CB2 receptor activation causes an ERK1/2-dependent inflammatory response in human RPE cells

A chronic low-level inflammation contributes to the pathogenesis of age-related macular degeneration (AMD), the most common cause of blindness in the elderly in Western countries. The loss of central vision results from attenuated maintenance of photoreceptors due to the degeneration of retinal pigment epithelium (RPE) cells beneath the photoreceptor layer. It has been proposed that pathologic inflammation initiated in RPE cells could be regulated by the activation of type 2 cannabinoid receptors (CB2). Here, we have analysed the effect of CB2 activation on cellular survival and inflammation in human RPE cells. RPE cells were treated with the selective CB2 agonist JWH-133 in the presence or absence of the oxidative stressor 4-hydroxynonenal. Thereafter, cellular viability as well as the release of pro-inflammatory cytokines and potential underlying signalling pathways were analysed. Our results show that JWH-133 led to increased intracellular Ca2+ levels, suggesting that RPE cells are capable of responding to a CB2 agonist. JWH-133 could not prevent oxidative stress-induced cell death. Instead, 10 µM JWH-133 increased cell death and the release of proinflammatory cytokines in an ERK1/2-dependent manner. In contrast to previous findings, CB2 activation increased, rather than reduced inflammation in RPE cells.

Excessive inflammatory processes in human retinal pigment epithelial (RPE) cells are associated with the development of age-related macular degeneration (AMD) 1,2 , the leading cause of visual impairment in the elderly in the Western world 3 . RPE cells form a single-cell layer located at the posterior part of the eye between the choroid and the photoreceptors, and are vital for the survival and the functionality of rods and cones. They regulate the visual cycle as well as the transport of nutrients from the choroid to the photoreceptors and the removal of waste products away from the retina 4,5 . RPE cells also renew photoreceptors by degrading their outer segments in the process called heterophagy, participate in the formation of the blood-retinal barrier, and maintain the ion balance and immune responses in the retina 1, [6][7][8][9] . Dysfunction of the RPE leads to the degeneration and death of photoreceptors, causing the distinctive loss of central vision in AMD 4,5 (reviewed in 6,10 ).
One protein receptor potentially capable of modulating inflammatory responses is the cannabinoid receptor type 2 (CB 2 ). The G-protein-coupled receptor is one of the two receptors targeted by pharmacologically active, plant-derived cannabinoids as well as the body's own endocannabinoids 11,12 . Another cannabinoid receptor is CB 1 , which is predominantly expressed in the central nervous system (CNS) 13 . Along with neuroprotective effects, the CB 1 receptor mediates the psycho-active effects of cannabinoids, such as increased appetite, hallucinations, and antiemesis 11,14 . In contrast, the CB 2 receptor is expressed predominantly in the periphery, especially on immune cells, and has been linked to many of the beneficial, anti-inflammatory effects of cannabinoids 13 .
Specific agonists of CB 2 have been developed to facilitate the studies of the receptor's effects and to avoid side-effects associated with CB 1 activation 15,16 . Studies utilizing these activators found that CB 2 activation reduced the production of IL-6 in lipopolysaccharide (LPS)-treated murine macrophages and reduced the severity of collagen-induced arthritis in mice 17 . However, many effects of CB 2  on the studied cell type, the culture conditions, and the agonist used 13 . Schmöle et al. found that the knock-out of CB 2 reduced the release of IL-6 from primary microglia upon LPS stimulation 18 . At the same time, others have found the activation of CB 2 to be anti-inflammatory 19,20 . CB 2 is expressed by RPE cells 21 and endocannabinoid levels are increased in the eyes of AMD patients 22 . In one study, CB 2 activation was found to reduce the hydrogen peroxide-induced death of ARPE-19 and primary human RPE cells, suggesting a beneficial effect of CB 2 receptor activation in the treatment or the prevention of AMD 21 . However, studies on the role of CB 2 receptor in human RPE cells are scarce and data about its effects on inflammation in other cells are inconsistent.
Here, we describe that JWH-133, a direct agonist of the CB2 receptor, increases the release of pro-inflammatory cytokines IL-6 and IL-8 from human RPE cells. This effect was associated with augmented ERK1/2 activation and increased intracellular Ca 2+ levels.

JWH-133 does not protect cells from HNE-induced cytotoxicity.
One previous report indicated that the activation of the CB 2 receptor in ARPE-19 cells is protective against oxidative stress-related cell death caused by hydrogen peroxide 21 . We have previously shown that the reactive aldehyde 4-hydroxynonenal (HNE), an abundant source of oxidative stress in the retina in vivo, increases cytotoxicity in human RPE cells 23 . Here, we tested whether the activation of CB 2 with JWH-133 would protect ARPE-19 cells from the cytotoxicity induced by HNE. We found that none of the studied concentrations of JWH-133 were able to prevent HNE-induced cytotoxicity (Fig. 1a). Instead, 10 µM JWH-133 proved to be cytotoxic, reducing the cell viability an additional 53% when compared to HNE-treatment alone (Fig. 1a). In both LDH and neutral red assays, 10 µM JWH-133 caused significant cytotoxicity also without HNE treatment (Fig. 1b).

JWH-133 activates inflammation in RPE cells.
In conjunction with the increased cytotoxicity induced by 10 µM JWH-133, it additionally increased the secretion of pro-inflammatory cytokines IL-6 and IL-8 from ARPE-19 cells exposed to HNE (Fig. 2). In accordance with our previous results 23,24 , HNE alone decreased the release of IL-6 and IL-8 from ARPE-19 cells, most likely due to the inhibition of nuclear factor κB (NF-κB). Despite the decreased IL-6 and IL-8 levels following the HNE treatment, an exposure of RPE cells to 10 µM JWH-133 still raised the cytokine levels by 62% and 64%, respectively (Fig. 2a). In cells that were not subjected to oxidative stress, 5 μM, JWH-133 slightly decreased IL-8 levels but had no effect on either IL-6 or HMGB1 (Fig. 2b). 10 µM JWH-133 alone significantly increased the levels of IL-6 and IL-8, as well as those of HMGB1 (Fig. 2b). from HNE-induced cytotoxicity, but is toxic at the concentration of 10 µM. Analysis of cell death and cell viability by the LDH and neutral red assay, show a moderate decrease in cell viability after an exposure to HNE (a), which could not be prevented by any of the studied concentrations of JWH-133. Instead, 10 µM JWH-133 was toxic to both untreated (b) and HNE-treated ARPE-19 cells (a). Results are shown as mean ± SEM and combined from 3-6 independent repetitions with 4-6 parallels per group. ns denotes not statistically significant, *denotes P < 0.05, ***denotes P < 0.001; Mann-Whitney U-test. This increase was comparable also in the absence of HNE, where JWH-133 by itself increased IL-6, IL-8, and HMGB1 levels (b). Results are shown as mean ± SEM and combined from 3 independent repetitions with 3-4 parallels per group. ns denotes not statistically significant, ***denotes P < 0.001; Mann-Whitney U-test.

Results obtained with the ARPE-19 cell line are repeatable in primary human RPE cells.
Repetition of our experiments in unpassaged hRPE cells also showed increased IL-6 and IL-8 secretion after an exposure to 10 µM JWH-133 (Fig. 5a). Inhibition of ERK1/2 with PD98059 decreased the levels of IL-6 and IL-8 by 52% and 54% respectively, efficiently reducing the levels of the inflammatory cytokines to control values (Fig. 5a). Neither JWH-133 treatment nor the addition of PD98059 was toxic to the studied primary RPE cells (Fig. 5b), which is in line with our previous findings that unpassaged primary hRPE cells are more resistant to cell death than ARPE-19 cells 28 .

Discussion
The CB 2 receptor is predominantly expressed by immune cells 12 . Its potential to modulate the immune response might be beneficial in the treatment of diseases associated with chronic low-level inflammation, such as atherosclerosis, diabetes, and AMD 1,29 . The CB 2 receptor is highly inducible, and its expression increases strongly when microglia and other immune cells become activated in response to inflammatory stimuli 12 . CB 2 receptor activation by a specific agonist could potentially control inflammatory responses and delay or prevent the onset of disease. The finding that CB 2 receptors are expressed by RPE cells and that their activation protected these cells from oxidative stress-induced damage led to the suggestion that CB 2 activation might be a possible new treatment strategy for AMD 21 .
Wei et al. were the first to show that the activation of CB 2 could protect RPE cells from hydrogen peroxide-induced cell death 21 . In contrast, we found that the CB 2 agonist JWH-133 had no protective effect on RPE cell survival after an exposure to the reactive aldehyde HNE. HNE is a product of lipid peroxidation and one of the most abundant oxidative stressors in the retina 30 . JWH-133 could not protect RPE cells from HNE-mediated death and even augmented the toxicity at a 10 µM concentration. At the same time, 10 µM JWH-133 increased the production of pro-inflammatory cytokines IL-6, IL-8, and HMGB1. This is in line with the results from Schmöle et al. who showed that CB 2 deletion reduced the production of IL-6 in LPS-treated microglia, suggesting that CB 2 can act as a pro-inflammatory factor under specific conditions 18 . CB 2 knockout mice also showed diminished inflammation in response to severe induced sepsis compared to wild type mice 31 . However, multiple other groups have shown that the activation of CB 2 leads to reduced, rather than increased production of pro-inflammatory cytokines. Activation of CB 2 reduced the release of pro-inflammatory cytokines in LPS-induced uveitis 19 , and JWH-133 reduced the production of IL-6 in both, TNFα-stimulated fibroblast-like synoviocytes 20 and a model of acute induced pancreatitis in mice 32 . Our results indicate that in RPE cells, CB 2 activation causes increased inflammation, which could aggravate the pathogenesis of AMD.
CB 2 modulation has resulted in contradictory findings in the past and the activation of CB 2 is known to cause different reactions in cells depending on the choice of agonist, the activation status of the cells, or the cell type 12,13 . CB 2 activation by the endocannabinoid 2-arachidonylglycerol induces migration in immune cells 33,34 , while other CB 2 activators, both chemical and biological, are known to inhibit this migration 33,35 . The knockout of CB 2 reduced the production of pro-inflammatory cytokines in LPS-stimulated CB 2 −/− microglia and reduced the levels of cytokines and infiltrating microglia in the brain in an Alzheimer's disease mouse model 18 . At the same time, in a model of controlled cortical impact injury, neuroinflammation was increased in CB 2 -knockout mice compared to wild-type animals 36 . It appears that both the activation and the inhibition of CB 2 can exert proinflammatory, as well as anti-inflammatory effects depending on the context and the local circumstances of the employed disease model 12 .
The complexity of CB 2 receptor activation extends to the signalling pathways underlying its immunomodulatory effects. Research has shown that CB 2 can influence different signalling pathways, including mitogen-activated protein kinase (MAPK) and cyclic adenosine monophosphate (cAMP) signalling 13 . To complicate matters, previous studies have suggested that CB2 activation can either increase 26,27 or decrease 35,37 the phosphorylation, and thereby the activity, of MAPK ERK1/2. CB 2 activation has also been shown to increase intracellular Ca 2+ levels 25 . In our study [Ca 2+ ] i is increased after the addition of JWH-133 to ARPE-19 cells, which is in line with previous observations 25 . Wei et al. were the first to report the expression of the CB 2 receptor in ARPE-19 and primary human RPE cells, showing both, mRNA and protein expression of CB 2 21 . Our results, indicating a Ca 2+ -response after JWH-133 treatment, provide further evidence for the presence of the CB 2 receptor in RPE cells. Additionally, CB 2 activation led to increased ERK1/2 phosphorylation, while the inhibition of ERK1/2 with a specific inhibitor reduced the JWH-133-induced secretion of IL-6 and IL-8 back to control levels in hRPE cells. This suggests that the activation of ERK1/2 is directly associated with the JWH-133-induced production of proinflammatory cytokines. We have previously shown that the inhibition of ERK1/2 can reduce inflammation in HNE-treated  23 , which is in line with our current results. Additionally, increased [Ca 2+ ] i after CB 2 activation could be involved in the release of pro-inflammatory cytokines. Calcium responses and ERK1/2 activation working in tandem, have been shown to be involved in the endothelin 1-induced production of IL-6 in human airway smooth muscle cells 38 , as well as in the production of IL-8 in oxysterol-treated monocytes 39 . Figure 6 illustrates a possible pathway of CB 2 activation-linked inflammation in RPE cells. Future studies analysing the benefits of calcium channel blockers on JWH-133-induced inflammation in RPE cells could shed further light on the importance of the observed calcium response.
It is worth noticing that CB 2 agonists are highly lipophilic compounds with a potential for unspecific binding 12 . However, increased intracellular Ca 2+ levels coupled with an increased ERK1/2 phosphorylation is in line with previous findings related to CB 2 activation 12,26,27 , indicating that our results are facilitated by CB 2 . Additional studies in different models, such as CB 2 -knockout mice could provide additional clarity concerning the role of CB 2 in RPE cell-associated inflammation.
In summary, our results show that the activation of the CB 2 receptor has detrimental effects on RPE cells, leading to increased pro-inflammatory cytokine production in an ERK1/2-dependent manner. Interestingly, endocannabinoid levels are increased in the retina of AMD patients 22 , which could suggest that CB 2 activation plays an important role in the chronic inflammation underlying the disease. Further studies are necessary to fully elucidate the role of (endo)cannabinoids and their receptors in AMD. Nevertheless, our findings suggest that CB 2 activation might contribute to AMD development rather than prevent it.

Materials and Methods
Cell culture. ARPE  Results are shown as mean ± SEM and combined from 3 independent repetitions with 2-4 parallels per group. ns denotes not statistically significant, *denotes P < 0.05, **denotes P < 0.01, ***denotes P < 0.001; Mann-Whitney U-test.  strong pigmentation and little to no contamination with fibroblast-like cells. Culture medium was changed for all cells twice per week during routine culture. All cells were incubated at +37 °C in a humidified atmosphere supplemented with 5% CO 2 .
Cell treatments. ARPE-19 cells were treated on 12-well plates to which they were seeded at a density of 200.000 cells/ml/well and incubated for 48 h until confluent. hRPE cells were treated on fully confluent 24-well plates. All cells were washed once prior to treatments with serum-free maintenance medium supplemented with 1% bovine serum albumin (BSA; Roche, Basel, Switzerland). Cells were treated with the known CB 2 agonist JWH-133 (Tocris Bioscience, Bristol, UK) and incubated for 24 h before the collection of serum and protein samples. Cell Viability Assays. Cell viability was assessed with the lactate-dehydrogenase assay (LDH), the neutral red assay, or the 3-(4,5-dimethyldiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, Sigma-Aldrich, St. Louis, MO, USA) assay. LDH was measured from medium samples according to the manufacturer's instructions using the commercially available CytoTox96 ® Non-Radioactive Cytotoxicity Assay (Promega, Fitchburg, WI, USA).
The neutral red assay was performed as described by Repetto et al. 41 using 96-well plates to which cells were split at a density of 15000 cells/100 µl medium/well. The MTT assay was performed according to our laboratory's standard protocol, which has been described before 42 . Calcium measurements. For Calcium imaging experiments, ARPE-19 cells were cultivated onto plastic glass bottom Petri dishes (Mattek Corp., USA; 3.5 cm in diameter) in 1:1 mixture of DMEM and Nutrient Mixture F12 medium (both obtained from Sigma, Steinheim, Germany) supplemented with 10% fetal calf serum (FCS) and 1% antibiotics (penicillin-streptomycin; Sigma, Steinheim, Germany) until complete attachment and full confluence in the presence of 5% CO 2 at 37 °C. The Petri dish central part had a 14 mm glass bottom, with the remaining surface made from cell culture plastic and was poly-d-lysine coated. For monitoring of cytosolic free calcium concentrations ([Ca 2+ ] i ), the ARPE-19 cells were loaded with AM ester of Fura-2 (Fura-2 AM; Invitrogen-Molecular Probes, USA). For loading, Fura-2 AM in DMSO was diluted in 2 ml culture medium to a final concentration of 3 μM and added to the cells. Cells were loaded in the incubator at 37 °C for 1 hr. After loading, the ARPE-19 cells were washed twice for 7 min with culture medium. The Petri dish with the ARPE-19 cells was then mounted onto an inverted Zeiss Axiovert S 100 microscope (Carl Zeiss AG, Oberkochen, Germany). In order to evoke calcium responses in ARPE-19 cells, either 5 μM cannabinoid receptor agonist JWH-133 or 1% bovine serum albumin (BSA) alone, as a control, were applied. The application as well as its washout from the bath was driven by the hydrostatic pressure of a 35 cm of water column and controlled manually. Image acquisition was done with a 12-bit cooled CCD camera SensiCam (PCO Imaging AG, Kelheim, Germany). The software used for the acquisition was WinFluor (written by J. Dempster, University of Strathclyde, Glasgow, UK), while the optical objective used was 63x ⁄ 1,25 oil Plan-NeoFluar (Zeiss), and the light source was XBO-75W (Zeiss) Xe arc lamp. The light intensity was attenuated when necessary with grey filters with optical densities 0.5, 1 and 2 (Chroma Technology Corp., Bellows Falls, VT, USA). The excitation filters used and mounted on a Lambda LS-10 filter wheel (Sutter Instruments Co.) were 360 and 380 nm (Chroma). Excitation with the 360 nm filter (close to the Fura-2 isosbestic point) allowed observation of the cells' morphology and of the changes in the concentration of the dye, irrespective of changes in [Ca 2+ ] i , while the 360⁄380 nm ratio allowed visualization of the [Ca 2+ ] i changes in the cytoplasm. Image acquisition, timing and filter wheel operation were all controlled by WinFluor software via a PCI6229 interface card (National Instruments, Austin, TX, USA). Individual image frames were acquired every 500 ms resulting in frame cycles which were 1 second long (two wavelengths).

Statistical Analysis.
Results from ELISA and cell viability assays were analysed using GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA). The data were tested for statistical significance by pairwise analysis of treatment groups using the Mann-Whitney U-test and a value of P < 0.05 was considered statistically significant.
Data Availability. All data generated or analysed during this study are included in this published article.