Divergent Effects of Endogenous and Exogenous Glucocorticoid-Induced Leucine Zipper in Animal Models of Inflammation and Arthritis

Authors


Abstract

Objective

Glucocorticoid-induced leucine zipper (GILZ) has effects on inflammatory pathways that suggest it to be a key inhibitory regulator of the immune system, and its expression is exquisitely sensitive to induction by glucocorticoids. We undertook this study to test our hypothesis that GILZ deficiency would exacerbate experimental immune-mediated inflammation and impair the effects of glucocorticoids on inflammation and, correspondingly, that exogenous GILZ would inhibit these events.

Methods

GILZ−/− mice were generated using the Cre/loxP system, and responses were studied in delayed-type hypersensitivity (DTH), antigen-induced arthritis (AIA), K/BxN serum–transfer arthritis, and lipopolysaccharide (LPS)–induced cytokinemia. Therapeutic expression of GILZ via administration of recombinant adeno-associated virus expressing the GILZ gene (GILZ-rAAV) was compared to the effects of glucocorticoid in collagen-induced arthritis (CIA).

Results

Increased T cell proliferation and DTH were observed in GILZ−/− mice, but neither AIA nor K/BxN serum–transfer arthritis was affected, and GILZ deficiency did not affect LPS-induced cytokinemia. Deletion of GILZ did not impair the effects of exogenous glucocorticoids on CIA or cytokinemia. In contrast, overexpression of GILZ in joints significantly inhibited CIA, with an effect similar to that of dexamethasone.

Conclusion

Despite effects on T cell activation, GILZ deficiency had no effect on effector pathways of arthritis and was unexpectedly redundant with effects of glucocorticoids. These findings do not support the hypothesis that GILZ is central to the actions of glucocorticoids, but the efficacy of exogenous GILZ in CIA suggests that further evaluation of GILZ in inflammatory disease is required.

Profound antiinflammatory actions of glucocorticoids are the basis of their widespread use in rheumatoid arthritis (RA), despite adverse effects reflecting their metabolic actions (1–3). Greater understanding of mechanisms may enable the development of therapies that mimick the antiinflammatory, but not the metabolic, effects of glucocorticoids. Glucocorticoid actions are integrated at the molecular level by the effects of the glucocorticoid–glucocorticoid receptor (GR) complex on gene transcription (4). The glucocorticoid–GR complex binds as a dimer to glucocorticoid response elements in target gene DNA, or inhibits (“transrepresses”) gene expression by tethering as a monomer to other transcription factors such as NF-κB and activator protein 1 (AP-1) (5). Among the genes most sensitive to transcriptional activation by glucocorticoids is glucocorticoid-induced leucine zipper (GILZ) (also called TSC22D3) (6). GILZ inhibits NF-κB through a physical interaction with the NF-κB p65 subunit which impedes nuclear translocation (7). GILZ also binds to the AP-1 components c-Jun and c-Fos (8) and inhibits Raf-1 (9) and Ras (10). GILZ silencing increased the severity of murine collagen-induced arthritis (CIA), and GILZ overexpression inhibited chemokine and cytokine expression in human RA synovial fibroblasts (RASFs) (11). These observations have led to the concept that GILZ is a critical mediator of the antiinflammatory effects of glucocorticoids (12, 13).

The majority of the effects of GILZ have been reported in forced overexpression studies. Although male infertility was recently reported in a GILZ-knockout (KO) mouse strain (14), the physiologic effects of GILZ on the immune system, and whether GILZ is required for the therapeutic effects of glucocorticoids, remain unknown. To investigate these issues and to probe the potential for GILZ-based RA therapies, we generated GILZ−/− mice and the means to locally overexpress GILZ in vivo. We report that although GILZ deficiency resulted in increased T cell activation, there was no effect on effector pathways of inflammation or on the expression of RA in 2 distinct models. Moreover, and unexpectedly, glucocorticoid effects on inflammation were retained in the absence of GILZ. In contrast, therapeutic overexpression of GILZ mimicked the inhibitory effects of glucocorticoids in a model of RA. These findings revise our understanding of the role of GILZ in immunity and inflammation and in the actions of glucocorticoids, yet they support the possibility of GILZ manipulation as a therapy for RA.

MATERIALS AND METHODS

Generation of GILZ-deficient mice.

GILZ-deficient mice were generated (in conjunction with Ozgene), by disruption of the Gilz gene in C57BL/6 mouse embryonic stem cells via homologous recombination using the Cre/loxP system (15). The targeting construct of the Gilz gene (see Supplementary Figure 1A, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37858/abstract) was built by using the in-house targeting vector backbone (FLSniper; Ozgene) with homologous sequences spanning ∼6 kb upstream and ∼6 kb downstream of the Gilz gene. The floxed genomic area was ∼2.2 kb, with two loxP sites flanking the last exon of the Gilz gene. The neomycin cassette was removed by crossing with Flp deletor C57BL/6 mice. The resulting heterozygous floxed animals (GilzfloxedΔNeo/wt) were crossed with a general Cre C57BL/6 deletor, and the resulting Gilzflox/wt, Cre+ mice were backcrossed to wild-type (WT) C57BL/6 mice to remove the Cre gene.

Genotypes of the resulting mice were determined by polymerase chain reaction (PCR) using genomic DNA isolated from mouse tails (see Supplementary Figure 1B, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37858/abstract). The reported infertility of male GILZ−/− mice (14) was confirmed; consequently, Gilzwt/KO females and Gilzwt/Y males were bred, and male offspring were genotyped to select Gilz-deficient mice (see Supplementary Figure 1B). GILZ−/− mice developed normally in terms of size and general development, had no difference in spleen neutrophil, monocyte, or total T cell numbers, but had a modest yet statistically significant increase in the ratio of CD4+ T cells to CD8+ T cells in the spleen (data not shown).

Induction of delayed-type hypersensitivity (DTH), antigen-induced arthritis (AIA), CIA, and K/BxN serum–transfer arthritis.

DTH was induced in WT and GILZ−/− mice with ovalbumin (OVA), and paw swelling was assessed at 24 hours using calipers as previously described (16). Lymph node T cells were stimulated with OVA (10–100 μg/ml) for 24 hours, and interferon-γ (IFNγ) and interleukin-17A (IL-17A) enzyme-linked immunospot (ELISpot) assays (BD Biosciences) were performed according to the manufacturer's protocol. DTH and AIA were induced in WT and GILZ−/− mice with methylated bovine serum albumin (mBSA)/Freund's complete adjuvant and assessed by measurement of joint swelling using calipers and histologic analysis, as previously described (17).

CIA was induced in DBA/1 mice, as previously described (11). Joints were scored daily for arthritis development as previously described, with a maximum possible score of 12 per mouse (11). Antigen-specific T cell proliferation was analyzed as previously described (17) using single-cell suspensions of lymph node or spleen cells stimulated with concanavalin A (Sigma-Aldrich) or mBSA. DNA synthesis (incorporation of 3H-thymidine [0.5 μCi; GE Healthcare]) was measured using a liquid scintillation beta counter (Cambridge Scientific). K/BxN serum–transfer arthritis was induced as previously described, and each limb was scored daily from 0 to 5, allowing for the greater degree of swelling and erythema observed in this model (18). Histologic severity was evaluated on 4-μm sections stained with Safranin O and counterstained with fast green/iron hematoxylin (11). DBA/1 mice developing CIA were treated from day 21 with liposome-encapsulated control and GILZ small interfering RNA (siRNA) sequences as previously described (11). Mice were injected intravenously with 200 μl of cationic liposome/nucleic acid formulation 3 times a week. All animal experiments were performed in accordance with the regulations of the Monash University Animal Ethics Committee.

Lipopolysaccharide (LPS)–induced cytokine production.

WT and GILZ−/− mice were injected intraperitoneally (IP) with 10 mg/kg LPS. Mice were bled via the tail vein at 0 and 30 minutes and at 1, 2, 4, and 24 hours, and serum was collected. In some experiments 1 mg/kg dexamethasone (DEX) or vehicle was injected IP 1 hour prior to LPS injection. Murine dermal fibroblasts were isolated as previously described (19) and maintained in complete RPMI 1640. Cells were seeded overnight and serum-starved in 0.5% fetal bovine serum (FBS)/RPMI 1640 prior to 24 hours of treatment with 10−7−10−10M DEX with or without 200 ng/ml LPS.

RASFs.

RASFs were cultured as previously described (20). RA patients fulfilled the American College of Rheumatology 1987 revised classification criteria (21). All experiments were approved by the Human Research Ethics Committee, Monash Medical Centre. RASFs were transfected with 100 nM control siRNA or GILZ siRNA for 24 hours using RNAiMAX (Invitrogen) as previously described (11). Cells were treated with 1 ng/ml tumor necrosis factor (TNF; BioSource International) with or without DEX (10−7−10−10M), and supernatants were collected 7 hours later.

Enzyme-linked immunosorbent assay (ELISA).

Anti–type II collagen (anti-CII) IgG, IgG1, and IgG2a and cytokine concentrations were measured by ELISA as previously described (11, 22). Murine TNF and IL-6 (R&D Systems), IL-1β (Endogen), and IFNγ and monocyte chemotactic protein 1 (MCP-1) (BD Biosciences) were detected according to the recommendations of the manufacturers. RASF supernatant cytokine concentrations were analyzed using a cytometric bead array (BD Biosciences) as recommended by the manufacturer.

Construction, production, and administration of GILZ-rAAV5.

GILZ-rAAV5 was constructed as described (23). Briefly, the mouse gene coding for GILZ flanked by adeno-associated virus type 2 (AAV2) inverted terminal repeats and under control of a cytomegalovirus promoter was packaged into the capsid from AAV5, resulting in the GILZ-rAAV vector. Vectors were produced with an adenovirus-free system in HEK 293 cells using a triple transfection method (24) and purified by density-gradient centrifugation. AAV titers were measured by quantitative PCR (qPCR), and final titers were 7 × 1013 vector genomes/ml (mock recombinant AAV [mock rAAV; empty vector]) and 3.85 × 1013 vector genomes/ ml (GILZ-rAAV). DBA/1 mice with CIA (as above) were treated with an intraarticular injection of phosphate buffered saline (PBS), mock rAAV, or GILZ-rAAV in both knee joints (5 μl, 2 × 1011 vector genomes) and both ankle joints (2.5 μl, 1 × 1011 vector genomes), administered in each mouse on the day of onset of arthritis (n = 10 per group).

Western blotting and qPCR.

Cell lysates were collected, and Western blotting was performed as previously described (11). Primary and secondary antibodies were rabbit anti-GILZ (FL134; Santa Cruz Biotechnology), mouse anti–β-actin (Sigma-Aldrich), Alexa Fluor 680–conjugated anti-rabbit IgG for GILZ (New England Biolabs), and Alexa Fluor 750–conjugated anti-mouse IgG for β-actin (New England Biolabs). Bands were detected using an Odyssey infrared imaging system (Li-Cor). GILZ messenger RNA (mRNA) expression was measured by qPCR using methods and primers previously described (11).

RAW264.7 macrophage cell lines.

The P3K cell line (25) derived from RAW264.7 macrophages expresses an NF-κB luciferase reporter construct. Cells were maintained in α-minimum essential medium (Gibco BRL) containing 10% FBS, 2 mM L-glutamine, and 1% penicillin–streptomycin–amphotericin B. GILZ expression was induced using GILZ-rAAV. After 72 hours cells were stimulated with 100 ng/ml RANKL for 6 hours, and luciferase activity was detected as described (25).

Statistical analysis.

Student's t-tests or Mann-Whitney tests were used for continuous or discontinuous variables, respectively. P values less than 0.05 were considered significant.

RESULTS

Effect of GILZ deficiency on T cell activation and inflammatory arthritis.

As inhibitory effects of GILZ overexpression on Th1 cell activation in response to OVA have been reported (26), we examined these responses in WT and GILZ−/− mice. Cutaneous DTH responses to OVA were significantly increased in GILZ−/− mice compared to WT mice (Figure 1A). This was accompanied by increased lymph node T cell IFNγ production as measured by ELISpot assay (Figure 1B). Lymph node T cell IL-17A production was also increased in GILZ−/− mice compared to WT mice (Figure 1C). To establish whether these effects would have an impact on inflammatory arthritis, we next examined the Th1-dependent mBSA-induced AIA model, which also allows examination of DTH (17). Cutaneous DTH responses to mBSA were significantly increased in GILZ−/− mice compared to WT mice (Figure 1D), and antigen-induced T cell activation, as measured by proliferation and IFNγ release, was significantly increased in lymph node cells from GILZ−/− mice (Figures 1E and F). In contrast, no difference in arthritis severity, measured by knee thickness or histologic examination, was observed between WT and GILZ−/− mice (Figures 1G and H). These findings suggest that endogenous GILZ is a physiologic inhibitor of T cell activation, but that the loss of this inhibition in GILZ−/− mice is not associated with activation of effector pathways that mediate joint inflammation.

Figure 1.

Effect of glucocorticoid-induced leucine zipper (GILZ) deficiency on delayed-type hypersensitivity (DTH) responses to ovalbumin (OVA) and T cell responses in antigen-induced arthritis (AIA). A–C, OVA/Freund's complete adjuvant (CFA) was injected into wild-type (WT) mice (n = 6) and GILZ−/− mice (n = 6), and footpad DTH was induced on day 10. A, DTH responses were measured after 24 hours. B and C, WT and GILZ−/− mouse lymph node T cells were restimulated with OVA or were not treated (NT) for 24 hours, and interferon-γ (IFNγ) (B) and interleukin-17A (IL-17A) (C) production in response to OVA was measured by enzyme-linked immunospot (ELISpot) assay. D–H, AIA was induced in WT mice (n = 16) and GILZ−/− mice (n = 14) with methylated bovine serum albumin (mBSA)/CFA, and footpad DTH was induced on day 27. D, DTH responses were measured after 24 hours. E, Lymph node T cells were restimulated with mBSA, concanavalin A (Con A), or were not treated for 48 hours prior to measurement of 3H-thymidine incorporation. F, IFNγ in T cell supernatants was measured by enzyme-linked immunosorbent assay. G, AIA severity was measured on day 28 by change in knee thickness. Values are the mean ± SEM. ∗ = P < 0.05 versus WT mice. H, Representative images of Safranin O–stained sections of knee joints from WT and GILZ−/− mice with AIA are shown.

To further explore this, we next examined the effect of physiologic expression of GILZ on the K/BxN serum–transfer arthritis model, which is T cell independent but depends on chemokine-mediated recruitment of myeloid effector cells (27). No significant difference in clinical or histologic arthritis severity was observed between WT and GILZ−/− mice (Figures 2A and B). Finally, we investigated the effect of physiologic expression of GILZ on LPS induction of TNF, IL-1β, and IL-6. LPS increased serum cytokine levels in both WT and GILZ−/− mice, and no significant differences were observed (Figures 2C–E). These observations were confirmed in vitro using thioglycollate-induced peritoneal macrophages (22) from WT and GILZ−/− mice (data not shown). Taken together, these data suggest that despite its effects on T cell activation, physiologic expression of GILZ may not exert dramatic inhibitory effects on B cell activation, myeloid effector cell recruitment, or the expression of proinflammatory effector cytokines, and therefore not on models of RA that depend on these pathways.

Figure 2.

Effect of GILZ deficiency on K/BxN serum–transfer arthritis and lipopolysaccharide (LPS)–induced cytokinemia. A and B, K/BxN serum–transfer arthritis was induced in WT mice (n = 5) and GILZ−/− mice (n = 4), and clinical (A) and histologic (B) scores were measured as described in Materials and Methods. C–E, WT and GILZ−/− mice were injected intraperitoneally with LPS (10 mg/kg). Serum was collected at the indicated time points, and concentrations of tumor necrosis factor (TNF) (C), IL-1β (D), and IL-6 (E) in WT and GILZ−/− mice were measured by enzyme-linked immunosorbent assay. Values are the mean ± SEM. See Figure 1 for other definitions.

Effects of glucocorticoids in GILZ−/− mice.

The lack of effect of physiologic expression of GILZ on effector pathways suggested that GILZ may not be required for the actions of glucocorticoids on these events, and hence on arthritis. We therefore examined the effects of DEX on LPS-induced cytokines in WT and GILZ−/− mice. DEX inhibited LPS-induced TNF, IL-1β, and IL-6 in both WT and GILZ−/− mice, and no significant difference between WT and GILZ−/− mice was detected (Figures 3A–C). We next examined the effect of GILZ deficiency on glucocorticoid sensitivity in vitro. DEX at a concentration of 10−7M robustly induced the expression of GILZ in WT murine dermal fibroblasts over 4–8 hours (Figure 3D). The chemokine MCP-1 is abundantly produced by murine dermal fibroblasts (19). DEX dose-dependently inhibited basal and LPS-induced MCP-1 in both WT and GILZ−/− mouse cells, and no significant difference between WT and GILZ−/− mouse cells was observed (Figures 3E and F).

Figure 3.

Effect of glucocorticoid treatment in GILZ-deficient mice. A–C, WT mice (n = 11) and GILZ−/− mice (n = 13) were injected with vehicle or dexamethasone (DEX; 1 mg/kg) 1 hour prior to intraperitoneal injection of lipopolysaccharide (LPS; 10 mg/kg). Serum was collected at the indicated time points after LPS injection, and concentrations of tumor necrosis factor (TNF) (A), IL-1β (B), and IL-6 (C) in WT and GILZ−/− mice were measured by enzyme-linked immunosorbent assay (ELISA). Values are the mean ± SEM. † = P < 0.05; ††† = P < 0.001, DEX-injected WT mice versus vehicle-injected WT mice. ∗ = P < 0.05; ∗∗∗ = P < 0.001, DEX-injected GILZ−/− mice versus vehicle-injected WT mice. D, WT and GILZ−/− mouse dermal fibroblasts were treated with DEX (10−7M) for 4 and 8 hours, and GILZ expression was detected by Western blotting. E, WT and GILZ−/− mouse dermal fibroblasts were treated with DEX (10−7−10−10M) for 24 hours, and the concentration of monocyte chemotactic protein 1 (MCP-1) in supernatants was measured by ELISA. Mean ± SEM values are expressed relative to untreated controls. F, WT and GILZ−/− mouse dermal fibroblasts were pretreated with DEX (10−7−10−10M) for 1 hour prior to LPS stimulation for 24 hours, and the concentration of MCP-1 in supernatants was measured by ELISA. Mean ± SEM values are expressed relative to LPS-treated controls not treated with DEX (n = 4 independent experiments). ∗ = P < 0.05; ∗∗ = P < 0.01. Tx = treatment (see Figure 1 for other definitions).

As CIA in C57BL/6 mice is insufficiently severe to permit the examination of glucocorticoid effects, we examined the effects of glucocorticoids on CIA in DBA/1 mice depleted of GILZ using liposome-encapsulated siRNA that we have shown effectively silences GILZ expression in this model (11). Exacerbation of CIA severity by GILZ siRNA treatment in the absence of DEX was confirmed (Figure 4B and data not shown). In contrast, GILZ siRNA treatment had no significant effect on DEX inhibition of CIA (Figures 4A and B). Inhibition of DEX-induced GILZ expression in human RASFs by siRNA was demonstrated using Western blotting (Figure 4C). GILZ siRNA had no effect on DEX inhibition of RASF TNF, IL-8, and IL-6, either basally (data not shown) or in response to TNF (Figures 4D–F). These data do not support the notion that GILZ plays an essential role in the antiinflammatory effects of glucocorticoids on these phenomena.

Figure 4.

Effect of glucocorticoid treatment in the absence of GILZ. A and B, Collagen-induced arthritis (CIA) was induced in DBA/1 mice, and mice with CIA were treated 3 times a week from the onset of disease with control (ct) small interfering RNA (siRNA) (siCON) (n = 10) or with GILZ siRNA (siGILZ) (n = 10) and daily with vehicle or dexamethasone (DEX; 0.25 mg/kg subcutaneously) for 10 days. Arthritis scores were determined in control siRNA–treated mice and GILZ siRNA–treated mice undergoing DEX treatment (A). ∗ = P < 0.05; ∗∗ = P < 0.01, control siRNA plus DEX versus vehicle. Scores were expressed as the area under the curve (AUC)/days observed (B). ∗ = P < 0.05. C–F, Rheumatoid arthritis synovial fibroblasts were treated with control siRNA or GILZ siRNA using RNAiMAX. Cells were treated with DEX for 5 hours, and protein lysates isolated for GILZ expression were measured by Western blotting (C). Cells were cotreated with DEX (10−7−10−10M) and tumor necrosis factor (TNF) for 7 hours, and supernatant concentrations of TNF (D), IL-8 (E), and IL-6 (F) were measured using a cytometric bead array (n = 3 independent experiments). Values are the mean ± SEM. NS = not significant (see Figure 1 for other definitions).

Effect of therapeutic GILZ-rAAV on CIA.

The above-described experiments suggest that endogenous GILZ is not a significant inhibitor of effector pathways or mediator of the effects of glucocorticoids in arthritis. To test the hypothesis that therapeutic induction of GILZ could nonetheless suppress established disease, we induced CIA in DBA/1 mice, and we induced local expression of GILZ with GILZ-rAAV. Western blot analysis indicated that infection of RAW264.7 macrophages with GILZ-rAAV resulted in GILZ protein expression and inhibition of NF-κB luciferase activity (see Supplementary Figures 2A and B, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/doi/10.1002/art.37858/abstract). Compared to mock rAAV– or PBS-injected joints, intraarticular injection of GILZ-rAAV significantly increased GILZ mRNA in the distal paw and knee (Figure 5A). In contrast, there was no increase in GILZ mRNA in the spleen (Figure 5A). In mice developing CIA, a single GILZ-rAAV injection on the day of onset of clinical disease significantly reduced arthritis severity over the subsequent 15 days (Figure 5B).

Figure 5.

Effect of GILZ-rAAV on collagen-induced arthritis (CIA). CIA was induced in DBA/1 mice as described. Mice were injected intraarticularly in the knees and ankles with phosphate buffered saline (PBS), mock recombinant adeno-associated virus (mock rAAV; empty vector), or GILZ-rAAV from the time of onset of arthritis (n = 10 per group). A, GILZ mRNA was measured in the distal paw, knee synovium, and spleen from mice treated with PBS, mock rAAV, or GILZ-rAAV. ∗ = P < 0.05 versus mock rAAV. B, The CIA clinical score was determined in mice treated with PBS, mock rAAV, or GILZ-rAAV. ∗ = P < 0.05; ∗∗ = P < 0.01 versus mock rAAV. C, Arthritis scores were determined in mice with CIA treated with vehicle (control) or dexamethasone (DEX; 0.05 mg/kg/day). ∗ = P < 0.05 versus control. D, Clinical scores in GILZ-rAAV–treated and DEX-treated mice with CIA were expressed as the area under the curve (AUC)/days observed. ∗ = P < 0.05; ∗∗ = P < 0.01. E–G, Results from PBS-treated, mock rAAV–treated, or GILZ-rAAV–treated mice with CIA are shown. Representative images of ankle joints stained with Safranin O are shown (E). Serum anti–type II collagen (anti-CII) IgG, IgG1, and IgG2a antibody titers were measured by enzyme-linked immunosorbent assay (F). Lymph node T cells were restimulated with CII or were not treated for 72 hours prior to measurement of 3H-thymidine incorporation (G). Values are the mean ± SEM. NS = not significant (see Figure 1 for other definitions). Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.37858/abstract.

In separate experiments, mice developing CIA were treated from the same time point with daily subcutaneous injections of DEX. The requirement for daily injections limited the duration of this experiment to 7 days. DEX resulted in a significant reduction in arthritis severity that was similar in magnitude to the effect of GILZ-rAAV treatment (Figure 5C). Clinical scores calculated as the area under the curve (AUC) adjusted for the number of days observed were significantly reduced by both GILZ-rAAV and DEX treatment, and the effects of these treatments were not significantly different (mean ± SEM AUC 1.5 ± 0.3 for GILZ-rAAV versus 1.0 ± 0.2 for DEX; P = 0.26) (Figure 5D). Separate experiments in which mice were observed for up to 21 days after GILZ-rAAV injection demonstrated that the inhibition of arthritis persisted (data not shown). Histologic severity was also markedly attenuated in GILZ-rAAV–treated mice (Figure 5E). In contrast, GILZ-rAAV treatment did not affect serum levels of anti-CII IgG, IgG1, and IgG2a antibodies (Figure 5F) or CII-induced T cell proliferation in cells from draining lymph nodes (Figure 5G). These data indicate an inhibitory effect of therapeutically induced local GILZ expression on joint inflammation during CIA.

DISCUSSION

Since the discovery that dimerization of the GR is not required for inhibition of the NF-κB and AP-1 pathways (28), transrepression has been considered the dominant means through which glucocorticoids regulate inflammation. Recently, however, it has been shown that many effects of glucocorticoids on inflammation require homodimeric GR effects (29), and that selectivity for transrepression reduces effectiveness of GR ligands (30). This suggests that genes transcriptionally activated by glucocorticoids are essential for the control of immune–inflammatory responses. GILZ is among the most sensitively glucocorticoid-induced proteins. Together with its ability to directly inhibit NF-κB and AP-1, this suggests that GILZ is a potential mediator of the immune–inflammatory effects of glucocorticoids. It has not previously been established whether endogenous GILZ is required for the effects of glucocorticoids or whether it acts as a constitutive inhibitory regulator of the immune response like other glucocorticoid-induced antiinflammatory proteins (31, 32).

Our first experiments demonstrated increased OVA-induced DTH responses and IFNγ production in GILZ−/− mice, consistent with previous findings in T cells of mice transgenic for GILZ (33). In addition, a previously unsuspected effect of GILZ on Th17 cell activation was suggested by increased IL-17A production in the setting of GILZ deficiency. The mechanism of these effects may depend on the reported effects of GILZ on Ras (10) or on other pathways such as NF-ATc1, as both affect T cell IL-17A expression (34). Unexpectedly, increased T cell activation in GILZ−/− mice was not accompanied by increased AIA severity. Although AIA requires the activation of T cells (31, 32), activation of effector cells resulting in the production of proinflammatory cytokines is also required, and such effects may vary between tissues. Our observations suggest that physiologic expression of GILZ is not a major regulator of effector responses to T cell activation, that increased T cell activation in the absence of GILZ was insufficient to exacerbate joint disease, or that GILZ deficiency resulted in a paradoxical insensitivity of local effector cells to T cell–driven activation.

To exclude the possibility that the lack of effect of endogenous GILZ on arthritis severity was specific to the relatively mild AIA model, we studied the more robust K/BxN serum–transfer arthritis model, which is mediated by chemokine-dependent recruitment of myeloid cells (27). Here we also found no significant difference between WT and GILZ−/− mice in clinical or histologic arthritis severity. Finally, LPS-induced production of the cytokines TNF, IL-6, and IL-1β was unaffected by the absence of GILZ. Although multiple pathways are involved in the pathogenesis of arthritis models, one possible explanation for our findings is that the effects of physiologic expression of GILZ on T cells are not accompanied by effects on effector events such as antibody production, leukocyte recruitment, or cytokine expression that are essential for the development of arthritis.

Glucocorticoids act therapeutically via effects on both T cell and effector leukocyte function. The lack of effect of GILZ deficiency in these experiments called into question the hypothesized requirement for GILZ in the effects of glucocorticoids, suggested by previous in vitro studies using GILZ silencing (35, 36). Therefore, we examined the requirement for GILZ in the effects of glucocorticoids, using GILZ−/− mice and cells and in vivo siRNA silencing. In contrast to expectations, LPS-induced serum cytokines were inhibited in vivo by DEX to an equivalent degree in WT and GILZ−/− mice, and dose-dependent effects of DEX on cytokine release in vitro were equivalent in WT and GILZ−/− mouse cells. Moreover, the attenuation of CIA by DEX was unaffected by treatment with a GILZ siRNA regimen that we have previously shown to inhibit GILZ expression in vivo (11). Finally, GILZ silencing in RASFs did not impair DEX inhibition of cytokine production by these cells. Taken together, these data do not support the hypothesis that GILZ is required for glucocorticoids to exert their therapeutic effects on these aspects of the immune response.

Assessment of a potential therapeutic molecule based on results observed in its absence may be incompletely informative. Previous studies have demonstrated that T cell overexpression of GILZ leads to reductions in severity of experimental colitis and spinal cord injury in vivo (26, 37). Therefore, to determine if GILZ could exert therapeutic effects in CIA despite the redundancy demonstrated by its deletion, we used rAAV as a means to induce the synthesis of GILZ in the joints (38). AAV serotype 5 has been shown to have tropism for dendritic cells, synovial fibroblasts, and endothelial cells in vitro, and to successfully deliver genes to the inflamed joint in vivo (23, 39). Here, GILZ-rAAV injection at the onset of disease in CIA successfully induced local GILZ expression, accompanied by significant attenuation of clinical and histologic joint inflammation. The reduction of joint disease in response to GILZ-rAAV, despite the lack of effect on T cell proliferation and anti-CII antibody production, is consistent with the restriction of GILZ overexpression to the joint and with the fact that T and B cell responses against CII would have been established prior to the time GILZ-rAAV was administered, i.e., after disease was already evident. These effects also differ from those observed in AIA in the context of GILZ deficiency, in which increased T cell activation was observed without increased joint disease. These 2 observations, that systemic deletion of GILZ during the development of immune responses resulted in increased T cell activation but not arthritis, while local GILZ overexpression inhibited joint inflammation without affecting T cell responses, suggest differences between endogenous and supraphysiologic, and between local and systemic, effects of GILZ.

Despite the effects of exogenous GILZ, we demonstrated redundancy of endogenous GILZ with glucocorticoid antiinflammatory effects. No reported studies inform as to the potential redundancy of GILZ in the metabolic effects of glucocorticoids. However, GILZ has been described to promote osteoblast differentiation (40), suggesting the possibility that therapeutic GILZ induction may not transduce proresorptive signals of glucocorticoids. A wide array of studies is needed to probe the participation of GILZ in the metabolic effects of glucocorticoids, and the availability of GILZ-deficient mice such as those we have generated here will enable such studies for the first time.

Several warnings apply to the interpretation of the results presented here. The mechanisms through which physiologic expression of GILZ affects T cell activation, and the functional consequences of these effects, require detailed examination. The lack of a phenotype in AIA in GILZ−/− mice is contradictory with the previously reported exacerbation of CIA in DBA/1 mice by GILZ silencing (11), which was confirmed by our current results with GILZ siRNA. Our GILZ−/− mice are on the C57BL/6 background, a strain that is relatively resistant to the induction of CIA (41). Backcrossing the GILZ−/− mouse strain onto the DBA/1 background would assist in addressing the effects of GILZ on CIA; however, the infertility of male GILZ−/− mice renders this infeasible. Moreover, it remains possible that a GILZ-based therapy would be insufficiently broad in its effects to be efficacious in inflammatory disease. For example, the ability of GILZ to mimic the recently highlighted direct GR-mediated repression of gene expression (42) is unknown.

In conclusion, we have demonstrated the unexpected finding that despite attenuation of CIA in response to therapeutic GILZ overexpression, and increased T cell activation in GILZ−/− mice, GILZ is not a significant physiologic inhibitor of effector pathways of arthritis. Moreover, our results indicate that GILZ is redundant with the inhibitory effects of glucocorticoids on CIA and on LPS-induced cytokine production. These findings suggest that, rather than being essential to the actions of glucocorticoids, the expression of GILZ may represent a “backup” pathway, wherein the effects of GILZ on transcription factors such as NF-κB and AP-1 operate in parallel with direct GR transrepression of these pathways. Given the lethal effects of inflammation in the absence of endogenous glucocorticoids (43), the existence of such redundancy in glucocorticoid effects could confer a survival advantage. Nonetheless, the inhibitory effect observed in response to local GILZ overexpression in CIA suggests that further exploration of potential glucocorticoid-mimicking therapeutic effects of GILZ is warranted. Understanding of the contribution of GILZ to metabolic effects of glucocorticoids would be required prior to advancing such an application to the clinic.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Morand 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 conception and design. Ngo, Beaulieu, Gu, Leaney, Santos, Xu, Escriou, Loiler, Vervoordeldonk, Morand.

Acquisition of data. Ngo, Beaulieu, Gu, Leaney, Santos, Fan, Yang, Kao.

Analysis and interpretation of data. Ngo, Beaulieu, Gu, Leaney, Santos, Yang, Kao, Vervoordeldonk, Morand.

ADDITIONAL DISCLOSURES

Authors Loiler and Vervoordeldonk are employees of Arthrogen BV.

Acknowledgements

The authors wish to thank Dr. Qiang Cheng for his helpful ideas throughout this work, Steven Lim for his assistance in tissue culture, and Dr. Camden Lo for his assistance with imaging.

Ancillary