The cytosolic DNA sensors, IFI204 and STING, are upregulated in OA-like mouse chondrocytes.
We first screened the mRNA levels of cytosolic DNA sensors, cytosolic RNA sensors, and related molecules from microarray data obtained from OA-like chondrocytes. OA chondrocytes were mimicked by treating primary-culture mouse chondrocytes with the pro-inflammatory cytokine, IL-1β [25], or an adenovirus expressing a critical cellular mediator of OA pathogenesis, such as HIF-2α (Ad-HIF-2α) [7] or ZIP8 (Ad-ZIP8) [8]. Our microarray analysis revealed that the mRNA levels of interferon-inducible gene IFI204 (the murine ortholog of human IFI16) and stimulator of interferon genes (STING) were markedly increased in OA-like chondrocytes (Fig. 1A). (q)RT-PCR analysis further confirmed that IL-1β treatment or overexpression of HIF-2α or ZIP8 in chondrocytes significantly increased the mRNA levels of IFI204 and STING (Fig. 1B ~ E). STING is not only a cytosolic DNA sensor, but also an essential mediator of innate immune responses initiated by cytoplasmic DNA [9, 11, 26]. We, therefore, focused on the possible functions of upregulated STING in chondrocytes during OA pathogenesis and OA-associated pain behavior in mouse.
Similar to the results obtained when we examined the mRNA levels, IL-1β treatment or overexpression of HIF-2α or ZIP8 increased the protein levels of STING in primary-culture mouse chondrocytes (Fig. 2A). The protein levels of cGAS, which is an upstream molecule of STING [11], were also increased in these chondrocytes (Fig. 2A). Consistently, STING protein levels were markedly elevated in post-traumatic OA cartilage caused by DMM surgery in mouse (Fig. 2B). Similarly, compared to an undamaged region of the same cartilage tissue, a damaged part of human OA cartilage exhibited markedly elevated STING protein levels (Fig. 2C). These results suggest that activation of innate immune response by the cGAS-STING pathway in chondrocytes plays a role in OA pathogenesis.
The cGAS-STING pathway activates an innate immune response by sensing cytosolic DNA as a DAMP signal [10, 11]. Endogenous cytosolic DNA can come from damaged mitochondrial DNA or nuclear DNA that is leaked/damaged by chromosome instability and cell damage [11, 26]. We, therefore, evaluated DNA damage in OA-like chondrocytes and OA cartilage by detecting γH2AX, a marker for DNA damage [27]. The protein level of γH2AX was markedly increased in primary-culture mouse chondrocytes stimulated with IL-1β or overexpressing HIF-2α or ZIP8 (Fig. 3A). γH2AX was also found to be upregulated in DMM-operated post-traumatic OA cartilage of mouse (Fig. 3B) and damaged parts of human OA cartilage compared to the corresponding undamaged regions of the same cartilage tissues (Fig. 3C). These results suggest that catabolic factors such as IL-1β, HIF-2α, or ZIP8 induce DNA damage in OA chondrocytes, and activates cGAS-STING pathway during OA pathogenesis.
Genetic ablation of STING in mouse ( Sting1 −/− ) mitigates post-traumatic OA and pain behavior.
The role of STING in OA pathogenesis was directly examined by DMM surgery in WT and Sting1−/− mice. Sting1−/− mice are viable and exhibited normal development of skeletal elements (Supplementary Fig. 1). Compared with WT mice, Sting1−/− mice exhibited significant reduction in cartilage destruction at both 6 and 8 weeks after DMM surgery. For instance, At 6 weeks after DMM surgery, median OARSI grade in WT mice [3.29 (IQR 2.11-5.00)] was decreased to 1.28 (IQR 0.83–3.25, P = 0.0191) in Sting1−/− mice (Fig. 4A−D). Similarly, at 8 weeks after DMM surgery, 3.53 (IQR 3.08–3.94) median OARSI grade in WT mice was decreased to 2.11 (IQR 1.44–2.50, P = 0.0007) in Sting1−/− mice (Fig. 4A−D). Thickening of the subchondral bone plate (SBP), an indicator of subchondral bone sclerosis [8, 14, 22], was also inhibited in DMM-operated Sting1−/− mice. SBP thickness in WT mice were 99.47 ± 5.03 µm (95% CI [76.03-121.81]) and 108.26 ± 3.85 µm (95% CI [92.51-132.24]) at 6 and 8 weeks after DMM surgery. These values were significantly decreased to 74.90 ± 3.69 µm (95% CI [59.19-103.05], P = 0.0003) and 74.62 ± 2.94 µm (95% CI [63.87–91.19], P < 0.0001) in Sting1−/− mice at 6 or 8 weeks, respectively (Fig. 4A−D). However, DMM-induced osteophyte formation was not modulated by genetic ablation of STING at either time point (Fig. 4A−D).
Although the function of STING in OA-associated joint pain was previously unknown, STING was recently shown to regulate nociception via type-I interferon (IFN-I) signaling in peripheral sensory neurons [13]. We, therefore, examined whether STING regulates OA associated pain behavior in WT and Sting1−/− mice. Our von Frey assay in DMM-operated mice revealed that Sting1−/− mice were less sensitive to von Frey filaments before the initiation of the surgery and showed significant pain relief from 4 to 8 weeks after DMM surgery (Fig. 4E). These findings indicate that STING deficiency plays protective roles in DMM-induced post-traumatic OA and pain behavior in mice.
Stimulation Of The Sting Pathway Exacerbates Oa Pathogenesis And The Associated Pain Behavior
Next, we investigated whether stimulation of the STING signaling pathway in joint tissues modulates OA pathogenesis and pain behavior. For this purpose, cGAMP, a natural agonist of STING [17], was IA injected into the knee joints of mice with or without DMM surgery. IA injection of cGAMP alone did not cause any OA-like change in the joint tissues at 3 and 8 weeks post-IA injection (Fig. 5A; Supplementary Fig. 2A). However, IA injection of cGAMP in DMM-operated knee joints significantly exacerbated cartilage destruction. Compared to vehicle-treated group, IA injection of 10 or 20 µg of cGAMP increased median OARSI grade from 2.03 (IQR 1.09–2.71) to 2.73 (IQR 2.32–3.21, P = 0.0193) and 2.57 (IQR 2.32–3.32, P = 0.0430), respectively (Fig. 5B and C; Supplementary Fig. 2B). IA injection of cGAMP in DMM-operated mice also caused more significant SBP thickening. For instance, SPB thickness in PBS-injected mice was 75.95 ± 2.33 µm (95% CI [64.83–94.85]), whereas these value was increased to 103.93 ± 3.49 µm (95% CI [83.00-122.52], P < 0.0001) and 112.35 ± 2.03 µm (95% CI [99.37-122.36], P < 0.0001) by IA injection of 10 and 20 µg of cGAMP, respectively (Fig. 5B and C). Contrast to OARSI grade and SBP thickening, osteophyte formation was not modulated by IA injection of cGAMP in DMM-operated mice (Fig. 5B and C). Together with the data obtained from Sting1−/− mice, our results indicate that STING regulates OA cartilage destruction and subchondral bone sclerosis, but not osteophyte formation, in DMM-operated mice.
Consistent with the observation of pain relief in DMM-operated Sting1−/− mice, stimulation of the STING signaling pathway via IA injection of cGAMP in DMM-operated knee joints significantly exacerbated OA-associated pain behavior in mice (Fig. 5D). These results are consistent with the view that inhibition of STING yields a chondro-protective function and pain relief in DMM-operated mice (Fig. 4A−E).
STING regulates the expression of pain-sensitization molecules.
In an attempt to elucidate the regulatory mechanisms underlying the pain-relief impact of STING, we examined whether STING regulates the expression of molecules involved in pain sensitization, such as TRPV1 [28, 29], CGRP [30, 31], and NGF [32, 33]. These molecules play a key role in peripheral sensitization by activating and sensitizing nociceptors, which contribute to the development of OA-associated pain [34, 35]. We compared the expression levels of these molecules in the cartilage, meniscus, synovium, subchondral bone, and periosteum of DMM-operated Sting1−/− mice and WT. Immunofluorescence microscopy of knee joint sections revealed that the protein levels of TRPV1 were significantly reduced in the synovium and meniscus of Sting1−/− mouse knee joints (Fig. 6A and B; Supplementary Fig. 3). CGRP protein levels were also significantly decreased in the synovium and meniscus of Sting1−/− mice (Fig. 6C and D; Supplementary Fig. 3]. In contrast, the expression levels of these molecules were not modulated by STING deficiency in other examined tissues, such as cartilage, subchondral bone, and periosteum (Fig. 6A ~ D; Supplementary Fig. 3 and Fig. 4). Unlike the findings for TRPV1 and CGRP, the expression levels of NGF in the examined tissues were not modulated in DMM-operated Sting1−/− mice compared to WT (Fig. 6E and F; Supplementary Fig. 3 and Fig. 4]. These results suggest that STING may regulate OA-associated pain by decreasing the expression of peripheral sensitization molecules in the synovium and meniscus of mouse knee joints.