Article Text
PDF
Abstract
Objectives Interleukin (IL)-38 is a newly characterised cytokine that belongs to the IL-1 family. This cytokine is expressed in the rheumatoid arthritis (RA) synovial tissue and IL-38 deficient mice have exacerbated arthritis. Here, we analysed the effect of IL-38 overexpression in the joints of arthritic mice, in human macrophages and synovial fibroblasts in vitro.
Methods Articular injections of an adeno-associated virus (AAV) 2/8 encoding IL-38 were performed in collagen-induced arthritis (CIA), K/BxN serum transfer-induced arthritis (STIA) and antigen-induced arthritis (AIA) in mice. The effect of IL-38 overexpression was evaluated through clinical scores, immunohistochemistry, microCT, Luminex and RT-qPCR analysis. THP-1 macrophages were transduced with a lentiviral vector to overexpress IL-38.
Results Clinical inflammatory scores were significantly decreased after AAV IL-38 injection in joints of mice with CIA and STIA, but not AIA. This decrease was accompanied by reduced macrophage infiltration and a decreased expression of Th17 cytokines (IL-17, IL-23, IL-22) and TNFα. However, IL-38 overexpression had no effect on cartilage or bone destruction. In vitro, the THP-1 monocytic cell line expressed less IL-6, TNFα and IL-23 after IL-38 overexpression. Conditioned media from these cells, containing released IL-38, also exert an anti-inflammatory effect on human primary macrophages and synovial fibroblasts from patients with RA.
Conclusions This study shows for the first time that IL-38 overexpression attenuates the severity of experimental arthritis. IL-38 may exert its anti-inflammatory effects by decreasing the production of proinflammatory cytokines by macrophages and synovial fibroblasts. This effect can lead to the development of novel treatment strategies in arthritis.
- Arthritis
- Cytokines
- Inflammation
Statistics from Altmetric.com
Introduction
The interleukin (IL)-38 gene was cloned and its product first characterised in 2001 as an IL-1 family member.1 IL-38 is a 17–18 kDa weight protein devoid of signal peptide or caspase-1 cleavage site. It was shown that apoptotic cells were able to produce a mature, truncated form of IL-38, but the exact cleavage site and the protease responsible for this maturation are not yet identified.2 Recently we have shown that IL-38 is expressed by keratinocytes, synovial fibroblast from patients with rheumatoid arthritis (RA), as well as by human monocytes and type I macrophages (M1) polarised in vitro.3 IL-38 gene polymorphisms are associated with RA, psoriatic arthritis, ankylosing spondylitis and heart disease.4–8 Moreover, patients with systemic lupus erythematosus, chronic hepatitis B, myocardial infarction or childhood asthma have higher levels of serum IL-38 than control populations.7–13 IL-38 levels are also increased in the synovial membrane and sera from patients with RA compared with healthy controls.13 ,14 We have previously shown that IL-38 mRNA levels are enhanced in synovial membranes of patients with RA and in colons of patients with Crohn's disease. In contrast, IL-38 expression was decreased in psoriatic skin.3 IL-38 protein levels were also enhanced in the synovial fluid of patients with RA.3
IL-38 could bind three different receptor chains, IL-1R1,1 IL-36R15 and the orphan receptor IL-1RAPL1.2 In peripheral blood mononuclear cells (PBMCs) stimulated with IL-36γ, IL-38 decreases IL-8 expression. Moreover, IL-38 is able to decrease IL-22 and IL-17 expression from Candida Albicans stimulated PBMCs.15 PBMCs from healthy donors treated with siRNA targeting IL-38 produced significantly more proinflammatory mediators after stimulation with toll-like receptors 7 and 9 ligands.9 IL-38 depletion also exacerbated the expression of IL-6 and IL-8 in cultured macrophages.2 Interestingly, full length recombinant IL-38 induced IL-6 production by macrophages, whereas truncated IL-38 decreased IL-6 expression after IL-1RAPL1 binding.2
In the collagen-induced arthritis (CIA) model in mice, IL-38 was mainly expressed during the resolution phase of inflammation suggesting that IL-38 exerts anti-inflammatory effects.3 Moreover, C57BL/6 mice deficient for IL-38 have a more severe K/BxN serum transfer-induced arthritis (STIA) together with a higher expression of IL-1β and IL-6 in the joints.14 Thus, IL-38 appears to be a potential negative regulator of inflammation during arthritis, but its exact biological effect in vitro and in vivo needs to be elucidated. Because the exact mature form of IL-38 is currently unknown, we designed here viral vectors (adeno-associated virus (AAV) and lentivirus) encoding the full length IL-38 cDNA. We investigated the effects of IL-38 overexpression in mice with CIA, K/BxN STIA, and antigen-induced arthritis (AIA), as well as in THP-1 macrophages in vitro.
Methods
Mouse models of arthritis
The full length mouse IL-38 cDNA (amino acid 1–152, NM_153077) was subcloned into the SSV9scCMV plasmid, under the control of the cytomegalovirus (CMV) promoter, to generate a double stranded AAV, serotype 2/8 (AAV IL-38). This recombinant virus was produced together with the control virus AAV GFP (enhanced Green Fluorescent Protein) in the gene therapy facility (Inserm UMR 1089, Nantes, France). This serotype is known to have a stable expression after intra-articular injection.16
For CIA, 8-week-old male DBA/1 mice were immunised as described previously.3 Each ankle was intra-articularly injected with 25×109 vector genomes (vg) of AAV at the first signs of inflammation. Each ankle was considered independently and, if the two ankles did not develop arthritis at the same time, the second ankle was included and injected only if arthritis appeared within the following 2 days. Inflammatory symptoms were assessed by a daily hind paw clinical scoring and mice were sacrificed around day 3 (peak phase) or day 11 (resolution phase). Other arthritis models are described in additional methods (K/BxN STIA and AIA models).
Histological and immunohistochemistry analysis
Ankles were fixed, decalcified and embedded.17 H&E, toluidine blue or tartrate-resistant acid phosphatase (TRAP) staining was performed as described.18
Immunostaining was performed as described previously19 ,20 with the following primary antibodies: goat anti-ionized calcium-binding adapter molecule 1 (Iba1; Abcam, Cambridge, UK), goat anti-CD3 (Abcam), rat anti-CD45R (BD Biosciences, San Jose, California, USA) rat anti-Ly6G (Abcam).
microCT analysis
Bone architecture was analysed using the high-resolution SkyScan-1076 X-ray microCT system (SkyScan, Kartuizersweg, Belgium).20
Cell culture
The full length human IL-38 cDNA (amino acid 1–152, NM_032556) was subcloned into the pWPI lentiviral plasmid under the control of the elongation factor 1-alpha (EF-1 alpha) promoter. pWPI was a gift from Didier Trono (Addgene plasmid # 12254). This bicistronic vector allows for simultaneous expression of the inserted transgene and an EGFP marker to facilitate tracking of transduced cells. The EGFP marker cDNA is inserted downstream of encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES). Lentivirus were produced in human embryonic kidney (HEK) 293T cells as described.21
THP-1 monocytic cells (American Type Culture Collection, Manassas, Virginia, USA) were transduced with lentiviral particles encoding IL-38-GFP or GFP alone using a multiplicity of infection of 5. The two cell lines obtained will be named THP-1-IL-38 and THP-1-CT. Parental cells are named THP-1-P. THP-1 were stimulated with phorbol 12-myristate 13-acetate (PMA; Sigma) at 0.1 µM during 72 hours to induce differentiation into active macrophages-like cells.22
M1 macrophages were obtained from PBMCs of five different healthy donors, RA synovial fibroblasts (RA-SF) from six different patients, as described.3
RT-qPCR analysis
Mouse tissue samples were homogenised using a DI25 Ultra-Turrax homogeniser (IKA, Staufen, Germany) in TriReagent (Molecular Research Center, Cincinnati, Ohio, USA). Cell cultures were scraped in TriReagent. The reverse transcription (RT)-qPCR was carried out as described.3 Analysis was performed using hypoxanthine-guanine phosphoribosyltransferase as invariant control and results were expressed as 2–ΔCt.
Western blot, luminex and ELISA
Mouse tissue samples were homogenised using a DI25 Ultra Turrax homogeniser in lysis buffer (Promega, Madison, Wisconsin, USA). Cells in culture were lysed in radioimmunoprecipitation assay (RIPA) buffer. Samples were loaded on precast 12% Bolt mini gels (Life technologies, Waltham, Massachusetts USA). Primary antibodies (anti-mIL-38, R&D Systems; anti-hIL-38, R&D Systems) were used to detect proteins of interest as described.20 ,21
ELISA tests for the detection of IL-6 and IL-23 (R&D Systems) and Luminex assays (Merck Millipore, Billerica, Massachusetts, USA) were performed according to the supplier's instructions.
Statistical analysis
Differences between groups were evaluated by the Mann-Whitney test using GraphPad Prism V.6 software (La Jolla, California, USA). p Values <0.05 (*) were considered statistically significant. In figure 1F, Student's t-test was used.
Effects of adeno-associated virus (AAV)-mediated interleukin (IL) 38 overexpression in collagen-induced arthritis (CIA). (A) enhanced Green Fluorescent Protein (eGFP) and IL-38 mRNA expression was assessed in the peak and resolution phases of arthritis in ankle extracts of CIA mice injected with AAV GFP (grey columns) or AAV IL-38 (black columns) (n=3 per group and per time point), CT, non-arthritic control, ND, not detectable. (B) IL-38 mRNA expression in AAV injected and non-injected ankles was compared with expression in inguinal lymph nodes, spleen or liver at the resolution phase of arthritis (n=3 per organ). mRNA expression is represented as 2−ΔCT relative to hypoxanthine-guanine phosphoribosyltransferase (HPRT). Results are shown as mean±SEM (A and B). (C) IL-38 protein expression in ankle extracts of CIA mice (n=3 per group) at the resolution phase of arthritis was detected by western blot. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as invariant control. (D) IL-38 protein expression in the serum of mice injected intra-articularly with AAV GFP or AAV IL-38 was analysed by western blot. n=3 for AAV GFP and n=5 for AAV IL-38. Recombinant mouse IL-38 (rmL-38) was used as a control. (E) Clinical scores for individual hind paws are shown after CIA induction in mice injected intra-articularly at the first signs of inflammation (day 0) with AAV GFP (grey symbols) or IL-38 (black symbols), (n=8 per group and per time point). This experiment is representative of four independent experiments with similar results. Results are shown as mean±SEM,*p<0.05, **p<0.01 compared with AAV GFP injected mice, as assessed by Mann-Whitney test. (F) Clinical scores for individual hind paws in CIA mice during the peak and resolution phases of arthritis were pooled from four independent experiments. AAV GFP injected mice are represented in grey and AAV IL-38 injected mice in black. Results are shown as box (lines for median, 25th and 75th centiles; crosses for mean) and whiskers (minimum and maximum values). p<0.0001 was assessed using Student's t-test.
Results
Local overexpression of IL-38 in AAV injected mice
As shown in figure 1A, GFP or IL-38 mRNAs were stably expressed in the joints of mice with established CIA after intra-articular injection of AAV GFP or IL-38. Moreover, IL-38 was overexpressed only locally in injected ankles and was present at much lower levels in inguinal lymph nodes, spleen and liver (figure 1B). At the protein level, IL-38 was present in ankles of mice with CIA injected with AAV IL-38 but was not detected in their sera (figure 1C, D). Taken together, these results showed that our AAV were able to induce stable overexpression of IL-38 locally in the joint of mice during arthritis, but not at the systemic level.
IL-38 overexpression decreases clinical score in CIA and K/BxN STIA
To investigate the impact of IL-38 overexpression on arthritis course, we injected AAV IL-38 or GFP (control group) into the ankle joints of mice with CIA at the onset of arthritis. The administration of AAV IL-38 significantly reduced the clinical score from the first day after injection. This anti-inflammatory effect was only partial (25%–45% of inhibition depending on the time point) and observed at the peak and resolution phases of arthritis (figure 1E, F). Early administration of AAV IL-38 1 day before the immunisation boost slightly reduced arthritis incidence at early time points but this effect did not reach statistical significance (see online supplementary figure S1A). When analysing ankles that developed arthritis, the early administration of AAV IL-38 significantly reduced the clinical inflammatory score from day 5 onwards (see online supplementary figure S1B).
supplementary data
In K/BxN STIA, the late administration of AAV IL-38 at the onset of arthritis also had an anti-inflammatory effect. This effect appeared independent of the mouse strain, since the same results were obtained in C57BL/6 (see online supplementary figure S1F) or DBA/1 (see online supplementary figure S1G) mice. In a third model of arthritis, the AIA model, injection of AAV IL-38 had no significant impact on knee diameter or clinical score (see online supplementary figure S1C, D) although IL-38 was strongly overexpressed locally in injected knees (see online supplementary figure S1E). Thus, late (see online supplementary figure S1C, D) and early (data not shown) AAV IL-38 injections had no significant impact in AIA while they significantly reduced inflammation in CIA and K/BxN STIA.
The anti-inflammatory effect of IL-38 overexpression is associated with a reduction in the monocyte/macrophage infiltrate
Following AAV IL-38 injection, the histological inflammatory score was significantly reduced during the resolution phase of inflammation (figure 2A, B). This effect was only local at the site of AAV injection (tarsus and tibiotalar joints), no effect was observed in metatarsophalangeal joints (see online supplementary figure S3). In contrast, no effect was observed on bone destruction and cartilage erosion at any time point (figure 2A). Similarly, IL-38 overexpression had no effect on TRAP+ osteoclasts in the pannus (figure 2A). MicroCT analysis confirmed that bone erosions were similar when mice were injected with AAV GFP or IL-38 (figure 3A, B).
Histology and IHC of ankle sections in collagen-induced arthritis (CIA) mice after adeno-associated virus (AAV) interleukin (IL) 38 injection. (A) Histological and IHC scores were evaluated during the peak and resolution phases of arthritis on ankle sections of AAV GFP (black dots) or AAV IL-38 (white dots) injected mice. Inflammatory, cartilage erosion and bone destruction scores were attributed for each tarsus on a scale from 0 to 3 (H&E, toluidine blue (TB)). Within the inflammatory synovial area (exclusion of bone, bone marrow and fibrous tissue): CD45R+ (B lymphocytes) and CD3+ (T lymphocytes) cells were counted manually and are presented as number per mm2; tartrate-resistant acid phosphatase (TRAP)+ (osteoclasts), ionized calcium-binding adapter molecule 1 (Iba1)+ (monocytes/macrophages) and Ly6G+ (granulocytes) surfaces were determined using ImageJ software (NIH, Bethesda, MD) and are presented as % of the region of interest. Results are shown as mean±SEM, *p<0.05 compared with AAV GFP injected mice, as assessed by Mann-Whitney test. n=5–8 for each group and time point. (B) Sections of CIA ankles injected with AAV GFP or AAV IL-38 were stained with H&E (upper panels) or with an anti-Iba1 antibody (lower panels). Representative images are shown. Scale bar=500 µm. Enlarged images correspond to the respective boxed areas.
MicroCT analysis of collagen-induced arthritis (CIA) mice after adeno-associated virus (AAV) interleukin (IL) 38 injection. (A) Three-dimensional reconstructions of paws from CIA mice using the Skyscan CtVol software. Representative non-arthritic and non-injected mice (CT) and arthritic AAV GFP or AAV IL-38 injected paws are shown. (B) Bone destruction scores were attributed from the microCT analysis (erosion score between 0 and 4 for the tibia, talus and navicular joints according to median, medial and lateral plans for a total score of 36), n=7 per group, results are shown as mean±SEM.
Within the inflammatory synovial area, granulocytes (Ly6G+), B lymphocytes (CD45R+) and T lymphocytes (CD3+) were not affected by IL-38 overexpression (figure 2A). However, the density of Iba1+ monocytes/macrophages was significantly reduced during the resolution phase of inflammation (figure 2A, B). In addition, there was no difference in the serum titres of anticollagen type II antibodies when IL-38 was overexpressed (see online supplementary figure S2).
Thus, the anti-inflammatory effect of IL-38 overexpression was mainly associated with a decrease of the macrophage infiltrate.
IL-38 overexpression reduces Th17 cytokine expression
We analysed mRNA expression of major inflammatory cytokines. Tumor necrosis factor alpha (TNFa) expression was not altered during CIA, but IL-1β and IL-6 mRNA levels were induced at the peak of inflammation (day 3) and IL-38 overexpression did not reduce their expression in inflamed joints. Interestingly, IL-38 overexpression significantly reduced expression of Th17 cytokines (IL-17A, IL-23p19, IL-22), chemokine (C-X-C motif) ligand 1 (CXCL1) (neutrophil chemoattractant) and receptor activator of nuclear factor kappa-B ligand (RANKL) (osteoclastic cytokine) but not osteoprotegerin (OPG) (RANKL decoy receptor) (figure 4A). IL-38 overexpression had variable effects on expression of IL-36 cytokines (IL-36α, β, γ and Ra) or Th1 cytokines (interferon gamma (IFNγ) and IL-12p35), and did not induce anti-inflammatory cytokines such as IL-10, IL-4 or transforming growth factor beta (TGFβ) (see online supplementary figure S4A). Rather, IL-38 overexpression reduced IL-10 mRNA levels (see online supplementary figure S4A). At the serum protein level, IL-6 expression was not altered in AAV IL-38 injected mice but TNFα, IL-22 and IL-10 expressions were significantly reduced (see figure 4B and online supplementary figure S4B). IL-23 and IL-17A serum levels were below the detection limits.
Cytokines expression in paws and serum of arthritic mice after adeno-associated virus (AAV) interleukin (IL) 38 injections. (A) mRNA expression of various inflammatory mediators was assessed in ankles from AAV GFP (black square) or AAV IL-38 (white square) injected mice with collagen-induced arthritis (CIA) at the peak and resolution phases of arthritis, and in non-arthritic and non-injected mice (CT). The results represent IL-1β, tumor necrosis factor alpha (TNFa), IL-6, IL-17A, IL-23p19, IL-22, chemokine (C-X-C motif) ligand 1 (CXCL1), receptor activator of nuclear factor kappa-B ligand (RANKL) and osteoprotegerin (OPG) mRNA expression relative to hypoxanthine-guanine phosphoribosyltransferase (HPRT), shown as mean±SEM. *p<0.05 compared with AAV GFP injected mice, as assessed by Mann-Whitney test. (B) IL-6, TNFα and IL-22 protein expression were assessed by Luminex assay in serum of AAV GFP (black dots) or AAV IL-38 (white dots) injected mice with CIA at the peak and resolution phases of arthritis. *p<0.05 compared with AAV GFP injected mice, as assessed by Mann-Whitney test.
IL-38 overexpression in THP-1 cells also triggers an anti-inflammatory effect
Together, the results obtained in mouse models of arthritis suggested that IL-38 overexpression could mainly impact macrophage infiltration and their production of key cytokines and chemokines implicated in the Th17 pathway. Thus, we next analysed whether IL-38 overexpression in the THP-1 monocytic cell line in vitro also impacted its cytokine production. THP-1 cells were transduced with lentiviral particles to overexpress human IL-38 (figure 5A, B). Overexpression of IL-38 in THP-1 cells significantly reduced IL-6 and IL-23p19 mRNA expression after lipopolysaccharide (LPS) stimulation (see online supplementary figure S5A). IL-38 overexpression also significantly reduced LPS-induced IL-6, TNFα, IL-23 and IL-10 protein secretion by 2–2.5-fold (figure 5C, D) but IL-1β was not affected (figure 5D).
Effects of lentiviral interleukin (IL) 38 overexpression in THP-1 cells. (A) IL-38 mRNA expression was measured in THP-1-CT and THP-1-IL-38 cells differentiated with phorbol 12-myristate 13-acetate (PMA). mRNA expression is represented as 2−ΔCT relative to hypoxanthine-guanine phosphoribosyltransferase (HPRT) and results are shown as mean±SEM. (B) IL-38 protein expression in PMA-differentiated THP-1 cells was assessed by western blot. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (C) IL-6 secretion was assessed by ELISA in culture supernatants of PMA-differentiated THP-1-CT (grey columns) or THP-1-IL-38 (black columns) cells stimulated for 24 hours with lipopolysaccharide (LPS) (0.1; 1; 10; 100 ng/mL). (D) TNFα, IL-23, IL-1β and IL-10 expressions were assessed by Luminex assay in culture supernatants of PMA-differentiated THP-1-CT (grey columns) or THP-1-IL-38 (black columns) cells stimulated for 24 hours with LPS at 100 ng/mL. Results are shown as mean±SEM.*p<0.05 compared with THP-1-CT in the same conditions as assessed by Mann-Whitney test (C and D).
Conditioned media from IL-38-transduced cells contain biologically active IL-38
We next asked whether IL-38 could be released from lentivirally transduced THP-1 cells to exert anti-inflammatory effects. IL-38 (17–18 kDa) was detected in the culture supernatants of THP-1-IL-38 cells, but not of THP-1-CT cells (figure 6A). Release of IL-38 was not modified by macrophage differentiation induced by phorbol 12-myristate 13-acetate (PMA) or LPS stimulation (data not shown).
Effects of THP-1 conditioned media containing interleukin (IL) 38 on different cell types. (A) The release of IL-38 into the culture supernatant by PMA-differentiated THP-1-CT and -IL-38 cells after 24, 48 or 72 hours and by human embryonic kidney (HEK)-CT and HEK-IL-38 cells after 24 hours was detected by western blot. (B) IL-6, IL-23, TNFα and IL-1β secretion was assessed by Luminex assay in culture supernatants of parental THP-1 (THP-1-P) cells incubated in presence or absence of LPS (10 ng/mL) and in presence of conditioned media from THP-1-CT or THP-1-IL-38 cells (dilution 1:4) for 24 hours. ND, not detectable. (C) IL-6 and IL-23 secretions were assessed by ELISA in culture supernatants of M1 macrophages differentiated from CD14+ monocytes of five different healthy donors (IL-6) or three different donors (IL-23) and incubated in presence or absence of LPS (10 ng/mL) and in presence of conditioned media from THP-1-CT or THP-1-IL-38 cells (dilution 1:1) for 24 hours. (D) IL-6 secretion was assessed by ELISA in culture supernatants of rheumatoid arthritis synovial fibroblasts (RA-SF) of six different patients incubated in presence or absence of IL-1β (100 ng/mL) and in presence of conditioned media from THP-1-CT or THP-1-IL-38 cells (dilution 1:1) for 48 hours. ELISA results are shown as mean±SEM. * p<0.05 compared with control conditioned media as assessed by Mann-Whitney test (B). a p<0.05, b p<0.01 compared with control conditioned media as assessed by the paired Mann-Whitney-Wilcoxon test (C-D). (E) Proposed model of IL-38 action in arthritic joint. Overexpressed IL-38 reduces the production of several proinflammatory cytokines on macrophages and synovial fibroblasts. This could lead to reduced generation of Th17 cells and thus reduced level of IL-17, IL-22, RANKL, and so on. This inhibition of the Th17 axis would limit osteoclast formation and bone resorption (indicated in black). However, overexpressed IL-38 does not impact in vivo level of pro-osteoclastogenic IL-1β and IL-6, reduced the antiosteoclastogenic cytokine IL-10 and could also have direct effects on osteoclasts (indicated in red). Overall, IL-38 overexpression does not affect bone destruction.
(Continued)
Conditioned media from THP-1-IL-38 cells significantly reduced IL-6, TNFα and IL-23 secretion of THP-1-P stimulated with LPS but IL-1β secretion was not altered (figure 6B). Moreover, conditioned media from the epithelial HEK cell line transduced to overexpress IL-38 also reduced IL-6 secretion by THP-1-P cells (see online supplementary figure S5B).
These results indicated that overexpressed IL-38 was efficiently released by different cell types and that part of the anti-inflammatory effect of IL-38 could be autocrine or paracrine. To expand these results to cells more relevant to RA pathogenesis, we next treated M1 macrophages from healthy donors and RA-SF in primary culture with IL-38-containing conditioned media. Again, the addition of conditioned media from THP-1-IL-38 cells reduced the secretion of IL-6 and IL-23 by LPS stimulated M1 macrophages (figure 6C), as well as IL-6 secretion by RA-SF stimulated with IL-1β (figure 6D). IL-23 expression was below the detection limit for two M1 macrophage donors. In RA-SF, only the IL-23p19 subunit is expressed and total IL-23 (p19+p40) could not be detected.23
Discussion
Previously, it has been shown that IL-38 deficient mice exhibit more severe STIA than wild type mice.14 Our present study expands these results and demonstrates for the first time that AAV-mediated IL-38 overexpression exerted moderate but significant anti-inflammatory effects in mice with CIA or STIA. AAV IL-38 appeared to have similar effects after late or early administration and to act only locally on the inflammatory infiltrate. This effect was associated with a decreased number of Iba1+ monocytes/macrophages in the synovial tissue, whereas other immune cell populations were not significantly altered. The levels of endogenous IL-38 were increased during the resolution of acute inflammation3 as previously shown for IL-1Ra,24 thus suggesting that IL-38 may act naturally as an anti-inflammatory cytokine.
In addition to the reduced macrophage number, a significant decrease in the expression of Th17 cytokines (IL-17, IL-22, IL-23), TNFα and CXCL1 was observed in mice with CIA injected with AAV IL-38. IL-17 and IL-22 are mainly produced by Th17 cells, while IL-23 is produced by macrophages and induce the polarisation and maintenance of Th17 cells.25 IL-17 also induces the expression of proinflammatory cytokines such as TNFα, IL-1β or IL-6 and of chemokines such as CXCL1 or CCL20.26 Of note, IL-38 overexpression did not induce the production of other anti-inflammatory cytokines but reduced significantly IL-10 expression. Our in vitro experiments indicate that IL-38 inhibited the production of cytokines such as IL-6, TNFα and IL-23 by THP-1 cells, M1 macrophages and RA-SF but did not alter IL-1β expression. These effects could indirectly influence the expression of Th17 cytokines (figure 6E). Overall, these results are in accordance with previous studies showing that IL-38 restricts macrophage-dependent generation of Th17 cells2 and that TNFα and IL-6 are not regulated as IL-1β in macrophages.27 The fact that IL-38 overexpression does not reduce IL-6 expression in CIA mice could be explained by the numerous different cell types that produce this pleiotropic cytokine in vivo, some of which being possibly not responsive to IL-38. Moreover, IL-38 overexpression did not affect cartilage and bone destruction even if RANKL expression was reduced after AAV IL-38 injection in mice with CIA. This could be explained by the very destructive character of the CIA model and the moderate effect of full length IL-38 overexpression. Furthermore, other cytokines whose global expression is not altered in vivo by IL-38 overexpression, such as IL-1β or IL-6, are known to sustain cartilage and bone destruction during arthritis.28 Reduced expression of the antiosteoclastogenic cytokine IL-10 by IL-38 would also sustain osteolysis29 and we cannot rule out a direct effect of IL-38 on osteoclast formation or on cartilage destruction (figure 6E). Considering IL-38 as an anti-inflammatory cytokine, these contrasting and counterintuitive effects now need to be better dissected.
IL-38 overexpression also reduced clinical scores in STIA. Interestingly, the pathogenesis of STIA is independent of B cells and T cells,30 which is consistent with the role of IL-38 on innate immune cells such as macrophages. In contrast, IL-38 overexpression had no effect in AIA, an acute and monoarticular model of arthritis.31 Although we have no definitive explanation for this finding, IL-1Ra is also not efficient in reducing the severity of AIA, in contrast to CIA,32 indicating that mouse models of arthritis have variable dependency on IL-1 family members.
It remains unclear if IL-38 needs to be processed in order to have a full biological activity. We confirmed that full length recombinant IL-38 had no anti-inflammatory effect on multiple cell types such as RA-SF, THP-1 or macrophages (data not shown). It was recently demonstrated that its maturation in apoptotic cells (cancer cells, macrophages, neutrophils) allowed to potentiate its anti-inflammatory action, but the exact cleavage site and the proteases involved are unknown.2 In the same study, the authors used a truncated form of IL-38 designed based on in silico prediction of potential cleavage site (AA 20–152, from Adipogen, Switzerland).33 In contrast to full length IL-38, this truncated form of IL-38 proved to have an anti-inflammatory effect, especially on macrophages.2 Unfortunately, truncated recombinant IL-38 was not commercially available for our experiments. Here, THP-1 macrophages and HEK epithelial cells overexpressing full length IL-38 release large amounts of biologically active IL-38 (μg/mL range). This form of IL-38 comigrates with recombinant full length IL-38 at an apparent molecular weight of 17–18 kDa. However, it remains possible that a minor truncated form of IL-38 is produced, but not easily detected by western blot, especially if only a few amino acids are removed.2
It has been shown that IL-38 exerts anti-inflammatory effects via binding to IL-36R and neutralisation of IL-36 cytokine signalling.15 However, subsequent studies rather indicated that IL-38 binds to IL-1RAPL1 to limit cytokine production in a broader inflammatory context, such as LPS stimulation of macrophages.2 Here we showed that IL-38 present in conditioned media reduced the production of inflammatory cytokines by LPS treated macrophages and IL-1β stimulated RA-SF. These findings further indicate that the effects of IL-38 are not specifically limited to IL-36 inhibition.
This study is the first to highlight the anti-inflammatory effect of IL-38 overexpression in vivo. This cytokine appears to act on macrophages in the inflammatory infiltrate, reducing the expression of proinflammatory cytokines with potent indirect effects on Th17 cytokines. Whether IL-38 should be considered as a new therapeutic option in arthritis or other inflammatory diseases deserves further experiments.
References
Footnotes
Handling editor Tore K Kvien
FB and BLG are senior coauthors.
Contributors M-AB performed mouse arthritis models, in vitro study and associated analysis and wrote the manuscript, AN did some in vitro experiments and GB performed some histology and microCT analysis. RB contributed to Luminex experiments. ST provided synovial biopsies. VT contributed to AAV design. PL contributed to study design. CG and GP provided K/BxN serum, lentiviral plasmids and contributed to the elaboration of the manuscript. BLG performed mouse arthritis models, planned studies, analysed data and wrote the manuscript. FB planned studies, analysed data and wrote the manuscript. All authors read and approved the final manuscript.
Funding This work was supported by Inserm and in part by the Arthritis Foundation and by the French Society of Rheumatology. M-AB was a recipient from a fellowship from the French Ministry of Research. CG is supported by grants from the Swiss National Science Foundation (310030_152638), the Rheumasearch Foundation, the Uniscientia Foundation and the Institute of Arthritis Research.
Competing interests None declared.
Ethics approval Local ethics committee of the Nantes Hospitals and the French Research Ministry (no. 2008-402). All research involving animals is conducted following institutional guidelines and has been approved by the French ethical committee CEEA.PdL (License 2012.86) and by local veterinary services.
Provenance and peer review Not commissioned; externally peer reviewed.