Endocannabinoids Prevent β-Amyloid-mediated Lysosomal Destabilization in Cultured Neurons*

Neuronal cell loss underlies the pathological decline in cognition and memory associated with Alzheimer disease (AD). Recently, targeting the endocannabinoid system in AD has emerged as a promising new approach to treatment. Studies have identified neuroprotective roles for endocannabinoids against key pathological events in the AD brain, including cell death by apoptosis. Elucidation of the apoptotic pathway evoked by β-amyloid (Aβ) is thus important for the development of therapeutic strategies that can thwart Aβ toxicity and preserve cell viability. We have previously reported that lysosomal membrane permeabilization plays a distinct role in the apoptotic pathway initiated by Aβ. In the present study, we provide evidence that the endocannabinoid system can stabilize lysosomes against Aβ-induced permeabilization and in turn sustain cell survival. We report that endocannabinoids stabilize lysosomes by preventing the Aβ-induced up-regulation of the tumor suppressor protein, p53, and its interaction with the lysosomal membrane. We also provide evidence that intracellular cannabinoid type 1 receptors play a role in stabilizing lysosomes against Aβ toxicity and thus highlight the functionality of these receptors. Given the deleterious effect of lysosomal membrane permeabilization on cell viability, stabilization of lysosomes with endocannabinoids may represent a novel mechanism by which these lipid modulators confer neuroprotection.

Neuronal cell loss underlies the pathological decline in cognition and memory associated with Alzheimer disease (AD). Recently, targeting the endocannabinoid system in AD has emerged as a promising new approach to treatment. Studies have identified neuroprotective roles for endocannabinoids against key pathological events in the AD brain, including cell death by apoptosis. Elucidation of the apoptotic pathway evoked by ␤-amyloid (A␤) is thus important for the development of therapeutic strategies that can thwart A␤ toxicity and preserve cell viability. We have previously reported that lysosomal membrane permeabilization plays a distinct role in the apoptotic pathway initiated by A␤. In the present study, we provide evidence that the endocannabinoid system can stabilize lysosomes against A␤-induced permeabilization and in turn sustain cell survival. We report that endocannabinoids stabilize lysosomes by preventing the A␤-induced up-regulation of the tumor suppressor protein, p53, and its interaction with the lysosomal membrane. We also provide evidence that intracellular cannabinoid type 1 receptors play a role in stabilizing lysosomes against A␤ toxicity and thus highlight the functionality of these receptors. Given the deleterious effect of lysosomal membrane permeabilization on cell viability, stabilization of lysosomes with endocannabinoids may represent a novel mechanism by which these lipid modulators confer neuroprotection.
Alzheimer disease (AD) 2 is a debilitating illness of the brain defined by the progressive deterioration of cognition and memory as a result of selective neuronal loss in the hippocampus and surrounding areas of the cerebral cortex (1). There is substantial evidence to suggest that at least a subset of neurons in the AD brain die by apoptosis (2). The principal neuropathological hallmark of the disease, ␤-amyloid peptide (A␤), has been shown to induce apoptosis in neuronal cells in vivo and in vitro (3,4) through a variety of enzymatic pathways that include activation of caspase-3 (5), calpain (6,7), and lysosomal cathepsins (8,9).
Recently, the lysosomal system has been implicated in AD pathogenesis (9,10). Neurons of AD patients demonstrate alterations in the lysosomal system, including the cellular pathways that converge on it, namely endocytosis and autophagy (10,11). Such alterations include an increase in the size and number of endosomes (10,12), autophagosomes (13) and lysosomes (10) and an increase in the gene expression and synthesis of all classes of lysosomal hydrolases, including cathepsins (14). In addition to their role in the digestion of cellular waste, it has become clear that partial and selective lysosomal membrane permeabilization (LMP), followed by the release of lysosomal enzymes into the cytosol, can induce apoptotic cell death (15). Cathepsins D and L are among the lysosomal proteases that have been implicated in apoptosis by virtue of their ability to activate apoptotic effectors, such as mitochondrial uncoupling and caspases (16).
Among the agents that are capable of destabilizing lysosomes, A␤ has emerged as an inducer of LMP (17). Cultured primary neurons are able to internalize A␤ from the culture medium where it accumulates within lysosomes, resulting in the loss of lysosomal membrane integrity and activation of the apoptotic cascade (18). We have recently identified that A␤ destabilizes lysosomes early in the apoptotic cascade in a manner involving the tumor suppressor protein, p53, and its transcription target, Bax (9). However, the precise mechanism by which A␤ causes lysosomes to become vulnerable to LMP has not yet been resolved.
The endocannabinoid system has emerged as a promising new target for neuroprotective therapy in AD (19). This system comprises the G protein-coupled cannabinoid (CB) receptors, CB 1 and CB 2 , their endogenous ligands, anandamide (AEA) and 2-arachidonoylglycerol (2-AG), and their degradative enzymes (20). Among their numerous functions, endocannabinoids are believed to play a role in the cell death/survival decision and thus govern cell fate (21). Numerous studies have identified neuroprotective roles of the endocannabinoid system against excitotoxicity (22), oxidative stress (23), and inflammation (24), all key pathological events in the AD brain. Moreover, endocannabinoids have been directly implicated in the protection of neurons against A␤ toxicity (25,26). Recently, it was reported that CB 1 receptors can be targeted to lysosomes via the adaptor protein, AP3, and are capable of mediating signal transduction while located at the lysosomal compartment (27). This spatial compartmentalization suggests a mechanism for diversity in CB 1 receptor signaling and indicates a functional role for CB 1 receptors at the lysosome. This has prompted us to investigate herein whether the neuroprotective capacity of the endocannabinoid system is based on its ability to stabilize lysosomes and thus block the lysosomal branch of the apoptotic pathway.
The results from this study indicate that endocannabinoids have the ability to stabilize lysosomes against A␤-induced LMP and thus may represent a novel target for therapeutic intervention in AD.

EXPERIMENTAL PROCEDURES
Culture of Cortical Neurons-Primary cortical neurons were established as described previously (28). Briefly, 1-dayold male Wistar rats were decapitated in accordance with institutional and national ethical guidelines, and their cerebral cortices were removed. The dissected cortices were incubated in phosphate-buffered saline (PBS) containing trypsin (0.3%) for 25 min at 37°C. The tissue was then triturated (X5) in PBS containing soybean trypsin inhibitor (0.1%) and DNase (0.2 mg/ml) and gently filtered through a sterile mesh filter. Following centrifugation, 2000 ϫ g for 3 min at 20°C, the pellet was resuspended in neurobasal medium, supplemented with heat-inactivated horse serum (10%), penicillin (100 units/ml), streptomycin (100 units/ml), and GlutaMAX (2 mM). Suspended cells were plated out at a density of 0.25 ϫ 10 6 cells on circular 13 mm diameter coverslips, coated with poly-L-lysine (60 g/ml), and incubated in a humidified chamber containing 5% CO 2 , 95% air at 37°C. After 48 h, 5 ng/ml cytosinearabino-furanoside was included in the culture medium to prevent proliferation of non-neuronal cells. Culture medium was exchanged every 3 days, and cells were grown in culture for 5 days prior to treatment.
Drug Treatment-A␤(1-40) (BioSource International Inc.) was made up as a 200 M stock solution in PBS and doubledeionized water and allowed to aggregate for 48 h at 37°C. For treatment of cortical neurons, A␤ was diluted to a final concentration of 2 M in prewarmed neurobasal medium. Control cells were exposed to the reverse peptide, A␤(40 -1) (2 M), which we have previously demonstrated to lack toxicity (8). Cells were exposed to A␤ in the presence or absence of the caspase-3 inhibitor, Z-DEVD-FMK (10 M) (29) (Tocris Bioscience), to assess whether the apoptotic pathway was activated in A␤-induced cell death. Cells were then exposed to A␤ in the presence or absence of the endocannabinoids, AEA (10 nM; Sigma-Aldrich) or 2-AG (10 nM; Sigma-Aldrich) to assess their neuroprotective capacity. The dose of AEA and 2-AG used was based on dose response experiments using the TUNEL technique. 10 nM was subsequently chosen because that concentration had no effect on basal cell viability. In some experiments, the hydrolysis of AEA and 2-AG was prevented by incubation with an inhibitor of fatty acid amide hydrolase, URB 597 (1 M) (30) (Cayman). The role of calpain in A␤-induced toxicity was assessed using an inhibitor of its activity, MDL 28170 (10 M) (6) (Merck). The involvement of CB 1 receptors in endocannabinoid-mediated neuroprotection was investigated using various antagonists, SR 141716A (1 M) (31) (gift from Dr. David Finn at The National University of Ireland, Galway. Original source: The National Institute of Mental Health's Chemical Synthesis and Drug Supply Program), hemopressin (10 M) (32, 33) (Tocris), and AM 251 (10 M) (34) (Tocris). In all cases, the concentration of inhibitors used was in agreement with the literature or based on previous publications, as indicated.
The water-soluble CB 1 receptor agonist, O-2545, was used as a positive control to demonstrate the ability of hemopressin to block surface CB 1 receptor activity. Finally, a neutralizing cathepsin L antibody (1 M) (Merck) was used to assess the toxicity of cathepsin L released into the culture medium.
Cells were viewed under ϫ63 magnification using a confocal microscope (Zeiss LSM 510 META). The multitrack FITC/rhodamine channel configuration was selected (for Alexa 488, excitation at 488 nm and emission at 520 nm; for LysoTracker TM probe, excitation at 543 nm and emission at 599 nm).
Lysosomal Integrity Assay; Acridine Orange (AO) Relocation-The lysosomal integrity assay was carried out as described previously (9). Briefly, cells were exposed to prewarmed supplemented neurobasal medium containing acridine orange (5 g/ml; Invitrogen) for 10 min at 37°C. Cells were rinsed in neurobasal medium, exposed to cell treatments for 6 h, and viewed by confocal microscopy. Visualization of the fluorophore was achieved using the 488-nm argon laser in the mode, where emission over the 499 -670 nm range was collected. The configuration parameters were as follows: 1) filters (Ch3-BP, 585-615 nm; Ch2-BP, 505-530 nm, and ChS1, 499.3-670.7 nm); 2) beam splitters (HFT 488 nm); 3) scan zoom 1. For each digital image, 512 ϫ 512 pixels were used. The leakage of AO from the lysosome produces a decrease in the 633 nm emission, and this parameter was used as an index of lysosomal integrity, as reported previously (35). The average cellular fluorescence at 633 nm was measured from at least 200 cells for each treatment, from at least four independent experiments.
Cathepsin L Activity-Cathepsin L activity was measured using a commercially available fluorogenic assay under conditions that inactivate other thiol proteases, such as cathepsin B. Culture medium was used from cells treated with A␤ in the presence or absence of AEA and 2-AG and incubated with the cathepsin L substrate, Z-Phe-Arg, conjugated to aminofluorocoumarin (200 M; Alexis Biochemicals) and the reaction buffer (20 mM NaOAc, pH 5.0, 4 mM EDTA, 8 mM DTT, and 0.4 M urea) for 1 h at 37°C in a reaction volume of 100 l. The amount of free aminofluorocoumarin produced was assessed by spectrofluorometry (excitation, 400 nm; emission, 505 nm).
Calpain Activity-Calpain 1 and 2 activity was assessed as described previously (8). Briefly, cleavage of the fluorogenic calpain substrate, Suc-LLVY-AMC, was used to assess calpain activity in cell lysates. Samples were incubated with Suc-LLVY-AMC for 15 min at room temperature, and the amount of AMC released upon cleavage with calpain was measured fluorometrically at an excitation wavelength of 360 nm and an emission wavelength of 440 nm.
TUNEL-Apoptotic cell death was assessed using the Dead End TM colorimetric apoptosis detection system (Promega Corp.). Cells were treated with A␤ for 48 h in the presence or absence of Z-DEVD-FMK, AEA, 2-AG, or URB 597. Cells were then fixed with 4% paraformaldehyde and permeabilized with Triton X-100 (0.1%), and the biotinylated nucleotide was incorporated at 3Ј-OH DNA ends using the enzyme terminal deoxynucleotidyltransferase. Horseradish peroxidase-labeled streptavidin then bound to the biotinylated nucleotide, and this was detected using the peroxidase substrate H 2 O 2 and the chromogen diaminobenzidine. Cells were viewed under light microscopy at ϫ40 magnification, where the nuclei of TUNEL-positive cells stained brown. Apoptotic cells (TUNEL-positive) were counted and expressed as a percent-age of the total number of cells examined. To exclude the possibility that the number of living cells present on the coverslip had an affect on the TUNEL-positive ratio, the same number of cells (ϳ500) were counted for each treatment.
Statistics-Data are reported as the mean Ϯ S.E. of the number of experiments indicated in each case. Statistical analysis was carried out using one-way analysis of variance (ANOVA) followed by the post hoc, Student-Newman-Keuls, test when significance (at the p Ͻ 0.05 level) was indicated. When comparisons were being made between two treatments, an unpaired Student's t test was performed, and p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001 were considered significant.
Endocannabinoids Prevent A␤-mediated Destabilization of Lysosomes and Cathepsin Release-Lysosomal destabilization has been reported as an upstream event in A␤-mediated neurotoxicity (9). We tested the hypothesis that endocannabinoids could thwart the apoptotic pathway and confer neuroprotection by preventing lysosomal destabilization. Cells were loaded with AO prior to treatment with A␤ (2 M; 6 h) with or without 2-AG (10 nM), AEA (10 nM), or URB 597 (1 M) (i.e. prior to onset of DNA fragmentation), and emission at 633 nm was assessed as an index of lysosomal integrity ( Fig.  2A). A␤ significantly reduced the fluorescence emission at 633 nm from 133 Ϯ 18 fluorescence units (mean Ϯ S.E.; arbitrary units) to 52 Ϯ 10 (p Ͻ 0.05, ANOVA, n ϭ 5), indicative of a loss in lysosomal integrity. Exposure to A␤ in the presence of AEA or 2-AG restored the fluorescence emission at 633 nm to 124 Ϯ 10 and 117 Ϯ 10 fluorescence units, respectively (p Ͻ 0.05, ANOVA), demonstrating a stabilizing effect of both endocannabinoids on lysosomes. The fatty acid amide hydrolase inhibitor, URB 597, also restored lysosomal integrity because emission at 633 nm was comparable with control values in cells exposed to A␤ in the presence of URB 597 (130 Ϯ 18 fluorescence units, n ϭ 5). Thus, in control cells (Fig. 2B, i), the AO staining displayed a punctate pattern of fluorescence due to its accumulation within acidic lysosomal vesicles, indicative of intact lysosomal membranes. In contrast, the pattern of AO staining in A␤-treated cells (Fig. 2B, ii) had shifted to a diffuse cytosolic green fluorescence, reflecting the release of AO from lysosomes into the cytosol. In cells exposed to A␤ in the presence of 2-AG (Fig. 2B, iii), AEA (Fig. 2B, iv), or URB 597 (Fig. 2B, v), the punctate pattern of AO staining was restored.
The stabilizing effect of endocannabinoids on lysosomes was confirmed with BODIPY FL-pepstatin A staining (Fig.  2C), which demonstrates that the punctate staining in control cells (Fig. 2C, i), indicative of intact lysosomes containing cathepsin D, becomes more diffuse in cells exposed to A␤ (Fig.  2C, ii), indicating a loss of lysosomal cathepsin D into the cytosol. In cells exposed to A␤ in the presence of 2-AG (Fig. 2C, iii), AEA (Fig. 2C, iv), or URB 597 (Fig. 2C, v), the punctate pattern of BODIPY FL-pepstatin A staining was restored. The frequency distributions of BODIPY FL-pepstatin A pixel intensity were compared for control and A␤-treated cells. Four data sets per group were averaged and statistically compared with the Kolmogorov-Smirnov test. Exposure to A␤ was associated with a flattening of the high intensity frequency peak, and the curves were significantly different (p Ͻ 0.001), representing the A␤-induced change in cathepsin D distribution (data not shown).
To further demonstrate a loss of lysosomal stability, we measured cathepsin L activity in the culture medium of treated cells (Fig. 2D) and thus examined its extracellular release. A␤ (2 M, 6 h) significantly increased cathepsin L activity in the culture medium from 3.7 Ϯ 0.9 relative fluorescence units (RFU; mean Ϯ S.E.) to 7.6 Ϯ 1.2 RFU (p Ͻ 0.05, ANOVA, n ϭ 5), and this was inhibited when cells were treated with A␤ in the presence of AEA (p Ͻ 0.05, ANOVA, compared with cells treated with A␤, n ϭ 5) or 2-AG (p Ͻ 0.05, ANOVA, compared with cells treated with A␤, n ϭ 5). To assess whether the cathepsin L release evoked by A␤ could be neurotoxic to neighboring neurons, the medium from A␤primed cells was incubated with naive cells, and apoptosis was assessed (Fig. 2E). The medium from A␤-primed cells evoked a significant increase in the percentage of apoptotic cells from 7 Ϯ 1% (mean Ϯ S.E.) to 22 Ϯ 2% (p Ͻ 0.001, ANOVA, n ϭ 5), and this was significantly reduced to 14 Ϯ 2% apoptotic cells (p Ͻ 0.05, ANOVA, n ϭ 5) following co-treatment with the cathepsin L-neutralizing antibody (1 M). This suggests that the cathepsin L released from the lysosomes by A␤ is still active and capable of inducing apoptosis in neighboring cells, thus highlighting the importance of blocking this lysosomal aspect of A␤ neurotoxicity. A, the involvement of the apoptotic pathway in A␤-induced cell death was assessed by the TUNEL technique in cultured cortical neurons exposed to A␤ (2 M) for 48 h in the presence or absence of the caspase-3 inhibitor Z-DEVD-FMK (10 M). In cells exposed to A␤, the percentage of DNA fragmentation was significantly increased (***, p Ͻ 0.001 versus control), and this was prevented by Z-DEVD-FMK (***, p Ͻ 0.001 versus A␤). B, 2-AG, AEA (**, p Ͻ 0.01 versus A␤), and URB 597 (*, p Ͻ 0.05 versus A␤) significantly prevented the A␤-induced increase in DNA fragmentation (***, p Ͻ 0.001 versus control). C, representative TUNEL images of cortical neurons exposed to control (i), A␤ (ii), A␤ and 2-AG (iii), A␤ and AEA (iv), and A␤ and URB 597 (v). The arrows indicate cells with fragmented DNA (scale bar, 10 m). D, A␤ significantly increased caspase-3 activity (**, p Ͻ 0.01 versus control), and this was prevented by 2-AG, AEA, and URB 597 (ϩϩ, p Ͻ 0.01 versus A␤). n ϭ 5. Error bars, S.E.

Endocannabinoids Prevent the A␤-mediated Up-regulation of p-p53 Ser-15 and p-p53 Ser-15 -Lysosomal Interaction-We
have recently reported that A␤ destabilizes lysosomes in a manner involving the interaction of p-p53 Ser-15 with the lyso-somal membrane. In Fig. 3, we demonstrate that the A␤-mediated up-regulation of p-p53 Ser-15 (Fig. 3A) and association of p-p53 Ser-15 with the lysosomes (Fig. 3B) is prevented by the endocannabinoids, AEA and 2-AG. In Fig. 3A, the expression FIGURE 2. Endocannabinoids prevent A␤-mediated destabilization of lysosomes and cathepsin release. A, cells were loaded with AO and exposed to A␤ (2 M) for 6 h in the presence or absence of 2-AG (10 nM), AEA (10 nM), or the fatty acid amide hydrolase inhibitor, URB 597 (1 M), and lysosomes were viewed with confocal microscopy. A decrease in emission at 633 nm was indicative of LMP. In cells exposed to A␤, emission at 633 nm was significantly decreased (*, p Ͻ 0.05 versus control), and this was prevented by 2-AG, AEA, and URB 597 (ϩ, p Ͻ 0.05 versus A␤). B, representative AO-stained neurons depicting intact lysosomes (pink) in control cells (i) and A␤-induced leakage of AO into the cytosol (diffuse green) (ii), which is prevented by 2-AG (iii), AEA (iv), and URB 597 (v). The arrows indicate cells with intact lysosomes (scale bar, 10 m). C, representative images of cortical neurons loaded with BODIPY FL-pepstatin A depicting punctate staining in control cells (i), indicative of intact lysosomes containing cathepsin D and A␤-induced redistribution of cathepsin D into the cytosol (ii), which is prevented by 2-AG (iii), AEA (iv), and URB 597 (v). D, cathepsin L activity was significantly increased in the culture medium of cells exposed to A␤ (*, p Ͻ 0.05 versus control), and this was prevented by 2-AG and AEA (ϩ, p Ͻ 0.05 versus A␤). E, the A␤-primed medium, rich in cathepsin L, induced a significant increase in DNA fragmentation in cortical neurons (***, p Ͻ 0.001 versus control), and this was prevented upon the addition of a cathepsin L-neutralizing antibody (*, p Ͻ 0.05 versus A␤-primed cells). n ϭ 5. Error bars, S.E. DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 of p-p53 Ser-15 is low in control cells (Fig. 3A, i), whereas p-p53 Ser-15 immunoreactivity is increased in cells exposed to A␤ (2 M, 6 h; Fig. 3A, ii). The A␤-mediated increase in p-p53 Ser-15 immunoreactivity is blocked by both 2-AG (Fig.  3A, iii) and AEA (Fig. 3A, iv). In Fig. 3B, cells are loaded with LysoTracker TM Red to observe lysosomes and probed for p-p-p53 Ser-15 immunoreactivity. In control cells (Fig. 3B, i), lysosomes are intact, and p-p53 Ser-15 immunoreactivity is low. In A␤-treated cells (Fig. 3B, ii), p-p53 Ser-15 immunoreactivity is increased, and some of the immunoreactivity is co-localized with LysoTracker TM Red (shown in purple), in accordance with our previous findings (9). The co-localization of p-p-p53 Ser-15 with lysosomes is abolished by 2-AG (Fig. 3B, iii) and AEA (Fig. 3B, iv). These results suggest that the A␤-mediated up-regulation and association of p-p53 Ser-15 with lysosomes are prevented by the endocannabinoid system.

Endocannabinoids Prevent Lysosomal Destabilization
Endocannabinoids Prevent A␤-induced Calpain Activation-In Fig. 4A, we demonstrate that the A␤-mediated increase in p-p53 Ser-15 expression is a calpain-dependent event. Cells were treated with the selective calpain inhibitor, FIGURE 3. Endocannabinoids prevent the A␤-mediated up-regulation of p-p53 Ser-15 and p-p53 Ser-15 -lysosomal interaction. Cortical neurons were exposed to A␤ (2 M) for 6 h in the presence or absence of 2-AG (10 nM) and AEA (10 nM) and assessed for p-p53 Ser-15 expression (A). A␤ increased the expression of p-p53 Ser-15 (ii) compared with control (i), and this was prevented by 2-AG (iii) and AEA (iv). The arrows indicate p-p53 Ser-15 immunoreactivity (scale bar, 10 m). B, loading of cortical neurons with LysoTracker TM Red demonstrated the A␤-induced co-localization of p-p53 Ser-15 with lysosomes (ii) compared with control (i). Inset, arrows indicate areas of colocalization (purple). This was also prevented by 2-AG (iii) and AEA (iv). Scale bar, 10 m. n ϭ 5.
In order to assess whether inhibition of calpain could also prevent the effects of A␤ on lysosomal membrane integrity, cultured cortical neurons were loaded with AO prior to treatment with A␤ (2 M; 6 h) with or without MDL 28170 (10 M). To demonstrate the importance of calpain activation in the apoptotic pathway evoked by A␤, DNA fragmentation was also assessed in cells exposed to A␤ (2 M; 48 h) in the presence or absence of MDL 28170. Fig. 4C demonstrates that the apoptotic pathway evoked by A␤ is dependent upon activation of calpain. Thus, in control cells, DNA fragmentation was 3.8 Ϯ 0.5% (mean Ϯ S.E.), and this was significantly increased to 24.1 Ϯ 1.4% in cells exposed to A␤ (p Ͻ 0.001, ANOVA, n ϭ 5). However, in the presence of MDL 28170, DNA fragmentation was significantly decreased to 5.5 Ϯ 0.4%.
Finally, we investigated whether the neuroprotective proclivity of endocannabinoids relied on their ability to prevent calpain activation. Calpain activity was assessed in cultured cortical neurons exposed to A␤ (2 M; 6 h) in the presence or absence of 2-AG (10 nM), AEA (10 nM) or URB 597 (1 M). Fig. 4D demonstrates that modulation of the endocannabinoid system can prevent the A␤-induced increase in calpain activation. Thus, A␤ significantly increased calpain activity from 5.8 Ϯ 0.3 RFU (AMC release/min/mg) in control cells to 8.1 Ϯ 0.9 RFU (mean Ϯ S.E.). This was significantly decreased to 4.8 Ϯ 0.4, 4.7 Ϯ 0.9, and 5.2 Ϯ 0.2 RFU by co-treatment with 2-AG, AEA, and URB 597, respectively. These results demonstrate the role of calpain activation in A␤-induced apoptosis and the ability of the endocannabinoid system to prevent this phenomenon.
Receptor Target for AEA-To investigate the involvement of the CB 1 receptor in endocannabinoid-mediated neuroprotection, cortical neurons were exposed to A␤ (2 M; 48 h) with or without 2-AG (10 nM) or AEA (10 nM) in the presence or absence of the CB 1 receptor antagonist, AM 251 (10 M). Fig.  5A demonstrates that AEA, but not 2-AG, provides neuroprotection against A␤-induced DNA fragmentation in a CB 1 receptor-dependent manner. Thus, when neurons were exposed to A␤ and 2-AG in the presence of AM 251, 2-AG maintained its ability to confer neuroprotection. However, AEA was unable to provide neuroprotection against A␤ while in the presence of AM 251. Subsequent experiments on the role of CB 1 receptors in endocannabinoid-mediated neuroprotection thus focused solely on AEA. To determine whether surface or intracellular CB 1 receptors were required for the neuroprotective effects of AEA on A␤-induced lysosomal destabilization and apoptosis, we used the cell-permeable CB 1 receptor antagonist, SR 141716A (1 M), which can target both intracellular and plasma membrane CB 1 receptors, and the cell-impermeable CB 1 receptor antagonist, hemopressin (10 M), which can target only plasma membrane CB 1 receptors (27). Cortical neurons were exposed to A␤ (2 M, 48 h) and AEA (10 nM) in the presence or absence of SR 141716A or hemopressin and assessed for DNA fragmentation. Fig. 5B demonstrates that the neuroprotective effect of AEA against A␤induced DNA fragmentation was blocked by SR 141716A. In contrast, AEA maintained the ability to confer neuroprotection against A␤-induced toxicity while in the presence of hemopressin, thus providing evidence for the role of intracellular CB 1 receptors in AEA-mediated neuroprotection. The water-soluble CB 1 receptor agonist, O-2545, acted as a positive control to demonstrate the ability of hemopressin to block surface CB 1 receptor activity. Thus, cortical neurons exposed to O-2545 expressed a higher level of p-ERK Tyr-205 phosphorylation compared with control cells, and this was prevented following co-treatment with hemopressin (Fig. 5B,  inset).
Similarly, SR 141716A, but not hemopressin, abolished the ability of AEA to conserve the integrity of the lysosomal membrane (arbitrary units; Fig. 5C). Representative images of AO-stained neurons treated with hemopressin control (i), hemopressin, A␤ and AEA (ii), SR 141716A control (iii), and SR 141716A, A␤ and AEA (iv) are shown in Fig. 5D.

DISCUSSION
Endocannabinoids Prevent A␤-induced Apoptosis-In this study, we demonstrate that endocannabinoids, both directly and through pharmacological enhancement, are able to protect primary cerebral cortical neurons from A␤-induced apoptosis in vitro. Thus, 2-AG, AEA, and URB 597 prevented the A␤-induced increase in DNA fragmentation and caspase-3 activation, both of which are hallmarks of apoptosis. Whether apoptosis is responsible for all of the neuronal cell loss observed in AD is a subject of some controversy. The findings of our study would suggest that at least a subset of neurons in the AD brain are vulnerable to cell death by apoptosis. This is in agreement with a number of studies that have reported an increase in DNA fragmentation (37,38) and caspase-3 activation (39) in the AD brain as well as the proclivity of A␤ to induce apoptosis in neuronal cells in vivo (3).
The findings of our study also suggest that modulation of the endocannabinoid system could be therapeutically advantageous in preventing the neuronal cell loss associated with AD. We demonstrate that both direct applications of 2-AG and AEA and their pharmacological enhancement using URB 597 are able to provide protection against A␤ toxicity. The relevance of the endocannabinoid system to AD treatment has been contemplated for many years. Several studies have reported on its potential neuroprotective capabilities in various models of AD (25,26,40) as well as highlighting diseaserelated alterations that occur over time (40 -42). However, certain discrepancies also exist regarding the benefits of endocannabinoid therapeutics to AD research, with a number of DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49

JOURNAL OF BIOLOGICAL CHEMISTRY 38549
studies reporting that it has no effect on AD neuropathology (43,44). Many of these inconsistencies can be explained by the use of different AD models, application of a diverse range of endocannabinoid system-modulating drugs (e.g. phyto-, endo-, and synthetic cannabinoids), and different drug concentrations used. Contradictions in results may also be due to variable timing of drug application, as was suggested by van der Stelt et al. in 2006 (26). Thus, our findings are specific to an in vitro model of AD consisting of primary cerebral cortical neurons exposed to toxic concentrations of A␤ and the neuroprotective proclivity of co-applied low concentrations of endocannabinoids. The concentration of AEA and 2-AG used in this study was based on a dose-response experiment using the TUNEL technique, whereby 10 nM had no effect on cell viability. However, in this experiment, higher concentrations (Ͼ2 M for 2-AG and Ͼ10 M for AEA) induced a significant In the presence of AM 251, 2-AG maintained the proclivity to prevent A␤-induced DNA fragmentation (888, p Ͻ 0.001 versus A␤). However, in the presence of AM 251, the neuroprotective effect of AEA on A␤-induced apoptosis was abolished, and a significant increase in DNA fragmentation was observed (###, p Ͻ 0.001 versus A␤ and AEA). B, to determine the involvement of intracellular and plasma membrane CB 1 receptors, cortical neurons were exposed to A␤ (2 M; 48 h) and AEA (10 nM) in the presence of the cell-permeable CB 1 receptor antagonist, SR 141716A (1 M), or the cell-impermeable antagonist, hemopressin (10 M), and assessed for DNA fragmentation. DNA fragmentation was significantly increased in cells exposed to A␤ (***, p Ͻ 0.001 versus control), and this was prevented by AEA (ϩϩϩ, p Ͻ 0.001). In the presence of hemopressin, AEA maintained the proclivity to prevent A␤-induced DNA fragmentation (888, p Ͻ 0.001 versus A␤). However, in the presence of SR 141716A, the neuroprotective effect of AEA was abolished, and a significant increase in DNA fragmentation was observed ( § § §, p Ͻ 0.001 versus A␤ and AEA). Inset, the water-soluble CB 1 receptor agonist, O-2545, was used as a positive control to demonstrate the inhibitory effect of hemopressin on p-ERK Tyr-205 expression. Representative images for p-ERK Tyr-205 immunoreactivity are displayed in cells treated with control (i), hemopressin (ii), O-2545 (iii), and hemopressin and O-2545 (iv). C, emission at 633 nm was significantly decreased in cells exposed to A␤, indicative of lysosomal destabilization (***, p Ͻ 0.001 versus control), and this was prevented by AEA (ϩϩϩ, p Ͻ 0.001 versus A␤). In the presence of hemopressin, AEA maintained the proclivity to prevent A␤-induced lysosomal destabilization (888, p Ͻ 0.001 versus A␤). However, in the presence of SR 141716A, the stabilizing effect of AEA on lysosomes was abolished (#, p Ͻ 0.05 versus A␤ and AEA). D, representative images of AO-stained neurons depicting hemopressin control-treated (i); hemopressin-, A␤-, and AEA-treated (ii); SR 141716A control-treated (iii); and SR 141716A-, A␤-, and AEA-treated (iv) neurons. The arrows indicate cells with intact lysosomes (scale bar, 10 m). n ϭ 5. Error bars, S.E. increase in DNA fragmentation compared with control cells, thus representing the neuroprotection/neurotoxicity paradox of endocannabinoid modulation, which has been the focus of a recent review (45).
Consistent with our finding, other studies have reported on the neuroprotective properties of the endocannabinoid system in hippocampal and cortical neurons against a number of toxic insults, including excitotoxicity (22,46), ischemia (47), and oxidative damage (23), all of which are known to induce cell death by apoptosis. This endocannabinoid-mediated neuroprotection has been reported to occur through a variety of mechanisms, including activation of prosurvival signaling pathways like inositol triphosphate (48), PI3K (49), focal adhesion kinase (50), and ERK (49 -50) as well as through inhibition of calcium currents and opening of potassium channels (51)(52)(53). The non-psychoactive component of cannabis, cannabidiol, has a number of additional characteristics that highlight the benefits of using cannabinoid-based therapeutics for the treatment of AD, such as its ability to scavenge reactive oxygen species (23), reverse Tau hyperphosphorylation (54), and reduce activation of the inflammatory transcription target, nuclear factor-B (54). This study demonstrates for the first time that endocannabinoid-mediated neuroprotection can also occur through stabilization of the lysosomal membrane.
Endocannabinoids Prevent A␤-induced Lysosomal Membrane Permeabilization-Our finding that endocannabinoids stabilize lysosomes against A␤ toxicity is novel and represents an early event in the apoptotic pathway that may be targeted by the endocannabinoid system to confer neuroprotection. LMP, as a distinct event in the apoptotic pathway, is now widely accepted (16,55,56) and is believed by many to occur upstream of mitochondrial outer membrane permeabilization and caspase-3 activation (57,58). This is consistent with our findings, given that LMP was evoked after 6 h of exposure to A␤, whereas DNA fragmentation and caspase-3 activation typically do not occur at a significant level until 48 h post-A␤ treatment.
Although it is possible that A␤ increased the luminal pH of lysosomes, leading to the redistribution of AO into the cytosol, our finding that A␤ promotes the translocation of the lysosomal enzymes cathepsin D to the cytosol and cathepsin L to the extracellular matrix supports our proposal that A␤ evokes permeabilization of the lysosomes. Both of these enzymes have been implicated in the initiation and execution of cell death pathways (15). For example, cathepsin D has the proclivity to activate the proapoptotic protein, Bax, in a Bidindependent (59) and Bid-dependent (60) manner, resulting ultimately in mitochondrial outer membrane permeabilization, whereas cathepsin L is capable of activating caspase-dependent apoptosis through the cleavage of the proapoptotic protein, Bid, and the antiapoptotic proteins Bcl-2 and Bcl-xL (61).
The diffuse appearance of BODIPY FL-pepstatin A in Fig.  2C (ii) demonstrates the ability of A␤ to translocate cathepsin D from the lysosomal lumen into the cytosol. It is still unknown whether all lysosomes within a cell are uniformly susceptible to LMP; however, there is some evidence to suggest that larger lysosomes may be more vulnerable than smaller ones (62). This discrepancy may account for the small amount of punctuate staining that is visible in the A␤-treated neurons. Another possibility is that alterations in the amount of A␤ loading into lysosomes may in turn affect the extent of LMP that occurs. However, it should be noted that a study by Roberg et al. (63) demonstrated that microinjection of cathepsin D enzyme into the cytosol is sufficient to trigger mitochondrial outer membrane permeabilization and thus highlights the impact of A␤ toxicity on a single lysosome.
We also report in this study that cathepsin L activity was significantly increased in the cell culture medium concomitant with LMP. Furthermore, the A␤-primed cells, rich in cathepsin L, were sufficient to induce apoptosis in fresh neurons, thus suggesting that the secreted cathepsin L is active and capable of initiating apoptosis in neighboring cells. This could be of relevance to AD, a disease characterized by neuronal cell loss, because it implies that loss of lysosomal integrity in one cell could be enough to trigger cell death in a whole population of adjacent neurons. Most cells secrete small amounts of newly synthesized procathepsin L, but under certain physiological conditions, mature cathepsin L has also been detected in cell culture medium (64). The mechanism of this secretion remains to be determined, but there is some evidence to suggest that endosomes can become enriched with cathepsin L. Under certain stimuli, cathepsin L is then activated and secreted via a distinctive targeting pathway (64). Of interest, when cathepsin L is up-regulated, targeting to lysosomes continues at a constant level, whereas the additional enzyme produced enters a different targeting pathway (e.g. endosomes) (64,65). We have previously demonstrated that A␤ can promote a time-and dose-dependent release of cathepsin L into the cytosol, with cathepsin L release being induced 6 h after A␤ treatment (8). Thus, it is possible that the cytosolic release of cathepsin L is the result of A␤-induced LMP, whereas the cathepsin L detected in our cell culture medium is a result of its up-regulation and subsequent targeting to secretory endosomes. Using a cathepsin L-neutralizing antibody, we determined that cathepsin L in our cell culture medium was inducing apoptosis in neighboring cells. However, the effect of the neutralizing antibody was only partial and not sufficient to return DNA fragmentation to control levels. This suggests that other lysosomal enzymes may be released alongside cathepsin L following exposure to A␤ (e.g. the cysteine protease cathepsin B).
Precisely how A␤ induces LMP remains to be clarified. Findings from our laboratory would suggest that it may evoke structural alterations in lysosome-associated membrane proteins, which in turn cause lysosomes to become vulnerable to LMP (data not shown). In this study, we report that this harmful effect of A␤ on lysosomal integrity is dependent upon the phosphorylation of p53 Ser-15 .
Our finding that endocannabinoids have the proclivity to stabilize lysosomes is novel and may represent a vital mechanism by which these lipid modulators confer neuroprotection. Our results suggest that this stabilizing effect of endocannabinoids on lysosomal membranes is based upon their ability to prevent A␤-induced alterations in p53  . Given that the DECEMBER 3, 2010 • VOLUME 285 • NUMBER 49 neuroprotective effect of endocannabinoids may rely upon their early administration (26), and because LMP occurs early in the apoptotic pathway, lysosomal integrity may be a pivotal point at which endocannabinoids decide between cell death and cell survival.

Endocannabinoids Prevent Lysosomal Destabilization
Endocannabinoids Prevent the A␤-induced Up-regulation of p-p53  and Its Association with the Lysosomal Membrane by Inhibiting Calpain Activation-We report that endocannabinoids stabilize lysosomes by inhibiting the A␤-induced up-regulation of p-p53 Ser-15 and its subsequent translocation to the lysosomal membrane. Based on our previous findings that LMP evoked by A␤ is dependent upon p-p53 Ser-15 (9), the inhibiting effect of endocannabinoids on p-p53 Ser-15 up-regulation may account for their neuroprotective proclivities at the lysosomal membrane. Precisely how p-p53 Ser-15 impacts on the lysosome to induce LMP remains to be elucidated. There is some evidence to suggest that p53 forms a complex with the adaptor protein, LAPF (lysosome-associated and apoptosis-inducing protein pleckstrin homology and FYVE domains) to promote LMP in a transcription-independent manner (66). p53 has also been reported to associate with the lysosomal protein DRAM (damage-regulated autophagy modulator), whereby it induces autophagy (67). Given that A␤ destabilizes lysosomes, an induction of autophagy at such a time would probably culminate in lysosomal burden, rendering the lysosomes vulnerable to LMP. Thus, it is interesting to speculate that endocannabinoids inhibit p-p53 Ser-15 up-regulation and lysosomal association in order to prevent an induction of autophagy.
We also report on the involvement of calpain in the apoptotic pathway evoked by A␤. Thus, we demonstrated that A␤ significantly increased calpain activity in our cultured cortical neurons, whereas inhibition of calpain prevented A␤induced phosphorylation of p-p53 Ser-15 and the subsequent effects on lysosomal integrity and DNA fragmentation. Consistent with our findings that calpain activation may be pivotal to A␤ toxicity, it has been reported that inhibition of calpain improves memory and synaptic transmission in a mouse model of AD (68). With respect to calpain involvement in neuronal cell death, several studies have reported on its role in the apoptotic pathway (7, 69 -71) and its ability to catalyze LMP (72,73). Calpains are non-lysosomal cysteine proteases that are activated in response to an increase in intracellular Ca 2ϩ concentration leading to the proteolysis of its target proteins and activation of signal transduction pathways. Among the protein substrates for calpain is p53. We have demonstrated in our study that calpain activation is pivotal for p-p53 Ser-15 phosphorylation and thus vital for the effects of p-p53 Ser-15 at the lysosomal membrane. This is in agreement with other studies that have demonstrated pharmacological inhibition of calpain abrogates LMP (72).
It has been known for some time that intracellular Ca 2ϩ levels are increased in the aged brain, and on the basis of this observation, it has been suggested that dysregulation of Ca 2ϩ homeostasis may underlie the neurodegeneration that occurs in AD. A␤-induced increases in intracellular Ca 2ϩ concentrations have been reported in a number of studies (74 -76). Thus, a possible scenario emerges whereby exposure of corti-cal neurons to A␤ results in an increase in intracellular Ca 2ϩ concentration, which in turn leads to calpain activation and phosphorylation of p53 Ser-15 , resulting ultimately in LMP and cell death.
A well established neuroprotective property of endocannabinoids is their ability to reduce Ca 2ϩ influx through the CB 1 receptor by blocking the activity of voltage-dependent N-, P/Q-, and L-type Ca 2ϩ channels and through the presynaptic inhibition of glutamate release (22). Because a small change in intracellular Ca 2ϩ concentration is sufficient to activate calpains, it is a possibility that the capacity of endocannabinoids to confer neuroprotection is through their inhibition of Ca 2ϩ influx. Indeed, we have demonstrated that endocannabinoids prevent the A␤-induced increase in calpain activity. Given that endocannabinoids are produced in response to an increase in intracellular Ca 2ϩ concentration, it is possible that under normal circumstances, this leads to the inhibition of calpain and its downstream affects on p53 Ser-15 and lysosomal integrity. However, this regulatory effect may be lost in the AD brain. Thus, therapies that modulate endocannabinoid tone may be advantageous based on their ability to prevent a rise in intracellular Ca 2ϩ . In accordance with this theory, we have demonstrated that URB 597 can prevent the A␤-induced increase in calpain activation, destabilization of the lysosomal membrane, and ensuing cell death.
AEA Targets Intracellular CB 1 Receptors-Our findings also suggest that AEA, but not 2-AG, mediates its neuroprotection in a CB 1 receptor-dependent manner. Furthermore, the neuroprotective and lysosome stabilizing effects of AEA were dependent upon intracellular and not plasma membrane CB 1 receptors. Thus, when cortical neurons were exposed to hemopressin, a CB 1 receptor antagonist that cannot cross the plasma membrane and therefore can only target plasma membrane CB 1 receptors, AEA still had the proclivity to protect against the A␤-induced increase in DNA fragmentation and LMP. In contrast, in the presence of SR 141716A, a cell-permeable CB 1 receptor antagonist, AEA could no longer confer neuroprotection. Hemopressin is a naturally occurring peptide that is capable of functionally antagonizing the activity of CB 1 receptors. Recent studies have reported that hemopressin's mechanism of action reflects those of the well characterized CB 1 inverse agonists, AM 251 and SR 141716A (32,33).
In this study, we demonstrated the ability of hemopressin to prevent p-ERK Tyr-205 signaling by the water-soluble CB 1 agonist O-2545 and thus exhibit the proclivity of hemopressin to antagonize this receptor. Further evidence that hemopressin selectively antagonizes CB 1 receptors comes from the recent finding that it can prevent agonist-induced receptor internalization into endosomes (32). However, it should be noted that Gomes et al. (77) have identified that N-terminally extended forms of hemopressin can function as CB 1 receptor agonists, activating a signal transduction pathway distinct from that activated by endocannabinoids.
It has recently been reported that the intracellular pools of CB 1 receptors can directly associate with lysosomes and moreover that they retain their ability to couple to G␣ i to activate ERK (27). Given our finding that intracellular CB 1 receptors are required for AEA-mediated neuroprotection, it is possible that direct signaling of CB 1 receptors at the lysosome can be attributed to this neuroprotective effect. Furthermore, the finding that CB 1 receptors are trafficked to lysosomes via AP-3 (27), the same adaptor protein responsible for trafficking lysosome-associated membrane proteins 1 and 2 to lysosomes, suggests that a far more intimate relationship may exist between the endocannabinoid and lysosomal systems. Moreover, given that lysosomes are stores for Ca 2ϩ (78), it is an intriguing possibility that endocannabinoids govern lyosomal integrity by preventing A␤-induced lysosomal calcium release, a small amount of which would be sufficient to activate calpain.
In 2005, Ramirez et al. (40) reported that CB 1 receptor expression was decreased on neurons remote from amyloid plaques in post-mortem AD brains and suggested that they might be extremely vulnerable to the toxic species generated by microglia. Similarly, autoradiographic studies with [ 3 H]CP55.940 in patients with AD demonstrated that CB 1 receptor density was reduced in the entorhinal cortex, hippocampus, and caudate of the latter (41). However, in this study, reductions in CB 1 receptor density were not associated with the characteristic neuropathological hallmarks of AD. Recent findings from Cudaback et al. (79) suggest that cannabinoid receptor expression may determine their efficacy for activating proapoptotic and prosurvival signals, whereby low levels of cannabinoid receptor expression in astrocytoma subclones couple to apoptotic pathways, whereas high levels of expression also couple to prosurvival pathways. Thus, it is a possibility that endocannabinoids have a direct modulating effect on the expression of cannabinoid receptors, which in turn couple to prosurvival signaling pathways to confer neuroprotection. Moreover, given that endocytic processes are required for the axonal targeting of CB 1 receptors (80), which in turn depend upon an efficient lysosomal system, endocannabinoid-mediated stabilization of lysosomal membranes may in turn affect the expression level of cannabinoid receptors within a given cell. CB 2 receptor expression is increased on microglia within amyloid plaques (40,42), and evidence is emerging in support of the neuroprotective actions of CB 2 receptor agonists for the treatment of neurodegenerative diseases (81)(82)(83). Although the results generated from these experiments were obtained with primary cultured neurons, we cannot exclude a small contributing role for glia. However, in this study, the use of AM630, a selective CB 2 receptor antagonist, had no effect on the proclivity of 2-AG or AEA to confer neuroprotection (data no shown).
In contrast to AEA, the mechanism of 2-AG-mediated neuroprotection was not CB 1 receptor-dependent. Over the last decade, other receptor targets and oxygenation pathways for endocannabinoids have been identified. For example, numerous studies have now reported on the ability of 2-AG and AEA to activate peroxisome proliferator-activated receptors (84 -86) and putative cannabinoid receptors, such as GPR55 (87). In addition, 2-AG has been demonstrated to be a natural substrate for cyclooxygenase-2 (COX-2) (88, 89); thus, the possibility cannot be ruled out that downstream metabolites of 2-AG oxygenation account for the neuroprotective effects of 2-AG in our model of AD. Further experiments are required to decipher precisely how 2-AG confers neuroprotection. In light of these results, we propose that stabilization of lysosomes by endocannabinoids may represent a key mechanism by which these lipid molecules can confer neuroprotection.