G protein-coupled receptor kinase 5 deletion suppresses synovial inflammation in a murine model of collagen antibody-induced arthritis

G protein-coupled receptor kinase 5 (GRK5) regulates inflammatory responses via the nuclear factor-kappa B (NF-κB) pathway. This study investigated the functional involvement of GRK5 in the pathogenesis of inflammatory arthritis. Immunohistochemically, rheumatoid arthritis (RA) synovium had a significantly higher proportion of GRK5-positive cells in the synovial lining layer than healthy control synovium. Gene expression and NF-κB activation in lipopolysaccharide-stimulated human SW982 synovial cells were significantly suppressed by silencing of the GRK5 gene. Similarly, GRK5 kinase activity inhibition in human primary RA synovial cells attenuated gene expressions of inflammatory factors. In a murine model of collagen antibody-induced arthritis, arthritis scores and serum IL6 production of GRK5 knockout (GRK5-/-) mice were significantly lower than those of wild-type mice. Histologically, the degree of synovitis and cartilage degeneration in GRK5-/- mice was significantly lower than in wild-type mice. In in vitro analyses using activated murine macrophages and fibroblast-like synoviocytes, gene expression of inflammatory factors and p65 nuclear translocation were significantly lower in GRK5-/- mice compared to wild-type mice. In conclusion, our results suggested that GRK5 is deeply involved in the pathogenesis of inflammatory arthritis, therefore, GRK5 inhibition could be a potential therapeutic target for types of inflammatory arthritis such as RA.

GRK5 knockdown with siRNA attenuates NF-κB transcriptional activation and the inflammatory gene expression in the human SW982 synovial cell line. The functional effects of GRK5 on NF-κB transcriptional activity and gene expressions were analysed in human SW982 cells transfected with small interfering RNA (siRNA) for GRK5. SW982, a human synovial cell line, is a useful tool for investigating the expression of inflammatory cytokines or MMPs in response to various stimuli 23 . GRK5 gene expression was successfully decreased with siRNA ( Fig. 2A). Lipopolysaccharide (LPS)-induced NF-κB transcriptional activity was significantly attenuated by GRK5 knockdown (Fig. 2B). GRK5 knockdown significantly attenuated LPS-induced gene expressions of IL6, MCP1, and MMP3 compared to siRNA for control (Fig. 2C).
GRK5 knockdown with siRNA and GRK5 inhibitor attenuates LPS-stimulated the inflammatory gene expression in human primary RA synovial cells. In human primary RA synovial cells, LPS, TNFα, IL6, and IL17 did not affect GRK5 gene expression (Fig. 2D). GRK5 knockdown with siRNA significantly attenuated LPS-induced gene expressions of GM-CSF compared to siRNA for control (Fig. 2E). Amlexanox, a selective GRK5 inhibitor 24 , was used to analyse the effect of GRK5 kinase activity inhibition. Amlexanox In a CAIA model, GRK5 knockout mice had less joint swelling, synovitis, and cartilage degradation. To investigate the in vivo effect of GRK5 deletion on inflammatory arthritis, we generated a CAIA model using GRK5 knockout and wild-type (WT) mice. There were no differences between GRK5 knockout and WT mice during development and growth (Fig. 3A). After LPS injection on day 3, the body weight of both GRK5 knockout and WT mice was slightly reduced by the same amount (Fig. 3B). Joint swelling in WT mice started on day 4 and increased over time. On day 10, the arthritis score of WT mice reached 13.8 ± 1.6 (out of 16) (Fig. 3C,D). In contrast, in GRK5 knockout mice, only minor joint swelling was observed on day 5. The arthritis score on day 10 was 3.5 ± 0.2, which was significantly lower than in WT mice (p < 0.001). Divided into front and hind paws, each arthritis score was also significantly lower in GRK5 knockout mice than in WT mice (Fig. 3E,F). The ankle joint of WT mice on day 10 showed moderate to severe synovitis, such as multilayered lining cell layers, high cellularity, and subsynovial lymphatic infiltration (Fig. 4A). GRK5 knockout mice had less severe synovitis histologically. GRK5 knockout mice had significantly lower average synovitis scores than WT mice on day 10 (p < 0.01) (Fig. 4B). In the cartilage of WT but not GRK5 knockout mice, significant loss of Safranin O staining was observed (Fig. 4C). Cartilage degeneration score was significantly lower in GRK5 knockout mice than in WT mice (p < 0.01) (Fig. 4D). In CAIA WT mice, GRK5 was positively immune-stained in the lining cells but not in the sublining cells (Fig. 4E).
GRK5 knockout suppresses serum IL6 production in CAIA mice and gene expressions of Il6, Tnfα, Il1β, Mip2, and factor B in macrophages. To explore the effect of GRK5 knockout on cytokine production in the initiation of CAIA, serum IL6 and TNFα concentration were examined by ELISA. At 2 day after LPS injection, serum IL6 concentration, but not TNFα (data not shown), significantly increased in WT and GRK5 knockout mice. The increase in serum IL6 concentration significantly suppressed in GRK5 knockout mice compared to WT mice (p < 0.001) (Fig. 5A).

Discussion
This study is the first to elucidate the characteristics of GRK5 expression in human synovium and the effect of GRK5 knockout on the development of arthritis in a murine model of CAIA. GRK5 was more highly expressed in the synovial lining layer of patients with RA than healthy controls. GRK5 knockdown in a human synovial cell line clearly attenuated LPS-induced expression of inflammatory genes and NF-κB transcriptional activity. Similarly, GRK5 kinase activity inhibition in human primary RA synovial cells suppressed LPS-induced inflammatory gene expressions. In a CAIA model, GRK5 knockout mice had significantly suppressed development of arthritis, with less severe synovitis and cartilage degeneration compared to WT mice. Suppressed serum IL6 production seemed as a causal phenomenon for less severity of CAIA in GRK5 knockout mice. Consistently, murine macrophages from GRK5 knockout mice had downregulated gene expression of Il6, Tnfα, Il1β, Mip2, and factor B. In FLSs from GRK5 knockout mice, downregulated inflammatory gene expression was accompanied by attenuation of p65 nuclear translocation. These results suggested that GRK5 plays a pathogenic role in the development of synovitis in inflammatory arthritis via NF-κB signalling.
In general, GRKs consist of seven isoforms that share structural and functional similarities in the phosphorylation and desensitisation of GPCRs 25 . GRK2, GRK3, GRK5, and GRK6 are ubiquitously found in cells throughout the body. In contrast, GRK1, GRK4, and GRK7 have limited tissue distribution 26,27 . Lack of GRK2 expression results in embryonic lethality due to cardiac abnormalities 28 , but mice with knockouts of other GRKs develop normally. www.nature.com/scientificreports/ Importantly, GRKs have been reported to have disease-specific functions by interacting with intracellular substrates other than GPCRs [29][30][31] . In particular, GRK5 has been reported to activate NF-κB signalling through IκBα phosphorylation and p65 nuclear translocation 12,20,32 . In macrophages, Patial et al. demonstrated that GRK5 is a positive regulator of the TLR4-induced IκBα-NFκB pathway as well as a key modulator of LPS-induced inflammatory responses 20 . In myocytes, overexpression of GRK5 increases the levels of NF-κB p50 and p65 in vitro and in vivo, whereas loss of GRK5 expression resulted in lower cardiac NF-κB levels 32 . In chondrocytes, GRK5  www.nature.com/scientificreports/ overexpression significantly increases NF-κB transcriptional activation in humans and GRK5 deletion reduces IκBα phosphorylation and p65 nuclear translocation in mice 12 . Consistent with these findings, the present study showed that active synovial lining cells from patients with RA expressed substantial amounts of GRK5 protein, and LPS-induced NF-κB transcriptional activity and p65 nuclear translocation in synovial cells were significantly attenuated by downregulation of GRK5.
In this study, the murine CAIA model was used as a model of arthritis because it has provided the most reproducible induction of synovitis and consequent cartilage degradation in the study of inflammatory arthritis [33][34][35] . The collagen-induced arthritis (CIA) model is another traditional murine model of arthritis, but there is a lower incidence of arthritis in C57BL/6 background mice. The in vivo results from this study demonstrated that severe arthritis can be successfully induced in a CAIA model, as determined by the macroscopic score and histological scores of synovium and cartilage in WT mice. Previously, Tarrant et al. examined the effect of GRKs knockout on the pathogenesis of arthritis in the K/BxN model in which a spontaneously erosive arthritis with similarities to RA occurs 36 . In their study, GRK6 and GRK2 knockout mice demonstrated severe arthritis and weight loss due to the loss of anti-chemotaxis property during the development of the K/BxN model, whereas there was no significant difference in the response to K/BxN serum-transfer between GRK5 knockout and WT mice. Accordingly, GRK5 knockout did not show any significant phenotype in the K/BxN arthritis model. Although there is no clear explanation for the discrepancy with our findings, the difference in the pathogenesis between the CAIA and the K/BxN model might be a potential influential factor. CAIA, but not K/BxN model, needs a LPS-injection to induce arthritis after sensitization by antibody. In our study, LPS significantly increased NF-κB transcriptional activation, therefore, GRK5 deletion contributed to suppress inflammatory responses via NF-κB signaling.
Primary activation of innate immune cells including macrophages plays an pivotal role in the development of CAIA based on functional complement system 37 . In the present study, GRK5 knockout markedly attenuated serum IL6 concentration after LPS injection at the initiation of CAIA. The in vitro results from this study confirmed that macrophages from GRK5 knockout mice showed a less LPS-response in gene expressions of inflammatory factors. In complement elements, not C4 but factor B gene expression was significantly attenuated in GRK5 knockout macrophages, indicating that alternative complement pathway, which is essential in the pathogenesis of CAIA 37 , was involved. Collectively, these findings suggested that GRK5 functions as a positive intermediator of LPS-stimulation in macrophages, therefore, GRK5 knockout significantly delays the onset of arthritis and suppresses the initial progression down to a mild level in a CAIA.
FLSs, also termed type B synoviocytes, are the most abundant resident cell type in synovial tissue 38 . Numerous lines of evidence support the potential contribution of FLSs to the pathogenesis of inflammatory arthritis 3,4,39,40 . Activated FLSs are responsible for excessive matrix degradation, which destroys cartilage and causes permanent joint damage in RA [41][42][43] . The in vitro results of this study showed that not only LPS-induced but also IL-1β and TNFα-induced inflammatory cytokines and chemokines are markedly attenuated in FLSs from GRK5 knockout www.nature.com/scientificreports/ mice, a possible mechanism by which GRK5 knockout results in less severe joint destruction in a murine CAIA model. Regarding other disorders, in the pathogenesis of OA, GRK5 knockout mice have less cartilage degradation compared to WT mice in a model of surgically induced OA. Chondrocytes in GRK5 knockout mice have a smaller response to IL-1β stimulation 12 . In the pathogenesis of sepsis, Packiriswamy et al. reported that mortality due to induced polymicrobial sepsis was prevented in GRK5 knockout mice; therefore, GRK5 is an important regulator of inflammation in polymicrobial sepsis 44 . Research in the field of tumour biology showed that GRK5 phosphorylates p53 and inhibits DNA damage-induced apoptosis in cultured osteosarcoma cells and mice 45 . In the heart, GRK5 overexpression was reported to worsen heart failure and cardiac hypertrophy by functioning as a nuclear HDAC kinase irrespective of GPCRs 46,47 .
This study has some limitations. First, only inflammatory arthritis was included in the CAIA model, although there are several models of arthritis such as CIA, adjuvant-induced arthritis, and glucose-6-phosphate isomeraseinduced arthritis. Second, the effect of GRK5 knockout in this murine CAIA model was evaluated at two time points, 5 and 10 days after CAIA induction. Additional analysis at different time points might provide further information about the functional effects of GRK5 on the pathogenesis of arthritis. Third, GRK5 knockout mice used in this study were global knockout mice. FLSs-specific and macrophages-specific GRK5 knockout animals would be useful to analyze further specific function of GRK5.
In conclusion, our results demonstrated that GRK5 protein is highly expressed in inflammatory synovium and GRK5 deletion suppresses inflammatory responses in vivo and in vitro. GRK5 is a positive regulator of inflammatory responses; therefore, GRK5 inhibition could be a potential therapeutic target for types of inflammatory arthritis such as RA.

Materials and methods
Ethical approval. All animal studies were approved by the Committee of Ethics on Animal Experimentation of the Faculty of Medicine, Kyushu University (Fukuoka, Japan) and carried out in accordance with relevant rules and regulations. All animal studies were also carried out in compliance with the ARRIVE guidelines. Human synovium from individuals were obtained from healthy controls at autopsy under approval of the Scripps Human Subjects Committee or patients undergoing total knee arthroplasty under the approval of the Ethics Committee of Kyushu University Hospital. Informed consent was obtained from all patients prior to surgery. All experimental procedures were performed under the guidelines of Kyushu University (Fukuoka, Japan).
Clinical samples. Human normal synovium was harvested from three donors (aged 21-47 years, mean ± SD = 37.7 ± 14.5 years) with no history of joint disease. The inflammatory grade was 0. Human OA synovium samples were obtained from three donors (aged 60-76 years, mean ± SD = 67.7 ± 8.0 years) with grade III-IV OA. Human RA synovium samples were obtained from six donors (aged 37-66 years, mean ± SD = 56.5 ± 12.9 years). All patients with RA satisfied the 2010 American College of Rheumatology criteria for RA 48 . Immunohistochemistry of human synovium samples. Immunohistochemistry was performed on all human synovium tissue sections. Synovium samples were fixed in 4% paraformaldehyde for 2 days, delipidised, cut into sections that were 4 μm thick, and embedded in paraffin. Antigen retrieval was performed overnight with 1 mM EDTA at pH 8.0. Endogenous peroxidase activity was blocked with 3% hydrogen peroxidase in methanol for 30 min. For the blocking procedure, each specimen was placed in normal horse serum (VEC-TASTAIN Universal Elite Kit; Vector Laboratories, Burlingame, CA) for 20 min and then incubated for 1 h at room temperature with primary anti-GRK5 antibody (Proteintech, Rosemont, IL). Finally, the samples were counterstained with haematoxylin. Immunofluorescence staining of human RA synovium samples. Sections were stained with primary antibodies at room temperature for 1 h, then incubated with Alexa Fluor-conjugated secondary antibodies (Thermo Fisher Scientific, Waltham, MA). The following primary antibodies were used: GRK5 (Abcam and Proteintech), Cadherin-11 (GENETEX), and CD68 (Dako).

Quantification of positive cells in human synovium samples. GRK5 localisation in each synovial
Human RA synovial cells isolation and culture. Tissue samples were minced into pieces and incubated for 2 h at 37 °C in 5% CO 2 with 4 mg/mL of collagenase (Collagenase; Wako, Osaka, Japan) in DMEM/F12 (Gibco, Langley, OK). Dissociated cells were centrifuged at 500 g and re-suspended in DMEM containing 10% FBS (Gibco). The medium was changed every 3-4 days, and remaining adherent cells were used for subsequent experiments. siRNA transfection. We used human SW982 cells and human primary RA synovial cells for following procedures. Cells were seeded in 12-well plates at a density of 0.5 × 10 5 cells/well with DMEM and 10% FBS. After the cells reached subconfluence, they were transfected with siRNA (5 nM) against GRK5 (siGRK5, Santa Cruz Mice. GRK5 knockout mice and WT mice with a C57BL/6 background were used in all animal experiments.
These GRK5 knockout mice were general knockout mice. Mice were housed in a specific pathogen-free facility with a 12-h light, 12-h dark cycle and given free access to food and water. We have previously reported that there were no differences during growth and development between GRK5 knockout and WT mice 12 .

CAIA in mice.
Both GRK5 knockout and WT mice were used with the same protocol. As male mice have been reported to have greater susceptibility to CAIA induction than female mice 49 , we only used male mice in this study. Male mice were injected intraperitoneally with a five-clone cocktail of collagen type II antibodies (5 mg/mouse; Chondrex, Redmond, WA, USA) to induce arthritis at 12 weeks of age. Three days after antibody administration, 25 μg of LPS were injected intraperitoneally. Daily clinical scoring was undertaken using the manufacturer's protocol. The score for each leg was assessed on a scale of 0 to 4 (0, no change; 1, swelling and redness of one joint; 2, moderate swelling and erythema of ≧ two joints; 3, severe swelling and erythema of all joints; 4, extensive swelling and deformity of all joints); the total clinical arthritis score ranged from 0 to 16 34,50 . Arthritis was initially observed on day 4 and peaked around day 9-10. Thus, all mice were sacrificed on day 10, and paws were collected (n = 4 per each group). Paws were fixed in 4% paraformaldehyde for 2 days. After decalcification, murine ankle joints were cut into sections that were 4 μm thick along the sagittal plane and embedded in paraffin. Next, sagital ankle sections were stained with haematoxylin and eosin (H&E) and safranin-O and fast green (Safranin O) (n = 4 per each group). The severity of synovitis was quantified using a histopathological synovitis grading system previously reported by Krenn et al 51 . Cartilage degeneration (decrease in Safranin O) was quantified using the following semiquantitative grading scale: 0 = no pathological changes; 1 = minimal (minimal changes, or lesions involving < 25% of the whole section); 2 = slight (obvious changes, or lesions affecting 25-50% of the whole section; 3 = moderate (relatively severe changes, or lesions involving 50-75% of the whole section); 4 = severe (very severe changes, or lesions affecting > 75% of the whole section) 52 . Both scores were evaluated by two blinded independent observers. Scores were averaged to minimise observer bias. Immunohistochemistry was also performed on CAIA ankle sections.
Serum preparation and enzyme-linked immunosorbent assay (ELISA). In a CAIA model, murine blood samples were collected at day 0 (before a five-clone cocktail of collagen type II antibodies injection) and day 5 (paws swelling started to be appeared after LPS injection) (n = 3 per each group). Serum was prepared by centrifugation of coagulated blood samples at 3,000 rpm for 15 min and frozen at − 80 °C. The concentrations of IL6 and TNFα in collected serum samples were measured using murine IL6 and TNFα ELISA kits (R&D Systems, Minneapolis, MN, USA).
Murine bone marrow derived macrophages (BMDMs) culture. Murine BMDMs were obtained from both GRK5 knockout and WT mice. Whole bone marrow was flushed from the femurs and tibiae, and cells were collected by centrifugation at 400 g for 5 min at 4 °C. Cells were resuspended in RPMI 1640 (Gibco) supplemented with 10% FBS and 20 ng/ml murine recombinant GM-CSF (R&D Systems), then seeded and cultured at 37 °C in 5% CO2 for 7 days. Non-adherent cells were removed by washing, and only adherent cells were used for subsequent experiments.
Murine FLSs isolation and culture. Tissue samples were minced into pieces (1-3 mm) and incubated for 2 h at 37 °C in 5% CO 2 with 4 mg/mL of collagenase in DMEM/F12. Dissociated cells were centrifuged at 500 g and re-suspended in DMEM containing 10% FBS. After 24 h, non-adherent cells were removed. The remaining adherent cells were cultured in DMEM with 10% FBS. Cultures were maintained at 37 °C and 5% CO 2 . The medium was changed every 3-4 days. FLSs from passages 3 to 8 were used for subsequent experiments. Murine FLSs were isolated from hip joint tissue from both GRK5 knockout and WT mice as previously described 53 .