**3. Results**

#### *3.1. RASF Express CB1 and CB2*

In our experiments, RASF were treated with TNF (10 ng/mL) for 72 h before we conducted our experiments and, although the expression of CB1 and CB2 in SF was already documented [28], their regulation by TNF was only investigated by our group but under different experimental conditions [45]. We found little CB1 expression at the cell surface but high intracellular levels that were not significantly regulated by TNF (Figure 1). CB2 was exclusively found at the plasma membrane and it was upregulated by TNF (*p* = 0.05) (Figure 1).

**Figure 1. Flow cytometric detection of CB1 and CB2 in and on RASF.** RASF were incubated with TNF (10 ng/mL) for 72 h, and CB1 and CB2 levels were determined thereafter. Upper panel: histogram; detection of CB1 at the plasma membrane (PM) and intracellularly and CB2 at the plasma membrane. Lower panel: violin plots; quantification of CB1 and CB2. *t*–test was used for comparisons. *p* = 0.05 was the level of significance.

#### *3.2. THC Increases Intracellular Calcium in RASF Primed with TNF*

Target receptors for THC include TRP ion channels [3] and CB1 and activation [46] of either is coupled to elevations or reductions in intracellular calcium levels, respectively. Without TNF priming, THC (0.1–25 μM) did not modulate intracellular calcium levels (Figure 2A). However, when RASF were treated with TNF 72 h prior to THC addition, we detected a significant increase (up to ~200%) in intracellular calcium levels in response to THC (*p* < 0.001; 5–25 μM) (Figure 2B). This increase was not inhibited by the CB1 antagonist/inverse agonist rimonabant (Figure 2C) and slightly modulated by the CB2 antagonist COR170 (Figure 2D). TRPA1 is strongly upregulated by TNF [47] and, since it is also a receptor for THC, we inhibited this channel with ruthenium red (RR) (Figure 2E). RR not only increased basal intracellular calcium levels (*p* < 0.001), but also reduced THC-induced intracellular calcium levels (5 μM and 10 μM THC; *p* = 0.032 and *p* < 0.001, respectively). Previous results from our group sugges<sup>t</sup> that TRPA1 is located intracellularly [45,47] and, since RR cannot actively cross the plasma membrane [48], we also employed specific lipophilic TRPA1 antagonists (Figure 2F,G). We found that both HC030031 (Figure 2F) and A967079 (Figure 2G) inhibited the stimulatory effects of THC (*p* < 0.001 for 5 μM and 10 μM THC (Figure 2F); *p* < 0.001 for 5 μM–25 μM THC (Figure 2G)) on intracellular calcium levels over a wide range of THC concentrations. The latter was more potent, since, in contrast to HC030031, it also reduced calcium elevations in response to the highest concentration of THC (25 μM, *p* < 0.001). We also conducted these experiments with PBS (Figure 2H–M) instead of HBSS, establishing a calcium-free extracellular environment. Under these conditions, alterations in intracellular calcium levels can only be elicited by emptying intracellular stores. Similar results compared to the HBSS groups were obtained but, under these conditions, the TRPA1 antagonist A967079 (Figure 2M) completely abrogated all effects of THC on intracellular calcium.

**Figure 2. Mean intracellular calcium level changes in RASF in response to THC.** (**A**) Intracellular calcium mobilization without TNF pre–stimulation. (**B**,**C**) Intracellular calcium level regulated by THC (0.1−25 μM) with TNF (10 ng/mL) pre–stimulation for 72 h and extracellular calcium (HBSS; (**B**–**G**)) or without calcium (PBS; (**H**–**M**)). \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05 for differences between antagonist treatment and control (THC only). (**B**,**H**) Comparisons of different THC concentrations versus control (no THC). Significant values are given in the graph. ANOVA with Bonferroni post hoc test was used for all comparisons. Rimonanbant, CB1 inverse agonist; COR170, CB2 inverse agonist; A967079, TRPA1 antagonist; HC-030031, TRPA1 antagonist; RR = Ruthenium Red, general TRP inhibitor.

#### *3.3. THC Enhances PoPo3 Uptake in RASF Primed with TNF*

In a previous study, we established PoPo3 as a surrogate marker for drug uptake [47], which was coupled to intracellular calcium levels and, therefore, we also assessed the ability of THC (0.1–25 μM) to modulate PoPo3 uptake. Without TNF priming, PoPo3 uptake was only slightly increased by THC (Figure 3A), whereas, after TNF stimulation, THC robustly increased PoPo3 uptake (Figure 3B; *p* < 0.001 for [c] 1 μM−25 μM of THC). Rimonabant modulated PoPo3 uptake only at low THC concentrations (0.1–1 μM), but the magnitude of uptake at these concentrations was rather small (Figure 3C). The CB2 inverse agonist COR170 also reduced PoPo3 uptake by THC (Figure 3D; *p* < 0.001; 1 μM, 10 μM, 25 μM THC). RR modulated PoPo3 levels to all but the highest concentration of THC (Figure 3E), but increased rather than decreased its uptake. Like intracellular calcium, PoPo3 uptake was almost completely inhibited by the TRPA1 antagonists HC030031 (Figure 3F; *p* < 0.001 for all [c] of THC except 0.1 μM) and A967079 (Figure 3G, *p* < 0.001 for all [c] of THC except 0.1 μM). Decynium-22 (D22), an inhibitor of organic cation transporters [47], inhibited PoPo3 uptake alone (Figure 3H, blue line, *p* = 0.018) and together with THC (Figure 3H, *p* < 0.001 for all [c] of THC). Lastly, we investigated whether THC itself can block subsequent effects of added THC in higher concentrations and we found that it indeed inhibited further PoPo3 uptake by higher concentrations of THC (Figure 3I; *p* < 0.001, 10 μM and 25 μM). We also assessed the ability of THC to induce PoPo3 uptake without extracellular calcium in PBS (Figure 3J–Q). We confirmed our findings from the HBSS groups, but the CB2 antagonist COR170 showed a higher efficacy in calcium-free conditions (Figure 3L; *p* < 0.001, for all [c] of THC). RR, HC030031, A967079 and D22 also inhibited Popo3 uptake almost completely (Figure 3M–O; *p* < 0.001 for all [c] of THC, except 0 μM and 0.1 μM in the A967079 and D22 group). THC itself also reduced PoPo3 uptake, but the effect was attenuated compared to the conditions with extracellular calcium (Figure 3Q).

**Figure 3. Mean PoPo3 uptake by RASF in response to THC.** (**A**) PoPo3 uptake without TNF pre– stimulation, (**B**,**C**) by THC with TNF (10 ng/mL) pre–stimulation for 72 h and extracellular calcium (HBSS; (**B**–**I**)) or without calcium (PBS; (**J**–**Q**)). \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05 for differences between antagonist treatment and control (THC only, Figure 2B,J). (**B**,**J**) Comparisons of different THC concentrations versus control (no THC). Significant values are given in the graph. ANOVA with Bonferroni post hoc test was used for all comparisons. Rimonanbant, CB1 inverse agonist; COR170, CB2 inverse agonist; A967079, TRPA1 antagonist; HC-030031, TRPA1 antagonist; RR = Ruthenium Red, general TRP inhibitor, Decynium-22, organic cation transport inhibitor.

#### *3.4. THC Reduces Cytokine Production Only at High Concentrations*

Besides intracellular calcium and PoPo3 uptake, we assessed whether THC (0.1–25 μM) also modulates cytokine production by RASF. We identified TRPA1 as an important target receptor for THC and, therefore, we induced its expression by stimulating RASF for 72 h

with TNF before adding THC. THC did not modulate IL-6 or IL-8 production significantly but it blunted MMP-3 levels either alone or in combination with A967079 or rimonabant at 25 μM (Figure 4C, *p* < 0.001). IL-8 production was only reduced by THC (25 μM) when combined with A967079 (*p* < 0.001) or rimonabant (*p* = 0.005). Cell viability was slightly enhanced by THC at 5 μM (*p* = 0.022) but extensive cell death occurred at 25 μM (*p* < 0.001) (Figure 4D). In addition, cell viability was slightly increased when THC (1 μM, 5 μM and 10 μM) was combined with rimonabant or A967079, but the magnitude was small.

#### *3.5. THC Has Negligible Effects on PBMC Cytokine, IgM and IgG Production*

In synovial tissue, endocannabinoids are abundantly produced not only by RASF, but immune cells are also capable of producing anandamide and 2-AG [28,29,49]. Since lymphocytes and macrophages are also present in RA synovial tissue where these cells closely interact with RASF [27], we investigated the impact of THC (1 and 10 μM) on peripheral blood mononuclear cells (PBMC) alone or in co-culture with RASF (Figure 5). In co-culture with RASF, THC did not modulate IL-6 and IL-10 production but decreased TNF production when PBMC/RASF were stimulated with CpG or IFN-γ (1 μM THC, *p* = 0.027 and *p* = 0.010, respectively) (Figure 5C). Immunoglobulin G production induced by CpG was further enhanced by 1 μM and 10 μM THC, but it did not reach significance (*p* = 0.077 and *p* = 0.085, respectively) (Figure 5E). Without RASF, 10 μM THC reduced IL-10 levels in response to IFN-γ (*p* = 0.011) and CpG (*p* = 0.03) in PBMC (Figure 5G). Immunoglobulin

G production was fostered by 1 μM THC without any additional stimulus (*p* = 0.026) (Figure 5J).

**Figure 5. Cytokine production by human PBMC monoculture and co-culture with RASF in response to THC.** (**A**,**B**) IL-6; ( **C**,**D**) IL-10; (**E**,**F**) TNF; ( **G**,**H**) immunoglobulin M (IgM) and (**I**,**J**) immunoglobulin G (IgG) production by PBMC and PBMC/RASF co-culture over 7 days. Cells were concomitantly stimulated with THC and the respective activation stimulus (IFN-γ or CpG). In coculture experiments, RASF were stimulated with 10 ng/mL IFN-γ 72 h prior to PBMC addition to induce MHC II expression and induce an allogeneic T cell response. ANOVA with Bonferroni post hoc test was used for all comparisons. \* *p* < 0.05.
