0.2).

The mean open duration (τ_op) of spontaneous channels was 1.4 ± 0.2 ms (4068 openings from 19 patches). Upon application of Aβ_1–42, τ_op significantly increased to 2.1 ± 0.9 ms (11920 openings from 9 patches; one-way ANOVA, P = 0.02), with a distribution made up of three exponential components (Fig. 4B) with time constants τ_o1, τ_o2 and τ_o3 given in Table 1 (see also Fig. 4B). ACh-induced openings showed comparable τ_op values (2.7 ± 0.7 ms; 9032 openings from 10 patches; P = 0.43) and channel open times distribution (Table 1). Neither the opening frequency nor the τ_op of spontaneous channel were significantly altered when patches were exposed to Aβ_40–1 (1 µm, 4/1) (data not shown).

Muscle nAChRs are blocked by Aβ_1–42

In other experiments, we examined whether Aβ_1–42 functionally modulates the α-BuTx-sensitive mouse muscle γ- or ε-nAChRs, expressed in transiently transfected BOSC 23 cells. We chose this cell expression system as it yields γ- and ε-nAChR-channels with functional properties matching those of native muscle fibres (Grassi, 1999), while this is not the case for Xenopus oocytes (Kullberg et al. 1990). By itself, Aβ_1–42 (up to 1 µm) did not affect baseline current, nor did co-application of Aβ_1–42 together with ACh alter the current response (data not shown). However, when cells were pre-treated with Aβ_1–42 (100 nm) for 60–120 s, I_ACh was partially blocked (Fig. 5A). The effect of Aβ_1–42 developed within the first 120 s of application (Fig. 5B) and was not further increased by prolonged exposure to the peptide (Fig. 5C). The reduction of the peak current amplitude (to about 60 % of control) was accompanied by the acceleration of I_ACh decay and was similar for γ- and ε-nAChRs (Table 2), indicating that the two muscle receptors are comparably susceptible to block by Aβ_1–42.

Figure 5. Aβ_1–42 blocks muscle nAChRs in transiently transfected BOSC 23 cells.

A, typical inward currents evoked by ACh (1 µm) in BOSC 23 cell expressing γ-AChR (left) or ε-AChR (right), before (C) or after (Aβ) 120 s of application of Aβ_1–42 (100 nm). Note the accelerated decay of I_ACh during Aβ_1–42 application. B, time course of I_ACh block by Aβ_1–42 or Aβ_42–1 (both 100 nm) in two different cells expressing γ-AChR. Bars, Aβ applications. Note the lack of recovery 15 min after Aβ_1–42 wash-out, as compared to the prompt recovery upon Aβ_42–1 removal. ACh concentration, 1 µm. I_ACh normalised to control current amplitude. C, plot of I_ACh amplitude (normalised to the control in each cell) vs. the duration of Aβ_1–42 (100 nm) application. The continuous line represents the linear regression of the data, with a slope of −0.005 % s⁻¹ (R = 0.05), indicating that the block of I_ACh is independent of the duration of Aβ_1–42 application. All the data obtained for γ-AChR, irrespective of ACh concentration (0.2–20 µm), are included in this plot. D, activation of γ-nAChR-channels by ACh (1 µm) in an outside-out patch, before (trace C), during (Aβ) and after (W) the application of Aβ_1–42. τ_cl, 26 ms (C), 184 ms (Aβ), 60 ms (W). Aβ_1–42 (100 nm) application began 60 s before recording trace Aβ, terminated 5 min before recording trace W. Traces were filtered at 200 Hz for display purposes. Insets, expanded traces beginning 750 ms after ACh application (filter, 1 kHz). Single-channel conductance (35.7 pS) and open duration (3.5 ms) were not affected by Aβ_1–42. E, plot of the average NP_op, normalised to the control value of each patch, in 5 outside-out patches, measured over 1-s intervals during the first 10 s of ACh (1 µm) application. Data were sampled before (□), in the continuous presence of Aβ_1–42 (30–120 s preincubation, (•), or 30–120 s after Aβ_1–42 wash-out (○).

Table 2.

Effects of Aα_1–42 (100 nm) on whole-cell I_ACh in transiently transfected BOSC 23 cells

	Current amplitude			Half-decay
Cells	IAch (nA)	Reduction (%)	Recovery (%)	T0.5 (s)	Reduction (%)	Recovery (%)
γ-AChR	−5.1 ± 1.5 (12)	59 ± 5* (12)	66 ± 6 (8)	0.94 ± 0.23 (12)	60 ± 6** (9)	92 ± 7 (8)
ɛ-AChR	−0.9 ± 0.3 (3)	69 ± 3* (3)	81(2)	1.27 ± 0.37(3)	45 ± 1** (3)	55(2)

Results are given as means ± S.E.M. (number of cells). Membrane potential, −70 mV; ACh concentration, 1 μM. Aβ_1–42 was applied for 80–300 s. Recovery was calculated >300 s after Aβ_1–42 washout. *, ** Statistically not different (P > 0.15).

In most cells, the amplitude of I_ACh did not recover to control even 10 min after Aβ_1–42 withdrawal (e.g. Fig. 5B), whereas T_0.5 showed a more complete recovery (Table 2). For both γ-and ε-nAChRs, the reduction of I_ACh amplitude and T_0.5 was not statistically different when changing ACh concentration in the range 0.2 to 20 µm (one-way ANOVA, P > 0.1). Increasing the concentration of Aβ_1–42 from 100 nm to 1 µm did not enhance the block of I_ACh (3 cells tested, data not shown), suggesting that maximal inhibition of I_ACh is already induced by the peptide at the concentration of 100 nm. As for α7 nAChRs in oocytes, the block of I_ACh was voltage independent in the range −30 to −90 mV (data not shown).

The effect of Aβ_42–1 on I_ACh was also similar to the findings in oocytes. The peptide (100 nm) reduced the amplitude of I_ACh to 75 % (n = 4, γ-nAChR) and accelerated current decay. This reduced block was reversible within min of Aβ_42–1 wash-out (e.g. Fig. 5B), at variance with the effect of Aβ_1–42.

The effects of Aβ_1–42 (100 nm) on the single-channel properties of muscle nAChRs were investigated in five outside-out patches from cells expressing γ-nAChRs. Neither the conductance (39.7 ± 1.4 pS) nor the τ_op (3.2 ± 0.5 ms) of the events evoked by ACh (1 µm) were affected by applying Aβ_1–42 (100 nm) for 60–120 s. After this pre-treatment, application of ACh in the continuous presence of the peptide elicited single-channel openings with a conductance of 40.5 ± 1.5 pS and τ_op of 3.1 ± 0.5 ms (e.g. Fig. 5D, inset). During long lasting ACh applications, channel opening frequency markedly decreased, while channel conductance and τ_op remained stable. Channel closed time (τ_cl) increased from 2- to 10-fold within 30–60 s of Aβ_1–42 exposure (e.g. Fig. 5D). Given the non-stationary behaviour of channel activity in these patches, NP_op was measured over 1-s intervals during the first 10 s of ACh application. In good agreement with whole-cell data, after 30–120 s of Aβ_1–42 application, NP_op was reduced to about 45 % control and the rate of NP_op decrease was accelerated by about 50 % (Fig. 5E), indicating that Aβ_1–42 promotes the block of ACh-evoked channels. The effects of Aβ_1–42 were only partially reversible by 30 s wash-out (see Fig. 5E). Longer washes (>120 s) could be performed in only two patches, and yielded almost full recovery.

To examine whether Aβ_1–42 modulates muscle nAChRs via indirect pathways, the peptide was applied to the extra-patch membrane while recording γ-nAChR single-channel activity under cell-attached conditions. In the four patches examined, the unitary channel conductance remained unchanged during Aβ_1–42 applications lasting 2–6 min (33 pS, data not shown). Channel opening frequency reversibly decreased to 30 % of control in one patch; channel mean open time decreased to 85 % in another patch. Taken together with the results of outside-out recordings, these data are consistent with the hypothesis that Aβ_1–42 exerts its effects by directly binding to the nAChR molecule.

DISCUSSION

A high-affinity association of the amyloid peptide Aβ_1–42 with α7 nAChRs has recently been observed in amyloid plaques and in the neurons of AD patients (Wang et al. 2000a,b, 2002; Nagele et al. 2002). However, the modulation of the α7 nAChR function has only been described in chick and rodent preparations (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002; Tozaki et al. 2002). In this paper, we give evidence that Aβ_1–42 is able to functionally block the human neuronal α7 nAChR, in a poorly reversible manner, with a potency comparable with that previously described for native and reconstituted rat preparations (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002). Moreover, mouse muscle γ- and ε-nAChRs, other types of α-BuTx-sensitive nAChR, were blocked by Aβ_1–42 in a manner rather similar to the block of α7 nAChRs, the main difference being that Aβ_1–42 accelerates the rate of current decay for muscle but not neuronal nAChRs, as already described in rat hippocampal cultures (Liu et al. 2001). Thus, blockade of α-BuTx-sensitive nAChRs by Aβ_1–42 appears to be a rather general property, although other studies have found it to be fully reversible (Liu et al. 2001; Pettit et al. 2001).

It has been reported that picomolar concentrations of Aβ_1–42 elicit current responses from oocytes expressing rat WT α7 nAChRs (Dineley et al. 2002), even though no activation was seen in rat hippocampal neurones exposed to similar concentrations of Aβ_1–42 (Liu et al. 2001). In our hands, human WT α7 nAChRs were not activated by Aβ_1–42 over a wide range of concentrations, although, in parallel experiments, the L248T mutant α7 receptor did respond to the peptide. Thus, Aβ_1–42-induced activation of WT α7 nAChRs appears to be strongly dependent on the receptor type, the cell system, or the experimental procedure.

It might be argued that our data lack specificity, as the inverse peptide, Aβ_42–1, does inhibit I_ACh. To the best of our knowledge, this is the first report of the biological effects of Aβ_42–1. In particular, in the papers investigating the interaction between Aβ_1–42 and α7 nAChRs, the only peptide used as a control was Aβ_40–1 (Wang et al. 2000a,b; Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002). In our hands, Aβ_40–1 is ineffective in inhibiting WT α7 nAChRs, and only marginally capable of activating L248T mutant α7 nAChRs when used at very high concentrations (1 µm). Comparably small effects of Aβ_40–1 have been observed when measuring α-BuTx binding to α7 nAChRs (Wang et al. 2000a) or current block (Liu et al. 2001), and were considered negligible. Thus, the effects of Aβ_1–42 on muscle and neuronal nAChRs reported here can be claimed to be as specific as those previously reported. Two questions remain open: what causes the reverse peptide Aβ_42–1 to be active, and why is Aβ_42–1 effective while Aβ_40–1 is not. The hypothesis that the effects are due to peptide contaminants is rather unlikely, since we used peptides of different origin and in different solvents. It must be noted that similarities in the neurotoxic action of Aβ_1–40 and Aβ_40–1 have been reported (Giordano et al. 1994), indicating that reverse peptides are not entirely biologically inactive. It has been shown that the smaller fragment Aβ_12–28 is able to mimic the action of the Aβ_1–42 peptide on muscle and WT α7 nAChRs (Wang et al. 2000b; Pettit et al. 2001; authors’ unpublished observations), indicating that a binding epitope for nAChRs resides in this peptide region, which comprises an α-helix and a ‘kink’ region (Coles et al. 1998). That the binding epitope is conserved in the reverse peptide is quite unlikely, but it might be possible that the reverse peptide contains another binding site for nAChRs, causing a weaker block. In line with this hypothesis, the block by Aβ_42–1 is fully reversible, while the effect of Aβ_1–42 is not, suggesting differential interactions of the two peptides with nAChRs. It may be speculated that the two very hydrophobic terminal amino acids (an isoleucine and an alanine) present in Aβ_42–1, but not in Aβ_40–1, favour the interaction of the longer peptide with the cell membrane, thus enhancing the probability of an interaction with nAChRs. Understanding the interaction between Aβ_42–1 and nAChRs is, however, beyond the scope of this paper, especially because the effect of Aβ_1–42, being stronger than that of Aβ_42–1, is likely to be biologically relevant.

In agreement with other studies (Liu et al. 2001; Pettit et al. 2001; Dineley et al. 2002), we report that the Aβ_1–42-induced block of I_ACh requires a few minutes of preincubation, both for α7-expressing oocytes and for γ- and ε-nAChR-expressing BOSC 23 cells. This might suggest the involvement of pathways mediated by second messengers. Several pieces of evidence argue against this hypothesis. First, in BOSC 23 cells, the reduced amplitude and accelerated decay of whole-cell I_ACh upon application of Aβ_1–42 matches the reduced NP_op and faster desensitisation of γ-nAChR-channels observed in cell-free outside-out patches, where the cytosolic components are lost. Second, cell-attached recordings in intact cells, with a fully preserved cytoplasmic environment, failed to reveal any indirect effect of Aβ_1–42 on γ-nAChR-channel activity. Third, Aβ_1–42 behaves as an agonist of the L248T mutant α7 nAChR, as do many other α7 nAChR antagonists whose direct actions on nAChRs are very firmly established. Fourth, this agonist action is also seen in excised patches, again ruling out the requirement for cytoplasmic components. It is noteworthy, however, that the agonist action of Aβ_1–42 on L248T nAChRs is rapid, both on intact oocytes and in outside-out patches (our data and Dineley et al. 2002). It is possible that a simple gating process activates the mutant α7 nAChR, whereas blockade of WT α7 and muscle receptors requires the slow stabilisation of an inactive state. The poor reversibility of the inhibition is also compatible with the hypothesis of Aβ_1–42 driving the nAChRs into a long-lived closed (or blocked) conformation.

The significance of this interaction between Aβ_1–42 and α7 nAChRs for the aetiology or the pathogenesis of AD is unclear. Recent work shows a preferential accumulation of Aβ_1–42 in neurons expressing α7 nAChRs (Wang et al. 2000a,b, 2002) and evidence has been provided that intracellular accumulation of Aβ_1–42 may be facilitated by α7 nAChRs (Nagele et al. 2002), thus implying a relevant physio-pathological role for the interaction. This raises the possibility that the binding of Aβ_1–42 to muscle ε-nAChRs might be related to the initiation of plaque deposition in IBM and/or in the muscles of AD patients, which show an increased content of Aβ_1–42 (Kuo et al. 2000b). The functional modulation of muscle nAChRs by Aβ_1–42 strengthens the similarity between AD and IBM, further suggesting that the two diseases share at least some pathogenic mechanisms.

The question remains whether the observed Aβ_1–42-induced nAChR functional changes affect synaptic transmission. We and others (Pettit et al. 2001; Dineley et al. 2002) have shown that Aβ_1–42 affects I_ACh with an IC₅₀ around 100 nm (that is, about 450 ng ml⁻¹), although an IC₅₀ of about 7.5 nm has been described for rat hippocampal neurones (Liu et al. 2001). The concentrations of Aβ_1–42 in the plasma and cerebrospinal fluid of control and AD humans are uncertain, reported values ranging between 0.04 ng ml⁻¹ (i.e. 0.01 nm, Mehta et al. 2000) and 20 ng ml⁻¹ (i.e. 5 nm, Kuo et al. 2000a). These values are lower than the observed IC₅₀, but functional modulation of α7 nAChR in vivo might ensue because the neurones are tonically exposed to Aβ_1–42, that is, for times much longer than have been tested in experimental studies.

A possible link between Aβ_1–42 binding to α7 nAChRs and cognitive impairments in AD was recently suggested by a paper (Dineley et al. 2001) showing that Aβ_1–42 is able to promote MAP kinase activation by inducing Ca²⁺ influx through α7 nAChRs, thereby interfering with long term potentiation processes. That study, however, was conducted in mice heterozygous for the mutant L250T α7 nAChR and we show here that Aβ_1–42 does not activate the human WT α7 nAChR. The fact that human WT α7 nAChRs is not activatable by Aβ_1–42 rules out the likelihood that memory loss in AD is caused by the suggested mechanism. Nevertheless, the activation of L248T α7 nAChRs by Aβ_1–42 raises the possibility of a correlation between genetic variations of α7 nAChRs and AD. The hypothesis that an allelic variant, a 2 bp deletion, of the partially duplicated human gene encoding the α7 subunit induces susceptibility to AD has recently been tested and dismissed (Liou et al. 2001). To our knowledge, other mutations have not been investigated.

In conclusion, we give evidence that Aβ_1–42 alters the gating of α-BuTx-sensitive nAChRs, blocking human WT α7 nAChRs and mouse muscle nAChRs, while activating the human mutant L248T α7 nAChR. The functional impairment of nAChRs might be responsible, at least in part, for the cognitive deficits known to appear well before plaque formation both in mouse models (Moechars et al. 1999) and in AD patients (for review, see Neve & Robakis, 1998; Smith, 2002). The loss of synaptic input to cortical areas might underlie AD progression from the medial temporal lobe to the whole cerebral cortex (Smith, 2002). Further research should elucidate this point.