Attenuation of arthritis in rodents by a novel orally-available inhibitor of sphingosine kinase

Abstract Pro-inflammatory cytokines like TNF-a acti- vate sphingosine kinase (SK). Therefore, inhibition of SK is a potential molecular target for the treatment of rheu- matoid arthritis.

Aims The primary goal of this study was to assess the efficacy of ABC249640 (a selective SK-2 inhibitor) in two models of rodent arthritis. A secondary goal was to eval- uate the pharmacological profile of ABC294640, when given in combination with methotrexate.

Methods The efficacy of ABC294640 was determined by paw diameter/volume measurements, histological evalua- tions, and micro-CT analyses.

Results ABC294640 attenuated both collagen-induced arthritis in mice, as well as adjuvant-induced arthritis in rats. With the adjuvant arthritis model, the prophylactic efficacy profile of ABC294640 was similar to indometha- cin. Of note, ABC294640 reduced the bone and cartilage degradation, associated with adjuvant-induced arthritis. Rats treated with a suboptimal dose of MTX (50 lg/kg/day) in combination with ABC249640 (100 mg/kg/day) had better anti-arthritis effects in the adjuvant model, than treatment with either agent alone.

Conclusion Our results suggest that ABC249640 is an orally available drug candidate with a good pre-clinical efficacy profile for the prevention and/or treatment of RA.

Keywords : Sphingosine kinase · Arthritis · Rodents · ABC-294640

Introduction

Rheumatoid arthritis (RA) is a chronic, systemic disease that is characterized by synovial hyperplasia, massive cellular infiltration, erosion of the cartilage and bone, and an abnormal immune response (Anthony and Haqqi 1999). Studies on the etiology and therapy of rheumatoid arthritis have been greatly facilitated by the development of animal models that mimic the clinical and immunopathological disorders seen in humans (Anthony and Haqqi 1999; Luross et al. 2002; Oliver and Brain 1996; Matsukawa and Yoshinaga 1998; Goldring 1999; Brand 2005). From studies in these models, it is clear that the full manifesta- tions of RA are dependent on synergy between the humoral and cellular immune responses (Anthony and Haqqi 1999; Oliver and Brain 1996; Brand 2005). Two commonly used animal models for identifying potential therapeutic agents for RA are the rat adjuvant-induced arthritis (AIA) model, as well as the mouse collagen-induced arthritis (CIA) model (Schopf et al. 2006; Hegen et al. 2008). Neither of these models fully mimics either the initiating events or chronicity of human RA. However, there are distinct features of each model (Hegen et al. 2008; Holmdahl et al. 2002). Importantly, other investigators have emphasized the relevance of examining potential therapeutic agents in both the AIA and CIA models of RA (Hegen et al. 2008). Therefore, we utilized this pharmacological rationale in our studies (see below).

The rat AIA model is T lymphocyte and neutrophil dependent, with no documented role for B lymphocytes (Hegen et al. 2008). In contrast, both T and B cells play a central role in the pathogenesis of murine CIA. The major role of B-lymphocytes is the production of arthritogenic anti-collagen type II antibodies (Holmdahl et al. 2002). This is an attractive and clinically relevant feature of the CIA model of RA. As typically employed, the murine CIA model has a slower onset and a more prolonged duration than the rat AIA model (Hegen et al. 2008).

Of relevance to this study, both the AIA and CIA models are responsive to anti-tumor necrosis factor (TNF-a) agents, and involve the influx of various inflam- matory cells (macrophages, and neutrophils; Hegen et al. 2008; Holmdahl et al. 2002; Griffiths et al. 1995). The notion that immune cells, especially neutrophils, and cytokines play critical roles in the pathogenesis of arthritis is well established (Matsukawa and Yoshinaga 1998; Goldring 1999). However, the mechanisms by which this occurs are not fully elucidated.

Sphingosine kinases (SKs) catalyze the production of Sphingosine-1-phospate (S1P) in cells (Olivera and Spiegel 2001). Activation of SK is critical for intracellular signal- ing responses. Specifically, the ability of TNF-a to induce adhesion molecule expression in human umbilical vein endothelial cells via activation of NFjB is mimicked by S1P and is blocked by the SK inhibitor dimethylsphingo- sine (Xia et al. 1998). Similarly, S1P mimics the ability of TNF-a to induce the expression of COX-2 and the syn- thesis of PGE2 in fibroblasts. Moreover, knock-down of SK by RNA interference blocks these responses to TNF-a but not S1P (Pettus et al. 2003). S1P is also a mediator of Ca2? influx during neutrophil activation by TNF-a and other stimuli, leading to the production of superoxide and other toxic radicals (MacKinnon et al. 2002). These results suggest that SK could play a critical role in mediating the pathogenesis of various inflammatory diseases such as RA (Kee et al. 2005; Taha et al. 2006).

Recently, S1P levels in synovial fluid were shown to be significantly higher in patients with rheumatoid arthritis as opposed to osteoarthritis (Lai et al. 2008). These same investigators showed that i.p. administration of dimeth- ylsphingosine (DMS) significantly reduced disease severity associated with murine collagen-induced colitis (Lai et al.2008). A similar reduction in disease severity was found in mice treated with a sphingosine kinase knock-down approach via small interfering RNA (siRNA; Lai et al. 2008). These data provided a possible proof of concept for testing new SK inhibitors in models of rheumatoid arthritis. We have recently identified non-lipid small molecule inhibitors of SK activity, and demonstrated that these compounds have in vivo anticancer activity (French et al. 2003, 2006). The current lead SK inhibitor is 3-(4-chlorophenyl)-adamantane-1-carboxylic acid (pyri- din-4-ylmethyl) amide (ABC 294640), which had an in vitro IC50 of approximately 6 lM for inhibiting S1P for- mation via the SK pathway (Smith et al. 2008). Our latest data suggest that this compound is a SK-2 selective inhibitor with a Ki of approximately 30 lM (French et al. 2010). Moreover, we have shown that ABC294640 is an effective anti-inflammatory compound that attenuates the intestinal inflammation associated with experimental IBD in mice (Maines et al. 2008, 2010). Overall, ABC294640 and its congeners provide novel drug candidates for dis- eases that involve excessive activity of pathways activated by inflammatory cytokines. Such, is the case in RA. Therefore, the primary goal of this study was to assess the efficacy of ABC249640 in two well-established models of rodent arthritis. Both prophylactic and therapeutic para- digms were used to assess the efficacy of ABC294640 in these arthritis models. In addition, we evaluated the pharmacological profile of ABC294640, when given in combination with methotrexate (MTX). MTX is exten- sively used for the treatment of RA in humans (Magari et al. 2003). Specifically, MTX has become the gold- standard of disease-modifying anti-rheumatic drugs (DMARDs), due to its favorable efficacy and safety profile, as well as its ability to be used in combination with other DMARDs (Feely et al. 2009). Some of the proposed pharmacological actions of MTX (e.g., increasing endog- enous adenosine concentrations) are different than those of ABC294640 (SK2 inhibition; French et al. 2010; Wessels et al. 2008). Therefore, a clear pharmacological rationale existed for performing the combination drug treatment study in the rat AIA model.

Materials and methods

Materials

The methods for the synthesis of ABC294640, as well as the chemical structure of this sphingosine kinase inhibitor, have been previously reported (Maines et al. 2008). Indometha- cin was obtained from Sigma Chemical Co. (St. Louis, MO). Type II chicken collagen was obtained from Chon- drex (Redmond, WA). Incomplete and Complete Freund’s adjuvant, as well as heat-killed Mycobacterium butyricum, was purchased from Difco Microbiology (Lawrence, KS). Mineral oil was obtained from Sigma.

Animals

All rodents used in these arthritis studies were housed in appropriate cages with a 12-h light dark cycle. The arthritis protocols were approved by the IACUC at the Penn State College of Medicine.

Methods

Murine collagen-induced arthritis model

Female DBA/1 mice were injected subcutaneously in the tail with chicken immunization-grade type II collagen emulsified in complete Freund’s adjuvant. Twenty-one days later, the mice received a collagen booster in incomplete Freund’s adjuvant and were monitored daily thereafter for arthritic symptoms. Once mice showed signs of arthritis, as determined by their increased paw thickness and clinical score (study day 31), they were randomized into the following treatment groups: ABC294640 (100 mg/ kg po, 69 week) [n = 7], or vehicle (PEG 400) [n = 8]. Treatment began on study day 31 (designated as dosing day 1). Mice were dosed from day 1 to day 11 and then euthanized on day 12 (study day 42). Body weights were measured routinely during the 11-day dosing period. The severity of disease in each animal was quantified by mea- surement of the hind paw diameter with a digital caliper. Additionally, each paw was scored for inflammation, in which 0 = normal; 1 = mild, but definite redness and swelling of the ankle or wrist, or apparent redness and swelling limited to individual digits, regardless of the number of affected digits; 2 = moderate redness and swelling of the ankle and wrist; and 3 = severe redness and swelling of the entire paw including digits (maximum score = 12). After anaesthesia, the mice were euthanized and their hind limbs were removed, stripped of skin and muscle, formalin-fixed, decalcified, and paraffin-embed- ded. The limbs were then sectioned and stained with hematoxylin/eosin. Tibiotarsal joints were evaluated his- tologically for severity of inflammation and synovial hyperplasia.

Rat adjuvant-induced arthritis model

Age- and weight-matched female Lewis rats were used in this study. Briefly, rats were randomized to receive either mineral oil (non-adjuvant control) or complete Freund’s adjuvant containing 0.3 mg of heat-killed Mycobacterium butyricum suspended in 0.1 ml of light mineral oil. All intradermal injections were given into the right plantar area of ketamine/xylazine anaesthetized animals. Following recovery from anesthesia, rats were returned to their cages and provided standard laboratory chow and water ad libi- tum. Bedding material was placed in the cages to limit pain due to the paw swelling that resulted from adjuvant injections.

Two treatment paradigms were utilized with this arthritis model: (1) A prophylactic treatment paradigm, in which dosing was started before the overt appearance (typically on study days 11–14) of arthritis in the left (contralateral) limb. (2) A therapeutic paradigm, in which dosing was started after the overt appearance of arthritis in the contralateral limb (study day 14).

For the prophylactic paradigm, on study day 7, 15 rats were randomized to receive the following treatments by oro-gastric gavage: (1) Sham/mineral oil (n = 3), (2)
Sham/adjuvant (n = 3), (3) ABC294640 50 mg/kg/adju- vant (n = 6) and (4) Indomethacin-1.5 mg/kg/adjuvant (n = 3). Dosing was twice daily (9 a.m. and 5 p.m.), so that rats received 100 mg/kg/day of ABC294640 and 3 mg/kg/ day of indomethacin. ABC294640 was dosed in gelatin capsules for this study, due to an improved pharmacoki- netic profile in this dose form (unpublished observations). Dosing was from study day 7 to day 20, for a total of 28 doses. Subsequently, on study day 21, the rats were euthanized by inhalation of CO2. The left hind limbs were removed for the assessment of damage to the tibiotarsal (ankle) joints by micro-CT scans and histology (see below).

For the therapeutic paradigm, on study day 14, 14 rats were randomized to receive the following treatments by oro-gastric gavage: (1) Sham/mineral oil (n = 3), (2)
Sham/adjuvant (n = 3), (3) ABC294640 50 mg/kg/adju- vant (n = 5) and (4) Indomethacin-1.5 mg/kg/adjuvant (n = 3). Dosing was twice daily (9 a.m. and 5 p.m.), so that rats received 100 mg/kg/day of ABC294640 and 3 mg/kg/ day of indomethacin. ABC294640 was also dosed in gel- atin capsules for this study. Dosing was from study day 14 to day 20, for a total of 14 doses. Then rats were euthanized on day 21.
The total body weights, as well as paw severity scores, were monitored routinely throughout the study period. Each rear limb was scored for inflammation/swelling by the following scale: 0 = normal; 1 = mild, but definite redness and swelling of the ankle or wrist, or apparent redness and swelling limited to individual digits, regardless of the number of affected digits; 2 = moderate redness and swelling of the ankle and wrist; 3 = severe redness and swelling of the entire paw including digits; and 4 = max- imally inflamed limb with involvement of multiple joints. Rear paw volumes were measured on certain days (7, 14,17, 21) with a plethysmometer (UGO Basile-Stolting Co.). Both the limb injected with adjuvant (right hind paw and ankle joint) and the contra-lateral (left hind paw and ankle joint) were evaluated. On day 21, total paw scores were also determined using the aforementioned scoring system (maximal total paw score = 16). After euthanasia, the spleens were removed and weighed. Additionally, the stomach was also examined for the presence of macro- scopic gastric lesions (Pendley et al. 1993).

Rat adjuvant-induced arthritis model: drug combination study

In the drug combination study, we used a similar paradigm to that described above, with slight modifications. On study day 7, 15 rats were randomized to receive the following treatments by oro-gastric gavage: (1) Vehicle (PEG400)/ mineral oil (n = 6), (2) Sham/adjuvant (n = 8), (3) ABC294640 50 mg/kg/adjuvant (n = 10), (4) Methotrex-
ate (MTX) 50 lg/kg/adjuvant (n = 10), (5) ABC294640 50 mg/kg/adjuvant ? Methotrexate (MTX) 50 lg/kg (n = 8), (6) MTX 100 lg/kg/adjuvant (n = 5). Dosing was once daily for methotrexate, and twice daily (9 a.m. and 5 p.m.) for ABC249640, except for day 7 (one dose only). The overall dosing period was from study day 7 to day 21. Left paw volume measurements were as described above on study days 7, 14, 17, and 22. Rats were euthanized on day 22.

Micro CT analyses: rat ankle joints

After euthanasia, the left hind limbs were severed proximal to the femoral condyles, formalin fixed, and then secured in full extension in a polypropylene tube filled with 70% EtOH. The principal axis of the limb was centered along the Z axis of the microCT field. The entire specimen distal to the tibio-fibular junction was scanned using a vivaCT 40 (Scanco Medical AG, Bru¨ttisellen, Switzerland). Data were acquired as 1,000 projections at energy of 55 keV and a current of 145 lA with an integration time of 225 ms. Image slices were reconstructed as 2 K 9 2 K arrays of 12.5 lm isotropic voxels. Bone was segmented from soft tissue using the Scanco IPL software set to thresh- old = 300, sigma = 0.8, and support = 1, as determined empirically from the control specimens. Total bone volume in mm3 was calculated for the entire paw beginning at the slice in which the calcaneus first appeared and proceeding to the distal end.

Histological analyses: rat ankle joints

Subsequently, the limbs were stripped of skin and muscle, formalin-fixed in 10% neutral buffered formalin, decalcified with 8% formic acid (Surgipath Decalcifier I, Surgipath Medical Industries, Inc., Richmond, Il), and bisected sag- ittally through the tibiotalar joint and mid-foot region. One- half of each specimen was processed and paraffin embed- ded, and two 6-micron-thick sections were stained with haematoxylin/eosin and with Safranin-O/fast green. Tibio- tarsal joints were evaluated histologically for severity of inflammation, boney changes, synovial hyperplasia, and changes in cartilage, according to criteria described previ- ously (Hegen et al. 2003). The histology score was based on five parameters, each scored on a scale of 0–5. The five parameters are shown here: (1) composite inflammation (average of cellular infiltrate, oedema and joint/tendon effusion); (2) composite bony change (average of periosteal new bone formation and osteolysis); (3) synovial alteration, hyperplasia; (4) pannus (fibrinocellular debris/granulation tissue within the joint space); (5) cartilage degeneration, fibrillation. Each parameter was scored by a trained his- tologist (EEF) on coded slides, using the following severity scale: 0 = within normal limits; 1 = minimal; 2 = mild; 3 = moderate; 4 = marked; 5 = severe. Therefore, the total histology severity score ranged from 0 to 25 (Hegen et al. 2003).

Statistics

All data are presented as the mean ± SEM. Statistical analyses were conducted with a Graph-Pad prism software program (San Diego, CA). For parametric data (e.g., paw volumes, body weight) an unpaired test analysis was used. For non-parametric data, (e.g., paw severity scores) the Mann–Whitney test was used in the analysis. For multiple group comparisons (combination drug study) we used One-Way ANOVA, and the Newman Keul’s test for determining differences between individual treatment groups. Differences between linear regression lines were analyzed by ANCOVA analyses. For these statistical analyses, p \ 0.05 was considered to be statistically significant.

Results

Effects of ABC 294640 on collagen-induced arthritis in mice

As illustrated in Fig. 1, treatment with this SK inhibitor dramatically slowed the development of paw inflammation and oedema, when measured as either the average clinical score (Fig. 1a) or the average hind paw diameter (Fig. 1b). Significant decreases began at days 4 to 6 of treatment, for both endpoints. By the end of the dosing period (designated as Day 12), ABC294640-treated animals had a 90% reduction in the increase in hind paw thickness, as well as a 67% reduction in clinical score, compared with vehicle- treated mice.

During the dosing period, mice in both the vehicle and ABC294640 groups lost weight. The body weight loss was similar in both groups of mice (&-1.8 g). The final body weights (mean ± SEM) were: 18.6 ± 0.3 g (vehicle treatment) and 18.2 ± 0.6 g (ABC294640 treatment).

By histological analysis, collagen-induced arthritis resulted in a severe phenotype compared with non-induced mice. This histological pattern manifested as severe inflammation and synovial cell infiltration, as well as sig- nificant bone resorption (Fig. 1c). Mice that had been treated with ABC294640 had significantly reduced histo- logical damage (Fig. 1d).

Effects of ABC 294640 on adjuvant-induced arthritis in rats

Prophylactic paradigm

The mean body weight in sham/adjuvant treated animals continued to decrease throughout the study period (day 7–21) of the prophylactic paradigm. In contrast, after a slight transient decrease in body weight during the first week, indomethacin-treated rats gained weight (p \ 0.05 vs. sham/adjuvant treatment). Animals treated with the SK inhibitor lost weight up to study day 17, but then showed a slight increase in body weight by day 21 (data not shown). As shown in Fig. 2a, the mean left paw volume in the sham/adjuvant group increased over the study period from approximately 1.5 ml on day 7 to 3.0 ml on day 21. In contrast, a significant attenuation in the mean left paw On the indicated day of treatment, the average clinical score (a) and the average hind paw diameter (b) were determined, as shown.

Fig. 1 Effects of ABC294640 on disease progression and histolog- ical pathology in the CIA model in mice. Female DBA/1 mice were injected with collagen, boosted after 3 weeks, and then monitored for symptoms of arthritis. Upon disease manifestation, groups of mice were treated for 12 days as follows: filled triangle ABC294640 (100 mg/kg given orally each day for 6 days per week, n = 7); or filled square vehicle (PEG400 given under the same schedule, n = 8).

Fig. 2 Effects of ABC294640 or indomethacin on contralateral (left) paw volume during adjuvant arthritis in rats (prophylactic dosing paradigm). On day 7, Female-Lewis rats were randomized into one of four groups: filled diamond Sham/mineral oil (n = 3); filled square Sham/adjuvant (n = 3); filled triangle ABC294640 (100 mg/kg/day)/ adjuvant (n = 6) or filled down pointing triangle Indomethacin (3 mg/kg/day)/Adjuvant (n = 4). a Rats were dosed orally with the drugs over a 14 day period and left paw volumes were measured by plesthymometry on the study days indicated. *p \ 0.05 versus Sham/ adjuvant group. b The graph illustrates changes in left paw volumes of adjuvant treated rats during the study period. Filled black square Sham/Adjuvant; filled gray square ABC294640 (100 mg/kg/day)/ Adjuvant or opened square Indomethacin (3 mg/kg/day)/Adjuvant.

The reduced paw volume was maintained throughout the remainder of the study. The efficacy profile of the SK- inhibitor was similar to that observed with indomethacin treatment. At the time of randomization (day 7) individual rats had slightly different paw volumes. Therefore, the contralateral paw volume data were also calculated as the change in volume during the study period (Fig. 2b). As shown in this figure, the observed paw volume increase (&1.25 ml) in sham-treated rats was significantly greater at both time intervals than the increase in both ABC- 294640 and indomethacin-treated animals. Similar to the paw volume data, left paw severity scores were signifi- cantly smaller (p \ 0.05 vs. sham/adjuvant) in drug-treated rats on day 21. The median left paw scores were 0 (Sham/ mineral oil), 3 (sham/adjuvant), 1 (ABC294640/adjuvant), and 0 (indomethacin/adjuvant).

The right (injected) paw volume in sham-treated rats slightly increased over the 14-day study period. In contrast, by day 21, there were significant reductions in the absolute paw volumes for both the indomethacin and ABC294640 pretreatment groups. The effect was more pronounced with indomethacin. When the data were plotted as the change in volume, a similar trend was observed (data not shown). However, the modest reduction in volume with the SK-inhibitor did not reach statistical significance, while a statistically relevant reduction (p \ 0.05 vs. sham) was found with indomethacin treatment. The median right paw scores were 0 (sham/mineral oil), 3 (sham/adjuvant), 3 (ABC294640/adjuvant), and 1 (indomethacin/adjuvant).

On study day 21, mild or moderate swelling was also noticed on the front limb(s) of many adjuvant-treated rats.Specifically, final total paw scores (mean ± SEM) were 0 ± 0 (sham/mineral oil), 9.3 ± 0.3 (sham/adjuvant), 5.8 ± 0.4 (ABC294640/adjuvant), and 4.0 ± 1.0 (indo- methacin/adjuvant). Total paw scores were significantly reduced (p \ 0.05 vs. sham/adjuvant) in both the ABC294640 and indomethacin-treated rats.

The dorsal micro-CT scan of the left ankle joint showed significant areas of bone erosion (boxed area) in the sham/ adjuvant-treated animal (Fig. 3). All segments of the ankle joint were affected in this rat, from the distal tibia-fibula region to the proximal portion of the metatarsals (panel a). In the sham-treated animal, marked degenerative changes were particularly apparent to the naviculus, cuboidal, and cuniform bones. In contrast, only limited focal areas of erosions (arrows) were found in the ankle joint of an ABC294640/adjuvant treated rat (panel b).

In order to better show any damage to the calcaneus, ventral views of the left ankle joint are also shown in Fig. 3g–l. Prominent bone erosions were evident on the calcaneus of a sham/adjuvant treated rat (panel g). The areas of erosion also affected the cuboidal and naviculus bones and extended to the metatarsals. In contrast, only modest erosions were seen in these areas (panel h), within the ankle joint of the ABC294640-treated rat. Micro-CT scans from one indomethacin-treated rat (panels c and i) are shown for the sake of comparison. Overall, this animal showed only very mild evidence of erosions (arrows) in ankle joint.

When the bone volume density was quantified, there was a significant reduction (p \ 0.05 vs. mineral oil) in the bone volume of sham/adjuvant-treated rats. This attenua- tion in bone volume density was nearly completely reversed by ABC249640 treatment (Fig. 4a). As shown in Fig. 4b, there was a significant (p = 0.0036) inverse association (r = -0.792) between the left paw volume and left ankle joint bone density measurements, with the pro- phylactic treatment paradigm.

Representative joint histology results are shown in Fig. 5. A sham/adjuvant-treated rat had clear evidence of substantial joint pathology (panels a and b). Specifically as shown in panel b, there was severe inflammation (arrows), synovial (Syn) hyperplasia, and pannus (P) formation. In addition, moderate cartilage degeneration (Car) and clear evidence of bone deformation were also apparent. The total joint histology score for this rat was 22. Pictures from the ankle joint of a rat that received ABC249640 (panels c and d) showed evidence of modest inflammatory infiltra- tion (arrow), moderate synovial hyperplasia (Syn), and mild cartilage (Car) degeneration (total joint histol- ogy score = 9). The tibiotarsal joint histology scores (mean ± SEM) for all the treatment groups were 0 ± 0 (sham/mineral oil), 22.0 ± 0 (sham/adjuvant), 10.7 ± 2. and ABC-294640 (100 mg/kg/day) treatment groups. b Linear regression analysis shows a significant (p = 0.0036) inverse associ- ation (r = -0.792) between paw volume and bone volume in the rat AIA model. MO denotes mineral-oil treated rats, Adjuvant denotes sham/adjuvant-treated animals, and ABC indicates ABC294640 (100 mg/kg/day)/adjuvant-treated rats (ABC294640/adjuvant), and 14.0 ± 1.5 (indomethacin/ adjuvant). The sham/adjuvant-treated rats had significantly increased (p \ 0.05) joint histology scores compared with all the other treatment groups.

Prominent splenomegaly (&a three-fold increase in spleen weight) was evident in adjuvant treated rats, as compared with animals treated with mineral oil. Specifically, the spleen weights (mg/gram body weight) were 1.8 ± 0.1 (sham/mineral/oil), 5.6 ± 0.2 (sham/adjuvant), 4.0 ± 0.4 (ABC294640/adjuvant), and 5.1 ± 0.4 (indomethacin/ adjuvant). Treatment with ABC294640 significantly (p \ 0.05 vs. sham/adjuvant) reduced this splenomegaly, while indomethacin did not. Gastric damage (small lesions) was evident in 67% of the indomethacin-treated rats. In contrast, gastric damage was less evident (16% incidence) or non-existent (0% incidence) in the stomachs of rodents undergoing a sham dosing procedure, or treated with ABC294640, respectively.

Therapeutic paradigm

The mean body weights (grams) on day 14 were 207 ± 1 (sham/mineral oil), 182 ± 1 (sham/adjuvant), 183 ± 1 (ABC294640/adjuvant), and 182 ± 3 (indomethacin/adju- vant). The final body weights on day 21 were 215 ± 0.1 (sham/mineral oil), 177 ± 4 (sham/adjuvant), 182 ± 3 (ABC294640/adjuvant), and 191 ± 4 (indomethacin/adju- vant). Therefore, the mean body weight was increased in the sham/mineral oil and indomethacin/adjuvant treatment group, but not in sham or ABC294640-treated adjuvant rats (data not shown).

The left paw volumes were reduced in both indometh- acin- and ABC294640-treated rats by day 21 (Fig. 6a). The attenuation in paw volume with this NSAID was more substantial than for treatment with the SK inhibitor. Moreover, when compared with the baseline value (day 14), there was a significant reduction (-1.32 ± 0.49 ml) in the mean paw volume with indomethacin treatment, but only a modest reduction (-0.38 ± 0.31 ml) in the paw volume with ABC294640 treatment (Fig. 6b).

The swelling (paw volume) in the injected right limb was also attenuated (&-1.0 ml) by treatment with either ABC294640 or indomethacin. However, statistically rele- vant effects were not generally found because of the variable response in sham-treated rats.
As was the case with the prophylactic paradigm, prominent splenomegaly was evident in adjuvant-treated rats (data not shown). Treatment with ABC294640 reduced this splenomegaly, while indomethacin did not. Gastric damage (small lesions) was evident in 33% of the indo- methacin treated rats, while damage was non-existent in the stomachs of other rats.

Combination drug treatment paradigm

When ABC249640 was administered as a solution, it was still quite effective for attenuating the adjuvant-induced increase in left paw volume between days 7 and 22 (Fig. 7a). MTX (50 lg/kg/day) moderately but effectively prevented this change in left paw volume. Interestingly, combined treatment with ABC249640 (100 mg/kg/day) and MTX (50 lg/kg/day) virtually completely prevented the increase in left paw volume (Fig. 7a). In this regard, the efficacy profile of the combination treatment group was similar to that of MTX (100 lg/kg/day), which was used as a positive control drug in this study (Magari et al. 2003).

Fig. 6 Effects of ABC294640 or indomethacin on contralateral (left) paw volume during adjuvant arthritis in rats (therapeutic dosing paradigm). On day 14, Female-Lewis rats with evidence of arthritis were randomized into one of four groups: filled square Sham/mineral oil (n = 3); filled triangle Sham/adjuvant (n = 3); filled down pointing triangle ABC294640 (100 mg/kg/day)/adjuvant(n = 5) or filled diamond Indomethacin (3 mg/kg/day)/Adjuvant (n = 3). a Rats were dosed orally with the drugs over a 7-day period, and left paw volumes were measured by plesthymometry on the study days indicated. b The graph illustrates the changes in left paw volumes of all adjuvant-treated groups of rats during the study period. *p \ 0.05 versus Sham/adjuvant group.

In order to further show that combination treatment enhanced the anti-arthritic effects compared with either treatment alone, the time course profiles of the left paw volumes were analyzed, as shown in Fig. 7b and c. As demonstrated in these figures, when the time courses for the development of arthritis in the left paw were compared, significant differences (p \ 0.05) in the regression lines were detected in the combination treatment group, as compared with either single drug treatment. This data show that the development of arthritis was significantly attenu- ated in adjuvant-treated rats that received treatment with both MTX and ABC294640.

Representative micro-CT scans from MTX 50 or 100 lg/kg/day-treated rats, as well as an animal treated with MTX 50 lg/kg/day plus ABC249640 100 mg/kg/day, are shown in Fig. 3. The dorsal micro-CT scans from the MTX-(alone or in combination) treated rats (panels d through f) showed relatively small focal erosions (arrows). In particular, the ankle joints from rats receiving either the combination therapy, or the higher dose of MTX (panels e and f, respectively), had only minor evidence of bone damage. A somewhat similar visual damage pattern was observed with the ventral CT scans (panels j through l). Moderate areas of erosions (arrows) were apparent in the joint of the MTX 50 lg/kg/day-treated rat (panel j). In comparison, only small focal erosions were found in the animal treated with MTX 50 lg/kg/day plus ABC249640 100 mg/kg/day (panel k), or the MTX 100 lg/kg/day-treated rat (panel l). Overall, these micro-CT results (Fig. 3) suggest the same type of pharmacological profile that was found for the paw volume data (Fig. 7).

As was the case with our other studies, prominent splenomegaly (&a 2.5 fold increase in spleen weight) was evident in adjuvant-treated rats (data not shown). Treat- ment with all the drug treatments (alone or in combination) significantly attenuated the splenomegaly. Visible gastric damage was not evident in this experiment.

Discussion

Previous studies have shown that oral administration of ABC294640 to rodents effectively attenuates the disease phenotype in experimental models of inflammatory bowel disease (Maines et al. 2008, 2010) or diabetes (Maines et al. 2006), as well as cancer (French et al. 2003, 2006, 2010). In the present studies, we have expanded these studies to include two models of RA that are commonly used to test drug efficacy. In the first case, collagen- induced arthritis in mice is an extensively studied murine model of arthritis because it emulates many immunological and pathological features of human RA. This arthritis is primarily an autoimmune disease of joints, which requires both T and B lymphocyte immunity to autologous type II collagen for disease manifestation (Kannan et al. 2005); and it is sensitive to agents that antagonize TNF-a (Saito et al. 2007).

Therefore, collagen-injected DBA/1 mice were dosed orally with vehicle or ABC294640 in a therapeutic-type paradigm. Specifically, treatments were begun during the early active phase of arthritis (average paw severity score = 3, Fig. 1). In vehicle-treated mice, the arthritis severity progressively increased throughout the study per- iod. However, treatment of the animals with ABC294640 effectively attenuated the progression of arthritis, as determined by both the clinical scores and more quantita- tive paw thickness measurements. The inhibitory profile of the SK inhibitor was similar for both of these parameters. These data extend the previous report of Lai et al. (2008), who showed that intraperitoneal injection of a SK inhibitor (DMS) attenuates inflammatory cytokine production and disease progression associated with CIA in mice.
To confirm these initial findings in another well-estab- lished rodent model of RA, we next evaluated the efficacy of ABC294640 in the rat AIA model. The time course of contralateral paw swelling in vehicle-treated rats (Fig. 2a) corresponds very well to the onset (Day 14) and plateau phase (Days 17–21) described in the literature (Schopf et al. 2006). With this model, we utilized both a prophy- lactic paradigm and a therapeutic paradigm to test the efficacy of ABC29460.

Nonsteroidal anti-inflammatory drugs (NSAIDs) are effective in relieving symptoms (e.g. pain, swelling) of RA, but are not disease modifying (Feely et al. 2009). Despite this limitation, a non-selective NSAID (indomethacin) is often used as a positive therapeutic control drug in the rat AIA model, because it effectively attenuates the paw swelling associated with adjuvant administration (Hamada et al. 2000; Ochi and Goto 2002). Therefore, in our AIA studies, we initially compared the efficacy of ABC294640 to indomethacin. Nevertheless, the use of indomethacin in experimental models of arthritis, or in patients with RA, can result in an increased incidence of gastric damage (see below; Kato et al. 2002).

Using a prophylactic treatment paradigm, the efficacy profile of the SK inhibitor for limiting the increase in both the contralteral (left) paw severity score and the left paw volume was similar to indomethacin (see Fig. 2). As shown in Fig. 3, when comparing the microCT scans of ABC249640- and sham-treated rats, there was an obvious and impressive drug-induced reduction in the observed areas of left ankle joint bone erosions. Upon quantification of this anti-erosion effect, the calculated bone volume density, which was significantly reduced in sham-treated animals, was nearly normalized in rats treated with ABC249640 (Fig. 4). Moreover, histological evaluations showed clear evidence of significantly less joint inflam- mation, synovial hyperplasia, pannus formation, cartilage destruction, and bone deformation in the tibiotarsal joints of rats treated with ABC249640, when compared with sham-treated animals (Fig. 5).
Interestingly, the right (injected) paw data also sug- gested that the SK inhibitor could also regress (to some degree) established inflammation in the injected paw. However, the observed volume reductions were not as prominent as those observed with indomethacin treatment. As a whole, these results suggested a very good anti- arthritis effect of ABC294640, when this SK inhibitor was administered in a prophylactic mode, as well as the pos- sibility that the compound could exhibit some efficacy in a more stringent therapeutic paradigm.

Therefore, in the next study, animals with well-estab- lished arthritis (left paw volumes of approximately 3 ml) were randomized to receive treatments over a 1-week period. As shown in Fig. 6, vehicle-treated animals expe- rienced a slight further increase in paw volume (&0.5 ml). In contrast, a modest reduction in left paw volume was observed in animals treated with the SK inhibitor. Indo- methacin was quite effective in this paradigm, since the left paw volume was significantly decreased by [1 ml.
In our studies, ABC294640 treatment had certain advantages, when compared with indomethacin. As expected from the relevant literature (Kato et al. 2002), indomethacin-treatment caused visible damage (small erosions) in the rat stomach, which was greatly reduced or non-existent in the other treatments groups. These data suggest a gastric safety advantage for ABC294640, as compared with this non-selective NSAID. Interestingly, this novel SK inhibitor also significantly attenuated splenomegaly, which is a salient feature of adjuvant arthritis (Gugasyan et al. 1997). In contrast, indomethacin treatment did not significantly alter the spleen weight in this study.

MTX has become the overall DMARD of choice for RA. Moreover, it is often used in combination treatment regimens to enhance the effects of other drugs (Feely et al. 2009). Of relevance, previous investigators found that MTX (100 lg/kg/day) was very effective for preventing paw swelling in adjuvant treated rats (Magari et al. 2003). Therefore, we thought it was important to evaluate the efficacy profiles of ABC294640 and MTX in the AIA model, when the SK inhibitor was administered either alone or in combination with this DMARD. Similar results to those of Magari and colleagues were found in this study (Fig. 7a). In their study, the calculated ED50 for preventing swelling was 55 lg/kg (Magari et al. 2003). Therefore, we also dosed a cohort of rats with a suboptimal dose of MTX (50 lg/kg/day) in combination with ABC249640 (100 mg/ kg/day). Although these agents alone significantly reduced the development of left paw swelling (Fig. 7a), the overall pharmacological effect was more pronounced when both drugs were administered to rats (Figs. 3, 7). The enhanced efficacy profile, which was observed with combination therapy, is consistent with the different pharmacological mechanisms of the two agents (French et al. 2010; Maines et al. 2008; Magari et al. 2003). The pharmacological profile for combination therapy with ABC294640 and MTX (Fig. 7) is more consistent with additive (as opposed to synergistic) effects in the rat AIA model.

The use of SK inhibition as a pharmacological approach for arthritis has only recently been addressed in the liter- ature. In one study, collagen-induced arthritis was not altered in SK-1 knockout mice (Michaud et al. 2006). In this regard, it should be noted that SK1-deleted transgenic mice have normal phenotypes, and serum concentrations of S1P are reduced by only 50% (Keohane et al. 2004). Additionally, levels of S1P within a variety of tissues are not different from those of control mice, indicating func- tional replacement of SK1 by SK2 in normal joints (Keohane et al. 2004). This is likely the reason that SK1 knockout mice demonstrate a normal inflammatory response during collagen-induced arthritis (Michaud et al. 2006). In contrast, other investigators showed a reduction in disease severity by administration of siRNA targeting SK1 in mice (Lai et al. 2008), suggesting that SK2 may not fully compensate for reduction of SK1 activity in adult joints. More recently, Lai et al. (2009) reported that administration of siRNA targeting SK2 potentiated the arthritis disease phenotype in mice. As in their prior study, siRNA targeting SK1 effectively reduced murine arthritis (Lai et al. 2008, 2009). These results again suggest the possibility that SK2 may not fully compensate for the reduction of SK1 activity in adult mice.

Within the context of the previous data (Keohane et al. 2004; Michaud et al. 2006; Lai et al. 2008, 2009), the current results with ABC-294640 (a selective SK-2 inhibitor) suggest that the roles of the individual SK isoforms in arthritis are still unresolved. Nevertheless, our data are important because they show the first anti- arthritis effects of a chemical SK-2 inhibitor in relevant rodent models of RA. Finally, it is also possible that ABC249640 may also more directly antagonize other inflammatory targets, in addition to inhibiting SK-2 (French et al. 2010). In support of this possibility, previ- ous results from our laboratory showed that ABC2494640 potently inhibits the in vitro activation of NF-jB by TNF- a (Maines et al. 2008). Inhibition of both of these important cellular targets could thereby contribute to the observed attenuation of disease pathogenesis in rodent models of RA.

In summary, the orally available small molecule inhib- itor of SK-2 (ABC294640) demonstrates efficacy in two well-established rodent arthritis models. With the rat adjuvant arthritis model, the efficacy profile of ABC- 294640, when administered in a prophylactic paradigm, was similar to that of indomethacin. ABC-294640 also showed some anti-inflammatory effects when administered in a therapeutic paradigm. Importantly, ABC-294640 had a better gastric safety profile in the adjuvant arthritis model than did indomethacin. Based on this pre-clinical efficacy profile, we conclude that this novel SK inhibitor could be useful as a therapeutic and/or maintenance therapy for human RA.