Galantamine-induced Amyloid-β Clearance Mediated via Stimulation of Microglial Nicotinic Acetylcholine Receptors*

Reduction of brain amyloid-β (Aβ) has been proposed as a therapeutic target for Alzheimer disease (AD), and microglial Aβ phagocytosis is noted as an Aβ clearance system in brains. Galantamine is an acetylcholinesterase inhibitor approved for symptomatic treatment of AD. Galantamine also acts as an allosterically potentiating ligand (APL) for nicotinic acetylcholine receptors (nAChRs). APL-binding site is located close to but distinct from that for acetylcholine on nAChRs, and FK1 antibody specifically binds to the APL-binding site without interfering with the acetylcholine-binding site. We found that in human AD brain, microglia accumulated on Aβ deposits and expressed α7 nAChRs including the APL-binding site recognized with FK1 antibody. Treatment of rat microglia with galantamine significantly enhanced microglial Aβ phagocytosis, and acetylcholine competitive antagonists as well as FK1 antibody inhibited the enhancement. Thus, the galantamine-enhanced microglial Aβ phagocytosis required the combined actions of an acetylcholine competitive agonist and the APL for nAChRs. Indeed, depletion of choline, an acetylcholine-competitive α7 nAChR agonist, from the culture medium impeded the enhancement. Similarly, Ca2+ depletion or inhibition of the calmodulin-dependent pathways for the actin reorganization abolished the enhancement. These results suggest that galantamine sensitizes microglial α7 nAChRs to choline and induces Ca2+ influx into microglia. The Ca2+-induced intracellular signaling cascades may then stimulate Aβ phagocytosis through the actin reorganization. We further demonstrated that galantamine treatment facilitated Aβ clearance in brains of rodent AD models. In conclusion, we propose a further advantage of galantamine in clinical AD treatment and microglial nAChRs as a new therapeutic target.

Alzheimer disease (AD) 4 is characterized by progressive cognitive impairment as a consequence of extensive neuronal loss (1). Pathological hallmarks of AD include the development of senile plaques by the accumulation of extracellular amyloid-␤ (A␤) and neurofibrillary tangles formed by intraneuronal accumulation of hyperphosphorylated tau. According to the amyloid hypothesis, accumulation of A␤ in the brain parenchyma is the primary event influencing other AD pathologies such as the formation of neurofibrillary tangles and neuronal cell death (2). Therefore, reduction of brain A␤ has been proposed as a primary therapeutic target for AD. In mouse models of AD, active immunization with A␤ peptides reduced brain A␤ and restored cognitive functions (3)(4)(5). One proposed mechanism of this A␤ reduction is the promotion of microglial A␤ phagocytosis via Fc receptors (6,7). A case report of an A␤ immunization clinical trial also suggested the contribution of microglia in A␤ clearance (8).
Deficits of cholinergic function have been demonstrated in AD brain (9 -12). These findings provide the rationale for cholinergic enhancement as an approach to improve cognitive function in AD, and acetylcholinesterase (AChE) inhibitors are currently approved for AD treatment. Their therapeutic benefits are commonly regarded as purely symptomatic, and they are considered only a partial answer for AD treatment. However, most of them exhibit effects that are inconsistent with purely symptomatic actions, and it has been suggested that some AChE inhibitors induce molecular and cellular changes that directly influence AD pathologies.
Galantamine is a centrally acting AChE inhibitor. Although its AChE inhibitory activity appears to be much weaker than other clinically available AChE inhibitors, its therapeutic effects on cognitive function in AD are comparable with the other agents (13)(14)(15). Furthermore, the latest long term ex-tension clinical trial of galantamine suggested that the cognitive benefits of galantamine are sustained for at least 36 months (16). Several lines of evidence suggest that the therapeutic effects of galantamine are mediated by additional mechanisms apart from AChE inhibition. Galantamine actually acts as an allosterically potentiating ligand (APL) for nicotinic acetylcholine receptors (nAChRs) (17)(18)(19)(20) and exhibits neuroprotective effects in vitro (21) and in vivo (22).
A recent study has shown that microglia express the ␣7 subunit of nAChRs (␣7 nAChRs) (23) and that treatment with nicotine modulates secretion of tumor necrosis factor from microglia, resulting in neuroprotection (24). It therefore seems feasible that galantamine acts as an APL on microglial nAChRs to modulate microglial function. In the present study, we examined the effect of galantamine on microglial A␤ phagocytosis.
Human Materials and Preparation of Brain Sections-All of the experiments using human materials were performed in accordance with the guidelines of Kyoto Pharmaceutical University Ethical Committee. Informed consent was obtained from the subject. The tissues of frontal cortices from a patient who was clinically and histopathologically diagnosed as having AD (67 years of age; Consortium to Establish a Registry for Alzheimer's Disease criteria C; Braak stage IV) were used. The dissected tissue blocks were fixed in 10% formalin and then were cut into 20-m sections on a cryostat.
Rat Microglial Culture-Primary-cultured rat microglia were prepared as previously described (7,27,28). Forebrains from newborn Wistar rats were minced and filtrated through a 70-m-diameter nylon mesh. The cells were cultured in DMEM with 10% fetal bovine serum at 37°C in humidified 5% CO 2 , 95% air. At 21 days, floating microglia were isolated from the cell cultures. Purified microglia were treated with A␤42 alone or simultaneously with galantamine and/or nicotine. Antagonists or inhibitors were added individually to microglia 10 min prior to treatment with A␤42.
Rodent Models of AD and Drug Treatment-All of the animal experiments were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the protocols were approved by the Committee for Animal Research at Kyoto Pharmaceutical University. The rats were anesthetized with sodium pentobarbital, and 1 g of human A␤42 was injected into the hippocampus as previously described (29). Starting the day after the A␤42 injection, the rats received 1 or 5 mg/kg of galantamine or 1 ml/kg of PBS intraperitoneally once daily for 2 weeks. On the day after the final drug treatment, the hippocampus was dissected and was prepared for ELISA as described below.
Hemizygous APdE9 mice expressing chimeric mouse/human amyloid precursor protein APPswe (mouse APP695 harboring a human A␤ domain and mutations K594N and M595L linked to Swedish familial AD pedigrees) and human mutated presenilin 1-dE9 (deletion of exon 9) were purchased from Jackson Laboratory (30). Starting at 9 months of age, groups of APdE9 mice were orally administrated 1 mg/kg (n ϭ 7) or 5 mg/kg (n ϭ 7) of galantamine or 10 ml/kg of PBS (n ϭ 7) once daily for 2 months (56 days). Groups of age-matched nontransgenic littermates received 5 mg/kg of galantamine (n ϭ 7) or 10 ml/kg of PBS (n ϭ 7). On the day after the final drug treatment, the mice were killed by cervical dislocation, and the brains were removed. One cerebral hemisphere was prepared for ELISA as described below, and the other was post-fixed in 4% paraformaldehyde for histochemical analysis.
Time Lapse Imaging Assay-In the time lapse imaging assay for Ca 2ϩ , primary-cultured rat microglia were incubated with Fluo 4-AM in Ca 2ϩ (Ϫ) DMEM, and then the culture medium was replaced with Ca 2ϩ (Ϫ) DMEM supplemented with 1.8 mM Ca 2ϩ (Ca 2ϩ (ϩ) DMEM). Fluorescence of Fluo 4 was monitored at a rate of 1 scan/s using LSM510META. In the middle of the laser scanning, a drop of concentrated galantamine or nicotine was added to obtain a final concentration of 1 M galantamine or 1 mM nicotine, respectively. Subsequently, the images were reconstructed into supplemental videos using the software iMovie (Apple Inc.).
Biochemical Analysis-Western blot analysis was performed as previously described (7,27,28) using anti-␣7 nAChR antibody (1:1,000) or FK1 antibody (1:5,000). The amounts of A␤ phagocytosed by rat microglia and those in the brains of A␤-injected rats and APdE9 mice were measured by ELISA. Briefly, at 12 h after the treatment of A␤42 to rat microglia, culture medium was collected. The rat microglia were rinsed with PBS and then lysed with PBS containing 0.1% Triton X-100. The brains of A␤-injected rats and APdE9 mice were homogenized with Tris-buffered saline (TBS). The homogenate was centrifuged at 100,000 ϫ g for 1 h, and the supernatant was collected as the TBS-extracted fraction. 70% formic acid (FA) was added to the pellet, which was homogenized again. The homogenate was incubated for 1 h at 4°C and then centrifuged at 100,000 ϫ g for 1 h at 4°C. The resultant supernatant was collected as the FA-extracted fraction, which was neutralized with a 20-fold volume of 1 M Tris buffer (pH 11.0). Subsequently, the amounts of A␤ in the culture medium, cell lysate, TBS-extracted fraction, and neutralized FA-extracted fraction were measured using a human A␤42 or A␤40 specific ELISA kit (IBL) according to the manufacturer's instructions.
RT-PCR-Total microglial RNA was extracted using the Isogene RNA isolation kit (Nippon Gene), and first strand cDNA was synthesized using a RT-PCR kit (GE Healthcare) according to the manufacturer's instructions. Second strand cDNA synthesis/PCR amplification was performed with ␣7 nAChR subunit-specific primer set (forward, 5Ј-TTT CTG CGC ATG AAG AGG CCC GGA GAT-3Ј and reverse, 5Ј-ACC TCC TCC AGG ATC TT-3Ј). PCR products were electrophoresed on 2% agarose gels and stained with ethidium bromide.
Water Maze Testing-Morris water maze (MWM) examination was performed in a circular pool (diameter, 120 cm; height, 30 cm). The pool was filled with water (maintained at 25.0 Ϯ 1.0°C) and was located in a large room with various distal visual cues. An invisible (transparent) circular platform (diameter, 10 cm) was submerged ϳ1.0 cm below the surface of the water and placed in the center of the south quadrant. The swimming activity of each mouse was monitored via a video camera mounted overhead, which relayed information including latency to find the platform, total distance traveled, and time and distance spent in each quadrant to a video tracking system (Muromachi). Each mouse was given three acquisition trials/day for 3 consecutive days to locate and climb on to the hidden platform. An acquisition trial was initiated by placing the mouse in the water with its nose directly facing the pool wall. The starting position varied between five constant locations at the pool rim. For each trial, the mouse was allowed to swim a maximum of 120 s to find the platform. When successful, the mouse was allowed a 30-s rest period on the platform. If unsuccessful within the allotted time period, the mouse was given a score of 120 s and then physically placed on the platform and also allowed the 30-s rest period. One day following the last acquisition trial, a single probe test was conducted for each study subject to measure spatial bias for previous platform location. This was accomplished by removing the platform from the pool and measuring the time spent in the previous platform quadrant location for 100 s. Starting 4 days and 3 days before the first and final drug treatment, respectively, each mouse was given two sets of the MWM examination.
Statistical Evaluation-All of the data represent the means Ϯ S.E. In the ELISA, the statistical significance of differences among groups was determined by an analysis of variance with a Bonferroni/Dunn test. In the MWM, the mean sources during acquisition trials were analyzed using two-way repeated measures analysis of variance, with genotype and the test day of acquisition trials or treatment group and the test day of acquisition trials as sources of variation. As the multiple comparison among APdE9 mice treated with vehicle and galantamine (1 and 5 mg/kg), Fisher's protected least significant difference (Fisher's PLSD) test was used to examine post hoc differences. In probe tests, the search bias in the target quadrant was compared between vehicle-treated nontransgenic littermates and vehicle-treated APdE9 mice by using Student's t test. The search bias among APdE9 mice treated with vehicle and galantamine (1 and 5 mg/kg) was evaluated by an analysis of variance with a Bonferroni/Dunn test.

Expression of ␣7 nAChRs in Human AD Brain and Rat
Microglia-We first analyzed the expression of nAChRs on microglia in human AD brain. Microglia labeled with anti-Iba1 antibody (Fig. 1B) accumulated on the A␤ plaque probed with FSB ( Fig. 1A), an A␤-binding fluorescent dye. The immunoreactivities of ␣7 nAChRs (Fig. 1C) were co-localized with the microglia (Fig. 1D, arrows).
We next prepared primary cultures of rat microglia, which expressed the mRNA of ␣7 nAChRs (Fig. 1E). In Western blot analysis, we also detected the protein of ␣7 nAChRs in rat microglia with primary cultures of rat neuron (21) and rat liver homogenate (31) as positive and negative controls, respectively ( Fig. 1F). In laser confocal microscopic analysis, immunoreactivity of ␣7 nAChRs (Fig. 1H) was detected in rod-shaped rat microglia (Fig. 1G), and the merged image ( Fig. 1I) showed that they were strongly expressed on the microglial cell surface.
Enhancement of Rat Microglial A␤ Phagocytosis by Galantamine-In previous studies, we reported that primarycultured rat microglia markedly phagocytose A␤ (7,27,28). In the present study, we treated rat microglia with 1 M A␤42 in the presence or absence of galantamine or nicotine. The amount of A␤ phagocytosed by microglia was then measured by ELISA at 12 h after treatment. Galantamine enhanced microglial A␤ phagocytosis in a concentration-dependent manner, and phagocytosis reached the maximum at 1 M galantamine ( Fig. 2A). Similarly, nicotine, a classical acetylcholine competitive agonist, increased microglial A␤ phagocytosis, and the phagocytosis was significantly enhanced by treatment with 1,000 M nicotine (Fig. 2B).
Microglia were next pretreated with acetylcholine competitive nAChR antagonists. Pretreatment with mecamylamine (a nonselective nAChR antagonist) or MLA (an ␣7 nAChR-specific antagonist) prevented the galantamine-enhanced ( Fig.  2C) or nicotine-enhanced ( Fig. 2D) microglial A␤ phagocytosis. In contrast, atropine, a muscarinic AChR antagonist, did not inhibit the enhanced phagocytosis (Fig. 2, C and D). Confocal microscopic analysis supported this result (Fig. 2E). In the presence of A␤, microglia changed from rod-shaped ( Fig.  1G) to round amoeboid-shaped (Fig. 2E, red) morphology and markedly phagocytosed A␤ (Fig. 2E, green). Treatment with galantamine increased the intensity of A␤ inside microglia, and this increase was inhibited by pretreatment with mecamylamine or MLA, but not with atropine. We subsequently measured the amount of extracellular A␤ in microglial culture medium (Fig. 2F). Extracellular A␤ amounts were significantly reduced by simultaneous treatment with galantamine plus A␤ or nicotine plus A␤, and the reductions were inhibited by pretreatment with mecamylamine or MLA, but not with atropine. Thus, we found that microglial A␤ phagocytosis is enhanced by galantamine or nicotine, at least in part through ␣7 nAChRs.
Expression of APL-binding Site for nAChRs in Human AD Brain and Rat Microglia-Galantamine acts as an APL for nAChRs (17)(18)(19)(20). The APL-binding site is located close to but distinct from that for acetylcholine on nAChRs, and FK1 anti-FIGURE 1. ␣7 nAChRs in human AD brain and on primary-cultured rat microglia. A-D, in the AD brain, immunoreactivities of ␣7 nAChRs (C, green) were co-localized (D, arrows) with microglia (B, Iba1; red) accumulated on A␤ plaque (A, FSB; blue). Scale bar, 20 m. E, RT-PCR demonstrated the expression of mRNA of ␣7 nAChRs in rat microglia. F, in Western blot analysis, proteins of ␣7 nAChRs were detected in rat microglia and neurons but not in rat liver. ␣7, ␣7 nAChRs. G-I, rhodamine-conjugated phalloidin, a fluorescent probe for actin cytoskeleton, revealed the rod-shaped rat microglia (G, red). ␣7 nAChR (H, green) was expressed on the microglia (I). In I, the blue area indicates microglial nucleus labeled with Hoechst 33258. Scale bar, 20 m. body specifically binds to the APL-binding site without interfering with the acetylcholine-binding site (25,26). Therefore, we next analyzed expression of the APL-binding site for nAChRs using the FK1 antibody in human AD brain. The A␤ plaque (Fig. 3A) and activated microglia (Fig. 3B) were probed with FSB and anti-Iba1 antibody, respectively. Immunoreactivity of APL-binding sites appeared as dot-like areas throughout the AD brain parenchyma (Fig. 3C), and areas of intense immunoreactivity were also observed (Fig. 3C, arrows). A merged image (Fig. 3D) revealed that areas with strong immunoreactivity to FK1 antibody were co-localized with the microglia accumulated on the A␤ plaque in human AD brain (Fig. 3D, arrows).
In Western blot analysis using FK1 antibody, we detected expression of the APL-binding site for nAChRs in primary cultures of rat microglia and neurons, but not in rat liver (Fig.  3E). We further confirmed the co-localization of ␣7 nAChRs and APL-binding sites in the rod-shaped untreated microglia (Fig. 3, F-H) and amoeboid-shaped A␤-treated microglia (Fig.  3, I-K) by laser confocal microscopy.
APL-binding Site for nAChRs in Galantamine-enhanced A␤ Phagocytosis-Using FK1 antibody, we examined the involvement of the APL-binding site in galantamine-or nicotineenhanced microglial A␤ phagocytosis (Fig. 4). Because FK1 antibody is a mouse IgM antibody, the treatment with mouse IgM isotype control (Sigma) instead of FK1 antibody was included as a negative control. Although galantamine-enhanced microglial A␤ phagocytosis was significantly inhibited by FK1 antibody (1:100, ϳ1 g/ml IgM), FK1 antibody alone did not influence the magnitude of A␤ phagocytosis in the absence of galantamine. Mouse IgM isotype control had no inhibitory effects on galantamine-enhanced A␤ phagocytosis or on A␤ phagocytosis in the absence of galantamine. Thus, FK1 anti-  body abolished galantamine-enhanced microglial A␤ phagocytosis. However, treatment with FK1 antibody (1:100) did not inhibit nicotine-enhanced A␤ phagocytosis. These results suggest that galantamine enhances microglial A␤ phagocytosis by acting on the APL-binding site for nAChRs, whereas nicotine increases phagocytosis by directly interacting with the acetylcholine-binding site.
Requirement of Choline for Galantamine-enhanced A␤ Phagocytosis-APLs interact with a binding site on nAChRs that is distinct from the acetylcholine-binding site, and from there APLs sensitize nAChRs to respond to acetylcholine or acetylcholine competitive agonists (19,32). Thus, APLs act as modulatory ligands for nAChRs rather than as agonists. In this context, galantamine may need acetylcholine or some acetylcholine competitive agonists to induce activation of microglial nAChRs. In fact, MLA, the acetylcholine competitive antagonist for ␣7 nAChRs, inhibited galantamine-enhanced microglial A␤ phagocytosis (Fig. 2C). Therefore, we checked whether our microglial culture medium contained acetylcholine or acetylcholine competitive agonists and found 28 M choline. Choline functions as a relatively selective and competitive agonist for the acetylcholine-binding site on ␣7 nAChRs (33)(34)(35). Therefore, we next examined how choline affected microglial A␤ phagocytosis in the presence or absence of galantamine (Fig. 5, A and B). In the absence of galantamine, the levels of microglial A␤ phagocytosis were almost the same in choline (Ϫ) DMEM and choline (Ϫ) DMEM supplemented with 28 M choline [choline (ϩ) DMEM] (Fig.  5A). However, enhanced A␤ phagocytosis induced by galantamine in choline (ϩ) DMEM was abolished in choline (Ϫ) DMEM. In contrast, microglial A␤ phagocytosis was enhanced by nicotine in a choline-independent manner.
We further examined concentration-dependent effects of choline and nicotine on microglial A␤ phagocytosis (Fig. 5, B and C, respectively). In the absence of galantamine, choline and nicotine (both 100 M-1 mM) increased A␤ phagocytosis in a concentration-dependent manner, and the response peaked at 1 mM choline and nicotine. In contrast, in the presence of galantamine, phagocytosis was increased already at 1 M choline or 0.03 M nicotine and peaked at 10 M choline or 1 M nicotine. These results suggest that choline acts as an acetylcholine competitive agonist for microglial nAChRs and enhances A␤ phagocytosis. Thus, we found that galantamine dramatically sensitizes microglial nAChRs to choline and nicotine.
Ca 2ϩ Influx and Calmodulin-dependent Pathway in Galantamine-enhanced A␤ Phagocytosis-The ␣7 nAChRs are known to function as ligand-gated Ca 2ϩ channels (36). To test for this activity in microglia, a time lapse Ca 2ϩ imaging study using Fluo 4-AM, a Ca 2ϩ -sensitive fluorescent dye, was conducted in the choline (ϩ) DMEM ( Fig. 6A and supplemental Videos S1-S3). Faint Fluo 4 fluorescence was observed in resting rat microglia before drug treatment. Vehicle treatment did not influence the intensity of Fluo 4 fluorescence (Fig. 6A, top panels, and supplemental Video S1). On the other hand, Fluo 4 fluorescence was increased transiently in the microglial cytoplasm ϳ30 s after the addition of galantamine (Fig. 6A, middle panels, and supplemental Video S2) or nicotine (Fig.   6A, bottom panels, and supplemental Video S3) and returned to the basal level within ϳ1 min. These results suggest that nAChRs in microglia indeed function as Ca 2ϩ channels.
We next investigated whether the Ca 2ϩ influx correlates with enhanced microglial A␤ phagocytosis. In this assay, we used Ca 2ϩ (Ϫ) DMEM containing 28 M choline and cap- tured any remaining Ca 2ϩ by adding 0.1 mM EGTA. In the absence of galantamine or nicotine, the levels of microglial A␤ phagocytosis were almost the same in Ca 2ϩ (Ϫ) DMEM and Ca 2ϩ (ϩ) DMEM. Although galantamine (Fig. 6B) and nicotine (Fig. 6C) significantly enhanced microglial A␤ phagocytosis in Ca 2ϩ (ϩ) DMEM, the effect was not observed in Ca 2ϩ (Ϫ) DMEM. Furthermore, 20 M W-7 (an inhibitor of calmodulin (CaM)) and 10 M KN93 (an inhibitor of Ca 2ϩ / CaM-dependent protein kinase II (CaMKII)) significantly inhibited galantamine-enhanced microglial A␤ phagocytosis (Fig. 6D). This result suggests that Ca 2ϩ influx through nAChRs and subsequent activation of CaM-CaMKII signaling cascade are involved in enhanced microglial A␤ phagocytosis.
In platelet, CaM directly binds to Ras-related C3 botulinum toxin substrate 1 (Rac1) in a Ca 2ϩ -dependent manner and is involved in Rac1 activation (37). Rac1 is a Rho (Ras homologous) family small GTPase and regulates actin reorganization (38). We next examined the effects of Toxin B (a broad inhibitor of Rho family small GTPases including Rac1) and NSC23766 (a Rac1-specific inhibitor) in Ca 2ϩ (ϩ) DMEM containing 28 M choline. Pretreatment with 5 ng/ml Toxin B and 100 M NSC23766 significantly inhibited galantamineenhanced microglial A␤ phagocytosis (Fig. 6E). This result suggests that Ca 2ϩ -dependent CaM-Rac1 signaling cascade is also involved in galantamine-enhanced microglial A␤ phagocytosis.
Cell signaling cascades such as Janus activated kinase-2 (JAK2)-phosphoinositide-3 kinase (PI3K)-Akt cascade and Fyn-PI3K-Akt cascade have been reported to be involved in neuroprotection conferred by nAChR stimulation in neurons (21,39). In addition, Fyn generally activates mitogen-activated protein kinase kinase (MAPKK). Consequently, we further investigated whether these cell signaling cascades are involved in galantamine-enhanced microglial A␤ phagocytosis. Microglia were treated with AG490, PP2, LY294002, and PD98059 (inhibitors of JAK2, Fyn, PI3K, and MAPKK, respectively) in Ca 2ϩ (ϩ) DMEM containing 28 M choline, but they did not inhibit galantamine-enhanced A␤ phagocytosis (Fig. 6F). These results suggest that rather than JAK2, Fyn, PI3K, or MAPKK, the activation of CaM, CaMKII, and Rac1 by Ca 2ϩ influx plays a crucial role in the cell signaling cascades linked to the promotion of microglial A␤ phagocytosis induced by nAChR stimulation.
Effect of Galantamine on A␤ Clearance in A␤-injected Rat Brain-Subsequently, we investigated the effect of galantamine on in vivo A␤ clearance in the rat brain. We first mi- croinjected 1 g of human A␤42 into the rat hippocampus (ϳ100 mg of wet weight). It is theoretically estimated that 2222 pmol/g of tissue of human A␤42 existed in the hippocampus immediately after the microinjection. A previous study measured endogenous levels of rat A␤40 and A␤42 produced in brains of wild-type rats by ELISA and estimated the concentrations to be ϳ50 and 13 pmol/g of tissue, respectively (40). It suggests that the endogenous levels of rat A␤40 and A␤42 might be ϳ2.3 and 0.6% of the exogenously microinjected human A␤42, respectively. In addition, we used a human A␤42-specific ELISA kit for the measurement of exogenously microinjected human A␤42 (cross-reactivity: rat A␤40 Յ 0.1%, rat A␤42 Յ 0.1%). Therefore, we considered that the influence of endogenous A␤ was negligible in the measurement of the exogenously microinjected human A␤42. In the next step, rats that had received intrahippocampal A␤42 injection received galantamine (1 or 5 mg/kg) intraperitoneally once daily for 14 consecutive days. As shown in Fig.  7, the amount of A␤ was significantly reduced by galantamine in a dose-dependent manner, compared with the vehicletreated rats. We previously analyzed that in the A␤-injected rat brain, microglial A␤ phagocytosis has been proposed to be a major mechanism of A␤ clearance (29). Therefore, this result suggests that galantamine may enhance microglial A␤ phagocytosis and then promote the A␤ clearance from the A␤-injected rat brain.
Effect of Galantamine on Spatial Learning and Memory and A␤ Clearance in APdE9 Mice-Before the drug treatment, we tested 21 APdE9 mice and 14 nontransgenic littermates in MWM at 9 months of age. In the acquisition trials, no difference was found in swimming speed (F 1,66 ϭ 0.32, p ϭ 0.5744). In the escape latency in the acquisition trials, although it was not significant, a weak genotype effect was observed (F 1,66 ϭ 3.13, p ϭ 0.0863; Fig. 8A). In the probe test, APdE9 mice showed significantly less retention time in the former platform location (target quadrant) than nontransgenic littermates (Fig. 8C). These results indicate that APdE9 mice displayed a weak decline in spatial learning and memory at 9 months of age.
Starting at 9 months of age, groups of APdE9 mice were orally administrated 1 or 5 mg/kg of galantamine or 10 ml/kg of PBS once daily for 56 days. Groups of age-matched nontransgenic littermates received 5 mg/kg of galantamine or 10 ml/kg of PBS. Three days before the final drug treatment (at 11 months of age), a second block of acquisition trials and a single probe test were started. In the acquisition trials, no difference was found in swimming speed among all groups (F 4,60 ϭ 1.59, p ϭ 0.2017). In the escape latency between vehicle-treated APdE9 mice and vehicle-treated nontransgenic littermates, there was a significant genotype effect (F 1,24 ϭ 16.83, p ϭ 0.0015; Fig. 8B) in the acquisition trials. In the probe test, vehicle-treated APdE9 mice predictably showed significantly less retention time in the target quadrant than vehicle-treated nontransgenic littermates (Fig. 8D). Thus, vehicle-treated APdE9 mice displayed a significant decline in spatial learning and memory at 11 months of age. In the comparison among APdE9 mice, a significant drug effect was detected in the escape latency (F 2,36 ϭ 9.37, p ϭ 0.0016; Fig. 8B) in the acquisition trials. Post hoc analysis revealed that in the acquisition trials, both low and high doses of galantamine significantly improved the performance in the 11-month-old APdE9 mice. Similarly, in the probe test, galantamine treatment increased retention time, and the performance was significantly improved by treatments of low and high doses of galantamine (Fig. 8D). The typical swimming trajectories of the 11-month-old mice are shown in Fig. 8E. Thus, galantamine treatment improved behavioral performance on the spatial learning and memory in aged APdE9 mice.
After the MWM experiment, we finally examined the A␤ burden in the brains of APdE9 mice. In the cortices and hippocampi of vehicle-treated APdE9 mice, A␤ plaques were markedly formed (Fig. 9A). Moderate accumulations of A␤ were detected in APdE9 mice treated with 5 mg/kg galantamine (Fig. 9B). Microglia accumulated on the A␤ plaques in the vehicle-treated (Fig. 9, C and E) and galantamine-treated (Fig. 9, D and F) APdE9 mice were expressed in ␣7 nAChRs (Fig. 9, C and D) and the APL-binding site (Fig. 9, E and F). The amounts of A␤40 (Fig. 9G) and A␤42 (Fig. 9H) in brains of APdE9 mice were measured by ELISA. Although the amounts of soluble A␤40 and A␤42 in TBS-extracted fractions were almost the same between the drug treatments, those of insoluble A␤40 and A␤42 in FA-extracted fractions were significantly decreased by the treatment of galantamine. Thus, in two different models of AD, such as A␤-injected rats and transgenic mice, galantamine promoted A␤ clearance in their brains.

DISCUSSION
Galantamine has been reported to have multiple actions on neurons. One of the most interesting effects is exerted by AChE inhibition, resulting in enhanced cholinergic neurotransmission caused by the increased acetylcholine level. Another effect is the neuroprotection exerted through allosteric modulation of nAChRs on neurons (21). In the present study, we demonstrated for the first time that galantamine allosterically modulates microglial nAChRs and enhances microglial A␤ phagocytosis. This effect may also contribute to neuropro- tection because neurotoxicity of A␤ has been demonstrated (41,42), and A␤-dependent formation (43,44) and/or maturation (45) of neurofibrillary tangles have been suggested. Therefore, the effect of galantamine on microglia could be related to the long term cognitive benefits of galantamine, as reported in AD patients (16). Thus, we propose a further advantage of galantamine as a therapeutic drug for AD.
In the galantamine-enhanced microglial A␤ phagocytosis, we found that at least two sites on nAChRs, the acetylcholinebinding site (Fig. 2C) and the APL-binding site (Fig. 4), contribute. In other words, galantamine may require some natural or acetylcholine competitive agonists to enhance microglial A␤ phagocytosis. Indeed, galantamine utilized choline, an acetylcholine competitive and relatively selective full agonist for ␣7 nAChRs (33)(34)(35), to enhance A␤ phagocytosis (Fig. 5A). Furthermore, the concentration-dependent curves of choline and nicotine on microglial A␤ phagocytosis (Fig. 5, B and C, respectively) revealed that in the absence of galantamine, choline and nicotine at concentrations of ϳ100 M and higher enhanced A␤ phagocytosis. On the other hand, 1 M galantamine dramatically increased choline or nicotine sensitivity of microglial nAChRs, and 1 M choline or 0.03 M nicotine was enough to enhance microglial A␤ phagocytosis.
Although further studies are required to clarify the reason why nicotine was more sensitive than choline, galantamine dramatically sensitized microglial nAChRs to choline and nicotine. A previous study measured extracellular choline in the mouse dorsal hippocampus by microdialysis and estimated the concentration to be 4.36 M (46). These findings suggest that choline alone at a physiological concentration is unlikely to activate nAChRs on microglia in the brain. However, in the presence of galantamine, such a low concentration of choline may suffice to activate nAChRs on microglia and consequently increase microglial A␤ phagocytosis in the brain.
The ␣7 nAChRs are ion channels with high Ca 2ϩ permeability that exhibit a low affinity for nicotine and are desensitized rapidly in neurons (36). In the present study, the influx of extracellular Ca 2ϩ (Fig. 6, A-C, and supplemental Videos S1-S3) and subsequent activation of the CaM-CaMKII and CaM-Rac1 pathways (Fig. 6, D and E, respectively) were involved in promoting microglial A␤ phagocytosis. However, Suzuki et al. (24) have reported that microglial ␣7 nAChRs drive a phospholipase C-inositol trisphosphate pathway and elicit Ca 2ϩ release from an intracellular Ca 2ϩ store. Therefore, the possibility remains that intracellular Ca 2ϩ is involved in the activation of the FIGURE 8. Galantamine improved spatial learning and memory in APdE9 mice. A and B, in MWM examination, APdE9 mice and wild-type littermates were given three acquisition trials/day for three consecutive days at 4 days before the drug treatments (A) and at 3 days before the final administrations (B). Wild-type littermates (n ϭ 14) were divided into a vehicle-treated group (n ϭ 7) and a 5 mg/kg galantamine-treated group (n ϭ 7). APdE9 mice were divided into a vehicle-treated group (n ϭ 7), a 1 mg/kg of galantamine-treated group (n ϭ 7), and a 5 mg/kg of galantamine-treated group (n ϭ 7). **, p Ͻ 0.01 versus vehicle-treated wild-type mice. † †, p Ͻ 0.01; † † †, p Ͻ 0.001 versus vehicle-treated APdE9 mice. Gal, galantamine. C and D, 1 day after the last acquisition trial, a single probe test was conducted for each study subject to measure spatial bias for previous platform location at 4 days before the drug treatments (C) and at 3 days before the final administrations (D). *, p Ͻ 0.05 versus wild-type mice (C) and vehicle-treated wild-type mice (D). †, p Ͻ 0.05; † †, p Ͻ 0.01 versus vehicle-treated APdE9 mice. E, the red area indicates the target quadrant, and typical swimming trajectories in the last probe test are shown. In each circle, the black dot indicates the starting point, and the red dot shows the final position of the mouse at the end of the probe test.
In neurons, CaMKII is an abundant synaptic protein that has been shown to regulate bundling of actin filaments leading to structural modifications of synapses (47). At resting synapse, actin filaments are bundled by the nonactivated CaMKII. When CaMKII is activated by neuronal activity and the resultant Ca 2ϩ influx, CaMKII detaches from actin filaments and allows actin reorganization by small GTPases. In this study, we also revealed the involvement of Rac1 activity in the galantamine-enhanced microglial A␤ phagocytosis. Rac1 is a small GTPase and regulates actin assembly through the activation of Wiskott-Aldrich syndrome protein family verprolin-homologous protein (WAVE) (48). In addition, we suggested previously that actin reorganization regulated by WAVE participates predominantly in microglial A␤ phagocytosis (28). Based on the evidence presented, we postulate that as downstream mechanisms of Ca 2ϩ influx through ␣7 nAChRs, the activation pathways for the regulation of actin reorganization, such as CaM-CaMKII and CaM-Rac1-WAVE, are critically and simultaneously involved in the galantamineenhanced microglial A␤ phagocytosis.
In the present study, we were able to clearly demonstrate an increased clearance of A␤ in the brain of A␤-injected rat by subchronic (14 days) intraperitoneal administration of galantamine (1 and 5 mg/kg) (Fig. 7). Recent in vivo studies using Tg2576 transgenic mice, an animal AD model that develops A␤ plaques in the brain, have revealed that nicotine reduces the brain A␤ level (49,50). These findings support our hypothesis that modulation of ␣7 nAChRs by galantamine effectively increased A␤ clearance in A␤-injected rat brain. However, Unger et al. (51) reported that subchronic subcutaneous treatment with galantamine (2 mg/kg) in Tg2576 mice had no effect on brain A␤ levels. It is possible that the duration of galantamine treatment (10 days) was not long enough to detect a difference in brain A␤ levels, because a massive amount of A␤ is continuously produced in aged Tg2576 mice. Therefore, we further investigated whether the chronic (56 days) administration of galantamine to a transgenic mouse model of AD (APdE9 mouse) is effective on the A␤ clearance in the brain. Indeed, the chronic treatment of galantamine improved the impaired spatial learning and memory in aged APdE9 mice (Fig. 8) and significantly reduced amounts of A␤ in FA-extracted fractions prepared from the brains of APdE9 mice (Fig. 9). These data suggest that the promotion of A␤ clearance in the brains of A␤-injected rats and APdE9 mice may be induced by galantamine treatment through the enhancement of microglial A␤ phagocytosis as shown in the study of primary-cultured rat microglia ( Fig. 2A). We further speculate that the long term clinical treatment of galantamine to patients with AD may also give rise to a promotional effect on the A␤ clearance through the microglial A␤ phagocytosis, and this effect of galantamine on microglia may, at least partially, contribute to its long lasting cognitive benefits revealed in the latest clinical trial (16).
A recent long term follow-up study of patients with AD who were immunized with A␤42 peptide (AN1792) suggested that the immunization is associated with a reduction in fibrillar A␤ plaques (52). Unfortunately, this study could not detect the inhibitory effect on the time to severe dementia in the AN1792 group compared with the placebo group. Thus, this study also suggested a necessity of the immunization as early as possible. In addition, anti-A␤ antibody produced by the immunization with A␤ peptides induced not only microglial A␤ phagocytosis and drainage of brain A␤ into systemic circulation but also disaggregation of fibrillar A␤ (53). Therefore, another possible explanation for follow-up study results is that immunization failed to reduce the concentration of oligomeric A␤, a synaptotoxic form of A␤ (54), and might even have increased it during the active phase of disaggregation of fibrillar A␤ plaques (55). Microglia phagocytose and degrade oligomeric A␤ (56). In the present study, although we could not detect a significant reduction of A␤ in the TBSextracted fractions, the amounts of A␤ were not increased by the treatment of galantamine to APdE9 mice. Therefore, enhanced microglial A␤ phagocytosis by galantamine may exert A␤ clearance from brains barring diffusion of toxic oligomeric A␤.
In conclusion, we have provided new evidence that modulation of microglial nAChRs by galantamine or stimulation of nAChRs by nicotine enhances A␤ phagocytosis in primarycultured rat microglia. Galantamine sensitizes microglial nAChRs to choline and nicotine by binding to the APL-binding site on nAChRs. Thus, galantamine requires extracellular choline or other acetylcholine competitive agonists to enhance microglial A␤ phagocytosis. Nicotine alone directly induces enhanced microglial A␤ phagocytosis. Furthermore, we also suggest that the Ca 2ϩ signaling cascade followed by CaM-CaMKII and CaM-Rac1 pathways for the regulation of actin reorganization may be involved in the phagocytic enhancement in microglia. Subsequently, we postulate that this mechanism may also be involved in galantamine-enhanced A␤ clearance observed in the brains of A␤-injected rats and APdE9 mice. All of the results presented here suggest a further advantage of galantamine as a therapeutic drug for AD and significant therapeutic potential of microglial nAChRs, such as ␣7 nAChRs, in the treatment of AD.