Introduction
As new biologics enter the clinical arena and advances in synovial histopathology identify divergent mechanisms of arthritis progression, it is essential to understand how the cytokine network governs the pattern of synovial inflammation [
1],[
2]. Innate sensing mechanisms involving pattern recognition receptors are increasingly implicated in autoimmunity and promote cytokine responses associated with rheumatoid arthritis. Although these pathways represent promising therapeutic targets, further investigation is required to understand the expression and functional contribution of pattern recognition receptors in autoimmunity.
Pattern recognition receptors were initially characterized as sensors of microbial products of bacterial, fungal or viral infection. These include the Toll-like receptors (TLR); nucleotide-binding domain and leucine-rich repeat containing receptors (NLR), Rig-I-like receptors (RLR) and C-type lectins [
3],[
4]. Activation of these receptors promotes the inflammatory regulation of various interleukins, tumour necrosis family members and type-1 interferons [
3]. They therefore represent innate sensing mechanisms, which shape the adaptive immune response to chronic disease, allergy, cancer and infection. As a consequence, various processes have evolved to protect against the prolonged activation of these receptors. For example, interleukin (IL)-10 limits the duration and intensity of their signaling in myeloid cells [
5],[
6]. Here, IL-10 inhibits pattern recognition receptor signaling through mechanisms, which include downregulation of MyD88 expression [
7], and the ubiquitination and subsequent degradation of MyD88-dependent signaling molecules such as TRAF6 [
8]. In experimental models of inflammatory arthritis, IL-10 is protective and mice deficient in IL-10 show exacerbated joint inflammation [
9],[
10]. These data are consistent with the characterization of IL-10 as a cytokine synthesis inhibitory factor, which acts as an immunomodulatory cytokine affecting both innate and cellular immunity [
11],[
12]. For example, IL-10 inhibits nuclear factor kappa B (NF-κB) signaling in response to TLR agonists to block expression of certain proinflammatory mediators associated with arthritis progression. Interleukin-10 is abundantly expressed in synovial fluids of RA patients and has been linked with the control of bone resorption through inhibition of osteoclastogenesis
in vitro[
13]-[
15].
While IL-10 is renowned for its ability to inhibit TLR signaling, its impact on innate sensing receptors, including the NLR family, is less documented. Here, caspase 1 activity is essential for the processing of cytokine precursors (for example, pro-IL-1β, pro-IL-18 and potentially pro-IL-33) into functionally active mature forms [
16]. Activation of caspase 1 by the NLR family member NLRP3, acting in association with its adaptor protein ASC, leads to the secretion of IL-1β and IL-18 [
16]. Each of these cytokines affect arthritis progression [
17],[
18], which suggests that innate sensing complexes (termed the inflammasome) must be regulated during the course of disease. Various particulate and crystalline agonists activate the NLRP3 inflammasome. For example, monosodium urate crystals and calcium pyrophosphate dihydrate crystals trigger arthritis symptoms in inflammatory forms of gout and pseudogout [
19]-[
21], while basic calcium phosphate (hydroxyapatite) crystals are seen in 70% of osteoarthritis (OA) cases [
22],[
23]. A role for the NLRP3 inflammasome in arthritic disease is illustrated by several
in vitro studies, which show that crystals from the joints of OA patients and basic calcium phosphate crystals induce IL-1β production by macrophages [
24],[
25]. Also, a recent report shows that there is modulation of the NLRP3 inflammasome in peripheral blood mononuclear cells in RA patients and that single nucleotide polymorphisms (SNPs) in
NLRP3 are associated with disease severity [
26]. However, little is known about the regulation and activation of inflammasome components in inflammatory arthritis. We now show that the exacerbated joint pathology seen in IL-10KO mice during antigen-induced arthritis (AIA) is associated with increased synovial expression of NLRP3 inflammasome components and a localized expression of IL-1β at sites of focal bone erosions. Our data supports a role for IL-10 as a negative regulator of the inflammasome and highlights a role for the inflammasome in osteoclastogenesis during inflammatory arthritis.
Materials and methods
Mouse strains
Inbred C57BL/6 mouse strains from The Jackson Laboratory (Bar Harbor, ME, USA) were bred and maintained in-house under high barrier and pathogen-free conditions. All animal studies were performed in the United Kingdom. Experiments were performed on eight- to twelve-week-old male mice in accordance with UK Home Office Project License PPL-30/2361 and 30/2928. The ethical approval of these licenses covers all aspects of the study and all experiments conducted.
Induction of murine AIA
Mice were immunized (subcutaneous (s.c.)) with an emulsion containing 1 mg/ml methylated bovine serum albumin (mBSA) in phosphate-buffered saline (PBS) and Freud's complete adjuvant (CFA) (Sigma-Aldrich, St Louis, MO, USA). Concurrently, mice were injected (intraperitoneal (i.p.)) with 200 ng of heat-inactivated Bordetella pertussis toxin adjuvant (Sigma-Aldrich, Poole, UK). The immune response was boosted one week later with a second injection (s.c.) of mBSA emulsified in CFA. Arthritis was induced two weeks later with an intra-articular (i.a.) injection of 10 μl of mBSA (10 mg/ml) into the right knee joint. Arthritis progression was monitored using a micrometer to measure changes in knee joint swelling.
Radiology
A Kodak in vivo Imaging System FX was used to take radiographs of the mouse knee joints. Both arthritic (right) and non-arthritic joints (left) were compared. Radiographic scores were independently assigned by an orthopedic registrar and based on visible bone erosions (0; normal, 1; mild, 2; moderate, 3; severe).
Histology
Joints were fixed in neutral-buffered formalin saline, decalcified with formic acid at 4°C and embedded in paraffin. Midsagittal sections (8 μm) were stained with haematoxylin, safranin-O and Fast Green. Two independent observers scored histology sections for subsynovial inflammation (0 = normal, to 5 = ablation of adipose tissue due to leukocyte infiltrate), synovial exudate (0 = normal, to 3 = substantial number of cells with large fibrin deposits), synovial hyperplasia (0 = normal 1 to 3 cells thick, to 3 = over three layers thick with overgrowth onto joint surfaces with evidence of cartilage/bone erosion), cartilage/bone erosion (0 = normal, 3 = destruction of a significant part of the bone). Cartilage integrity was determined in histological sections using a Mankin scoring system. Two independent observers evaluated cartilage irregularity and cleft formation (0 = normal, to 6 = complete disorganization of glycoproteins with clefts into the cartilage), cellularity (0 = normal, to 3 = hypocellularity), proteoglycan depletion (0 = normal, to 4 = complete proteoglycan degradation with no dye apparent) and tidemark integrity (0 = intact, to 1 = tidemark crossed by blood vessels). The total sum of these scores resulted in a maximum score of 14. For detection of tartrate-resistant acid phosphatase (TRAP) activity, slides were rehydrated after decalcification, incubated with TRAP staining solution (0.2 M acetate buffer, 50 mM sodium tartrate, 0.5 mg/ml naphthol AS-MX phosphate, 1.1 mg/ml Fast Red Violet LB salt) and counterstained with haematoxylin.
Immunohistochemistry
Antigen retrieval was performed on paraffin-embedded sections using either trypsin (0.1%) for 30 mins at 37°C, or 10 mM citrate buffer (pH 6) for 40 mins at 95°C. Endogenous peroxidase and biotin activity was blocked using 3% H202 and an avidin/biotin blocking kit (Vector Laboratories, Burlingame, CA, USA) respectively. Sections were incubated in 10% (v:v) rabbit serum for 1 hour before staining with rat anti-mouse F4/80 (1:50 dilution, Santa Cruz Technology, Santa Cruz, CA, USA). Antibody binding to sections was detected with rabbit anti-rat biotin-conjugated secondary antibody and streptavidin-horseradish peroxidase (HRP) complex (Vector Laboratories). Diaminobenzidine substrate (Dako, Glostrup, Denmark) was used to develop sections and haematoxylin was used as a counterstain.
Image analysis
Immunohistochemistry was viewed with a Leica DMLB light microscope (Milton Keynes, UK). Analysis across five random fields of view was performed using the Leica digital image capture program. Values are expressed as a percentage of total immunoperoxidase staining.
Osteoclast cell culture and TRAP stain
Bone marrow cells from femurs of WT and IL-10KO mice were re-suspended in alpha minimum essential medium (αMEM) supplemented with 10% (v:v) foetal calf serum (FCS) and seeded at a density of 6.4 × 106 cells/ml in 24-well plates. Following adhesion culture media was supplemented with macrophage colony-stimulating factor (MCSF) and receptor activator of NF-κβ ligand (RANKL). IL-10 and inflammasome inhibitors were added as indicated in figure legends. TRAP-positive cells were detected after seven days and total RNA isolated from subsequent analysis.
Quantitative real-time PCR (qPCR)
Synovial membranes were dissected from the underpinning cartilage of knee joints [
27]. Total RNA was extracted from samples using TRI Reagent (Sigma-Aldrich) and cDNA derived from 1 μg of total RNA using a reverse transcription kit (Primer Design, Southampton, UK) [
27]. Gene expression analyses were performed on triplicate samples with SYBR Green (Invitrogen, Thermo Fisher Scientific, Carlsbad, CA, USA) using an ABI Prism 7900HT instrument (Applied Biosystems, Thermo Fisher Scientific, Carlsbad, CA, USA). Details of oligonucleotide primer sequences are presented in Additional file
1. Data analysis was performed using the Sequence Detection System Version 2.3 software (Applied Biosystems).
Enzyme-linked immunosorbent assay (ELISA)
Serum dickkopf-1 (DKK1) was quantified using a commercial enzyme-linked immunosorbent assay (ELISA) (R&D Systems Inc, Minneapolis, MN, USA) as per the manufacturer's instructions.
Statistical analysis
Data was evaluated using the non-parametric Mann-Whitney U test and an unpaired Student t test. In all cases, P <0.05 was considered significant.
Discussion
Innate sensing mechanisms were traditionally linked with the recognition of bacterial, fungal and viral infections. This viewpoint has, however, changed and many act as sensors of both endogenous and exogenous danger signals [
32]-[
34]. Members of the NLR family are intrinsic to caspase-activating complexes termed the inflammasome. These receptors recognize certain infectious pathogens, particles (for example, microcrystals), metabolic anomalies (for example, hyperglycaemia, ATP) and chemicals, and contribute to the pathology of various autoinflammatory diseases [
32]. This has led to the novel application of drugs that target IL-1β (for example, anakinra, canakinumab and rilonacept) in conditions such as periodic fever syndromes, Still's disease, Schnitzler's syndrome, and gouty arthritis where conventional anti-inflammatory drugs fail to provide long-lasting relief [
35]. Here, we show that IL-10 negatively regulates the expression of NLRP3 inflammasome components within the inflamed synovium of experimental arthritis and provide a link to degenerative bone erosion.
Several lines of evidence support an involvement of the inflammasome in inflammatory and degenerative arthritis. While NLRP3 is expressed in RA and OA synovium [
36],[
37] it is difficult to comment on the functional significance of these findings as quantification of transcript levels provide minimal information on inflammasome activity. An involvement of the inflammasome in joint pathology is illustrated by studies of calcium crystals, which accumulate through biomechanical stress or altered mechanisms of calcification. Here, the ectopic deposition of hydroxyapatite crystals in synovial fluids from osteoarthritis patients is associated with disease progression [
38]. Hydroxyapatite crystals activate the NLRP3 inflammasome to promote IL-1β and IL-18 release by lipopolysaccharide (LPS)-primed macrophages [
24]. Studies in an air pouch model of synovitis also confirmed the activation of NLRP3/ASC/caspase 1 by hydroxyapatite crystals and supported a role in controlling neutrophil infiltration [
24]. Analysis of synovial inflammation in AIA-challenged ASC-KO mice showed reduced disease severity [
39]. This response was not, however, seen in NLRP3-KO and caspase1-KO mice, where both genotypes showed similar pathology to that observed in WT controls [
39]. These results are consistent with our own, where we see similar disease pathology between the WT and IL-10KO mice at these early time points. We now show that at the later stages of inflammatory arthritis, in the absence of IL-10, joint inflammation results in a temporal increase in synovial IL-1β, which corresponds with enhanced pathology and co-localization to areas rich in F4/80-postive cells that cluster at sites of focal bone erosion. Such changes in IL-1β may reflect the capacity of IL-10 to inhibit TLR control of
Il1β (pro-IL-1β) and the expression of NLRP3 inflammasome components. At this stage we are, however, unable to ascertain what inflammatory events (for example, TLR or cytokine-driven outcomes) regulate the induction of these inflammasome genes.
Our results implicate the involvement of the inflammasome in osteoclastogenesis, with IL-10 deficiency causing an increase in both synovial IL-1β and IL-33 expression, but not IL-18. Interleukin-1β is considered pro-osteoclastogenic and drives an imbalance in chondrocyte responses through regulation of prostanoids, nitric oxide and other free radicals, and the induction of degradative enzymes that impact collagen and proteoglycan turnover [
40],[
41]. In contrast, IL-33 and IL-18 are considered anti-osteoclastogenic and protect against tumour necrosis factor alpha (TNFα)-mediated bone loss [
42],[
43]. Thus, IL-33 often opposes the activities of IL-1β, which raises a question about the regulation of these cytokines by the NLRP3/caspase 1 system. While pro-IL-1β and pro-IL-18 are processed by caspase 1 into mature active forms, full-length IL-33 is already biologically active and is released as a consequence of cell damage [
44]. Further processing by caspase 1 causes inactivation of IL-33 [
44],[
45]. IL-33 is also modified by the activity of elastase and cathepsin G, which are secreted by infiltrating neutrophils and enhance IL-33 bioactivity [
46]. Here, IL-33 acts as an endogenous danger signal (alarmin), which alerts innate immune cells to sites of infection or injury. Thus, the elevated expression of
Il33 in IL-10KO mice may simply reflect the overall increase in synovial inflammation. Moreover, the blockade of osteoclastogenesis by CRID3 and glibenclamide suggests a role for caspase 1 in bone turnover. Targeting the inflammasome may therefore benefit joint pathologies allied with IL-1β production, such as OA, Muckle-Wells syndrome (an autoinflammatory disorder linked with mutations in NLRP3) and gout [
20],[
35],[
47],[
48]. Such an approach may provide an added benefit over traditional biologic interventions including anakinra, which in clinical trials of OA offered no improvement in symptoms [
49]. To provide preclinical evidence to support this notion, attempts were made to treat AIA-challenged IL-10KO mice (i.a.) with either CRID3 or glibenclamide. However, administration (i.a.) of vehicle alone as a control promoted a robust inflammatory response, which prevented the direct investigation of this
in vivo.
Our results indicate that IL-10 has the capacity to inhibit expression of NLRP3 and certain components of the inflammasome. Here, neutralization of IL-10 receptor signaling in WT mice enhanced synovial
Nlrp3 and
Casp1 expression and addition of endogenous IL-10 inhibited the formation of TRAP+ osteoclasts
in vitro. Significantly, human trials with recombinant IL-10 showed no improvement in disease activity [
50], while IL-10 responses in synovial macrophages from RA patients appear dysregulated [
51]. Such alterations in IL-10 responsiveness results in a loss of its anti-inflammatory properties and a concomitant acquisition of interferon-gamma (IFNγ)-like activities [
51].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CJG, GWJ, MAN and ASW performed all animal experiments, tissue processing, histology scoring, and contributed to the drafting of the manuscript. CJG and ZN performed all real-time PCR and immunoassays. AKH, ANM and MAN participated in the radiographic studies and design of scoring criteria. CJG, AKH, FLC, ACB, ASW and MR performed all osteoclast studies including data analysis and assay development. AKH participated in the statistical evaluation of data. RC, LAO, MAC and IRH provided necessary research reagents and AABR synthesized the CRID3. RC, LAO and MAC provide necessary technical expertise relating to the inhibition of the inflammasome and interpretation of results. PRT, IRH, ASW and SAJ designed experiments and evaluated all results. CJG, GWJ and SAJ wrote the manuscript. All authors read and approved the manuscript.