Article Text
Abstract
Objectives To define the cell type (myeloid vs other cells) specific effect of interleukin 1 (IL-1) receptor antagonist (IL-1Ra) deficiency on the acute inflammatory phase of arthritis.
Methods Arthritis was induced by K/BxN serum transfer in wild-type (WT), IL-1Ra-deficient (IL-1Ra−/−) and conditional knockout mice. In the latter, IL-1Ra production was specifically targeted in myeloid cells (IL-1RaΔM) or in both hepatocytes and myeloid cells (IL-1RaΔH+M). Arthritis severity was clinically evaluated and ankle sections were scored for synovial inflammation and cartilage erosion. Quantitative RT-PCR, western blot and immunohistochemical analyses measured expression, localisation and cellular sources of the different IL-1Ra isoforms in arthritic joints.
Results Total and myeloid cell-specific IL-1Ra deficiency was associated with increased arthritis severity, although disease incidence was similar to that of WT mice. Increased clinical scores were associated with exacerbated synovial inflammation. All IL-1Ra isoforms, except for intracellular (ic)IL-1Ra2, were expressed in arthritic joints of WT mice. In contrast, production of secreted (s)IL-1Ra and icIL-1Ra3 isoforms was markedly decreased in arthritic joints of both IL-1RaΔM and IL-1RaΔH+M mice. Immunohistochemical and western blot analyses suggested that the icIL-1Ra1 isoform is produced primarily by synovial fibroblasts.
Conclusion Myeloid cell-derived IL-1Ra, including both sIL-1Ra and icIL-1Ra3 isoforms, controls articular inflammation during the acute phase of K/BxN serum transfer-induced arthritis.
Statistics from Altmetric.com
Introduction
Interleukin-1 receptor antagonist (IL-1Ra) is a well-known anti-inflammatory cytokine belonging to the IL-1 family. Four IL-1Ra isoforms are produced from the same IL-1RN gene.1 One isoform is glycosylated and secreted (sIL-1Ra) with a molecular weight that varies between 17 and 22 kDa.1 The three others have no leader peptide and thus remain intracellular (icIL-1Ra1, 2 and 3). icIL-1Ra1 is generated by alternative splicing into the sequence encoding the leader peptide of sIL-1Ra and has a molecular weight of 18 kDa.2 IcIL-1Ra2 mRNA contains an extra 63 bp coding region unique to this isoform, producing a predicted 26 kDa protein.3 IcIL-1Ra3 is the smallest isoform with a molecular weight of 16 kDa and is generated by alternative translation initiation from sIL-1Ra mRNA.4 The expression of these IL-1Ra isoforms differs depending on cell type and stimulus.2 5,–,8
Extensive data have been described over the past 25 years demonstrating that the major function of extracellular IL-1Ra is to modulate pleiotropic proinflammatory effects of IL-1. IL-1Ra binds to cell-surface IL-1 receptors (IL-1Rs) without inducing any intracellular response and prevents thereby the interaction between IL-1 and IL-1Rs, thus competitively inhibiting its biological effects.8 9 Intracellular isoforms can also be released after cell lysis or under particular circumstances, and then bind to IL-1Rs to interfere with IL-1 activities.4 10 11 It has also been suggested that they exert regulatory roles within cells in vitro.12,–,15
The presence of an excess of IL-1 in joints leads to the development of arthritis characterised by synovial proliferation as well as cartilage and bone erosion.16 17 In contrast, blockade of endogenous production or activity of IL-1 by gene deletion, gene therapy, or administration of recombinant IL-1Ra, anti-IL-1 antibodies, or soluble IL-1 receptors (sIL-1Rs), attenuates joint inflammation and tissue damage in several experimental models of arthritis.18,–,22 The beneficial effects of recombinant IL-1Ra administration in some patients with rheumatoid arthritis also supports a contribution of the IL-1/IL-1Ra axis to the pathogenesis of this disease.23 24
Recently, we provided the first in vivo evidence that myeloid cell-derived IL-1Ra plays a key role in the control of the development and severity of collagen-induced arthritis (CIA) by modulating T cell-mediated immunity in lymphoid organs upstream of local inflammatory responses.25 Myeloid cells, such as neutrophils, are known to massively infiltrate arthritic joints.26 However, using the CIA model, we could not distinguish the relative contribution of myeloid cell-derived IL-1Ra to the initiation phase, including T cell responses, and to the articular inflammatory phase of the disease. In addition, we recently demonstrated that myeloid cells and hepatocytes are the two major sources of systemic IL-1Ra in response to different inflammatory stimuli in vivo.27
Thus, the objective of this study was to determine the role of total IL-1Ra deficiency and the specific contributions of hepatocyte- and myeloid cell-derived IL-1Ra during the effector phase of arthritis by using the K/BxN serum transfer-induced model, which bypasses adaptive immune responses.
Materials and methods
Materials
Recombinant human IL-1β and mouse sIL-1Ra were obtained from R&D Systems (Abingdon, UK), and purified lipopolysaccharide (LPS) from Fluka (Escherichia coli 055:B5, Buchs, Switzerland). Recombinant mouse icIL-1Ra1 was generated as previously described.28
Mice
Wild-type (WT) C57BL/6J mice were obtained from Janvier (Le Genest-St-Isle, France). Conditional myeloid cell-specific IL-1Ra-deficient mice (IL-1RaΔM mice) and hepatocyte–myeloid cell-specific IL-1Ra-deficient mice (IL-1RaΔH+M mice) were generated, backcrossed into the C57BL/6J genetic background and then characterised in our laboratory.27 IL-1Ra-deficient C57BL/6J mice (IL-1Ra−/−) were originally obtained from Dr MJ Nicklin (Division of Molecular and Genetic Medicine, University of Sheffield, Sheffield, UK).29 PCR genotyping of IL-1RaΔM, IL-1RaΔH+M and IL-1Ra−/− mice has been previously described.25 27 These mice were maintained under conventional conditions in the animal facility of the Geneva University School of Medicine, and water and food were provided freely. KRN T-cell receptor transgenic mice (K/B) were kindly provided by C Benoist and D Mathis (Harvard Medical School, Boston, Massachusetts, USA) and genotyped as previously described.30 Non-obese diabetic (NOD/ShiLtJ) mice were purchased from Jackson Laboratory (Bar Harbour, Maine, USA). Arthritic K/BxN mice were generated by crossing K/B mice with NOD/ShiLtJ mice in ventilated cages and then housed at 5 weeks old under conventional conditions. Animal studies were approved by the Animal Experimentation Ethics Committee and the Geneva Veterinarian Office and were performed according to the appropriate codes of practice.
K/BxN serum transfer-induced arthritis and clinical scoring
K/BxN serum was collected from 9-week old arthritic K/BxN mice. The serum samples were pooled and stored at −80°C until use. K/BxN serum transfer-induced arthritis was induced in adult female IL-1RaΔM, IL-1RaΔH+M, IL-1Ra−/− and WT C57BL/6 mice by a single intraperitoneal injection (200 μl) of K/BxN serum. Mice were scored clinically every day for the development of arthritis using a semiquantitative scoring system.30 The mice were killed on day 4 after serum transfer owing to the occurrence of severe arthritis in some mice among the different IL-1Ra-deficient mouse lines.
Histological scoring of arthritis
At sacrifice, ankles (right) were dissected and fixed in 10% formalin, decalcified in 15% EDTA and embedded in paraffin. Serial sections (3 μm) were stained with haematoxylin and eosin for evaluation of inflammation or with toluidine blue to analyse cartilage damage. Sections were scored for inflammation and cartilage erosion, as described elsewhere.30 Scoring was performed by a pathologist (CAS) in a blinded manner—that is, not knowing the different conditions of the mice joints.
Determination of serum levels of IL-1Ra and IL-6
Blood was collected at sacrifice 4 days after serum injection. Serum samples were obtained after blood coagulation and subsequent centrifugation. Serum concentrations of IL-1Ra and IL-6 were quantified by ELISA using commercial Duoset ELISA Development Systems from R&D Systems (Abingdon, UK).
Real-time PCR analysis
Ankles (left) were collected 4 days after serum transfer. Total RNA was extracted with Trizol reagent (Invitrogen AG, Basel, Switzerland). Levels of sIL-1Ra and icIL-1Ra1 mRNA were examined by RT-quantitative PCR, using the following primers: 5′-TTGCTGTGGCCTCGGGATGG-3′ (forward sIL-1Ra), 5'-AGACACTGCCTGGGTGCTCCT-3' (forward icIL-1Ra1) and 5′-GTTTGATATTTGGTCCTTGTAAG-3′ (reverse for both isoforms). IL-6 was amplified using the following primers: 5′-TGAACAACGATGATGCACTTGCAGA-3′ (forward IL-6), and 5′-TCTGTATCTCTCTGAAGGACTCTGGCT-3′ (reverse IL-6). Relative levels of IL-1Ra isoforms and IL-6 mRNA expression were normalised to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA levels using a comparative method (2–ΔCt). The annealing temperature was 60°C. Non-reverse-transcribed RNA samples and water were included as negative controls.
Western blot analysis
Total proteins were extracted from wrist joints of mice at sacrifice. Samples from IL-1RaΔM (n=6), IL-1RaΔH+M (n=7) and WT (n=4) mice were analysed. Whole protein lysates (50–60 μg protein) were fractionated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (18%) and electrotransferred onto a polyvinylidene fluoride membrane (Macherey-Nagel, Düren, Germany) in a semi-dry transfer cell. IL-1Ra proteins were detected with a polyclonal goat anti-mouse IL-1Ra antibody (diluted 1/10 000) recognising all described IL-1Ra isoforms, and revealed with a donkey anti-goat IgG horseradish peroxidase (Santa Cruz Biotechnology, Heidelberg, Germany) at a dilution of 1/10 000. Protein loading was assessed by the determination of GAPDH expression. Bands obtained for icIL-1Ra1 and icIL-1Ra3 were quantified using an image analysis software (Multi Gauge, version 3.2, Fujifilm). Expression of icIL-1Ra isoforms in each sample was normalised to GAPDH protein levels. Results obtained were expressed in relative values as compared with expression of the same icIL-1Ra isoform in WTmice.
Immunohistochemistry
IL-1Ra expression was studied by immunohistochemistry on paraffin-embedded knee sections. Sagittal sections (3 μm) were deparaffinised in xylol and rehydrated through graded concentrations of ethanol. Endogenous peroxidase was blocked using 0.6% H2O2 for 10 min and tissue sections were then boiled in citrate-based antigen unmasking solution (10 mM, pH 6) for 3 min. Tissue sections were incubated with 5.8 μg/ml of polyclonal goat anti-mouse IL-1Ra antibody (capture antibody of mouse IL-1Ra Duoset ELISA kit; R&D Systems) for 2 h. The sections were rinsed and then incubated for 1 h with 1.6 μg/ml of donkey anti-goat IgG horseradish peroxidase (Santa Cruz Biotechnology). Colour was developed with diaminobenzidine, and tissues were counterstained with haematoxylin.
Resident peritoneal macrophages and mouse primary synovial fibroblasts
C57BL/6 mice were killed by CO2 asphyxiation. Peritoneal macrophages were collected after washing the peritoneal cavity with phosphate-buffered saline. The cells were then centrifuged, resuspended in RPMI supplemented with 10% fetal calf serum, and plated in 5 cm culture dishes. After 2 h non-adherent cells were removed and culture media was replaced with the addition of 1 μg/ml LPS for 24 h. For the preparation of synovial fibroblasts, knee joints were dissected under microscope, and synovial explants were minced and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. After a few days of culture, synovial fibroblasts started to emerge from the explants and grow in culture. At passage 2, the cells were plated in 5 cm culture dishes and cultured for 24 h with 1 ng/ml IL-1β. Peritoneal macrophages and synovial fibroblasts were lysed in Trizol reagent (Invitrogen, Basel, Switzerland), and total proteins were prepared according to the manufacturer's recommendations and then used for western blot analysis.
Statistical analysis
One-way analysis of variance and an unpaired two-tailed Student t test were used for statistical analysis. A comparison between two groups using the unpaired two-tailed Student t test was made only when the one-way analysis of variance test yielded statically significant results. Incidence of arthritis was compared using the Kaplan–Meier statistic. p Values <0.05 were considered significant.
Results
Increased severity of K/BxN serum transfer-induced arthritis in mice lacking total IL-1Ra or myeloid cell-derived IL-1Ra
Mice lacking total IL-1Ra (IL-1Ra−/−) exhibited an earlier onset of arthritis, starting on day 1 after K/BxN serum transfer, as well as a more severe form of the disease, as indicated by raised clinical scores (figure 1A,B). Both onset of arthritis and evolution of clinical severity scores in IL-1RaΔM, IL-1RaΔH+M and IL-1Ra−/− mice had a similar profile, showing no significant difference. In contrast, WT mice exhibited a slower onset of disease (day 2) and attenuated arthritis severity. In addition, the number of affected paws was higher in IL-1Ra−/−, IL-1RaΔM and IL-1RaΔH+M mice than in WT mice (figure 1C). Similarly, circulating levels of IL-6, used as a marker for inflammation, were significantly increased in IL-1Ra−/− mice as compared with WT mice (figure 1D). Of note, the plasma levels of IL-1Ra tended to be lower in IL-1RaΔM (46 ± 80 pg/ml) and IL-1RaΔH+M (30 ± 54 pg/ml) mice than in WT mice (120 ± 121 pg/ml), suggesting that myeloid cells, but not hepatocytes, contribute to the production of circulating IL-1Ra in K/BxN serum transfer-induced arthritis (data not shown).
Total or myeloid cell-specific IL-1Ra deficiency accelerates joint inflammation after K/BxN serum transfer
Consistent with the clinical severity, synovial inflammation was markedly increased in IL-1RaΔM, IL-1RaΔH+M and IL-1Ra−/− mice as compared with WT mice 4 days after serum transfer (figure 2A,C (left panels)). These observations correlated with increased IL-6 mRNA levels in arthritic joints of total and conditional IL-1Ra−/− mice (figure 2D). In contrast, no significant differences were seen for IL-1β mRNA and protein levels between the four genotypes (data not shown). Cartilage damage scores were low and no significant differences were seen between the different genotypes on day 4 (figure 2B,C (right panels)). More extensive cartilage damage was observed on day 6, but again without any significant differences between the groups (WT=2.0±1.7, IL-1RaΔM=2.4±1.8, IL-1Ra−/−=2.3±1.7). The presence of a large variability in the development of cartilage erosions in this model does not allow conclusions to be reached about the role of IL-1Ra deficiency on structural damage.
Cellular sources of different IL-1Ra isoforms in the joint
Articular total protein extracts of IL-1RaΔM, IL-1RaΔH+M, IL-1Ra−/− and WT mice were analysed by western blot. We observed mainly the presence of icIL-1Ra1 and icIL-1Ra3 in arthritic joints of WT mice, probably reflecting an accumulation of these intracellular proteins in joint cells, as illustrated for two mice per group in figure 3A. Of note, icIL-1Ra3 expression was reduced in protein extracts of the conditional IL-1Ra−/− mice, whereas icIL-1Ra1 protein levels were similar in IL-1RaΔM, IL-1RaΔH+M and WT mice (relative icIL-1Ra isoform expression as compared with levels in WT (n=4) mice for icIL-1Ra1: IL-1RaΔM (n=6)=0.91±0.57, IL-1RaΔH+M (n=7)=0.78±0.86; for icIL-1Ra3: IL-1RaΔM=0.26±0.22, IL-1RaΔH+M=0.01±0.01. In addition, we observed a marked decrease of sIL-1Ra mRNA levels in ankle extracts of IL-1RaΔM and IL-1RaΔH+M as compared with WT mice (figure 3B), whereas icIL-1Ra1 mRNA levels were not significantly different (figure 3C). Immunohistochemical analysis demonstrated the production of IL-1Ra by myeloid cells, including neutrophils, in arthritic joints of WT mice but not in conditional IL-1Ra−/− mice (figure 3D). Taken together, the results indicate that myeloid cells represent the major source of local sIL-1Ra and icIL-1Ra3 in acute arthritis. In addition, IL-1Ra was also detected in the synovial tissue of IL-1RaΔH+M mice, suggesting that resident cells such as synovial fibroblasts could represent the main cellular source of the icIL-1Ra1 isoform (figures 3C,D). Consistent with this hypothesis, we observed that icIL-1Ra1 is produced by cultured synovial fibroblasts in response to IL-1 stimulation (figure 4).
Discussion
The transfer of K/BxN serum into recipient mice induces an acute joint-specific inflammatory response. In this experimental arthritis model, the effector phase of the disease occurs independently of adaptive immune responses, since both T- and B-lymphocytes are dispensable. However, its initiation and progression require contributions from several cell types of the innate immune system such as neutrophils and macrophages.31,–,33 Myeloid cells are known to be prominent participants in synovial inflammation in both human rheumatoid arthritis and mouse arthritis models, but the extent of their role in the control of inflammatory responses is poorly understood. In this study, we demonstrated that myeloid cells can regulate the development of acute arthritis by the production of IL-1Ra, thus inhibiting the local proinflammatory effects of IL-1. Of note, the contribution of IL-1 to the effector phase of arthritis was already confirmed by the total absence of clinical signs of disease in IL-1RI−/− mice in response to K/BxN serum transfer.33 Since neutrophils are the most abundant cell type in inflamed joints, constituting 90% of cells present in the synovial fluid of patients with inflammatory arthritis and in the synovium of arthritic animals, we suggest that they represent the major cellular source of local IL-1Ra.17 34
The production of icIL-1Ra3 occurs by alternative translation initiation from sIL-1Ra mRNA.4 Both sIL-1Ra and icIL-1Ra3 proteins were detected in transgenic mice overexpressing sIL-1Ra but not icIL-1Ra1 mRNA.35 Accordingly, we observed that both sIL-1Ra and icIL-1Ra3 were absent in arthritic joints of myeloid cell-specific IL-1Ra-deficient mice. Consistent with our results, icIL-1Ra3 was previously detected by western blot analysis in cell lysates of LPS-stimulated neutrophils and monocytes, and in small amounts in joint of mice with CIA.8 36 Like icIL-1Ra1, icIL-1Ra3 lacks a leader sequence and remains within the intracellular space. To date, no IL-1RI independent effect of icIL-1Ra3 has been reported. However, icIL-1Ra3 is probably released by dying cells within the synovial membrane, and can then bind to the IL-1 receptors. It is thus conceivable that icIL-1Ra3 is produced and released by myeloid cells in synovial fluid under inflammatory conditions, and contributes to the regulation of joint-specific inflammatory response by inhibiting the local effects of IL-1. However, previous reports showed that recombinant icIL-1Ra3 binds to IL-1RI with a four- to fivefold lower affinity than recombinant sIL-1Ra and icIL-1Ra1, and is two- to fourfold less active than the other two isoforms.4 Thus, the respective roles of icIL-1Ra3 and sIL-1Ra in the control of the acute arthritis are still unclear.
We also examined a potential contribution of hepatocytes, as a source of systemic IL-1Ra, in the control of arthritis. Recently, we demonstrated that hepatocytes and myeloid cells are the two major sources of circulating and hepatic IL-1Ra during systemic inflammatory events.27 Indeed, hepatocytes were the main source of IL-1Ra in response to an intraperitoneal injection of IL-1, with as consequence a regulatory effect on systemic inflammatory responses such as serum IL-6 and chemokine levels. The production of IL-1Ra by hepatocytes was also upregulated as an acute-phase protein in response to local tissue damage and inflammation.7 Low levels of circulating IL-1Ra were detected during K/BxN serum transfer-induced arthritis in WT mice. However, circulating IL-1Ra levels, as well as development and severity of arthritis, were similar in both IL-1RaΔM and IL-1RaΔH+M mice, indicating that hepatocyte-derived IL-1Ra does not play a major role in modulating the effector phase of arthritis.
The transcriptional regulation of sIL-1Ra and icIL-1Ra1 isoforms is under the control of two different promoters, leading to cell type and stimulus-specific expression.7 37,–,39 Indeed, sIL-1Ra is detected in activated monocytes/macrophages, neutrophils and hepatocytes, whereas icIL-1Ra1 is constitutively produced by keratinocytes and endothelial cells, and expressed also in activated human dermal and synovial fibroblasts.1 40 41 Interestingly, it has been shown that the proximal region of the icIL-1Ra1 promoter lacks regulatory elements controlling expression of this isoform in LPS-stimulated macrophages.35 Accordingly, we provide in vivo evidence that myeloid cells do not contribute to icIL-1Ra1 production in inflammatory joints. Furthermore, both immunohistochemistry and western blot analysis suggest that synovial fibroblasts within inflamed joints produce icIL-1Ra1 (figures 3 and 4). However, in spite of the large amount of local icIL-1Ra1 detected in the joint, our results suggest that this isoform does not play any significant role in the control of acute inflammatory arthritis. However, our findings do not exclude the possibility that icIL-1Ra1 exerts a modulatory role during the later phase of arthritis, thus leading to the resolution of synovitis, as suggested by data obtained in the more chronic CIA model.36
In conclusion, our results indicate that myeloid cell-derived IL-1Ra, including sIL-1Ra and icIL-1Ra3 isoforms, exerts a critical regulatory role in the development of acute arthritis.
Acknowledgments
We thank Vanessa Bochet for expert technical assistance and Axel Finckh, MD (Division of Rheumatology, University Hospitals of Geneva, Switzerland) for statistical analysis.
References
Footnotes
-
Funding This work was supported by the Swiss National Science Foundation grants 310030-135195 (to CG) and 310030-134691 (to GP), the Rheumasearch Foundation,and the Institute of Arthritis Research.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.