The Thyroid Hormone Receptors Inhibit Hepatic Interleukin-6 Signaling During Endotoxemia

Decreased thyroidal hormone production is found during lipopolysaccharide (LPS)-induced endotoxic shock in animals as well as in critically ill patients. Here we studied the role of the thyroid hormone receptors (TRs) in activation of STAT3, NF-κB and ERK, which play a key role in the response to inflammatory cytokines during sepsis. TR knockout mice showed down-regulation of hepatic inflammatory mediators, including interleukin 6 (IL-6) in response to LPS. Paradoxically, STAT3 and ERK activity were higher, suggesting that TRs could act as endogenous repressors of these pathways. Furthermore, hyperthyroidism increased cytokine production and mortality in response to LPS, despite decreasing hepatic STAT3 and ERK activity. This suggested that TRs could directly repress the response of the cells to inflammatory mediators. Indeed, we found that the thyroid hormone T3 suppresses IL-6 signalling in macrophages and hepatocarcinoma cells, inhibiting STAT3 activation. Consequently, the hormone strongly antagonizes IL-6-stimulated gene transcription, reducing STAT3 recruitment and histone acetylation at IL-6 target promoters. In conclusion, TRs are potent regulators of inflammatory responses and immune homeostasis during sepsis. Reduced responses to IL-6 should serve as a negative feedback mechanism for preventing deleterious effects of excessive hormone signaling during infections.

increasing evidence that the thyroid hormones could modulate immune responses 13,14 , but the molecular mechanisms involved in the cross-talk between thyroid hormone and immune function have not yet been clarified. On the other hand, illness results in a strong reduction in thyroid hormone production and in changes in thyroid hormone metabolism, a condition called the 'sick euthyroid syndrome' or 'nonthyroidal illness syndrome' (NTIS) 15 . LPS administration serves as a model for NTIS in mice 16 . During sepsis, thyroidal T4 and T3 synthesis is down-regulated by cytokines 17 , expression of deiodinases responsible for thyroid hormone metabolism is altered 16,18,19 and TRβ and RXR expression is reduced 20,21 .
We have previously shown that TRs can antagonize NF-κ B activation in a ligand-dependent manner in pituitary cells 22,23 and that they could modulate STAT3, NF-κ B and ERK activation in the skin in vivo 24 . This suggests that TRs could regulate inflammatory responses. In this work we demonstrate that these receptors control hepatic cytokine production and signalling during sepsis in vivo and that the thyroid hormone directly antagonizes IL-6 signalling in cultured hepatocarcinoma cells and macrophages, leading to reduced STAT3 transcriptional activity.

STAT3 and ERK over-activation in TR KO mice liver.
To analyse whether TRs could modulate the activity of signalling pathways involved in immune responses, we used mice lacking TRα 1 and TRβ , the main thyroid hormone binding isoforms 12 . These animals are extremely resistant to the actions of the thyroid hormones, presenting very high circulating levels of these hormones due to the lack of the feedback mechanism by which high thyroid hormone levels suppress pituitary thyroid-stimulating hormone 25,26 . Since liver inflammation and carcinogenesis show a clear gender disparity 27 , we measured the levels of total and phosphorylated STAT3 in male and female mice, finding that pSTAT3 levels were higher in the livers of KO animals of both sexes (Fig. 1A). Strongly increased ERK phosphorylation, without changes in total levels of the MAPK, and a weaker increase in p65/NF-κ B phosphorylation with minor changes in total p65 and Iκ Bs were also observed in TR KO mice.
The role of TRs in the liver response of these pathways to endotoxic shock was examined in male WT and TR KO mice sacrificed at 45 min or 4 h after injection of 5 mg/kg LPS (Fig. 1B). LPS treatment increased pSTAT3 and pERK levels and this induction was stronger in mice lacking TRs. LPS also enhanced p65 phosphorylation and caused a transient Iκ Bα reduction, indicating NF-κ B activation in both groups, although pp65 levels were slightly higher and Iκ Bα levels were still low at 4 h in KO animals. This suggests that these signalling pathways could be could be more sensitive to LPS in TR KO mice. Therefore, we next examined the liver response to a lower LPS dose (1 mg/kg). This LPS concentration only induced significant phosphorylation of STAT3 and ERK in TR deficient mice, confirming that endogenous liver TRs exert a repressive effect in the activity of the examined pathways (Fig. 1C).
Since these differences could be due to altered cytokine production and IL-6 and TNFα appear to be key components of the hepatic response to LPS, we measured transcript levels of these cytokines in the livers of WT and KO mice. No significant differences were found between untreated WT and KO animals and the same occurred with the anti-inflammatory IL-10 cytokine or with SOCS3 mRNAs. In addition, treatment with 5 mg/kg LPS ( Fig. 2A) or with 1 mg/kg LPS (Fig. 2B) resulted in the expected increase of TNFα and IL-6 transcripts, although IL-6 transcripts were lower in KO mice after 4 h of treatment with the higher LPS concentration. This was unexpected, since STAT3 phosphorylation is increased in these animals and IL-6 appears to be a major contributor for STAT activation under these conditions. We next determined the hepatic levels of chemokines and proinflammatory cytokines after 4 h treatment with 5 mg/kg LPS using cytokine arrays. TNFα , IL-6 and IL-10 are barely detectable in the arrays, but other inflammatory mediators show clear changes between WT and KO mice ( Fig. 2C and Supplementary Fig. S1A). Thus, the levels of Cluster of Differentiation 54 (CD54), several interleukins or CXCL9 and CXCL10 were reduced in the livers of untreated TR KO mice. Cytokine and chemokine expression was strongly stimulated by LPS, and under these conditions differences between both groups were less marked, although decreased expression of some proteins was still observed in KO mice. These results show that despite presenting increased STAT3 and ERK phosphorylation, TR deficient mice have reduced hepatic levels of chemokines and pro-inflammatory cytokines. Since macrophages and other immune cells are the major cytokine-producing cells, we also measured the levels of circulating inflammatory cytokines ( Fig. 2D and Supplementary Fig. S1B). In concordance with liver data, serum levels of several cytokines and chemokines were reduced in the untreated KO mice. Again LPS induced a strong increase in the expression of many inflammatory mediators, and under these conditions no important differences between both groups were found. Of interest, serum levels of TNFα and IL-6 were strongly increased by LPS reaching similar levels in WT and KO animals.
Thyroid hormone administration reduces STAT3 and ERK activation in response to LPS. We next tested the possibility that excessive thyroid hormone signalling could be associated with the opposite phenotype to that found in TR KO mice. For this purpose, the response to LPS was analysed in euthyroid mice and in thyroid hormone treated mice. mRNA levels of Diodinase 1 (Dio 1), an accurate marker of hepatic thyroid hormone action were significantly higher in the thyroid hormone treated mice demonstrating tissue hyperthyroidism ( Supplementary Fig. S2A,B). Mice were treated with vehicle or LPS at two different doses, 5 mg/kg or 20 mg/kg BW. Of note, hyperthyroidism severely decreased survival during sepsis. Four out of six of the thyroid hormone-treated mice died within 2 h of injection with 20 mg/kg LPS, while all control animals appeared healthy and survived the treatment for at least 5 h. With the lower LPS dose, all euthyroid animals survived for 72 h, but hyperthyroid mice showed 100% mortality at this time (Fig. 3A). Therefore, only short-term experiments were performed to analyse the effect of hyperthyroidism in pathway activation and cytokine production in response to LPS. Opposite to the results obtained in TR-deficient mice, hyperthyroid livers showed reduced STAT3 and ERK phosphorylation in response to administration to either 5 mg/kg LPS (Fig. 3B) or 20 mg/kg LPS (Fig. 3C). NF-κ B activation was again demonstrated by rapid disappearance of Iκ Bα , which almost returned to normal levels after 5 h. Also in contrast with the TR-deficient mice, p65 phosphorylation was lower in the untreated hyperthyroid mice and they showed a strongly reduced response to LPS treatment. Therefore, these signalling pathways appear to be altered in an opposite way by TR-deficiency and by thyroid hormone excess.
Also opposite to the results obtained in TRs KO mice, hyperthyroidism resulted in increased hepatic levels of IL-6 and SOCS3 transcripts in response to 5 mg/kg LPS (Fig. 4A) or 20 mg/kg LPS (Fig. 4B), while TNFα and IL-10 mRNAs remained unchanged. Cytokine profiling showed that thyroid hormone treatment increased basal expression of CD54, IL-16, CXCL9 or CXCL10 that were reduced in the KO mice ( Fig. 4C and Supplementary Fig.  3A). Again, LPS treatment (5 mg/kg for 5 h) strongly induced the levels of different chemokines and cytokines. Under these conditions differences between both groups were attenuated, although increased levels of some of these immune mediators were still observed in hyperthyroid mice. Serum levels of several proteins including CXCL10, CXCL11, CCL2 or G-CSF were also increased in hyperthyroid mice before LPS treatment ( Fig. 4D and Supplementary Fig. S3B), although most of the differences between control and hyperthyroid animals again disappeared after the treatment. Despite difference of IL-6 mRNA expression in the liver, circulating IL-6 levels were similarly induced by LPS in both groups.

Increased macrophage infiltration in hyperthyroid mice.
To analyse if the observed changes in cytokine production and pathways activation were reflected in altered liver damage or infiltration of inflammatory cells, liver histology was performed in TR-deficient and hyperthyroid animals and their corresponding controls before and after treatment with LPS (5 mg/kg for 4-5 h). AST activity was not still enhanced at this early Furthermore, no liver damage that could explain early lethality in hyperthyroid mice treated with 20 mg/kg LPS was observed, discarding liver failure as responsible for death ( Supplementary Fig. S6).
T3 reduces Il-6 signalling in cultured cells. The discrepancy between cytokine production and pathway activation observed in vivo, suggested that the thyroid hormones could directly regulate cytokine signalling in hepatic cells. Therefore, we next tested STAT3 and ERK activation by IL-6 in cultured Hep3B cells (Fig. 5A). Phosphorylation of STAT3 and ERK was strongly and transiently induced by IL-6, but this induction was markedly blunted in T3-treated cells showing that the hormone inhibits IL-6 mediated activation of both signalling pathways. Similar results were obtained in cells cultured with 10% serum (Fig. 5A) and 0.5% serum ( Supplementary Fig. S7). As expected, IL-6 did not cause Iκ B degradation or p65 phosphorylation illustrating that it does not activate NF-κ B (Supplementary Fig. S7). To analyse the effect of T3 on NF-κ B activation cells were treated with TNFα . TNFα did not stimulate STAT3 phosphorylation, but caused ERK activation, a rapid and transient disappearance of IKB and a detectable phosphorylation of p65 that were not significantly altered by T3 (Fig. 5B). The lack of inhibitory effect of T3 on the NF-κ B response to TNFα was confirmed under low serum conditions ( Supplementary Fig. S7).
To test the effect of T3 in STAT3-dependent transcriptional activity, we performed transient transfection assays with a reporter plasmid bearing STAT-binding elements in Hep3B cells. IL-6 markedly stimulated the activity of the reporter, and T3 strongly antagonized this response (Fig. 5C). In contrast, the hormone did not reduce significantly activation by TNFα of a reporter plasmid containing NF-κ B binding sites (Fig. 5D). Since several genes encoding APPs contain STAT-responsive sites in their regulatory regions 2,9 , we next examined the effect of T3 in IL-6 mediated stimulation of APP transcript levels. As shown in Fig. 6A, IL-6 caused a significant increase on C-Reactive Protein (CRP), Haptoglobin and Hepcidin mRNA levels that was strongly antagonized in T3-treated cells. In Hep3B cells the expression of the Serum Amiloid A-1 (SAA1) or SOCS3 genes was not responsive to IL-6 and consequently T3 did not alter their levels. In addition, transcripts of the IL6 receptor or Glycoprotein 130 (Gp130), the common subunit of the type I cytokine receptor, essential for IL-6 signal transduction were not altered by T3. Finally, and in agreement with previous results β -Fibrinogen gene expression was stimulated by T3 28 and also by IL-6.
To evaluate STAT3 recruitment to target promoters in vivo, ChIP assays were performed with chromatin from Hep3B cells treated with IL-6 and/or T3 (Fig. 6B). Un-stimulated cells did not show significant STAT3 binding to the CRP or Hepcidin promoter regions containing STAT3 response elements 2,3 . However, STAT3 was recruited to the promoters in IL-6-treated cells, and T3 significantly reduced this response in agreement with the repression of endogenous gene activation. The abundance of acetylated histone H4, a marker for transcriptional activation, followed a similar pattern since T3 also suppressed IL-6 dependent increase of histone acetylation (Fig. 6B).
As macrophages are major cytokine-producing cells, playing a key role in the response to endotoxemia, we also examined how the hormone affected LPS, IL-6 or TNFα signalling in these cells. T3 strongly reduced LPS-induced STAT3 phosphorylation in primary cultures of macrophages and in the murine RAW264.7 macrophage cell line, while having minor effects in ERK or NF-κ B activation (Fig. 7A). This reduction was not due to a lower cytokine production in response to the endotoxin, since TNFα and IL-6 mRNA induction by LPS was similar in control and T3-treated cells (Fig. 7B). T3 also markedly reduced pSTAT3 levels in response to IL-6, but did not alter the NF-κ B response to TNFα in RAW264.7 macrophages (Supplementary Fig. S8). Therefore, T3 is also a strong antagonist of IL-6 signalling in macrophages.

Discussion
Here we report a novel role for TRs and their ligands, the thyroid hormones, in the activation of signalling pathways that play an essential role in liver homeostasis and inflammation. STAT3 phosphorylation, which mediates dimerization, nuclear translocation and transcriptional stimulation 8 , was increased in the livers of genetically modified mice lacking TRs. Hepatic ERK phosphorylation was also enhanced and a modest increase of p65 phosphorylation was also detected. Augmented activation of these signalling molecules was also found after LPS treatment, the archetypical inflammatory stimulus causing sepsis and endotoxic shock. In view of these results, our initial working hypothesis was that TRs could play a repressor role in the production of cytokines and chemokines that are crucial in the recruitment of inflammatory cells and that are in turn prominent amongst STAT3, ERK and NF-κ B targets are genes 29 . Surprisingly, induction of IL-6, the main cytokine responsible for hepatic STAT3 activation during sepsis, was reduced in TR deficient mice. An increase in circulating IL-6 that could reach the hepatic cells and account for the increased STAT3 activation was also dismissed. Although STAT3 is predominantly activated by IL-6, other cytokines such as IL-10 also activate this transcription factor 30,31 . However, no changes in IL-10 were observed and therefore increased STAT activity in TR-deficient livers cannot be attributed to increased production of this cytokine. Furthermore, when hepatic expression of an array of cytokines and chemokines, as well as their circulating levels, was measured it was found that their levels were generally lower in TR-deficient mice. These results indicate that increased STAT3 and ERK phosphorylation detected in these animals in response to LPS is most definitely not due to increased cytokine production, and they emphasize the important role of these receptors as endogenous repressors of these signalling pathways since they are overactive even in the presence of abnormally low levels of cytokines. Intriguingly, we have previously shown that increased cytokine content is found in the skin of TR-deficient mice in response to treatment with 12-O-Tetradecanoylphorbol-13-acetate 24 . Therefore, the effect of TRs on cytokine production and inflammatory responses is a complex phenomenon that appears to differ depending on the tissue and type of inflammatory insult.
Excessive thyroid hormone signalling led in general to opposite hepatic changes to those observed in the TR KO mice. Increased production of several cytokines and chemokines was found in the liver and serum of thyroid hormone-treated animals. However, circulating levels of most of these proteins reached similar levels in euthyroid and hyperthyroid mice after endotoxic shock, probably reflecting that extrahepatic cytokine production is not greatly influenced by thyroid hormone treatment. Under these conditions, hepatic expression of the IL-6 gene is more sustained in LPS treated hyperthyroid mice, despite presenting similar circulating levels of this cytokine. Again, there was a clear dissociation between the levels of inflammatory mediators and hepatic STAT3, ERK or NF-κ B activity, as phosphorylation of these signalling molecules was not increased but rather significantly reduced in hyperthyroid livers. Although the LPS dosages used were well tolerated by all control mice, thyroid hormone-treated mice showed increased lethality, indicating that they are more sensitive to the endotoxic shock caused by cytokine hyper-production. TNFα production is believed to account for the fatal shock syndrome, which includes a prominent hepatic component 32 . Interestingly, lethality was associated with the specific up-regulated hepatic expression of IL-6, but not TNF-α in hyperthyroid mice. However, no changes in the hyperthyroid livers that could explain early lethality were found and therefore an extra-hepatic mechanism that remains to be defined should be responsible for the reduced survival to endotoxic shock. Although STAT3 activation in hepatocytes could attenuate lethality in sepsis 33 , hypersensitivity to LPS induced endotoxic shock has been related to increased IL-6 production 34 . Therefore, reduced STAT3 activation might be involved in the increased mortality in hyperthyroid mice and a reduced response of the hepatic and/or extra-hepatic cells to this cytokine could represent a defence mechanism by which these animals could try to constrain sepsis-induced mortality. This is in line with the observation that thyroid hormone production by the thyroid gland is reduced not only after LPS administration in mice 16 but also in NTIS in humans 15 . Interestingly, IL-6 appears to be involved in the development of this syndrome, since the drop in circulating thyroid hormones is reduced in IL-6 KO mice 35 .
The dissociated effects of the thyroid hormones in cytokine levels and the activation status of their signalling pathways suggested that these hormones could act directly in the liver cells or in other cells to repress cytokine signalling. We demonstrate that indeed liganded TRs directly interfere with inflammation-based cell signalling pathways in cultured Hep3B cells. T3 suppresses IL-6 signalling by inhibiting the activation of the main downstream targets of the cytokine: STAT3 and ERK. This was not due to reduced expression of the IL-6 receptors and therefore TRs appear to inhibit STAT3 activation downstream of binding of IL-6 to its cognate receptors. Furthermore, STAT3 phosphorylation in response to LPS was reduced by T3 in primary cultures of macrophages, as well as in the murine RAW264.7 macrophage cell line, although the hormone did not reduce the induction of IL-6 gene expression by the cytokine. STAT3 activation in response to incubation with IL-6 was also suppressed by T3 in macrophages, indicating that inhibition of IL-6 signalling by the thyroid hormones also occurs in these cells that have potent regulatory functions during infection and inflammation.
IL-6 is known to be a strong stimulator of APP gene expression in liver cells 9 . In addition to T3 being a counter-regulator of IL-6 stimulated STAT3 phosphorylation, the hormone also antagonized induction of APP genes in response to the cytokine, reducing STAT3 binding to their regulatory regions. Stimulation of transcription by STAT factors involves the recruitment of coactivators with histone acetyltransferase activity 36 , and T3 also decreased the abundance of acetylated histone H4, a mark for transcriptional activation. Some of the examined genes such as SAA1 or even SOCS3 that contain STAT response elements and are well-known targets of this cytokine were not responsive to IL-6 in Hep3B cells, indicating the importance of the cell context and the limitations of the cultured cell models. In any case, our results suggest that the thyroid hormone could counteract the response of liver cells to the cytokine excess produced in vivo under hyperthyroid conditions by reducing its transcriptional actions and serving as a negative feedback defence mechanism. Besides this mechanism, inhibition of thyroidal secretion 16 , as well as the reduced expression of TR and its heterodimeric partner RXR 20,21 should contribute to limit the deleterious effects of excess thyroid hormone signalling during LPS-induced sepsis.
The changes in p65 phosphorylation found in the livers of TR KO and hyperthyroid mice suggested that the liganded receptors could also antagonize NF-κ B activation that plays a crucial role in the response to pro-inflammatory cytokines. We have previously shown that T3 impairs NF-κ B activation by TNFα or by serum deprivation in pituitary tumour cells, causing a significant decrease of NF-kB-dependent transcription 22,23 . However, T3 did not reduce NF-kB activation by LPS in macrophages or by TNFα in Hep3B cells. In pituitary cells this inhibition involves the induction of the MAPK phosphatase, Dual Specificity Phosphatase 1. It is possible that this mechanism does not operate in macrophages or Hep3B cells. Therefore, more studies will be needed to clarify a possible antagonistic effect of TRs in NF-kB-activation in the liver in vivo.
The liver is a central organ in maintaining metabolic homeostasis and in controlling inflammatory responses during infection. The known functions of TRs in the liver have expanded from their well-known roles in regulating liver metabolism 37 , to also participate in liver regeneration 26 , senescence 38 , fibrosis 39 and hepatocarcinogenesis 40 . In this study we show that these receptors are also potent regulators of immune homeostasis during sepsis, showing a link between thyroid hormone signalling and STAT3 and ERK activation. Therefore, TRs may represent a confluence point of metabolic and inflammatory responses in the liver, suggesting that they could be important targets for developing new therapeutic strategies for the treatment of liver diseases. New thyromimetic drugs are being developed with the expectation that they could have beneficial effects in dyslipidemia, non-alcoholic fatty liver disease or even fibrosis. Our present results showing increased mortality to sepsis in thyroid hormone treated animals stress the importance of discovering novel ligands that could dissociate their beneficial metabolic effects from their potential danger in case of infections. Protein extraction and western-blotting. Livers were homogenized and total proteins were extracted in Ripa Buffer (50 mM Hepes pH 7.5, 150 mM NaCl, 10% Glicerol, 1 mM EGTA, 1% Triton-X100, 1% Sodium Doxicolate, 0,1% SDS). Cells were harvested and lysed in triple-detergent lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate). In all cases, phosphatase and protease inhibitor cocktail tablets (Roche, Basel, Switzerland) were added. Protein lysates were mixed with Laemmli sample buffer, boiled and loaded onto 8% or 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Western analysis was performed using the antibodies and dilutions included in the Antibody Table. Cytokine and Chemokine Quantification. R&D Systems Mouse Cytokine Array, Panel A (Catalog # ARY006, Minneapolis, MN), was used to simultaneously detect the levels of 40 different cytokines and chemokines in 100 μ l serum or 400 μ g of whole-cell protein extracts obtained from livers from 3-4 animals per group and following the manufacturer specifications. Signals were detected by chemiluminiscence and were quantitated by scanning the X-ray films after different exposures (between 1 and 6 min) with ImageJ.

Methods
AST activity. Serum AST activity was measured in 32 μ l of a 1:3 dilution of sera with the Reflotron test (Roche), following the manufacturer's instructions.
Real-time quantitative RT-PCR (qRT-PCR). RNA extraction, reverse transcription and quantitative PCR were performed as described previously 38 . The sequences of the oligonucleotides used in this study are listed in the Supporting Information.
Histology and Immunohistochemistry. Liver samples were fixed in 4% buffered formalin and embedded in paraffin. Deparaffinized tissue sections were stained with H&E using standard procedures and images were obtained using a high-resolution Leica DC200 digital camera mounted on an Olympus DMLB microscope. Immunohistochemistry of the macrophage marker F4/80, a kind gift of Dr. Antonio Castrillo, was performed on 4 μ m deparaffinized-rehydrated sections with 1:100 antibody dilution. Antigen retrieval was carried out with citrate buffer and endogenous peroxidase activity was inhibited with 3% H 2 O 2 . Samples were blocked and incubated overnight with the antibody, and signal was amplified with the ABC Kit (Vectastin). Slides were revealed with DAB (Vector), counterstained with Hematoxylin and mounted with DePeX (Serva).
Cell culture. Hep-3B cells were grown in DMEM and RAW264.7 cells in RPMI with 10% FCS. Before the experiments cells were transferred to medium containing serum depleted of thyroid hormones by treatment with AG1-X8 resin. Bone marrow macrophages were obtained from the tibia and femur of 8 weeks-old C57-BL6 mice, plated and grown as previously described in DMEM and 30% medium conditioned by L929 fibroblasts 41 . After 6 days the medium was changed to DMEM containing 10% resin-treated serum and treatments started.

Transient transfections and luciferase assays.
Hep-3B cells were plated in DMEM with 10% FCS in 24-well plates. After 24 h, cells were transfected in serum free medium with 300 ng of luciferase reporter plasmids containing consensus NF-kB and STAT3 binding sites 23,42 and 50 ng of pRL-TK-Renilla (Promega, San Luis, CA) as a normalizer control, using Lipofectamine 2000 (InVitrogen, Carlsbad, CA). After 6 h cells were transferred to medium supplemented with 0.5% thyroid hormone-stripped serum by treatment with AG1-X8 (Bio-Rad, Hercules, CA) resin. Cells were incubated with 5 nM T3 for 36 h in the presence and absence of IL-6 (10 ng/ml) or TNFα (10 ng/ml) for the last 5 h. Luciferase activity was determined with the Dual Luciferase Assay System (Promega). Experiments were performed in triplicate and repeated three times. Data are mean ± s.d and are expressed as fold induction over the values obtained in the untreated cells.
ChIP assays. Hep-3B cells were plated in 150 mm dishes and after 24 h were washed and treated with 5 nM T3 for 36 h and/or 10 ng/ml IL6 for 2 h in medium containing 0.5% thyroid hormone-depleted serum. Cells were fixed and lysed following specifications of 17-295 Upstate kit, and sonicated in a Bioruptor UCD-200TM (Diagenode, Liège, Belgium). In each immunoprecipitation approximately 3 × 10 6 cells and 5 μ g of STAT3 antibody (sc-482(c-20)x), 1 μ g of acetylated histone H4 antibody (Upstate 06-866) or 3 μ g of normal rabbit IgG (sc-2027) were used. DNA was purified and precipitated. The promoter region of the C-Reactive Protein (CRP) promoter was amplified with the oligonucleotides 5′ -GAAATAATTTTGCTTCCCCTCTTCCC-3′ (forward) and 5′ -TCCTAGATCTCTTGCCTTAGAGCTACCTCC-3′ (reverse), and the promoter of the Hepcidine gene with the oligonucleotides 5′ -CCCCACCCCCTGAACACA-3′ (forward) and 5′ -ACCGAGTGACAGTCGCTTTT-3′ (reverse). Results were normalized and are presented as a fraction of the inputs after subtracting the values obtained with the control IgG, which was always below 1% of the input chromatin.
Statistical analysis. Statistics were calculated with the SPSS software. Statistical significance of data was determined by applying ANOVA followed by the Bonferroni post-test. Results shown in the figures are means ± s.e. in experiments with animals or means ± s.d. in the experiments performed in cultured cells. P-values between WT and KO mice, euthyroid and hyperthyroid mice or cells treated with and without T3 are represented in the figures with asterisks: * P < 0.05, * * P < 0.01 and * * * P < 0.001.