Address correspondence and reprint requests to Dr. H.-Y. Wang at R. W. Johnson Pharmaceutical Research Institute, Rm. 345, Research Bldg., McKean and Welsh Roads, Spring House, PA 19477-0776 U.S.A. E-mail: firstname.lastname@example.org
Abstract: We have recently reported evidence that a very high affinity interaction between the β-amyloid peptide Aβ1-42 and the α7 nicotinic acetylcholine receptor (α7nAChR) may be a precipitating event in the formation of amyloid plaques in Alzheimer's disease. In the present study, the kinetics for the binding of Aβ1-42 to α7nAChR and α4β2nAChR were determined using the subtype-selective nicotinic receptor ligands [3H]methyllycaconitine and [3H]cytisine. Synaptic membranes prepared from rat and guinea pig cerebral cortex and hippocampus were used as the source of receptors. Aβ1-42 bound to the α7nAChR with exceptionally high affinity, as indicated by Ki values of 4.1 and 5.0 pM for rat and guinea pig receptors, respectively. When compared with the α7nAChR, the affinity of Aβ1-42 for the α4β2nAChR was ∼5,000-fold lower, as indicated by corresponding Ki values of 30 and 23nM. The results of this study support the concept that an exceptionally high affinity interaction between Aβ1-42 and α7nAChR could serve as a precipitating factor in the formation of amyloid plaques and thereby contribute to the selective degeneration of cholinergic neurons that originate in the basal forebrain and project to the cortex and hippocampus.
Although it has been known for many years that a selective degeneration of cholinergic neurons that originate in the basal forebrain and project to the cortex and hippocampus is a characteristic feature of Alzheimer's disease (AD) (Bartus et al., 1982), the pathological basis for the selective vulnerability of these neurons has not been established. Accumulating evidence suggests the β-amyloid peptide Aβ1-42 is a causal factor in the neurodegenerative process (Selkoe, 1998). Aβ peptides, particularly Aβ1-42, have been reported to be neurotoxic (Yankner et al., 1989; Scheuner et al., 1996; Mattson, 1997), and recently Aβ1-42 at picomolar concentrations was reported to adversely affect cholinergic synaptic activity in rat cortical and hippocampal tissues but not striatal tissues (Kar et al., 1998). Amyloid peptides have been reported to bind with high affinity to several proteins. These include the scavenger receptors expressed in microglia and macrophages (El Khoury et al., 1996; Paresce et al., 1996), the receptor for advanced glycation end products (RAGE) (Yan et al., 1996, 1997a), α2-macroglobular protein (Du et al., 1997; Narita et al., 1997), apolipoprotein E (ApoE) (Pillot et al., 1999), and a novel intracellular protein discovered by Yan et al. (1997b) that they termed endoplasmic reticulum amyloid binding protein (ERAB). Recently, we have reported evidence that an interaction of Aβ1-42 with nicotinic acetylcholine receptors (nAChRs) that contain the α7 subunit (α7nAChRs) may be a key factor in the formation of AD plaques and degeneration of cholinergic neurons (Wang et al., 2000). The α7nAChR is a pentameric ion channel protein complex selectively permeable to Ca2+ and Na+ and is highly expressed in basal forebrain cholinergic neurons that project to the hippocampus and cortex (Breese et al., 1997). Within these cholinergic neurons, α7nAChR modulates calcium homeostasis (Seguela et al., 1993; Quik et al., 1997) and the release of acetylcholine from synaptic terminals (McGehee et al., 1995; Orr-Urtreger et al., 1997).
Our studies demonstrated that Aβ1-42 binds with femtomolar affinity to α7nAChR expressed in human SK-N-MC neuroblastoma cells and that Aβ1-42 is selectively toxic to SK-N-MC cells expressing high levels of α7nAChR. We further demonstrated α7nAChR is present in AD plaques and is closely associated with Aβ1-42. In the present study, we have used the subtype-selective nicotinic receptor ligands [3H]methyllycaconitine ([3H]MLA) and [3H]cytisine to examine the selectivity of Aβ1-42 for α7nAChR compared with α4β2nAChR in synaptic membranes prepared from rat and guinea pig cerebral cortex and hippocampus. To further evaluate the selectivity of Aβ1-42 for α7nAChR, the affinity of Aβ1-42 for muscarinic, serotonin 5-HT3, and NMDA receptors and receptors for neuropeptide Y, galanin, melanocortin, substance P, and neurokinin A (NKA) was determined.
MATERIALS AND METHODS
Amyloid peptides (Aβ1-42 and Aβ1-40), cytisine, epibatidine, peptide YY (PYY), and NDP-melanocortin-stimulating hormone (NDP-MSH) were purchased from Palomar Research (Encinitas, CA, U.S.A.). MLA, α-bungarotoxin (α-BTX), [SAR9,Met(O2)11]-substance P, and [β-Ala8]-NKA(4-10) were purchased from Research Biochemicals International (Natick, MA, U.S.A.). Nicotine and atropine sulfate were purchased from Sigma Chemicals (St. Louis, MO, U.S.A.). 125I-α-BTX, [3H]nicotine, 125I-substance, P, 125I-NKA, 125I-human galanin, 125I-PYY, 125I-NDP-MSH, and (I)-[3H]quinuclidinyl benzilate [3H]QNB were purchased from Amersham Pharmacia Biotech (Arlington Heights, IL, U.S.A.). [3H]Cytisine, 125I-MK-801, and [N-methyl-3H]GR65630 were purchased from NEN (Boston MA), and [3H]MLA was purchased from Tocris (Ballwin, MO, U.S.A.). A protease inhibitor cocktail was purchased from Boehringer Mannheim (Indianapolis, IN, U.S.A.). Bovine serum albumin and HEPES (1 M solution) were purchased from Sigma.
Male Long-Evans rats were purchased from Charles River Laboratories (Portage, MI, U.S.A.). Male Hartley guinea pigs were purchased from Charles River Laboratories or Covance (Denver, PA, U.S.A.). The rats and guinea pigs were housed in groups of three to five at an ambient temperature of 21-23°C with an automated 12:12-h light/dark cycle and access to water and a commercial rodent food ad libitum. Rats weighing 280-320 g or guinea pigs weighing 400-600 g were asphyxiated with carbon dioxide gas and were cervically dislocated. The whole brain was immediately excised, placed on an ice-cold Petri dish, and dissected into specified regions. The tissue was either frozen at -80°C or used fresh. All procedures involving animals were approved by our institutional animal care and use committee and conform to USDA guidelines.
Preparation of rat and guinea pig synaptic membranes
Fresh or frozen tissue was homogenized in ∼50 volumes of ice-cold 10 mM Na+-HEPES (pH 7.4) using a motor-driven Teflon pestle and a polished glass tube (0.25-mm clearance), and the homogenate was centrifuged at 42,000 g for 10 min. The supernatant was discarded, and the sedimented (pellet) material was washed by resuspending it in ∼40 volumes of the 10 mM Na+-HEPES solution and centrifuging at 42,000 g for 10 min. This washing step was performed at least three times, which was necessary to reveal picomolar affinity binding of Aβ1-42 to the α7nAChR. Subsequently, the sediment was resuspended into 25 volumes (1 g of original tissue into 25 ml) of a buffered incubation medium comprising Na+-HEPES buffer (10 mM, pH 7.4), 5 mM MgCl2, 0.01% bovine serum albumin, a cocktail of protease inhibitors (antipain, bestatin, E-64, leupeptin, pepstatin, aprotinin, chymostatin, and phosphoramidon), and 100 mM NaCl. This membrane suspension was used in the ligand-receptor binding experiments.
Radioligand binding to nicotinic receptors
Hippocampal and cortical tissues were chosen because preliminary experiments indicated these areas expressed high levels of α7nAChR (Court and Perry, 1995; Breese et al., 1997) or α4β2nAChR (Court and Perry, 1995) compared with whole brain, thereby making it possible to maximize specific binding (SB). The assays were designed to maximize the ratio of SB to nonspecific binding (NSB) yet provide adequate total SB to ensure reliable data. Concentrations of radioligand below the KD values were used to minimize possible differences between the IC50 values obtained experimentally and theoretical Ki values (Cheng and Prusoff, 1973). The experiments were performed at 21-23°C to avoid possible effects on membrane fluidity that occur at lower temperatures, and the incubation time was sufficient to ensure that ligand binding reached equilibrium before the binding reaction was terminated.
Each sample tube contained 0.25 ml of the tissue membrane suspension, 0.4 ml of the HEPES-buffered incubation medium, 0.25 ml of ligand in the HEPES-buffered incubation medium, and 0.1 ml of test compound dissolved in water (Table 1). Ligand and substrate binding saturation curves were generated using 5-12 concentrations, and for each experiment, a given set of samples was tested in quadruplicate. The binding reaction was terminated by separating the membrane material from the incubation medium using vacuum filtration [Wallac Filtermat B (Turku, Finland), Brandel MLR-24 cell harvester (Gaithersburg, MD, U.S.A.)]. The samples were washed twice with 2 ml of ice-cold HEPES-buffered solution (10 mM, pH 7.4). Radioactivity was quantified by liquid scintillation spectrometry using a Wallac Betaplate counter at a counting efficiency of 44%. The total binding ranged between 800 and 1,200 dpm and NSB between 100 and 200 dpm. As the samples were counted for 5 min two or three times, the minimum total disintegrations counted in the absence of inhibition was ∼8,000. NSB was determined using a saturating concentration (10 μM) of non-labeled ligand or another compound that bound to the receptor (Table 1).
For muscarinic, 5-HT3, and NMDA receptor experiments, membrane suspensions were prepared from rat or guinea pig brain as described above. For other receptors, specified cell types (Table 1) were grown in culture by standard procedures. Membranes were prepared by homogenizing 1 g of harvested cells in 30 ml of the Na-HEPES solution (pH 7.4) as described previously for the brain tissue. The membranes were washed five times to remove soluble factors that could bind to Aβ1-42 and suspended (∼20 μg of protein) in a medium comprising 50 mM HEPES (neutralized with NaOH to a pH of 7.4), 0.01% bovine serum albumin, 2 mM MgCl2, 2 mM CaCl2, and the cocktail of protease inhibitors described previously. For the ligand binding reaction, each sample tube contained 100 μl of the cell membrane suspension, 45 μl of ligand in the HEPES-buffered incubation medium, and Aβ1-42 or a compound used to determine NSB (Table 1). The final volume was 150 μl, and samples were incubated in triplicate. The binding reaction was terminated by separating the membrane material from the incubation medium using a 96-well Brandel vacuum filtration unit. The samples collected on the filter plate were washed twice with 0.2 ml of ice-cold HEPES-buffered solution (10 mM, pH 7.4) and dried. Then 25 μl of Microscint (Packard, Meriden, CT, U.S.A.) was added to each well. Bound radioactivity was measured using a TopCount (Packard). Each sample was counted for 5 min two or three times. NSB was determined using a saturating concentration (10 μM) of nonlabeled ligand or another compound that bound to the receptor (Table 1). NSB accounted for 10-30% of total 125I bound.
Protein was determined by the method of Bradford as modified by Bio-Rad Laboratories (Technical Bulletin 1051; Bio-Rad Laboratories, Richmond, CA, U.S.A.).
The KD and Bmax values for the binding of [3H]MLA and [3H]cytisine to α7nAChR and α4β2nAChR, respectively, were calculated using a computer-assisted procedure described previously (Shank et al., 1990). In this procedure, NSB is calculated as a nonsaturable variable in the equation rather than being defined pharmacologically (NSB determinant; Table 1). IC50 values were calculated from a curve-fit analysis (PRISM; GraphPad Software, NJ, U.S.A.) of the SB obtained by subtracting the pharmacologically defined NSB. Ki values were calculated using the Cheng-Prusoff equation (Cheng and Prusoff, 1973). Although this equation is strictly applicable only to competitive inhibition, which may not be true for Aβ1-42, the actual difference between IC50 and Ki values was small: 20 and 12% for data obtained with [3H]MLA and [3H]cytisine, respectively.
KD and Bmax values for [3H]MLA and [3H]cytisine binding
A curve-fit analysis of the binding of [3H]MLA or [3H]cytisine to the rat and guinea pig membranes was performed using equations appropriate for either one or two saturable binding sites. For both ligands, the data fitted significantly better to the equation with two sites (p < 0.05). The calculated KD for the high-affinity binding of [3H]MLA to rat cortical/hippocampal and guinea pig hippocampal membranes was 2.8 and 3.0 nM, respectively, and the corresponding Bmax values were 144 and 187 fmol/mg of protein. For comparison, Davies et al. (1999), who performed their experiments at 4°C, reported a KD of 1.9 nM. Our Bmax for rat cortex/hipocampus is more than twofold higher than the Bmax reported by Davies et al. for whole rat brain (61 fmol/mg of protein). The calculated KD for the high-affinity binding of [3H]cytisine to rat cortical/hippocampal and guinea pig cortical membranes was 3.3 and 5.2 nM, respectively, and the corresponding values for the Bmax were 370 and 380 fmol/mg of protein.
The low-affinity binding sites for the binding of [3H]MLA and [3H]cytisine to rat and guinea pig membranes accounted for <20% of the SB. The calculated KD values for both were ∼700 nM, and the Bmax values were ∼6 pmol/mg of protein. The biochemical significance of these low-affinity sites is not known, but it may reflect a weak affinity for some other neurotransmitter receptors in neuronal membranes.
Affinity of Aβ1-42 for α7- and α4β2nAChRs: inhibition of [3H]MLA and [3H]cytisine binding to rat and guinea pig brain membranes
As noted in Materials and Methods, in preliminary experiments, we found that Aβ1-42 inhibited the binding of [3H]MLA to rodent membranes at picomolar concentrations only if the membrane preparations were washed at least three times. Notably, the number of times the membranes were washed did not significantly affect the kinetics of [3H]MLA binding. This suggests the possible existence of a soluble endogenous factor that binds to Aβ1-42 with high affinity. We have not identified this presumed binding factor, but it may be the ERAB protein characterized by Yan et al. (1997b).
A curve-fit analysis of the inhibition of [3H]MLA binding by Aβ1-42 indicated the data fitted significantly better to an equation with two saturable sites rather than one (Fig. 1). The calculated Ki for the site with highest affinity, which accounted for 60-70% of the inhibition of ligand binding, was 4.1 and 5.0 pM for rat and guinea pig, respectively (Table 2). The affinity of Aβ1-42 for the second site was ∼100-fold lower in both species (Table 2).
Table 2. Comparison of Kivalues for binding of Aβ1-42, α-BTX, MLA, cytisine, and epibatidine to α7nAChR ([3H]MLA) and α4β2nAChR ([3H]cytisine) in rat and guinea pig brain synaptic membranes
Ki (95% CL)a
Data were obtained from two to four experiments, each performed in triplicate or quadruplicate using 5-12 concentrations of the specified compound. ND, not determined.
aA competition curve-fit analysis was performed using the GraphPad Prism program. All data were analyzed assuming one or two saturable sites. The numbers in parentheses are the calculated 95% confidence limits (CL).
bThe curve-fit analysis suggested the existence of a second site with aKi of 440 pM (130-1,500).
cThe curve-fit analysis suggested the existence of a second site with aKi of 340 pM (120-960).
dThe curve-fit analysis suggested the existence of a second site with aKi of 360 pM (90-1,400).
eThe curve-fit analysis suggested the existence of a second site with aKi of 140 pM (16-1,060).
A similar analysis for the inhibition of [3H]cytisine binding by Aβ1-42 indicated that these sets of data also fitted significantly better to an equation with two saturable sites for both rat and guinea pig membranes (Fig. 2). The Ki value for the more prominent site, which accounted for 70-80% of ligand inhibition, was 30 and 23 nM for rat and guinea pig, respectively (Table 2). The affinity of Aβ1-42 for the second site was ∼100-fold higher, with a calculated Ki of 360 and 140 pM for rat and guinea pig, respectively (Table 2).
Affinity of α-BTX and epibatidine for α7- and α4β2nAChRs
A curve-fit analysis for the inhibition of [3H]MLA binding by α-BTX indicated the saturable inhibition was dominated by one site with a Ki of 18 and 22 nM in rat and guinea pig membranes, respectively (Fig. 1; Table 2). Epibatidine was 250-fold more potent as an inhibitor of [3H]cytisine binding than [3H]MLA binding (Table 2).
Affinity of Aβ1-42 for non-nicotinic receptors
Aβ1-42 was tested for binding affinity to muscarinic, NMDA, and 5-HT3 receptors and was inactive at 1 μM. As QNB binds nonselectively to muscarinic receptors, the results are relevant only to the dominant receptors in rat cortical membranes (M1 and M2). Aβ1-42 was tested at concentrations ranging from 1 to 10 μM for an ability to bind to eight neuropeptide receptors (Table 1). Aβ1-42 did not inhibit ligand binding to any of these receptors.
In recent years, evidence has accrued implicating nAChRs in the etiology of AD (Court et al., 1996; Kiraha et al., 1998; Hellström-Lindahl et al., 1999). We have recently reported immunohistochemical evidence that α7nAChR is present in AD neuritic plaques and is colocalized with Aβ1-42 in individual cortical neurons (Wang et al., 2000). Furthermore, α7nAChR and Aβ1-42 can be co-immunoprecipitated from AD brain tissue using antibodies specific to each protein, suggesting that α7nAChR and Aβ1-42 are tightly associated. Our studies further demonstrated that Aβ1-42 bound with femtomolar affinity (IC50∼0.01 pM) to α7nAChR in the membrane of human neuroblastoma cells overexpressing these receptors and that Aβ1-42 is toxic to these cells but not to normal human neuroblastoma (SK-N-MC) cells.
The results of the present study confirm and generalize our previous observation that human Aβ1-42 binds selectively and with exceptionally high affinity to the α7nAChR. This study also demonstrates that Aβ1-42 can bind to the α4β2nAChR, however, only at concentrations 100-5,000 times higher. Our observation that Aβ1-42 at micromolar concentrations did not bind to 5-HT3, NMDA, or muscarinic receptors or any of eight neuropeptide receptors indicates that a high-affinity interaction of Aβ1-42 with cell surface receptors is not a common characteristic of this peptide. As noted previously, Aβ1-42 and other amyloid peptides are known to possess a high affinity for several proteins. Most of these proteins are thought to serve as mediators for sequestering and eliminating amyloid peptides, thereby preventing their accumulation to toxic levels. These proteins include the scavenger receptors RAGE, α2-macroglobular protein, and ApoE. Yan et al. (1997b) reported evidence suggesting that the interaction between Aβ peptides and ERAB (the binding affinity of which was in the nanomolar range) might contribute to the neurotoxicity of the peptides. The exceptionally high affinity of the interaction between Aβ1-42 and α7nAChR, coupled with our observation that Aβ1-42 is selectively toxic to cells expressing high levels of α7nAChR, suggest that this interaction is unlikely to have physiological significance; however, it may reflect a pathological event associated with the development of AD. Specifically, we postulate that if any of several mechanisms that normally function to maintain Aβ peptides at very low levels become impaired, Aβ1-42 can accumulate within the synaptic cleft of cholinergic synapses in sufficient amounts to promote the formation of Aβ1-42/α7nAChR complexes, which can serve as a seed upon which Aβ1-42 aggregates. Such aggregates could impair the function of the α7nAChR and adversely affect cholinergic neurotransmission, and ultimately disrupt the integrity of the membrane and imperil the viability of the neuron (Auld et al., 1998). Over an extended period of time, many small aggregates might coalesce to form the plaques characteristic of AD (Gouras et al., 2000). A potential argument against our hypothesis is the lack of correlation between the concentration of Aβ peptides in CSF and the occurrence and severity of AD (Neve and Robakis, 1998); however, only a few Aβ peptides (e.g., Aβ1-42 and possibly Aβ1-43) may form insoluble aggregates with α7nAChR. Furthermore, concentrations of Aβ1-42 in cerebrospinal fluid may not reflect the concentration within the cleft of cholinergic synapses expressing α7nAChR (Younkin, 1998). The interaction of Aβ1-42 with α7nAChR may account for the various effects of Aβ peptides on cholinergic neurons observed by Kar et al. (1998), including their observation that striatal cholinergic neurons, which express very low levels of α7nAChR, were insensitive to Aβ peptides.