Amyloid-beta increases acetylcholinesterase expression in neuroblastoma cells by reducing enzyme degradation


Address correspondence and reprint requests to Stephen Brimijoin, Department of Molecular Pharmacology, Mayo Medical School, Rochester Minnesota, USA. E-mail:


Amyloid-beta (Aβ) is the principal protein constituent of ‘senile plaques’ and is a suspected mediator in Alzheimer's disease (AD). Senile plaques also contain acetylcholinesterase (AChE; EC, which may have a role in promoting Αβ-toxicity. We have found that Αβ can affect AChE expression in a neuron-like line, the N1E.115 neuroblastoma cell. When 1 µmΑβ 1–42 or 25–35 was added for 24 h to differentiating N1E.115 in culture, AChE activity increased 30–40% in adherent cells, and 100% or more in nonadherent cells. The changes in both tetrameric (G4) and monomeric (G1) AChE forms were comparable. Turnover studies indicated that the elevation of AChE activity reflected slowed AChE degradation rather than accelerated synthesis. With a similar time course, Αβ also increased the quantity of muscarinic receptors on the plasma membrane. Immunocytochemistry for a lysosomal membrane protein (LAMP-1) indicated no change in abundance or localization of lysosomes in treated cells. But decreased labeling by pH-sensitive fluorescent dye pointed to an impairment of lysosomal acidification. We consider that the alteration of AChE expression after Αβ-exposure could reflect lysosomal dysfunction, and might itself enhance Αβ-toxicity.

Abbreviations used

amyloid beta




Alzheimer's disease


amyloid precursor protein




lysosomal-associated membrane protein 1


pyridine-2-aldoxime methochloride


quinuclidinyl benzilate

The senile plaques in brains of patients with Alzheimer's disease (AD) contain aggregated amyloid-beta (Αβ) protein, suspected as a primary cause of neurodegeneration (Glenner and Wong 1984; Pike et al. 1992; Yuan and Yankner 2000). Both Αβ 1–42 and Αβ 25–35, a major AD species and an unnatural aggregating fragment, will induce apoptosis in neuronal culture (Pike et al. 1991; Busciglio et al. 1992). This toxicity is still poorly understood, but possible mechanisms include free radical damage and disrupted function of mitochondria or lysosomes (Mattson 1995; Schapira 1996; Nixon et al. 2000; Bahr and Bendiske 2002).

Acetylcholinesterase (AChE) is another component of senile plaques that, interestingly, will promote Αβ aggregation in vitro and enhance Αβ toxicity in neuronal tissue culture (Inestrosa et al. 1996; Alvarez et al. 1998). Recent work in our laboratory suggests that an analogous process can occur in the brain. We found more numerous plaques in hybrid transgenic mice that over-express human AChE and amyloid precursor protein (APP) than in mice expressing APP alone (Rees et al. 2003; Rees, unpublished data). Other groups have discovered that Aβ may in turn influence AChE expression. For example, addition of Aβ causes elevated AChE activity in embryonal carcinoma P19 cells (Sberna et al. 1997). Likewise, APP-expression in some transgenic mouse lines is linked with increased AChE activity in certain brain regions (Sberna et al. 1998). To investigate such effects further, we have now examined the apparent rates of AChE synthesis, degradation, and release in N1E.115 neuroblastoma cells during exposure to Aβ.

Materials and methods

Amyloid-beta 1–42 and 25–35 (US Peptides, Rancho Cucamonga, CA, USA) were reconstituted in 100 µL of dimethylsulfoxide (10 mg/mL) and stored at −20°C. Before addition to cultures, Αβ 1–42 was diluted 200× and incubated in distilled water at 37°C for 3 days to promote aggregation and toxicity (Lorenzo and Yankner 1994). Αβ 25–35 was used directly after reconstitution because it proved to be already neurotoxic. Dulbeco's modified Eagle's medium and fetal bovine serum were purchased from Invitrogen (Carlsbad, CA, USA); LysoSensor DN-189 and FITC-dextran were from Molecular Probes (Eugene, OR, USA); monoclonal FITC-conjugated antibody against lysosomal-associated membrane protein 1 (LAMP-1) was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Radiolabeled [3H]acetylcholine was from Dupont–NEN (Boston, MA, USA). Other materials were purchased from Sigma-Aldrich (St. Louis, MO, USA).

Cell culture and viability assays

Cells for study were wild-type N1E.115 neuroblastoma and a stable transfectant subline that expresses low levels of AChE under the influence of an antisense plasmid (Koenigsberger et al. 1997). The cells were initially maintained in 10% CO2 at 37°C in Dulbecco's modified Eagle's medium with 2.5% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin. After 24 h, this medium was replaced with serum-free medium containing Αβ or dimethylsulfoxide vehicle together with various drugs and test agents. Cell viability was assessed from mitochondrial reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide added directly to the culture wells at 1 mg/mL. Absorbance was monitored at 560 and 650 nm in a plate reader (Mossman 1983). A complementary index of cell death was obtained from the release of cytoplasmic lactate dehydrogenase. Lactate dehydrogenase activity in medium and cell lysates was determined spectrometrically in 50-µL aliquots by a CytoTox 96 Assay kit (Promega, Madison, WI, USA).

AChE assay

Cells were lyzed by freeze-thawing in 200 µL of 0.1% Triton X-100 in 10 mm Tris-buffer, pH 7.4, followed by centrifugation (10 000 × g, 10 min). AChE activity in 50-µL aliquots of lysate supernatants and medium was assayed by measuring release of [3H]acetate from [3H]acetylcholine over 90 min (Johnson and Russell 1975) under conditions ensuring linearity with the amount of sample. AChE kinetics were examined with the acetylthiocholine assay (Ellman et al. 1961), using substrate concentrations from 30 nm to 1 mm. All reactions ran at room temperature in the presence of a butyrylcholinesterase inhibitor (ethopropazine, 10−4 m). AChE activities were normalized in terms of total protein content determined by the BCA method (Pierce Chemical, Rockford, IL, USA), standardized with bovine serum albumin.

Echothiophate labeling and analysis of AChE isoforms

Under appropriate conditions, echothiophate inhibits cell-surface AChE without affecting intracellular AChE (Rotundo 1983). To use this effect for selective labeling, echothiophate was added to media at a concentration of 1 µm for 15 min at 4°C. Lysates of rinsed cells (see above) were assayed for AChE activity, either directly or after reactivation with 10−5 m pyridine-2-aldoxime methochloride (2-PAM, Sigma). The activity restored by the 2-PAM treatment was considered to reflect unmasked, echothiophate-sensitive AChE from the cell surface. Molecular forms of AChE, along with catalase as a sedimentation marker, were separated on 5-mL gradients of sucrose (5–20%) in 50 mm buffer (Tris-HCl, pH 7.4, 1 m NaCl, 0.2 m EDTA, 1% Triton X-100). After centrifugation for 16 h at 116 000 g (SWi55.1 rotor, Beckman Instruments, Palo Alto, CA, USA), consecutive 200 µL fractions were recovered from the bottoms of the tubes and assayed for AChE activity.

AChE synthesis and degradation

Levels of AChE mRNA (synaptic form) were determined by semiquantitative RT-PCR (Koenigsberger et al. 1998). The PCR primers were E3-UP 5′-CGGGTCTACGCCTACGTCTTTGAACACCGTGCTTC-3′ and E6-LP 5′-CACAGGTCTGAGCAGCGATCCTGCTTGCTG-3′. After 5 min at 94°C, the samples were amplified for 25 cycles: 94°C for 1 min, 55°C for 30 s, 72°C for 30 s, and final extension at 72°C for 10 min. Data were analyzed with reference to a standard curve generated with known amounts of template (cDNA prepared previously from control N1E.115 cells). De novo synthesis of AChE was determined from the reappearance of enzyme activity after irreversible inhibition by 10−7 m diisopropyl-flurophosphate (DFP). DFP was added for 15 min, then removed by three 5-min washes with serum-free Dulbecco's modified Eagle's medium. Washed cells were returned to their original media (± Aβ) and harvested for analysis 15, 30, 45, and 75 min later. AChE degradation was estimated after adding cycloheximide (10 µg/mL), which reduced [3H]leucine incorporation into total protein by 99%.

Lysosomal morphology and integrity

To assess lysosomal morphology, cells were fixed in 3.3% formaldehyde (15 min, 23°C), permeabilized with 0.1% Triton, and immunostained with FITC-conjugated LAMP-1 antibody (5 µg/mL, 1 h). To assess lysosomal acidity, the pH-sensitive fluorescent dye, LysoSensor DN-189, was added for the last 1 h of a 24-h exposure to Αβ. To assess lysosomal integrity (Yang et al. 1998), FITC-dextran (10 µg/mL) was added at the beginning of an Aβ-exposure or for the last 2 h. Cells incubated with dye or dextran were fixed immediately afterward with formaldehyde and observed under a Zeiss confocal microscope with rigidly standardized settings of relative gain and contrast.

Muscarinic receptors

A modified quinuclidinyl benzilate (QNB)-binding assay was used to quantitate muscarinic receptors on N1E.115 membranes. After cells were treated for 24 h with Αβ, medium was removed and replaced with 200 µL of fresh medium containing [3H]QNB. Non-specific binding was measured by pre-incubation with 10−4 m atropine. Following 1 h of QNB exposure at 4°C, cells were washed in 1 mL of phosphate-buffered saline and collected in 200 µL of phosphate-buffered saline; bound radiolabel in 80-µL aliquots was then filtered and quantitated by scintillation counting. A pilot study of control cells showed that 4 nm QNB was near-saturating and gave an adequate ratio of specific to non-specific binding (> 2×). This concentration was used later in single-point determinations to estimate abundance of muscarinic receptors.

Calculations and statistical analysis

Mean differences between groups were evaluated by t-test; p < 0.05 was deemed statistically significant. To derive kinetic constants for acetylthiocholine hydrolysis, measured reaction velocities were fitted directly to the hyperbolic Michaelis–Menten equation, using unweighted non-linear regression analysis with SigmaPlot 5.1 (Jandel Scientific, Temecula, CA, USA). SigmaPlot was also used to determine AChE degradation rates by fitting AChE activities, measured up to 8 h after inhibition of protein synthesis, to a model of first-order exponential decay: A = A0ekt.


Aβ-treatment increases AChE activity

Adherent, differentiating, wild-type N1E.115 cells exhibited a consistent 30–35% increase in total AChE activity after a 24-h exposure to 1 µm concentrations of aggregated Αβ 1–42 or freshly prepared Αβ 25–35 (n = 27, p < 0.0001). To test for isoform-specific changes, AChE tetramers (G4) and monomers (G1) were analyzed on sucrose density gradients. The activity of these two forms increased in proportion to the increase in total AChE activity, by 38 ± 9% and 35 ± 5%, respectively (p < 0.05 and p < 0.01; Fig. 1a). Similar increases of AChE activity, again with no change in G4/G1 ratio, also occurred under non-differentiating conditions, in serum-supplemented cultures (data not shown). Much larger increases of G4 and G1 activities (95 ± 40% and 120 ± 28%, p < 0.02) were found in nonadherent cells, tested as potential targets of greater Aβ toxicity (Fig. 1b). Aβ-treated cells did not release abnormal amounts of lactate dehydrogenase into the culture media, but they showed much less 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide-reduction (34 ± 3% of control, p < 0.01). Altogether the data suggest that Αβ-treatment increased AChE expression in cells that remained viable, but with compromised mitochondrial function.

Figure 1.

Effects of pre-incubated Aβ1–42 on AChE expression in N1E.115 cells. Cells were cultured for 24 h with 10% serum and then in serum-free media containing either 1 µm pre-incubated Aβ1–42 in dimethylsulfoxide, or vehicle alone. Isoform-specific changes were analyzed by sucrose density gradients. Shown are representative gradient profiles of AChE activity from adherent (a) and nonadherent (b) cells after Aβ treatment. Here and elsewhere, reaction velocities are in milliunits (1 mU = 1 nmol ACh hydrolyzed per min).

In these experiments, AChE activity was used as an index of enzyme expression, providing accurate quantitation of small changes in samples with little protein. Although AChE activity as a rule is tightly correlated to the amount of AChE protein (Brimijoin et al. 1987), it was important to determine whether apparent changes in AChE expression could reflect altered substrate kinetics. However, AChE from Αβ-treated cells and controls exhibited virtually the same Km values (144 ± 16 µm and 124 ± 17 µm, respectively). Therefore, the increased activity observed after Αβ-exposure was judged to reflect a real increase in AChE content.

Intracellular and extracellular AChE accumulation

The plasma membrane is largely impermeable to echothiophate, an irreversible, positively charged inhibitor of AChE. Hence, in intact cells, the echothiophate-sensitive pool of AChE primarily consists of enzyme anchored on the surface membrane (Rotundo 1983). Using echothiophate, we selectively inhibited this enzyme pool in adherent N1E.115 cells, then processed lysates with and without reactivation by 2-PAM. The activity specifically restored by 2-PAM was ascribed to echothiophate-sensitive AChE from the cell membrane, whereas activity in raw samples was ascribed to intracellular AChE. So defined, membrane-associated and intracellular AChE activities each increased by 30–40% in Aβ-treated cells (Fig. 2a). In samples fractionated by sucrose density gradient ultracentrifugation, similar increases were also noted for intracellular G1, membrane G1, and membrane G4 AChE (Fig. 2b).

Figure 2.

Effects of Aβ on surface and intracellular AChE. (a) Membrane-associated AChE was selectively inhibited by treating intact cells with 10−6 m echothiophate. Lysates were then reactivated with 10−5 m 2-PAM. Echothiophate-sensitive and resistant AChE were considered to be, respectively, surface and intracellular enzyme. Values, as percentage of vehicle control, are mean ± SEM of 10 independent determinations (*p < 0.05; **p < 0.01; ***p < 0.001). (b) Samples from six experiments were analyzed on sucrose density gradients. Representative profiles are shown for cell lysates assayed with 2-PAM reactivation or without (‘raw’).

AChE synthesis and degradation

We sought to determine whether the elevated AChE activity in Αβ-treated cells reflected increased synthesis or decreased degradation. Semi-quantitative RT-PCR for AChE mRNA was used as one indicator of synthetic potential (Fig. 3). The results showed that mRNA levels in Aβ-treated cells did not increase, averaging 83 ± 4% of those in controls. For another index of synthesis we followed the reappearance of AChE activity in cells pulsed with a cell-permeating organophosphate, DFP (10−7 m), which irreversibly inhibits all forms of AChE. Just after DFP washout, AChE activity was near zero. Activity steadily reappeared over the next few hours, but the recovery was no faster in cells exposed to Αβ than in cells exposed to vehicle alone (Fig. 4). Thus the evidence argued against an elevation of synthesis.

Figure 3.

Effect of Aβ on AChE mRNA levels. (a) Total cell-associated mRNA was reverse transcribed from control cells after 24 h of incubation, and a standard curve was constructed by PCR amplification of varying amounts of template (see Methods). Quantification of band intensity by NIH Image after gel staining with ethidium bromide indicated near-linear input–output relations over a wide range of template concentrations. (b) Semi-quantitative RT-PCR of total cellular mRNA (1× template concentration) after 24 h of treatment with vehicle (lanes 1–3) or Aβ (lanes 4–6). Band intensities are expressed as percentages of the vehicle-treated control values (shown are mean values ± SEM of four independent experiments). The apparent decrease after Aβ treatment was not statistically significant (p = 0.08).

Figure 4.

Effect of Aβ on AChE synthesis. Total cell-associated AChE was inhibited by 10−7 m DFP after 22 h of vehicle or Aβ-treatment, and the recovery of AChE activity was monitored after DFP was removed. Values show mean ± SEM of nine independent determinations. Slopes of the fitted linear regression curves did not differ significantly.

Since Αβ-exposure did not alter the apparent rate of AChE synthesis, we also assessed degradation. For this purpose, protein synthesis was blocked with cycloheximide and the ensuing decline of AChE activity was fitted to a model of first order exponential decay (Fig. 5). The data did not perfectly fit this simple model, perhaps because different molecular forms had different half-lives. Nonetheless, the results clearly indicated a substantial slowing of degradation. The overall AChE half-life, computed for purposes of comparison, was 56% longer in Αβ-treated cells (t1/2, 10.5 h) than in vehicle controls (t1/2, 6.7 h). This effect is sufficient to account for the increased equilibrium level of enzyme activity actually observed after 24 h of Aβ-treatment.

Figure 5.

Effect of Aβ on AChE degradation. After 20 h of incubation in either vehicle or Aβ, cycloheximide (10 µg/mL) was used to inhibit protein synthesis. AChE activities, determined in lysates after exposure to cycloheximide for the indicated times, were normalized as percentages of the mean control value at the 1 h point. Shown are mean ± SEM of 15 independent determinations (some error bars obscured by data symbols). Aβ-treated cells had statistically different AChE activities from control at 2 h after cycloheximide addition (*p < 0.01, **p = 0.001, ***p < 0.0001). Apparent half-lives were estimated from a first-order exponential decay model.

Slowed AChE degradation accompanied by decreased release

One potential pathway for loss of cellular AChE is release or secretion. To see if decreased release could account for the observed increase in half-life after Aβ-treatment, we measured the rate at which AChE activity appeared in culture media. Background activity was eliminated by transferring cells to serum-free Dulbecco's modified Eagle's medium, with or without Aβ, and exposing them briefly to echothiophate (see Methods). Immediately after this exposure, the medium contained no AChE activity, but modest amounts of activity appeared by 24 h (Fig. 6). Since cell viability remained high, this activity was taken to reflect direct cellular secretion. Additional activity emerged when the echothiophate-labeled AChE was unmasked with 2-PAM (see Methods); this activity was taken to reflect surface AChE shed into the medium after the introduction of echothiophate. According to this analysis, Αβ-treatment barely affected the direct secretion of AChE, but it markedly reduced shedding from the cell surface. In fact, slow-sedimenting forms of echothiophate-labeled AChE nearly disappeared after Αβ treatment (Fig. 6). But reduced release cannot explain the 36% fall in overall degradation rate, since the medium at 24 h contained less than 5% of the total cellular AChE activity.

Figure 6.

Effects of Aβ on surface AChE shedding. Echothiophate was used to label all surface AChE immediately before Aβ-treatment; 24 h later, medium was collected and frozen. Afterward, samples from six independent experiments were thawed, divided for additional treatment with or without 2-PAM reactivation, and finally pooled for analysis on sucrose density gradients. Open symbols represent samples without 2-PAM, and filled symbols represent samples after reactivation. Shown are representative sucrose density profiles.

Lysosomes and AChE degradation

Impaired lysosomal digestion would be a reasonable explanation for slowed degradation after Aβ-exposure. Since the role of lysosomes in AChE disposal was unknown, however, we examined AChE decay rates in the presence of agents that disrupt lysosomal function. Decay rates, especially during the first 4 h after cycloheximide, were slowed by chloroquine and leupeptin (Fig. 7a). Ammonium chloride, 5 mm, had a similar effect (not shown). Bafilomycin A1, a vacuolar H+-ATPase inhibitor with high specificity for lysosomes, worked too slowly to test in the cycloheximide paradigm; but given alone, it induced a dose-dependent rise in cell-specific AChE activity within 14 h (Fig. 7b). Thus, pharmacological data supported the view that lysosomes constitute a major pathway for AChE disposal.

Figure 7.

Effects of direct lysosomal inhibition on AChE degradation. (a) After 20 h incubation in serum-free medium, a lysosomotropic agent (chloroquine 1 µm) or protease inhibitor (leupeptin, 5 µm) was added simultaneously with cycloheximide (10 µg/mL). Shown are AChE activities as percentages of mean control value at zero time (mean ± SEM of nine independent determinations) in cell lysates prepared at the indicated times. (b) AChE activities (percentage of mean value from untreated control) measured 14 h after exposure to the indicated concentrations of bafilomycin A1 (mean ± SEM of six independent determinations). All treatment values were statistically different from control (p < 0.01) with the exception of chloroquine at 8 h and bafilomycin at 3.8 µm.

Aβ and lysosomes

In view of indications that AChE degradation may involve lysosomes, three approaches were taken to determine how a 24-h exposure to Aβ would affect this digestive organelle. First, we tested lysosomal distribution by immunocytochemistry for the marker protein, LAMP-1. However, no morphologic abnormality was apparent, and perinuclear LAMP-1 positive structures were similarly abundant in control and Aβ-treated N1E.115 cells (Figs 8a and b). Next, we tested lysosomal integrity with FITC-dextran. When dextran was presented for 2 h, the fluorescence label gathered in distinct punctate structures presumed to reflect lysosomes (Fig. 8c). When dextran was presented for a full 24 h, intense fluorescence from overlapping structures obscured details, but most of the label was still associated with organelles, even in cells exposed to Aβ (Fig. 8d). Hence the lysosomes were able to retain macromolecules. Finally we tested lysosomal acidity with the fluorescent, pH-sensitive dye, LysoSensor DN-189. Aβ-exposure caused a qualitative reduction in punctate perinuclear staining with this dye (Figs 8e–g). Because deprotonation quenches DN-189 fluorescence, these results point to a rise in lysosomal pH. Similar reductions in DN-189 fluorescence were also caused by 10 µm chloroquine (Fig. 8h), which is known for its effects on lysosomal pH (Homewood et al. 1972; Golde et al. 1992). The data thus indicate that Aβ did not destroy lysosomal integrity but did impair acidification and perhaps, therefore, proteolytic function.

Figure 8.

Effects of Aβ on lysosomes. Lysosomal localization and abundance were assessed by immunostaining for LAMP-1 after a 24-h treatment of cells with vehicle (a) or Aβ (b). Lysosomal membrane integrity was demonstrated by retention of fluorescent dextran captured during the last 2 h (c) of a 24-h exposure to Aβ, or during the entire treatment period (d). Lysosomal acidity was tested with Lysosensor DN189 added to cells exposed to vehicle (e) or Aβ (f). For comparison, LysoSensor labeling is also shown for untreated cells (g) and cells exposed to 10 µm chloroquine (h).

Aβ affects muscarinic receptors

With impaired lysosomal function as a likely toxic mechanism, many cell-surface proteins could be affected. We saw no change in total cellular protein content after Αβ-treatment, but to determine if the expression of particular membrane proteins was enhanced, we tested the abundance of muscarinic receptors on the N1E.115 cell surface. Αβ-treatment of these cells for 24 h caused more than a doubling in the amount of specific QNB-binding; a similar but smaller effect was seen in an N1E.115 line modified to express less AChE (Fig. 9).

Figure 9.

Effects of Aβ on surface muscarinic receptor density. QNB binding sites on the cell surface were assayed after 24-h treatment with Aβ or vehicle. Cells were exposed to 4 nm[3H]QNB, with and without atropine pre-adsorption, and the difference in retained radioactivity was taken as an index of specific binding to muscarinic receptors. Columns show mean ± SEM of six independent determinations of binding expressed as a percentage of the mean value from vehicle-treated controls for each cell line. Both Aβ-treated groups were statistically different from control, p < 0.01.


We have shown that Aβ-exposure enhances AChE expression in murine neuroblastoma cells, both internally and on the surface membrane. Although we do not exclude the possibility that Aβ may under some circumstances alter the substrate kinetics of AChE (Geula and Mesulam 1989), we saw no alteration in Km values, and we concluded that enzyme protein was increased. Our results are consistent with observations by Small's group (Sberna et al. 1997), who found a 2.5-fold elevation of AChE activity in embryonal carcinoma cells aggressively treated for 7 days with 10 µm Aβ 25–35. This effect appeared to depend on a rise in intracellular calcium, but the accompanying changes in AChE-turnover were not addressed. Our finding that AChE activity rises because of slower degradation, not faster synthesis, provides a new perspective and focuses attention on the pathways for AChE disposal.

Much of the available information on the cellular fate and disposal of AChE comes from work in avian myocytes. Rotundo and Fambrough (1980) first identified two catalytically active pools of AChE with different cellular fates in cultured chick embryo muscle cells. Significant amounts of soluble cytoplasmic enzyme, with an apparent half-life of 3–5 h, appeared to transfer directly from the trans-Golgi network for secretion into the surrounding medium. But the majority of the catalytically active AChE was membrane-bound and destined for expression on the cell surface, where its half-life was calculated at about 50 h. Rotundo found no evidence that cell-surface AChE was later shed into the medium, and he did not define the ultimate fate of this enzyme pool. Further studies in chick myocytes (Rotundo and Fambrough 1980; Rotundo 1988) and murine neuroblastoma (Lazar et al. 1984) revealed an inactive pool of AChE that was rapidly degraded, with a half-life of 1–2 h. This degradation was not affected by the lysosomotropic agent, ammonium chloride, or by the lysosomal peptidase inhibitor, leupeptin (Rotundo 1988). Hence the older literature gives good reason to conclude that lysosomes are not involved in this rapid turnover, for which, as recent work suggests, the proteasome is probably far more important. Broadly speaking, however, the specific pathways for turnover of catalytically active AChE remain unclear at present.

Our turnover studies in murine neuroblastoma cells extend previous information on the basic biology of AChE. One new feature is the appearance of echothiophate-labeled surface AChE in media. To the best of our knowledge, this is the first demonstration that some AChE is shed into the extracellular environment from the surface membranes of neuron-like cells. The minor amounts of enzyme involved would not readily be detected by isotopic labeling and could hardly affect overall turnover. Still, the process is of interest because Aβ had such a dramatic effect on the shedding of smaller AChE forms. We suggest that AChE shedding involves actions of proteases on the Proline Rich Membrane Anchor (PRiMA). In fact several extracellular matrix metalloproteases have ‘sheddase’ activities that serve to cleave proteins from the cell surface (Hooper et al. 1997; Sbarba and Rovida 2002). It is worth entertaining the possibility that Aβ interferes directly or indirectly with one or more such enzymes.

Since AChE shedding is only a minor pathway, its inhibition is unlikely to be significant for cholinergic function. The ability of Aβ to reduce overall AChE degradation, our central finding, could be much more important. We suggest that the reduced rate of AChE degradation reflects a disturbance of internalization or an impairment of lysosomal digestion, which might affect a wide range of membrane proteins. There is a clear need for direct studies of the endocytosis and digestion of AChE, muscarinic receptors, and other markers. Meanwhile, evidence is mounting that Aβ induces an array of abnormalities in protein trafficking or disposal. For example, it has been shown that Aβ associates with ApoE-containing beta-very-low-density lipoprotein to form a complex with an intracellular half-life four-times longer than that of the native counterpart (Scharnagl et al. 1999). Many studies point to lysosomes and protein turnover as key targets for Aβ toxicity. Thus, neuronally captured Aβ enters lysosomes, where it promotes the generation of free radicals, disrupts the proton gradient, and causes leakage (Bahr et al. 1998; Yang et al. 1998; Ditaranto et al. 2001; Ji et al. 2002). The outcome is an impaired digestion of polypeptides, probably including C-terminal fragments of APP (Estus et al. 1992; Golde et al. 1992). In short, there is a positive feedback loop with the potential to trigger a pathogenic cascade.

An unanswered question is whether lysosomal dysfunction can fully explain the increases in AChE activity and muscarinic receptor density in Aβ-treated neuroblastoma cells. Muscarinic receptors do accumulate in lysosomes upon short-term inhibition with ammonium chloride (Ray and Berman 1989), but it is difficult to see how this effect would increase the quantity of receptors on the cell surface. For AChE at least, elevation of surface expression must involve increased residence time. It is equally difficult to understand at present how such changes would be driven by elevations of intracellular calcium, as the data of Sberna et al. (1997) suggest. Still, we hold to the view that lysosomal impairment is a key to Aβ-toxicity and to the associated abnormalities in expression of AChE and muscarinic receptors.

Several studies with mouse models of AD show that Aβ can affect AChE expression not only in cell culture but also in the intact brain. Some transgenic animals expressing human Aβ, e.g. Tg2576 mice (Hsiao et al. 1995, 1996), show normal levels of brain AChE either by activity or immunoassay (W. Hu, unpublished data; Berdnar et al. 2002). However, mice that express human APP with the London mutation (V642I) exhibit selective elevation of AChE activity in the hippocampus and dentate gyrus (Bronfman et al. 2000). It may be significant that these mice develop higher levels of Aβ 42 than do other APP transgenics (Moechars et al. 1999). AChE is also altered in mice engineered to over-express the C-terminal fragment of APP (Sberna et al. 1998). Such mice are notable for outright neurodegenerative changes and ‘oddly shaped lysosomes’ (Oster-Granite et al. 1996). Along with those signs of lysosomal involvement, and in accord with our expectations, these mice have higher levels of neuronal AChE, especially the G1 form.

It is risky to extrapolate in vitro observations to patients with clinical AD. Nonetheless, it is becoming accepted that AD pathology involves lysosomal abnormalities that contribute to accumulation of amyloid in senile plaques (Cataldo et al. 1994, 1997). This concept fits our conclusion that Aβ-exposure affects AChE expression by impairing lysosomal digestion. Also, a subnormal AChE content in CSF and an increased G1 to G4 AChE ratio (Atack et al. 1983, 1988) are hints that AChE processing is disturbed in AD. On the other hand, the levels of active AChE and immunoreactive AChE protein in post-mortem AD brains are generally decreased, not increased (Hammond and Brimijoin 1988). That result makes sense, given the severe loss of cholinergic neurons in later stages of illness. The critical question is, what happens earlier on? AChE is elevated in brain tissue immediately surrounding the amyloid plaques (Arendt et al. 1992). Might this elevation mean that local neurons react to Aβ deposition by increasing AChE expression? If so a ‘vicious cycle’ could erupt, since AChE may further enhance the deposition of Aβ (Rees et al. 2003). More work is needed to elucidate the abnormalities of cellular functions in AD, especially the possibility of abnormal protein sequestration, and to determine if AChE can directly promote Aβ-toxicity or serves only as a passive bystander. Meanwhile, the interactions between Aβ and AChE may have implications for the design of therapeutic strategies in AD.