Overexpression of Interleukin-4 in the Thyroid of Transgenic Mice Upregulates the Expression of Duox1 and the Anion Transporter Pendrin

Abstract

Background: The dual oxidases (Duox) are involved in hydrogen peroxide generation, which is essential for thyroid hormone synthesis, and therefore they are markers of thyroid function. During inflammation, cytokines upregulate DUOX gene expression in the airway and the intestine, suggesting a role for these proteins in innate immunity. It was previously demonstrated that interleukin-4 (IL-4) upregulates DUOX gene expression in thyrocytes. Although the role of IL-4 in autoimmune thyroid diseases has been studied extensively, the effects of IL-4 on thyroid physiology remain largely unknown. Therefore, a new animal model was generated to study the impact of IL-4 on thyroid function.

Methods: Transgenic (Thyr-IL-4) mice with thyroid-targeted expression of murine IL-4 were generated. Transgene expression was verified at the mRNA and protein level in thyroid tissues and primary cultures. The phenotype of the Thyr-IL-4 animals was characterized by measuring serum thyroxine (T4) and thyrotropin levels and performing thyroid morphometric analysis, immunohistochemistry, whole transcriptome sequencing, quantitative reverse transcription polymerase chain reaction, and ex vivo thyroid function assays.

Results: Thyrocytes from two Thyr-IL-4 mouse lines (#30 and #52) expressed IL-4, which was secreted into the extracellular space. Although 10-month-old transgenic animals had T4 and thyrotropin serum levels in the normal range, they had altered thyroid follicular structure with enlarged follicles composed of elongated thyrocytes containing numerous endocytic vesicles. These follicles were positive for T4 staining the colloid, indicating their capacity to produce thyroid hormones. RNA profiling of Thyr-IL-4 thyroid samples revealed modulation of multiple genes involved in inflammation, while no major leukocyte infiltration could be detected. Upregulated expression of Duox1, Duoxa1, and the pendrin anion exchanger gene (Slc26a4) was detected. In contrast, the iodide symporter gene Slc5a5 was markedly downregulated resulting in impaired iodide uptake and reduced thyroid hormone levels in transgenic thyroid tissue. Hydrogen peroxide production was increased in Thyr-IL-4 thyroid tissue compared with wild-type animals, but no significant oxidative stress could be detected.

Conclusions: This is the first study to show that ectopic expression of IL-4 in thyroid tissue upregulates Duox1/Duoxa1 and Slc26a4 expression in the thyroid. The present data demonstrate that IL-4 could affect thyroid morphology and function, mainly by downregulating Slc5a5 expression, while maintaining a normal euthyroid phenotype.

Introduction

Follicular units are critical for the principal function of the thyroid gland: iodide concentration for biosynthesis and secretion of the thyroid hormones triiodothyronine (T3) and thyroxine (T4). Iodide uptake is supported by the Na⁺/I⁻ symporter (NIS) that localizes to the basolateral membrane of follicular cells, while apical iodide efflux is mediated by the recently characterized Ca²⁺-activated ion channel anoctamin-1 (1,2) and the anion exchanger pendrin encoded by SLC26A4 (3). Within the lumen, iodide binding to tyrosine residues on thyroglobulin (Tg) is catalyzed by thyroperoxidase (TPO) after oxidation of iodide by hydrogen peroxide (H₂O₂) (4). The thyroid H₂O₂-generating system is composed of the dual oxidases, Duox1 and Duox2, which are membrane-bound NADPH-dependent flavoproteins (5,6). These Ca²⁺-activated NADPH oxidases require additional factors, the DuoxAs, for appropriate maturation and translocation to the apical pole of the thyrocyte (7,8). The thyd mouse carries a biallelic Duox2 inactivating mutation, which results in a profound iodide organification defect, In contrast, Duox1 KO mice remain euthyroid (9,10). Complete inactivation of the Duoxa1 and Duoxa2 genes results in growth retardation and hypothyroidism that are related to the absence of correct Duox maturation (11). Mutations in DUOX2 and DUOXA2 in humans cause inherited hypothyroidism, and emphasize the functional importance of these proteins in thyroid hormonogenesis (12,13).

During thyroid embryogenesis, Duox expression emerges only during the late stages of cell differentiation when the follicular structure has developed (14,15). In rat thyroid cell lines, Pax8-mediated induction of Duox2 expression and increased Duox1 expression after functional inactivation of Nkx2.1 have been demonstrated (16,17). However, no significant thyrotropin (TSH)-dependent modulation of DUOX transcription has been shown in humans and mice (14,18). The thyroid oxidases are considered as essential thyroid-related markers, but they are not thyroid-specific genes. Duox protein expression has been documented on the luminal side of highly differentiated epithelia (e.g., airway, salivary glands, and intestinal tract), where these proteins play a potent role in innate host defense (19). During infection or chronic inflammation, cytokines secreted by helper T lymphocytes (Th)2 (IL-4 and IL-13) or Th1 (IFN-γ) increase DUOX1 or DUOX2 expression levels (20–22). In human thyrocytes, it has been demonstrated in vitro that IL-4 selectively upregulates Duox2 and DuoxA2 protein expression levels that were repressed by IFN-γ treatment (18).

Autoimmune thyroid diseases (AITD) are the most frequent organ-specific autoimmune disturbances, and affect between 3% and 10% of the world's population (23). Hashimoto's thyroiditis (HT) and Graves' disease (GD) are two extreme manifestations of AITD that cause hypothyroidism and hyperthyroidism, respectively. HT is characterized by cytotoxic destruction of the thyroid gland with autoantibodies against TPO and Tg. Stimulating antibodies directed against the TSH receptor (TSAb) are responsible for the hyperthyroid phenotype of GD. Reactive oxygen species (ROS)-mediated oxidative stress could also participate in the pathophysiology of AITD (24–26). Moreover, antioxidant supplementation (selenium or N-acetylcysteine) in combination with antithyroid drugs, especially methimazole, can reduce the severity of AITD symptoms (27,28). The exact source of ROS has not been clearly identified, but they could arise from either the infiltrated leukocytes and/or the thyrocytes themselves. In vitro studies using thyroid cell lines to assess how cytokines secreted during the inflammation process affect thyroid function demonstrated repression of Nis, TPO, and Tg expression following cytokine stimulation (IL-1α, IL-1β, IL-6, TGF-β, TNF-α, TNF-β, and IFN-γ) (29). Transgenic mouse models expressing IFN-γ, IL-12, or TGF-β have also been generated to understand the effect of these cytokines better on thyroid pathophysiology (30–32). The role of IL-4 during the inflammatory process has been extensively studied using murine models for experimental autoimmune thyroid disease (EATD). However, the effect of IL-4 in GD pathophysiology remains unclear mainly because none of the EATD models can recapitulate all the hallmarks of the human disease (33–35). Moreover, the direct effects mediated by IL-4 on thyroid physiology remain largely unknown.

Given the IL-4-mediated induction of DUOX/DUOXA gene expression and the relative absence of in vivo data concerning the potential impact of IL-4 on thyroid function, a new transgenic mouse model was generated that specifically expresses IL-4 in the thyrocytes. These transgenic mice had serum levels of T4 and TSH within the reference ranges, despite an iodide uptake defect and a lower thyroid hormone content. The follicular structure of these mice showed enlarged follicles with thyroid cells containing multiple colloid droplets, suggesting that the gland adapted in order to maintain sufficient thyroid hormone secretion. For the first time, in vivo upregulation of pendrin expression is also demonstrated, as well as Duox1 and Duoxa1 genes following transgenic overexpression of IL-4.

Materials and Methods

Generation of Thyr-IL-4 transgenic mice

The mIL-4 cDNA (NM_021283.2) was subcloned into the pIRES2-AcGFP1 vector (Clontech Laboratories) between the EcoRI and BamHI sites. The polymerase chain reaction (PCR) amplified fragment containing the mIL-4-IRES2-AcGFP1-SV40PolyA sequence was then cloned between the EcoRI and XhoI sites of the bTg-βGlobin intron-pBluescript SK+ vector downstream of the bovine thyroglobulin promoter (bTg) (36). The final transgene fragment including the bTg-βGlobin intron-mIL-4-IRES2-AcGFP1-SV40PolyA was excised from the vector by SstII-XhoI digestion, and the fragment was gel purified (QIAquick gel extraction kit; Qiagen), followed by a second purification step (QIAquick PCR purification; Qiagen). The transgene cassette was injected at the Université Catholique de Louvain (UCL; Prof. P. Jacquemin) Transgenesis Platform into the pronuclei of B6D2F2 zygotes as previously described (37). Transgenic animals were maintained by mating hemizygous founder animals to wild-type (WT) C57BL/6J mice in a conventional animal housing facility. All transgenic and WT animals used in this study were between 3 and 12 months of age and were from at least the F4 progeny. All animal experiments were approved by the University's ethical committee (CEBEA permit 505N).

The transgene copy number was determined from tail genomic DNA by real-time quantitative PCR (RT-qPCR) amplification using primers directed to the GFP transgenic sequence and the β-actin gene for DNA normalization (Supplementary Table S1; Supplementary Data are available online at www.liebertpub.com/thy; n = 6 mice) (38). Six pg of DNA was considered to be equivalent to one C57BL/6J mouse cell (39).

RNA extraction and RT-qPCR

Total RNA was extracted from thyroid tissues using the RNeasy kit (Qiagen) and reverse transcribed with random primer hexamers (Roche Applied Science) and the Superscript II reverse-transcriptase (Life Technologies) as described previously (18). RT-qPCR was performed on a 7500 fast RT-qPCR system using Power SYBR Green mix (Life Technologies) with gene-specific, intron spanning primers (Supplementary Table S1). Amplicon purity and identity were assessed by dissociation curve analysis and sequencing of the PCR products on an automated ABI Prism 3100 sequencer (Life Technologies). Two reference genes (Retention in Endoplasmic Reticulum 1 homolog [Rer1]) and TATA box Binding Protein [Tbp]) (40) were used to analyze relative RNA expression with the Pfaffl method (41).

RNA sequencing

Thyroid RNA samples were collected and pooled from three four-month-old transgenic mice (lines 30 and 52) as well as their WT littermates. Indexed cDNA libraries were obtained using the TruSeq Stranded mRNA preparation kit (Illumina) following the manufacturer's recommendations. The multiplexed libraries (9.5 pM) were loaded on a Rapid flow cell. Sequences were produced using a TruSeq Rapid PE cluster and SBS-kit (200 cycles) on a HiSeq 1500 (Illumina). Approximately 44.2, 41.6, 48.3, and 49.1 million paired-end reads corresponded, respectively, to WT lines 30 and 52 and transgenic lines 30 and 52. These reads were mapped against the mouse reference genome (GRCm38.p4/mm10) using STAR software (https://github.com/alexdobin/STAR/releases) to generate read alignments for each sample. Annotated Mus_musculus.GRCm38.79.gtf were obtained from ftp.Ensembl.org. After transcript assembly, gene level counts were obtained using HTSeq (www-huber.embl.de/HTSeq). EdgeR (https://bioconductor.org/packages/edgeR) was used to calculate the level of differential gene expression with biological and data replications with false discovery rate (FDR) values of <0.05 being selected. Genes that showed more than a twofold change in expression between WT and transgenic conditions was considered as being significantly regulated.

Protein isolation and immunodetection

Proteins were extracted in lysis buffer composed of 1% Nonidet P-40 (NP40), 0.05 M of Tris-HCl pH 8, 0.15 M of NaCl, and supplemented with the protease inhibitors Pefabloc (300 ng/mL; Stago BNL), leupetin (5 μg/mL; Molekula), and pepstatin (1 μg/mL), as well as a cOmplete tablet (Roche Applied Science). For thyroid tissues, the samples were first homogenized in the lysis buffer with a rotor stator using a PTFE pestle and glass tube before vortexing for 1 h at 4°C and centrifugation at 10,000 g for 10 min. The soluble protein fraction was quantified by a Pierce 660 nm Protein Assay (Thermo Fisher Scientific). Proteins (30 μg) were denatured in Laemmli buffer at 100°C for 1 min, separated on sodium dodecyl sulfate polyacrylamide gel electrophoresis, and transferred to a polyvinylidene difluoride membrane (Immobilon-FL, Millipore). After blocking for 2 h in Odyssey Blocking Buffer (LICOR, Biosciences), the membrane was incubated overnight at 4°C with an anti-GFP primary antibody (1:1000; Clontech). A fluorescent secondary antibody (1:10,000; IRDye800 from LI-COR) was used for image acquisition and quantification with the Odyssey infrared imaging system (LI-COR).

T4, TSH, and IL-4 measurements

Total T4 in 25 μL of serum was measured using a solid-phase ¹²⁵I radioimmunoassay (RIA; Coat-a-Count T₄, Siemens). TSH serum concentrations were evaluated with a TSH activity bioassay using stable JP26 CHO cells expressing the TSH receptor (42) as described previously (43). Results are expressed as pmol cAMP/50,000 cells. IL-4 quantification was performed by enzyme-linked immunosorbent assay (ELISA; BD OptEIA) on proteins from thyroid tissues, and from the medium of WT and transgenic thyrocytes cultured for 48 h.

Mouse thyroid primary cultures

Samples from between 8 and 10 C57BL/6 mice aged 6–12 months were pooled and used in each culture. Thyroid glands were removed, and the follicles were isolated by collagenase/dispase digestion followed by differential centrifugation (44). The follicles were cultured for 3 days in modified Ham's F-12 medium supplemented with 2.5 μg/mL of fungizone, 100 IU/mL of penicillin, 100 μg/mL of streptomycin (Life Technologies), 5 μg/mL of human apotransferrin, and 1 μg/mL of insulin (Sigma-Aldrich).

Immunohistochemistry

Thyroid lobes were fixed in buffered formalin, embedded in paraffin, sectioned (5 μm), mounted on glass slides, deparaffinized, and rehydrated. For histological analysis, sections were stained with hematoxylin and eosin (H&E) following a routine protocol. For Duox and GFP detection, thyroid tissues were embedded and frozen in Tissue-Tek OCT compound (Sakura). Frozen sections were fixed in acetone for Duox staining. Staining conditions are summarized in Supplementary Table S2. For negative controls, either no antibodies were used or pre-immune serum was substituted (Duox staining). Semi-thin sections and Toluidine Blue coloration were performed as previously described (45). Cyclin D1 positive cells among 1000 cells were counted, and the proliferation index was represented as the mean ± standard error of the mean (SEM; n = 3–5 mice). The anti-pendrin antibody (PDS 75-03) was generated by the group of Prof. D. Eladari (Université Paris-Descartes, France) (46). The anti-Duox antibody was generated in the laboratory (5).

Hydrogen peroxide measurements

Hydrogen peroxide accumulated in Krebs-Ringer Hepes (KRH) medium was quantified using a fluorimetric assay based on H₂O₂-dependent oxidation of 440 μM of homovanillic acid (3-methoxy-4-hydroxyphenylacetic acid; Sigma-Aldrich) in the presence of 0.1 mg/mL of horseradish peroxidase type II (Sigma-Aldrich) as previously described (14,47). The amount of H₂O₂ accumulated in the KRH medium over 3 h was normalized to the total RNA content of the cells and represented as μM H₂O₂/μg RNA.

Morphometric analysis by image segmentation

Embedded 5 μm sections were deparaffinized, rehydrated, and treated with Tris-EDTA buffer (10 mM of Tris, 1 mM of EDTA, 0.05% Tween 20, pH 9.0) for 40 min at 98°C before hematoxylin staining. For each group, three different thyroid sections separated by at least 50 μm taken from three to four mice were analyzed.

After acquisition using a Zeiss Axioimager Z1 microscope, images were processed to preserve the maximum intensity of the three primary colors (R, G, and B) in one new module called lightness. Each individual layer corresponding to the thyroid follicle lumen was labeled and analyzed to determine its surface and diameter (major axis). The program was developed by the Laboratory of Image, Signal Processing and Acoustics (ULB) in the Python language with the mahotas tool box (48).

Iodide uptake and iodide organification assays

Iodide uptake experiments were conducted as described previously, with minor modifications for mouse thyroid samples (49). Briefly, two lobes from different mice were pooled and pre-incubated for 30 min in Krebs-Ringer bicarbonate medium supplemented with 8 mM of glucose, 0.5 g/L of bovine serum albumin (BSA), and 100 μM of methimazole to block iodide organification. Thyroids were incubated with 0.1 μM of KI and 1 μCi/mL of radioactive ¹²⁵I for 1 h at 37°C. The uptake was calculated as the ratio (T/M) of ¹²⁵I counts/min/100 mg of tissue (T) to the ¹²⁵I counts/100 μL medium (M) (Wizard^™ 1470 Automatic Gamma Counter, Perkin Elmer).

For iodide organification assays, thyroid lobes were pre-incubated for 30 min in KRH medium with 8 mM of glucose and 0.5 g/L of BSA, and transferred to fresh medium supplemented with 1 μM of KI and ¹²⁵I (1 μCi/mL) for 2 h as described previously (50). Results were calculated as the ratio of the protein bound ¹²⁵I (PBI) and the total radioactivity incorporated into the thyroid lobes, and were expressed as the percentage of PBI relative to the total thyroidal ¹²⁵I.

Tg–T3 and Tg–T4 thyroid content in the thyroid gland

Thyroid lobes were dissected and weighed. Tg–T3 and Tg–T4 contents, corresponding to T3 and T4 associated with Tg, were measured following a previously described protocol (51). The Tg–T3 and Tg–T4 contents were calculated by subtracting the non-Tg-T3/T4, measured without the hydrolysis step, from the total T3 and T4 content.

Statistical analysis

Statistical analysis was performed using an unpaired, two-tailed Student's t-test (Prism 5.0, GraphPad Software, Inc.) and the non-parametric Mann–Whitney U-test. Data are presented as means ± SEM. Differences were considered significant at p < 0.05.

Results

Generation of a Thyr-IL-4 mouse model that overexpresses murine IL-4 in the thyroid

Murine IL-4 cDNA was expressed under the control of the bTg promoter that supports thyroid-specific transcription (36). The mIL-4 sequence was followed by an IRES and GFP-encoding cDNA (Fig. 1A). Five founders were obtained (lines 7, 21, 26, 30, and 52) bearing 6, 1, 12, 2, and 4 copies of the transgene, respectively. ELISA measurement of IL-4 protein levels showed 6- and 12-fold increases in the thyroid tissue from transgenic mouse lines 52 and 30, respectively, relative to WT littermates (Fig. 1B). The mice derived from the other founders showed no significant IL-4 overexpression as illustrated by line 7 mice (Fig. 1B and Supplementary Fig. S1A). IL-4 and GFP transcription in thyroid tissue was quantified by RT-qPCR (Supplementary Fig. S1A). GFP protein expression was also validated in transgenic thyroid tissues by immunoblotting and immunofluorescence (Supplementary Fig. S1B and C). IL-4 secretion was then estimated from primary cultures of WT and Thyr-IL-4 mice. After 48 h of culture, IL-4 accumulated in the culture medium of Thyr-IL-4 thyrocytes to a level of 65 pg/mL/μg of RNA, while only 3 pg/mL/μg of RNA was present in the medium of WT thyrocytes (Fig. 1C). No significant amount of IL-4 could be detected in the mouse sera, suggesting that IL-4 expression was confined to the thyroid gland and did not reach systemic circulation. In addition, there was no significant Il-4 mRNA overexpression in other major organs that were tested, including the lung, liver, heart, kidney, spleen, testis, and brain (data not shown). IL-4 transgenic animals were viable and reached adulthood without any growth retardation or fertility defect.

FIG. 1.

Generation and characterization of Thyr-IL-4 mice. (A) Schematic representation of the Thyr-IL-4 transgenic construct (IRES, internal ribosome entry site; Tg, thyroglobulin). (B) Thyroid expression of mIL-4. mIL-4 protein levels were measured by enzyme-linked immunosorbent assay of total protein extracts from wild type (WT) and transgenic thyroid tissues. Protein concentrations were normalized to the total protein content (30 μg; n ≥ 4 mice). (C) mIL-4 secretion by Thyr-IL-4 primo-cultured thyrocytes. Culture medium was collected 48 h after cell seeding. Inteleukin-4 (IL-4) concentration was normalized to the total RNA content of the cells (n = 4). Data are presented as means ± standard error of the mean (SEM).

Thyroid function is not strongly modified in transgenic Thyr-IL-4 mice

The thyroid status of 3- to 12-month-old mice was examined by measuring the serum levels of total T4 and bioactive TSH. Relative to WT littermates, adult Thyr-IL-4 mice from line 30 showed a slight decrease in T4 level, but the T4 amount was still within the reference range. No significant difference from WT littermates was observed for line 52 (Fig. 2A). Meanwhile, serum TSH showed no statistically significant difference between WT and the two transgenic mouse lines (Fig. 2B).

FIG. 2.

Thyroid status of adult (3- to 12-month-old) transgenic IL-4 mice. (A) Serum total thyroxine (T4) was measured by radioimmunoassay (μg/dL; n ≥ 6). (B) Serum thyrotropin levels were evaluated by bioassay and presented as pmol cAMP produced by 50,000 TSHr-expressing cells (n ≥ 18). Data are presented as means ± SEM.

Transgenic animals show altered thyroid follicular structure with enlarged follicles

The mean weight of thyroids from 10-month-old mice was 3.34 ± 0.79 mg for transgenic lines (n = 20) and 3.02 ± 0.82 mg for WT littermates (n = 28). Macroscopic evaluations revealed no visible goiter, although a superficially nodular appearance was detected in Thyr-IL-4 mice, particularly from the age of 10 months. This nodularity was likely the consequence of large peripheral follicles that were full of colloid that could be observed upon histological examination (Fig. 3A). The histological architecture of thyroid lobes from 4- and 10-month-old animals was also explored by H&E staining. The most relevant phenotype in the transgenic mice was the increased follicle size seen in a significant proportion of the follicle population in elderly animals. In these enlarged follicles, the thyrocyte shape was elongated with flattened nuclei usually found in hypoactive follicles. A small proportion of follicles presented with cuboidal cells with cytoplasm extensions and a reduced colloidal lumen. No significant increase in leukocyte infiltration could be detected in transgenic mice compared to WT mice either by H&E or anti-CD45 staining (data not shown). Transgene positive lines without IL-4 overexpression did not show these histological alterations, which supports the conclusion that the observed phenotype is IL-4 dependent. Toluidine Blue staining of semi-thin sections of 10-month-old Thyr-IL-4 thyroid tissues showed numerous endocytic vesicles represented by dark blue spots in enlarged follicles with flat or cuboidal cells that were absent in WT thyroids (Fig. 3B).

FIG. 3.

Enlarged thyroid follicles without goiter in Thyr-IL-4 mice. (A) Hematoxylin and eosin staining of thyroid sections (5 μm thick) from 4- and 10-month-old mice. Scale bars: 100 μm. (B) Toluidine Blue staining on thyroid semi-thin sections (0.5 μm thick) from 10-month-old animals. Scale bars: 10 μm. Arrows indicate endocytic vesicles. (C) Follicle diameter (major axis) distribution in 4- and 10-month-old Thyr-IL-4 and WT animals (n = 12 thyroid sections from three mice per genotype) expressed as number of follicles per 100 follicles. (D) Mean follicle area in 4- and 10-month-old Thyr-IL-4 and WT mice (n = 9–12 thyroid sections per genotype). Data are presented as means ± SEM.

To quantify these structural differences, morphometric analyses were conducted by measuring the follicular diameter (major axis) and the area of 7425 WT and 8060 Thyr-IL-4 follicles. Figure 3C shows the data concerning the follicular diameter presented as frequency per hundred follicles. Most follicles from WT mice had a diameter <100 μm (93 ± 1% at 4 months and 90 ± 2% at 10 months). At four months, a small but significant proportion of follicles with a diameter between 50 and 100 μm was present in Thyr-IL-4 tissues (45 ± 2% vs. 38 ± 2% for line 30 and 47 ± 3% vs. 37 ± 2% for line 52). By 10 months of age, transgenic follicles showed a significantly increased proportion of enlarged follicles with a diameter >100 μm (28 ± 5% vs. 9 ± 2% for line 30 and 20 ± 2% vs. 10 ± 2% for line 52). The mean follicle area in 10-month-old Thyr-IL-4 animals was also increased relative to WT littermates (4508 ± 557 μm² vs. 2714 ± 184 μm² for line 30 and 3393 ± 178 μm² vs. 2383 ± 185 μm² for line 52; Fig. 3D).

Gene profiling of Thyr-IL-4 thyroids reveals modulated expression of genes involved in thyroid function

Whole transcriptome sequencing was performed on thyroids of four-month-old animals. A total of 15,068 transcripts expressed in either WT or transgenic thyroids were identified. Genes showing a more than a twofold change from WT animals with a significant FDR (<0.05) were used to discriminate differentially expressed genes. Together, thyroids from transgenic lines 30 and 52 had 583 up- and 203 downregulated genes compared with their WT littermates (Supplementary Table S3). Pathway analysis of these differentially regulated genes using DAVID software (52,53) revealed involvement of several signaling pathways with 12 implicated in immunity, including one thought to be involved in AITD (Supplementary Table S4). The thyroid overexpression of the IL-4 transgene was confirmed by an 18-fold increase in expression seen in transgenic thyroids. Thyroid expression of the type II IL-4 receptor was confirmed by multiple reads corresponding to Il-4rα and Il-13rα1 genes measured in WT and transgenic tissues (54). Supplementary Table S5 shows a list of genes that are described in the literature to be regulated by the IL-4 pathway (55–58) and that were significantly modulated in Thyr-IL-4 thyroids. RT-qPCR experiments confirmed the modulated expression of Il-13rα2, a decoy IL-13 receptor, and Il-4rα transcripts (Fig. 4A). These data strongly suggest that mIL-4 was not only highly expressed in the thyroid, but also exerted a functional effect.

FIG. 4.

Aberrant thyroid IL-4 overexpression modulates Nis/Slc5a5, Duox1, and Duoxa1 gene expression in adult (3- to 12-month-old) transgenic mice. (A) mRNA expression of Il-4rα, and Il-13rα2 measured by quantitative reverse transcription polymerase chain reaction (RT-qPCR) of WT and Thyr-IL-4 thyroid tissues. Relative mRNA expression was normalized to the reference genes Rer1 and Tbp (n ≥ 6). (B) Duox1, Duox2, Duoxa1, and Duoxa2 mRNA relative expression (n ≥ 5). (C) Relative mRNA expression of Tshr, Nis/Slc5a5, and Tpo (n ≥ 6). Data are presented as means ± SEM.

Among genes related to thyroid function, including Tshr, Tpo, Iyd (iodotyrosine deiodinase), Duox1, Duoxa1, Duox2, Duoxa2, Slc26a4 (anion exchanger pendrin), Ano1 (Anoctamine-1), and Slc5a5 (Nis), the expression of only Duox1, Duoxa1, Pendrin, and Nis transcripts was significantly modulated (Supplementary Table S6). Previously, an IL-4-dependent positive modulation of both Duox1/Duoxa1 and Duox2/Duoxa2 was demonstrated in primary mouse thyrocyte cultures treated for 48 h with 100 ng/mL of mIL-4 (18). Here, it is shown that expression of mIL-4 in the thyroid in vivo selectively upregulates Duox1 and Duoxa1 gene expression (Fig. 4B). Interestingly, both Duox1/Duoxa1 and Duox2/Duoxa2 transcripts were upregulated in primary cultures of thyrocytes from Thyr-IL-4 mice compared with WT thyroid cultures (Supplementary Fig. S2). No Duox/Duoxa gene induction was observed in the IL-4 negative transgenic line 7 (Supplementary Fig. S3). Nis expression showed a significant three- to fivefold decrease as confirmed by RT-qPCR (Fig. 4C), while Tpo and Tshr expression did not differ between Thyr-IL-4 and WT animals.

Surprisingly, the thyroid marker showing the most robust modulation was pendrin, as evidenced by the more than a 1000-fold increase in mRNA levels seen with Thyr-IL-4 lines 30 and 52 relative to WT mice (Fig. 5A). Although a previous study demonstrated that the Slc26a4 gene encoding pendrin was induced in bronchial epithelium following IL-4 exposure (57), to date no data concerning IL-4-dependent pendrin regulation have been reported for the thyroid. The regulation of pendrin expression was confirmed in WT primary thyrocyte cultures stimulated for 48 h with mIL-4 (100 ng/mL; Fig. 5B). Moreover, pendrin protein staining was also enhanced in transgenic animals, particularly at the follicular apical membrane (Fig. 5C).

FIG. 5.

Pendrin overexpression in thyroids from transgenic animals. (A) Relative Slc26a4 (pendrin) mRNA expression measured by RT-qPCR in WT and Thyr-IL-4 thyroid tissue (n ≥ 4 mice). (B) Relative Slc26a4 mRNA expression in WT thyrocyte primary cultures after 48 h of stimulation with 100 ng/mL of mIL-4 (n = 4). (C) Thyroid sections (5 μm thick) from WT and transgenic mice were submitted to immunohistochemistry for pendrin expression. Pendrin apical staining was clearly increased in Thyr-IL-4 tissues. Scale bars: 50 μm (n = 3). Data are presented as means ± SEM.

Consequences of Duox1 and Duoxa1 upregulation in Thyr-IL-4 thyroid

Duox proteins are Ca²⁺-dependent NADPH oxidases that produce extracellular H₂O₂, a key step in thyroid hormonogenesis (4). To assess the functional consequences of Duox induction, the amount of extracellular H₂O₂ generated by thyroid tissues from WT and Thyr-IL-4 animals (a mixture of lines 30 and 52) was quantified. Under all conditions (i.e., basal, 2 μM of ionomycin, 10 μM of forskolin, or 5 μM of phorbol 12 myristate 13-acetate [PMA]), H₂O₂ levels were increased in transgenic tissues compared with WT (Fig. 6A). Forskolin alone was sufficient to stimulate H₂O₂ production in Thyr-IL-4 thyroids but not in WT animals. This result strongly suggests that Duox1 contributes to H₂O₂ production by Thyr-IL-4 thyroids. Moreover, upon stimulation with ionomycin and PMA, the H₂O₂ concentration generated from transgenic primary thyrocyte cultures was significantly increased by twofold compared with WT cells (data not shown), and was very similar to the amount of H₂O₂ produced by WT thyrocytes stimulated with exogenous mIL-4 (18).

FIG. 6.

Consequences of Duox1/Duoxa1 overexpression in Thyr-IL-4 thyroid tissue. (A) H₂O₂ production by thyroid tissues. Extracellular H₂O₂ was allowed to accumulate for 3 h in control (–) and stimulated (+) thyrocytes (2 μM of ionomycin, 10 μM of forskolin, or 5 μM of PMA; n = 3). H₂O₂ levels were normalized to the total RNA content. (B) Duox and 4-HNE immunodetection in thyroid sections (5 μm thick) from 10-month-old WT and transgenic mice. Duox and 4-HNE immunostaining were localized at the apical membrane and in the cytoplasm, respectively. Scale bars: 50 μm (n = 3). (C) Proliferation index of cyclin D1 staining quantified from thyroid sections (n ≥ 4 mice). Data are presented as means ± SEM.

Analysis of Duox protein expression by immunohistochemistry confirmed the apical membrane localization of these proteins in either small or large follicles. However, because the anti-Duox antibody was not isoform specific, an intensification of Duox staining in transgenic tissues could not be demonstrated (Fig. 6B). Instead, the oxidative stress marker, 4-hydroxy-2-nonenal (4-HNE), revealed heterogeneous staining of the follicles with no significant increase in Thyr-IL-4 mice (Fig. 6B). Moreover, expression of key antioxidant related genes, including heme oxygenase 1 (Hmox1), glutathione peroxidase 3 (Gpx3), thioredoxin reductase 1 (Txnrd1), and glutamate cysteine ligase (Gclm), was not significantly modulated in four- (whole transcriptome sequencing) and 10-month-old (RT-qPCR, data not shown) Thyr-IL-4 thyroids compared to WT.

To assess whether the enlargement seen in transgenic follicles is due to a higher proliferation rate, cyclin D1 immunostaining was performed. In Thyr-IL-4 thyroids, rare proliferating cells could be observed. Furthermore, the proliferation index was similar between WT and Thyr-IL-4 thyroids, with no particular proliferation capacity of IL-4 expressing thyrocytes seen (Fig. 6C).

Reduction of Nis expression results in impaired iodide uptake and lower thyroid T3 and T4 content

To evaluate the physiological consequences of partial suppression of Slc5a5 expression in Thyr-IL-4 mice, ex vivo iodide uptake experiments were conducted in thyroid lobes. For both transgenic lines 30 and 52, iodide uptake was decreased by threefold relative to WT glands, thus confirming the presence of an iodide transport defect in transgenic animals (Fig. 7A). In all animals tested, iodide uptake (T/M) could be abolished in the presence of NaClO₄, which supports a Nis-specific activity.

FIG. 7.

Iodide uptake is decreased in transgenic mice, but T4 synthesis is maintained. (A) ¹²⁵I uptake measured ex vivo on thyroid lobes and expressed as a T/M ratio (see Methods; n = 4). NaClO₄ was used as a control for Nis-dependent iodide transport. (B) T4 and Tg immunostaining on serial thyroid sections (5 μm thick) from 10-month-old WT, Thyr-IL-4 mice, and Duoxa^–/– mice. Scale bars: 100 μm (n = 3). (C) Iodide organification assays performed ex vivo on thyroid lobes expressed as protein bound ¹²⁵I percentage (see Methods; n = 4). (D) Tg–T3 and Tg–T4 content measured in thyroid lobes from young (3- to 4-month-old) and old (10- to 11-month-old) mice (n ≥ 4). Data are presented as means ± SEM.

Next, the T4 synthesis capacity of the enlarged follicles composed of elongated cells with flattened nuclei that were suggestive of hypoactive units was investigated. T4 immunostaining in thyroids of 10-month-old mice showed positive labeling in WT and transgenic animals in small and enlarged follicles with no visible differences (Fig. 7B). The Tg content also seemed to be similar between the two genotypes. No positive T4 staining was detected in the negative control, the Duoxa–/– mice, which have an iodide organification defect caused by impaired H₂O₂ generation (11). To explore the overall efficiency of the iodination process, the proportion of PBI was measured. Methimazole pretreatment inhibited iodide organification, showing a specific TPO activity. Thyr-IL-4 thyroids presented a similar proportion of PBI as that seen in WT tissues, suggesting no iodide organification defect (Fig. 7C). However, a minor, but not significant, diminution of the total amount of organified iodide per milligram of Thyr-IL-4 tissue could be detected (data not shown). The thyroid content of Tg–T3 and Tg–T4 (T3 and T4 associated with the Tg macromolecule) was quantified (51). Transgenic thyroid tissues contained significantly less Tg–T3 and Tg–T4 relative to WT counterparts either in young (three- to four-month-old) and old (10- to 11-month-old) animals (Fig. 7D).

Discussion

A role for inflammatory cytokines, including IL-4, has been widely documented in inflammation in general and suggested in AITD. Here, the in vivo pathophysiological consequences of chronic exposure of mouse thyroid glands to the Th2 cytokine IL-4 were investigated using the well-characterized bTg promoter (36). In this new mouse model, it was found that IL-4 expression was restricted to thyroid tissue and functional as evidenced by: (i) IL-4 transgene expression at the mRNA and protein level; (ii) extracellular secretion of IL-4 from Thyr-IL-4 thyroid cells; (iii) GFP localization in follicular cells; and (iv) gene profiling showing modulated expression of multiple genes involved in inflammation.

Recently, Th2 cytokine-mediated induction of DUOX and DUOXA expression in in vitro models of human (DUOX2/DUOXA2) and mouse (Duox1/Duoxa1 and Duox2/Duoxa2) thyrocytes was demonstrated. Here, positive regulation of only Duox1 and Duoxa1 genes in vivo in Thyr-IL-4 mice was clearly identified. However, primary cultures of transgenic thyrocytes recapitulated the increased expression of both Duox1/Duoxa1 and Duox2/Duoxa2 isoforms compared with WT cells. The apparent inconsistency between tissue and cell culture most likely resulted from cellular de-differentiation during the culture, as suggested by the strong downregulation of Duox2/Duoxa2 (Fig. 4B and Supplementary Fig. S2), Tpo, and Slc5a5 mRNA expression (data not shown). No significant increase in Ifn-γ expression was detected by RNA sequencing in Thyr-IL-4 thyroids, which could also explain the repression of Duox2 expression in thyroid tissue. A similar repression of these differentiation genes was previously reported in human thyroid cancer cell lines (59). Together, these results show the limitations that can be associated with experiments performed with thyrocyte cultures or cell lines as thyroid physiological models. Moreover, induction of DUOX1 expression via the IL-13 pathway was recently demonstrated in both human primary thyrocytes and the thyroid cell line Hthy-ori3.1 (60). These data clearly indicate that IL-4 or IL-13-dependent induction of DUOX gene transcription is species- and cell type-dependent and can be markedly influenced by the differentiation status of the cell.

In the human thyroid, the DUOX2 mRNA content is two- to fivefold higher than DUOX1 (61). Thyroid mouse RNA profiling in this study also showed an increased Duox2/Duox1 ratio (Supplementary Table S6), suggesting that the Duox2 isoform is also a major component of the murine thyroid H₂O₂-generating system. However, Duox2/Duox1 expression levels were very low relative to other thyroid function markers such as Tpo and iodotyrosine deiodinase (Iyd), which were the most abundant transcripts after Tg mRNA (Supplementary Table S6). In Thyr-IL-4 thyroids, despite the Duox1 overexpression, Duox2 remained the major Duox isoform. This could explain the absence of a significant increase in the oxidative stress marker 4-HNE in the Thyr-IL-4 transgenic mice.

The Slc26a4 gene encodes the transmembrane protein anion exchanger pendrin that exchanges chloride for bicarbonate, iodide, and thiocyanate (SCN⁻) (62). The pendrin protein is expressed not only in the thyroid, but also in many other tissues, including the kidney, the airways, the mammary gland, and the inner ear (63). The role of pendrin in thyroid apical iodide transport is still a matter of debate. Pendred syndrome, an autosomal recessive disorder, is mainly characterized by deafness, but some patients also present with hypothyroidism (63). No thyroid phenotype has been reported for Slc26a4 KO mice (64), suggesting that pendrin plays a non-essential role in iodide transport, at least in the mouse, and that alternate transporters, most likely the Ca²⁺-activated ion channel Anoctamin-1, can mediate apical iodide efflux (1,2). Using Thyr-IL-4 transgenic mice, this study showed for the first time an IL-4-dependent induction of Slc26a4 transcription, which was associated with increased apical expression of pendrin. TSH can rapidly stimulate pendrin membrane translocation via the PKA pathway (65), although the Thyr-IL-4 mice showed no significant increase in serum TSH level relative to WT mice (Fig. 2B). The IL-4-mediated pendrin expression observed here suggests that pendrin may have a role in thyroid physiology during inflammation. This possibility is consistent with a previous study in bronchial epithelial cells showing that IL-4/IL-13-mediated SLC26A4 induction via direct promoter transactivation by the JAK1-STAT6 pathway (66,67) is associated with increased SCN⁻ apical transport (57). In the presence of H₂O₂ produced by Duox, SCN⁻ is oxidized in the microbicide compound hypothyocyanite (OSCN⁻) by mucosal lactoperoxidase, which plays an important role in innate immunity (68,69).

Although Thyr-IL-4 mice show neither growth retardation nor hypothyroidism, a clear repression of Nis expression associated with impaired iodide uptake (reduced T/M) was seen. This result suggests that an increase in Duox1-mediated ROS could inactivate Nis function in Thyr-IL-4 mice to inhibit iodide uptake further (70). In Thyr-IL-4 thyroid tissue, iodide organification was not compromised (as shown by a similar PBI), but the thyroid content of T3 and T4 bound to Tg was significantly reduced. Moreover, the lack of significant induction of iodotyrosine dehalogenase, which is involved in iodide recycling, suggests that this dehalogenase does not contribute to the maintenance of the normal serum T4 levels in these transgenic animals (71). The most striking phenotype observed in Thyr-IL-4 thyroid tissue was the increased proportion of enlarged follicles in old mice. Despite their hypofunctional aspects (e.g., elongated cells with flattened nuclei), these enlarged follicles still stained positive for T4. This phenotype is most likely the result of specific biological effects of IL-4 because its severity was related to the level of IL-4 expression in the thyroid gland, which was higher in transgenic line 30 than in line 52, and not detected in line 7. Numerous intracellular dense bodies were present in the cytoplasm of Thyr-IL-4 thyrocytes. These colloid droplets could reflect an intense endocytosis of luminal Tg to maintain normal levels of thyroid hormone secretion despite decreased amounts of iodinated Tg contained in the gland (lower Tg–T3 and Tg–T4) (72). However, the exact molecular mechanism behind this IL-4 mediated phenotype is unclear and awaits further characterization in future studies.

Overexpression of the anion exchanger pendrin could partially compensate for the lower basal Nis-dependent iodide uptake by facilitating the apical efflux of iodide. Pendrin may also be implicated in the alkalinization of the follicular lumen by exchanging Cl⁻ for HCO₃⁻ to promote generation of follicular H₂O₂ (50,73,74). However, no significant increase in iodide organification was seen in transgenic thyroid tissues. A thyroid mouse model with overexpression of the proto-oncogene pituitary tumor transforming gene-binding factor also showed modified thyroid architecture that featured enlarged follicles (75), but this phenotype, which was associated with hyperproliferative lesions, was not observed in Thyr-IL-4 mice. As such, the follicular volume may have resulted from a bidirectional transepithelial transport of basal to apical Cl⁻ and an opposite Na⁺ flux (76). IL-4 and IL-13 have been reported to modulate these Na⁺Cl⁻ currents in certain cell types by regulating the expression of channels and transporters, including aquaporins (62). However, pendrin overexpression would favor the reduction of the colloid space by promoting reabsorption of luminal chloride. As such, a cross of Thyr-IL-4 mice with Slc26a4 KO animals (64) would likely provide valuable information about the contribution of pendrin to the Thyr-IL-4 phenotype.

The common simplified view that separates HT and GD on the basis of the Th1 and Th2 inflammatory response, respectively, was recently reviewed (77). RT-qPCR performed on GD and HT thyroid tissue samples showed an increase in IL-4 transcripts in GD compared with HT thyroids, but also augmented IFN-γ expression in GD compared with control patients (78). An IL-4 gene haplotype is associated with pediatric GD, and more recent plasma analysis of a GD cohort demonstrated increased IL-4 protein levels in recurrent GD patients (79,80). However, data obtained from various EATD murine models were even more complex. Il-4^–/– BALB/c mice immunized with TSHr-expressing cells are resistant to GD development and showed a delayed antibody response (34). Meanwhile, another model involving immunization of BALB/c mice with TSHr-expressing adenovirus showed that coinjection of an adenovirus expressing IL-4 significantly reduced TSAb levels and incidence of hyperthyroidism (35). No significant increase in leukocyte infiltration could be detected in Thyr-IL-4 thyroid glands, which was likely due to the absence of immunization experiments and also to the genetic background of the mice (C57BL/6J) that are known to be EATD resistant (81). However, differential expression of multiple genes involved in inflammatory pathways in Thyr-IL-4 thyroid samples was observed (Supplementary Table S3), suggesting that Thyr-IL-4 mice could be used to investigate the role of IL-4 in AITD when appropriate immunization protocols are used after crossing these mice with other AITD permissive mice, such as the recently reported TSHr-transgenic NOD.H2^h4 mice that spontaneously develop anti-TSHr antibodies (82).

In GD patients, the main source of IL-4 is the thyroid lymphocyte infiltrate, which is generated by the thyrocytes themselves in Thyr-IL-4 mice. The effects associated with uncontrolled ectopic production of IL-4 do not reflect the actual situation in GD patients who present with thyroid inflammation, TSHr-autoantibodies, and thyroid hyperstimulation; the latter would likely compensate for the IL-4-mediated Nis repression observed in the transgenic animals in the present study. In contrast, enhanced thyroid apical expression of pendrin was also reported in GD patients (83–86). Moreover, autoantibodies against pendrin have been found in the sera of AITD patients, and SLC26A4 has been proposed to be a new AITD susceptibility gene (87,88).

In conclusion, the development of a novel mouse model with targeted IL-4 expression in the thyroid is herein described. Characterization of these mice showed that ectopic expression of IL-4 in the thyroid positively regulates Duox1, Duoxa1, and Slc26a4 expression while inhibiting Slc5a5 expression. The Thyr-IL-4 mice had impaired thyroid iodide uptake, increased follicular size, and lower thyroidal T3 and T4 content, but serum levels of T4 and TSH were within reference ranges. No leukocyte infiltration was clearly established.

Acknowledgments

We gratefully acknowledge Dr. H. Grasberger for the Duoxa KO mice, Prof. D. Eladari for the pendrin antibody 75-03, and Prof. P. Jacquemin of the UCL transgenesis platform. We thank Dr. B. Contempré and Prof. R. Beauwens for their interest in this work and their advice. We thank J.M. Vanderwinden for the use of the microscope facility. We extend special thanks to C. Degraef, B. Bournonville, C. Massart, M. de Bournonville, and C. de Ville de Goyet for their excellent technical assistance, and to F. Libert and A. Lefort for RNA sequencing. This work was supported by the Fonds de la Recherche Scientifique (FRS-FNRS), the Fund Doctor J.P. Naets, managed by the King Baudouin Foundation, the Fondation Rose et Jean Hoguet, the Fonds Lekime-Ropsy, and grant DK15070 from the U.S. National Institutes of Health to S.R. Z.E. is a recipient of a fellowship from the Fonds pour la Formation à la Recherche dans l'Industrie et l'Agriculture.

Author Disclosure Statement

No competing financial interests exist.