Proinflammatory role of the Th17 cytokine interleukin-22 in collagen-induced arthritis in C57BL/6 mice

Authors


Abstract

Objective

To investigate the role of interleukin-22 (IL-22) in collagen-induced arthritis (CIA), an animal model of rheumatoid arthritis.

Methods

C57BL/6 mice were immunized with type II collagen (CII) in Freund's incomplete adjuvant with added Mycobacterium tuberculosis, and levels of IL-22 and its specific receptor, IL-22 receptor type I (IL-22RI), were measured in sera and tissue by enzyme-linked immunosorbent assay and real-time quantitative polymerase chain reaction analysis. Clinical and histologic signs of arthritis were recorded and compared with those in C57BL/6 mice deficient in the IL-22 gene (IL-22–/–). Humoral and cellular immune responses against CII were analyzed. In vitro osteoclastogenesis assays were performed on splenocytes.

Results

Upon immunization with CII in Freund's incomplete adjuvant plus heat-killed Mycobacterium tuberculosis, sera from C57BL/6 mice were found to contain high levels of IL-22, and the specific IL-22RI was expressed in lymphoid tissue, including splenocytes. IL-22–/– mice were less susceptible to CIA than were wild-type mice, as evidenced by their decreased incidence of arthritis and decreased pannus formation. Remarkably, the less severe form of arthritis in IL-22–/– mice was associated with increased production of CII-specific and total IgG antibodies, whereas cellular CII responses were unchanged. In vitro, IL-22 was found to promote osteoclastogenesis, a process that might contribute to its proinflammatory activity in CIA.

Conclusion

Endogenous IL-22 plays a proinflammatory role in CIA in C57BL/6 mice. Our data also indicate that IL-22 promotes osteoclastogenesis and regulates antibody production.

Interleukin-22 (IL-22), a cytokine belonging to the IL-10 superfamily (1), has been demonstrated to act as an effector cytokine of the Th17 lineage (2). IL-22 is primarily produced by activated T cells and natural killer cells (3), and IL-22 signal transduction is mediated through a receptor complex consisting of the specific IL-22 receptor type I (IL-22RI) and the common IL-10R2 subunits (1, 4). IL-22 acts on various cell types, including hepatocytes, epithelial cells, keratinocytes, and fibroblasts, and induces an acute-phase response in vivo and chemokines and matrix metalloproteinases (MMPs) in vitro (3).

Upon intradermal injection in mice, IL-22 induces the expression of proinflammatory cytokines (3). In addition, increased levels of IL-22 have been demonstrated in patients with rheumatoid arthritis (RA) (5) and in patients with psoriasis (3), and the levels were correlated with disease severity, indicating the importance of IL-22 in these autoimmune diseases. Targeting of IL-22 in vivo in animal models with the use of neutralizing anti–IL-22 antibodies demonstrated the importance of IL-22 in the host defense against bacterial infections (6) and in psoriasis (7). IL-22–/– mice were found to be susceptible to experimental autoimmune encephalomyelitis (8), indicating that IL-22 is not directly involved in the pathogenesis of this condition.

In the present study, we investigated the role of IL-22 in collagen-induced arthritis (CIA), an animal model of RA, using C57BL/6 mice with a deletion in the IL-22 gene (IL-22–/–) (8) and their wild-type (WT) counterparts. Although it was initially believed that the induction of CIA was restricted to mice bearing the class II major histocompatibility complex H-2q or H-2r, but not H-2b, gene (9), it was recently demonstrated that C57BL/6 mice of the H-2b background are susceptible to arthritis induction by chicken type II collagen (CII) through an increase in either the concentration of mycobacteria in Freund's incomplete adjuvant or the dose of CII (9, 10). We first verified whether IL-22 and IL-22RI are expressed in immunized mice. The endogenous role of IL-22 in CIA was then investigated by comparing the arthritis symptoms in IL-22–/– and WT mice and by analyzing the humoral and cellular autoimmune responses and osteoclastogenesis.

MATERIALS AND METHODS

Mice.

IL-22–/– mice, backcrossed for 16 generations on a C57BL/6 strain, were generated by homologous recombination as described previously (8). Mice were bred in the animal facility of the Ludwig Institute for Cancer Research under specific pathogen–free conditions. Experiments were performed on age-matched female mice at the animal facility of Leuven University. Studies were approved by the Local Ethics Committee.

Induction of CIA and histologic assessments.

For the induction of CIA, we followed the procedure described by Campbell et al (9). Mice were examined daily beginning on day 15 for clinical symptoms of arthritis, and a mean arthritis score was calculated as described previously (11). Histologic features of arthritis were scored as described previously (11).

Detection of IL-22, CII-specific and nonspecific antibodies, and delayed-type hypersensitivity (DTH).

IL-22 levels were measured by enzyme-linked immunosorbent assay (ELISA) using a kit from Antigenix America (Huntington Station, NY). CII-specific (11) and nonspecifc antibodies were detected according the manufacturer's instruction (Jackson ImmunoResearch, De Pinte, Belgium). DTH was recorded as the percentage of swelling upon injection of CII into the ear (11).

In vitro osteoclastogenesis, tartrate-resistant acid phosphatase (TRAP) staining, and pit-forming assay.

In vitro osteoclastogenesis was performed by stimulating splenocytes with macrophage colony-stimulating factor (M-CSF; 20 ng/ml) plus RANKL (100 ng/ml) in the presence or absence of recombinant murine IL-22 or 4% supernatants from HEK 293 cells that had been transfected with mouse IL-22 complementary DNA (cDNA) or empty vector (12). On day 6, cells were stained for the presence of TRAP (11). Osteoclasts were identified as TRAP+ multinucleated (≥3 nuclei) cells. The pit-forming assay on quartz substrates coated with a calcium phosphate film was performed as described previously (13).

Real-time quantitative polymerase chain reaction (PCR) analysis.

RNA extraction, cDNA synthesis, and real-time quantitative PCR for IL-22, IL-22RI, IL-1β, IL-6, tumor necrosis factor α (TNFα), MMP-9, and IL-17 (assay nos. Mm00444241_m1, Mm00663697_m1, Mm00434228_m1, Mm00446190_m1, Mm00443258_m1, Mm00600163_m1, and Mm00439619_m1, respectively; Applied Biosystems, Lennik, Belgium) were performed as described elsewhere (13).

Statistical analysis.

Differences between groups were analyzed by Mann-Whitney U test. P values less than 0.05 were considered significant.

RESULTS

Detection of IL-22 and IL-22RI in immunized C57BL/6 mice.

First, we investigated whether IL-22 and IL-22RI were expressed in arthritic mice. Since symptoms of arthritis in C57BL/6 mice typically appear between day 24 and day 28 after induction (9, 10), this time point was chosen for analysis. No detectable levels of IL-22 were found in the sera of naive or immunized mice. Therefore, in a subsequent experiment, mice were challenged with anti-CD3 antibody. IL-22 levels were found to be higher in sera from immunized mice as compared with sera from naive animals (Figure 1A).

Figure 1.

Increased levels of interleukin-22 (IL-22) after immunization, and detection of IL-22 receptor type I (IL-22RI) in C57BL/6 mice that were either not immunized (naive [NI]) or were immunized with type II collagen (CII) in Freund's incomplete adjuvant plus heat-killed Mycobacterium tuberculosis, to produce collagen-induced arthritis (CIA). A, On day 28 postimmunization, mice were challenged with 10 μg of anti-CD3 antibody, and serum IL-22 levels were determined 5 hours later by enzyme-linked immunosorbent assay (ELISA). Values are the mean and SEM of 6 mice per group. B, Lymphocytes were isolated from naive and immunized mice and were stimulated in vitro with medium (med) alone, anti-CD3 (aCD3; 1 μg/ml), IL-23 (10 ng/ml), Mycobacterium tuberculosis (MTb; 100 μg/ml), or chicken CII (cCII; 100 μg/ml). Supernatants were collected after 48 hours and tested for the presence of IL-22 by ELISA. Vertical broken line separates data from naive and CIA mice. Values are the mean and SEM of 5 mice per group. ∗ = P < 0.05 versus medium alone, by Mann-Whitney U test. ND = not detectable. C, Real-time quantitative polymerase chain reaction analysis was performed on cDNA derived from the isolated organs of immunized mice (day 27 postimmunization). Vertical broken line separates data for the left and right y axes. Values are the relative quantities of IL-22RI mRNA normalized to the quantity of 18S RNA.

Lymphocytes from naive and immunized mice were then stimulated in vitro with anti-CD3 and IL-23, which are known triggers of IL-22 (2), as well as with Mycobacterium tuberculosis or CII, and supernatants were analyzed for the presence of IL-22. No IL-22 was detectable in lymphocytes from naive mice or in unstimulated lymphocytes from immunized mice (Figure 1B). Stimulation with anti-CD3 or IL-23 resulted in the production of IL-22. Stimulation with M tuberculosis, but not CII, elicited production of IL-22 (Figure 1B).

To investigate the spectrum of tissues that might respond to IL-22, we surveyed IL-22RI expression in various organs of immunized mice by real-time quantitative PCR. Apart from the skin, where levels of IL-22RI messenger RNA (mRNA) were high, considerable amounts were found in the liver, kidney, and lung (Figure 1C). Low levels of IL-22RI mRNA were present in the thymus, lymph nodes, splenocytes, pancreas, and synovium (Figure 1C).

Reduced incidence of CIA in IL-22–/– mice.

To analyze the role of IL-22 in CIA, IL-22–/– and WT C57BL/6 mice were immunized and examined. Five of the 9 WT, but only 2 of the 8 IL-22–/– mice, developed arthritis (Figure 2A), resulting in a significantly lower mean arthritis score in IL-22–/– mice (mean ± SEM 2.1 ± 0.9 in WT mice versus 0.3 ± 0.3 in knockout mice; P < 0.05). When the analysis was restricted to the mice with CIA, the severity of arthritis was still lower in the IL-22–/– mice, although the difference was not significant (Figure 2B). These results were confirmed in 2 independent experiments (incidence 20 of 33 WT mice versus 9 of 32 IL-22–/– mice). In one of these experiments, mice were killed for histologic evaluation of the joints. Scoring the severity of arthritis in mice with CIA showed that the reduced arthritis severity in IL-22–/– mice was associated with a significantly lower degree of pannus formation (Figure 2C). Furthermore, lower numbers of mRNA copies of IL-1β, IL-6, TNFα, and MMP-9 were found in pooled synovium samples from immunized IL-22–/– mice (Figure 2D). IL-17 mRNA was present in the synovium of immunized WT mice but was undetectable in the synovium of IL-22–/– mice (data not shown).

Figure 2.

Reduced incidence of collagen-induced arthritis (CIA) in interleukin-22–knockout (IL-22–/–) mice. CIA was induced in wild-type (WT) and IL-22–/– mice according to the method described by Campbell et al (9). Results shown are from 1 of 3 independent experiments that were performed. A and B, Mice were examined for symptoms of arthritis beginning on day 15, and the cumulative incidence of arthritis (A) and mean arthritis scores (B) in the arthritic mice over the course of the experiment were determined (maximum score 16 per mouse). Values in B are the mean and SEM. C, On day 36 postimmunization, histologic analysis of the joints was performed. The average histologic scores for infiltration (inf), hyperplasia (hyp), and pannus formation and bone destruction (pan) in both front limbs and both hind limbs from 4 arthritic mice were calculated. Values are the mean and SEM. * = P < 0.05 versus WT mice. D, Synovium samples were obtained from immunized WT and IL-22–/– mice (day 27 after immunization), and real-time quantitative polymerase chain reaction analysis was performed on cDNA pooled from 3 immunized WT and 3 immunized IL-22–/– mice. Values are the relative quantities of interleukin-1β (IL-1β), IL-6, tumor necrosis factor α (TNFα), and matrix metalloproteinase 9 (MMP-9) mRNA normalized to the quantity of 18S RNA.

Association of the reduced incidence of arthritis in IL-22–/– mice with increased humoral, but similar cellular, responses to CII.

To verify whether the less-severe arthritis in IL-22–/– mice resulted from modulation of the humoral and/or cellular immune responses against CII, sera from immunized WT and IL-22–/– mice were analyzed for the presence of CII-specific total IgG, IgG1, IgG2b, and IgG2c antibody isotypes. Significantly higher levels of anti-CII total IgG antibodies were present in the sera of IL-22–/– mice as compared with WT mice (Figure 3A), a finding that could be explained by the significantly higher levels of the IgG2c isotype that were present (Figure 3B). Levels of non–CII-specific IgG were also significantly higher in the sera of IL-22–/– mice as compared with WT mice (mean ± SEM 30.1 ± 5.1 mg/ml versus 22.1 ± 1.4 mg/ml; P < 0.05).

Figure 3.

Changes in humoral and cellular immune responses in wild-type (WT) and interleukin-22–knockout (IL-22–/–) mice with collagen-induced arthritis (CIA). A and B, Sera from mice with CIA were obtained on day 28 after immunization and analyzed for total IgG antibody (A) and IgG1, IgG2b, and IgG2c antibody isotypes (B) directed against type II collagen (CII). Values are the mean and SEM. ∗ = P < 0.05 versus WT mice, by Mann-Whitney U test. Delta OD = change in optical density. C, Delayed-type hypersensitivity responses to CII injection in the ear were measured as the percentage of ear swelling and determined at the indicated time points after injection. Values are the mean and SEM. D, Splenocytes were isolated on day 24 from immunized WT mice and were cultured for 6 days in medium containing 20 ng/ml of macrophage colony-stimulating factor plus 100 ng/ml of RANKL in the presence of no stimulus (control [con]), 100 ng/ml of IL-22, or 4% supernatant from HEK 293+ or HEK 293− cells to determine osteoclastogenesis. Cultures were fixed and stained for tartrate-resistant acid phosphatase (TRAP), and the numbers of TRAP+ multinucleated (≥3 nuclei) cells in each cup were counted and identified as osteoclasts. Values are the mean and SEM of quadruplicate cultures. ∗ = P < 0.05 versus control, by Mann-Whitney U test. E, TRAP-stained cultures of splenocytes were treated as described in D and showed significantly higher numbers of TRAP+ multinucleated osteoclasts after stimulation with IL-22. Representative photomicrographs are shown (original magnification × 20). F, A pit-forming assay of splenocytes was performed using the same stimuli as described in D. Resorption pits (white areas) on quartz substrates coated with a calcium phosphate film were increased in number and size after stimulation with IL-22. Representative photomicrographs are shown (original magnification × 20).

To measure the cellular immune response against CII, DTH analysis was performed. The lower susceptibility of IL-22–/– mice to CIA was not associated with a decreased cellular immune responsive to CII (Figure 3C).

IL-22–induced increase in osteoclastogenesis by splenocytes from immunized WT mice.

Our data shown in Figure 1C demonstrated detectable expression of IL-22RI on splenocytes from immunized WT mice. Analysis of enriched splenocyte populations by real-time quantitative PCR showed that the presence of IL-22RI was attributable to the CD11b+ fraction (data not shown), which contains osteoclast precursors (13).

Subsequently, we verified whether the decreased pannus formation and bone destruction in the IL-22–/– mice could be explained by a stimulatory effect of IL-22 on osteoclastogenesis. Splenocytes from immunized WT mice were stimulated with M-CSF plus RANKL in the presence or absence of IL-22 or with supernatants from mammalian HEK 293 cells that had been transfected with mouse IL-22 cDNA or empty vector. Significantly higher numbers of TRAP+ multinucleated osteoclasts were observed in cultures of IL-22–stimulated splenocytes (Figure 3D). Figure 3E shows representative photomicrographs of the splenocyte/osteoclast cultures.

To verify osteoclast activity, a pit-forming assay was performed using the same stimuli. Increased numbers of pits and increased surface area were found in the IL-22–stimulated cultures (Figure 3F).

DISCUSSION

IL-22, a member of the IL-10 cytokine superfamily, is produced by Th17 cells (2). Since its expression has been demonstrated in RA (5), we verified whether IL-22 contributes to the development of CIA. IL-22 was found to be up-regulated upon immunization of WT mice, suggesting a role of IL-22 in CIA. Through stimulation of lymphocytes with anti-CD3 and IL-23, both of which are inducers of IL-22 (2), the production of IL-22 was evident. M tuberculosis triggered the production of IL-22, as potently as anti-CD3. Since anti-CD3 and M tuberculosis failed to induce IL-22 in naive mice, specific adaptive immune cells are probably the source of IL-22.

The IL-22R complex consists of IL-22RI paired with IL-10R2 (1). Using real-time quantitative PCR analysis, high levels of IL-22RI were detected in the skin, whereas moderate levels were detected in nonimmune organs, thus confirming previous findings (1). Low levels of IL-22RI mRNA were also present in immune organs (thymus, lymph nodes, and splenocytes) from the immunized mice. IL-22RI was detected in the synovium, a finding consistent with that from a study by Ikeuchi et al (5) in which the presence of IL-22RI was identified in synovial tissue from RA patients.

To assess the importance of IL-22 in CIA, we studied IL-22–/– mice. We confirmed the susceptibility of C57BL/6 WT mice to CIA (9, 10). IL-22–/– mice were found to be less susceptible (not statistically different), as evidenced by their lower incidence of arthritis and decreased arthritis scores. The lower degree of inflammation in IL-22–/– mice was confirmed by histologic features and by the finding of lower levels of mRNA for IL-1β, IL-6, TNFα, MMP-9, and IL-17 in the synovium. When considering only the arthritic mice, no significant differences in clinical arthritis scores were observed, a finding that could be explained by the similar extent of infiltration of mononuclear and polymorphonuclear immunocompetent cells in their synovium samples. These data indicate that IL-22 is not involved in the chemotaxis of immunocompetent cells in the joint. Histologic analysis of arthritic mice revealed less severe synovial tissue hyperplasia and significantly less pannus formation in IL-22–/– mice as compared with their WT counterparts. Interestingly, Ikeuchi et al (5) reported that IL-22 was a factor in the proliferation of synovial fibroblasts obtained from RA patients.

In order to account for the decreased pannus formation and bone destruction in IL-22–/– mice, we evaluated the role of IL-22 on osteoclastogenesis. Since we had detected IL-22RI in the CD11b+ fraction of splenocytes, a known osteoclast precursor population (13), we determined the effect of IL-22 on osteoclastogenesis in vitro. Significantly higher numbers of osteoclasts were observed in IL-22–stimulated splenocyte cultures. Since TRAP staining is not a specific marker of osteoclasts (14), we also performed a pit-forming assay using the same stimuli. Increased pits and surface area were found in IL-22–stimulated cultures. These data demonstrate the stimulatory role of IL-22 in osteoclast differentiation and activity.

The pathogenesis of CIA is considered to be dependent on humoral and cellular immune responses to CII. The lower susceptibility of IL-22–/– mice to CIA was not associated with differences in cellular immune responsiveness to CII. An intriguing observation was the increased anti-CII antibody response in IL-22–/– mice despite their decreased susceptibility. Anti-CII antibodies of the IgG2a isotype could not be measured, since the gene that encodes IgG2a is deleted in C57BL/6 mice (15). Instead, C57BL/6 mice produce antibodies of the IgG2c isotype. In our experiments, differences in anti-CII total IgG were explained by significantly higher levels of IgG2c in the sera of IL-22–/– mice. The mechanism by which IL-22 promotes IgG2c production needs further investigations, but it probably results from an indirect effect, since B lymphocytes do not express IL-22RI (3). Importantly, the increase in antibody production seen in IL-22–/– mice was not CII-specific, since levels of total IgG were also higher in the sera of IL-22–/– mice.

In conclusion, our study provides evidence of a proinflammatory role of endogenous IL-22 in CIA. Our findings also indicate that IL-22 promotes osteoclastogenesis and regulates antibody production and suggest that this field might be an interesting area for further investigation.

AUTHOR CONTRIBUTIONS

Dr. Geboes had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Geboes, Renauld, Matthys.

Acquisition of data. Geboes, Dumoutier, Kelchtermans, Schurgers, Mitera.

Analysis and interpretation of data. Geboes, Dumoutier, Kelchtermans, Schurgers, Renauld, Matthys.

Manuscript preparation. Geboes, Kelchtermans, Matthys.

Statistical analysis. Geboes.

Provision of knockout mice. Dumoutier, Renauld.

Acknowledgements

The authors would like to thank C. Dillen for technical assistance.

Ancillary