Adrenergic receptors in osteoarthritic and rheumatoid synovial broblasts: Identication of β3 as novel player to modulate IL-6 and p38 activation

Background: Rheumatoid arthritis (RA) is inuenced by the activity of the sympathetic nervous system (SNS). In animal models of RA, the SNS promotes severity of the disease and its manipulation modulates experimental arthritis depending on timing of the intervention. Synovial broblasts (SF) are major contributors to RA pathology but their modulation by the SNS has been rarely investigated. In this study we assessed the expression and function of adrenergic receptors in RA and osteoarthritis (OA) synovial broblasts and investigated their downstream signaling. Methods: We used western blot and quantitative PCR (qPCR) to determine protein and mRNA of adrenergic receptors in OASF/RASF. Furthermore we determined α 1a and β 2 protein in synovial tissue by immunouorescence. ELISA was employed to determine IL-6 production. p38 kinase activation and translocation was analyzed by cell-based ELISA and immunouorescence. Results: We detected α 1a , α 2b , β 1 , β 2 and β 3 protein in OASF/RASF and α 1a and β 2 protein in synovial tissue of OA and RA patients. The pro-inammatory cytokines IFN-γ and TNF downregulated β 3 adrenergic receptor. Activation of α 1a , α 2b , β 2 and β 3 increased production of TNF-induced IL-6 which was inhibited by specic antagonists. Furthermore, β 3 agonism enhanced p38 phosphorylation and translocation to the nucleus. Conclusion: Among a comprehensive characterization of the adrenergic system of OASF/ RASF, we report for the rst time β 3 expression and demonstrated that this adrenergic receptor participates in the inammatory response of synovial broblasts. Therefore, modulation of β 3 might pose a new therapeutic opportunity to modulate synovial broblast function in patients with RA.


Introduction
Rheumatoid arthritis (RA) is an autoimmune disease which is characterized by in ammation of the joints, autoantibody and proin ammatory cytokine production. The immune reaction in RA is supported by the sympathetic nervous system (SNS), since sympathectomy before the induction of experimental arthritis in mice results in less severe symptoms [1]. The SNS supports migration of immune cells to lymph nodes and enhances IFN-γ production in the spleen [2,3]. However, in the late phase of the disease the in uence of the SNS changes from pro-to anti-in ammatory, which might be explained by the appearance of IL-10 producing, regulatory B cells which are generated under the in uence of norepinephrine [4]. In addition, tyrosine hydroxylase (TH) positive immune cells, capable of producing own catecholamines, appear in in amed tissue [5,6], which is considered as a compensatory mechanism, since SNS bre density decreases in in amed tissue [7,8]. As a consequence, late sympathectomy further aggravates experimental arthritis, most likely due to deletion of these TH + cells [9].
While the effects of norepinephrine on T cells, B cells and monocytes/macrophages are well described, the situation is less clear in rheumatoid arthritis synovial broblasts (RASF) [10]. RASF are a major contributor of joint destruction and they produce copious amounts of proin ammatory cytokines such as IL-6 and matrix degrading enzymes [11]. RASF migration and cytokine production are in uenced by dopamine [12,13], which is synthesized by blood lymphocytes [14]. Norepinephrine is released by sympathetic nerve terminals but also by TH + cells [15] and it in uences cytokine production of RASF [16,17]. Although the effects of norepinephrine on RASF have been described it is still unclear which adrenergic receptors (AR) in particular are involved. Wu et.al showed the surface expression of β 2 adrenergic receptors [17], but it is not known whether β 1 and β 3 -AR are also expressed. Similarly, αadrenergic stimulation by low-dose norepinephrine in uences cytokine levels in RASF, but the respective receptor subtypes have not been identi ed [16].
In this study, we determined the AR subtypes expressed by RASF and osteoarthritis synovial broblasts (OASF) and determined the effects of AR receptor stimulation on the production of the pro-in ammatory cytokine IL-6 and on activation of the MAP kinase p38, which has been shown to be a downstream target of β 2 AR [18].
Since antibodies against AR can be unspeci c [19], we also employed quantitative PCR to detect α 1a , α 2b , β 1 and β 3 mRNA expression in RASF (Fig. 1E). Here, we con rmed expression and the downregulation of AR by cytokines as demonstrated in western blot analyses. In addition, we detected β 1 AR mRNA which veri es the presence of this AR as shown on the protein level (Fig. 1E). β 2 was not investigated by PCR since we used several different antibodies and techniques ( ow cytometry, western blotting, immuno uorescence) to detect and verify this AR (data not shown). α and β AR in synovial tissue Besides isolated OASF and RASF we were also interested in the expression of AR in OA and RA synovial tissue since it might be possible that in vitro culture conditions alter the levels of AR on synovial broblasts. We detected α 1a , β 1 (RA only, Fig.2B) and β 2 AR in OA and RA synovial tissue and these AR costained with CD55, a marker for synovial broblasts ( Fig. 2A). The antibodies against α 2b and β3 used in western blot were not suitable for immuno uorescence. In addition, we also co-stained the AR with CD68 (marker for macrophages, suppl. Fig. 2A) and CD3 (marker for T cells, suppl. Fig. 2B) and found that macrophages and some T cells also express α 1a and β 2 AR. Furthermore, we detected α 1b AR to be expressed by a few CD3 low cells in synovial tissue ( Fig. suppl. 2C), suggesting that a T cell subpopulation expresses this AR.
Modulation of IL-6 release from OASF and RASF stimulated with TNF by AR agonists TNF, a proin ammatory cytokine, is one important factor in RA and anti TNF treatment ameliorates disease activity in around 30% -50% of patients [20]. TNF is a potent inducer of IL-6 expression [21], and, therefore we assessed the in uence of AR agonists on IL-6 production. We found that 1h preincuabtion with the α 1 agonist phenylephrine, the α 2 agonist dexmedetomidine hydrochloride and the unselective β agonist isoproterenol increased TNF-induced IL-6 production by OASF (Fig. 3A, B, C), whereas dexmedetomidine and BRL increased IL-6 production by RASF (Fig. 3G, I). These effects were speci c since the effects by AR agonists were inhibited by their respective antagonists in OASF (Fig. 3A

AR agonists increase IL-6 without further stimulation with proin ammatory cytokines
We demonstrated that AR ligands can further increase TNF-induced IL-6 production by OASF and RASF. However, we were also interested whether AR stimulation increases IL-6 levels without concomitant cytokine stimulation. For this purpose, we incubated OASF and RASF with the β 2/3 agonist BRL37344 ( Fig. 3K) and found that BRL alone was able to increase IL-6 in RASF (~50% increase at 10 -10 M) (Fig. 3K).

Activation of β 3 AR regulates p38 phosphorylation and translocation into the nucleus
Our results show that the β 3 AR is expressed and functional in OASF and RASF and the importance of this receptor is underscored by the potential to upregulate IL-6 in RASFs without further cytokine stimulus. Therefore, we focused on this receptor for further characterization of downstream signaling. Since it is known that β 3 agonists foster p38 activation in cardiomyocytes and adipocytes [22,23], we investigated p38 phosphorylation and translocation to the nucleus in OASF and RASF by cell-based ELISA and immunocyto uorescence. Cell-based ELISA demonstrated a time-dependent increase in p38 phosphorylation in OASF and RASF (Fig. 4A). The activation and translocation of p38 induced by BRL37344 (10 -7 M) was visualized in Fig. 4B and C. In OASF, the signal for phosphorylated p38 appeared already after 1min and remained for at least one hour (Fig. 4B). In RASF, phosphorylated p38 was detectable after 3 minutes and nuclear translocation was still evident after 1h following stimulation with BRL ( Fig. 4C).

Discussion
In this study, we assessed for the rst time the protein expression of ARs in OASF and RASF. Besides the already described β 2 AR, we detected α 1a , α 2b , β 1 and β 3 AR protein and mRNA in OASF and RASF lysates.
The discovery of β 3 expression in broblasts was unexpected since this receptor is only expressed by very few cell types such as adipocytes [23]. Overall, stimulation of AR resulted in increased IL-6 production, which is usually considered as proin ammatory since IL-6 neutralization is one important therapeutic intervention in RA [24] and is required for the successful induction of collagen-induced arthritis in mice [25]. However, proin ammatory properties of IL-6 are mediated by trans signaling (stimulation of cells devoid of own IL-6 receptor by acquiring soluble IL-6 receptor and signaling transducer gp130) whereas classical signaling is mostly anti-in ammatory [26]. IL-6 is important for metabolic control as IL-6 -/mice develop metabolic disturbances [27]. Therefore, β AR stimulation might increase IL-6 to control metabolism and energy expenditure since one major function of β ARs is to mobilize energy rich substrates [28]. This is bene cial in short term in ammatory episodes but detrimental in chronic in ammation [28]. On the other hand, IL-6 production induced by α ARs might directly support proin ammatory immune cells. Of note, we did not investigate the effect of α AR stimulation without the addition of TNF. It might be that activation of α ARs in the absence of in ammation does not in uence IL-6 production. We found that α and β ARs both increase IL-6 production and this might not depend on G protein but β-arrestin signaling which can be employed by both α and β ARs [29,30]. In line with this, βarrestin signaling can be pro-or anti-in ammatory, but under TNF-stimulated conditions, proin ammatory effects predominate [31].
Since β 3 ligation alone entailed upregulation of IL-6 in RASF without further cytokine stimulus, we investigated downstream signaling of this AR and found that BRL37344 was able to induce p38 phosphorylation and nuclear translocation in OASF and RASF.
Western blotting, quantitative PCR and immuno uorescence in OA and RA synovial tissue showed the expression of α 1a , α 2b , β 1, β 2 and β 3 in OASF, RASF but also in other immune cell populations present in synovial tissue emphasizing the important role of adrenergic mechanisms for regulation of joint in ammation [10]. Western blotting further revealed that Cortisol increased the expression of α 1a (in RASF) , α 2b (in OASF), β 2 (in OASF) and β 3 (in OASF) protein. This is in line with early results from Lefkowitz who showed a stimulatory effect of cortisol on β AR [32]. In addition, upregulation of the α 1b AR by the glucocorticoid dexamethasone was demonstrated due to enhanced transcription of the respective mRNA [33]. Since only OASF showed an upregulation of α 2b , β 2 , and β 3 in response to cortisol, RASF might have lost their ability to adequately respond to this glucocorticoid. In fact, it has been demonstrated that some loss-of function polymorphisms in the glucocorticoid receptor α gene are present in RA patients [34]. In addition, some RA patients have a relative preponderance of glucocorticoid β over glucocorticoid α receptors, which is considered a proin ammatory signal, since glucocorticoid β receptors antagonize the DNA binding of the α receptor subtype [35]. We do not consider drug therapy with glucocorticoids as contributing factor for this difference between OASF and RASF, since SFs were used after several passages and therefore any acute effect of GC therapy would not be relevant anymore. Nevertheless, there might be epigenetic alterations by GC therapy [36]. We also observed decreased β 3 (and β 2 in OASF) protein levels after stimulation with TNF or IFN-γ. Similar effects are known for b2 AR since it was shown that TNF blunts the ability of the unselective β agonist Isoproterenol to relax smooth muscle cells by desensitization and therefore possible downregulation of the receptor [37]. Although western blot results for β 1 AR were inconclusive due to incorrect molecular weight of detected bands, immuno uorescence showed expression of this receptor in RASF. However, as discussed below, antibodies against β AR in general are not utterly speci c and the molecular weight of β AR can vary widely [19].
The functional impact of AR stimulation was assessed by analyzing IL-6 production by OASF and RASF. We found that α 1a , α 2b , β 1/2 and β 3 activation increased TNF-induced IL-6 production by OASF and RASF (albeit weaker) which was inhibited by respective antagonists, except for the unselective β agonist isoproterenol. Similarly, β 3 activation without additional TNF also increased IL-6 levels in RASF. This is in line with data from Burger et al. and Tanner et al. who demonstrated enhanced IL-6 production in cardiac broblasts solely in response to norepinephrine in an α and β AR dependent manner [38,39]. Also, Raap et al. showed a stimulatory or inhibitory effect of norepinephrine on IL-6 and IL-8 production by OASF and RASF, respectively, without further cytokine stimulation and depending on used norepinephrine concentration [16].
Interestingly, when we combined the β 3 -AR agonist BRL37344 with the unselective β-AR agonist Isoproterenol, we detected a decrease rather than an increase of IL-6 production by RASF. This suggests some degree of antagonism between the three types of β-AR, since activation of β 3 -AR alone increased IL-6. This might be related to differential signaling induced by individual β-AR. While e.g. the β 2 -AR couples to the PKA activating G protein Gαs with a switch to inhibitory Gαi after prolonged incubation, β 3 -AR can bind both Gαs and Gαi simultaneously leading to distinct signaling events [40].
One important kinase involved in IL-6 production by synovial broblasts is p38 [41], and therefore we investigated the activation of this map kinase in response to β 3 -AR activation. We found that BRL37344 induced p38 phosphorylation and translocation to the nucleus in OASF and RASF. Similar results have been obtained in adipocytes, where p38 was identi ed as downstream target of β 3 -AR [42]. In addition, p38 was also involved in β 3 -AR signaling in cardiomyocytes [22]. The activation of p38 is not restricted to the β 3 -AR, since earlier studies by our group also demonstrated p38 phosphorylation by ligation of β2-AR [18].
Limitations of our study One major challenge detecting speci c isoforms of α and β AR is the lack of speci city with a lot of the commercially available antibodies. In a study by Hamdani it was shown that antibodies raised against β 1 or β 2 recognized all three β isoforms without distinction [19]. The same was true for antibodies raised against β 3, which also labelled β 1 and β 2 [43]. Similar problems have been demonstrated by using an antibody against α 1 AR, which detected all α 1 subtypes including additional non-speci c bands [44].
Although we did not use any of the α/β AR antibodies investigated in the above mentioned studies, there is still the possibility that the antibodies used in our experiments are not speci c. Therefore, we also con rmed the expression of respective ARs by quantitative PCR. Similar problems might arise with the ligands used in our study. Although e.g. BRL37344 is sold as a speci c β 3 AR agonist, it also binds to β 2 with similar a nity [45]. However, we also used the β 3 antagonist L-748,337, which is 20 fold (vs β 2 ) and 45 fold (vs β 1 ) more selective at β 3 AR [46].

Conclusion
In this study, we investigated for the rst time in a comprehensive manner, which ARs are present on OASF and RASF, respectively. A completely novel nding is the presence of β 3 AR on OASF and RASF which might turn out to be a major responder to sympathetic stimuli in the joint, as this receptor was able to modulate IL-6 without further cytokine stimulus. In addition, we found that α and β AR stimulation modulates IL-6 production and also revealed that β 3 activation induces p38 phosphorylation and translocation into the nucleus. These data suggest that intervention with the AR system especially the β 3 AR poses a therapeutic possibility to dampen proin ammatory activity of synovial broblasts in RA.

Synovial broblast and tissue preparation
Samples from RA and OA synovial tissue were isolated and prepared as described previously [48]. After opening of the knee joint capsule, synovial tissue samples were obtained immediately. Synovial tissue of 9 cm 2 was excised, part of which was cut off and stored in a protective freezing medium at −80°C until further use (Tissue Tek, Sakura Finetek, Zoeterwoude, The Netherlands). The other part was chopped and treated overnight at 37 ° C with liberase (Roche Diagnostics, Mannheim, Germany). The resulting suspension was ltered (70 µm) and centrifuged at 300 g for 10 minutes. The pellet was then treated with erythrolysis buffer (20.7 g NH 4 Cl, 1.97 g NH 4 HCO 3 , 0.09 g EDTA ad 1L H 2 O) for 5 minutes, and centrifuged again for 10 minutes at 300 g. Cells were resuspended in RPMI-1640 (sigma Aldrich, St. Louis, USA) with 10% FCS. The number of cells was calculated by a Neubauer cell counting chamber. A total of 1,000,000 cells were transferred to a 75 square centimeter tissue culture ask. After overnight culture, cells were supplemented with fresh medium.
Stimulation of OA and RASFs 5000 cells were seeded onto 96 well microtiter plates, grown for three days and were then incubated with or without TNF (10 ng/ml) and AR agonists and antagonists for 24h in RPMI medium containing 2% FCS to minimize proliferation; for all assays. Cell-free supernatants were collected (18-24h after TNF-α stimulation).

IL-6 ELISA
Cell culture supernatants were used for ELISAs 24 h (IL-6) after addition of related AR ligands. The test was carried out according to the supplier's description (BD, OptEIA, Heidelberg, Germany). The coe cient of variation between and within batches was less than 10%.
Samples were covered with ProLong Gold Antifade Mountant (Thermo Fisher) and visualized. Isotype IgG was used as negative control.

Statistical analysis
All the data are presented from at least of three independent experiments. Statistical analysis was performed with GraphPad Prism (GraphPad software Inc, California, USA) and SPSS 25 (IBM, Armonk, USA). The statistic tests used are given in the gure legends. The level of signi cance was p < 0.05. When data are presented as line plots, the line represents the mean. When data are presented as bar charts, the top of the bar represents the mean and error bars depict the standard error of the mean (sem).

Declarations
Competing interests: The authors declare that they have no competing interests.
Ethics approval: This study was approved by the local ethics committee of Düsseldorf (2018-87-KFogU) Funding: This work was supported by an unlimited grant of the Hiller Foundation and the Deutsche Forschungsgemeinschaft (DFG, PO801/8-1).
Disclosure: The authors have nothing to disclose. Data availability statement: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.  Table   Table 1 was not provided with this version.