The β₂-Adrenergic Agonist Salbutamol Is a Potent Suppressor of Established Collagen-Induced Arthritis: Mechanisms of Action¹ Free

Interleukin-12 secreted by myelomonocytic cells is a critical factor promoting the development of Th1 cells (1). Salbutamol, a β₂-adrenergic agonist widely used in asthma as a bronchodilator, has recently sparked renewed interest as a potential immunomodulatory agent since it was shown that β₂-agonists inhibit IL-12 production by macrophages and dendritic cells stimulated with LPS or via interaction of CD40-CD40 ligand, thus preventing Th1 development (2, 3). This action was mediated via the β₂-adrenoreceptor and correlated with increased intracellular cAMP levels (3).

Elevation of intracellular cAMP can be achieved either by direct stimulation of adenylate cyclase, e.g., by PGE₂, or via activation of the adrenoreceptors, or by inhibition of phosphodiesterase inhibitor (PDE),⁴ a cAMP-degrading enzyme. PDE inhibitors such as pentoxifylline inhibit IL-12 production via cAMP up-regulation (4). There are reports that accumulation of intracellular cAMP reduces the secretion of TNF by monocytes/macrophages (5, 6, 7) and augments their production of the antiinflammatory cytokine IL-10 (8). In addition, cAMP is involved in regulating the production of reactive oxygen species (9) and of nitric oxide (10). Taken together, all these characteristics make drugs that increase intracellular cAMP attractive candidates for the treatment of Th1-driven chronic inflammatory autoimmune diseases.

CIA, a murine model for RA is elicited by immunizing mice with type II collagen in CFA and is thought to be predominantly a Th1-mediated disease. The disease is characterized by joint inflammation and Th1 responses against type II collagen (11, 12). IL-12 knockout mice or mice that have been treated with neutralizing anti-IL-12 Abs before onset of disease develop little or no arthritis (13, 14). Suppression of the inflammatory process by blocking TNF with mAb has proved to be a successful treatment of CIA (15, 16). We recently observed that combining anti-IL-12 with anti-TNF synergistically suppressed clinical progression of CIA and prevented joint damage.⁵ Because of the reported effects of salbutamol on both TNF and IL-12 production, we decided to investigate the therapeutic potential of this compound in CIA. We report here that salbutamol attenuates ongoing CIA and that it might be promising in the treatment of Th1-mediated diseases.

Bovine CII was purified from hyaline cartilage, as previously described (16). Male DBA/1 mice (8–12 wk old) were immunized with 200 μg of CII emulsified in CFA (Difco, Detroit, MI) by intradermal injection at the base of the tail.

From the 15th day after immunization and onwards, mice were examined daily for onset of CIA using two clinical parameters, paw swelling and clinical score (16). Paw swelling was assessed by measuring the thickness of the affected hind paws with 0–10 mm calipers (Kroeplin, Schluchtern, Germany). For the clinical score a four-point scale was used: 0 = normal; 1 = slight swelling and erythema; 2 = pronounced edema; 3 = joint rigidity. Each limb was graded, resulting in a maximal clinical score of 12 per animal. The arthritis was monitored for 10 days after which the mice were sacrificed.

Treatment commenced at the onset of disease, and salbutamol (Sigma, Dorset, U.K.) was administered i.p. either daily or every other day, at the specified doses, until day 10 of arthritis. In each experiment, a group of mice was injected with PBS alone, which served as a control.

Hind paws and knee joints were removed postmortem on the 10th day of arthritis, fixed in formalin, and decalcified in 5% EDTA. Paws and knees were then embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Arthritic changes in the ankle, the metatarsophalangeal joints, the proximal interphalangeal joints, and the distal interphalangeal joints were scored by an observer blinded from treatment received. The histological changes were graded as mild (mild synovial hyperplasia), moderate (pannus formation and erosions limited to the cartilage-pannus junction), or severe (extended bone and cartilage erosions with loss of joint architecture).

To localize and identify mast cells, entire tissue sections were stained with safranin O and examined under an Olympus B_H2 microscope. To assess the proportion of degranulated/activated cells from the total number of mast cells, automated image analysis (AnalySIS, Soft Imaging System GmbH, Munster, Germany) was applied. All sections were assessed under ×400 magnification. Three microscopic images per specimen (in distal interphalangeal, proximal interphalangeal, or metatarsophalangeal joints) were recorded with a Sony three-chip color camera. Quantitative analysis was performed according to optical density of the mast cells and their size. Staining values were corrected for the background staining in each specimen and mast cells with reduced staining intensity and a cell size >18 μm were enumerated as activated cells. Mast cell degranulation was assessed in nine randomly picked mice of the PBS-control group and in nine randomly chosen salbutamol-treated mice (200 μg every other day). Additionally, nine mice successfully treated with anti-TNF (mAb cV1q⁵) were evaluated for mast cell activation. All this was done by an observer, blinded from the treatment received.

Mice were sacrificed at day 3 after disease onset. Inguinal lymph nodes were excised, teased apart to make a single-cell suspension, washed, and then cultured in 96-well plates at a density of 1 × 10⁶ cells/ml (200 μl/well) in complete medium comprising RPMI (BioWhittaker, Verviers, Belgium), 10% heat-inactivated FCS, 1% glutamine, penicillin (100 U/ml) (BioWhittaker), streptomycin (100 μg/ml) (BioWhittaker), and 2-ME (2 × 10⁻⁵ M). Cells were cultured with or without bovine CII (50 μg/ml) solubilized in Tris-buffered saline, pH 7. Supernatants were collected after 72 h and stored at −20°C until cytokine measurement.

Ten-week-old DBA/1 mice were injected i.p. with 1% starch in PBS (2 ml). Mice were sacrificed 4 days later, and the peritoneum was rinsed. Lavage fluid was collected and centrifuged, and the cell pellet was resuspended at 4 × 10⁶ cells/ml. Cells were plated in 24-well culture plates at 1 ml/well and incubated for 2 h at 37°C. The cells were washed to remove nonadherent cells. The adherent cells were 90% macrophages, as shown by FACS analysis with the F4/80 marker (17) and the ER-BMDM1 marker (18) (data not shown). Macrophages were then pretreated with salbutamol, in a concentration range from 0.1 to 100 mM for 4 h, after which IFN-γ (13 ng/ml) was added overnight. The cells were subsequently stimulated with LPS (100 ng/ml) for 24 h. Supernatants were collected and frozen at −20°C until cytokine measurement.

As previously described (14, 19), mice were sacrificed at day 10 of arthritis, and the knee joints were removed. Synovial membranes were excised under a dissecting microscope and digested with collagenase A (1 mg/ml) (Boehringer-Mannheim, Mannheim, Germany) and DNase type IV (150 μg/ml) (Sigma) at 37°C for 20 min, in the presence of polymyxin B (33 μg/ml) (Sigma). The cells were then washed extensively and cultured in 96-well plates at a density of 2 × 10⁶ cells/ml (100 μl/well) in complete medium with or without salbutamol. Supernatants were collected after 24 h for TNF analysis, or at 72 h for IL-10 analysis and stored at −20°C until measured for cytokines.

The levels of immunoreactive TNF, IL-12p70, IL-10, and IFN-γ were measured by sandwich ELISA as have also previously been described (11, 14). The Ab pairs used were as follows (listed by capture/detection) TNF, rat mAb TX3 (obtained from the American Type Culture Collection (ATCC), Manassas, VA, courtesy of Dr. J Abrams, DNAX, Palo Alto, CA)/rabbit polyclonal Ab (kindly provided by Dr. W. Buurman, University of Limburg, Maastricht, The Netherlands); IL-12, rat mAbs 9A5/5C3 (gifts from Dr. D. Presky, Hoffman La Roche, Nutley, NJ); IL-10, rat mAbs 2A5/SXC-1 (ATCC and J. Abrams); IFN-γ, R4-6 A2/XMG1.2 (ATCC and J. Abrams). The sensitivity of TNF and IL-12p70 ELISAs was 10–40 pg/ml. The detection limit was 15 pg/ml for IL-10 and 10 pg/ml for IFN-γ.

The Mann-Whitney U test to compare nonparametric data for statistical significance was applied on all clinical results and cell culture experiments. The χ² test was used to analyze histological data.

Salbutamol was administered at a dose of 20 or 200 μg on alternate days for 10 days. It significantly slowed down the progression of clinical arthritis when compared with control mice treated with PBS, as demonstrated by paw thickness (Fig. 1,A) and clinical score (Fig. 1,B). The optimal effect was obtained with the higher dose. Increasing the frequency of injecting salbutamol 200 μg to once a day in a group of mice in the same experiment did not yield a better result than administering this dose every other day (Fig. 2). To test whether the drug might possibly act via a “hit and run” mechanism, another group was injected with 2 pulses of salbutamol at day 1 and day 3 of arthritis only, and the response was clinically monitored for another 8 days. This treatment protocol was equally potent in maintaining a low clinical score for 10 days (Fig. 2,A), whereas the effect on paw swelling was somewhat variable (Fig. 2 B), suggesting that this treatment does have sustained effects although maximal control of arthritis requires continuous dosing. With the treatment protocols used, no adverse effects were noticed.

Observer-blinded histological analysis of the joints in all the hind paws of the mice in the control group showed that 46% were severely destroyed, 46% were moderately or mildly affected, and only 8% were found to be normal (Table I), in keeping with previous studies (14). Salbutamol, 200 μg every other day, completely protected 38% of the joints whereas another 38% were only mildly affected and only 10% were destroyed (p < 0.02). As for the clinical assessment, the effect of salbutamol was dose dependent, and one-tenth the dose offered less protection against joint damage. Daily administration of the drug, while lacking additional clinical benefit, clearly protected the joints even better, with 50% of them found to be normal and only 10% severely destroyed (p < 0.02). On the other hand, injecting salbutamol just twice at the beginning of arthritis gave a suppression of the clinical score for a 10-day period, but histological analysis at the end of this period revealed that most joints were severely destroyed (50%), and only 12% were unaffected (Table I).

Stabilizing the mast cell membrane is a well-established property of β₂ agonists (20). Therefore, we decided to see whether salbutamol treatment prevented mast cell degranulation in the mice, in that this would provide a good indication of the in vivo activity of salbutamol. The total number of mast cells associated with one particular joint was counted for nine randomly picked mice in the PBS control group; all nine joints were scored as either moderately or severely affected. This number was remarkably constant: 23.8 ± 2.3 mast cells on average per joint (Table II). The same count was performed in nine randomly picked salbutamol-treated mice; all these joints were scored as either normal or mildly affected. The number of mast cells per joint was not different from the control mice: 24.8 ± 5.3 (Table II). There was, however, a striking reduction in the number of those mast cells that were actually degranulated: 10.2 ± 3.3 per joint in the treated mice (=41%) vs 18.1 ± 1.9 per joint (=76%) in the control mice. To test whether this finding was specific for salbutamol-treated animals, we also scored nine joints of successfully anti-TNF-treated mice from a previous study (19); all these joints were also scored as either normal or mildly affected. We found a reduced number of mast cells associated with these joints (Table II): 15.2 ± 2.6 per joint but, as in the PBS controls, 78% of these cells were degranulated (11.9 ± 2.8). Thus, salbutamol did not reduce mast cell numbers but did diminish degranulation, whereas anti-TNF reduced mast cell numbers without affecting degranulation.

To address the question of whether salbutamol can directly affect the function of murine macrophages, peritoneal macrophages were elicited by injection of 1% starch. Macrophages were pretreated with salbutamol, 0.1–100 μM, primed with IFN-γ and then costimulated with LPS 24 h later. Supernatants were assayed for IL-12, TNF, and IL-10. It was found that salbutamol directly suppresses the release of TNF and IL-12 in a dose-dependent manner (Fig. 3) without affecting cell viability as assessed by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (not shown). IL-10 was not detected in these cultures (not shown).

The most relevant site of cytokine production in arthritis is the synovial membrane, and we have described that synovial cells from arthritic DBA/1 mice, like those from human RA joints, spontaneously release bioactive TNF when cultured in vitro (14, 21). We tested whether salbutamol was able to block this spontaneous release when added in vitro. Synovial cells were cultured from six individual mice at day 10 of arthritis. Fig. 4 shows a dose-dependent suppression of TNF production by salbutamol, 1–100 μM. None of these synovial cell cultures produced IL-10; neither did those cultured in the presence of salbutamol (data not shown).

We wanted to address the question whether the ability of salbutamol to suppress IL-12 release by macrophages in vitro is relevant in the in vivo situation and leads to a down-regulation of the Th1 response to CII in treated animals. Mice were injected with salbutamol, 200 μg on day 1 and day 3 of arthritis. Six hours later, the mice were sacrificed and the CII-specific IFN-γ release by cultured inguinal LNC was measured. It was found that the CII-specific IFN-γ production was completely abrogated in the salbutamol-treated mice (Fig. 5). No concomitant rise in CII-specific IL-5 release was found, even by intracellular FACS analysis (not shown).

From earlier studies, it was postulated that β₂-agonists might be potentially useful in the treatment of Th1-mediated diseases such as multiple sclerosis or RA (2, 3). This hypothesis was based on the observation that β₂-agonists in general and salbutamol in particular inhibit the secretion of IL-12 from human monocytes and dendritic cells and block the development of Th1 cells from neonatal lymphocytes (3). Moreover, it was shown recently that activated Th1 cells, but not Th2 cells, express the β₂-adrenoreceptor and that IFN-γ production is suppressed on activation of this receptor (22, 23). A prior study had suggested a therapeutic effect of salbutamol in chronic relapsing EAE (24).

The aim of this study was to test this hypothesis in CIA, a murine model for RA, and a Th1-mediated disease (11, 12, 13). It was found that salbutamol, administered i.p. every other day after onset of clinical disease, had a pronounced therapeutic effect, both on clinical arthritis and in preventing joint damage. This result was somewhat surprising in view of the short half-life of salbutamol. Increasing the frequency of administration to once daily offered an even better protection against joint erosions. When salbutamol was injected twice only at the beginning of arthritis and treatment was then discontinued, clinical disease seemed to be attenuated quite effectively, but histological analysis after 10 days of arthritis revealed that joints were severely damaged.

In vitro studies revealed that in vivo administration of salbutamol resulted in blunting of the Th1 response against CII, as seen in a reduced CII-specific production of IFN-γ by LNC from treated mice sacrificed on day 3 of arthritis (Fig. 5). Thus, after only two injections (on day 1 and on day 3 of arthritis), salbutamol was able to abrogate an established Th1 response against CII. As predicted from the recent literature, salbutamol exerted a dose-dependent suppression on the IL-12p70 release by IFN-γ/LPS-stimulated peritoneal macrophages. In addition, salbutamol blocked the release of TNF-α, as shown here and described in the literature for β₂- agonists (7). These effects may contribute to the blockade of a Th1 response. We have recently described that anti-TNF treatment attenuates the Th1 response against CII and moreover that a marked synergy exists between anti-TNF and anti-IL-12 in diminishing CII-stimulated IFN-γ release by LNC from arthritic mice and that there is a synergy in the therapeutic effect of these two Abs in CIA. Moreover, blockade of bioactive TNF release by synovial cells from arthritic mice by salbutamol (Fig. 4) contributes to its antiarthritic effect through suppression of inflammation. The antiinflammatory and therapeutic effects of anti-TNF in CIA (16) and in RA (25) have been extensively described.

IL-10 was not detected in culture supernatants of control or salbutamol-treated peritoneal macrophages, but this is probably due to the fact that the cells were costimulated with IFN-γ, a potent inhibitor of IL-10 expression (26). There is a cAMP-responsive element in the mouse IL-10 promoter (27) and β₂-agonists have been reported to enhance IL-10 production by murine macrophages (8). This property of salbutamol might further contribute to its antiinflammatory and joint-protecting actions. IL-10 has been postulated to be part of a negative feedback that controls inflammation (28) and has been shown to have a therapeutic effect in CIA (29). We could not demonstrate salbutamol-induced stimulation of IL-10 production by arthritic synovial cells in vitro, but this might be explained by the fact that the cells were activated in vivo, and the salbutamol was subsequently added upon culture in vitro. It has been described that the IL-10-increasing effect of β-adrenergic agonists on LPS-stimulated macrophages was observed only when the cells were pretreated before stimulation with the β-adrenergic agonists (8).

β₂-Adrenoreceptors are widely expressed by different cells in the organism; hence, it is likely that multiple mechanisms operate to result in the rapid and sustained therapeutic response to salbutamol observed in this model. We opted for looking at prevention of mast cell degranulation as a marker for salbutamol function in vivo. Mast cells are a relatively abundant cell type in the synovial infiltrate in CIA and in RA (30, 31, 32); therefore, we studied whether salbutamol might stabilize those cells locally in the arthritic joints. Mast cell degranulation might contribute to the acute inflammatory phase of arthritis by enhancing vascular permeability (33). We found an inhibition of mast cell degranulation in the salbutamol-treated mice, which could account for the rapid clinical response we observed (reduction of paw edema and inflammation). A role for mast cells in RA has been proposed by the work of Woolley and colleagues, who showed that mast cells contribute to inflammation and joint erosion (34, 35). Studies in animal models have also implicated mast cell degranulation in joint erosion (32, 36, 37). This may thus explain why daily administration of salbutamol, although not augmenting the clinical benefit, offered a better protection against joint destruction than alternate day treatment. Moreover, injecting salbutamol just twice at the onset of disease prevents spreading of arthritis but lacks protection against joint damage in the already affected paw. Thus, it seems that an early blunting of the Th1 response, coupled to TNF blockade, controls clinical disease (inflammation, spreading of the arthritis to other limbs), but a continued daily administration of salbutamol is required to keep joint erosions under control.

It has been described previously that catecholamines and salbutamol exacerbate adjuvant arthritis in the rat via β₂-adrenoreceptors (38, 39). These results are in apparent contrast to our findings, but this is probably due to the fact that in these studies the compounds were administered in continuous infusion, which is known to desensitize β₂-adrenoreceptors (40, 41). In 1980, it was described that the combination of salbutamol and aminophylline, a weak PDE inhibitor, can prevent adjuvant arthritis (42). PDE inhibitors also block the production of TNF (6) and IL-12 (4), and we have recently described that rolipram, a specific PDE-IV inhibitor, works via these pathways to suppress CIA (43).

In conclusion, salbutamol has a profound therapeutic effect on CIA through its antiinflammatory actions and through a specific effect on the Th1 response. In this murine system, we did not observe any adverse effects of the drug. Extending the use of salbutamol to the human situation, however, will depend on the dosage. If it turns out that equally high doses are needed in humans, then the possible cardiotoxic side effects might create problems. These problems could possibly be overcome by combining salbutamol with PDE-IV inhibitors, both in lower doses, given that both reagents are likely to synergize in elevating cAMP.

Our study also indicates that the in vitro findings that salbutamol specifically inhibits the Th1 pathway (3) are relevant in vivo and therefore that the consequences of its chronic use for the treatment of asthma should be seriously investigated. Evidence that long term use of salbutamol in the end may have a detrimental effect on asthma has been reported (44).

The study has shown that cAMP-elevating agents are effective in CIA and confirms the results of a previous study that demonstrated amelioration of arthritis by rolipram (43). Taken together, these findings suggest that cAMP elevating agents may be effective in RA.

We thank Paul Warden and his team in our Biological Services Unit, for their valuable assistance, and Richard Williams, for helpful suggestions during the performance of this study.