Intra-articular administration of IκBα kinase inhibitor suppresses mouse knee osteoarthritis via downregulation of the NF-κB/HIF-2α axis

Activation of NF-κB signaling promotes osteoarthritis (OA) through the transcriptional induction of Hif-2α and catabolic enzymes. This study sought to examine whether inhibiting IκBα kinase (IKK) could suppress the development of surgically-induced OA of the knee in a mouse model. We employed BMS-345541 (4(2′-aminoethyl) amino-1, 8-dimethylimidazo (1,2-a) quinoxaline) as a selective inhibitor of the subunits of IKK. OA was created by resecting the medial collateral ligament and the medial meniscus in the knees of mice. The mice were then treated with an intra-articular injection of BMS-345541 (50 nM to 500 µM) or vehicle three times a week for 8 weeks. We found that the intra-articular administration of 500 nM and 5 µM BMS-345541 significantly suppressed OA development. In the BMS-345541-treated cartilage, there was a decrease in the phosphorylation of IκBα and the expression of Hif-2α, Mmp13, and Adamts5. In human articular chondrocytes, the IL-1β-enhanced expression of Hif-2α and catabolic factors were decreased by BMS-345541 treatment in dose-dependent manner. We conclude that the intra-articular administration of BMS-345541 at some concentrations may suppress the development of OA by downregulating signaling through the NF-κB–Hif-2α axis.

Osteoarthritis (OA) is the most common joint disorder in older people. Although joint pain can be partially controlled, the development of OA cannot be prevented, and degenerated articular cartilage cannot be regenerated by current therapeutics.
Since the development of surgically-induced mouse models of OA, various molecules and signaling pathways have been shown to be involved in its pathophysiology 1,2 . Among them, hypoxia-inducible factor-2a (Hif-2α) plays essential roles in OA development through the transcriptional induction of matrix metallopeptidase 13 (Mmp13), a representative cartilage-degrading enzyme 3,4 . Recently, we revealed that the activity of the nuclear factor kappa-B (NF-κB) signaling pathway is also tightly associated with the development of OA, along with other factors upstream of Hif-2α 5 .
The NF-κB family includes v-rel reticuloendotheliosis viral oncogene homolog A (Rela, also known as p65), Relb, Rel, p105/p50 and p100/p52. Inhibitors of NF-κB (IκB) proteins bind to some NF-κB family members in the cytoplasm 6 . Following the activation of IκB kinases (IKKs) in response to several signals, IκB proteins are phosphorylated and degraded. This frees NF-κB complexes and permits their translocation from the cytoplasm into the nucleus, leading to target gene transactivation 7,8 . In articular cartilage, IκBα is phosphorylated and Rela is translocated into the nucleus during OA development 5 . Intriguingly, studies have shown that homozygous knockout of Rela in adult chondrocytes accelerates OA due to enhanced chondrocyte apoptosis, whereas heterozygous knockout suppresses OA development through a downregulation of Hif-2α and Mmp13 5  mouse primary chondrocytes have confirmed that Rela is indispensable for chondrocyte survival via the transcriptional induction of anti-apoptotic genes, and that activation of NF-κB signaling leads to induction of Hif-2α and catabolic factors 5 . Thus, NF-κB signaling must be tightly regulated for articular cartilage homeostasis and the prevention of OA.
Here, we employed BMS-345541 (4(2′-aminoethyl) amino-1, 8-dimethylimidazo (1,2-a) quinoxaline) as a selective inhibitor of the subunits of IKK 9 , and examined whether inhibiting IKK could suppress OA development in vivo. We performed intra-articular injections of BMS-345541 for 8 weeks into the knee joints of mice surgically treated to induce OA, and then histologically analysed the development of OA and changes in the expression of OA-related proteins. We further examined the suppressive effects of the NF-κB-Hif-2α axis by BMS-345541 in human primary articular chondrocytes.

Suppression of OA development by intra-articular injection of BMS-345541.
We first examined the effects of BMS-345541 in a surgically-induced mouse OA model. Periodic injections of BMS-345541 or HBC (vehicle control) were administered into the knee joints of mice three times a week for 8 weeks after surgery (Fig. 1a). The OARSI scores of the 500 nM and 5 µM BMS-345541 groups were lower in the first examination (Fig. 1b,c), so we additionally replicated treatment with intra-articular injections of vehicle, 500 nM and 5 µM BMS-345541 (Fig. 2a). Histological analyses and OARSI grading at 8 weeks after surgery revealed that the joints treated with 500 nM and 5 µM of BMS-345541 showed significantly less OA as compared with the vehicle control group (Fig. 2a), and the responsiveness was excellent (effect size, 1.509, 1.841; 95% confidence interval, 0.876-2.143 and 1.278-2.403 for 500 nM and 5 µM BMS-345541, respectively). The OARSI scores decreased in the animals in a dose-dependent manner up to 5 µM; however, its protective effect was not observed over 50 µM (Fig. 1b,c). Over the 8 weeks, we found no obvious alteration in body weight among the six groups (Supplementary Table S1).
Suppression of the NF-κB-Hif-2α pathway by BMS-345541 treatment. We next analysed the expression of NF-κB-related proteins and catabolic enzymes by immunohistochemistry. At 8 weeks, in addition to Hif-2α, the expression of Mmp13 and Adamts5 was decreased significantly in BMS-345541-treated cartilage (Fig. 2b,c). To examine the activity of NF-κB at an early stage of OA, we additionally prepared 4 mice per group, and performed histological analyses after 2 -weeks of BMS-345541 treatment. Although OA development was mildly observed at 2 weeks after surgery, there was no significant difference between any of the groups ( Supplementary Fig. S1). Phosphorylation of IκBα was markedly suppressed in the BMS-345541-treated cartilage, as was the expression of Hif-2α at 2 weeks (Fig. 3a,b).

Effects of BMS-345541 in human primary articular chondrocytes.
To examine the molecular mechanisms underlying the cartilage protective effect of BMS-345541, we performed in vitro experiments using human articular chondrocytes derived from surgical specimens. We first examined the effects of BMS-345541 on cell proliferation and apoptosis. Cell proliferation was not affected by BMS-345541 concentrations of 500 nM or lower, but was moderately decreased by 5 µM BMS-345541, and decreased by concentrations over 50 µM (Fig. 4a). Most chondrocytes died upon treatment with 50 µM or 500 µM BMS-345541 until 48 h ( Supplementary Fig. S2), and TUNEL staining indicated significantly enhanced apoptosis upon treatment with 50 µM or 500 µM BMS-345541 (Fig. 4b,c).
To examine whether BMS-345541 exerts a suppressive effect against the NF-κB-Hif-2α pathway in humans as well as in mice, the cells were treated with IL-1β with or without BMS-345541. We found that the induction of catabolic enzymes including MMP13 and ADAMTS5 by IL-1β was could be suppressed by BMS-345541 treatment in a dose-dependent manner, when administered simultaneously or at 3 hours after IL-1β stimulation ( Fig. 5a,b, Supplementary Fig. S3). Alterations to the expression of COL2A1, a representative cartilage matrix protein, were not consistent (Fig. 5a,b, Supplementary Fig. S3). The phosphorylation of IκBα and Rela, and the increase in HIF-2α expression in IL-1β-treated cells could be inhibited by 5 µM BMS-345541 (Fig. 5b). Finally, using the NF-κB reporter luciferase construct, we confirmed that BMS-345541 suppressed the activation of NF-κB by IL-1β in a dose-dependent manner (Fig. 5c).

Discussion
Here, we showed that the intra-articular administration of an IKK inhibitor could regulate the development of surgically-induced OA of the knee in mice. Histological analyses indicated that the activity of NF-κB was decreased in articular cartilage treated with the IKK inhibitor before OA development, and that the subsequent induction of catabolic factors was also decreased. In addition, we showed that similar chondro-protective effects were observed in human primary articular chondrocytes treated with the IKK inhibitor. OA was most effectively  9 . It has been widely used as a highly selective inhibitor of IKK in many experimental studies. McIntyre and colleagues previously reported that the oral administration of BMS-345541 is efficacious against collagen-induced arthritis (CIA) in mice 10 . Meanwhile, 15 years has passed since the first publication with this compound, and there is almost no possibility for its clinical applicability. In the present study, we selected the compound among several IKK inhibitors to examine whether OA could be suppressed by IKK inhibition, because the selectivity and efficacy of the compound have been confirmed by many previous studies [9][10][11][12][13][14][15][16] . Although the compound will not be used clinically, the present data indicates that IKK may be a potent therapeutic target to suppress OA via inhibition of the IKK-NF-κB-HIF-2α axis.
Although IKK inhibition may be useful for the prevention or treatment of various diseases, including OA, the side effects of such inhibition should be carefully examined for its clinical utility, because the IKK-NF-κB pathway is an important pathway in numerous cellular processes, such as cell adhesion, differentiation, proliferation, autophagy, senescence, and apoptosis 17 . According to a previous study 10 , mice treated with BMS-345541 perorally daily for 6 weeks showed no toxicological changes in their vital organs on gross or histological observation, even at higher doses of 100 mg/kg/day, which is about 10 6 times the amount of BMS-345541 used in the present study. However, Yang et al. showed a 14% loss of body weight in melanoma-inoculated nude mice receiving high doses of oral BMS-345541 (75 mg/kg/day) 13 . In the present study, we could suppress OA development using a localized delivery of a much lower amount of BMS-345541 without causing body weight changes or obvious health impairment. The IKK inhibitor may be better suited to treating OA rather than systemic inflammatory joint diseases, because OA usually occurs in selective joints that can be independently targeted. However, in addition to efficacy, safety should be carefully examined when we target the IKK-NF-κB pathway through a novel approach.
In conclusion, the intra-articular administration of BMS-345541 at some concentrations may suppress the development of surgically-induced mouse knee OA through downregulation of the NF-κB-HIF-2α signaling axis. This inhibitory effect is also replicated in human cells in vitro. These results indicate the potential beneficial effects of IKK inhibitors for the prevention of OA development, assuming that treatment regimens adhere to suitable lower doses and localized modes of delivery.  (Fig. 1a). OA severity was assessed at 8 weeks after surgery. The Osteoarthritis Research Society International (OARSI) scores of the 500 nM and 5 µM BMS-345541 groups were lower in the first examination, so we replicated treatment with intra-articular injections of vehicle, 500 nM and 5 µM BMS-345541 using 7 mice per group for 8 weeks. The number of the mice for the OA analyses decreased for several reasons that were independent of BMS-345541 treatment: water leakage, severe injury due to fighting, and failures during tissue preparations. Finally, the number of mice evaluated was 5, 4, 4, 4, 6, 5 (1st) and 7, 0, 5, 7, 0, 0 (2nd) for vehicle, 50 nM, 500 nM, 5 µM, 50 µM, and 500 µM, respectively. To examine histological changes at the early stage of treatment, we also performed OA surgery in 24 mice, divided them into 6 groups, and analysed them after 2-week treatment.
Histological analyses. Knees were fixed in 4% paraformaldehyde prepared in phosphate-buffered saline (PBS), decalcified in ethylenediaminetetraacetic acid (EDTA), and embedded in paraffin. To quantify the severity of knee OA, all samples were quantified using the OARSI scoring system (Supplementary Table S2) 18    Statistical analyses. For multiple comparisons, statistical significance for differences between groups was determined with a one-way analysis of variance using JMP 13 (SAS Institute, Cary, NC, USA). In one-way ANOVA, the assumptions of the analysis were assessed by the Shapiro-Wilk test of normality and Levene's test for homogeneity of variance. If the result was significant, statistical analysis was performed with the Kruskal-Wallis and Dunn's post hoc test for unequal variance. ANOVA followed by Dunnett's post hoc test was employed if there was no significant result. For the results presented in all figures, each group was compared with the control. P values less than 0.05 were considered significant.