Characterization of the Endocannabinoid System in Human Neuronal Cells and Proteomic Analysis of Anandamide-induced Apoptosis*

Anandamide (AEA) is an endogenous agonist of type 1 cannabinoid receptors (CB1R) that, along with metabolic enzymes of AEA and congeners, compose the “endocannabinoid system.” Here we report the biochemical, morphological, and functional characterization of the endocannabinoid system in human neuroblastoma SH-SY5Y cells that are an experimental model for neuronal cell damage and death, as well as for major human neurodegenerative disorders. We also show that AEA dose-dependently induced apoptosis of SH-SY5Y cells. Through proteomic analysis, we further demonstrate that AEA-induced apoptosis was paralleled by an ∼3 to ∼5-fold up-regulation or down-regulation of five genes; IgG heavy chain-binding protein, stress-induced phosphoprotein-1, and triose-phosphate isomerase-1, which were up-regulated, are known to act as anti-apoptotic agents; actin-related protein 2/3 complex subunit 5 and peptidylprolyl isomerase-like protein 3 isoform PPIL3b were down-regulated, and the first is required for actin network formation whereas the second is still function-orphan. Interestingly, only the effect of AEA on BiP was reversed by the CB1R antagonist SR141716, in SH-SY5Y cells as well as in human neuroblastoma LAN-5 cells (that express a functional CB1R) but not in SK-NBE cells (which do not express CB1R). Silencing or overexpression of BiP increased or reduced, respectively, AEA-induced apoptosis of SH-SY5Y cells. In addition, the expression of BiP and of the BiP-related apoptotic markers p53 and PUMA was increased by AEA through a CB1R-dependent pathway that engages p38 and p42/44 mitogen-activated protein kinases. Consistently, this effect of AEA was minimized by SR141716. In conclusion, we identified BiP as a key protein in neuronal apoptosis induced by AEA.

appears to be mediated by divergent pathways, each potentiated, reduced, or unaffected by cannabinoid or TRPV1 receptors (6 -9). They can also be dependent on membrane cholesterol content (10,11), generation of reactive oxygen species (12,13), or even release of ethanolamine from AEA catalyzed by FAAH (14). In this context, it should be stressed that data on ECS and AEA-induced apoptosis in human neuronal cells are still very scant. Furthermore, a proteomic analysis of AEA-induced apoptosis has never been performed, leaving open the question of which proteins (if any) might be modulated at the level of expression upon induction of programmed cell death by this endocannabinoid. On this basis, we sought to characterize through functional and immunochemical assays and confocal microscopy the ECS components in human neuroblastoma SH-SY5Y cells. In addition, we performed a proteomic analysis of AEA-induced apoptosis in these cells that identified five proteins whose expression was modified. Among these proteins, one (BiP/GRP78) was shown to be modulated by AEA in a CB1Rdependent manner, in parallel with apoptosis. Through silencing and overexpression experiments, it was found to be a key player in the death program. Rabbit anti-CB1R polyclonal antibodies were purchased from Affinity Bioreagents (Golden, CO). Rabbit anti-CB2R, anti-NAPE-PLD, and anti-MAGL polyclonal antibodies were from Cayman Chemical Co. (Ann Arbor, MI). Rabbit anti-DAGL antibodies were from Frontier Science Co. Ltd. (Hokkaido, Japan), and rabbit anti-TRPV1 and anti-␤-actin polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-BiP polyclonal antibodies were from Abcam (Cambridge, UK). Mouse monoclonal anti-cleaved PARP (Asp-214) antibodies were purchased from Cell Signaling Technology (Danvers, MA). Goat Alexa Fluor conjugates secondary antibodies, Image-iT TM FX signal enhancer, and Prolong antifade kit were from Molecular Probes (Eugene, OR). Goat anti-rabbit and goat anti-mouse antibodies conjugated to horseradish peroxidase (GAR-HRP and GAM-HRP, respectively) were from Santa Cruz Biotechnology.

Materials and Antibodies-Anandamide
Cell Culture and Determination of Apoptosis-Human neuroblastoma SH-SY5Y cells were grown in Dulbecco's modified Eagle's medium, supplemented with 15% inactivated fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin/streptomycin, 1 mM sodium pyruvate, 1 mM Hepes, and 1 mM nonessential amino acids. Human neuroblastoma SK-NBE cells and human neuroblastoma LAN-5 cells were grown in RPMI 1640 medium, supplemented with 20% inactivated fetal bovine serum, 1 mM sodium pyruvate, 1 mM Hepes. All cell lines were maintained at 37°C in a humidified 5% CO 2 atmosphere. SH-SY5Y cells were settled on a glass coverslip at a density of 2 ϫ 10 4 cells/cm 2 , and after 1 day they were treated for 24 h with different concentrations of AEA (in the range 0 -12 M), with 2-AG (12 M), or with vehicle alone (controls). SK-NBE and LAN-5 cells were treated for 24 h with 10 M AEA under the same experimental conditions. After each treatment, cell viability was measured by trypan blue exclusion (15). For quantitation of apoptosis, cells were fixed for 10 min at room temperature with 3% paraformaldehyde and then were permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 2 min at 4°C. After a blocking step in Image-iT TM FX signal enhancer for 30 min, cleaved PARP was identified by mouse anti-PARP antibody (diluted 1:500 in Image-iT TM FX signal enhancer) after incubation of cells for 1 h at room temperature. Secondary antibody conjugated to Alexa Fluor 488 was diluted 1:200 in Image-iT TM FX signal enhancer and was incubated with the specimens for 30 min at room temperature. Cell nuclei were counterstained with DAPI (0.1 g/ml in PBS). After washing, the coverslips were mounted using the antifade prolong Gold reagent and were visualized by Leica TCS SP confocal microscope. Pictures were taken using the LAS AF program (Leica Microsystems, Wetzlar, Germany), and then processed with Adobe Photoshop (Mountain View, CA) for brightness and contrast adjustments. The number of cells with apoptotic morphology was assessed in five randomly chosen fields. For each treatment, 200 -300 cells were counted. In SK-NBE and LAN-5 cells, and in BiP-silenced or overexpressing cells, apoptosis was quantified by fluorocytometric analysis. Cells were trypsinized and collected by centrifugation. Pellets were fixed with 70% ethanol for 12 h at Ϫ20°C, were centrifuged again, and were incubated in RNase solution (200 g/ml) at room temperature for 10 min, and then with DNA-staining solution with propidium iodide (50 g/ml) for a further 10 min. Then samples were scanned using a Coulter Epics XL (Beckman Coulter, Fullerton, CA) under conditions allowing the measurement of propidium iodide fluorescence. Apoptotic cells appeared as a hypodiploid (sub-G 1 ) peak because of nuclear fragmentation and loss of DNA (16).
Cell Transfection-To construct expression plasmids, total RNA from human neuroblastoma SH-SY5Y cells was extracted, and cDNA was synthesized by reverse transcription (see under "qRT-PCR Analysis"). A full-length C-terminal hemagglutinin epitope-tagged BiP was amplified by using the specific forward and reverse primers that included NheI and XhoI restriction sites, respectively. The PCR product was purified and cloned into pcDNA3 mammalian expression vector (Invitrogen). For transient transfection, SH-SY5Y cells were plated on 12-well plates (Iwaki, Tokyo, Japan), and after 24 h they were transfected with plasmid encoding hemagglutinin-tagged BiP, or empty vector as control, by using Lipofectamine 2000 (Invitrogen). Briefly, 1.6 g of plasmid and 4 l of Lipofectamine 2000 were mixed in a total volume of 200 l and were added to cells. After 6 h, 10 M AEA was added to the medium, alone or in the presence of 1 M SR141716, and 24 h later cells were harvested and subjected to either apoptosis quantitation or total protein extraction for Western blotting analysis.
Functional Assays of ECS, Metabolism of AEA-The synthesis of AEA through NAPE-PLD was assayed in SH-SY5Y cell homogenates (200 g of protein/test), by measuring the release of [ 3 H]AEA from 100 M [ 3 H]N-arachidonoylphosphatidylethanolamine by reverse phase-HPLC, coupled to on-line scintillation counting (17). The hydrolysis of AEA by FAAH in SH-SY5Y cell homogenates (50 g of protein/ test) was evaluated by reverse phase-HPLC, using 10 M [ 3 H]AEA as substrate (18). Both NAPE-PLD and FAAH activities were expressed as picomoles of product released per min per mg of protein.
Metabolism of 2-AG-To evaluate the synthesis of 2-AG by DAGL, SH-SY5Y cell homogenates (200 g of protein/test) were incubated at 37°C for 30 min with 500 M [ 14 C]1-stearoyl-2-arachidonoyl-sn-glycerol, as reported (19). Then a mixture of chloroform/methanol (2:1, v/v) was added to stop the reaction, and the organic phase was dried and fractionated by TLC on silica, using polypropylene plates with chloroform/methanol/ NH 4 OH (94:6:0.3, v/v) as eluent. The release of [ 14 C]2-AG was measured by cutting the corresponding TLC spots, followed by scintillation counting. The hydrolysis of 2-AG by MAGL was assayed in SH-SY5Y cell supernatants (100 g of protein/test), obtained at 39,000 ϫ g, and incubated with 10 M [ 3 H]2-oleoylglycerol at 37°C for 30 min (20). The reaction was stopped with a mixture of chloroform/methanol (2:1, v/v), and the release of [ 3 H]glycerol in the aqueous phase was measured in a ␤-counter (PerkinElmer Life Sciences). Both DAGL and MAGL activities were expressed as picomoles of product per min per mg of protein.
CBR and TRPV1 Receptor Binding-For cannabinoid receptor studies, SH-SY5Y, SK-NBE, or LAN-5 cells were resuspended in 2 mM Tris/EDTA, 320 mM sucrose, 5 mM MgCl 2 , pH 7.4, and then they were homogenized in a Potter homogenizer and centrifuged as reported (18). The resulting pellet was resuspended in assay buffer (50 mM Tris-HCl, 2 mM Tris/EDTA, 3 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, pH 7.4) to a protein concentration of 1 mg/ml. The membrane preparation was divided into aliquots that were stored at Ϫ80°C for no longer than 1 week. These membrane fractions (100 g of protein/test) were used in rapid filtration assays with the synthetic cannabinoid [ 3 H]CP55940 (400 pM), as described (18). Also binding of the TRPV1 agonist [ 3 H]RTX (500 pM) to SH-SY5Y cells was evaluated by rapid filtration assays (18). Unspecific binding was determined in the presence of cold agonists (1 M CP55940 or 1 M RTX) and was further corroborated by selective antagonists (0.1 M SR141716 for CB1R, 0.1 M SR144528 for CB2R, and 1 M I-RTX for TRPV1), as reported (18).
Fluorescence Microscopy Studies of ECS-SH-SY5Y cells were plated on collagen-coated glass coverslips and were fixed 24 h later using 3% paraformaldehyde plus 4% sucrose for 20 min at room temperature. Fixed cultures were permeabilized with 0.1% Triton X-100 in PBS for 10 min at room temperature. After a blocking step with 5% bovine serum albumin (BSA) in PBS for 45 min at room temperature, cells were incubated for 2 h with the following primary antibodies: anti-CB1R (1:400), anti-CB2R (1:400), anti-TRPV1 (1:100), anti-NAPE-PLD (1:400), anti-FAAH (1:200), anti-DAGL (1:200) or anti-MAGL (1:400), diluted in PBS with 5% BSA. Secondary antibodies conjugated with Alexa Fluor 488 were diluted 1:200 in PBS with 5% BSA and were incubated with the specimens for 1 h at room temperature. After washing, the coverslips were mounted using the antifade Prolong Gold Reagent and were visualized by Leica TCS SP confocal microscope (Leica Microsystems, Wetzlar, Germany). Pictures were taken using the LAS AF program (Leica Microsystems) and then were processed with Adobe Photoshop (Mountain View, CA) for adjustments of brightness and contrast.
RNA Interference-Synthetic ready-to-use small (21 nucleotides) interfering RNA (siRNA), complementary to a region of BiP, and nonsilencing control siRNAs were custom-synthesized by Qiagen (Tokyo, Japan). SH-SY5Y cells were transfected with 560 pmol of BiP siRNA using the Lipofectamine 2000 reagent. Briefly, 560 pmol of siRNA and 28 l of Lipofectamine 2000 were diluted with Opti-MEM to a final volume of 3 ml, mixed, and added to SH-SY5Y cells grown to 60% confluency in 10-mm diameter plates. After 6 h, cells were washed and cultured in fresh media for 24 h. The cells were then washed and cultured for a further 24 h in media containing 10 M AEA, alone or in the presence of 1 M SR141716, and finally qRT-PCR analysis and Western blotting were performed.
Immunochemical Analysis-Western blotting was performed according to standard procedures (18). The following antibody dilutions were used to immunodetect specific proteins: anti-CB1R ( (18).
Proteomic Analysis, Two-dimensional Gel Electrophoresis, and Image Analysis-SH-SY5Y cell pellets were sonicated in the presence of protease inhibitors in 0.1 M Tris-HCl, pH 7.2; for each sample 100 g of proteins were solubilized for 2 h in 350 l of rehydration buffer (4.4 M urea, 1.8 M thiourea, 3.5% w/v CHAPS, 1.4% carrier ampholytes, pH 3-10, 85 mM dithiothreitol, 0.02% w/v bromphenol blue). For isoelectric focusing, the Ettan IPGphor system (Amersham Biosciences) and pH 3-10 IPG strips (18 cm, nonlinear, Amersham Biosciences) were used according to the manufacturer's recommendations. The IPG strips were rehydrated for 1 h at 0 V and for 8 h at 30 V in rehydration buffer containing protein samples. Focusing was performed at 18°C, 50 A/strip, in different steps as follows: 1 h gradient to 300 V, 1 h at 300 V, 3 h gradient to 3500 V, 3 h at 3500 V, 1.5 h gradient to 8000 V, plateau at 8000 V until 50,000 V-h. The focused gel strips were equilibrated for 20 min in equilibration buffer (50 mM Tris-HCl, pH 8.8, 6 M urea, 2% w/v SDS, 30% v/v glycerol, 0.02% w/v bromphenol blue), containing 2% dithiothreitol, and for further 20 min in equilibration buffer containing 2.5% iodoacetamide. SDS-PAGE was performed with gradient gels (9 -16% acrylamide/ bisacrylamide), using the PROTEAN II xi Cell system (Bio-Rad). Each gel (18 ϫ 20 ϫ 1.5 cm) was run at 50 mA and 10°C until the bromphenol blue frontier reached the gel bottom. At least three replicates were performed for each sample. After fixing, the protein spots in analytical gels were visualized by Vorum Silver Staining (modified for MS) (21), and the stained gels were scanned with a UMAX Image Scanner (Amersham Biosciences). Image analysis was done using the PDQuest software (version 7.1.1; Bio-Rad). A reference gel was selected from one of the experimental gels, and unmatched protein spots of the remaining gels were manually added to this reference gel to create the Master Gel, a virtual image comprehensive of all matched spots derived from all analyzed samples. Subsequent to automatic spot detection and spot filtering, the matching of spots between gels was manually reviewed and adjusted as necessary. The abundance of each protein was estimated by the spot intensity. Only those with significant (at least 3-fold) and reproducible changes were considered to be differentially expressed proteins. Statistical analysis was performed apply-FIGURE 1. Endocannabinoid system in human SH-SY5Y cells. A, Western blotting of the ECS elements in SH-SY5Y cells. The expected molecular mass (M r ) of each protein is shown on the right-hand side. B-H, characterization of the ECS elements by immunofluorescence. SH-SY5Y cells were grown on glass coverslips to 40 -50% confluence for 24 h and then were stained with the indicated antibodies for confocal laser-scanning microscopy. Nonspecific immunofluorescence in these cells was negligible for all tested antibodies (data not shown). The subcellular localization of the receptors revealed weak cytoplasmic and nuclear staining for CB1R (B) and CB2R (C) and to a larger extent for TRPV1 (D). Immunoreactivities for the enzymes NAPE-PLD (E), FAAH (F), DAGL (G), and MAGL (H) were mainly restricted to internal membranes, possibly those of the endoplasmic reticulum. Pictures were taken using a confocal microscope equipped with a ϫ63 objective (numerical aperture ϭ 1.4). Images are representative of at least three independent experiments, and five fields were examined for each treatment. Bars, 10 m. ing the Student's t test with a confidence level of 95% to estimate the significance of the expression differences of protein spots. Protein Identification-Protein spots of interest were excised from two-dimensional gel electrophoresis gels, and in-gel digestion was carried out as reported (22) with a solution of porcine trypsin (Promega, Madison, WI) in 50 mM ammonium bicarbonate at 37°C for 16 -18 h. The reaction was stopped by adding a final concentration of 0.1% trifluoroacetic acid. For the mass spectrometric analysis of tryptic digests, samples were concentrated and desalted by reverse phase extraction using ZipTip C18 (Millipore, Bedford, MA). Elution and spotting on MTP 384 Ground Steel target plate (Bruker Daltonik GmbH, Bremen, Germany) were obtained using a 0.5 g/liter solution of ␣-cyano-4-hydroxycinnamic acid in 50% acetonitrile, 0.1% trifluoroacetic acid. Samples were analyzed with a Ultraflex III MALDI-TOF/TOF mass spectrometer (Bruker Daltonik GmbH), operating in a positive ion mode with a reflectron setup and equipped with a LIFT-cell; fragment ion analysis was performed by tandem mass spectrometry (MS/MS) after laserinduced dissociation. Mass spectra were externally calibrated with peptide standards from Bruker and internally calibrated with autodigest peaks of trypsin. Data obtained from both MS and MS/MS were used to search against the National Center for Biotechnology Information (NCBInr) human data base, using the peptide search routine MASCOT (Matrix Science, London, UK). Carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionine was allowed as variable modification of the peptides. The query was performed with the maximal tolerance of 100 ppm for peptide mass fingerprinting and a mass tolerance of 0.5 Da for MS/MS ion search. For positive identification protein scores Ͼ65 and MS/MS ion scores Ͼ35 were considered significant (p Ͻ 0.05), as calculated by the MASCOT scoring algorithm.
Statistical Analysis-Data reported in this paper are the means Ϯ S.D. of at least three independent experiments, each performed in triplicate. Unless stated otherwise, statistical analysis was performed by the nonparametric Mann-Whitney U test, elaborating experimental data by means of the InStat 3 program (GraphPad Software for Science, San Diego).

ECS in SH-SY5Y Cells-
The presence of an endocannabinoid system in human SH-SY5Y cells was demonstrated at the transcriptional, translational, and functional level. First of all, qRT-PCR analysis demonstrated mRNA transcripts for the main elements of the ECS, including CB1R, CB2R, TRPV1, NAPE-PLD, FAAH, DAGL, and MAGL. In particular, TRPV1 and MAGL transcripts were the most abundant, followed by NAPE-PLD, FAAH, and DAGL, whereas both cannabinoid receptors were expressed the least (Table 1). Next, translation of ECS mRNAs into the corresponding proteins was assessed by Western blotting (Fig. 1A) and immunofluorescence (Fig. 1, B-H). Western blotting showed that all ECS elements detected by qRT-PCR were also expressed at the translational level, although TRPV1 and MAGL proteins were not abundant as their transcripts (compare Table 1 and Fig. 1A). Confocal microscopy further supported the presence of ECS components in SH-SY5Y cells, where intense and specific immunostaining for FAAH, NAPE-PLD, TRPV1, and to a lower extent for CB1R, CB2R, DAGL, and MAGL was seen (Fig. 1,  B-H). Finally, the activity of the ECS elements was tested by biochemical assays, demonstrating that the enzymes and receptors detected in SH-SY5Y cells were all functional ( Table 2). In particular, it was found that the binding of the synthetic cannabinoid [ 3 H]CP55940, a CB1R and CB2R agonist (23), was displaced to ϳ50% of the controls by either 0.1 M SR141716, a selective CB1R antagonist, or 0.1 M SR144528, a selective CB2R antagonist (23). These data, shown in Table 2, indicate that both cannabinoid receptor subtypes were functional in human SH-SY5Y cells.
Induction of Apoptosis by AEA-AEA has been shown to trigger apoptosis at micromolar concentrations (10 -15), i.e. those found centrally and peripherally in a number of neurological disorders (see Ref. 24 and references therein). In effect micromolar concentrations of AEA are able to trigger significant neuronal loss (25). Here treatment of SH-SY5Y cells for 24 h with different amounts of AEA (in the range 0 -12 M) was able to reduce dosedependently cell viability, to ϳ60% of controls at 10 M (not shown). To ascertain whether the observed cytotoxicity was indeed because of programmed cell death, PARP cleavage, a hallmark of apoptosis, was measured. Immunofluorescence analysis revealed that AEA induced dose-dependently SH-SY5Y cell apoptosis, with ϳ40% of apoptotic cells at 10 M AEA ( Fig.  2A). Furthermore, 1 M SR141716 rescued the apoptosis induced by 10 M AEA, whereas SR144528 or I-RTX was ineffective at the same concentration (Fig. 2, B and C). Interestingly, 2-AG was unable to induce any apoptosis of SH-SY5Y cells even at 12 M ( Fig. 2A), further extending previous data on other cell systems (15).
Protein Expression Patterns of Control and AEA-treated SH-SY5Y Cells-To map the differential protein expression levels upon 24 h of exposure of human SH-SY5Y cells to 10 M AEA, alone or in the presence of 1 M SR141716, we performed a proteomic analysis. Proteins were separated by two-dimensional gel electrophoresis and visualized by silver staining (Fig.  3). Experiments were performed in triplicate. An average of 372 Ϯ 3 (mean Ϯ S.D.) spots were resolved on the SH-SY5Y gels, and 326 Ϯ 5 spots and 320 Ϯ 2 spots were visible for AEAand AEA ϩ SR141716-treated cell maps, respectively. After digital image analysis, the abundance of each single protein spot in gels from different cell preparations was estimated by its intensity, and the results were evaluated in terms of spot units of signal intensity (absorbance) expressed as parts/million (ppm). The ratios of the mean normalized spot intensity of treated versus control gels were calculated for spots identified in every three replicates of each sample. Five proteins were chosen for further analysis, because they were present in AEAtreated cells with at least a 3-fold (p Ͻ 0.05) up-or down-regulation of absorbance values versus controls (Fig. 4). The selected  proteins were in-gel digested with trypsin and the extracted peptides were analyzed by either MALDI-TOF-MS mass fingerprint or MALDI-TOF/TOF-MS/MS peptide sequencing. For both approaches, the acquired spectral data were compared with the theoretical molecular weight and fragmentation pattern of peptide sequence in a data base to identify the proteins by means of probabilistic assignments (p Ͻ 0.05). This analysis identified BiP (GRP78), stress-induced phosphoprotein-1 (STIP1), and triose-phosphate isomerase-1 (TPI1) as the three proteins whose expression was up-regulated by more than 3-fold over controls and actin-related protein 2/3 complex subunit 5 (ARP5) and peptidylprolyl isomerase-like protein 3 isoform PPIL3b (PPIL3b) as the two proteins whose expression was down-regulated by more than 3-fold to the controls (Table  3). We performed a literature search to find out the function(s) of these proteins, and the results are summarized in Table 4. Remarkably, we found that only the effect of AEA on BiP protein expression was significantly reduced by SR141716 (Fig. 4), under the same conditions that minimized AEA-induced apoptosis (Fig. 2, B and C).
mRNA Expression of Proteins Modulated by AEA in Apoptotic Cells-To corroborate the proteomic analysis data, we evaluated the transcriptional abundance of the five proteins modulated upon AEA-induced apoptosis of SH-SY5Y cells. We found that mRNA levels followed the same trend of protein expression (Fig. 5). In particular, only the increase of BiP mRNA induced by AEA was significantly reduced by SR141716, in parallel with the reduced apoptosis. To further extend these data, we used two other human neuroblastoma cell lines, SK-NBE and LAN-5, after checking for the presence of functional CB1 receptors. Although LAN-5 cells express the receptor at the mRNA and protein level and show CB1R binding comparable with that of SH-SY5Y cells, SK-NBE cells do not (Fig. 6A). After treatment with 10 M AEA, in the presence or absence of 1 M SR141716, apoptosis (Fig. 6B) and BiP mRNA expression (Fig.  6C) in LAN-5 cells but not in SK-NBE cells were found to be similar to that observed in SH-SY5Y cells, further supporting that AEA-induced apoptosis and up-regulation of BiP were both mediated by CB1R.
Involvement of BiP in AEA-induced Apoptosis-Thus a role of BiP in the AEA-induced apoptosis of human neuronal cells seems likely. Therefore, we further explored whether the modulation of BiP upon AEA treatment was linked by a cause-effect relationship to the induction of programmed cell death, by silencing or overexpressing BiP gene in SH-SY5Y cells. Transfection of these cells with BiP-siRNA or with a plasmid encoding hemagglutinin-tagged BiP efficiently reduced or enhanced, respectively, endogenous BiP mRNA (Fig. 7A) and protein level (Fig. 7B). The modulation of BiP expression was specific, as suggested by the lack of effect on the amount of ␤-actin in the same cells (Fig. 7B). In addition, nonsilencing siRNAs, used as a control, did not affect either mRNA or protein levels of BiP under the same experimental conditions (data not shown). Treatment with 10 M AEA significantly enhanced BiP mRNA level in SH-SY5Y cells where BiP had been silenced, whereas 1 M SR141716 counteracted this effect of AEA (Fig. 7A). Instead 10 M AEA did not increase any further the BiP mRNA level in SH-SY5Y cells that overexpressed it (data not shown), possibly due to a saturating effect. Next, we investigated the effect of BiP   silencing or overexpression on apoptosis, by fluorocytometric analysis (Fig. 7C). We found that after the administration of 4 M AEA in cells where BiP was silenced (siBiP), the apoptosis was almost ϳ4-fold higher than that in untreated siBiP cells (ϳ20% versus ϳ5%), whereas the same amount of AEA only doubled apoptosis (ϳ4% versus ϳ2%) in control cells. Administration of 1 M SR141617 minimized AEA-induced apoptosis in both cases (Fig. 7C). On the other hand, 4 M AEA failed to induce apoptosis in cells that overexpressed BiP (data not shown). We further extended our study to two BiP-related apoptotic markers, p53 and PUMA. As shown in Table 5, the level of these gene products increased after treatment of SH-SY5Y cells with 10 M AEA, and again this effect of AEA was reversed by 1 M SR141716.

Related information of the five proteins differentially expressed in AEA-treated SH-SY5Y cells and identified on the two-dimensional gels by MALDI-TOF/TOF-MS and/or MS/MS
Mechanism of AEA-induced Apoptosis and BiP Expression through CB1R-CB1R activation by AEA triggers two common signaling pathways that engage p38 and p42/44 MAPK (26,27). Therefore, we used selective inhibitors of p38 and p42/44 MAPKs at concentrations known to inhibit their target enzymes (28,29). Our results show that SB203580, a selective inhibitor of p38 MAPK (28), and PD98059, a selective inhibitor of p42/44 MAPK (29), were able to fully prevent the AEA-in- duced increase of mRNA expression of BiP and of the BiPrelated apoptotic markers p53 and PUMA (Table 5). Altogether, these data suggest that the apoptotic activity of AEA is mediated by CB1R, through a signaling cascade that triggers p38 and p42/44, leading to increased expression of the ER-stress sensor BiP, and hence of the apoptotic markers p53 and PUMA.

DISCUSSION
In this study we demonstrate the presence of a fully functional endocannabinoid system in human neuroblastoma SH-SY5Y cells. In addition, we report the first proteomic analysis of the protein expression modulation in these cells upon induction of apoptosis by anandamide, which identifies BiP as a key protein in the apoptotic program induced by AEA in human neuronal cells.
Human SH-SY5Y cells were chosen for this study, because they have been widely used to investigate neuronal cell damage and death. For instance, they served as an in vitro model for Parkinson disease and associated neuronal apoptosis (30 -35) to evaluate the protective effects of anti-apoptotic substances in a cellular model of Alzheimer disease (36) and to replicate in vitro some pathological changes observed in the brain of mice affected by the latter disease (37,38). In short, human SH-SY5Y cells are largely employed to evaluate neurotoxic conditions that lead to apoptosis (39 -42). Remarkably, endocannabinoid signaling has been shown to play a central role in both Parkinson and Alzheimer diseases, as well as in other major human neurodegenerative or neuroinflammatory disorders (for a recent review see Ref. 43). In addition, the ECS tightly controls the cell choice between life and death (6 -8), with a therapeutic impact for the treatment of cancer (44). Therefore, the demonstration of a complete and functional ECS should allow a better understanding of the role of ECS-related compounds and ECS-targeted drugs, in models of neurotoxicity based on SH-SY5Y cells.
Another major outcome of this study is the identification of five different proteins, whose expression is up-regulated (BiP, STIP1, and TPI1) or downregulated (ARP5 and PPIL3b) more than ϳ3-fold over controls in AEA-induced apoptotic SH-SY5Y cells. This is the first proteomic approach to the understanding of AEA pro-apoptotic activity. By the way, the first ever proteomic analysis on the effects of the cannabis-derived ⌬ 9 -tetrahydrocannabinol in chronic users (45), and in mouse cerebellum (46), appeared during the preparation of this manuscript.
It seems noteworthy that the three proteins that are up-regulated by AEA are known to act as anti-apoptotic agents. In fact, BiP/GRP78 is a molecular chaperone of the heat shock protein 70 (Hsp70) family and belongs to a defense mechanism against endoplasmic reticulum (ER) stress induced by accumulation of unfolded protein aggregates (unfolded protein response). BiP normally leads to a general inhibition of protein translation, to an increase in chaperone transcription and translation, and to a cell cycle arrest; these processes allow rapid restoration of protein homeostasis and avoid cell death due to the initiation of apoptotic pathways (47,48). In addition, up-regulation of BiP is mediated by the ERK1/2 pathway (49), and it stabilizes mitochondrial permeability in cells treated with ER stress inducers (50,51). These observations suggest that BiP may inhibit apo- ptosis via a mitochondria-dependent route (52). In line with this, AEA induces apoptosis in neuronal cells through an ERK1/2-and mitochondria-dependent pathway (11). Here we have shown that programmed cell death and BiP expression are both dependent on p38 and p42/44 MAPK signaling triggered through CB1R. Therefore, the increase of BiP in apoptotic SH-SY5Y cells might represent a neuroprotective mechanism, aimed at counteracting AEA-induced cell death. For instance, BiP is up-regulated in neurons of Alzheimer disease brains, where this protein has indeed been shown to play a neuroprotective role (53,54). Consistently, we observed that overexpression of BiP confers resistance to AEA-induced apoptosis in human neuronal cells. Conversely, programmed cell death induced by AEA was paralleled by the increased expression of two marker genes that are strictly related to ER stress-induced apoptosis, p53 and PUMA. The latter is known to be significantly up-regulated after prolonged exposure to classical ER stress inducers (55), and it has been previously identified as a p53 target (56,57). Human cells deficient in PUMA expression show a dramatically reduced apoptosis in response to ER stress (55), and induction of PUMA has been observed also in primary neurons and in response to cerebral ischemia in vivo, altogether suggesting that regulation of PUMA expression may be a common pathway during ER stress-induced cell death in mammals. Therefore, the death program triggered by AEA seems to engage typical executioners of ER stress-induced cell death (p53 and PUMA), along with the specific defense mechanism that counteracts it through BiP. This effect seems to extend well beyond a single cell type and to be true also for human neuroblastoma cells that express CB1R.
On the other hand, other four proteins were affected by AEA at the transcriptional and translational level, although their modulation did not depend on CB1R, at variance with AEAinduced apoptosis. STIP1 is an Hsp70/Hsp90-organizing protein (58), and it regulates protein folding and proteasomal degradation upon induction of neuronal apoptosis by staurosporine (42). Therefore, the up-regulation of STIP1 in apoptotic cells seems to be neuroprotective similarly to that of the chaperone BiP. TPI1 is a homodimeric enzyme, which catalyzes the interconversion of dihydroxyacetone phosphate and glyceraldehyde 3-phosphate (59), and it has an important role in glycolysis, gluconeogenesis, fatty acid synthesis, and the pentose cycle (60). Several reports have shown that TPI1 can be induced as a stress protein in mouse brain capillary endothelial cells (61) and can be up-regulated through an AP-1 (activator protein-1)dependent pathway (62). It seems noteworthy that inhibition of TPI1 induces progressive and selective death of cortical and striatal neurons (63), and that in an in vitro model of acute pancreatitis it was overexpressed, suggesting an anti-apoptotic role (64). Therefore, also the up-regulation of this protein might contribute to neuroprotection against cell death induced by AEA. On the other hand, the role of the down-regulation of ARP5 and PPIL3b remains to be clarified. In fact, the ARP complex is required for the formation of branched networks of actin filaments in the cell cortex (65). It has been implicated in abnormal cellular signal transduction, abnormal morphology of cancer cells, and even cancer development and progression (66). No information is yet available on the functional role of PPIL3b, but this protein arises from a splicing variant of the cyclophilinlike gene PPIL3, whose mRNA was found to be ubiquitously transcribed in adult human tissues, low expressed in human brain tissues, and overexpressed in human glioma tissues (67,68). Recently, a direct binding between PPIL3 and the Apoptin protein was reported in cancer cells, where a PPIL3-controlled nuclear-cytoplasmic distribution of Apoptin suggested a possible role in apoptosis (69).
In conclusion, we have demonstrated the presence of a complete and fully functional endocannabinoid system in human SH-5YSY cells, a widely used model for the study of neuronal cell damage and death, as well as for major human neurological disorders. In addition, we performed a proteomic analysis that identifies BiP as a key protein in AEA-induced apoptosis of human neuronal cells. Incidentally, a recent study has documented the proapoptotic and potential antitumoral action of (endo)cannabinoids in glioma stem-like cells (70), which is in line with our work. We believe that an investigation on the regulation of cell survival and death by endocannabinoids in stem-derived neuronal cells is worthwhile also for therapeutic exploitation. In fact, endocannabinoid-targeted drugs have been considered for the treatment of neurodegenerative diseases (43) and cancer (44), where apoptosis plays a central role and stem cells hold the promise of opening novel avenues of intervention.