CXC chemokine ligand 10 (CXCL10) plays a pivotal role in the self-perpetuation of the inflammatory processes in patients with autoimmune thyroid disease. Treatment with methimazole (MMI) reduces serum CXCL10 in patients with Graves’ disease. In isolated human thyrocytes, tumor necrosis factor (TNF)α demonstrates a potent synergistic effect on interferon (IFN)γ-induced CXCL10 secretion. We investigated the mechanism underlying the synergism between IFNγ and TNFα and the effect of MMI on CXCL10 secretion in human thyrocytes. A peroxisome proliferator-activated receptor γ agonist, rosiglitazone (RGZ), a known inhibitor of T helper 1 (Th1)-mediated responses, was also studied for comparison. Experiments were carried out in human thyrocytes isolated from internodular parenchyma of thyroid tissues derived from patients who had undergone surgery for multinodular goiter. ELISA was used to measure CXCL10 levels in culture supernatant. Flow cytometry was used to assess IFNγ membrane receptor expression. Specific mRNA analysis was performed by Taqman real-time PCR. Immunofluorescence was performed to detect nuclear translocation of nuclear factor-κB (NF-κB). In human thyrocytes, the synergistic effect of TNFα with IFNγ on CXCL10 secretion is due to the upregulation of IFNγ receptor expression. MMI decreased cytokine-induced CXCL10 secretion by reducing TNFα-induced upregulation of the IFNγ receptor. RGZ decreased the cytokine-induced CXCL10 secretion by impairing NF-κB translocation, without affecting IFNγ receptor. MMI and RGZ targeted thyrocytes with the same pharmacological potency, likely acting throughout different mechanisms. Targeting T helper 1-mediated autoimmune thyroid disease with drugs that impair different intracellular pathways could be a novel pharmacological tool.
In the pathogenesis of human autoimmune thyroid disorders (AITDs), Hashimoto’s thyroiditis and Graves’ disease (GD), intrathyroidal lymphocytes seem to play a central role since thyroid antigen recognition is an essential step to T- and/or B-cell stimulation (Weetman & McGregor 1994). The T-cell pattern involved in GD might change throughout the disease course, with T helper 1 (Th1) as the predominant subtype of CD4+T cells in patients with recent disease onset and Th2 as the predominant subtype in patients with longer disease duration (Aniszewski et al. 2000).
In AITDs, a wide range of Th1-associated cytokines, such as interleukin (IL)-1, IL-2, IL-6, interferon (IFN)γ, and tumor necrosis factor (TNF)α, are produced. The cytokine production in AITDs has been ascribed to infiltrating T cells, macrophages (Weetman 2003), thyrocytes (Watson et al. 1995, Weetman 2004), and endothelial cells (Romagnani et al. 2002). The active phase of the disease is characterized by the presence of proinflammatory and Th1-derived cytokines in the thyroid gland, whereas Th2-derived cytokines do not seem to be involved (Wakelkamp et al. 2003).
Chemokines, a group of low-molecular-weight peptides belonging to the cytokine family, are known to induce the chemotaxis of different leucocyte subtypes (Zlotnik & Yoshie 2000). The CXC chemokines inducible by IFNγ – CXCL9, CXCL10, and CXCL11 – are associated with Th1-mediated immune responses (Antonelli et al. 2006a). CXCL10, in particular, has been identified as a prototypic chemokine involved in the pathogenesis of glandular autoimmunity (Romagnani et al. 2002, Rotondi et al. 2003, Antonelli et al. 2004, 2005). Indeed, CXCL10 and its CXCR3 receptor play a pivotal role in the initial phases of AITDs (Romagnani et al. 2002, Kemp et al. 2003). In patients with GD, CXCL10 was detected in thyrocytes and endothelial and inflammatory cells, which also expressed peculiarly a large amount of CXCR3 (Garcia-Lopez et al. 2001, Romagnani et al. 2001, Aust et al. 2002, Kemp et al. 2003, Antonelli et al. 2004, 2005). Furthermore, CXCL10 serum levels were significantly increased in GD patients with recent disease onset (Romagnani et al. 2002) and with active Graves’ ophthalmopathy (Antonelli et al. 2006d). The CXCL10 secretion was synergistically induced by IFNγ and TNFα in isolated thyrocytes in vitro (Garcia-Lopez et al. 2001, Antonelli et al. 2006d).
Anti-thyroid drugs (ATDs) used in the treatment of hyperthyroidism due to GD, in addition to its effect on thyroperoxidase, inhibited cell-mediated and humoral immune reactions (Volpé 2001, Laurberg 2006). While some investigators supported an immunosuppressive effect on immune system cells (Mc Gregor et al. 1980), other authors favored a direct effect exerted primarily on the thyroid cells, with secondary effects on the immune system via reduced thyrocyte–immunocyte signaling (Volpé 2001, Laurberg 2006).
Methimazole (MMI), in particular, has been reported to interfere with immunological signals associated with GD hyperthyroidism (Mc Gregor et al. 1980, Laurberg 2006) and decrease CXCL10 serum levels (Antonelli et al. 2006b,c). The reduction in circulating CXCL10 levels induced by MMI might suggest that this drug could target thyrocytes as a source of CXCL10.
To our knowledge, the synergistic effect of TNFα on IFNγ-induced CXCL10 secretion in human thyrocytes has yet to be clarified and it has not been elucidated whether MMI is able to exert any direct effect on these cells. Since we hypothesized that MMI might hamper the cytokine-induced biological pathway(s) leading to CXCL10 secretion, we investigated the mechanisms underlying TNFα+IFNγ synergistic effect on CXCL10 secretion in cultured human thyrocytes, and the modulation, if any, by MMI on CXCL10 secretion in the cells stimulated with Th1-type cytokines. We utilized for comparison a peroxisome proliferator-activated receptor γ (PPARγ) agonist, rosiglitazone (RGZ), a drug for type 2 diabetes treatment, since it has been recently reported to suppress IFNγ+TNFα-induced CXCL10 secretion in human thyroid follicular cells of GD patients (Antonelli et al. 2006b,d ).
Materials and Methods
Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F-12 medium (1:1) with and without phenol red, Ca2+/Mg2+-free PBS, BSA fraction V, glutamine, antibiotics, collagenase type IV, NaOH, Bradford reagent, and MMI were from Sigma–Aldrich Corp. The protein measurement kit was obtained from Bio-Rad Laboratories Inc. Fetal bovine serum was purchased from Unipath (Bedford, UK). IFNγ, TNFα, and ELISA kit for CXCL10 measurement were from R&D Systems (Minneapolis, MN, USA). Mouse monoclonal anti-thyreoglobulin antibody was from Cell Marque Corporation (Hot Spring, AR, USA). Goat polyclonal anti-Pax8 antibody was from Abcam plc (Cambridge, UK). For flow cytometry analysis, PE-conjugated anti-CD119 (GIR-208, mouse IgG1) mAb was from BD Biosciences (Mountain View, CA, USA), the PE-conjugated anti-TNFR2 (22235.311, mouse IgG2a) mAb was purchased from R&D Systems, and conjugated isotype-matched control Abs were from Southern Biotechnology Associated Inc. (Birmingham, AL, USA; mouse IgG1:clone 15H6, mouse IgG2a:clone HOPC-1). β-Mercaptoethanol was purchased from Fluka Biochemika Ultra (Buchs, Switzerland). For RNA extraction, RNeasy Mini reagent kit was from Quiagen Italy. TaqMan Reverse Transcription Reagents kit, all primer/probe mixes (Taqman Gene Expression Assays), CXCL10 (ID number Hs00171042-m1), IFNγR (ID number Hs00166223-m1), Pax8 (ID number Hs00247586-m1), and 1× Universal Master Mix were from Applied Biosystems (Forster City, CA, USA). Quantitative PCR human reference total RNA was purchased from Stratagene (La Jolla, CA, USA). RGZ was from Glaxo (Welwyn, UK). For immunofluorescence, rabbit polyclonal antihuman primary antibody against nuclear factor-κB (NF-κB) p65 (C-20) was from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Alexa Fluor 488 goat anti-rabbit conjugate antibody was from Molecular Probes (Eugene, OR, USA). Plasticware for cell cultures and disposable filtration units for growth media preparation were purchased from Corning (Milan, Italy).
Primary cultures of thyrocytes were obtained from inter-nodular parenchyma of thyroid tissues derived from 15 patients who underwent surgery for multinodular goiter (10 females: age range 37–83 years and 5 males: age range 32–81 years). Certificates of consent were obtained. Patients did not receive any specific treatment for thyroid disease; thyroid hormones and thyroid autoantibody measurements were in normal range. Thyrocytes were prepared as previously described (Garcia-Lopez et al. 2001) with some modifications. Briefly, tissues were minced to fragments as small as possible and treated with 2 mg/ml bacterial collagenase in PBS for 45 min at 37 °C. Digested tissues were mechanically dispersed until a homogeneous suspension was obtained. After washing with PBS, the cell suspension was cultured in DMEM/Ham’s F-12 medium (1:1) supplemented with 10% FBS, 2 mmol/l glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin in a fully humidified atmosphere of 95% air–5% CO2 in plastic 100 mm dishes. Cells began to emerge within 48 h and were used within the fifth–sixth passage. Human thyrocytes expressed paired box (Pax)8, a thyroid-specific transcription factor (Kang et al. 2001), and were positively stained for thyroglobulin and Pax8 (data not shown).
CXCL10 secretion assay
For CXCL10 secretion assays, 4000 cells were seeded onto 96-well plates in growth medium. After 24 h, the growth medium was removed and cells were washed in PBS and incubated in phenol red- and serum-free medium. After 24 h, different stimuli were added in phenol red- and serum-free medium with 0.1% BSA (200 μl/well). Cells in phenol red- and serum-free medium containing 0.1% BSA and vehicle (absolute ethanol, 0.47%, vol/vol) were used as control. For dose–response assays, cells were incubated for 24 h with TNFα at 0.1, 1, 10, 100, and 500 ng/ml or IFNγ at 10, 100, 1000, 5000, and 10 000 U/ml, alone or combined with 10 ng/ml of TNFα; or with a combination of IFNγ (1000 U/ml)+TNFα (10 ng/ml) in the presence or absence of MMI (1, 2.5, 5, 10, 25, 50, 100, 200, 300, 500, and 1000 ng/ml) or RGZ (0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 20, and 30 μM). The supernatant was harvested and kept frozen at −20 °C until CXCL10 ELISA was performed. Experiments were performed in triplicate or quadruplicate with 4–7 different cell preparations.
CXCL10 levels were measured in culture supernatants (diluted 1:20–1:30) using commercially available kits, according to the manufacturer’s recommendations. The sensitivity ranged from 0.41 to 4.46 pg/ml; mean minimum detectable dose was 1.67 pg/ml. The intra-and inter-assay coefficients of variation were 3.1% and 6.7% respectively. Samples were assayed in triplicate or quadruplicate. Quality control pools of low, normal, or high concentrations for all parameters were included in each assay. The obtained results, expressed as pg/ml, were normalized by total cell protein amount. Protein extraction was performed on the cell layer in the 96-well plates using 1 M NaOH for 5 min, followed by Bradford reagent for 10–15 min. Protein concentration was measured by spectrophotometric analysis at 595 nm.
Flow cytometry analysis
For cytometry analysis, cells were seeded onto 100 mm dishes in growth medium. Since the cultures were near confluent (about 106), the growth medium was removed and cells were washed in PBS and incubated in phenol red- and serum-free medium for 24 h. Thereafter, cells were stimulated for 24 h with IFNγ (1000 U/ml) or TNFα (10 ng/ml) in phenol red- and serum-free medium with 0.1% BSA. Cells in phenol red- and serum-free medium containing 0.1% BSA and vehicle (absolute ethanol, 0.47%, vol/vol) were used as control. Flow cytometry analysis on cell suspensions was performed as detailed elsewhere (Annunziato et al. 2002). Briefly, 105 cells were incubated with the specific or the isotype control mAb at +4 °C for 30 min; cells were then washed with PBS (pH 7.2) containing 0.5% BSA and analyzed on a BDLSRII cytofluorimeter using the Diva software (BD Biosciences). Ten thousand events for each sample were acquired. The area of positivity was determined using an isotype-matched mAb. Experiments were performed at least four times with 4–8 different cell preparations.
For mRNA analysis, cells (500 000 in 60 mm dishes) were maintained in phenol red- and serum-free medium for 24 h. Then, cells were incubated in phenol red- and serum-free medium containing 0.1% BSA with different stimuli, according to the following protocol: IFNγ (1000 U/ml) or TNFα (10 ng/ml) alone or combined; TNFα (10 ng/ml) combined with MMI (300 ng/ml) or RGZ (5 μM). Cells in phenol red- and serum-free medium containing 0.1% BSA and vehicle (absolute ethanol, 0.47%, vol/vol) were used as control. Thereafter, cells were trypsinized, washed twice in PBS, and pelleted before lysis. Cell pellet (stored at −80 °C) was resuspended in 350 μl of RLT plus β-mercaptoethanol (10 μl β-mercaptoethanol per ml of RLT buffer). Total RNA from the cells was extracted with the RNeasy Mini reagent kit according to the manufacturer’s recommendations. RNA concentration and quality were measured by NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Experiments were performed three/four times with different cell preparations.
Total RNA (400 ng) was reverse transcribed using TaqMan Reverse Transcription Reagents kit. Reverse transcription was performed in a final volume of 80 μl containing 500 mM KCl, 0.1 mM EDTA, 100 mM Tris–HCl (pH 8.3), 5.5 mM MgCl2, 500 μM of each dNTP, 2.5 μM random examers, 0.4 U/μl RNase inhibitor, and 1.25 U/μl Multiscribe Reverse Transcriptase. The reverse transcription reaction was performed at 25 °C for 10 min, 48 °C for 30 min, and 95 °C for 3 min. Measurement of gene expression was performed by quantitative real-time PCR (TaqMan). The amount of target, normalized to an endogenous reference (18s, pre-developed TaqMan Assay Reagents) and relative to a calibrator (Quantitative PCR human reference total RNA), was given by 2−ΔΔCt calculation (Livak & Schmittgen 2001). The formula applied is ΔΔCt = (Cttarget gene−Ct18s)− (Cttarget gene calibrator−Ct18s calibrator). For each sample, 12.5 ng of cDNA were added to 10 μl of PCR mix containing each primer/probe mix and 1× Universal Master Mix. The samples were then subjected to 40 cycles of amplification at 95 °C for 15 s and 60 °C for 60 s in the ABI Prism 7700 Sequence Detector (Applied Biosystems).
Ten thousand cells were seeded onto glass coverslips in growth medium for 24 h and then incubated with serum-free medium overnight, before treatment with TNFα (10 ng/ml) alone or combined with IFNγ (1000 U/ml), in the presence or absence of MMI (300 ng/ml) or RGZ (5 μM) for additional 24 h. Cells in phenol red- and serum-free medium containing 0.1% BSA and vehicle (absolute ethanol, 0.47%, vol/vol) were used as control. For method specificity, slides lacking the primary antibodies or stained with the corresponding nonimmune serum were processed. After washing twice with PBS, cells were fixed with 3.7% paraformaldehyde (pH 7.4) for 10 min and then permeabilized for 10 min with PBS with 0.1% Triton X-100. Immunostaining was performed as described previously (Vannelli et al. 1995) using primary antibody against NF-κB p65 (1:100), followed by Alexa Fluor 488 conjugate secondary antibody (1:200). The slides were examined with a phase contrast microscope (Nikon Microphot-FX microscope, Nikon, Tokyo, Japan). Experiments were performed thrice with different cell preparations.
The statistical analysis was performed using SPSS 12.0 software package (SPSS for Windows 12.0; SPSS Inc. Chicago, IL, USA). The Kolmogorov–Smirnov test was used to test for normal distribution of the data. One-way ANOVA was applied. A P value < 0.05 was considered significant and was corrected for comparisons using the Dunnett’s post hoc test. Data are expressed as mean ± s.e.m. The computer program ALLFIT (NIH, Bethesda, MD, USA; De Lean et al. 1978) was used for analysis of sigmoid dose–response curves to obtain estimates of IC50 of MMI and RGZ.
In human thyrocytes, 24-h treatment with increasing concentration of IFNγ (10–10 000 U/ml) significantly increased CXCL10 protein secretion in the supernatant over undetectable basal levels in control cells (P < 0.01 or P < 0.05 versus control; Fig. 1A). This effect was dose dependent and did not achieve plateau (125.0 ± 16.96 pg/μg protein, mean value at highest dose). Treatment with increasing doses of TNFα (0.1–100 and 500 ng/ml) enhanced CXCL10 secretion over control levels as well (P < 0.01 or P < 0.05 starting from 10 ng/ml; Fig. 1B), although the secreted amount of chemokine was low (13.49 ± 4.49 pg/μg protein, mean value at plateau).
The simultaneous addition of a fixed dose of TNFα (10 ng/ml) to IFNγ (10–10 000 U/ml) magnified the IFNγ-induced CXCL10 secretion until the maximal cell response was evoked (2866.35 ± 229.43 pg/μg protein, mean value at plateau) at each tested dose (P < 0.01 or P < 0.05 versus control and versus IFNγ alone corresponding dose), confirming the strong synergistic effect of the combined cytokines (Fig. 1C). Combination of TNFα (10 ng/ml) and IFNγ (1000 U/ml) yielded the highest response, very much in agreement with previous studies (Antonelli et al. 2006d) and, therefore, it was utilized for all subsequent experiments. Taqman real-time PCR analysis confirmed that either IFNγ (1000 U/ml) or TNFα (10 ng/ml) similarly upregulated CXCL10-specific mRNA, while cytokine combination resulted in about a 23-fold increase in CXCL10 mRNA, when compared with single cytokine-induced expression (P < 0.01 or P < 0.05 versus control, P < 0.05 versus INFγ-induced expression; Fig. 1D).
Cytokines are known to interact through the reciprocal modulation of their receptors. To verify whether the mechanism underlying TNFα and IFNγ synergism involved this mechanism in thyrocytes, we performed flow cytometry and mRNA analysis of IFNγ and TNFα receptors (IFNγR and TNFαR, type I or II) in thyrocytes stimulated with TNFα (10 ng/ml) or IFNγ (1000 U/ml). As shown in Fig. 2A upper panel, IFNγR protein membrane expression increased in thyrocytes treated with TNFα versus control cells (28.82 ± 3.11% vs 5.42 ± 0.63%, P < 0.01), while IFNγ did not show significant effect. Figure 2A lower panel depicts a representative experiment out of eight. No significant effect on TNFαRI or II (data not shown) was observed. The mRNA analysis of IFNγR also revealed a significant TNFα-induced increase in the IFNγR gene over basal level (P < 0.01 versus control) – although the increase in the protein was much higher – whereas IFNγ did not exert significant effect on the specific messenger, as shown in Fig. 2B.
To investigate the possible role of MMI in CXCL10 secretion induced by proinflammatory cytokines in human thyrocytes, cells were simultaneously incubated with a combination of IFNγ (1000 U/ml) and TNFα (10 ng/ml) and increasing concentrations of MMI (1, 2.5, 5, 10, 25, 50, 100, 200, 300, 500, and 1000 ng/ml) or RGZ (0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 20, and 30 μM) used for comparison. The first two lowest doses of each molecule (1 and 2.5 ng/ml for MMI; 0.025 and 0.05 μM for RGZ) were reported only in ALLFIT interpolation.
The drug concentrations were selected on the basis of their near therapy doses (300 ng/ml = 2.63 μM for MMI and 5 μM for RGZ respectively) according to their pharmacokinetics (Cmax and area under the time–concentration curve, AUC).
As shown in Fig. 3A, 24-h cell incubation with MMI decreased, in a dose-dependent manner, cytokine-induced CXCL10 secretion, significantly from 10 ng/ml (P < 0.01 or P < 0.05 versus IFNγ+TNFα-treated cells).
MMI and RGZ exerted a comparable inhibitory effect on CXCL10 secretion, as confirmed by the simultaneous fitting using ALLFIT program (De Lean et al. 1978) of the inhibitory curves, showing no statistically significant difference in 50% effective concentration (IC50 = 0.29 ± 0.05 μM, P = 0.21), as shown in Fig. 3C. To verify whether the inhibitory effect of both drugs somehow affected the mechanism of synergy between the two cytokines, we analyzed IFNγR expression induced by TNFα (10 ng/ml) in the presence of MMI (300 ng/ml) and RGZ (5 μM) respectively.
Flow cytometry demonstrated that the treatment of thyrocytes with MMI for 24 h significantly reduced IFNγR membrane protein expression induced by TNFα (17.4 ± 3.8% vs 30 ± 1.5% respectively, P < 0.05), while RGZ did not exert any significant effect (Fig. 4A). Real-time PCR analysis of specific mRNA confirmed that MMI significantly reduced IFNγR gene expression induced by TNFα (P < 0.01 versus TNFα-induced expression), although to a lower extent with respect to protein, while RGZ showed no effect (Fig. 4B).
In order to assess whether MMI or RGZ impaired the pathway of NF-κB, a pleiotropic transcriptional factor tightly associated with several cellular biological functions (Gilmore 2006) and specifically activated by TNFα, we set up a nuclear translocation assay.
Stimulation with TNFα alone (10 ng/ml; Fig. 5B) or combined with IFNγ (1000 U/ml; Fig. 5C) resulted in the same compact increase (nuclear staining with TNFα: 91.42 ± 2.13%; nuclear staining with TNFα+IFNγ: 89.82 ± 2.48%) of NF-κB translocation from the cytoplasmic to the nuclear compartment. In Fig. 5A, control cells are depicted, showing 1.14 ± 0.64% nuclear staining for NF-κB. IFNγ alone did not exert any effect on NF-κB (data not shown). Since the effect of each drug with TNFα alone or combined with IFNγ was the same, only the condition with cytokine combination has been reported.
Co-incubation with MMI (300 ng/ml) did not influence nuclear translocation as demonstrated by 83.23 ± 3.21% of nuclear stain cell positivity (not significant versus cytokine-treated cells; Fig. 5D). Conversely, simultaneous incubation with RGZ (5 μM) significantly reduced cytokine-stimulated NF-κB translocation, although more than 40% of cell nuclei still showed positivity (nuclear staining 42.13 ± 6.85%, P < 0.01 versus cytokine-treated cells; Fig. 5E). Columns in Fig. 5F summarize the results obtained with all treatments.
This study shows that in human thyrocytes the synergistic effect of TNFα with IFNγ on CXCL10 secretion is associated with a significant upregulation of IFNγR driven by TNFα, both at protein and gene levels. We, for the first time, provide evidence in thyrocytes that MMI decreases cytokine-induced CXCL10 protein secretion by reducing TNFα-induced upregulation of the IFNγ membrane receptor. RGZ similarly decreases cytokine-induced CXCL10 secretion without affecting IFNγR expression.
Interestingly, RGZ blocked TNFα-mediated nuclear translocation of NF-κB, pointing out that both drugs abrogated CXCL10 secretion likely by different pathways.
CXCL10 secretion potentiates the Th1 immunity-promoting inflammatory loop, driven by the recruited CD4+T lymphocytes at inflammation sites, supporting T-cell proliferation and IFNγ secretion (Campbell et al. 2004, Romagnani et al. 2004). It is well known that TNFα synergizes with IFNγ, modulating a number of different biological effects (Krakauer & Oppenheim 1993). In particular, the synergy between the two cytokines is essential in potentiating inflammatory and Th1 immune responses promoted by IFNγ. To date, previous studies reported that CXCL10 secretion is triggered only by IFNγ in different resident cell types (Rotondi et al. 2005, Antonelli et al. 2006d ), while in microvascular endothelium TNFα has been reported to upregulate cytokines other than CXCL10 (Sana et al. 2005).
In our study, TNFα alone elicited a dose-dependent CXCL10 protein secretion in thyrocytes, although to a low extent. Accordingly, TNFα alone was able to upregulate CXCL10 mRNA similarly to IFNγ, suggesting that TNFα, in addition to its synergistic effect with IFNγ, could contribute to initiating the Th1-mediated response.
Furthermore, the synergic action between the two cytokines in thyrocytes seems to be exerted mainly at a receptor level. A number of cytokines are known to amplify their biological responses both in vivo and in vitro via regulation of the expression of other cytokine receptors (Krakauer & Oppenheim 1993).
We clearly demonstrated that TNFα upregulated IFNγR at both protein and gene levels, while no significant effect was observed on TNFR type I or II expression (data not shown). Since dose–response curves with IFNγ did not reach a plateau, the increase, induced by TNFα, of IFNγR expression may account for the raised magnitude of the response to IFNγ, in terms of CXCL10 protein secretion. Accordingly, TNFα was able to potentiate the amount of CXCL10 protein secretion (ng/ml versus pg/ml) dose-dependently induced by IFNγ, until the maximal cell response was evoked. CXCL10 mRNA specific amount resulted significantly increased in the presence of both cytokines. Since cells with higher number of specific receptors are able to respond to lower concentrations of IFNγ, TNFα could be critical in the pathogenesis of AITDs, in line with previous work highlighting the relevance of the TNFα pathway in GD (Diez et al. 2002).
Our data support the thyroid itself as the main site of CXCL10 secretion perpetuating the Th1-mediated auto-immune cascade in AITDs (Garcia-Lopez et al. 2001, Romagnani et al. 2002, Kemp et al. 2003, Antonelli et al. 2004). This is very much in agreement with recently published studies in patients showing the reduction of high CXCL10 serum levels in patients with GD after thyrodectomy (Antonelli et al. 2006b) or after treatment with 131I (Antonelli et al. 2007).
Treatment with thionamidein AITDs, in addition to its effect on hormone synthesis, is accompanied by a gradual remission of the autoimmune aberration in the majority of patients (Pinchera et al. 1969, Mc Gregor et al. 1980, Laurberg 2006). A recent report describes a reduction in CXCL10 circulating levels in hyperthyroid patients with GD under MMI therapy (Antonelli et al. 2006c).
We showed that MMI decreased proinflammatory cytokine-induced CXCL10 secretion in thyrocytes in vitro, throughout a strong reduction in TNFα-induced IFNγR membrane protein expression. This effect, matching a reduction in specific amount of IFNγR mRNA, seems to be particularly relevant because it counteracts the magnitude of the CXCL10 secretion in response to IFNγ, as previously emphasized. Thus far, it turned out that MMI exerts its immunomodulatory effect on thyrocytes and this should not be surprising since it is well known that it concentrates in follicular cells and not in the surrounding interstitial tissue, where the infiltrating immunocompetent cells are present (Marchant et al. 1972).
Hyperthyroidism per se plays a minimal role in determining CXCL10 levels that, indeed, are strongly associated with the autoimmune process of the hyperthyroid phase in GD (Romagnani et al. 2002, Antonelli et al. 2006a), but not with toxic nodular goiter (Antonelli et al. 2007). In this light, MMI emerged as a particularly relevant therapy, since it targets the thyroid by dual action, both restoring euthyroidism and decreasing CXCL10 release by thyrocytes.
Concerning drug doses, we want to highlight that the concentrations used in our experiments were selected according to drug plasma or intraglandular concentration range following therapeutic doses, differently from previous in vitro studies that have used far higher drug concentrations than the concentration reached in the thyroid tissue in vivo (Mc Gregor et al. 1980).
Our studies on MMI were performed in comparison to RGZ, a pure PPARγ agonist, because it has been recently showed to reduce CXCL10 levels in thyroid follicular cells in GD patients (Antonelli et al. 2006d ). PPAR γ agonists are known to play an inhibitory role in the self-perpetuation of chemokine-mediated inflammatory processes in several human cell types (Su et al. 1999, Marx et al. 2000, Gosset et al. 2001, Antonelli et al. 2006d ).
RGZ exerted its inhibitory effect on Th1-polarized response in thyrocytes without affecting IFNγR expression. In human thyrocytes, RGZ significantly inhibited CXCL10 secretion at micromolar concentrations by inhibiting NF-κB activation, as demonstrated by the decrease of TNFα-induced NF-κB translocation in the cell nucleus. On the contrary, in other cellular models (colon cells transfected with an NF-κB–luciferase reporter element), the mechanism of CXCL10 decrease induced by PPARγ agonists seems to be NF-κB independent (Schaefer et al. 2005). Further studies are necessary to provide details on the RGZ mechanism of action, intended to elucidate the potential effect on Graves’ ophthalmopathy. Indeed, while some authors have suggested (Antonelli et al. 2006d ) its beneficial effect, there is experimental and clinical evidence that PPARγ agonists may impair the clinical course of Graves’ ophthalmopathy by adipogenesis stimulation (Valyasevi et al. 2002, Starkey et al. 2003).
In summary, we have demonstrated a novel anti-inflammatory effect of MMI that is able to revert Th1 cytokine-mediated CXCL10 secretion directly in thyroid cells by damping the mechanism underlying cytokine synergy. Although further investigations are needed to understand its signaling pathway, MMI targeted thyrocytes with the same pharmacological potency as RGZ, a regulator of the inflammatory response (Campbell 2005), likely acting through different mechanisms.
Given the pivotal role of IFNγ-induced chemokines in Th1-mediated autoimmune thyroid diseases (Rotondi et al. 2007), a combination of different drugs with the same inhibitory effect targeting different intracellular pathways could be a novel pharmacological tool, intended to decrease therapeutic doses and thus minimizing side effects, similar to immunosuppressive protocols for the treatment of other autoimmune diseases or allograft rejection.
The authors wish to thank Dr Sabino Scolletta, Department of Surgery and Bioengineering, University of Siena, Italy for his kind help with statistical analysis.
Funding This research was supported by TRESOR (Tuscany REgional Study On Rosiglitazone) and MIUR (Ministero dell’Istruzione, dell’Università e della Ricerca). We declare that there is no conflict of interest that would prejudice the impartiality of the research reported.
RotondiMet al.2005 Elevated serum interferon-γ-inducible chemokine-10/CXC chemokine ligand-10 in autoimmune primary adrenal insufficiency and in vitro expression in human adrenal cells primary cultures after stimulation with proin-flammatory cytokines. Journal of Clinical Endocrinology and Metabolism902357–2367.