Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The discovery of specific cannabinoid receptors and a family of endogenous cannabinoid ligands (endocannabinoids) has led to intensive research on the role of the cannabinoid system in physiology and in pathological conditions. The CB1 receptor subtype is distributed throughout the brain (Devane et al., 1988), whereas the CB2 receptor is located primarily in peripheral tissues (Munro et al., 1993). The brain also produces anandamide, 2-arachidonoylglycerol, and other endogenous cannabinoid ligands of the eicosanoid class (Devane et al., 1992; Stella et al., 1997). Clinically important responses to ⁹-tetrahydrocannabinol (⁹-THC) and synthetic cannabinoids include the attenuation of nausea and vomiting in cancer chemotherapy, stimulation of appetite in wasting syndromes, and reduction in intestinal motility (Robson, 2001). In addition, cannabinoids may have a role in the treatment of neurological disorders, including spasticity associated with multiple sclerosis or spinal cord injury, movement disorders, epilepsy, and pain (Robson, 2001).

Neuronal cannabinoid signaling seems to regulate cell death and survival. Cannabinoids protect the brain from insults such as excitotoxicity (Shen and Thayer, 1998), hypoxia and ischemia (Nagayama et al., 1999; Sinor et al., 2000), and trauma (Panikashvili et al., 2001). Conversely, under some conditions, cannabinoids can induce neuronal apoptosis (Campbell, 2001).

Various signal transduction pathways have been shown to be involved in the action of cannabinoids, which exert most of their known effects through the CB1 receptor. This G protein-coupled receptor signals inhibition of adenylate cyclase and of N-and P/Q-type channels (Pertwee, 1997; Twitchell et al., 1997; Wilson et al., 2001) and activation of mitogen-activated protein kinases (Bouaboula et al., 1995) and protein kinase B (Gomez del Pulgar et al., 2000). CB1 cannabinoid receptors may also signal independently of G_i/o proteins through an adaptor protein (neutral sphingomyelinase activation-associated factor, or FAN) that has been implicated in the proapoptotic pathway involving sphingomyelin breakdown and ceramide accumulation (Sanchez et al., 2001). In addition, cannabinoids exert some effects independently of the CB1 and CB2 receptors, such as through gap junctions (Venance et al., 1995), T-type Ca²⁺ channels (Chemin et al., 2001), the vanilloid receptor (Zygmunt et al., 1999) or non-CB1/CB2 cannabinoid receptors (Jarai et al., 1999; Breivogel et al., 2001; Hajos et al., 2001).

Microarray analysis is a powerful tool for detecting differential gene expression in neural tissues (Greenberg, 2001), including changes in gene expression associated with the effects of drugs (Thibault et al., 2000). One recent study described changes in the hippocampal expression of 49 of 24,456 arrayed cDNAs after short- (24 h) or long-term (7-21 days) administration of ⁹-THC to rats (Kittler et al., 2000). To better understand the molecular processes involved in cannabinoid action in vivo, which might suggest mechanisms through which these compounds control cell death and survival decisions, we examined changes in gene expression in the brain after short-term administration to mice of two different cannabinoid receptor agoniststhe naturally occurring plant alkaloid ⁹-THC and the synthetic aminoalkylindole R(+)-WIN 55,212-2. We used large-scale cDNA microarrays that contain probes for ~11,000 known genes and expressed sequence tags selected from a mouse brain cDNA library. Results revealed a total of 65 genes altered by ⁹-THC exposure, 41 genes altered by R(+)-WIN 55,212-2 exposure, and 20 genes altered by exposure to both drugs. Genes affected similarly by ⁹-THC and R(+)-WIN 55,212-2 are likely to be involved in cannabinoid receptor-mediated signaling, whereas genes uniquely affected by one or the other drug may reflect nonreceptor effects.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Drugs. Drugs were purchased from Sigma-Aldrich (St. Louis, MO). ⁹-THC was provided dissolved in 100% methanol at 1 mg/ml; methanol was evaporated and ⁹-THC was redissolved in dimethyl sulfoxide (DMSO) at 5 mg/ml. R(+)-WIN 55,212-2 was also dissolved in DMSO, at 0.5 mg/ml.

Animals and Drug Treatment. Animal experiments were carried out in accordance with National Institutes of Health guidelines and were approved by local committee review. Male CD1 mice (Charles River Laboratories) 2 to 3 months old and weighing 25 to 30 g were housed three per cage and maintained on a 12-h/12-h light/dark cycle, with food and water provided ad libitum, for 1 week before cannabinoid treatment. Mice were given a single 60-µl intraperitoneal injection of R(+)-WIN 55,212-2 (1 mg/kg), ⁹-THC (10 mg/kg), or DMSO vehicle. Compared with vehicle, both drugs decreased spontaneous locomotion during the first hour after short-term injection, and ⁹-THC transiently reduced rectal temperature by 1 to 2°C. When mice were killed 12 h later, drug- and vehicle-treated mice were indistinguishable. Whole brains were cut in half, the brainstem and cerebellum were removed, and hemi-forebrains were immediately frozen in dry ice and stored at 80°C until mRNA and protein extracts were prepared.

Microarrays. National Institutes of Mental Health/National Institute of Neurological Disorders and Stroke Brain Molecular Anatomy Project (BMAP) cDNA microarrays, containing 11,200 mouse genes isolated from neural tissues and 32 control genes, were prepared by the Buck Institute's Genomics Core. cDNAs were dissolved at 40 µM in 3× SSC buffer (1× SSC is 140 mM NaCl and 15 mM sodium citrate) and printed on AminoSilane-treated glass microscope slides (CEL Associates, Houston, TX) using a high-performance OmniGrid microarrayer (GeneMachines, San Carlos, CA). Two separate copies of the array were printed per slide. Printed slides were stored in a light-tight box in a bench-top desiccator at room temperature.

Probe Synthesis and Hybridization. Hemi-forebrains from six mice treated with R(+)-WIN 55,212-2, ⁹-THC, or vehicle were pooled and poly(A) RNA was isolated using Fast Track mRNA isolation kits (Invitrogen, Carlsbad, CA). Fluorescence-labeled probes were prepared by reverse transcription using a superscript II polymerase (Invitrogen) from Cy3-dCTP or Cy5-dGTP and 1 µg of poly(A) RNA. Cy3 was used to label probes from vehicle-treated mice and Cy5 was used to label probes from drug-treated mice. Probes were purified using Millipore columns (Millipore Corporation, Bedford, MA) vacuum-dried, and resuspended in 20 µl of hybridization buffer (10× SSC/50% formamide/0.02% SDS), heated to 90°C for 5 min, cooled to room temperature for 5 min, and applied to slides. The two separate arrays per slide were hybridized with probes from vehicle- and R(+)-WIN 55,212-2-treated or from vehicle- and ⁹-THC-treated mice. Arrays were separately cover-slipped and hybridized at 60°C for 16 h in a sealed chamber (Corning, Corning, NY), then washed in 0.5× SSC/0.01% SDS at room temperature for 15 min with gentle agitation and in 0.06× SSC/0.01% SDS at room temperature for 5 min. After rapid rinsing in 0.1× SSC, slides were dried by centrifugation at low speed before scanning.

Microarray Scanning, Quantitation, and Statistical Analysis. Microarrays were scanned for Cy3 and Cy5 fluorescence using a ScanArray 3000 microarray scanner (General Scanning, Watertown, MA). QuantArray software (GSI Lunonics, Watertown, MA) was used for quantitation. Spot intensity was corrected by subtracting the immediate background. Background-subtracted element signals were used to calculate Cy3/Cy5 ratios. Each experiment was performed three times and fold-changes in expression were averaged. The selection criteria for differentially expressed genes were a minimum expression ratio of 1.8 on three of three arrays and an average expression ratio of at least 2.0.

Analysis of Protein Expression. For Western blotting, MRG-1, GK, and actin protein expression was assayed in 100-µg protein samples from the hemi-forebrain not used for microarray studies. Samples from six mice treated with R(+)-WIN 55,212-2, ⁹-THC, or vehicle were pooled, and protein extracts were boiled in reducing SDS sample buffer for 10 min and electrophoresed on a 12% SDS-polyacrylamide gel, then transferred electrophoretically to polyvinylidene difluoridine membranes at 4°C and 200 mA overnight. Membranes were probed with goat polyclonal antibody against the amino terminus of human MRG-1 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), sheep polyclonal anti-GK (1:5000; a gift from Dr. M.A. Magnuson, Dept of Molecular Physiology and Biophysics, Vanderbilt University, Nashville, TN), or monoclonal anti-actin (1:500; Santa Cruz Biotechnology), and then with horseradish peroxidase-conjugated anti-goat (1:5000), anti-sheep (1:5000), or anti-mouse (1:2000) IgG secondary antibody (Santa Cruz Biotechnology), and visualized by chemiluminescence. For immunohistochemistry, sections were treated with 1% H₂O₂ for 15 min (for diaminobenzidine detection) and then placed in blocking buffer containing 5% horse serum and 0.2% Triton X-100 in PBS for 1 h at room temperature. Sections were incubated overnight at 4°C with the following primary antibodies: anti-MRG-1 (as above, 1:100), anti-GK (as above, 1:2000), rat monoclonal anti-PECAM-1/CD31, (1:50; Cymbus Biotechnology, Chandlers Ford, UK), or mouse monoclonal anti-neuronal nuclear antigen NeuN (1:500; Chemicon, Temecula, CA). Sections were washed in PBS containing 0.1% Tween 20 and immunostaining was completed using the following secondary antibodies: biotinylated anti-goat and anti-sheep IgG (1:200; Vector Laboratories, Burlingame, CA), fluorescein-conjugated anti-goat and anti-sheep IgG (1:200; Vector Laboratories), and rhodamine-conjugated anti-mouse and anti-rat IgG (1:200; Jackson ImmunoResearch, West Grove, PA). The peroxidase reaction was detected with 0.05% diaminobenzidine in PBS and 0.03% H₂O₂. As controls, alternating sections were incubated without primary antibody.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Patterns of Altered Gene Expression Induced by Cannabinoids. Eighty-six genes, representing somewhat less than 1% of the arrayed cDNAs, showed 2-fold or greater changes in expression in the brain after systemic administration of ⁹-THC or R(+)-WIN 55,212-2. Of these genes, 52% were affected only by ⁹-THC, 25% were affected only by R(+)-WIN 55,212-2, and 23% were affected by both cannabinoids. More genes were down-regulated (72%) than were up-regulated (28%), although this ratio varied across treatments. Thus, 93% of ⁹-THC-regulated genes and 65% of genes regulated by both ⁹-THC and R(+)-WIN 55,212-2, but only 33% of genes regulated by R(+)-WIN 55,212-2, exhibited reduced expression.

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Functional classes of genes showing cannabinoid-induced changes in expression. Each bar represents the percentage of all genes regulated by THC only, WIN only, or both drugs that belong to a given functional class of genes.

Genes Regulated by ⁹-THC. Forty-five genes were differentially expressed after ⁹-THC treatment, but not after treatment with R(+)-WIN 55,212-2 (Table 1). Only two of these genes were up-regulated: aspartyl-tRNA synthase and neuromedin U. Neuromedin U is a peptide that is expressed at high levels in the ventromedial hypothalamus, and its administration suppresses food intake and increases core body temperature (Howard et al., 2000). Because these effects are opposite those produced by ⁹-THC, enhanced expression of neuromedin U could represent a homeostatic response to the drug. Of the genes that showed ⁹-THC-induced decreases in expression, 3-fold changes were seen for the small inducible cytokine family D member 1 (neurotactin), mouse myelin proteolipid protein, neuron-specific gene family member 1, copine 6, APP-binding protein, and monoglycerate lipase. Neurotactin is a proinflammatory chemokine that is up-regulated in brain microglia in experimental autoimmune encephalomyelitis (Pan et al., 1997), suggesting that its down-regulation by ⁹-THC could contribute to the drug's anti-inflammatory effect. Copine 6, a calcium-dependent phospholipid-binding protein expressed in hippocampus, is up-regulated by stimuli that evoke long-term potentiation (Nakayama et al., 1998), which is impaired by cannabinoids.

View this table: [in this window] [in a new window]   TABLE 1 Genes specifically modified in mouse brain 12 h after  9-THC administration Expression levels are ratios of  9-THC-treated versus control samples. Each mRNA sample group was hybridized to three different arrays, and fold-change values are average of the triplicate measurements. Matches of clone sequences from cDNA arrays were based on the closest calculated expect value (E).

Genes Regulated by R(+)-WIN 55,212-2. Twenty-one genes were regulated only by R(+)-WIN 55,212-2 (Table 2). The greatest up-regulation was observed for the 140-kDa subunit of replication factor C, which promotes cell survival after DNA damage (Pennaneach et al., 2001). R(+)-WIN 55,212-2 also up-regulated Hsp27, which has been implicated in the neuroprotective effect of ischemic preconditioning (Currie et al., 2000), and cyclin-dependent protein kinase 1, which promotes neuronal survival from apoptotic insults (Courtney and Coffey, 1999), whereas the proapoptotic ubiquitin-protein ligase Nedd4 (Anan et al., 1998) was down-regulated.

Genes Regulated by both ⁹-THC and R(+)-WIN 55,212-2. Twenty genes showed changes in expression associated with both ⁹-THC and R(+)-WIN 55,212-2; in all cases, the direction of the change induced by both drugs was the same (Table 3). Genes down-regulated in common constituted 31% of those down-regulated by ⁹-THC and 35% of those down-regulated by R(+)-WIN 55,212-2; for up-regulated genes, the corresponding percentages were 70 and 33%. Because genes regulated by two structurally dissimilar cannabinoids seemed most likely to be affected through a specific, receptor-mediated mechanism, we focused most of our attention on this group of genes. Known genes regulated by both ⁹-THC and R(+)-WIN 55,212-2 included genes involved in neuronal signaling, neuronal growth and structure, myelination or glial differentiation, and metabolism. Considering the neuroprotective effects of cannabinoids and the proposed role of endogenous cannabinoid signaling in regulating cell death and survival, certain up-regulated genes were of particular interest, including the enzyme GK (Alvarez et al., 2002; Roncero et al., 2000) and the transcription factor MRG1 (Sun et al., 1998; Bhattacharya et al., 1999).

View this table: [in this window] [in a new window]   TABLE 3 Genes modified in mouse brain 12 h after both  9-THC and R(+)-WIN 55,212-2 administration Expression levels are ratios of R(+)-WIN 55,212-2- or  9-THC-treated versus control samples. Each mRNA sample group was hybridized to three different arrays, and fold-change values are average of the triplicate measurements. Matches of clone sequences from cDNA arrays were based on the closest calculated expect value (E).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Western analysis of cannabinoid-induced changes in the expression of MRG1 and GK proteins in rat brain. A, Western blots show MRG1, GK, and actin immunoreactivity after treatment with vehicle (V), R(+)-WIN 55,212-2 (Win), or 9THC (THC). B, bar graphs show the relative optical density of the MRG1 and GK bands under each treatment condition. Values were normalized to the optical density of the actin band and were expressed as a percentage of optical density in vehicle-treated mice. Data shown are means ± S.D. from three experiments.

View larger version (42K): [in this window] [in a new window]   Fig. 3.   Immunohistochemical localization of MRG1 (a-d) and GK (e-h) in the brains of cannabinoid-treated mice. MRG1 was localized to blood vessels (a, brown), and MRG1 (b, green) and the endothelial cell marker CD-31 (c, red) were colocalized (d, yellow); nuclei are stained with 4,6-diamidino-2-phenylindole (c, blue). GK was localized preferentially to the hypothalamus (e, brown), and GK (f, green) and the neuronal marker NeuN (g, red) were colocalized (h, yellow). Asterisks in e-h indicate the third ventricle. Data shown are representative fields from three experiments that gave similar results.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The major finding of this study is that the cannabinoids ⁹-THC and R(+)-WIN 55,212-2 induce distinct but overlapping transcriptional responses in mouse brain, measured 12 h after a single systemic dose of either drug. The overlapping responses are likely to reflect effects of cannabinoid receptor activation. CB1 is the predominant established cannabinoid receptor subtype in the brain and can bind both ⁹-THC and R(+)-WIN 55,212-2, so transcriptional responses shared by these drugs are likely to be triggered by CB1 activation. Non-CB1/non-CB2 ("CB3") cannabinoid receptors are also thought to exist in brain (Breivogel et al., 2001) but are insensitive to ⁹-THC. Therefore, although some or all of the genes induced only by R(+)-WIN 55,212-2 in our study might be induced through CB3 receptors, those genes induced by ⁹-THC alone or by both ⁹-THC and R(+)-WIN 55,212-2 are presumably not.

In a previously published study of ⁹-THC -regulated gene expression (Kittler et al., 2000), the genes identified were different from what we observed. There are several explanations for the discrepancy, including differences in species (mouse versus rat), brain region studied (entire forebrain versus hippocampus), time between drug administration and sacrifice (12 h versus 24 h to 21 days) and in the genes present on the different arrays (the Kittler study included approximately twice as many genes). We chose our conditions based on an interest in what genes might account for neuroprotective effects of cannabinoids in focal cerebral ischemia from middle cerebral artery occlusion (Nagayama et al., 1999), which affect the forebrain prominently and occur during the first 24 h after onset. In contrast, Kittler et al. were interested in genes involved in tolerance to cannabinoids (Kittler et al., 2000). A microarray study of genes induced by the cannabinoid agonist CP 55,940 in HL-60 promyelocytic cells transfected with the CB2 receptor also showed no overlap with the genes induced in our study (Derocq et al., 2000).

Cannabinoids affect a variety of tissues, which helps to account for the diversity of their actions. Central nervous system effects on motor activity, memory, and pain are related to interaction with neuronal CB1 receptors; hypotension seems to be mediated through vascular CB1 receptors; and CB2 receptors on mono- and polymorphonuclear leukocytes and macrophages have been implicated in immunosuppressive and anti-inflammatory actions. One prominent central effect of cannabinoids is the stimulation of appetite, which may underlie the reported therapeutic value of the cannabinoid drug dronabinol in the AIDS wasting syndrome (Robson, 2001). Intrahypothalamic injection of the endocannabinoid anadamide produces hyperphagia in rats that is blocked by the CB1 antagonist SR141716A, consistent with an effect mediated through CB1 receptors (Jamshidi and Taylor, 2001). SR141716A also reduces food intake in wild-type but not CB1 knockout mice, and the anorexigenic hormone leptin reduces hippocampal levels of two endocannabinoidsanandamide and 2-arachidonoyl glycerol (Di Marzo et al., 2001). Our finding that ⁹-THC and R(+)-WIN 55,212-2 increase GK protein expression in hypothalamus is of interest in this regard, considering that GK may have an important role in glucose sensing and metabolic regulation in the brain (Roncero et al., 2000; Alvarez et al., 2002).

Cerebrovascular effects of cannabinoids have also been observed, and these seem to be mediated through CB1 receptors on vascular smooth muscle and endothelial cells (Hillard, 2000). Cannabinoids are neuroprotective in conditions, such as cerebral ischemia (Nagayama et al., 1999) and trauma (Panikashvili et al., 2001), in which vascular factors play an important role, and a vascular component to cannabinoid-induced neuroprotection has been proposed (Panikashvili et al., 2001). Our finding that cannabinoids induce different genes in neurons (e.g., GK) and in blood vessels (e.g., MRG1) suggests that the molecular basis for vascularly and neuronally mediated neuroprotection may be different. More generally, if cannabinoids activate different signaling pathways in different types of cells, therapeutic approaches that target such downstream events may be capable of dissociating desirable from undesirable effects of cannabinoids.