Characterization of the Nuclear Factor-�BResponsiveness of the Human dio2 Gene
Aniko Zeold, Marton Doleschall, Michael C. Haffner, Luciane P. Capelo, Judit Menyhert, Zsolt Liposits,Wagner S. da Silva, Antonio C. Bianco, Imre Kacskovics, Csaba Fekete, and Balazs Gereben
Laboratory of Endocrine Neurobiology (A.Z., J.M., Z.L., C.F., B.G.), Institute of Experimental Medicine, HungarianAcademy of Sciences, and Department of Neuroscience (Z.L.), Faculty of Information Technology, Peter Pazmany CatholicUniversity, Budapest H-1083, Hungary; Department of Physiology and Biochemistry (M.D., I.K.), Faculty of VeterinaryScience, Szent Istvan University, H-1400 Budapest, Hungary; Division Medical Biochemistry (M.C.H.), Biocenter, InnsbruckMedical University, A-6020 Innsbruck, Austria; Thyroid Section (L.P.C., W.S.d.S., A.C.B.), Division of Endocrinology,Diabetes, and Hypertension, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts 02115;and Tupper Research Institute and Department of Medicine (C.F.), Division of Endocrinology, Diabetes, and Metabolism,Boston, Massachusetts 02111
Type 2 iodothyronine deiodinase (D2) activates T4 by deiodi-nation to T3, a process being the source of most T3 present inthe brain. In the mediobasal hypothalamus, expression of thedio2 gene is potently activated by administration of bacteriallipopolysaccharide (LPS), which in turn mediates the modi-fications in thyroid homeostasis typically observed in pa-tients with nonthyroidal illness syndrome. Here we show thatLPS-induced D2 expression is also observed in human MSTO-211H cells that endogenously express D2. Exposure to LPSrapidly doubled D2 activity by a mechanism that was partiallyblocked by the nuclear factor-�B (NF-�B) inhibitor sulfasala-zine. Next, the human dio2 5�-flanking region promoter assaywas used in HC11 cells and the p65/NF-�B responsivenessmapped to the 3� approximately 600-bp region of hdio2 5�-flanking region, with an approximately 15-fold induction.
Semiquantitative EMSA identified the strongest NF-�B bind-ing sites at the positions �683 bp (called no. 2) and �198 bp (no.5) 5� to the transcriptional starting site. Despite the very sim-ilar NF-�B binding affinity of these two sites, site-directedmutagenesis and promoter assay indicated that only site no.5 possessed transactivation potency in the presence of the p65subunit of NF-�B. Other cytokine mediators such as signaltransducer and activator of transcription-3 (STAT3) or signaltransducer and activator of transcription-5 (STAT5) did notinduce transcription of the dio2 gene. Our results indicatethat inflammatory signals regulate D2 expression predomi-nantly via the NF-�B pathway in a direct transcriptional man-ner and could contribute to the changes in thyroid economyobserved in nonthyroidal illness syndrome during infection.(Endocrinology 147: 4419–4429, 2006)
TYPE 2 DEIODINASE (D2) is an outer ring deiodinasethat activates T4 by generating T3, which can effectively
bind to the thyroid hormone receptor (1). The complex reg-ulation of dio2 transcription contributes to the maintenanceof the spatially and temporally regulated cellular T3 levels inboth developing and adult animals (2). Importantly, D2serves as the exclusive activating deiodinase in the humancentral nervous system (3). In the brain, astrocytes and ta-nycytes are the two major cell types expressing D2, as dem-onstrated in different species (4–7).
The negative regulation of the dio2 gene by T3 forms thebasis of D2-mediated homeostatic regulation of tissue T3levels (2). D2 expression in astrocytes of the cerebral cortexis regulated predominantly in a homeostatic manner, whichprevents changes in cortical T3 levels despite changes ofperipheral T4 levels (8–10). However, although D2 expres-
sion in tanycytes in the mediobasal hypothalamus is rela-tively insensitive to changes in thyroid hormone levels, itresponds robustly to stimuli during infection (11). Conse-quently, D2 induction in tanycytes and the resulting increasein T3 production could affect the hypothalamic-pituitary-thyroid axis via changing the T3 levels in the region of TRH-secreting neurons in the paraventricular nucleus of the hy-pothalamus. This negative feedback mechanism could playa role in the generation of nonthyroidal illness syndrome.
Using an infection model, we have previously shown thatbacterial lipopolysaccharide (LPS) treatment increases D2mRNA levels in rat tanycytes (12) independent of the LPS-induced fall in serum thyroid hormone levels (13). The LPS-induced increase of D2 mRNA in the hypothalamus was alsoobserved in mice, immediately followed by other changes inthyroid economy, e.g. decreased expression of thyroid re-ceptor �2, TSH� in the pituitary, and decreased D1 mRNAin the pituitary and liver (14).
LPS is recognized by Toll-like receptor (TLR) 4, whichcontains an extracellular leucine-rich domain and an intra-cellular Toll/IL-1 receptor signaling domain (15). TLR4 sig-nals through the adapter protein MyD88, similarly to otherTLRs (with the possible exception of TLR3) (16). TLR acti-vation leads to an early translocation of the nuclear factor-�B(NF-�B) with consequent up-regulation of proinflammatory
First Published Online May 25, 2006Abbreviations: CMV, Cytomegalovirus promoter; D2, type 2 iodo-
thyronine deiodinase; DMSO, dimethylsulfoxide; DTT, dithiothreitol;5�FR, 5�-flanking region; IFN, interferon; LPS, lipopolysaccharide; NF-�B, nuclear factor-�B; STAT, signal transducer and activator of tran-scription; TESS, Transcription Element Search System; TLR, Toll-likereceptor; TSS, transcriptional starting site.Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving theendocrine community.
cytokines, costimulatory molecules, and chemokines includ-ing IL-1�, TNF-�, interferon (IFN)-�, and IL-6 (17, 18). As aconsequence, NF-�B is a crucial effector of LPS or LPS-induced cytokines (e.g. TNF-�) and plays an important rolein the signaling of other cytokine receptors as well (19–21).Therefore, NF-�B is an eminent candidate for transcriptionfactors potentially involved in LPS-induced D2 up-regula-tion, along with signal transducer and activator of transcrip-tion (STAT) proteins, which are involved in the signaling ofsome cytokine molecules (IL-6, IFN�) and play a crucial rolein host defense in infection (22). Although we have previ-ously shown that D2 expression is induced by NF-�B, themolecular mechanism of this response has not been estab-lished (12).
We assessed the potency of NF-�B to increase human D2activity in a nonheterologous expression system, examin-ing the effect of LPS-induced NF-�B activation on endog-enous D2 expression of the human mesothelioma (MSTO-211H) cell line. Furthermore, we characterized the NF-�Bbinding sites of the human dio2 (hdio2) 5�-flanking region(5�FR) to obtain evidence for the direct transcriptionalregulation of the hdio2 gene by NF-�B. Finally, we studiedthe responsiveness of the dio2 gene to STAT3 and STAT5to identify additional effectors that could increase D2 ex-pression during infection.
Materials and MethodsExpression constructs
Luciferase reporter constructs were generated by inserting hdio2 pro-moter fragments into pGL3-basic vector (Promega, Madison, WI) aspreviously described (12, 23), followed by automated sequencing. Thehdio2–6.9-Luc, hdio2–2.1-Luc, and hdio2–584-Luc have been describedpreviously (12).
The hdio2-Luc constructs were generated by Vent polymerase, andthe hdio2 5�FR cassette was inserted between SacI and NheI into a pGL3-basic vector (hdio2-short-Luc) containing the 95-bp hdio2 short promoter(hdio2-short-Luc) between NheI and HindIII. Names of the constructscontain the positions of the fragments. Constructs generated via thisstrategy are as follows: hdio2–5�-Luc, �3880 to �6860; hdio2-PstI-Luc,�2079 to �3880 PstI fragment; hdio2-PstI/PacI-Luc, PstI-PacI fragment�2079 to �901; and the constructs containing mutagenesis of no. 1, 2,and 5 hdio2 binding sites for the p65 subunit of NF-�B. Oligonucleotidesused for hdio2–1,2,5wt-Luc and its derivatives were: hdio2–2,5wt-Luc,Bp383–385; hdio2–2,5Mut-Luc, Bp384–386; hdio2–2wt-5Mut-Luc, Bp38-Bp386; hdio2–2Mut-5wt-Luc, Bp384–385; hdio2–2wt-5RatMut-Luc,Bp383–402; hdio2–1,2,5wt-Luc, Bp381–385; hdio2–1Mut-2wt-5Mut-Luc,Bp382–386; hdio2–1Mut-2wt-5wt-Luc, Bp382–385 (see Table 2). Positionof the mutated fragments in the hdio2 promoter (used for studies shown;see Fig. 7) was depicted (see Fig. 3).
The hdio2-Luc constructs generated independently from hdio2-short-Luc were as follows: hdio2–117-Luc, �117 to transcriptional starting site(TSS) and hdio2–901-Luc, �901/PacI site/ - to TSS. The PacI-NheI regionof hdio2–6.9-Luc was mutagenized by overlap extension PCR to gen-erate hdio2–6.9–5Mut-Luc, and this DNA was used as template togenerate hdio2–6.9–2,5Mut-Luc. The mutated fragments were insertedbetween PacI-NheI of hdio2–6.9-Luc [inner oligos for no. 5 mutation:Bp452 and its reverse complementer; for no. 2 mutation: 2-MutB (Table1) and its reverse complementer].
The Renilla luciferase construct pRL-actin (24) was kindly providedby P. J. van den Elsen (Leiden, The Netherlands), the STAT3c (25) by J.Bromberg (New York, NY), the GAS-Luc by A. N. Hollenberg (Boston,MA) (26), the rdio2–3.6-Luc by P. R. Larsen (Boston, MA), and the humanp65 expression plasmid (27) by M. Naumann (Berlin, Germany).
DNA transfection and luciferase promoter assayNF-�B, STAT3
HC11 [a D2-expressing mouse mammary cell line (28) that has beensuccessfully used to analyze the NF-�B pathway (29)], HEK-293, andU87 cells were cultured as described and transfected by the polyethyl-enimine (HC11, HEK-293) or calcium phosphate method (U87) whenthey reached approximately 70% confluency in 6-well plates (12, 23, 30).Neither HEK-293 [a human kidney cell line with established potentialto study STAT-mediated pathways (28)] nor U87 (a human glioblas-toma) express D2 endogenously (28). In one well, 800 ng of thepGL3Basic constructs were cotransfected with 200 ng of cytomegalovi-rus promoter (CMV)-driven p65 subunit NF-�B (27) or the same amountof the empty pCI-neo vector (Promega) and 1800 ng of pGEM-T vector(Promega) as carrier plasmid DNA. To monitor transfection efficiency,4 ng hu-�-actin promoter-driven Renilla luciferase construct (pRL-actin)was cotransfected. The same cotransfection was performed for STAT3experiments, with the exception that the p65 vector was replaced by 200ng of a vector expressing STAT3c, a constitutively active form of STAT3(25).
For transfection of U87 cells with calcium phosphate, 2400 ngpGL3Basic construct, 600 ng p65 expression vector or empty pCI-neovector, and 20 ng pRL-actin were transfected. Luciferase activity wasassessed using the dual-luciferase reporter assay system (Promega) anda Luminoskan Ascent luminometer (Thermo Electron Corp. Labsystems,Vantaa, Finland). Each construct was transfected at least three times.
STAT5
To produce an active phosphorylated STAT5, prolactin was added tothe cells along with cotransfection of the prolactin receptor, as described(31). We used this pathway to activate STAT5 in COS-7 and HEK-293cells that were grown in 6-well-plates and transfected with TransFastTHtransfection reagent (Promega). COS-7, a monkey kidney cell line with-out endogenous D2 expression is an established system for studies onthe D2 promoter (28) and STAT mediated pathways (28). A total of 2200ng DNA comprising 500 ng prolactin receptor PRL-R, 500 ng STAT5a(31), 1000 ng of pGL3-based constructs /hdio2–6.9-Luc, hdio2-onlylinker, rdio2–3.6-Luc (12), and pGL3(�344/�1) containing the rat �-ca-sein promoter from �344 to �1 (31)/ and 200 ng of SV40 Renilla ex-pression constructs was used for each 6-well plate. Thirty-six hours aftertransfection, cells were treated with 5 �g/ml ovine prolactin (Sigma, St.Louis, MO) or vehicle alone. Twelve hours after hormone stimulation,cells were washed with ice-cold PBS and luciferase activity wasmeasured.
EMSA, supershift assay, and semiquantitative EMSA
HeLa (a cell line without endogenous D2 expression) and HC11 cellswere treated with 20 ng/ml TNF-� (Sigma) for 1 h followed by thepreparation of nuclear extracts with CelLytic nuclear extraction kit
TABLE 1. hdio2 5�FR oligonucleotides used for the NF-�B EMSA
Wild-type and mutated sense oligonucleotides are shown. Con-served nucleotides are shown in bold.
a Consensus based on the binding site matrix of Kunsch et al. (42)and the Transfac (37).
b As positive control the NF-� B binding site containing oligonu-cleotide of the Promega gel shift assay systems was used (36).
4420 Endocrinology, September 2006, 147(9):4419–4429 Zeold et al. • NF-�B Responsive Elements in the Human dio2
(Sigma). Sense and antisense oligonucleotides containing hdio2 putativeNF-�B binding sites were annealed to produce double-stranded DNA(Table 1). Labeled probes were generated with T4 polynucleotide kinaseusing 1 �l �-[32P]ATP (�5 �Ci) followed by purification through aSephadex mini Quick Spin column (Roche, Indianapolis, IN). Labeleddouble-stranded DNA was incubated with nuclear extract to achieveprotein-DNA binding. The binding buffer contained 4 mm HEPES (pH7.9), 20 mm KCl, 0.4 mm dithiothreitol (DTT), 0.2 mm EDTA, 0.5 mg/mlBSA, 50 �g/ml poly(dI-dC) (Sigma), 0.1% IGEPAL (Sigma), and 4%glycerol. Alternatively, reaction mixture contained unlabeled oligonu-cleotides for competition assay or antibody to the p65 subunit of NF-�B(H-286, Santa Cruz Biotechnology, Santa Cruz, CA) for supershift assay.For comparison of signal strength, 400,000 cpm of each probe wereadded to the mixture. Probes were run on a nondenatured [4% (wt/vol)]acrylamide gel in 0.5� Tris-borate EDTA at 18 C. SemiquantitativeEMSA was performed by adding excess of unlabeled probe to the bind-ing reaction mixture (see Fig. 6). After electrophoresis, gels were driedand exposed for several hours. For analysis of the semiquantitativeEMSA results, the signal intensities of the specific bands were deter-mined by Quantiscan (Biosoft, Cambridge, UK). Signal strength of eachlane was calculated as percentage of the value in the control lane (probewithout competitor) and plotted against the amount of the added un-labeled competitor DNA. Each gel shift experiment was performed atleast twice. Semiquantitative EMSAs were run in triplicate.
LPS treatment of MSTO cells and D2activity measurements
Human mesothelioma (MSTO-211H) cells with endogenous D2 ex-pression were cultured as described (32). The cells were treated with 1�g/ml LPS (026:B6) (Sigma) for 4 and 10 h (Fig 1A) and subjected to D2activity measurement as previously described (32). In short, cells wereharvested, washed, and sonicated in buffer of 0.1 m potassium phos-phate, 1 mm EDTA (pH 6.9) containing 10 mm DTT and 0.25 m sucrose.Fifteen micrograms cell homogenate were assayed for deiodination offreshly purified 2 nm 125I-labeled T4 in the presence of 20 mm DTT and1 mm propylthiouracil for 3 h at 37 C. D2 activity was reported asfemtomoles per hour per milligram protein. Sulfasalazine (Sigma) wasused as a NF-�B inhibitor (33). Sulfasalazine was dissolved in dimeth-ylsulfoxide (DMSO) and added at concentrations of 0.1, 0.3, and 0.5 mm(Fig. 1B). DMSO was added at concentrations of 0.02, 0.06, and 0.1%,respectively, as a control. Cells were incubated for 30 min with sul-fasalazine or DMSO. After their removal, cells were incubated with 1�g/ml LPS or PBS. Cells were harvested 4 h after treatment and samplesassayed for D2 activity as described above.
ResultsNF-�B increases D2 activity in MSTO cells
To study the effect of NF-�B in human cells, we used theMSTO-211H mesothelioma cell line, which endogenously
expresses both D2 (32) and NF-�B (34). Cells were exposedto 1 �g/ml LPS for 4 or 10 h to induce NF-�B. D2 activity wassignificantly increased after 4 h (Fig. 1A). Furthermore, 30min treatment with the NF-�B inhibitor sulfasalazine par-tially blocked the LPS- (4 h treatment) induced increase in D2activity (Fig. 1B). These data indicate that the LPS-NF-�Bsignaling pathway is capable of stimulating D2 expression inhuman cells endogenously expressing D2.
The responsiveness of the hdio2 promoter to the p65 subunitof NF-�B is restricted to the 3� region
To characterize the molecular basis of the NF-�B respon-siveness of the human dio2 gene, we assessed the effect of thep65 subunit of NF-�B on different regions of the hdio2 5�FR.Computer-assisted inspection, Transcription Element SearchSystem (TESS) (35) identified 12 putative NF-�B binding sitesat �5520, �5344, �4018, �3684, �2487, �2334, �878, �683,�546, �254, �198, and �111 in the 6.9-kb 5�FR (numbersindicate the position of the starting nucleotide of the element5� in relation to the TSS). Thus, we performed promoterassays with luciferase expression constructs containing thehdio2 5�FR fragments, with Renilla luciferase as internal con-trol. Coexpression of the human p65 subunit of NF-�B withthe 6.9-kb hdio2 5�FR fragment (hdio2–6,9-Luc, Fig. 2) inHC11 cells, a D2-expressing mouse mammary cell line (28)that has been successfully used to analyze the NF-�B path-way (29), resulted in a 140-fold increase in transcriptionalactivity, confirming previous results (12). This construct wasalso tested in the U87 glioblastoma and astrocytoma derivedhuman cell line without endogenous D2 expression (23), inwhich it was induced approximately 55-fold by p65coexpression.
To locate the p65-responsive regions of the hdio2 5FR, westudied the p65 responsiveness of different fragments of thehdio2 5�FR. A truncation series of the hdio2 5�FR was gen-erated, linked to the luciferase reporter gene (Fig. 2), andused for promoter assay after p65 coexpression in HC11 cells.Whereas hdio2 5�FR regions between �6860 and �901showed no p65 responsiveness (Fig. 2), fragments containingportions of the approximately 1 kb proximal to the hdio2 TSSwere clearly induced except for the 95-bp short hdio2 pro-moter (hdio2-short-Luc) and the 117-bp 3� region of the hdio2
FIG. 1. A, D2 activity of MSTO-211H cells after4 and 10 h treatment with 1 �g/ml LPS. **,Significantly different from corresponding PBS-(vehicle) treated control, P � 0.01; ***, P � 0.001.B, LPS (1 �g/ml) induced D2 activity in MSTO-211H cells in the presence of 0.1, 0.3, and 0.5 mMNF-�B inhibitor sulfasalazine (SULF). Cellswere preincubated for 30 min with SULF orDMSO (vehicle) followed by incubation with1ug/ml LPS or PBS (vehicle) for 4 h. The cellswere then harvested and assayed for D2 activity(mean � SEM). ***, P � 0.001; *, P � 0.05 byANOVA followed by Newman-Keuls.
Zeold et al. • NF-�B Responsive Elements in the Human dio2 Endocrinology, September 2006, 147(9):4419–4429 4421
5�FR (hdio2–117-Luc) (Fig. 2). These findings indicate thatthe proximal portion of the hdio2 5�FR contains the functionalp65 binding sites. Six NF-�B binding sites (no.1–6) werepredicted within this region (Fig. 3). The 901- and 584-bp Lucconstructs (hdio2–901-Luc and hdio2–584-Luc) showed nosignificant difference in p65 response (Fig. 2), indicating thatthe most potent binding sites responsible for p65-mediatedtranscriptional activation were located in the approximately600-bp proximal region. To test the NF-�B binding affinity,we subjected all six putative binding sites to EMSA includingsite no. 6, although it is located in the unresponsive 3� 117-bpregion.
No. 2 and no. 5 are the most potent NF-�B binding sites inhdio2 5�FR EMSA
We used EMSA and supershift assays to determine theNF-�B binding affinity of the six putative NF-�B binding
sites in this proximal approximately 0.9-kb hdio2 region (Fig.3). Nuclear extract of TNF-�-treated HeLa or HC11 cells wasincubated with double-stranded oligonucleotides labeledwith 32P (Table 1). As positive control we used the oligonu-cleotide containing a high-affinity NF-�B binding site (36),which showed two shifted bands in accordance with thefindings of the gel shift assay system (Promega) (Fig. 4, lane2) and no band in probes prepared without nuclear extract(Fig. 4, lane 1). The shifted bands were abolished by an excessof unlabeled control DNA (Fig. 4, lane 4), but only the lowerband could be supershifted with an anti-p65 NF-�B antibody(Fig. 4, lane 3, open arrow), confirming the p65-DNA binding.Based on these findings, we considered the lower shiftedband of the control probe to represent the authentic p65-DNA complex (Fig. 4, solid arrow). All six putative NF-�Bbinding sites in the 0.9-kb 5�FR of the hdio2 promoter regionwere tested for NF-�B binding by EMSA. Equal counts per
FIG. 2. Induction of human dio2 5�FR Luc constructs after cotransfection with a p65-expressing construct or a CMV containing empty vectorin HC11 cells. The 5� and 3� positions of the 5�FR regions in each construct are indicated in relation to hdio2 TSS. Data are shown as the mean �SEM of the Luc to Ren ratios of at least three separate experiments. *, P � 0.001 vs. hdio2–6.9-Luc by ANOVA followed by Newman-Keuls; �,nonsignificant difference.
FIG. 3. Schematic map of the proximal1-kb region of the human dio2 5�FR withsix putative NF-�� binding sites as in-dicated by TESS (TESS search) (35). Theposition of hdio2 5�FR fragments testedin promoter assay related to the putativebinding sites is indicated.
4422 Endocrinology, September 2006, 147(9):4419–4429 Zeold et al. • NF-�B Responsive Elements in the Human dio2
minute activity of each probe was loaded to make signalstrengths comparable. In the presence of the nuclear extract,the hdio2 site no. 1 showed a faint band, whereas no. 2 andno. 5 showed an intense shifted band migrating similarly tothe lower band of the positive control (Fig. 4, lanes 5, 7, and
13, respectively, solid arrow). The shifted band could be abol-ished by adding unlabeled control DNA (Fig. 4, lanes 6, 8,and 14). The specific bands of all three sites disappearedon incubation with an NF-�B antibody, whereas the su-pershifted band (open arrow) was visible only in the pres-
FIG. 4. Study of binding of NF-�� tothe labeled six putative NF-�� bindingsites of the proximal 901 bp of the hu-man dio2 5�FR by EMSA in the absence(lane 1) or presence (lanes 2–16) of nu-clear extract of TNF-�-treated HeLacells. As a positive control (C), an NF-��binding site with consensus sequence(see Table 1, positive control) was used.The positive control was supershiftedwith an anti-p65 NF-�� antibody(s.shift). Unlabeled control DNA wasused as competitor (comp.). Solid ar-row, Specific NF-��/ DNA complex;open arrow, supershifted control DNA.The inset shows the long run of theframed 5 and 6 lanes.
FIG. 5. Study of binding of NF-�� to the no. 2 (A) or no. 5 (B) NF-�� binding site of the human dio2 5�FR. A, The labeled site no. 2 containingDNA was incubated in the presence of nuclear extract of TNF-� untreated (lane 1) or treated (lane 2–7) HeLa cells, unlabeled competitor control(comp.C; lane 3), mutated no. 2 DNA (lane 4), or supershifted (s.shift) with an anti-p65 NF-�� antibody (lane 5). Labeled DNA containing mutatedsite no. 2 was incubated in the presence of nuclear extract of TNF-�-treated HeLa cells (lanes 6 and 7). Mutations introduced to no. 2 are indicatedin Table 1. Solid arrow, specific NF-��-DNA complex; open arrow, supershifted control DNA. B, The labeled site no. 5 containing DNA wasincubated in the presence of nuclear extract of TNF-�-treated HeLa (lanes 5 and 6) or HC11 cells (lanes 7 and 8). The no. 5-NF-�� complexwas supershifted with an anti-p65 NF-�� antibody (lanes 6 and 8). As a positive control (C), an NF-�� binding site with consensus sequence(see Table 1, positive control) was used (lanes 1–4) under conditions described for site no. 5. Solid arrow, Specific NF-��-DNA complex; openarrow, supershifted control DNA.
Zeold et al. • NF-�B Responsive Elements in the Human dio2 Endocrinology, September 2006, 147(9):4419–4429 4423
ence of the no. 2 and no. 5 probes (Fig. 5A, lane 5, and B,lanes 6 and 8, for the sites no. 2 and no. 5, respectively, notshown for no. 1).
To study the fast migrating intense band obtained with theno. 1 site (framed in Fig. 4, lane 5), we repeated lanes 5 and6 of Fig. 4 with longer run (see inset in Fig. 4). This indicatedthat the fast migrating band is a doublet (left lane of the inset),and the upper band of this could be clearly suppressed withconsensus competitor oligo (right lane of the inset).
In the presence of extract, the site no. 3 probe produced anintense band that migrated more slowly than the NF-�Bcomplex, the intensity of which could not be decreased byunlabeled control DNA (Fig. 4, lanes 9 and 10) or nuclearextract of TNF-� untreated HeLa cells (not shown), whereasno. 4 did not show any band corresponding the size of thepositive control (Fig. 4, lanes 11 and 12). The site no. 6 pro-duced a band of slightly bigger size than the upper band ofthe control, and it could be eliminated with unlabeled controlDNA (Fig. 4, lanes 15 and 16) but could not be supershifted(not shown).
To test the efficiency of TNF-� treatment on HeLa, the no.2 containing probe was incubated with nuclear extracts fromuntreated HeLa cells, resulting in no specific binding (Fig.5A, lane 1). To further analyze NF-�B binding affinity of thehdio2 promoter sequence at site no. 2, we generated muta-tions of this NF-�B binding site (Table 1, no. 2-MutA and no.2-MutB) and tested them by EMSA. Neither no. 2-MutA norno. 2-MutB showed NF-�B binding affinity, as evidenced bythe absence of the band (Fig. 5A, lanes 6 and 7) migratingsimilarly to the NF-�B-DNA complex (Fig. 5, solid arrow).Moreover, incubation of binding reaction complex with un-labeled version of no. 2-MutB oligonucleotide did not com-pete with labeled wild-type site no. 2 for NF-�B binding, asindicated by the presence of labeled NF-�B-DNA complex(Fig. 5A, lane 4).
The signals of shifted no. 2 and no. 5 probes were strongerthan those of no. 1 (Fig. 4), suggesting a higher NF-�B bind-ing affinity of the former two sites. To establish the affinityfor NF-�B binding of these three sites, we used semiquan-titative EMSA. Different amounts of no. 1, 2, and 5 unlabeledDNA were used to decrease binding of the positive controlprobe to the nuclear extract of HeLa cells treated with TNF-�(Fig. 6A). Quantitative analysis of this experiment is depictedin Fig. 6B. In line with the previous experiments (Figs. 4 and5), sites no. 2 and 5 interfered with binding of the positivecontrol to NF-�B more strongly than did no. 1. Although no.2 and 5 showed a very similar inhibitory potency, their bind-ing capacities were still weaker than that of the positivecontrol.
The no. 2 and 5 NF-�B binding sites of hdio2 showstrikingly different transactivation potency
The data presented above identified one weak (no. 1) andtwo strong (no. 2 and 5) NF-�B binding sites at the most 3�approximately 900 bp of the hdio2 5�FR. To test their trans-activation potency in response to p65, sites no. 1, 2, and 5were subjected to site-directed mutagenesis. Oligonucleo-tides used for generation of fragments for luciferase con-structs are listed in Table 2, and the positions of the generated
5�FR fragments in relation to the binding sites are depictedin Fig. 3. The mutant hdio2 fragments were inserted 5� to aD2 minimal promoter governing the luciferase reporter geneand tested in promoter assay in HC11 cells as describedabove.
The wild-type constructs hdio2–2,5wt-Luc and hdio2–1,2,5wt-Luc showed a 5- and 4-fold response to p65, respec-tively (Fig. 7). However, the construct harboring both no. 2and 5 mutations (hdio2–2,5Mut-Luc) remained unresponsiveto p65. The construct containing a mutation in binding siteno. 1, but not in sites no. 2 and 5 (hdio2–1Mut-2wt-5wt-Luc),was induced by p65 similarly to the wild-type, indicatingthat site no. 1 is not critical in NF-�B-induced transcription,in good accordance with the relatively weak NF-�B bindingof this site, as shown with EMSA. Interestingly, the hdio2–2Mut-5wt-Luc construct containing a mutant site no. 2 andwild-type no. 5 performed similarly to the hdio2–2,5wt-Luc.However, none of the constructs containing a mutated no. 5and wild-type no. 2 (hdio2–1Mut-2wt-5Mut-Luc; hdio2–2wt-5Mut-Luc) responded to p65. These data indicate that despitethe similar NF-�B binding affinity of no. 2 and 5 hdio2 sites,no. 5 is the primary NF-�B binding site of the hdio2 gene.Whereas the rat dio2 (rdio2) promoter completely lacks the no.2 binding site of the hdio2, the no. 5 site in the hdio2 is highlysimilar in both species, except that a T was replaced by C inthe rdio2 5�FR at �193 5� to the TSS. The rat-like site re-sponded to p65 similarly to the hdio2–2,5wt-Luc as assessedby promoter assay in HC11 cells (not shown).
The transactivation potency of no. 5 site was also assessedin the context of the approximately 6.9-kb-long hdio2 pro-moter in the presence of the p65 subunit of NF-�B. The no.5 site was mutated in hdio2–6.9-Luc, resulting in hdio2–6.9–5Mut-Luc. This construct contained the same no. 5 mutationas used above in the approximately 600-bp-long hdio2–2wt-5Mut-Luc. The hdio2–6.9–5Mut-Luc responded to p6514.815 � 2.24-fold (mean � sem, n � 6) in HC11 cells,whereas p65 induced the parallel assayed wild-type hdio2–6.9-Luc 219.3 � 26.2-fold (mean � sem, n � 5) (Fig. 8).Furthermore, the approximately 6.9-kb-long construct con-taining both mutated no. 2 and 5 sites (hdio2–6.9–2,5Mut-Luc) showed a 11.44 � 1.75 (mean � sem, n � 4)-fold re-sponse to p65 (hdio2–6.9–5Mut-Luc vs. hdio2–6.9-Luc P �0.001; hdio2–6.9–2,5Mut-Luc vs. hdio2–6.9-Luc, P � 0.001;hdio2–6.9–5Mut-Luc vs. hdio2–2,5Mut-6.9-Luc, P 0.05 byANOVA followed by Newman-Keuls). These results indicatethat the no. 5 site plays a key role in p65 response of the 6.9-kbhdio2 5�FR, whereas mutation of site no. 2 did not have anyadditional effect.
STAT3 and STAT5 did not induce human or ratdio2 transcription
Four putative STAT binding sites were indicated at posi-tions �5558, �3385, �1873, and �428 using TFSEARCH (37).We studied whether STAT5 and/or STAT3 acted as effectorsof inflammatory signals in D2 up-regulation. STAT5a re-sponsiveness of the hdio2–6.9-Luc was tested by promoterassay in COS-7 and HEK 293 cells. To produce an activephosphorylated STAT5, prolactin was added to the cellsalong with cotransfection of the prolactin receptor, as de-
4424 Endocrinology, September 2006, 147(9):4419–4429 Zeold et al. • NF-�B Responsive Elements in the Human dio2
scribed (31). Whereas the positive control construct, a rat�-casein promoter linked to luciferase readily responded toactivated STAT5 in both cell lines (6- and 5-fold, respec-tively), the hdio2 promoter showed no induction upon pro-
lactin mediated activation of cotransfected STAT5 (notshown). Furthermore, cotransfection of HC11 and HEK-293cells with a constitutively active STAT3c and hdio2–6.9-Lucor rdio2–3.6-Luc luciferase construct did not lead to tran-
FIG. 6. Semiquantitative EMSA for thecomparison of the affinities of NF-��binding to sites no. 1, 2, or 5 of the hdio25�FR. A, As a positive control (C), anNF-�� binding site with consensus se-quence was used (see Table 1, positivecontrol). Probe in first lane (no comp.)contained the labeled control probe andthe TNF-�-treated HeLa nuclear extractbut no unlabeled competitor oligo. Thenumbers above each lane are the unla-beled probe concentrations (nanomoles).Solid arrow, specific NF-��-DNA com-plex. B, Densitometric analysis of semi-quantitative EMSA shown in A estimat-ing the relative binding affinity of the no.1, 2, or 5 NF-�� binding sites of the hdio25�FR. Signal strength of each lane wascalculated as a percentage of the value inthe control lane (probe without compet-itor) and plotted against the amount ofthe added unlabeled competitor DNA.
TABLE 2. PCR primers used for generation of fragments for Luc constructs
Primer Orientation Sequence (5�–3�)
Bp381 (no. 1 wt) Sense atcgagctcAAAAAGGAAGATTCCTTAAAGTATBp382 (no. 1 Mut) Sense atcgagctcAAAAAGaAAGcTgCCTTAAAGTATBp383 (no. 2 wt) Sense atcgagctcGGAACTTGGATTTTTTTTTTTTTTTCATTAGAAGCTBp384 (no. 2 Mut) Sense atcgagctcGGAACTTGGATcTgTTTTTTTTTTTCATTAGAAGCTBp385 (no. 5 wt) Antisense ctagctagcGACAGTGGAATTTTGCTTTGTGATBp386 (no. 5 Mut) Antisense ctagctagcGACAGTGGcAgTgTtCTTTGTGATBp452 (no. 5 Mut) Sense ATCACAAAGaAcAcTgCCACTGTCBp402 (no. 5 RatMut) Antisense ctagctagcGACAGTGGgATTTTGCTTTGTGAT
Non-gene-specific tails or mutations are indicated in lowercase letters.
Zeold et al. • NF-�B Responsive Elements in the Human dio2 Endocrinology, September 2006, 147(9):4419–4429 4425
scriptional activation, whereas the GAS-Luc, a STAT3-re-sponsive luciferase construct used as positive control,showed a 3-fold induction (not shown).
Discussion
The present studies were performed to gain insight intothe mechanisms underlying an elevated D2 expression in thebrain during inflammation. Here we provide evidence thatthe transcription factor NF-�B, unlike STAT3 and STAT5,exerts a direct transcriptional effect on the human dio2 gene.This adds dio2 to the set of key genes regulated by NF-�Bessential for inflammatory processes and host defense in thecentral nervous system (20, 38). An important role of thispathway in other tissues should also be considered, as ex-posure to LPS significantly increased D2 activity in MSTO-211H mesothelioma cells by a mechanism that was partiallyblocked by the NF-�B inhibitor sulfasalazine (Fig. 1). Therelatively modest effect of sulfasalazine could be explainedby published data indicating that 0.1 mm sulfasalazine blocksNF-�B-mediated transcription by only approximately 20%(33).
A key player of NF-�B-mediated stimulation is the induc-tion by the heterodimer of p65 and p50 (39), and transienttransfection with p65 (RelA) is commonly used to test NF-�Bsensitivity (40). We studied p65 responsiveness of the humandio2 5�FR to assess the NF-�B responsiveness of differentregions of this gene. Promoter assays mapped the responseof the hdio2 5�FR to p65 to the proximal approximately 600bp using HC11 (Fig. 2), a D2-expressing mouse mammary
cell line (28) that has been successfully used to analyze theNF-�B pathway (29). Although upstream 5� regions of thepromoter remained unresponsive, the 600-bp region wassignificantly less responsive to p65 than the approximately6.9-kb fragment. This discrepancy could be explained by thepresence of an enhancer sequence in the more 5� region thatcould amplify the p65 response of hdio2.
Computer-assisted inspection/TESS (35) identified six pu-tative NF-�B binding sites at �878, �683, �546, �254, �198,and �111 in the proximal region of the hdio2 5�FR, indicatedin Fig. 3. as sites no. 1, 2, 3, 4, 5, and 6, respectively, showingsequence variability (Table 1). Various consensus sites ofNF-�B responsive regions have been identified, indicatingthe existence of a fine-tuning mechanism of the NF-�B sig-naling system (41–43). One explanation proposed for thewell-known phenomenon of variability among NF-�B bind-ing sites is the ability of NF-�B to modify its conformation(44). In accordance with this suggestion, the sequence dif-ferences between sites no. 2 and 5 did not alter NF-�B bindingaffinity as assessed by EMSA (Figs. 4, 6).
EMSA excluded site no. 3 as an NF-�B binding site basedon its lower migration speed, the independence of its shiftfrom TNF-� treatment of the cells used for preparing thenuclear extract and its resistance to suppression by the un-labeled NF-�B-positive control DNA in access (Fig. 4).
Site no. 4 was excluded due to the absence of a shifted bandmigrating similarly to the positive control, whereas no. 6 wasnot considered as an NF-�B binding site due to the lack ofsupershiftable NF-�B-DNA complex paralleled by unre-
FIG. 7. Effect of mutations in the no. 1, 2, or 5NF-�� binding sites in the context of the 3�region (see Fig. 3.) of the human dio2 5�FR. Lucconstructs were transiently coexpressed withp65 in HC11 cells. Data are shown as themean � SEM of the Luc to Ren ratios of p65 orCMV vector cotransfected cells in three sepa-rate experiments (mean � SEM). *, P � 0.05 vs.hdio2–2,5wt-Luc by ANOVA followed by New-man-Keuls.
FIG. 8. Effect of mutations in the no. 2, or no. 2 and 5 NF-�� binding sites in the context of the approximately 6.9-kb-long human dio2 5�FR.Luc constructs were transiently coexpressed with p65 in HC11 cells. Data are shown as the mean � SEM of the Luc to Ren ratios of p65 or CMVvector cotransfected cells in at least three separate experiments (mean � SEM). *, P � 0.001 vs. hdio2–6.9-Luc by ANOVA followed byNewman-Keuls; �, nonsignificant difference.
4426 Endocrinology, September 2006, 147(9):4419–4429 Zeold et al. • NF-�B Responsive Elements in the Human dio2
sponsiveness of this region in the promoter assay. In contrast,the no. 2 and 5 sites could be supershifted with an NF-�Bantibody (Fig. 5). The lack of supershift at site no. 1 is prob-ably due to the weak binding, also indicated by the semi-quantitative EMSA. Interestingly, whereas semiquantitativeEMSA indicated that the NF-�B binding of sites no. 2 and 5were very similar (Fig. 6), their transactivation by p65 wasstrikingly different, as shown by promoter assay using thehdio2–1,2,5wt-Luc and its derivatives, prepared to assess theeffect of point mutations on p65 responsiveness of hdio2. Thesemiquantitative EMSA also demonstrated that all sites stud-ied have shown a significantly lower affinity to NF-�B thanthe positive control containing a strong consensus NF-�Bsite.
The response to p65 of the approximately 600- to 800-bp-long hdio2–1,2,5wt-Luc and its derivatives was relativelylow, compared with that of hdio2–584-Luc (�5- vs. 15-fold),but their basal promoter activity remained unchanged (Fig.7). The former phenomenon could be explained by an un-identified enhancer in the region between �95 and �180, aregion absent from hdio2–1,2,5wt-Luc and its derivatives.Still, this system was suitable to clearly demonstrate theeffect of mutations introduced to sites no. 1, 2, or 5 in HC11cells. Whereas the no. 5 binding site was functional (becauseits mutation completely abolished transcriptional activationby p65), sites no. 1 and 2 turned out to be silent NF-�B sites.The upper band of the fast migrating doublet obtained withthe no. 1 site (Fig. 4, inset) might be a p50/p50 homodimerknown to be a repressor, as indicated for the �-casein gene(29). This potential repression along with weak p65 bindingto no. 1 might result in silence of this site. The lack of effectof no. 2 mutation on p65 response cannot be explained byretained transactivation potency of the mutagenized se-quence because EMSA confirmed the lack of binding ofNF-�B to the mutant site no. 2 (no. 2-MutB). These findingswere further supported by the lack of significant differencebetween the p65 responsiveness of hdio2–901-Luc andhdio2–584-Luc along with the unresponsiveness of con-structs containing a wild-type no. 2 and mutant no. 5 sites.
Because of the relatively low response of the approxi-mately 600- to 800-bp-long hdio2–1,2,5wt-Luc derivatives, wealso assessed the transactivation potency of no. 5 and 2 sitesin the context of the approximately 6.9-kb-long hdio2 pro-moter in the presence of the p65 subunit of NF-�B. Whereasmutation of the no. 5 site in hdio2–6.9–5Mut-Luc decreasedthe p65 response of the wild-type 6.9 kb hdio2 promoter byapproximately 93%, the additional mutation of no. 2 site(hdio2–6.9–2,5Mut-Luc) did not result in an additional de-crease indicating that presence of an intact no. 5 site wasinevitable for the robust response of the hdio2 gene to NF-�B,whereas the no. 2 site was not (Fig. 8). These data also dem-onstrate that there is no significant synergism between theno. 2 site and more 5� sites. The retaining approximately 7%response observed in the presence of a mutated no. 5 (and no.2 site) sites of the 6.9-kb-long 5� hdio2 flanking region couldbe the result of weak NF-�B binding sites potentiating eachother. This process then requires the whole 6.9-kb hdio2 5�FRbecause the putative NF-�B binding sites in the �6860 to�2079 region tested as hdio2–5�Luc (�6860 to �3880) andhdio2-PstI Luc (�3880 to �2079) remained unresponsive to
p65 in the promoter assay along with the most 3� 117-bpregion (hdio2–117-Luc) (Fig. 2). It is also possible that theretaining 7% response could be the indirect result of otherfactors induced by NF-�B. Together, these data indicate thatsite no. 5 is a major element in the response of hdio2 to p65.
Discrepancies are common between the binding affinity ofa site to a specific transcription factor and the transactivationpotency of this factor on the same binding site. For example,the C1 and D TTF-1 binding sites of the hdio2 5�FR aresimilarly involved in TTF-1 response, whereas their bindingcapacities were very different (23). The contradiction be-tween NF-�B binding and transactivation potency at sites no.2 and 5 (which is more proximal to the TSS) could also arisefrom a positional effect because the linear arrangement oftranscription factors and the three-dimensional organizationof enhanceosomes are critical for efficient transcription, assuggested for NF-�B binding sites in the promoter of thehuman I�B kinase-related kinase IKKi/IKK� gene (45) andby comparing binding sites HIV-�B and IFN�-�B (44).
To put the no. 2 and 5 hdio2 binding sites into phylogeneticcontext, we analyzed the rat dio2 5�FR and found that itcompletely lacked the no. 2 binding site, whereas site no. 5of hdio2 was highly similar in both species, showing a C-for-Treplacement in rdio2 5�FR at position �193 relative to therdio2 TSS. The same point mutation is present in the corre-sponding site of the mouse dio2 5�FR (46). The response to p65of the C/T mutated site no. 5 was similar to that of thehdio2–2,5wt-Luc, as assessed by promoter assay in HC11cells, reflecting the evolutionary conservation of the mostpotent p65 binding site of the dio2 gene.
D2 is normally expressed in astroglial cells in the brain(47), and it is the only deiodinase that generates T3 in thehuman central nervous system (3). We used the U87 humanglioma cell line without endogenous D2 expression (23) todetermine whether p65 can induce D2 expression in a glialenvironment. The approximately 55-fold induction of thehdio2–6.9-Luc in the presence of coexpressed p65 in U87indicates that this mechanism could be functional in thebrain.
LPS-induced D2 activation in the tanycytes of the rat me-diobasal hypothalamus indicated a mechanism culminatingin increased T3 generation, providing local negative feedbackin the hypothalamus and accounting for the impaired re-sponse of the hypothalamo-pituitary-thyroid axis observedin nonthyroidal illness syndrome (12). Tanycytes probablyserve as a cytoplasmic passage between the cerebrospinalfluid and the blood circulating in the area of the arcuatenucleus and median eminence (48). Thus, changes in D2-catalyzed T3 generation in tanycytes could have importantconsequences. Recent studies have connected this systemwith the nonthyroidal illness syndrome because LPS-treatedrats showed an enhanced D2 expression in tanycytes (12),increasing local T3 generation, which could suppress thehypothalamo-pituitary-thyroid axis.
Although the upstream signaling components of theNF-�B pathway have not yet been resolved for D2-express-ing cells in the brain, indirect evidence suggests that NF-�Bmediated up-regulation of D2 expression could be functionalin the mediobasal hypothalamus. It has been shown that thecomplex LPS-LPS binding protein binds CD14 at the cell
Zeold et al. • NF-�B Responsive Elements in the Human dio2 Endocrinology, September 2006, 147(9):4419–4429 4427
surface of myeloid cells followed by the association of theLPS/CD14 receptor complex with TLR4. This is the crucialsignal transducer that mediates the effect of LPS on NF-�Bactivation via activation of a complex cascade of cytoplasmicsignaling molecules (20, 49). Signaling via TLR4 hence resultsin an up-regulation of IFN-inducible genes via the adaptorprotein TICAM-1 (Toll-IL-1 receptor domain-containingadaptor molecule) or TRIF (TIR domain-containing adaptor-inducing IFN-�) (50). In addition, TLR4 can signal viaanother adaptor protein, TRAM (TRIF-related adaptor mol-ecule), and can induce a late NF-�B response (51). Impor-tantly, both key components of LPS/TLR4 signaling (i.e.CD14 and TLR4 mRNA) have been detected in the medianeminence of the mediobasal hypothalamus (21). It has alsobeen proven that systemic injection of LPS can up-regulateCD14 in the median eminence (52).
Alternatively (or parallel to this mechanism), the role ofTNF-� should also be taken into account because its expres-sion can be both the consequence and inducing factor thattriggers NF-�B activation (20, 53). Importantly, the p55 TNF-Ireceptor is expressed in the median eminence and its level(like that of CD14), is increased by TNF-� (54, 55).
Based on these findings, we speculate that LPS acts on D2expression in tanycytes directly via the CD14/TLR4 complexor indirectly via TNF-� released locally, followed in bothcases by NF-�B activation, which exerts a direct transcrip-tional effect on the dio2 gene.
The importance of STAT proteins in the brain during in-flammation has also been established (53), and members ofthe STAT family share a relatively similar binding site (56).To look for functional STAT binding sites in the hdio2 pro-moter, we used STAT5 and STAT3 as members of this proteinfamily that can be effectively activated in the used heterol-ogous expression system. Neither STAT3 nor STAT5 couldactivate the hdio2 promoter in the cell lines we used. Becausespecies-specific responsiveness of the dio2 gene has beenalready described (23), we extended this study to the rat dio2promoter. Again, no response was detected, indicating thatthese effectors are not relevant for D2 regulation in the sys-tems studied.
In conclusion, our findings show that NF-�B exerts a directtranscriptional effect on the human dio2 gene via the no. 5binding site and that this effect also exists in glial cells. Thepresented data demonstrate that NF-�B could be a majorfactor in the regulation of D2 activity by infection. This isparticularly relevant, given the fact that D2 is the primarysource of extrathyroidal T3 generation in the euthyroid hu-man (57). Other factors such as decreased D1 (58) or in-creased D3 activities (59) have also been reported in patientswith nonthyroidal illness syndrome, and tissue-specificchanges of D2 expression during infection should be alsotaken into account, indicating that further studies are neededto clarify the individual contributions of these enzymes to thepathogenesis of this syndrome.
Acknowledgments
We thank Dr. W. Doppler for helpful comments. The technical helpof Mrs. Gy. Kekesi is gratefully acknowledged.
Received December 20, 2005. Accepted May 15, 2006.
Address all correspondence and requests for reprints to: Dr. BalazsGereben, Institute of Experimental Medicine, Laboratory of EndocrineNeurobiology, Szigony u. 43, Budapest H-1083 Hungary. E-mail:[email protected].
This work was supported by Hungarian Scientific Research FundGrants OTKA T049081 (to B.G.), T049015 (to I.K.), and T046492 (to C.F.);Medical Research Council Grant ETT 481/2003 (to B.G.); and NationalInstitutes of Health Grant TW006467 and DK65055 (to A.C.B.).
The results of this work were presented in part at the 13th Interna-tional Thyroid Congress, Buenos Aires, Argentina, 2005.
The authors have nothing to declare.
References
1. Bianco AC, Salvatore D, Gereben B, Berry MJ, Larsen PR 2002 Biochemistry,cellular and molecular biology and physiological roles of the iodothyronineselenodeiodinases. Endocr Rev 23:38–89
2. Gereben B, Salvatore D 2005 Pretranslational regulation of type 2 deiodinase.Thyroid 15:855–864
3. Campos-Barros A, Hoell T, Musa A, Sampaolo S, Stoltenburg G, Pinna G,Eravci M, Meinhold H, Baumgartner A 1996 Phenolic and tyrosyl ring io-dothyronine deiodination and thyroid hormone concentrations in the humancentral nervous system. J Clin Endocrinol Metab 81:2179–2185
4. Guadano-Ferraz A, Obregon MJ, St Germain DL, Bernal J 1997 The type 2iodothyronine deiodinase is expressed primarily in glial cells in the neonatalrat brain. Proc Natl Acad Sci USA 94:10391–10396
5. Tu HM, Kim SW, Salvatore D, Bartha T, Legradi G, Larsen PR, Lechan RM1997 Regional distribution of type 2 thyroxine deiodinase messenger ribonu-cleic acid in rat hypothalamus and pituitary and its regulation by thyroidhormone. Endocrinology 138:3359–3368
6. Gereben B, Pachucki J, Kollar A, Liposits Z, Fekete C 2004 Ontogenic re-distribution of type 2 deiodinase messenger ribonucleic acid in the brain ofchicken. Endocrinology 145:3619–3625
7. Alkemade A, Friesema EC, Unmehopa UA, Fabriek BO, Kuiper GG, LeonardJL, Wiersinga WM, Swaab DF, Visser TJ, Fliers E 2005 Neuroanatomicalpathways for thyroid hormone feedback in the human hypothalamus. J ClinEndocrinol Metab 90:4322–4334
8. Leonard JL, Kaplan MM, Visser TJ, Silva JE, Larsen PR 1981 Cerebral cortexresponds rapidly to thyroid hormones. Science 214:571–573
9. Escobar-Morreale HF, Obregon MJ, Escobar del Rey F, Morreale de EscobarG 1995 Replacement therapy for hypothyroidism with thyroxine alone doesnot ensure euthyroidism in all tissues, as studied in thyroidectomized rats.J Clin Invest 96:2828–2838
10. Broedel O, Eravci M, Fuxius S, Smolarz T, Jeitner A, Grau H, Stoltenburg-Didinger G, Plueckhan H, Meinhold H, Baumgartner A 2003 Effects of hyper-and hypothyroidism on thyroid hormone concentrations in regions of the ratbrain. Am J Physiol Endocrinol Metab 285:E470–E480
11. Lechan RM, Fekete C 2005 Role of thyroid hormone deiodination in thehypothalamus. Thyroid 15:883–897
12. Fekete C, Gereben B, Doleschall M, Harney JW, Dora JM, Bianco AC, SarkarS, Liposits Z, Rand W, Emerson C, Kacskovics I, Larsen PR, Lechan RM 2004Lipopolysaccharide induces type 2 iodothyronine deiodinase in the medio-basal hypothalamus: implications for the nonthyroidal illness syndrome. En-docrinology 145:1649–1655
13. Fekete C, Sarkar S, Christoffolete MA, Emerson CH, Bianco AC, Lechan RM2005 Bacterial lipopolysaccharide (LPS)-induced type 2 iodothyronine deio-dinase (D2) activation in the mediobasal hypothalamus (MBH) is independentof the LPS-induced fall in serum thyroid hormone levels. Brain Res 1056:97–99
14. Boelen A, Kwakkel J, Thijssen-Timmer DC, Alkemade A, Fliers E, WiersingaWM 2004 Simultaneous changes in central and peripheral components of thehypothalamus-pituitary-thyroid axis in lipopolysaccharide-induced acute ill-ness in mice. J Endocrinol 182:315–323
15. Medzhitov R, Preston-Hurlburt P, Janeway Jr CA 1997 A human homologueof the Drosophila Toll protein signals activation of adaptive immunity. Nature388:394–397
16. Schnare M, Barton GM, Holt AC, Takeda K, Akira S, Medzhitov R 2001Toll-like receptors control activation of adaptive immune responses. NatImmunol 2:947–950
18. Luster AD 2002 The role of chemokines in linking innate and adaptive im-munity. Curr Opin Immunol 14:129–135
19. Hanada T, Yoshimura A 2002 Regulation of cytokine signaling and inflam-mation. Cytokine Growth Factor Rev 13:413–421
20. Li Q, Verma IM 2002 NF-�B regulation in the immune system. Nat RevImmunol 2:725–734
21. Laflamme N, Rivest S 2001 Toll-like receptor 4: the missing link of the cerebralinnate immune response triggered by circulating gram-negative bacterial cellwall components. FASEB J 15:155–163
22. Levy DE, Darnell Jr JE 2002 Stats: transcriptional control and biological im-pact. Nat Rev Mol Cell Biol 3:651–662
4428 Endocrinology, September 2006, 147(9):4419–4429 Zeold et al. • NF-�B Responsive Elements in the Human dio2
23. Gereben B, Salvatore D, Harney JW, Tu HM, Larsen PR 2001 The Human,but not rat, dio2 gene is stimulated by thyroid transcription factor-1 (TTF-1).Mol Endocrinol 15:112–124
24. Gobin SJ, Biesta P, Van den Elsen PJ 2003 Regulation of human �2-micro-globulin transactivation in hematopoietic cells. Blood 101:3058–3064
25. Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, Pestell RG, AlbaneseC, Darnell Jr JE 1999 Stat3 as an oncogene. Cell 98:295–303
26. Harris M, Aschkenasi C, Elias CF, Chandrankunnel A, Nillni EA, BjoorbaekC, Elmquist JK, Flier JS, Hollenberg AN 2001 Transcriptional regulation ofthe thyrotropin-releasing hormone gene by leptin and melanocortin signaling.J Clin Invest 107:111–120
27. Wolff B, Naumann M 1999 INK4 cell cycle inhibitors direct transcriptionalinactivation of NF-�B. Oncogene 18:2663–2666
28. Song S, Oka T 2003 Regulation of type II deiodinase expression by EGF andglucocorticoid in HC11 mouse mammary epithelium. Am J Physiol EndocrinolMetab 284:E1119–E1124
29. Geymayer S, Doppler W 2000 Activation of NF-�B p50/p65 is regulated in thedeveloping mammary gland and inhibits STAT5-mediated beta-casein geneexpression. FASEB J 14:1159–1170
30. Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman D, Demeneix B,Behr JP 1995 A versatile vector for gene and oligonucleotide transfer into cellsin culture and in vivo: polyethylenimine. Proc Natl Acad Sci USA 92:7297–7301
31. Mayr S, Welte T, Windegger M, Lechner J, May P, Heinrich PC, Horn F,Doppler W 1998 Selective coupling of STAT factors to the mouse prolactinreceptor. Eur J Biochem 258:784–793
32. Curcio C, Baqui MM, Salvatore D, Rihn BH, Mohr S, Harney JW, Larsen PR,Bianco AC 2001 The human type 2 iodothyronine deiodinase is a selenoproteinhighly expressed in a mesothelioma cell line. J Biol Chem 276:30183–30187
36. Lenardo MJ, Baltimore D 1989 NF-�B: a pleiotropic mediator of inducible andtissue-specific gene control. Cell 58:227–229
37. Heinemeyer T, Wingender E, Reuter I, Hermjakob H, Kel AE, Kel OV,Ignatieva EV, Ananko EA, Podkolodnaya OA, Kolpakov FA, PodkolodnyNL, Kolchanov NA 1998 Databases on transcriptional regulation: TRANSFAC,TRRD, and COMPEL. Nucleic Acids Res 26:364–370
38. Meffert MK, Baltimore D 2005 Physiological functions for brain NF-�B.Trends Neurosci 28:37–43
39. Urban MB, Schreck R, Baeuerle PA 1991 NF-�B contacts DNA by a het-erodimer of the p50 and p65 subunit. EMBO J 10:1817–1825
40. Molinero LL, Fuertes MB, Girart MV, Fainboim L, Rabinovich GA, CostasMA, Zwirner NW 2004 NF-�B regulates expression of the MHC class I-relatedchain A gene in activated T lymphocytes. J Immunol 173:5583–5590
41. Baeuerle PA, Baltimore D 1988 I�B: a specific inhibitor of the NF-�B tran-scription factor. Science 242:540–546
42. Kunsch C, Ruben SM, Rosen CA 1992 Selection of optimal �B/Rel DNA-binding motifs: interaction of both subunits of NF-�B with DNA is requiredfor transcriptional activation. Mol Cell Biol 12:4412–4421
43. Natoli G, Saccani S, Bosisio D, Marazzi I 2005 Interactions of NF-�B withchromatin: the art of being at the right place at the right time. Nat Immunol6:439–445
44. Berkowitz B, Huang DB, Chen-Park FE, Sigler PB, Ghosh G 2002 The x-raycrystal structure of the NF-�B p50.p65 heterodimer bound to the interferon�-�B site. J Biol Chem 277:24694–24700
45. Wang N, Ahmed S, Haqqi TM 2005 Genomic structure and functional char-acterization of the promoter region of human I�B kinase-related kinase IKKi/IKK� gene. Gene 353:118–133
46. Song S, Adachi K, Katsuyama M, Sorimachi K, Oka T 2000 Isolation andcharacterization of the 5�-upstream and untranslated regions of the mouse typeII iodothyronine deiodinase gene. Mol Cell Endocrinol 165:189–198
47. Leonard JL 1988 Dibutyryl cAMP induction of type II 5�deiodinase activity inrat brain astrocytes in culture. Biochem Biophys Res Commun 151:1164–1172
48. Krisch B, Leonhardt H, Buchheim W 1978 The functional and structuralborder between the CSF- and blood-milieu in the circumventricular organs(organum vasculosum laminae terminalis, subfornical organ, area postrema)of the rat. Cell Tissue Res 195:485–497
49. Wright SD, Ramos RA, Tobias PS, Ulevitch RJ, Mathison JC 1990 CD14, areceptor for complexes of lipopolysaccharide (LPS) and LPS binding protein.Science 249:1431–1433
50. Oshiumi H, Matsumoto M, Funami K, Akazawa T, Seya T 2003 TICAM-1, anadaptor molecule that participates in Toll-like receptor 3-mediated interfer-on-� induction. Nat Immunol 4:161–167
51. Yamamoto M, Sato S, Hemmi H, Uematsu S, Hoshino K, Kaisho T, TakeuchiO, Takeda K, Akira S 2003 TRAM is specifically involved in the Toll-likereceptor 4-mediated MyD88-independent signaling pathway. Nat Immunol4:1144–1150
52. Lacroix S, Feinstein D, Rivest S 1998 The bacterial endotoxin lipopolysac-charide has the ability to target the brain in upregulating its membrane CD14receptor within specific cellular populations. Brain Pathol 8:625–640
53. Rivest S 2001 How circulating cytokines trigger the neural circuits that controlthe hypothalamic-pituitary-adrenal axis. Psychoneuroendocrinology 26:761–788
54. Nadeau S, Rivest S 1999 Effects of circulating tumor necrosis factor on theneuronal activity and expression of the genes encoding the tumor necrosisfactor receptors (p55 and p75) in the rat brain: a view from the blood-brainbarrier. Neuroscience 93:1449–1464
55. Nadeau S, Rivest S 2000 Role of microglial-derived tumor necrosis factor inmediating CD14 transcription and nuclear factor �B activity in the brain duringendotoxemia. J Neurosci 20:3456–3468
56. Ehret GB, Reichenbach P, Schindler U, Horvath CM, Fritz S, Nabholz M,Bucher P 2001 DNA binding specificity of different STAT proteins. Compar-ison of in vitro specificity with natural target sites. J Biol Chem 276:6675–6688
57. Luiza Maia A, Kim BW, Huang SA, Harney JW, Larsen PR 2005 Type 2iodothyronine deiodinase is the major source of plasma T(3) in euthyroidhumans. J Clin Invest 115:2524–2533
58. Nagaya T, Fujieda M, Otsuka G, Yang JP, Okamoto T, Seo H 2000 A potentialrole of activated NF-�B in the pathogenesis of euthyroid sick syndrome. J ClinInvest 106:393–402
59. Boelen A, Kwakkel J, Alkemade A, Renckens R, Kaptein E, Kuiper G,Wiersinga WM, Visser TJ 2005 Induction of type 3 deiodinase activity ininflammatory cells of mice with chronic local inflammation. Endocrinology146:5128–5134
Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving theendocrine community.
Zeold et al. • NF-�B Responsive Elements in the Human dio2 Endocrinology, September 2006, 147(9):4419–4429 4429
