NF-κB Essential Modulator (NEMO) Is Critical for Thyroid Function*

The I-κB kinase (IKK) subunit NEMO/IKKγ (NEMO) is an adapter molecule that is critical for canonical activation of NF-κB, a pleiotropic transcription factor controlling immunity, differentiation, cell growth, tumorigenesis, and apoptosis. To explore the functional role of canonical NF-κB signaling in thyroid gland differentiation and function, we have generated a murine strain bearing a genetic deletion of the NEMO locus in thyroid. Here we show that thyrocyte-specific NEMO knock-out mice gradually develop hypothyroidism after birth, which leads to reduced body weight and shortened life span. Histological and molecular analysis indicate that absence of NEMO in thyrocytes results in a dramatic loss of the thyroid gland cellularity, associated with down-regulation of thyroid differentiation markers and ongoing apoptosis. Thus, NEMO-dependent signaling is essential for normal thyroid physiology.

The nuclear factor-B (NF-B) signaling pathway controls a variety of important biological functions, including immune and inflammatory responses, differentiation, cell growth, tumorigenesis, and apoptosis (1). Two distinct pathways of NF-B activation have been reported. The classical, canonical pathway is found in virtually all mammalian cells and depends on the presence of the IKK␥/NF-B essential modulator (NEMO) 3 protein (1). In this pathway, NF-B activation is mediated by the IKK complex, consisting of the IKK1/IKK␣ and IKK2/IKK␤ catalytic kinase subunits and the NEMO regulatory protein (1). Basically, NF-B transcription factors are kept inactive in the cytoplasm through binding to members of the I-B family of inhibitory proteins. Cell activation by a variety of stimuli results in the IKK-dependent phosphorylation of I-B proteins, followed by their polyubiquitination and their proteasome-dependent degradation, allowing NF-B dimers to enter the nucleus and catalyze transcription of target genes (1).
In specific lymphoid tissues cells, an alternative, non-canonical pathway has been described that relies on IKK␣-mediated phosphorylation of I-B molecules. This pathway seems to be independent of NEMO activity (1). Recent publications have also shown that diverse posttranslational modifications, including ubiquitination, sumoylation, and phosphorylation, regulate the function of NEMO in the IKK complex (2)(3)(4)(5)(6).
Gene-targeting experiments have shown that mice lacking p65/RELA, IKK2/IKK␤, or NEMO die during embryonic development due to liver apoptosis (7)(8)(9)(10)(11)(12). The evidence that NEMO-deficient mice exhibit a phenotype similar to that of p65/RELA-deficient mice demonstrates the essential role of NEMO in the canonical NF-B signaling. Experiments based on tissue-and/or organ-specific deletion of NEMO, obtained through Cre-mediated genetic recombination, have given diverse results regarding the physiological role of this protein.
In fact, ablation of NEMO in liver parenchymal cells causes spontaneous development of steatohepatitis and hepatocellular carcinoma (13). Instead, intestinal epithelial cell-specific deletion of NEMO results in severe chronic intestinal inflammation due to apoptosis of colonic epithelial cells, impaired expression of antimicrobial peptides, and translocation of bacteria into the mucosa (14). Central nervous system-specific ablation of NEMO results in no apparent abnormalities but rather ameliorates inflammatory and autoimmune pathologies (15). NEMO inactivation in the heart did not affect embryonic cardiac development but led to spontaneous and progressive impairment of cardiac function, progression to dilated cardiomyopathy, and heart failure (16). Finally, genetic abrogation of NEMO in podocytes did not affect normal glomerular development and function under non-stressed conditions (17). Thus, it appears that canonical NF-B signaling performs different roles and functions, depending on the type of tissue and organ.
In the present work, we have investigated the requirement for NEMO in thyroid development, differentiation, and function by generating a mouse model bearing a thyroid-specific genetic inactivation of NEMO. In these mice, development and differentiation of the thyroid gland appears normal. In contrast, after birth, thyroid-specific NEMO knock-out mice develop progressive loss of thyroid cellularity and thyroid markers, hypothyroidism, and reduced vitality due to extensive apopto-sis of thyroid cells. Thus, our data indicate that NF-B-dependent gene expression is essential for maintaining normal thyroid gland structure and function in the adult.

Experimental Procedures
Ethics-Procedures involving animals were conducted as indicated in the Italian National Guidelines (D.L. 100/2006 and D.L. 116/1992) and in the pertinent European Directives (EEC Council Directive 86/609, 1.12.1987), adhering to the Guide for the Care and Use of Laboratory Animals (United States National Research Council). All of the in vivo experimental activities were approved by the Animal Ethics Committee of Biogem (Italy) with ID number 4713.
Generation of NEMO TS-KO Mice-To inactivate the NEMO gene in thyroid, mice expressing Cre recombinase under the control of endogenous Pax8 promoter (18) and NEMO Flox/ϩ (12) were bred. NEMO Flox/Y /Pax8 Cre/ϩ mice were named NEMO TS-KO and used as the experimental group, whereas NEMO ϩ/Y /Pax8 Cre/ϩ littermates were used as controls. Genotyping was performed by PCR analysis of a tail biopsy. All experiments were conducted on age-and gender-matched animals.
Metabolic Measurements-Renal parameters were evaluated in 12-month-old mice using metabolic cages as described previously (19). Mice were housed individually in metabolic cages for 5 days at 23°C with a 12-h dark/light cycle. After 4 days of adjustment, parameters were registered on day 5. 24-h urine output was collected under mineral oil to prevent evaporation. Proteinuria was quantified by a Bradford assay, and urinary electrolytes and creatinine were evaluated using Vitrovet (Scil).
Histology and Immunohistochemistry-Mice were anesthetized by isoflurane and perfused through abdominal aorta with 4% paraformaldehyde. Blood, left kidney, and thyroid were collected before perfusion. The left kidney and one thyroid lobe were used for immunoblotting or PCR, whereas the right kidney and the other thyroid lobe were used for immunohistochemistry. After embedding in paraffin, 4-m-thick sections were stained with hematoxylin and eosin (Sigma-Aldrich) or with Masson's trichrome kit (Bio-Optica). For immunohistochemistry, sections were incubated overnight in xylene and then progressively in ethanol solution (99-96-70%). Endogenous peroxidase activity was quenched with 35% H 2 O 2 in methanol, and target retrieval was performed in TEG buffer, pH 9.2. Primary antibodies were incubated overnight at 4°C, and sections were then incubated with secondary anti-rabbit HRPconjugated antibody (Dako). Chromogenic reactions were carried out with 3,3Ј-diaminobenzidine (Vector Laboratories), and stained sections were mounted with Eukitt (Bio-Optica).
For immunofluorescence analysis, thyroid sections were incubated overnight at 4°C with anti-NIS antibody (kindly provided by Prof. M. De Felice). Sections were then incubated with secondary anti-rabbit-conjugated Alexa Fluor 488; DAPI was used as a nuclear counterstain. The primary antibodies used for kidney staining were as follows: anti-AQP2 and anti-NKCC2 provided by Prof. Frische; anti-Pax8, anti-TITF1, and anti-thyroglobulin provided by Prof. De Felice; and anti-cleaved caspase-3 purchased from Cell Signaling Technology. A Zeiss Axioplan 2 microscope was used for image acquisition. All stainings were done at least three times using different biological material as sources.
Culture of Primary Mouse Thyrocytes-Mice of different genotypes were anesthetized and sacrificed. Thyroid lobes were dissected aseptically and placed on a microscope slide containing a drop of Eagle's minimum essential medium. The lobes were disrupted mechanically, using two 25-gauge needles to obtain approximately 10 fragments from each lobe. The fragments were transferred to a 1.5-ml tube containing 1 ml of digestion medium, consisting of 112 units/ml type I collagenase (Sigma) and 1.2 units/ml dispase dissolved in Eagle's minimum essential medium. The enzymatic digestion was carried out for 45 min in a 37°C water bath. The follicles were seeded and cultured in F-12 medium (EuroClone) supplemented with 10% Nu-Serum IV (BD Biosciences), 10 ng/ml somatostatin (Sigma-Aldrich), and 2 ng/ml glycyl-L-histidyl-lysine acetate (Sigma-Aldrich) in a water-saturated incubator, under 95% air, 5% CO 2 at 37°C. Thyroid-stimulating hormone (TSH) and TNF␣ used for thyrocyte stimulation were obtained from Sigma-Aldrich.
Free T4 and TSH Measurement-Venous blood samples were collected in microtubes without anticoagulant. After clot formation, samples were centrifuged, and the serum fraction was kept at Ϫ80°C. Free T4 was measured using the ELISA kit from DiaMetra according to the manufacturer's instructions. TSH levels were determined by a rat TSH radioimmunoassay kit (Institute of Isotopes, Budapest, Hungary).
Statistics-Data were analyzed by Student's t test. A p value of Ͻ0.05 was considered significant.

Results
In order to understand the role of NEMO in thyroid development and function, we applied the Cre-LoxP strategy to specifically delete the NEMO locus in the thyroid gland. For this, females bearing a floxed Nemo allele (Nemo Flox/ϩ ), which do not show phenotype abnormalities (12), were crossed to the knock-in Pax8 Cre/ϩ mice, which express the Cre recombinase under the control of the endogenous Pax8 gene promoter and mediate efficient Cre recombination in thyroid (18). Mutant mice were observed at the expected Mendelian frequency and were named NEMO thyroid-specific knock-out (NEMO TS-KO ). As shown in Fig. 1A, site-specific recombination of the NEMO allele was observed only in NEMO floxed mice expressing the Cre recombinase. Efficient ablation of the NEMO locus was confirmed by immunoblotting analysis on whole-thyroid extracts (Fig. 1B). Functional inactivation of the IKK complex was demonstrated by lack of I-B␣ degradation in NEMO TS-KO thyrocytes following exposure to TNF␣ (Fig. 1C, left). On the other hand, expression of p65, IKK␣/␤, and p50 were not affected by genetic deletion of NEMO (Fig. 1C, right). Because Pax8 Cre/ϩ mice express Cre recombinase also in the mesonephros and metanephros (18), excision of the Nemo floxed allele was consistently detected also in kidney tissues (Fig. 1, D and E).
NEMO TS-KO mutant mice appeared undistinguishable from control mice at birth (data not shown). In contrast, starting from 2 months of age, mutant mice developed clearly smaller than control littermates and displayed a significantly reduced body weight ( Fig. 2A). NEMO TS-KO mice had a significantly shortened life span, and about 50% of mice die before 8 months of age (Fig. 2B). When compared with control mice, NEMO TS-KO mice presented significantly lower levels of serum-free T4 hormone (Fig. 2C) and increased TSH levels (Fig. 2D), which is indicative of hypothyroidism. Hence, body weight reduction and early lethality observed in these mice might be due, at least in part, to thyroid dysfunction.
To exclude the possibility that the premature lethality observed in NEMO TS-KO mice was due to renal failure, we carried out a histological and functional analysis on the kidneys of the mutant mice. The result of this analysis, shown in Table 1, indicates that partial loss of NEMO expression in specific kid-ney tissues was not associated with an overt kidney phenotype. In fact, monitoring of the renal function in NEMO TS-KO mice did not reveal any physiological alteration (Table 1). Kidney histology was assessed by H&E staining and immunohistochemistry for the kidney-specific markers AQP2 and NKCC2. Again, no alteration in immunoreactivity for these markers was detected in NEMO TS-KO kidneys with respect to control mice at 12 months of age (Fig. 3, A-C). From these results, we deduced that NEMO TS-KO do not suffer of renal dysfunction. This conclusion is consistent with a previous observation showing that podocyte-specific NEMO-deficient mice do not show overt changes in kidney morphology and functionality (17).
To verify whether NEMO inactivation affects thyroid embryonic development, mutant embryonic day 16.5 embryos were analyzed because Cre recombinase in Pax8 Cre/ϩ mice is active from embryonic day 8.5 (18). At embryonic day 16.5, no alteration in thyroid size and morphology could be detected (Fig.  4A), and similar results were obtained when the same analysis was performed in mice at 1 month of age (Fig. 4A). Hence, these data indicate that canonical NF-B signaling is not required for embryonic thyroid development and differentiation.
The effect of NEMO conditional inactivation on thyroid morphology and differentiation was further analyzed. Starting    (Fig. 4B). In the surviving 2-month-old NEMO TS-KO mice, significant cell loss was observed, and the thyroid follicular architecture appeared highly variable in diameter with irregular outlines (Fig. 4, B-F). Immunohistochemical and immunofluorescent analysis shows that the expression level of two thyroid-specific transcription factors, namely PAX8 and TTF1, steadily declined over time (Fig. 4, C and D). A similar decline occurred for thyroglobulin and for the Na/I symporter NIS (Fig. 4, E and F). NIS, for instance, appeared barely detectable in thyrocytes isolated from 2-month-old NEMO TS-KO mice (Fig. 4E). Masson's trichrome staining also revealed the presence of fibrotic material in mutant thyroids (Fig. 4G). In all staining, the progressive disorganization and degeneration of the thyroid parenchyma in NEMO TS-KO mutant mice was clearly visible. Also, reduction of cellularity and gland size in NEMO TS-KO mice was accompanied by massive apoptosis, as indicated by the staining for active caspase 3 (Fig. 4H).
Quantitative PCR analysis confirmed the reduced expression of the thyroid markers PAX8, TTF1, TPO, and NIS in NEMOdeleted thyroids (Fig. 4I). The mRNA expression levels of TSH receptor and thyroglobulin were also both significantly reduced (Fig. 4I). Despite its reduced expression, however, TSH receptor signaling does not appear compromised in NEMO TS-KO thyrocytes, as assessed by monitoring phosphorylation of CREB following TSH stimulation (Fig. 4J).
NF-B controls the expression of several genes, including Bfl-1/A1, Bcl-2, Bcl-xL, c-Flip, and Iap, that protect cells against apoptotic cell death (23)(24)(25)(26)(27)(28)(29). Consistently, all of these anti-apoptotic genes appear significantly down-regulated in NEMOdeficient thyrocytes (Fig. 5A). Thus, we verified whether NEMO TS-KO thyrocytes exhibit a greater sensitivity to apoptotic stimuli. As shown in Fig. 5, B-E, lack of NEMO dramatically sensitizes these thyrocytes to the cytotoxic action of both TNF␣ and TSH. Taken together, these data demonstrate that in adult mice, NEMO is required for thyrocyte survival and contributes to the maintenance of thyrocyte differentiation by regulating the expression of several thyroid markers.

Discussion
In this paper, we demonstrate that NEMO signaling is essential for normal postnatal thyroid gland structure and function. Although previous studies have clearly established a role for NF-B in thyroid tumor progression (30 -32), this is the first work that examines the role of canonical NF-B signaling in normal thyroid development and physiology. We show that mice with thyroid-specific ablation of NEMO develop a pronounced impairment of thyroid function, eventually leading to shortened life span. On the other hand, no gross alterations of thyroid gland localization, morphology, and differentiation were observed during embryonic development and in newborn mice, indicating that the NEMO-dependent NF-B signaling is not required for embryonic thyroid development. Our results indicate that in thyroid, NEMO is required for at least two essential aspects: (i) thyrocyte survival and (ii) maintenance of thyroid marker expression. The critical role of NF-B in cell survival is widely supported by the available literature. Genetargeting experiments have in fact shown that mice lacking p65/RELA, IKK2, or NEMO die during embryonic development due to liver apoptosis (7)(8)(9)(10)(11)(12). Thus, the extensive apoptosis observed in NEMO TS-KO thyrocytes confirms the requirement for NF-B in orchestrating cytoprotective pathways. Particularly interesting are the data showing the different behavior of WT and NEMO TS-KO thyrocytes when exposed to TSH. In fact, whereas both wild type and NEMO TS-KO thyrocytes proliferate following TSH stimulation, NEMO TS-KO thyrocytes also undergo extensive apoptosis (Fig. 5). This finding is consistent with previous evidence showing that TSH stimulation on thyrocytes in fact triggers both proliferative and apoptotic responses (33). Our work now shows that NF-B signal-ing could play a decisive role in determining the fate of thyrocytes following TSH stimulation.
An interesting aspect of our experiments is the evidence here provided that NF-B controls, directly or indirectly, the expression level of several thyroid markers. Although an involvement of NF-B in the regulation of Nis expression has been already reported (34), we now show that, in addition to NIS, NF-B controls the expression of a panel of thyroid markers, including TTF1, PAX8, TPO, and thyroglobulin. In this context, our work provides a molecular explanation for some phenotypical aspects of human disorders associated with mutations in NEMO. In humans, in fact, mutations in the X-linked NEMO gene cause two distinct genetic diseases. Mutations that completely disrupt the NEMO locus result in Incontinentia Pigmenti, a disease where male patients die in utero, whereas the phenotypic analysis of females is complicated by the fact that NEMO-deficient cells quickly disappear and are replaced by wild-type cells. A second disease, named hypohydrotic ectodermal dysplasia with immune deficiency (HED-ID), is characterized by hypomorphic NEMO mutations that do not completely abolish but rather reduce NF-B activation. HED-ID is characterized by impaired skin appendage development and severe immune deficiency. Strikingly, hypothyroidism is not uncommon in patients with HED-ID (35)(36)(37)(38), as predicted by the data we show here.
In conclusion, our data can be easily interpreted considering the crucial role that NEMO plays in activation of NF-B, which, in turn, regulates the expression of genes that are critical for cell survival. However, it should not be disregarded that NF-Bindependent functions have been ascribed to NEMO, and these latter are also involved in the control of cell proliferation and FIGURE 5. Increased sensitivity of NEMO TS-KO thyrocytes to apoptosis. A, quantitative PCR expression analysis of the indicated anti-apoptotic genes in thyrocytes isolated from 12-month-old NEMO TS-KO and control mice left untreated or stimulated with TNF␣ (10 ng/ml) for 7 h. Each RNA sample was obtained by pooling three thyroids/genotype. Data shown (mean Ϯ S.E., n ϭ 3) are representative of at least three independent experiments done in triplicate. B, thyrocytes from 6-month-old NEMO TS-KO and wild type mice were left untreated or treated with TNF␣ (10 ng/ml) or TSH (1 milliunit/ml) for 24 h, and cell viability was measured by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay. Data shown (mean Ϯ S.E., n ϭ 3) are representative of at least three independent experiments done in triplicate. C, phase-contrast micrographs of thyrocytes treated as in B. D, immunoblotting assay of caspase-3 from proteic lysates of thyrocytes treated as in B. Activation of caspase-3 can be inferred from processing of the p32 precursor. E, immunoblotting assay (left) and its densitometric analysis (right) of NEMO expression in purified thyrocytes used for the experiments shown above. Data shown are representative of at least three independent experiments. CTR, control. Error bars, S.E. cell death (39 -41). Hence, it is very possible that NF-B-independent functions of NEMO may contribute, at least in part, to the cellular degeneration observed in NEMO TS-KO thyrocytes.