Cortisol-mediated adhesion of synovial fibroblasts is dependent on the degradation of anandamide and activation of the endocannabinoid system




In rheumatoid arthritis (RA) synovial fluid, levels of the endocannabinoids anandamide (AEA) and 2-arachidonylglycerol are elevated. Since synovial fibroblasts (SFs) possess all of the enzymes necessary for endocannabinoid synthesis, it is likely that these cells contribute significantly to elevated endocannabinoid levels. While glucocorticoids initiate endocannabinoid synthesis in neurons, this study was undertaken to test whether cortisol also regulates endocannabinoid levels in mesenchymal cells such as SFs, and whether this interferes with integrin-mediated adhesion.


Adhesion was determined in 1-minute intervals over 60 minutes using an xCELLigence system. Slopes from individual treatment groups were averaged and compared to the control. Fatty acid amide hydrolase (FAAH) and cyclooxygenase 2 (COX-2) were detected by immunocytochemistry, and AEA was detected by mass spectrometry.


Cortisol increased the adhesion of RASFs and osteoarthritis SFs with a maximum of 200% at both 10−7M and 10−8M. When cortisol was administered together with either cannabinoid receptor 1 (CB1) antagonist (rimonabant; 100 nM), CB2 antagonist (JTE907; 100 nM), transient receptor potential vanilloid channel 1 (TRPV-1) antagonist (capsazepine; 1 μM), FAAH inhibitor, or COX-2 inhibitor, adhesion was reduced below the level in controls. Concomitant inhibition of FAAH and COX-2 reversed these effects. Mass spectrometry revealed the presence of AEA in SFs.


Our findings indicate that glucocorticoid-induced adhesion is dependent on CB1/CB2/TRPV-1 activation. Since AEA is produced in SFs, this endocannabinoid is the most likely candidate to mediate these effects. Since AEA levels are regulated by COX-2 and FAAH, inhibition of both enzymes along with low-dose glucocorticoids may provide a therapeutic option to maximally boost the endocannabinoid system in RA, with possible beneficial effects.

The endocannabinoid system is an evolutionarily old system that controls neurotransmitter release in the brain and immune functions in the periphery (1, 2). Activation of the endocannabinoid system plays a beneficial role in many autoimmune diseases and cancers by decreasing immune cell migration and cytokine production or by increasing apoptosis (3–6). Since endocannabinoids are synthesized on demand, circulating levels remain low (1). Levels of endocannabinoids are increased by inactivation of the main metabolizing enzymes, fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (7). However, recent studies also identified cyclooxygenase 2 (COX-2) as a major contributor to endocannabinoid metabolism, and some effects of COX-2 inhibitors are a result of the action of endocannabinoids (8–10). Anandamide (AEA) and 2-arachidonylglycerol have been detected in rheumatoid arthritis (RA) synovial tissue but not in healthy joints (11). A beneficial effect of the elevation of endocannabinoid levels by FAAH inhibition was previously demonstrated in collagen-induced arthritis in mice (12).

Synovial fibroblasts (SFs) have all of the enzymes necessary for the production of endocannabinoids, and they also express cannabinoid receptor 1 (CB1) and CB2 as well as the endocannabinoid-degrading enzyme FAAH (11, 13). In addition, SFs express transient receptor potential vanilloid channel 1 (TRPV-1), a nonselective cation channel that is activated intracellularly by AEA (14, 15). Furthermore, our laboratory recently demonstrated the expression of G protein–coupled receptor 18 (GPR-18) and GPR-55 in SFs and described the nuclear localization of CB1 in SFs (Lowin, et al: unpublished observations). Nuclear and intracellular CB1 receptors have different functions than CB1 located at the cell membrane and are activated only by AEA produced or applied intracellularly; exogenously added AEA has no effect (16).

Central levels of endocannabinoids are controlled by glucocorticoids via different neuronal feedback loops. Previous studies showed that glucocorticoids, via membrane-bound G protein–coupled receptors, lead to the generation of endocannabinoids which suppress excitatory transmissions within the paraventricular nucleus of the hypothalamus (17). This decreases the output of corticotropin-releasing hormone in corresponding neurosecretory cells. Consequently, overall cortisol levels are reduced. Endocannabinoids limit the activation of the hypothalamic–pituitary–adrenal axis, thereby reducing stress (18). Their inactivation or inhibition of CB1 exacerbates stress responses (19). Until now it has been unclear whether this interplay is also present in the periphery.

In SFs from RA patients (RASFs) and SFs from osteoarthritis patients (OASFs), glucocorticoids inhibit the production of proinflammatory cytokines, chemokines, and matrix metalloproteinases. It has been demonstrated that cortisol increases the adhesion of SFs from OA and RA patients to fibronectin, and that the effect on adhesion was in part due to an up-regulation of the major fibronectin-binding integrin subunit α5 (20).

Integrin-mediated adhesion of cells is one prerequisite for initiating migration and invasion (21). In some cases, however, enhanced adhesion as a result of increased integrin levels also reduces cell dissemination (22). Consistent with this observation, we previously found an increase in RASF migration in response to knockdown of α5 integrin (20). This demonstrates that adhesion needs to be tightly controlled and influences the migratory capacity of the cell. While an enhancement of adhesion by proinflammatory stimuli, such as chemokines, also increases migration and cartilage destruction, our group demonstrated that glucocorticoid-induced adhesion promotes an antiinflammatory phenotype by reducing these parameters (20, 23).

Integrins mediate adhesion of SFs to various components of the extracellular matrix (24). They consist of an α-subunit that determines ligand specificity and a β-subunit, which initiates intracellular signaling. Integrin binding and affinity is regulated via “outside-in” and “inside-out” signaling (22). The latter involves a complex series of intracellular signaling events that results in an active or inactive (depending on the stimulus) integrin conformation (25). While the active conformation promotes adhesion, the inactive conformation supports detachment and is associated with a quiescent phenotype (26). Due to an enhancement of proinflammatory signaling, integrins play a major role in the progression of RA (22).

In many cases, integrin affinity is controlled by the activation of G protein–coupled receptors and associated calcium flux. This is evident by the activation of Gαi-coupled chemokine receptors on leukocytes that rapidly change integrin affinity (26). The importance of CB2 in adhesion is emphasized by its role in immature B cell retention in bone marrow sinusoids (27). In T cells, however, adhesion is controlled by CB1 activation (28).

Since cell attachment is enhanced by cortisol, we hypothesize that cortisol mediates this effect indirectly by the generation of endocannabinoids. In this study, we demonstrate that cortisol-induced adhesion is dependent on the activation of CB1, CB2, and TRPV-1. We provide evidence that AEA is produced by SFs and that manipulation of its intracellular levels has profound influences on adhesion. These findings might be applied in clinical use, since concomitant inhibition of COX-2 and FAAH is necessary to fully exploit the therapeutic properties of endocannabinoids. This is especially important in inflammatory conditions where COX-2 is up-regulated due to the action of proinflammatory cytokines.



The study group comprised 41 patients with longstanding RA who fulfilled the American College of Rheumatology revised criteria (29) and 46 patients with OA. These patients underwent elective knee joint replacement surgery. Written informed consent was obtained from all patients. The study was approved by the Ethics Committee of the University of Regensburg. Basic clinical and laboratory data on the patients are presented in Table 1. C-reactive protein levels were measured by standard techniques.

Table 1. Characteristics of the patients*
 Patients with OA (n = 46)Patients with RA (n = 41)
  • *

    OA = osteoarthritis; RA = rheumatoid arthritis; CRP = C-reactive protein; NSAIDs = nonsteroidal antiinflammatory drugs; NA = not available.

Sex, no. female/ male29/1731/10
Age, mean ± SEM years68.1 ± 1.562.1 ± 1.7
CRP, mean ± SEM mg/liter4.1 ± 1.410.7 ± 3.5
 No. taking NSAIDs3225
 No. taking prednisolone119
 Daily prednisolone dosage, mean ± SEM mgNA6.3 ± 0.8
 No. taking methotrexateNA4
 No. taking sulfasalazineNA1

Preparation of SFs and synovial tissue.

Synovial tissue samples from patients with OA and patients with RA were obtained immediately after opening the knee joint capsule; samples were prepared as previously described (30). Pieces of synovial tissue of up to 9 cm2 were excised. One part of the tissue sample was cut, placed in protective freezing medium (Tissue-Tek; Sakura Finetek), and stored at −80°C until used. Another part of the synovial tissue specimen was minced and placed in Dispase I (Roche Diagnostics). Digestion was carried out for at least 1 hour at 37°C on a shaking platform. The resulting suspension was filtered (70 μm) and spun at 300g for 10 minutes. The pellet was then treated with erythrocyte lysis buffer (20.7 gm NH4Cl, 1.97 gm NH4HCO3, 0.09 gm EDTA, and 1 liter H2O) for 5 minutes and recentrifuged for 10 minutes at 300g. The pellet was resuspended in RPMI 1640 (Sigma-Aldrich) with 10% fetal calf serum (FCS). A total of 1,000,000 cells were transferred to a 75-cm2 tissue culture flask. After overnight incubation, cells were supplemented with fresh medium.

Stimulation of RASFs and OASFs.

For stimulation of SFs, 25,000 cells were transferred into 24-well culture plates (Fisher Scientific) and cultivated until >60% confluence was reached. Then, culture medium (RPMI 1640 with 100 units/ml penicillin, 100 μg/liter streptomycin, 10% FCS, 4 mM L-glutamine, 2.5 μg/ml amphotericin, and 12.5 mM HEPES) was exchanged for serum-free (without FCS) RPMI 1640 medium (Sigma-Aldrich) prior to stimulation. The cells were stimulated with cortisol (10−10–10−6M; Sigma-Aldrich), JNJ1661010 (FAAH inhibitor; 10−7M) (Tocris), nimesulide (COX-2 inhibitor; 10−6M) (Tocris), CP47.497 (CB1 agonist; 10−12–10−6M) (Cayman/Biozol), rimonabant (CB1 antagonist/inverse agonist; 10−7M) (BioTrend), GP1a (CB2 agonist; 10−12–10−7M) (Tocris), JTE907 (CB2 antagonist/inverse agonist; 10−7M) (Tocris), or capsazepine (TRPV-1 antagonist; 10−6M) (Tocris) for 24 hours. Thereafter, cells were detached with Accutase (PAA Laboratories), washed, resuspended in RPMI1640 with 10% FCS, and used for adhesion assays without further stimulation.

Flow cytometric analysis for the detection of α5 integrin and active β1.

SFs were detached with Accutase/phosphate buffered saline (PBS) (1:1; PAA Laboratories) and centrifuged at 300g. Cells were resuspended in 100 μl PBS/1% FCS with primary antibody (α5 integrin [clone SAM-1; 1:200] [Millipore] and active β1 [clone 2Q839; 1:300] [US Biologicals/Biomol]). After 30 minutes of incubation at room temperature, cells were washed (in PBS with 1% FCS) and secondary antibody (goat anti-mouse RPE; 5 ng/μl) (DakoCytomation) in 100 μl PBS/1% FCS was added. After 30 minutes, cells were washed and analyzed in the flow cytometer.

Monitoring adhesion with an xCELLigence system.

The xCELLigence system (Roche Applied Sciences) enables real-time monitoring of cellular events by using impedance measurements across interdigitated microelectrodes integrated on the bottom of tissue culture microtiter plates (E-plates). Impedance changes occur if cells attach to E-plates or change their size, shape, and number due to specific treatments.

For adhesion experiments, E-plates (Roche) were coated for 1 hour at room temperature with 100 μl of a solution of 10 μg/ml fibronectin (Becton Dickinson) in PBS. To block nonspecific binding, 1% bovine serum albumin in PBS was added for 30 minutes. Then, 2,000–5,000 cells that had been pretreated as described above were added to the E-plates, and an xCELLigence system was programmed to monitor adhesion every minute for 240 minutes. Within 60 minutes after the addition of cells, adhesion remained linear, so this time point was chosen as the end point for adhesion. Adhesion was quantified by averaging slopes from 1 minute to 60 minutes after the experiment started. Untreated cells were used as controls, and the value in controls was set to 100%. Experiments were performed in duplicate. A typical xCELLigence experiment is depicted in Figure 1. The plates were equilibrated with medium, and baseline cell index values were determined. For cell addition, E-plates were removed from the device (for ∼13 minutes). Adhesion was then monitored for 1 hour. For data analysis, averaged slopes over the time course of 60 minutes were compared. Experiments were repeated at least 5 times with different donor RASF or OASF.

Figure 1.

Adhesion of synovial fibroblasts as monitored by xCELLigence software over 1 hour. Each curve represents a different treatment. Bars show the average slope of 60 measurements per treatment. [c] = concentration.


For the detection of FAAH and COX-2, the following antibodies were used: for FAAH, rabbit polyclonal antibody raised against residues 561–579 (CLRFMREVEQLMTPQKQPS) (Cayman/Biozol); and for COX-2, goat polyclonal antibody raised against residues 578–596 (TINASSSRSGLDDINPTVL) (Abcam). Cells were fixed with 4% formaldehyde and permeabilized with ice-cold methanol. Slides were blocked with 1% BSA in PBS and incubated with primary antibody for 3 hours. Sections were washed and incubated with secondary Cy3-conjugated goat anti-rabbit antibody or rabbit anti-goat antibody for 2 hours (Cy3 was obtained from Jackson Immunoresearch). The following isotype controls were included: for COX-2, normal goat IgG; and for FAAH, normal rabbit IgG (both from R&D Systems).

AEA extraction.

RASFs were treated with 1 μM nimesulide, 100 nM JNJ1661010, and 10 nM cortisol for 2 hours in serum-free RPMI 1640. Then supernatant (6 ml) was transferred into a 50-ml plastic tube, and 6 ml methanol/chloroform (2:1 volume/volume) was added. The remaining cells were washed extensively with PBS. Cellular AEA was extracted by the addition of 2 ml ice-cold methanol and subsequent scraping of cells. The methanol cell extract was transferred into a 50-ml plastic tube and PBS/chloroform were added, yielding equivalent amounts of each solvent in supernatant and cell pellet samples. Tubes were inverted several times and punctured at the bottom, and chloroform phases were collected in 1.5-ml reaction tubes. Solvent was evaporated under vacuum. Lyophilized samples were reconstituted in 100 μl acetonitrile and stored protected from light at −20°.

Detection of AEA.

Analysis of AEA was performed by liquid chromatography tandem mass spectrometry using an Agilent 1200 SL HPLC system coupled to an Applied Biosystems Q-Trap 4000 mass spectrometer. An Atlantis T3 reversed-phase column (inner diameter 2.1 mm, length 150 mm, and particle size 3 μm; Waters) was used. Chromatographic separation was achieved with a binary solvent system consisting of mobile phase A (0.1% formic acid in water [v/v]) and mobile phase B (0.1% formic acid in acetonitrile [v/v]). Gradient elution started with 50% B, which was ramped to 95% B within 1 minute and held at 95% for 7 minutes. The percentage of B was reduced to 50% from 8–8.1 minutes and held at 50% for 2.9 minutes. The total run time was 11 minutes. The solvent flow was 300 μl/minute. The column was kept at 25°C, and a sample volume of 10 μl was injected. Detection was carried out by electrospray ionization in positive mode and multiple reaction monitoring using the transitions described below. The column flow was not split for detection. D8-AEA (Cayman/Biozol) was used as an internal control. For multiple reaction monitoring, a quadrupole 1 (Q1) mass of 348.3 daltons and a quadrupole 3 (Q3) mass of 62 daltons were selected for AEA, and a Q1 mass of 356.3 daltons and a Q3 mass of 63 daltons were selected for D8-AEA. A dwell time of 50 msec, a declustering potential of 52V, collision energy of 33V, and a cell exit potential of 10V were used for both AEA and D8-AEA.

Statistical analysis.

Statistical analysis was performed with SigmaPlot 11 (Systat Software). Group medians were compared by the nonparametric Mann-Whitney test. The generalized linear model (GLM) with Bonferroni correction was used to compare the effects of treatment over all doses. P values less than 0.05 were considered significant.


Cortisol increases the adhesion of OASFs and RASFs to fibronectin in a dose-dependent manner.

In previous studies we found that high concentrations of cortisol (>10−6M) stimulated adhesion to fibronectin in an α5 integrin–dependent manner (20). In the present study we investigated whether low to intermediate cortisol levels also exerted an influence on adhesion to fibronectin. Cortisol dose-dependently increased adhesion to fibronectin, reaching a maximum at doses between 10−8M and 10−7M in both OASFs and RASFs (Figure 2A). The effects of cortisol on adhesion were blocked by RU-486, a glucocorticoid receptor antagonist (Figure 2B). RU-486 (2 μM) administered alone did not change adhesion significantly (data not shown). Adhesion was not dependent on the regulation of integrins, as evidenced by the finding that α5 integrin levels did not increase with 10−8M cortisol and only moderately increased with 10−7M cortisol (Figure 2C). The number of β1 integrins in active conformation, which promote binding to fibronectin, was significantly reduced by 15% at 10−6M, 10−7M, and 10−10M cortisol but was unaltered at 10−8M and 10−9M cortisol (Figure 2C).

Figure 2.

Adhesion and integrin regulation of rheumatoid arthritis synovial fibroblasts (RASFs) and osteoarthritis SFs (OASFs) in response to cortisol. A, Adhesion of OASFs (n = 10) and RASFs (n = 7) to fibronectin after 24 hours of treatment with cortisol. P values were determined by generalized linear model with Bonferroni correction. B, Adhesion of OASFs (n = 6) and RASFs (n = 4) to fibronectin after 24 hours of treatment with cortisol and the glucocorticoid receptor antagonist RU-486. P values were determined by nonparametric Mann-Whitney test. C, Regulation of α5 integrin in OASFs (n = 5) and RASFs (n = 7) and active β1 integrin in OASFs (n = 5). Values are the mean ± SEM. ∗ = P < 0.05 versus control, by nonparametric Mann-Whitney test. [c] = concentration.

Inhibition of cortisol effects on adhesion by CB1, CB2, and TRPV-1 antagonism.

To investigate the contribution of the endocannabinoid system to adhesion, SFs were incubated with cortisol (10−10–10−6M) and either the CB1 antagonist/inverse agonist rimonabant (100 nM) (Figure 3A), the TRPV-1 antagonist capsazepine (1 μM) (Figure 3B), or the CB2 antagonist/inverse agonist JTE907 (100 nM) (Figure 3C). Rimonabant, capsazepine, and JTE907 completely blocked the stimulatory effect of cortisol on adhesion in RASFs and OASFs when entire curves were compared to cortisol alone (P ≤ 0.001 by GLM). Rimonabant (100 nM), capsazepine (1 μM), or JTE907 (100 nM) alone did not significantly influence adhesion in RASFs and OASFs (data not shown). To verify that CB1 activation had a stimulatory effect on attachment, a selective agonist for this receptor, CP47.497, was used and adhesion was assessed. CP47.497 increased adhesion in a dose-dependent manner, with a maximum of ∼150% adhesion with 10−8M CP47.497 in OASFs and RASFs (Figure 3D). At concentrations of <10−10M, CP47.497 reduced adhesion to ∼80%. OASFs responded to CP47.497 over a wider dose range than did RASFs. The influence of CB2 agonism on adhesion was tested with GP1a, a high-affinity CB2 agonist. Stimulation with GP1a significantly increased adhesion in RASFs at concentrations of 10−12M, 10−11M, and 10−9M and in OASFs at concentrations of 10−11M, 10−10M, and 10−9M. Thus, cortisol-induced adhesion resembled adhesion induced by CB1 and CB2 agonists (compare Figure 2A to Figures 3D and E).

Figure 3.

Adhesion of OASFs and RASFs to fibronectin. A, Influence of cannabinoid receptor 1 (CB1) antagonism on cortisol-induced adhesion in OASFs (n = 7) and RASFs (n = 9). B, Influence of transient receptor potential vanilloid channel 1 (TRPV-1) antagonism on cortisol-induced adhesion in OASFs (n = 5) and RASFs (n = 6). C, Influence of CB2 antagonism on cortisol-induced adhesion in OASFs (n = 7) and RASFs (n = 9). D, Influence of CB1 agonism on adhesion in OASFs (n = 7) and RASFs (n = 6). E, Influence of CB2 agonism on adhesion in OASFs (n = 7) and RASFs (n = 7). Values are the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.001; ∗∗∗ = P < 0.001 versus control, by nonparametric Mann-Whitney test. See Figure 2 for other definitions.

Blocking of the cortisol-induced increase in adhesion in OASFs and RASFs by inhibition of AEA degradation.

Before assessing the effects of FAAH and COX-2 inhibition on adhesion, we verified that both proteins were expressed in OASFs and RASFs (Figure 4A). COX-2 is able to degrade endocannabinoids, including AEA, and metabolites from this pathway activate several receptors distinct from classic G protein–coupled cannabinoid receptors (31). To unmask the contribution of AEA to adhesion, SFs were stimulated with cortisol (10−10–10−6M) and either the FAAH inhibitor JNJ1661010 (100 nM), the COX-2 inhibitor nimesulide (1 μM), or both inhibitors. FAAH inhibition blocked the increase in adhesion induced by cortisol at intermediate concentrations (10−7–10−8M) (Figure 4B). The effect was more pronounced in OASFs, where FAAH inhibition blocked the action of 10−7M and 10−8M cortisol, whereas in RASFs only the effects of 10−7M cortisol were significantly inhibited (Figure 4B) (see Figure 2A for comparison). Nimesulide was more potent and reduced adhesion over all concentrations compared to cortisol alone (P ≤ 0.001 by GLM) (Figure 4C). When nimesulide (1 μM) and JNJ1661010 (100 nM) were applied simultaneously along with cortisol, the suppressive effect of each inhibitor on adhesion was reversed (Figure 4D). At 10−8M cortisol, concomitant COX-2 and FAAH inhibition restored adhesion to levels observed with cortisol alone (Figure 4D).

Figure 4.

Visualization of fatty acid amide hydrolase (FAAH) and cyclooxygenase 2 (COX-2) expression in OASFs and RASFs and adhesion of OASFs and RASFs to fibronectin in response to COX-2 and FAAH inhibition. A, Visualization of FAAH and COX-2 protein in OASFs and RASFs by immunofluorescence. Original magnification × 400. B, Influence of FAAH inhibition on cortisol-induced adhesion in OASFs (n = 7) and RASFs (n = 10). ∗ = P < 0.05; ∗∗∗ = P < 0.001 versus controls, by nonparametric Mann-Whitney test. C, Influence of COX-2 inhibition on cortisol-induced adhesion in OASFs (n = 5) and RASFs (n = 6). D, Influence of concomitant FAAH and COX-2 inhibition on cortisol-induced adhesion in OASFs (n = 8) and RASFs (n = 10). In B–D, values are the mean ± SEM. In C and D, P values were determined by generalized linear model with Bonferroni correction. See Figure 2 for other definitions.

JNJ1661010 (100 nM) without cortisol increased adhesion to 159% (P = 0.035) in OASFs (n = 11) but did not significantly increase adhesion in RASFs (n = 12) (data not shown). Nimesulide (1 μM) alone did not significantly change adhesion to fibronectin in OASFs (n = 8) but increased it to 109% in RASFs (P = 0.035; n = 6) (data not shown).

Stimulation of AEA production in SFs by cortisol.

Bioinformatics analysis revealed that SFs have the principal capacity to synthesize AEA (13). To investigate whether this actually occurs during cortisol treatment, RASFs were incubated with 10−8M cortisol, the FAAH inhibitor JNJ1661010 (100 nM) and the COX-2 inhibitor nimesulide (1 μM) for 2 hours. While no AEA was detected in medium control samples, extracts from treated cells were found to contain AEA (Figure 5). Although no exact quantification was possible due to its intracellular location, estimates from AEA standard samples suggest amounts between 100 pg and 300 pg per ∼700,000 cells (0.15–0.3 fg/cell).

Figure 5.

Chromatogram of a sample obtained from methanol/chloroform extraction of rheumatoid arthritis synovial fibroblasts. The large peak represents anandamide. cps = counts per second.


In this study, we demonstrated for the first time that glucocorticoids influence the endocannabinoid system in the periphery. We showed that cortisol-induced adhesion of OASFs and RASFs to fibronectin is regulated by intracellular endocannabinoids and that manipulation of their levels by inhibition of either FAAH or COX-2 decreases adhesion, while the administration of both inhibitors together reverses the effects of each inhibitor alone. AEA was detected in RASFs, and this endocannabinoid is a likely candidate for mediating cortisol-induced adhesion via the subsequent activation of CB1, CB2, and TRPV-1.

In previous experiments, our laboratory found that cortisol in concentrations of >1 μM increased the adhesion of OASFs and RASFs to fibronectin, and this was partly dependent on the up-regulation of α5 integrin (20). In the present study, the effects of lower cortisol concentrations on adhesion were investigated. We found that cortisol in concentrations below the threshold necessary to influence integrin levels maximally increased the adhesion of OASFs and RASFs.

Adhesion is not only influenced by surface integrin levels but also by intracellular proteins, such as Ras GTPases and kindlin-3, that regulate the affinity of integrins to ligand (32, 33). We confirmed integrin affinity to be unaltered or even slightly reduced at high cortisol concentrations. Since integrin levels and activation state are stable, an increase in avidity by integrin clustering might be responsible for the increase in adhesion (34, 35). Thus, endocannabinoids might assist in the formation of focal adhesions and rearrangement of the actin cytoskeleton in SFs. One important second messenger that affects integrin avidity and affinity directly and indirectly via adaptor proteins is calcium (36).

Chemokine receptors are coupled to Gαi, and they mediate rapid up-regulation of integrin affinity by raising intracellular calcium levels. Another prerequisite for chemokine-induced adhesion is the influx of extracellular calcium (37), a signaling molecule that may also be responsible for the inhibiting effects of the CB1 antagonist rimonabant and the TRPV-1 antagonist capsazepine on cortisol-induced adhesion.

Like chemokine receptors, CB1 is coupled to Gαi, and we recently identified its intracellular location in SFs (Lowin, et al: unpublished observations). The intracellular CB1 receptor pool, however, is associated with calcium mobilization from the endoplasmic reticulum, lysosomes, or the nuclear envelope (16, 38). Analogous to chemokine-induced adhesion, cortisol-induced adhesion also required the entry of extracellular calcium. Previous studies have shown that AEA mediates extracellular calcium entry by activating TRPV-1 at its intracellular binding site (39). This study provides evidence that TRPV-1 influences cortisol-induced adhesion, since antagonism of this nonselective cation channel by capsazepine reduced adhesion.

Therefore, we propose the following mechanism for cortisol-induced adhesion. First, cortisol induces the generation of AEA by a nongenomic mechanism, since glucocorticoids increase adhesion to a similar extent after 2 hours (data not shown) and after 24 hours. This nongenomic mechanism involves classic glucocorticoid receptors, as indicated by the fact that adhesion was inhibited by RU-486, which is not efficacious at membrane glucocorticoid receptors (40). Second, AEA activates intracellular CB1 in SFs. This is mimicked by the synthetic CB1 agonist CP47.497 and also by the use of AEA as stimulus (Lowin, et al: unpublished observations). Activation of CB1 leads to an increase in intracellular calcium levels, and since calcium is an inducer of endocannabinoid synthesis, more AEA is synthesized (7). Finally, when intracellular levels of AEA become high enough to activate TRPV-1, extracellular cation/calcium influx occurs and integrin affinity is increased. Furthermore, there is spillover of AEA in the supernatant where CB2 is activated, supporting adhesion. Assuming an average amount of AEA of 0.22 fg/cell with a cell body diameter of 10 μm, the intracellular AEA concentration then equals 400 μg/liter (1.15 μM). This is consistent with previously reported concentrations of AEA needed for the activation of TRPV-1 (41). A high level of intracellular AEA is maintained by binding to specialized fatty acid binding proteins. By this means, AEA is stored inside the cell and reaches a high intracellular concentration (42). The mechanism for cortisol-induced adhesion proposed here is further supported by the fact that exogenous AEA increases adhesion in a dose-dependent manner, with a bell-shaped distribution, similar to cortisol (Lowin et al: unpublished observations).

Furthermore, this study revealed that the levels of AEA and possible metabolites have to be tightly controlled to facilitate proper adhesion. FAAH is the main metabolizing enzyme of AEA, but COX-2 is gaining recognition as a major contributor to the action of AEA since it generates J-series prostaglandins that act as potent proapoptotic compounds in various cancers (5, 10). In this study, we demonstrated that both enzymes are constitutively expressed in OASFs and RASFs. While COX-2 expression has been linked to the generation of J-series prostaglandins that activate peroxisome proliferator–activated receptor γ in SFs, this is the first study to identify FAAH protein in SFs (43, 44). Degradation of AEA by FAAH yields ethanolamine and arachidonic acid, and the latter is preferentially used to generate N-arachidonylglycine, an endocannabinoid that activates the putative cannabinoid receptor GPR-18 (45, 46).

If AEA degradation is partially blocked by FAAH inhibition, AEA reaches higher intracellular levels and degradation is redirected to COX-2. Consequently, adhesion may be influenced by prolonged or enhanced activation of CB1, CB2, and TRPV-1; an increase in COX-2 metabolites of AEA; or, depending on degradation by COX-2, a possible spillover of AEA or COX-2 metabolites into the extracellular space, where different receptors are available for binding.

In the case of COX-2 inhibition, AEA is solely degraded by FAAH. If N-arachidonylglycine is formed in SFs due to the enhanced availability of arachidonic acid derived from AEA, activation of GPR-18 might explain the decrease in adhesion, since agonism at this receptor decreases the adhesion of SFs (Lowin, et al: unpublished observations).

When AEA degradation was completely blocked by FAAH and COX-2 inhibition, adhesion was reconstituted. During AEA synthesis and accumulation in the cell, time-dependent spillover of AEA into the extracellular space also occurs (39). If both degrading enzymes are inhibited, spillover is maximal, and extracellular AEA might reach concentrations necessary for a maximal and long-lasting activation of CB2; activation of this receptor promotes adhesion (27). Extracellular AEA might also activate the putative cannabinoid receptor GPR-55, which is also linked to integrin clustering and affinity regulation (47).

The findings presented here demonstrate for the first time that RASFs are able to synthesize AEA and that endocannabinoid signaling is used by cortisol to mediate adhesion. SFs might therefore be responsible for the elevated levels of AEA in OA and RA synovial tissue (11). Furthermore, we demonstrated that adhesion mediated by the generation of intracellular endocannabinoids is sensitive to COX-2 and FAAH inhibition, suggesting that endocannabinoid levels or endocannabinoid metabolites are relevant. This is important since COX-2 is present throughout the RA synovium and might therefore contribute significantly to endocannabinoid metabolism (48). The action of COX-2 inhibitors has been linked to an increased endocannabinoid tone, and this might also be a major effect of COX-2 therapy in RA patients (49). The therapeutic value of COX-2 inhibitors and glucocorticoids in RA might be enhanced with concomitant FAAH inhibition, since inhibition of this enzyme was beneficial in mouse models of arthritis (12). Treatment of RA with low-dose glucocorticoids would boost AEA production in SFs while concomitant COX-2 and FAAH inhibition would steadily increase its levels. Furthermore, high endocannabinoid levels promote adhesion of SFs, and this might decrease the migratory potential of these cells. Our findings shed new light on the role of the endocannabinoid system and demonstrate its role as an important intracellular messenger system that is activated by glucocorticoids.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Lowin had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Lowin, Straub.

Acquisition of data. Lowin, Zhu.

Analysis and interpretation of data. Lowin, Zhu, Dettmer-Wilde.


The authors thank Angelika Gräber for excellent technical assistance.