Address correspondence and reprint requests to H. A. Pearson, School of Biomedical Sciences, University of Leeds, Leeds LS2 9JT, UK. E-mail: firstname.lastname@example.org
The effects of amyloid β protein on voltage-gated K+ channel currents were studied using the whole-cell patch-clamp technique. The 1–40 amino acid form of amyloid β protein was applied to primary cultures of rat cerebellar granule and cortical neurones for 24 h. Both the unaggregated and aggregated forms of the peptide, which have differing biological activities, were used. In cerebellar granule neurones, 24-h pre-incubation with 1 µm unaggregated amyloid β protein resulted in a 60% increase in the ‘A’-type component of K+ current. Increased delayed rectifier activity was Cd2+-sensitive and was presumed to be secondary to an increase in voltage-gated Ca2+ channel current activity. Unaggregated amyloid β protein had no effect on any component of the K+ channel current in cortical neurones. One micromolar of aggregated amyloid β protein had no effect on K+ channel current in either cell type but reduced cell survival within 24 h as measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and terminal deoxynucleotidyl transferase-mediated dUTP nick end labelling (TUNEL) assays. The unaggregated form of amyloid β protein had no neurotoxic effects when applied to either neurone type for up to 72 h. These data indicate that the unaggregated, non-pathological form of amyloid β protein causes changes in the ion channel function of neurones, possibly reflecting a physiological role for the peptide.
sodium dodecylsulfate polyacrylamide gel electrophoresis
terminal deoxynucleotidyl transferase
terminal deoxynucleotidyl transferase-mediated dUTP nick end labelling.
Alzheimer's disease (AD) is a progressive neurodegenerative condition affecting the memory and cognitive abilities of 5–10% of the population over the age of 65. Brains of AD patients are characterized by three diagnostic hallmarks: senile plaques, neurofibrillary tangles and prominent cortical neurone loss (Dickson 1997). The senile plaques are insoluble proteinaceous deposits, principally composed of a 38–43 amino acid polypeptide known as amyloid β protein (Aβ). It is widely believed that the cellular actions of Aβ are responsible for the neurodegeneration observed in AD. The key to the cellular toxicity of Aβ appears to be in its aggregation state (Pike et al. 1993). Aggregated Aβ has been shown to have clear neurotoxic effects when applied to a variety of cultured neurone systems (Lorenzo and Yankner 1994; Howlett et al. 1995; Ueda et al. 1997; Pike 1999). Understandably, the mechanism by which aggregated Aβ promotes cell death has been the subject of intense study. Recent findings have suggested that a neurotoxic disruption of intracellular ion homeostasis may be the result of an action by Aβ on voltage-gated ion channel activity. Both voltage-gated K+ and Ca2+ channels have been implicated in these Aβ-induced changes (Colom et al. 1998; Yu et al. 1998; MacManus et al. 2000). Recent studies by Yu et al. (1998) on primary cultures of rat cortical neurones, and Colom et al. (1998) on the mouse septal cell line SN56, suggest that 10 µm aggregated Aβ may stimulate apoptotic death in these cell types via an increase in the delayed rectifier component of K+ channel current. An increase in Ca2+ channel activity was also implicated because the neurotoxic actions of 10 µm aggregated Aβ could be attenuated using the selective l-type Ca2+ channel antagonist nimodipine, the non-selective blocker cobalt and by removing extracellular Ca2+ (Weiss et al. 1994; Ueda et al. 1997). Although there seems to be clear evidence of a neurotoxic role for the aggregated form of Aβ, the unaggregated non-toxic form of Aβ has no clearly defined function (Vassar et al. 1999; Wolfe et al. 1999). This is surprising, as soluble Aβ has been detected at nanomolar concentrations in the cerebrospinal fluid of healthy human subjects (Seubert et al. 1992) and is secreted into growth media by cultured neurones (Haass et al. 1992). The presence of soluble Aβ under normal conditions therefore suggests the possibility of a physiological role for Aβ. Our previous observation that the soluble, unaggregated form of Aβ could give rise to a modulation of Ca2+ channel current in central neurones (Price et al. 1998) suggests that this may be just such a role and has prompted us to investigate the effect of Aβ on other ionic conductances. We therefore investigated the effects of Aβ1−40 on the K+ channel currents in primary cultures of central neurones from the rat.
Whole-cell patch-clamp of primary cultures from the cortex and cerebellum were used to measure voltage-gated K+ channel activity. Cultures from these two different brain regions were used as neurones from these regions show different susceptibility to neurodegeneration during AD. Parallel experiments measured the toxicity of the peptide application regimens and both the unaggregated and aggregated forms of the peptide were used.
Materials and methods
Culturing of rat central neurones
All experiments were performed using primary cultures of rat cerebellar granule and cortical neurones. Cells were obtained by enzymatic and mechanical dissociation as previously described (Held et al. 1998). Briefly, tissue was removed from 6–8-day-old rat pups (cerebellum) or 16–18-day fetal rats (cerebral cortex) and triturated following a 15-min trypsin (EC 220.127.116.11, 2.5 mg/mL in phosphate-buffered saline) digestion. Trypsin digestion was halted by the addition of a solution containing soybean trypsin inhibitor (0.1 mg/mL). After centrifugation for 1 min at 100 g, cells were resuspended in minimal essential medium (MEM) supplemented with 10% fetal calf serum, 2.5% chick embryo extract, 26 mm glucose, 19 mm KCl, 2 mm l-glutamine and penicillin/streptomycin (50 IU/mL/50 µg/mL). The cells were plated out at a density of 0.25 × 106 cells per well on circular 10-mm diameter poly l-lysine-coated coverslips. Multiwells were incubated in a humidified atmosphere containing 5% CO2 : 95% air at 37°C. After 48 h, the culture medium was exchanged for one consisting of MEM supplemented with 10% horse serum, 2.5% chick embryo extract, 26 mm glucose, 19 mm KCl, 2 mm l-glutamine, penicillin/streptomycin (50 IU/mL/50 µg/mL) and 80 µm fluorodeoxyuridine to prevent proliferation of non-neuronal cells. Culture media were exchanged every 3 days and cells were grown in culture for up to 14 days. All recordings were made from cells between days 5 and 12.
Electrophysiological recording of potassium channel currents
Whole-cell K+ currents were recorded from cells using the patch-clamp technique. Glass micropipettes (3–5 MΩ) were fabricated from borosilicate glass and filled with solution containing KCl 140 mm, CaCl2 0.5 mm, EGTA 5 mm, K2ATP 2 mm, HEPES 10 mm; pH was adjusted to 7.2 with KOH and osmolarity to 320 mOsm with sucrose. The external solution comprised NaCl 120 mm, KCl 2.5 mm, MgCl2 2 mm, CaCl2 0.5 mm, HEPES 10 mm and tetrodotoxin 0.0005 mm; pH was adjusted to 7.4 with NaOH and osmolarity to 320 mOsm with sucrose. For all electrophysiological measurements, serial resistance and capacity transients were electronically compensated.
To record K+ channel current–voltage relationships, cells were held at a potential of − 70 mV and depolarized to potentials ranging from − 60 to + 60 mV beginning 30 s after seal rupture. Each depolarization lasted 85 ms and followed a 250-ms pre-pulse to − 140 or − 50 mV. The steps were repeated every 7.5 s. For perfusion experiments, cells were depolarized to the + 50 mV potential following alternate pre-pulse potentials to − 140 mV or − 50 mV, every 7.5 s. Five leak subtraction steps were made prior to each depolarization to allow off-line removal of linear leak and residual capacity artefacts. In addition to these measurements, the resting cell conductance at − 70 mV was also measured to ensure that changes in leak did not affect the results.
Analysis of electrophysiological recordings
It has been previously shown that the voltage-gated K+ channels in cultured cerebellar granule neurones can be functionally separated by their inactivation properties (Watkins and Mathie 1996) into inactivating (IKA) and the non-inactivating (IKV) components of current. To distinguish between these components, measurements of the peak current, the current at the end of the test step and the difference between these two measurements (peak-end) were made in cells that had been pre-pulsed to a potential of − 140 mV. Peak current was assumed to consist of both IKA and IKV, end current of mostly IKV and peak-end of mostly IKA. Currents could be further separated by applying pre-pulses to − 50 mV, instead of − 140 mV. At this pre-pulse potential, only IKV could be activated.
Current recordings were analysed using the Patch v6.0 program by Cambridge Electronic Design (Cambridge, UK) following leak subtraction using a P/5 subtraction protocol. Further analyses were performed using Microsoft Excel 97 and Microcal Origin v4.1. All data are given as mean ± standard error of the mean. Current density in cells at a given potential was found to be normally distributed using the Kolmogorov–Smirnov test (p < 0.01). Therefore, parametric tests were used to distinguish statistically significant differences between currents evoked in cells. A repeated measures anova (with Tukey–Kramer's post-test) was used to compare differences between entire current–voltage (I–V) relationships, whereas Student's t-test (unpaired) was used to compare points on different curves, but activated by stepping to the same potential. p-Values of less than 0.05 were considered significant. All current recordings were normalized to whole cell capacitance to give current density. Peak current was measured as the maximal current observed during the depolarizing step. Voltage errors due to series resistance were calculated and never came to more than − 2–3 mV. Junction potential error was measured to be + 3 mV. No adjustments for these errors were therefore made.
Application of amyloid β protein
Aβ1−40 was applied in three different ways; acute extracellular exposure, acute intracellular exposure and 24–72 h extracellular incubation. Acute extracellular application of Aβ1−40 was achieved by microperfusing cerebellar granule neurones using a manually switched pressurized perfusion system. Neuronal cultures were perfused with the appropriate bath solution containing 1 µm Aβ1−40 or Aβ40−1. The immediate intracellular effects of Aβ1−40 on K+ channel current were investigated by adding 1 µm Aβ1−40 to the micropipette solution. For incubation of cells, Aβ1−40 was added to the cell culture medium to give a final concentration of between 1 µm and 10 nm and cells were left for 24–72 h before recording or measurement of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide dye (MTT)-reducing activity. As controls, another group of cells from the same batch were incubated with the reverse sequence peptide Aβ40−1 over the same period. No Aβ was present in the solution bathing the incubated cells at the time of electrophysiological recording.
Detection of amyloid β aggregation state
In most experiments, unaggregated Aβ was used. However, when necessary undiluted aliquots were incubated at 37°C for 5–7 days to aggregate the peptides, aggregation state of the stock peptide solutions was assessed by studying protein migration patterns following 16% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE) and staining with Coomassie brilliant blue.
Assessment of peptide aggregation following addition of Aβ to culture media was carried out by western blot. Following 16% SDS–PAGE of samples of culture media to which Aβ had been added, minigels were transferred to polyvinylidene difluoride membranes. Membranes were probed with 6E10 monoclonal antibody (Senetek, New York, NY, USA) raised against amino acids 1–17 of Aβ and a peroxidase-conjugated sheep anti-mouse IgG as the secondary antibody (Amersham Life Science, Buckinghamshire, UK). Both antibodies were used at a 1 : 1000 dilution. Immunolabelled bands were detected using enhanced chemiliminesence.
Neurotoxicity was assessed using the tetrazolium salt MTT assay, which reflects cellular enzymatic activity. MTT is in standard use as a colourimetric substrate for measuring cell viability (Mosmann 1983). When a cell is injured, the redox activity is altered such that they are unable to reduce the dye. Following neurotoxic challenge with Aβ, cells were washed twice with 1 mL phosphate-buffered saline (PBS) to remove culture medium and phenol red colouring. Cells were then incubated in PBS with 20% MTT (5 mg/mL) in 5% CO2 : 95% air at 37°C. After 3 h, a volume of acid propanol (0.04 m HCl in propan-2-ol) equal to that of PBS was added to dissolve blue formazan crystals. Absorbance was measured at a test wavelength of 630 nm and reference wavelength of 550 nm.
The neurotoxic effects of Aβ on central neurones were further investigated using the terminal deoxynucleotidyl transferase-mediated dUTP nick end labelling (TUNEL) assay for DNA fragmentation. Following incubation with Aβ, coverslips were fixed in 10% formalin for 45 min. Cells were initially pre-treated with H2O2 (3% in methanol) for 5 min to block endogenous peroxidase activity. Rat mammary tissue injected with a cancer-inducing virus was used as a positive control. Cells were then washed in Tris buffer for 5 min and the equilibration buffer was applied to re-buffer the tissue and ensure pH and salt concentrations were optimal for the terminal deoxynucleotidyl transferase (TDT) enzyme to work. The TDT enzyme was applied to the positives for 1 h at 37°C in a humidity chamber. Negative controls were incubated under the same conditions with TRIS buffer diluted 1 : 3 in reaction buffer. The reaction was terminated by the application of the stop/wash buffer and incubated for 10 min at room temperature. Following a 5-min wash in Tris, the antidigoxigenin conjugate was applied to each coverslip for 30 min at room temperature in a humidity chamber. After another 8-min wash in Tris, 3,3′-diaminobenzidine (DAB) was added to each coverslip for 6 min at room temperature. Following this, slides were washed and counterstained with haematoxylin. For each treatment group, the percentage of apoptotic cells was measured using a double blind approach. DAB stained cells in a minimum of 30 random fields from three different coverslips were counted for each experimental condition.
Three batches of Aβ1−40 were obtained from Sigma (Poole, UK), Bachem (Saffron Walden, UK) and Glaxo-Wellcome Research (Stevenage, UK). Aβ 25–35 was obtained from Sigma. Stock peptides were stored as powder at − 70°C. Aβ was dissolved in deoxygenated, deionized water at 100 µm (1–40) or 1 mm (25–35), aliquoted to 10 µL, and stored at − 20°C until needed. The reverse sequence peptides Aβ40−1and Aβ35−25 were obtained from Sigma, dissolved in deoxygenated, deionized water at 100 µm (1–40) or 1 mm (25–35), aliquoted to 10 µL and stored at − 20°C.
Tetraethylammonium chloride and 3,4-diaminopyridine were purchased from Sigma. All culture reagents were obtained from Gibco-BRL (Paisley, UK), except chick embryo extract which was purchased from Imperial Laboratories (Andover, UK). TUNEL assays were performed using the ApopTag® Plus assay kit purchased from Intergen (Oxford, UK).
Acute application of unaggregated Aβ1−40 has no effect on K+ channel current
The whole-cell configuration of the patch-clamp technique was used to study the acute effects of Aβ1−40 on K+ channel current in primary cultures of cerebellar granule neurones. There was no effect on cerebellar granule neurone peak, end (IKV) or inactivating (IKA) K+ channel current when cells were microperfused with bath solution containing 1 µm unaggregated Aβ1−40 for a period of 8 min (n = 5). Peak current values after 8-min application were 103 ± 7% of control values (Fig. 1a). Whole-cell leak current at the − 70 mV holding potential was also measured during recordings and was found to be unchanged by Aβ (before Aβ 75.4 ± 9.7pA, during Aβ application 68.0 ± 8.3pA, after Aβ application 85.6 ± 15.5pA, n = 5 no statistically significant differences, one way anova). Adding 1 µm unaggregated Aβ1−40 to the micropipette solution allowed us to observe the immediate intracellular effects of the peptide (Fig. 1b). Again there was no effect of unaggregated Aβ1−40 on IKA or IKV following pre-pulse potentials of −140 mV or −50 mV during a recording period of up to 20 min (I/It = 0 control 0.90 ± 0.07, n = 11; Aβ1−40 0.95 ± 0.14, n = 12). These data indicate that unaggregated Aβ has no acute affect on K+ channels in cerebellar granule neurones, whether applied extracellularly or intracellularly.
Twenty-four-hour application of unaggregated Aβ1−40 enhances K+ channel current in cerebellar granule neurones
In contrast to the lack of effect of acute application of Aβ, pre-incubation of cerebellar granule neurones with 1 µm unaggregated Aβ1−40 for 24 h resulted in an increase in mean K+ channel current when compared with controls (p < 0.001, repeated measures anova). Peak K+ current increased to 140 ± 11% of control values at a potential of + 50 mV following a pre-pulse to −140 mV (n = 36 for controls and n = 44 for Aβ-treated cells. p < 0.001, Student's unpaired t-test, Fig. 1c). IKA (peak-end current) in Aβ-treated cells was increased to 151 ± 13% of control values. IKV (end current) was also significantly increased to 132 ± 13% of control values (p < 0.01). This augmentation in current density was also observed when currents were recorded following a pre-pulse to −50 mV, a protocol that removes the majority of the IKA component of current (Fig. 1d). Mean cell capacitance was unchanged by Aβ1−40 pre-treatment, suggesting that pre-incubation with Aβ1−40 did not cause an increase in cell size (control 2.36 ± 0.09 pF, n = 36; Aβ1−40 2.23 ± 0.08, n = 44). There was no effect of Aβ treatment on resting whole-cell conductance values, an observation which indicates that under these conditions Aβ does not form non-specific cation channels (Aripse et al. 1994) in the cell membrane (control cell conductance 1.4 ± 0.3nS, n = 13; Aβ1−40-treated cell conductance 1.4 ± 0.2nS, n = 16). Three separate batches of Aβ1−40, from different suppliers, were used in independent experiments and similar results were obtained in each case. Data from all three batches of Aβ1−40 were therefore pooled. The reverse sequence of the amyloid fragment (Aβ40−1) which was used as a control had no effect on K+ current when compared with a group of cells pre-incubated with vehicle. For example, at the + 50 mV potential following a pre-pulse potential to − 140 mV, vehicle K+ channel current density was 1.34 ± 0.07 nA/pF (n = 23) compared with 1.32 ± 0.06 nA/pF in the Aβ40−1 treated neurones (n = 22).
Taken together these data indicate that chronic, but not acute, exposure to unaggregated Aβ1−40 can have a modulatory effect on K+ channel currents in these cells. The effect is concentration-dependent and does not appear to be due to an increase in cell size or a non-specific change in cell conductance.
Aβ1−40 enhances K+ channel current in cerebellar granule neurones in the presence of CdCl2
As unaggregated Aβ1−40 has previously been shown to increase Ca2+ channel currents in cerebellar granule neurones (CGN; Price et al. 1998) we investigated the possibility that the enhanced K+ current observed in Aβ1−40-treated cells may be due to an increase in Ca2+-activated K+ channel activity. Ten micromolars of CdCl2 were added to the bath solution during recording of K+ channel currents from cells pre-incubated for 24 h with either unaggregated Aβ1−40 or the reverse sequence of the peptide. At this concentration, Cd2+ blocks 90% of the Ca2+ channel current in these cells (Pearson et al. 1993). In the presence of CdCl2, peak K+ current was increased to 136 ± 12% at the + 50 mV step following a pre-pulse potential to − 140 mV in Aβ1−40-treated cells when compared with controls (Fig. 2a, n = 13 control cells and n = 16 Aβ1−40-treated cells, p < 0.01). The enhancement of current was specific for the IKA component which was increased from 0.63 ± 0.04 nA/pF in control cells to 0.98 ± 0.07 nA/pF (p < 0.001) in Aβ1−40-treated cells, whereas there was no increase in IKV (control 0.42 ± 0.04 nA/pF; Aβ1−40 0.45 ± 0.03 nA/pF). Furthermore, there was no significant difference in either the peak or end current following a pre-pulse to − 50 mV (Fig. 2b). We went on to measure the concentration-dependence of this effect of Aβ to increase the K+ channel current. Figure 2(c) plots the percentage change in IKA and IKV against Aβ concentration for currents recorded in the presence of 10 µm Cd2+ to block any indirect effects via an increase in voltage-gated Ca2+ channels. Pre-incubation of granule neurones with concentrations of Aβ as low as 10 nm produced statistically significant and concentration-dependent increases in IKA. No effect on IKV was seen at any of the Aβ concentrations used and no effect on cell size or resting cell conductance could be observed. These results suggest that only inactivating ‘A’-type K+ channel activity is directly altered by Aβ pre-incubation, whereas changes in the non-inactivating current are secondary to previously observed changes in Ca2+ channel activity. Furthermore, increased K+ channel activity is concentration-dependent and occurs at concentrations of Aβ that have been measured in the cerebrospinal fluid of healthy patients.
In cortical neurones Aβ1−40 has no effect on K+ channel current
The cell death associated with AD is highly region-specific. The cerebellum is largely unaffected by AD, whereas cortical regions suffer from a marked loss of neurones. This may be due to regional differences in exposure to Aβ, or may be due to differential effects of Aβ on cells from cortex and cerebellum. We therefore tested the regional specificity of the Aβ effect by repeating our experiments in cultured rat cortical neurones.
Cortical neurones have lower current densities than granule neurones and characteristically display a much smaller component of inactivating current (Ramsden et al. 1998; see, e.g. Fig. 2d). Mean values of peak K+ current at the + 50 mV test potential following a pre-pulse potential to − 140 mV were 0.45 ± 0.05 nA/pF (control, n = 23). This compared with a mean current density of 0.50 ± 0.05 nA/pF (n = 23) in cells treated for 24 h with unaggregated Aβ1−40. These values were not signifcantly different from each other. As might be expected, there was no effect of Aβ on IKV when evoked in the absence of IKA by use of a pre-pulse to − 50 mV (control 0.26 ± 0.03 nA/pF, Aβ1−40 0.3 ± 0.03 nA/pF, + 50 mV test potential). As with the granule neurones, there was no effect of Aβ on cell conductance (control 0.91 ± 0.15 nS, n = 10; Aβ1−40 0.95 ± 0.22 nS, n = 9). However, when cell capacitance was measured, an increase in capacitance was observed (control capacitance 4.1 ± 0.3 pF, n = 24; Aβ1−40 capacitance 5.5 ± 0.6 pF n = 20, p < 0.05). This suggests that, in these cells, Aβ pre-incubation gave rise to an increase in cell size. However, the lack of effect of Aβ on current density shows that this increase in cell size was not associated with an even greater increase in the K+ current.
K+ channel currents in granule and cortical neurones have a differential pharmacology
The lack of effect of unaggregated Aβ on the K+ channel current in cortical neurones suggests that the effect seen in granule neurones is specific either to the cell type or to the channels expressed by the cells. Therefore the pharmacology of the K+ channel currents in both CGN and cortical neurones were compared using 3,4 diaminopyridine (DAP) and tetraethylammonium ions (TEA+) which classically display a partially selective block of ‘A’-type and delayed rectifier channels, respectively.
The time-course of the effect of 10 mm TEA+ on the K+ current in cerebellar granule neurones and in cortical neurones is shown in Fig. 3(a,b, respectively) and the results are summarized in Fig. 3(c). In both cerebellar granule neurones and cortical neurones there was little inhibition of the peak current activated by a depolarization to + 50 mV from a pre-pulse potential of − 140 mV (cortical 13.2 ± 0.7% inhibition, n = 6, CGN − 0.33 ± 0.02% inhibition, n = 6). In contrast, there was a substantial inhibition of the current at the end of the step (cortical 53 ± 2% inhibition; CGN 30 ± 2% inhibition), consistent with the open channel block that is characteristic of this compound (Yellon et al. 1991). This resulted in an apparent increase in the inactivating component of current, again consistent with an open channel block mechanism of action. The current at the end of the pulse was inhibited to a greater degree in the cortical neurones than in CGN (p < 0.01). A greater percentage of current was blocked following a pre-pulse to − 50 mV where the current evoked is mainly IKV. For instance, current at the end of the pulse was blocked by 73 ± 7% in cortical neurones and by 62 ± 13% in granule neurones. (Fig. 3c). There were no statistically significant differences between block of K+ currents by TEA+ in cortical neurones compared with CGN following a pre-pulse to −50 mV.
The K+ channel current was inhibited in both cell types by 5 mm DAP. Following a pre-pulse potential of − 140 mV, cortical neurone peak K+ current was inhibited by 35 ± 2% (n = 5, Fig. 4b). This inhibition was significantly greater than the inhibition of CGN peak current by DAP (10.0 ± 0.4% inhibition, n = 6, Fig. 4a,c). Inhibition of the K+ current in cortical neurones was mainly due to an effect on the inactivating component which was blocked by 61 ± 13% (Fig. 4c), whereas in granule neurones following a pre-pulse to − 140 mV, a non-inactivating component of current was blocked (see example in Fig. 4a). When a pre-pulse to −50 mV was used, however, the residual inactivating component of current in granule neurones was significantly more susceptible to inhibition than that in the cortical neurones (p < 0.005, Fig. 4c). No other statistically significant differences between DAP inhibition of CGN and cortical neurones were observed at the − 50 mV pre-pulse potential. These data indicate that, although small in amplitude, the inactivating current in cortical neurones displays classical properties of the ‘A’-type current in that it is susceptible to block by DAP and inactivation by pre-depolarizations. The inactivating current in the granule neurones, however, is relatively resistant to the classic ‘A’-type current blocker DAP.
In summary, there are clear differences between CGN and cortical neurone K+ currents, both in terms of the relative contributions of IKA and IKV to the overall current, and in terms of the profile of their block by TEA+ and DAP.
Aβ-induced changes in ion channel currents are not correlated with neurotoxicity
Previous studies have indicated that an Aβ-induced increase in K+ currents is responsible for the neurotoxicity of the peptide (Colom et al. 1998; Yu et al. 1998:). To determine the relevance of the changes in ion channel activity observed in this study to the neurotoxicity of the Aβ peptide, the effects of both unaggregated Aβ1−40, and Aβ1−40 that had been allowed to aggregate were tested on cell viability using the MTT assay. Unaggregated Aβ1−40 (1 µm) was not toxic to either cerebellar granule neurones or cortical neurones during an incubation period of up to 72 h (Fig. 5a,b). Thus, the enhanced CGN IKA observed with the unaggregated form of Aβ is not associated with cell death. When neurones were incubated with Aβ1−40 (1 µm) which had been allowed to aggregate, cell viability was significantly decreased. Similar results were obtained for both cerebellar granule (Fig. 5a) and cortical neurones (Fig. 5b). Aggregated Aβ caused significant reductions in cell viability as early as 24 h. As the MTT assay may be considered to measure changes in cell enzyme activity rather than cell death per se (Liu et al. 1997), results using MTT were corroborated using the TUNEL assay for DNA fragmentation. Application of 1 µm aggregated Aβ1−40 to cortical neurones for 24 h resulted in a statistically significant increase (p < 0.001) in the percentage of TUNEL-positive cells when compared with cells treated with the reverse sequence of the peptide (Fig. 5c). In contrast, cells pre-treated with unaggregated Aβ1−40 showed no difference in the percentage of TUNEL-positive cells compared with controls. We also tested the effect of 10 µm Aβ25−35, the toxic fragment of the peptide. Twenty-four hours of pre-incubation of the same batch of cells with both the unaggregated and aggregated forms of this peptide fragment resulted in significant increases in TUNEL-positive cells when compared with neurones pre-treated with the reverse sequence of the peptide (Fig. 5c). The results of these two different assays for cell viability indicate that the unaggregated form of Aβ1−40 was not toxic to cells within the time periods used in this study, whereas the aggregated form has clear neurotoxic properties. In the light of these results we went on to test the effect of aggregated Aβ on the K+ channel current in both cell types.
Aggregated Aβ1−40 has no effect on K+ channel currents
Aβ was aggregated before application to cells by incubation at 37°C for 5 days. The reverse sequence of the peptide (Aβ40−1) was treated in the same manner. Cerebellar granule neurones were then pre-incubated with either aggregated Aβ1−40 or Aβ40−1 for 24 h prior to recording. Under these conditions, Aβ1−40-treated cells had the same K+ channel current densities when compared with controls (aggregated Aβ40−1). Neither the peak, the end nor the inactivating components of current were any different from each other when measured following a pre-pulse to − 140 mV (Fig. 6a). Furthermore, there was no difference in any component of current following a pre-pulse potential of − 50 mV (Fig. 6b) or in whole-cell capacitance (control 2.7 ± 0.1 pF, n = 20; Aβ1−40 2.7 ± 0.1 pF, n = 18). A similar lack of effect of aggregated Aβ1−40 was seen when the peptide was applied to cortical neurones. No effect was observed on any component of cortical neurone K+ channel current following a pre-pulse potential to either − 140 or − 50 mV (Fig. 6c,d). Again, there was no effect of aggregated Aβ on cortical cell capacitance (control 5.4 ± 0.2 pF, n = 23; Aβ1−40 5.5 ± 0.3 pF, n = 23). Thus the neurotoxic, aggregated form of Aβ1−40 has no effect on the K+ channel current in either cell type.
Assessment of Aβ solubility state
On SDS–PAGE gels, stock samples of unaggregated Aβ dissolved in either H2O or dimethyl sulfoxide showed a broad band of staining at 7 kDa and high molecular weight bands (approx. 34 kDa) were observed in samples of H2O-dissolved Aβ that had been aggregated (Fig. 7a). To ensure that unaggregated samples of Aβ did not undergo significant aggregation during the 24-h pre-incubation with cells, aggregation state of the Aβ peptide in culture media was assessed by western blotting. Figure 7(b) shows western blots of culture media immediately after addition of 1 µm Aβ and following incubation of Aβ-containing media for 24 h at 37°C in presence of neuronal cultures. Aβ was dissolved in either H2O or dimethyl sulfoxide to give stock solutions at a concentration of 1 mm. Some samples of the H2O stock were incubated in an equal volume of PBS at 37°C to induce aggregation. The most prominent band in all samples corresponded to the monomeric peptide. No differences were observed between unaggregated samples that had been dissolved in either H2O or dimethyl sulfoxide. Samples that had been allowed to aggregate prior to dissolution in culture media, however, showed additional high molecular weight bands (> 39 kDa) which probably represent stable aggregates of Aβ. Incubating samples for 24 h in the presence of neurones did not substantially alter the migration profile of unaggregated or aggregated Aβ. These data indicate that aggregation of Aβ gives rise to formation of a high molecular weight band and that during the 24-h pre-incubation of cells with unaggregated Aβ, no detectable formation of high molecular weight peptide aggregates takes place.
The dominant histological hallmark of AD is the occurrence of senile plaques, which are protein deposits composed primarily of the Aβ peptide. When Aβ is allowed to aggregate it becomes insoluble and forms plaques. Although not proven, it is widely believed that aggregated Aβ gives rise to the symptomology and toxicology observed in AD (Selkoe 1996). Not surprisingly, interest in Aβ has centred on its role as a possible neurotoxin in AD. Here, we report for the first time an action of Aβ that does not appear to be related to its toxicity but which may relate to a physiological function for a peptide known to be secreted by healthy neurones under normal conditions.
In this study we investigated the effects of unaggregated Aβ1−40 and Aβ1−40 that had been allowed to aggregate, on K+ channel currents in neurones from the cerebellum and the cortex. Unaggregated Aβ specifically increases IKA in cerebellar granule neurones but has no effect on any component of cortical neurone IKV. The sensitivity of the K+ channel current present in both these cell types to K+ channel blockers was studied and it was found that currents in the two cell types were differentially affected by IKA blocker 3,4-DAP. When the amyloid fragment was allowed to aggregate, there was no effect of Aβ on any component of IKV in either cell type. The efficacy of Aβ in inducing cell death was found to be dependent on aggregation.
The lack of effect of acutely applying Aβ1−40 by extracellular perfusion to CGN indicates that the mode of action has a longer time course than the 8 min used in these experiments. This suggests the involvement of uptake mechanisms or of other mediators in a cascade of events. The lack of effect of internal application via the micropipette solution on the K+ channel current would suggest that internalization of Aβ is not the rate-limiting factor. Furthermore, it suggests that a direct interaction of Aβ with the K+ channels is unlikely.
A longer incubation period of 24 h induced distinct effects on the K+ channel current, indicating a clear time-dependence to the effect in common with observations in other cell types (Yu et al. 1998; Taylor et al. 1999). Peak K+ channel currents in cerebellar granule neurones were increased by approximately 40% when compared with controls. This enhancement of peak K+ channel current was due to an effect of Aβ on IKA which was increased by over 50%. However, the delayed rectifier K+ channel current was also increased by 30%. When experiments were repeated in the presence of cadmium, a non-selective blocker of Ca2+ channels, the effect of Aβ1−40 was confined solely to the inactivating component of current. Furthermore, there was no increase in any component of current following a pre-pulse potential of − 50 mV (when only IKV was evoked). We have previously shown that 24 h incubation with 1 µm Aβ1−40 increases Ca2+ channel current in cerebellar granule neurones (Price et al. 1998). The concentration of CdCl2 used here will block 90% of the Ca2+ channel current (Pearson et al. 1993). This suggests that the apparent increase in IKV was due to increased Ca2+ activated K+ channel activity which itself was secondary to an increased Ca2+ influx through N-type Ca2+ channels. On the other hand, there appeared to be a specific effect of Aβ to increase IKA. This proclivity of Aβ1−40 to specifically increase IKA in CGN was found to be highly reproducible. Similar effects were seen using three different batches of Aβ from three different suppliers.
Although the cultured rat cerebellar granule neurones are a useful and widely used model for studying the function of central neurones, the cerebellum remains unaffected until the very latest stages of AD. In contrast, cortex is affected relatively early. To examine whether there were differences in the action of Aβ between neurones cultured from different brain regions, the effects of Aβ on cortical neurones were studied. When we repeated our incubation studies using pyramidal neurones from cortical cultures, we observed no effect of unaggregated Aβ on any component of K+ channel current. Cortical neurones characteristically displayed a much smaller component of inactivating K+ channel current than granule neurones and the K+ current densities are three- to fourfold lower. This may reflect differences in the importance of voltage-gated K+ channels in these cell types and/or a different functional role for IK in the two brain regions. The K+ channel currents were further characterised using 3,4-DAP and TEA+. Again there were definite differences in the susceptibility to block by DAP in the two types of neurone. The inactivating component of K+ channel current was susceptible to inhibition by DAP only in the cortical neurones, whilst the delayed rectifier K+ channel current was equally affected in both brain preparations. The fact that the cortical and granule neurones are differentially affected by DAP supports the notion that these brain regions express different subtypes of voltage-gated K+ channels. This could explain the differential effects of Aβ on the K+ channel currents in these two cell types. Moreover, these data suggest that the actions of Aβ on K+ channel currents in central neurones is via a specific action, perhaps on a subunit that is present or accessible in granule neurones but not in cortical neurones. It could also be argued that the cell type-dependent effects of Aβ1−40 support the idea that Aβ1−40 action on granule neurone IKA is a genuine alteration of channel activity rather than a phenomenom or peculiarity.
It has been suggested that Aβ may be responsible for the neurodegeneration observed during the progression of AD (Selkoe 1999). We incubated central neurones with unaggregated Aβ1−40 to investigate whether the effects of the peptide on K+ channel currents were associated with cell death. After an incubation period of 24 h, there was no effect of unaggregated Aβ1−40 on cell survival in either CGN or cortical neurones using two different cell viability assay methods (TUNEL and MTT assays). Thus the increase in the inactivating component of K+ channel current in CGN following 24 h incubation with unaggregated Aβ1−40 is not associated with cell injury. Furthermore, the effects on IKA do not appear to precede reductions in cell viability, as there is no apparent cell injury when cells are incubated with Aβ1−40 for 72 h. The lack of effect of Aβ1−40 incubation on cortical neurone K+ channel current is not associated with a decrease in cell viability, indicating an absence of Aβ-induced cell injury which might reduce K+ channel activity. In AD, neuronal death in cortex occurs well in advance of any cell death in cerebellum. In this study, cerebellar granule and cortical neurones showed a similar vulnerability to Aβ toxicity. This may suggest that neuronal death in AD is dependent on regional differences in Aβ production/deposition rather than a selective vulnerability of neurones. However, protein expression in these neonatal and fetal cultured neurones is likely to be different to that occurring in adults in vivo. Furthermore, these are dissociated cultures which do not retain the same complex neuro-neuronal interconnections that exist in vivo, although they can be induced to release glutamate (Huston et al. 1993) and display a degree of functional connectivty (Mellor and Randall 2001).
Recent studies suggest that the toxicity of Aβ is dependent on solubility state and that toxicity is associated with amyloid aggregation (Pike et al. 1993; Howlett et al. 1995; Puttfarcken et al. 1996; Estus et al. 1997). We investigated the effect of aggregated Aβ1−40 on cell viability. The aggregated Aβ1−40 caused a reduction in viability of around 30% compared with controls, in both CGN and cortical neuornes after a 72-h incubation. MTT data were supported by data obtained using the TUNEL assay which indicates DNA fragmentation, one of the markers of apoptosis. These data support current dogma which states that Aβ only becomes toxic to central neurones when it has aggregated. These data also showed that the aggregated Aβ1−40, which had no detectable effect on IK at this time point, was biologically active and may induce apoptosis. To confirm that the different biological activities of aggregated and unaggregated Aβ (i.e. effects on IKA and toxicity) were due to aggregation of the peptide, we studied the different size-forms present throughout the incubation period by western blot. High molecular weight size forms of Aβ were present in cultures where Aβ had been aggregated prior to application, but were absent in cultures where the unaggregated form had been applied. The high molecular weight aggregates were therefore associated with toxicity, whereas in their absence Aβ had modulatory effects on the K+ current that were not associated with toxicity.
Two recent studies have shown that Aβ can cause apoptotic cell death via an increase in the delayed rectifier K+ channel current (Colom et al. 1998; Yu et al. 1998). Apoptotic death induced by various Aβ species could be attenuated with TEA which is a known blocker of IKV. These studies differed from ours in several important methodological respects. Firstly, higher Aβ concentrations (10–20 µm) were used. Second, all peptides used in the earlier studies were aggregated prior to incubation with cells and no experiments with the non-toxic unaggregated forms were carried out. Third, there were differences in species (mouse neurones as opposed to rat) and, in one study, differences in culture type (cell lines as opposed to primary culture). Although the primary conclusions that Aβ can alter ion channel activity of neurones are broadly similar, differences arise when the channel subtypes affected by Aβ are compared. In both the studies using aggregated Aβ, a non-inactivating, delayed rectifier type current was increased, an effect associated with cell death (Colom et al. 1998; Yu et al. 1998). In our study using the unaggregated form of Aβ, the inactivating ‘A’ type K+ current is increased, but no simultaneous reduction of viability was seen. It is clear that, in all three studies, the effects of Aβ on cell survival are dependent on the aggregation state of the peptide.
Soluble Aβ has been detected at concentrations of 1–10 nm in the CSF of healthy human subjects and is secreted into the growth media by cultured neurones (Haass et al. 1992; Seubert et al. 1992; Shoji et al. 1992; Van Nostrand et al. 1992). This concentration range might therefore be considered the lower physiological range, as CSF removal in these studies was carried out by lumbar puncture, whereas the concentration near release sites and in the brain would likely be higher. In this study we typically used a concentration of 1 µm Aβ as this made results clearer and easier to measure, but we also saw the same effects of Aβ at concentrations between 1 and 100 nm. All the concentrations used were much lower than those used by others reporting a change in ion channel currents.
The presence of Aβ in the CSF of healthy individuals suggests the possibility of a physiological role for Aβ. Moreover, APP is ubiquitously expressed (Hardy 1997) and there is a sequential enzymatic cascade in place to yield amyloid species. Although unaggregated Aβ has been found to be constantly anabolysed and catabolysed under normal conditions (Iwata et al. 2000; Vekrellis et al. 2000), its definite in vivo function remains a mystery (Vassar et al. 1999; Wolfe et al. 1999). That Aβ may have a more important role in the brain than other areas of the body is suggested by the particularly high level of expression of APP in neurones. As the activity of voltage-gated ion channels is essential for the functioning of neurones in detecting, amplifying and reshaping electrical messages, it is possible that the altered activity of ion channels by Aβ represents a normal physiological function. This is supported by reports of a modulatory effect of other APP secretion products on ion channel currents in neurones (Furukawa et al. 1996).
The changes in ion channel activity that we have observed are not neurotoxic, while the increase in delayed rectifier observed by other groups using aggregated Aβ precedes neurodegenerative changes in the cell cultures. It is possible that the difference in neurotoxicity of the two forms of the peptide (aggregated or unaggregated) hinges on changes in the subtype of ion channel affected. However, it is also possible to speculate that Aβ in the unaggregated form increases the activity of inactivating, ‘A’ type K+ channels but has no neurotoxic effects. When the Aβ peptide aggregates, it may alter the activity of a separate class of ion channel, or induce other cellular cascades, leading to neurotoxic effects on the neurones. If this is the case, it raises the intriguing possibility that Aβ has a normal functional role in modulating the activity of K+ channels in central neurones. Such an hypothesis is supported by our data showing that this occurs at concentrations of Aβ that can be found in the CSF of healthy individuals (10 nm).
So what would one expect the physiological consequences of modulating IKA to be? The inactivating A-type current is thought to be the predominant K+ current in many neurones (Bardoni and Belluzzi 1993). As ‘A’ type K+ channels alter dendritic signal proporgation (Hoffman et al. 1997), substantial effects of soluble Aβ1−40 on both synaptic efficacy and action potential frequency might be expected. Voltage-gated K+ channels are essential to the function of all excitable cells and have been suggested to govern the discharge pattern of action potentials (Song et al. 1998; Kanold and Manis 1999). K+ channels are involved in shaping action potentials, modulating the frequency of action potentials (Giles and Shibata 1985), setting the resting membrane potential and functioning in the cellular mechanisms of learning and memory (Alkon et al. 1982; Klein et al. 1982; Siegelbaum et al. 1982). Enhancement of K+ channel current may alter cell functions by attenuating membrane potential changes and preventing Ca2+ entry and Ca2+ signalling, as demonstrated in other cell models (Philipson et al. 1994). Alteration of neuronal K+ current may lead to profound changes in excitability (Lockery and Spitzer 1992), and this in turn could have cause regional differences in levels of neurotransmitter release and excitotoxicity.
A physiological role for Aβ is further supported by the observation that soluble Aβ has been shown to modulate neurotransmitter release, an extremely important process in the brain (Arias et al. 1995; Kar et al. 1996). Moreover, Aβ has been shown to suppress long-term synaptic plasticity in the hippocampus at subneurotoxic concentrations (Chen et al. 2000). In the present study, soluble Aβ1−40 had affects on ion channels currents and these modulations were not associated with neurotoxicity.
Thus, alterations in K+ channel activity by soluble Aβ1−40 may be important in CNS development and neuronal function. During normal ageing, the metabolism of this peptide may be subtly altered or mismanaged, resulting in the deposition of excess Aβ1−40 as non-toxic diffuse plaques in the aged-brain. Other modulations in Aβ production, such as increased generation of Aβ1−42, reduced clearance or the development of specifically located microenvironments that promote aggregation, may lead to the development of neuritic plaques and the subsequent neurodegeneration that characterises AD.
In summary, we have shown that unaggregated Aβ can alter K+ channel activity in primary cultures of central neurones. These effects appear to be specific to cell type and channel subtype. Furthermore, they do not lead to cell death. It is possible that the effects seen in this study may be indicative of a normal functional role for Aβ in the CNS.
We thank Drs T. Rupniak and I. K. Anderson of Glaxo-Wellcome Research for Aβ1−40, Dr C. Shukla for assistance with TUNEL assays and Dr V. A. Campbell for critical discussion of the manuscript. Supported by the Wellcome Trust and the MRC.