Cannabinoid Receptors in Conjunctival Epithelium: Identification and Functional Properties

Abstract

purpose. Preservation of the ocular surface barrier requires complex control of epithelial cell proliferation and inflammation mechanisms. The endocannabinoid system may be regulating these processes. Therefore, the authors explored the presence and properties of cannabinoid receptors (CB1 and CB2) in conjunctival epithelial cells.

methods. The authors used immunohistochemistry to detect CB1 and CB2 in normal mouse conjunctiva, human conjunctival cryosections and impression samples, and IOBA-NHC cells, a human conjunctiva-derived cell line. The presence of CB1 and CB2 proteins and transcripts was studied in IOBA-NHC cells by Western blot and RT-PCR, respectively. The authors also used this cell line to assay cannabinoid ligand-induced changes in cAMP levels, cell growth, and tumor necrosis factor-α (TNF-α)–induced activation of c-jun N-terminal kinase (JNK) and nuclear factor-κB (NF-κB).

results. Mouse and human conjunctival epithelial cells displayed CB1 and CB2 proteins and transcripts. Cannabinoid receptor activation decreased cAMP levels in IOBA-NHC cells, and specific CB1 and CB2 antagonists canceled this effect. Cannabinoid ligands also increased cell growth and blocked stress pathways activated by TNF-α in vitro.

conclusions. Cannabinoid receptors are present in mouse and human conjunctival cells. Functional responses, such as decreased cAMP levels, proliferation, and modulation of stress signaling pathways, were mediated by CB1 and CB2 stimulation. Thus, these receptors might be involved in the regulation of epithelial renewal and inflammatory processes at the ocular surface.

The endocannabinoid system consists of cannabinoid receptors, endogenous cannabinoids (endocannabinoids), and the enzymes that synthesize and degrade endocannabinoids. The cannabinoid receptor family includes two types, CB1 and CB2, belonging to the seven transmembrane–spanning superfamily of G protein–coupled receptors. ¹ CB1 receptor messenger RNA was first identified in cell lines and regions of the brain that have cannabinoid receptors. ² It has also been found among testis transcripts. ³ CB2 was later recognized in spleen macrophages and other cells of the immune system. ⁴ CB1 and CB2 are widely distributed in neural and nonneural tissues. CB1 predominates in brain, whereas CB2 predominates in peripheral tissues. ^{5 6} Epithelial cells, such as prostate PC-3 cells, ⁷ human bronchial epithelial cells, ⁸ skin keratinocytes, ^{9 10} and rat salivary gland cells, ¹¹ display both receptor types. 

The widespread tissular localization of cannabinoid receptors and their various roles as modulators in various physiological and pathologic processes have been described in several excellent reviews. ^{12 13} Cannabinoids regulate cell proliferation and apoptosis in normal cells ¹⁴ and cancer cells. ¹² They can lead to antitumor effects, ¹⁵ or they can increase the proliferation of cancer cells. ¹⁶ Cannabinoids have also been associated with defensive, anti-inflammatory, and immunomodulatory epithelial functions. ^{6 17}  

Although each cannabinoid receptor has specific ligands, CB1 and CB2 display similar pharmacologic and biochemical properties. ¹⁸ Inhibition of adenylyl cyclase (AC) is a well-characterized response in brain tissue and neuronal cells expressing CB1. It also occurs in human lymphocytes and mouse spleen cells expressing CB2 receptors. ¹ In some experimental models, however, CB1 signaling can also stimulate AC. ¹  

Both cannabinoid receptors activate the mitogen-activated protein kinases (MAPKs) extracellular signal-regulated kinase (ERK)1/2 and p38 and the Akt/PKB survival pathway. Cannabinoid effects vary according to cell phenotype. ¹⁹ They decrease the nuclear expression of c-fos and c-jun in primary splenocytes, ²⁰ but they activate c-Jun N-terminal kinase (JNK) and p38 mitogen-activated protein kinase in Chinese hamster ovary cells. ²¹ In the hippocampus, cannabinoid ligands activate p38 MAPK but not JNK. ²²  

The best-known endocannabinoids, anandamide (AEA), 2-arachidonoyl glycerol (2-AG), and palmitoylethanolamide (PEA), are lipid molecules released in response to various pathologic events. ²³ The rat retina has a functional endocannabinoid system, ²⁴ and most ocular tissues display 2-AG, PEA, and AEA. ²⁵ In the trabecular meshwork, CB1 and CB2 signaling pathways have been associated with differential changes in aqueous humor outflow and intraocular pressure. ^{26 27} No specific information about the conjunctiva is available, but cannabinoid receptors may be active in ocular surface epithelia. 

To investigate this hypothesis, we examined protein and mRNA expression of CB1 and CB2 in mouse conjunctiva, human conjunctival sections and exfoliated cells, and a human conjunctiva–derived cell line (IOBA-NHC). ²⁸ To ascertain receptor function, we evaluated the effect of cannabinoid agonists on forskolin-induced AC activity and cell growth in vitro. We also studied the effects of a cannabimimetic on tumor necrosis factor-α (TNF-α)–induced activation of c-jun N-terminal kinase (JNK) and nuclear factor-κB (NF-κB) signaling cascades because this cytokine is increased in tear fluids, and the conjunctival epithelium is subjected to inflammatory conditions. ^{29 30 31}  

Materials and Methods

Mouse Tissues

Male Balb-c mice were handled according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. They were deeply anesthetized (chloral hydrate, 800 mg/kg) and perfused through the heart with a 4% paraformaldehyde solution in 0.1 M phosphate buffer (pH 7.3). Eyes were enucleated and cryopreserved in graded sucrose solutions and sucrose-OCT compound mixes. 

Human Conjunctiva

The procurement of human conjunctival samples adhered to the tenets of the Declaration of Helsinki (52nd WMA General Assembly, Edinburgh, Scotland, October 2000, http://www.wma.net/e/policy/b3.htm) and was approved by our Institutional Research Board. 

Informed consent was obtained from three patients prescheduled for strabismus ocular surgery and free from conjunctival or corneal lesions. A small peri-incisional conjunctival sample was obtained and fixed in 4% paraformaldehyde. Impression specimens were obtained from healthy donors (n = 6) who gave informed consent. After the application of topical anesthesia (1% proparacaine; Alcon, Tortuguitas, Argentina), two pieces of cellulose ester paper (22 μm; Millipore, Billerica, MA) were applied to the inferior tarsal and bulbar conjunctiva of each eye. The paper pieces were applied for approximately 10 seconds, and, after gentle pressure with the blunt end of the forceps, the pieces were peeled off and immediately transferred to tubes and fixed in 70% ethylic alcohol. ³²  

Cell Cultures

Cell from the human conjunctival line IOBA-NHC ²⁸ were grown on plastic tissue culture flasks or dishes. Proliferation medium was Dulbecco modified Eagle medium (DMEM)/F-12 medium (11330-032; Gibco, Invitrogen, Carlsbad, CA) with 10% inactivated fetal bovine serum (Notocor, Córdoba, Argentina), 40 μg/mL gentamicin (Biol, Buenos Aires, Argentina), and 2.5 μg/mL amphotericin (Richet, Buenos Aires, Argentina). After confluence, they were further cultured for 24 hours in the same medium but with a lower serum supplement (1.5%). 

Immunohistochemistry

Tissue blocks were frozen in N₂-cooled acetone and sectioned at 7 μm. Cryosections, impression papers, and monolayers were incubated overnight with different primary antibodies. Specimens were developed using biotinylated secondary antibodies followed by avidin-biotinylated peroxidase complex (Vectastain Elite ABC-peroxidase kit; Vector Laboratories, Burlingame, CA). A color reaction was obtained using nickel-enhanced diaminobenzidine staining. ³³ Negative blanks without the primary antibody were simultaneously processed with test samples. Samples from three or more donors or cultures were used for each antibody. Cryosections from salivary glands ¹¹ were used as positive controls. 

Several rabbit antisera against cannabinoid receptors were tested. CB1 was detected with CB1-CAY against the peptide MKSILDGLADTTFR (dilution 1:500; 101500; Cayman Chemical, Ann Arbor, MI) and CB1-ABR (dilution 1:500; OPA1-15297; Affinity BioReagents, Golden, CO), against a synthetic C-terminus peptide. CB1-N or CB1-C, obtained from rabbits immunized with fusion proteins containing 74 amino acids from either the N-terminal or the C-terminal of rat CB1 (dilutions 1:1500 and 1:500, respectively), were also used. ³⁴ CB2-CAY, made against the sequence NPMKDYMILSGPQK (dilution 1:500; 101550; Cayman Chemical) and CB2-ABR, made against a fusion protein containing the first 32 amino acid residues from rat CB2 (dilution 1:1000; PA1-746A CB2; Affinity BioReagents), were tested against CB2. 

Activation of signaling pathways was studied in IOBA-NHC cultures using the following antibodies: SAPK/JNK antibody (JNK, 9252), phospho-SAPK/JNK (Thr183/Tyr185) antibody (p-JNK, 9251), and phospho-p38 MAPK (Thr180/Tyr182) antibody (p-p38, 9211), all from Cell Signaling Technology (Danvers, MA). NF-κB p-65 was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). All these antibodies were diluted 1:200. 

Western Blot Analysis

IOBA-NHC cell cultures (35-mm multiwell plates, grown until confluence, 3 wells/extract) were washed in DMEM/F-12 medium without supplements and were homogenized in an extraction buffer containing 10 mM Tris-HCl with 2 mM EDTA, 150 mM NaCl, 1% Triton X-100, and protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). Extracts (15 μg protein/lane) were separated by SDS-PAGE and were transferred (100 V, 75 minutes) onto a nitrocellulose filter using standard techniques. Equal sample loading for electrophoresis was confirmed by the Bradford protein assay (Sigma-Aldrich), and equal transfer to the membrane was confirmed by Ponceau S staining. ³⁵ Extracts and blottings were made from three different cultures. 

After incubation with primary antibodies, membranes were further incubated with biotinylated anti–rabbit IgG and extravidin-alkaline peroxidase. Immunoreactive protein was detected by enhanced chemiluminescence (CPS160; Sigma-Aldrich). Optical densities were measured with NIH Image J software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). 

Immunoblot detection of cannabinoid receptors was performed with CB1-CAY and CB2-CAY (dilution 1:100). JNK and p-JNK were detected using antisera 9252 and 9251 (dilutions 1:500) from Cell Signaling Technology. 

RT-PCR Analysis

IOBA-NHC cells (35-mm multiwell plates, grown until confluence, 3 wells/extract) were washed as before. Total RNA was extracted (RNeasy Lipid Tissue Mini Kit; Qiagen, Valencia, CA). RT-PCR Beads (Illustra Ready-To-Go; Amersham Biosciences, Piscataway, NJ) were used for cDNA synthesis and PCR (2 μg RNA/bead). Oligo-dT primers were used for retrotranscription, and the following specific primers were used for PCR: CB1 sense, 5′-GATGTCTTTGGGAAGATGAACAAGC-3′; CB1 antisense, 5′-AG ACGTGTCTGTGGACACAGACATGG-3′; CB2 sense, 5′-TTTCCCACTGATCCCCAATG-3′; CB2 antisense, 5′-AGTTGATGAGGCACAGCATG-3′. 

PCR conditions were melting at 95°C for 1 minute, annealing at 60°C (CB1) and 52°C (CB2) for 1 minute, extension at 72°C for 2 minutes, and 35 amplification cycles. PCR products were analyzed by electrophoresis on a 2% agarose gel, and the expected sizes of amplicons were 309 bp for CB1 and 337 bp for CB2. Negative controls to evaluate DNA contamination were prepared inactivating the reverse transcriptase from the RT-PCR beads. Extracts were obtained from four different cultures. 

Cyclic Adenosine Monophosphate Assay

Changes in the levels of intracellular cAMP were assayed in ethanol extracts of confluent IOBA-NHC cells grown on 24-mm multiwell plates. The highly specific cAMP antibody, kindly provided by Albert F. Parlow of the National Hormone and Peptide Program at Harbor-UCLA Medical Center (Los Angeles, CA), was used for radioimmunoassay (RIA). Sensitivity of the assay was 0.061 pmol/mL, with intra-assay and interassay coefficients of variation of 8.1% and 10.5%, respectively. ¹¹  

Confluent cell monolayers were washed in DMEM/F12 without supplements and incubated with 0.1 to 10 × 10⁻⁶ M forskolin (FRSK) and 0.5 × 10⁻³ M isobutyl-1-methylxanthine (IBMX; both from Sigma-Aldrich Chemical) to increase cAMP content. Drugs were diluted in the same medium, and incubation lasted 15 minutes. Experiments with CB ligands were made with 0.5 × 10⁻⁶ FRSK because this was the minimal concentration inducing significant increases of intracellular cAMP. Cannabinoid agonists, antagonists, or their combinations were added, and incubation continued for another 30 minutes. Medium was discarded, and ethanol extraction was immediately performed. Each experimental point represented the average of four to six dishes. Triplicate samples were assayed from each dish. Results were expressed as picomoles of cAMP per dish ± SE. Commercial software (GraphPad Prism version 4.00 for Windows; GraphPad Software, San Diego, CA; www.graphpad.com) was used for statistical comparisons. 

Cannabinoid receptors were stimulated with the endocannabinoid AEA (Sigma-Aldrich Chemical) or with (−)-cis-3-[2-hydroxy4-(1,1-dimethylheptyl) phenyl]-trans-4-(3-hydroxypropyl)cyclohexanol (CP55,940; Tocris, Ellisville, MO). CP55,940 is a potent cannabinoid analogue, acting on CB1 and CB2 receptors, that does not bind to vanilloid receptors. ¹³ These compounds were first diluted in ethyl alcohol (1 × 10⁻² M), and further dilutions were made in tissue culture medium without serum. Preliminary experiments (not shown) indicated that 10⁻⁶ M AEA or 10⁻⁶ M CP55,940 induced reproducible decreases of cAMP, and this concentration was further used for evaluation of CB receptor inhibitors. 

The specific CB1 and CB2 inhibitors AM251 ([N-(piperidin-1-yl)-1-(2,4-dichlorophenyl)-5-(4-chlorophenyl)- 4-ethyl-1H-pyrazole-3-carboxamide], 10 × 10⁻⁶ M) and AM630 (6-Iodo-2-methyl-1-[2-(4-morpholinyl)ethyl-1H-indol-3-yl](4-methoxyphenyl)methanone, 10 × 10⁻⁶ M) were also obtained from Tocris (Ellisville, MO). AM251 and AM630 were dissolved in dimethyl sulfoxide (2 × 10⁻³ M), and further dilutions were made in tissue culture medium without serum. They were added 5 minutes before the agonists. 

Cell Viability Assay

The 3-(4,5-dimethylthiazol-2-yl)2,5-diphenyl tetrazolium bromide (MTT) cleavage assay ³⁶ is based on the ability of viable cells, but not dead cells, to convert MTT to a colored formazan product. Thus, optical density (OD) of formazan generated in vitro is proportional to the number of viable cells in culture. 

Aliquots (100 μL) of IOBA-NHC cell suspensions in proliferation medium were dispensed into 96-well tissue culture plates. Linearity of formazan production was tested at seeding densities between 10 and 50 × 10³ cells/well. Experiments evaluating the effect of cannabinoid receptor ligands on cell growth were made at 20 × 10³ cells/well. 

At the end of the incubation period, medium was aspirated and replaced with 0.5 mg/mL MTT (V13154; Invitrogen) in sterile phosphate-buffered saline. Plates were further incubated for 4 hours at 37°C, and 150 μL isopropyl alcohol/HCl 0.04 N was added to solubilize the formazan product. Aliquots (200 μL) were transferred to fresh 96-well plates, and MTT conversion (A550-A690) was measured with a microplate absorbance reader (Benchmark; Bio-Rad, Hercules, CA). 

Each experiment was conducted three times, and each experimental point was made in triplicate. CP55,940 (0.01–10 × 10⁻⁶ M) was added at the time of cell plating. The specific CB1 and CB2 inhibitors AM251 and AM630, respectively, were added to cell suspensions 20 minutes before plating. Cultures were grown for 1, 2, or 3 days before measurement of MTT conversion. 

Activation of Signaling Pathways

JNK phosphorylation was evaluated by comparison of p-JNK and JNK immunoreactivities in parallel Western blot analyses. Confluent IOBA-NHC monolayers were washed in nonsupplemented medium. Monolayers were challenged with 10 ng/mL TNF-α ³⁷ (G5241; Promega, Madison, WI) for 15, 30, 60, or 240 minutes in serum-free culture medium. To test the effects of cannabinoid receptor stimulation, CP55,940 (0.01–10 × 10⁻⁶ M) was added 10 minutes before TNF-α, and cell extracts were made 60 minutes after addition of this cytokine. 

Changes in the subcellular localization of p-JNK, p-p38, and NF-κB after a proinflammatory stimulus were made in similar cultures treated with TNF-α for 30 or 60 minutes, with or without the previous addition of CP55,940 (0.01–10 × 10⁻⁶ M). At the end of the experiment, monolayers were fixed with 2% paraformaldehyde in phosphate buffer. Experiments were repeated three times, with duplicate wells for each antibody. 

Results

Demonstration of Cannabinoid Receptors

Immunohistochemistry of Mouse Ocular Surface.

Cannabinoid receptor immunostaining was found in the conjunctival epithelium but could not be detected in the conjunctival stroma. Intense CB1 immunoreactivity was present in every cell layer of the conjunctival epithelium (Figs. 1A 1C) . Immunostaining was highest at the fornix, where epithelial outermost cells showed stronger immunoreactivity than basal cells. This difference disappeared in regions with thinner conjunctival epithelia. 

CB2 immunoreactivity exhibited a similar pattern (Figs. 1B 1D) . Strong immunostaining was found in almost every conjunctival epithelial cell. All tested anti-CB1 and anti-CB2 antisera displayed the same staining pattern. 

CB1 and CB2 Immunohistochemistry in Human Conjunctival Cells.

CB1 and CB2 antisera strongly stained the conjunctival epithelium in cryosections of human samples. CB1 and CB2 immunoreactivity appeared in the different epithelial layers, including goblet cells. In addition, CB2 antiserum immunostained blood vessels and scattered cells within the stroma (Figs. 2A 2B 2C) . 

Impression cytology showed conjunctival cells appearing as regular arrays of polygonal epithelial cells with some goblet cells. Exfoliated epithelial cells were also strongly immunostained by CB1 and CB2 antibodies (Figs. 2D 2E) . A punctate pattern was present in some preparations. Similar results were obtained from bulbar and tarsal samples. 

IOBA-NHC monolayers contained polygonal cells with large nuclei. These cells exhibited strong cytoplasmic immunoreactivity after incubation with CB1 and CB2 antisera. Most cells showed a fine punctate pattern, perhaps reflecting receptor localization in cytoplasmic vesicles and membrane domains (Figs. 3A 3B) . 

CB1 and CB2 immunoblotting of IOBA-NHC cell extracts (n = 3) showed single bands at MW 60 kDa and 45 kDa, respectively (Fig. 3C) . 

CB1 and CB2 Transcripts.

To evaluate the expression of CB1 and CB2 mRNA in IOBA-NHC cells, total RNA from these cells was extracted and analyzed by classic RT-PCR (Fig. 3D) . No bands were observed in negative controls, where the reverse transcription step was omitted. The lengths of amplicons representing CB1 and CB2 mRNA were 309 bp and 337 bp, respectively. IOBA-NHC RNA extracts (n = 4) contained both sequences. 

Functional Properties of Cannabinoid Receptors

Changes of cAMP Levels Induced by Cannabinoid Ligands.

Inhibition of phosphodiesterase by IBMX (0.5 × 10⁻³ M) did not induce changes in intracellular cAMP levels of IOBA-NHC cells. cAMP levels increased when AC was simultaneously activated by FRSK (Fig. 4A) . These increases became significant in the presence of 0.5 × 10⁻⁶ M FRSK. Addition of AEA (10⁻⁶ M) significantly reduced intracellular cAMP levels in the presence of 0.5 × 10⁻³ M IBMX and 0.5 × 10⁻⁶ M FRSK (Fig. 4B) . These conditions were used in the subsequent experiments. 

The specific CB1 and CB2 antagonists AM251 (10 × 10⁻⁶ M) and AM630 (10 × 10⁻⁶ M) reversed the effects of AEA (Fig. 4C) . Both antagonists produced similar reversals, suggesting that CB1 and CB2 receptors would be involved in the regulation of cAMP levels. 

Under the same conditions of AC stimulation, incubation with the selective cannabimimetic CP55,940 (10⁻⁶ M) also reduced intracellular cAMP levels (Fig. 4D) . Given that this compound does not bind to vanilloid receptors, results indicate that declines of cAMP levels would be mediated by cannabinoid receptors. As with AEA, CB1 and CB2 specifically reversed CP55,940 effects. No additive effects of AM251 and AM630 could be detected (not shown). 

Cannabinoid-Induced Promotion of Cell Proliferation.

To determine whether formazan generated in vitro was proportional to cell numbers, we evaluated parallel cultures plated with 10 to 50 × 10³ cells/well (n = 5 wells per experimental point) in proliferation medium. Formazan OD was linearly correlated with plating densities. Linear regression coefficients (r ²) were 0.87, 0.89, and 0.90 for 1, 2, and 3 days in vitro, respectively. Thus, the number of recovered viable cells per well was proportional to the number of growing cells (Fig. 5A) . 

The following experiments were made in cultures plated at 20 × 10³ cells/well and grown for 1, 2, and 3 days in proliferation medium. We tested different CP55,940 concentrations (0.01–10 × 10⁻⁶ M; n = 4 wells for each experimental point; three experiments). Cultures containing CP55,940 contained larger cell numbers than untreated cultures (Fig. 5B) . After 1 day in vitro, the lowest CP55,940 concentrations (0.01 and 0.10 × 10⁻⁶ M) induced significant increases, whereas higher concentrations showed no proliferative effect. Larger increases of cell numbers appeared after 2 and 3 days in vitro, when most prominent effects were induced by 0.01 to 1.00 × 10⁻⁶ M CP55,940. The largest increase in cell numbers was caused by 0.10 × 10⁻⁶ M CP55,940. Cell numbers observed after incubation in 10 × 10⁻⁶ M CP55,940 were not different from those found in control wells. 

The specific CB1 and CB2 antagonists AM251 and AM630 were tested at two concentrations, 0.1 × 10⁻⁶ M and 1 × 10⁻⁶ M (n = 4 wells for each experimental point; three experiments). In cultures without CP55,940, cannabinoid receptor antagonists slightly decreased cell numbers after 1 day in vitro. A significant difference, however, appeared only in cultures treated with 1 × 10⁻⁶ M AM630. After 2 and 3 days in vitro, CB1 and CB2 antagonists decreased cell numbers. Low and high AM251 and AM630 concentrations induced significant reductions of cell growth. 

CB1 and CB2 antagonists decreased cell numbers in cultures treated with 0.1 × 10⁻⁶ M CP55,940. In 1-day-old culture, the higher AM251 and AM630 concentrations (1 × 10⁻⁶ M) completely reversed CP55,940-induced increases in cell numbers. The lower antagonist concentrations, equimolar to CP55,940 concentration, only induced a partial reversal of the cannabinoid effect. In 2- and 3-day-old cultures, the higher antagonist concentrations not only reversed the CP55,940 effect, they reduced cell growth to the levels observed in cultures without CP55,940. The lowest concentrations also produced significant decreases of cell numbers, but these were still above those found in cultures without CP55,940 (Fig. 6) . 

Effects of Cannabinoid Receptor Stimulation on Stress Markers.

Immunoblotting recognized two JNK isoforms, p54 and p46, and the corresponding phosphorylated molecules, p-p54 and p-p46. In confluent, nonstimulated IOBA-NHC cultures, the phosphorylated isoforms were below detection level. Phospho-p46 appeared after just 15 minutes of incubation with TNF-α, whereas p-p54 could be detected after 30 minutes. Both phosphorylated isoforms remained elevated for at least 60 minutes. After 4 hours, p-p46 had decreased and p-p54 could no longer be detected. Quantitative analysis indicated that differences with control cultures were significant. Levels of nonphosphorylated isoforms did not change within this period (Figs. 7A 7B) . 

Phospho-JNK levels induced after 60 minutes of TNF-α incubation were strikingly reduced when CP55,940 was added 10 minutes before TNF-α. A reduction appeared at the lowest CP55,940 concentration (0.01 × 10⁻⁶ M). Higher concentrations (1 or 10 × 10⁻⁶ M) completely abolished JNK activation (Figs. 7C 7D) . 

Phospho-JNK immunoreactivity was almost nil in control confluent monolayers. By contrast, after 30 minutes of TNF-α incubation, cell cytoplasm showed strong p-JNK immunostaining. This immunoreactivity decreased after longer incubation times and never appeared within cell nuclei. Phospho-JNK immunoreactivity induced by a 30-minute incubation with TNF-α disappeared when CP55,940 was added 10 minutes before the cytokine (Fig. 8A) . Strong p-p38 immunoreactivity was observed under basal conditions, and no changes could be detected after TNF-α treatment. 

Confluent IOBA-NHC cells displayed cytoplasmic NF-κB immunoreactivity. Monolayers fixed 30 minutes after TNF-α stimulation showed nuclear translocation of this immunoreactivity. However, no nuclear translocation could be detected when TNF-α stimulation was performed in the presence of the cannabinoid agonist CP55,940 (Fig. 8B) . 

Discussion

Two cannabinoid receptors, CB1 and CB2, were detected in mouse and human conjunctival epithelia using various antisera directed to different epitopes of each receptor. RT-PCR amplification of receptor transcripts confirmed the expression of CB1 and CB2 receptors in a human-derived conjunctival cell line. 

CB1 and CB2 in the Conjunctival Epithelium

In mouse specimens, every conjunctival surface showed CB receptors. The highest immunoreactivity appeared in the outermost cells of the fornix region, whereas palpebral conjunctiva displayed the lowest immunoreactivity. The conjunctival epithelium includes a progenitor compartment along the basal and suprabasal layers and a nonproliferating compartment in the layers above. ³⁸ Thus, the high immunoreactivity of outermost layers suggests that changes in receptor distribution might accompany differentiation. On the other hand, since conjunctival epithelial stem cells are mainly located in the palpebral epithelium of Wistar rats, ³⁹ the lower CB1 and CB2 immunoreactivities in the mouse palpebral conjunctiva could be related to the presence of proliferating cells. 

In human specimens, cannabinoid receptor immunoreactivity was present in nongoblet and goblet epithelial cells. The role of cannabinoids in goblet cell secretion remains to be studied; however, we have previously demonstrated endocannabinoid inhibition of rat salivary secretion. ¹¹ In addition, Δ⁹-tetrahydrocannabinol reduces MUC5AC mRNA and mucous cell metaplasia in airway epithelia exposed to influenza infection. ⁴⁰  

Functional Responses

Agonist-induced reduction of cAMP levels demonstrated linkage of conjunctival cannabinoid receptors to AC. Similar decreases appeared after treatment with the endocannabinoid AEA and specific agonist CP55,940, proving that cAMP reduction would not be mediated by vanilloid receptors. Reversal of these responses by the specific CB1 and CB2 antagonists AM251 and AM630 confirmed the involvement of cannabinoid receptors in the regulation of cAMP levels. Control of the AC cascade could be directly involved in cannabinoid proliferative ⁴¹ and anti-inflammatory responses. ⁴²  

The role of endocannabinoids and exocannabinoids in proliferation has been studied primarily in cancer cells. Different outcomes have been described, depending on the nature of the target cell and its proliferative or differentiation status. Growth inhibition seems more evident in cancer than in noncancer cells. ^{19 43} Inhibition of cell proliferation is probably a pharmacologic effect because it is usually observed after high doses of exocannabinoids. By contrast, cancer ¹⁶ and normal human B cells ⁴⁴ proliferate after cannabinoid stimulation at nanomolar concentrations. Our findings in the conjunctival cell line IOBA-NHC confirm that proliferation responses to cannabinoids critically depend on agonist concentration because they disappeared when high CP55,940 concentrations (10 × 10⁻⁶ M) were used. 

Remarkably, incubation with the antagonists AM251 and AM630 increased cAMP to higher levels than those observed in control cultures. Similarly, they decreased proliferation below control levels. Although activation of receptors is usually considered to be a ligand-dependent phenomenon, various G protein–coupled receptors (GPCRs), including CB1 and CB2, display significant constitutive activity. This would explain the inverse agonist effects of AM251 and AM630. ¹³ Active GPCRs are constitutively endocytosed from the plasma membrane and often exhibit intracellular localization. ⁴⁵ Punctate distribution of CB1 and CB2 immunoreactivities that reflects a vesicular localization could be a consequence of constitutive activation. A tonically active endocannabinoid system could perhaps explain the strong basal expression of p-p38 in IOBA-NHC cells. 

Activation of CB1 and CB2 Blocked Stress-Activated Cascades

CP55,940 blocked TNF-α–induced activation of JNK and NF-κB in conjunctival epithelial cells. Downregulation of these stress-activated cascades implies an anti-inflammatory role of cannabinoid ligands. Such downregulation could be important in the management of ocular surface inflammatory diseases. TNF-α increases in the conjunctival epithelium ²⁹ of patients with dry eye keratoconjunctivitis. It is also rapidly upregulated in epithelial cells after endotoxin-induced conjunctivitis ⁴⁶ or hyperosmolar stress. ⁴⁷  

Although p38 was also involved in conjunctival inflammatory phenomena, ³¹ a cannabinoid-induced activation of this cascade could not be detected in IOBA-NHC cells. Rather, this probably was a consequence of their high basal levels of p-p38 immunoreactivity. 

IOBA-NHC cell cytoplasm displayed NF-κB-p65 immunoreactivity under basal conditions. A similar distribution has been described in normal mouse conjunctival epithelium in vivo. ⁴⁸ Nuclear translocation of this transcription factor has been observed in cells of keratoconjunctivitis samples or after experimental scraping. ⁴⁸ Thus, the cannabinoid-induced reversal of NF-κB-p65 nuclear translocation would take part in the control of conjunctival inflammatory conditions. 

Attenuation of TNF-α–induced responses by CB1 or CB2 stimulation has been demonstrated in various other systems. ^{8 49} Complex interactions occur between endocannabinoids and stress/inflammatory mechanisms. Thus, TNF-α stimulation increases endocannabinoid production, whereas cannabinoid stimulation can decrease TNF-α secretion induced by various inflammatory conditions. ⁵⁰ The net result, however, would afford protection. During intestinal inflammation, enhanced endocannabinoid tone, acting at least in part through cannabinoid receptors, diminishes epithelial damage. ⁵¹  

Conclusions

In summary, we would like to suggest that conjunctival cannabinoid receptors play an important role in the maintenance of this epithelium. Further studies are required to understand whether they would simply participate in normal epithelial turnover or whether their activation could be also involved in the development of epithelial metaplasia, a frequent finding in conjunctival diseases. In addition, their counteracting effects on TNF-α response suggest that they may play a modulatory role in ocular surface inflammation. 

 Supported by grants from Universidad Austral, Universidad de Buenos Aires (UBACYT 069/2004), CONICET (PIP 2006/6149), and SECYT (BID 1728 OC-AR PICT 14264). MI, VT, and JPP are Research Fellows of CONICET. VR and AMS are members of the Research Career of CONICET.

 Submitted for publication October 12, 2007; revised April 14, May 14, and June 4, 2008; accepted August 15, 2008.

 Disclosure: M. Iribarne, None; V. Torbidoni, None; K. Julián, None; J.P. Prestifilippo, None; D. Sinha, None; V. Rettori, None; A. Berra, None; A.M. Suburo, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunohistochemical demonstration of the cannabinoid receptors CB1 and CB2 in sections through mouse conjunctival sac. (A) This low-power photographic montage through the upper conjunctival sac shows that CB1 immunoreactivity appeared in the tarsal, fornical, and bulbar conjunctival regions. CB1 immunoreactivity was strongest in the fornical sac (f) and the bulbar conjunctiva, where every conjunctival layer displayed immunoreactivity. In the tarsal conjunctiva, CB1 immunoreactivity was mainly present in basal cells. Note the accumulation of immunoreactive cells at the transitional zone of the eyelid margin. s, sclera. (B) Another section from the same specimen demonstrates a similar distribution of CB2 immunostaining. Strong CB2 immunoreactivity was present in the fornix, the bulbar conjunctiva, and the eyelid margin. (C) The fornix is shown at high magnification. All cell layers displayed CB1 immunoreactivity, but the strongest immunoreactivity appeared in thin cytoplasmic lamellae forming the conjunctival outermost layers. (D) Fornical CB2 immunoreactivity displayed a similar pattern, with strong staining of outer cytoplasmic lamellae. Calibration bars: (A, B) 100 μm; (C, D) 25 μm.

Figure 2.

 

Cannabinoid receptors CB1 and CB2 in human bulbar conjunctival sections (A–C) and exfoliated cells (D, E). (A) This photomontage illustrates the presence of CB1 immunoreactivity in every epithelial layer of the human conjunctival epithelium. g, goblet cells. (B) Another section through the same conjunctival specimen shows a similar pattern of epithelial CB2 immunoreactivity. However, CB2 immunoreactivity was also present in vascular endothelium (v) and stromal cells (arrow). (C) A section through a goblet cell illustrates strong CB1 immunoreactivity in the cytoplasmic rim surrounding the mucous droplet. Note that surrounding epithelial cells showed similarly strong CB1 immunoreactivity. (D) Exfoliated cell preparations displayed polygonal epithelial cells exhibiting strong CB1 immunoreactivity. (E) Similar strong CB2 immunoreactivity appeared in conjunctival exfoliated cells. Some specimens, such as the one shown here, showed a punctate pattern of cannabinoid receptor immunostaining. Calibration bars: (A, B) 50 μm; (C–E) 25 μm.

Figure 3.

 

Expression of cannabinoid receptors CB1 and CB2 in IOBA-NHC monolayers. (A) Every cell in confluent cultures exhibited CB1 immunoreactivity. Immunostaining usually showed a punctate pattern. (B) CB2 immunostaining was also found in every cell of confluent IOBA-NHC monolayers and displayed the same punctate pattern. Calibration bars: 10 μm. (C) Immunoblot analysis of CB1 and CB2 protein expression in IOBA-NHC cells. A single positive band (MW 60 kDa and 45 kDa, respectively) was obtained for each receptor. The blots are representative of three independent experiments. (D) RT-PCR analysis of cannabinoid receptors CB1 and CB2 mRNAs in IOBA-NHC cells. The three lanes shown in this ethidium bromide–stained gel correspond to MW standards (500–300 bp), and the CB1 and CB2 amplicons (309 and 337 bp, respectively) after 35 PCR cycles. The image is representative of four independent experiments.

Figure 4.

 

Intracellular cAMP concentrations in IOBA-NHC monolayers, expressed as picomoles of cAMP/dish ± SE. In all experiments, the incubation medium contained the phosphodiesterase inhibitor isobutyl-1-methylxanthine (IBMX, 0.5 × 10⁻³ M). Experimental points represent the average of four to six dishes. Triplicate samples were assayed from each dish. (A) We used the AC activator FRSK to increase intracellular cAMP levels. Increasing FRSK concentrations from 0 to 10 × 10⁻⁶ M were tested. Statistical evaluation showed significant posttest results (<0.01) for linear trend. One-way ANOVA, followed by Tukey multiple comparison test, showed significant increases with 0.5 × 10⁻⁶ M or higher FRSK concentrations (a, indicates comparison with control values, without FRSK; a, P < 0.05; aaa, P < 0.001). (B) Incubation with AEA (1 × 10⁻⁶ M) decreased cAMP concentrations. A significant decrease, however, was observed only after AC activation by 0.5 × 10⁻⁶ M FRSK (t-test; a, P < 0.05). This FRSK concentration was used in the following experiments. (C) Addition of CB1- and CB2-specific antagonists AM251 (10 × 10⁻⁶ M) or AM630 (10 × 10⁻⁶ M) reversed the cAMP decrease induced by 1 × 10⁻⁶ M AEA. C, control without AEA. Independent t tests demonstrated significant differences (a, comparisons with control values; b, comparison with AEA as single treatment; a and b, P < 0.05). (D) The selective cannabimimetic CP55,940 (10⁻⁶ M) also produced a significant decrease of intracellular cAMP levels. C, control without CP55,940. The cannabinoid-induced decrease was not observed in the presence of AM251 (10 × 10⁻⁶ M) or AM630 (10 × 10⁻⁶ M) (t-test; a, comparisons with control values; b, comparison with CP55,940 as single treatment; a and b, P < 0.05).

Figure 5.

 

Formazan detection of cell growth in culture. (A) Formazan production was evaluated at 1, 2, and 3 days in vitro in proliferation medium after plating at different cell densities (n = 5 wells/experimental point). Linear correlation between formazan and number of plated cells (r ² = 0.87, 0.89, and 0.90 for 1, 2, and 3 days in vitro, respectively) indicated that cell growth was proportional to the number of plated cells. (B) We tested different CP55,940 concentrations (0.01–10 × 10⁻⁶ M; gray columns) on cell growth in proliferation medium after plating at 20 × 10³ cells/well (n = 4 wells/experimental point; three experiments). Number of cells/well was measured by formazan production after 1, 2, and 3 days in vitro. Selected statistical comparisons (two-way ANOVA, with Bonferroni posttests) are shown: a, difference with no CP55,940; b, c, d, differences with 0.01, 0.10, and 1.0 × 10⁻⁶ M CP55,940, respectively. a, P < 0.05. aa, P < 0.01. aaa, P < 0.001. Similar values apply to the other symbols. Slight but significant increases in cell numbers appeared after 1 day in vitro. Cell numbers significantly increased above control values after 2 and 3 days in vitro in 0.01 to 1 × 10⁻⁶ M CP55,940. No changes were detected in wells incubated with 10 × 10⁻⁶ M CP55,940. div, days in vitro.

Figure 6.

 

Reversal of cell growth by CB1 and CB2 antagonists in monolayers cultured in proliferation medium. Untreated cultures (white columns) and 0.1 × 10⁻⁶ M CP55,940-treated cultures (gray columns) were compared (n = 4 wells/experimental point; three experiments). Selected statistical comparisons (one-way ANOVA followed by Tukey multiple comparison test) are shown. In untreated cultures: a, difference with control; b and c, differences between low and high concentrations of AM251 and AM630, respectively. In CP55,940-treated cultures: d, difference with control; e and f, differences between low and high concentrations of AM251 and AM630, respectively. a, P < 0.05. aa, P < 0.01. aaa, P < 0.001. Similar values apply to the other symbols. After 1 day in vitro, only the highest AM630 concentration showed a significant effect on untreated cultures. In CP55,940-treated cultures, both antagonists completely reversed cannabinoid-induced increases in cell numbers. After 2 and 3 days in vitro, AM251 and AM630 decreased cell growth in untreated cultures. A larger effect appeared in 2- and 3-day-old CP55,940-treated cultures, in which both antagonists, even at their lowest concentrations, reversed the proliferative effect of the cannabinoid ligand. div, days in vitro.

Figure 7.

 

Effects of CP55,940 on TNF-α–induced activation of JNK. (A) Western blot analysis showed activation of p-JNKs after TNF-α stimulation of IOBA-NHC cells. Phosphorylated isoforms could not be detected in unstimulated cultures (t = 0), but p-p46 appeared 15 minutes after stimulation and p-p54 showed at 30 minutes. Maximal activation of both isoforms occurred between 30 and 60 minutes. After 240 minutes, p-p54 almost disappeared and p-p46 was considerably reduced. Nonphosphorylated isoforms (p54 and p46) showed no changes within this period. Blots are representative of three independent experiments. (B) Bars represent the mean ± SE of the sum of p-p46 and p-p54 for each time period. Statistical comparisons (one-way ANOVA followed by Tukey multiple comparison test) with controls without TNF-α are shown (aaa, P < 0.001). (C) Blotting of cell extracts activated by TNF-α showed that pretreatment with CP55,940 decreased or abolished JNK phosphorylation according to its concentration (0.01 to 10 × 10⁻⁶ M). C, control without CP55,940 or TNF-α. Blots are representative of three independent experiments. (D) The bar graph compares mean ± SE of the sum of p-p46 and p-p54 for each treatment. Statistical comparisons (one-way ANOVA followed by Tukey multiple comparison test) with controls without TNF-α (aaa, P < 0.001) and TNF-α without CP55,940 (b, P < 0.05; bbb, P < 0.001) are shown.

Figure 8.

 

Localization of p-JNK and NF-κB immunoreactivities in three parallel IOBA-NHC cell cultures incubated under control conditions (left), with TNF-α (middle), and TNF-α plus the cannabinoid agonist CP55,940 (right). (A) Under control conditions, IOBA-NHC cells showed little or no p-JNK immunoreactivity. Immunostaining strikingly increased after 30 minutes of incubation with TNF-α. Phospho-JNK did not increase when TNF-α stimulation was performed in the presence of CP55,940. (B) Under control conditions, NF-κB was localized in the cytoplasm but was translocated to the nucleus after TNF-α stimulation. Incubation with CP55,940 prevented nuclear translocation of NF-κB. Calibration bar for all figures: 25 μm.

We thank Guillermo Gastón and Soledad Arregui for their skillful technical assistance.