TSH Compensates Thyroid-Specific IGF-I Receptor Knockout and Causes Papillary Thyroid Hyperplasia

Abstract

Although TSH stimulates all aspects of thyroid physiology IGF-I signaling through a tyrosine kinase-containing transmembrane receptor exhibits a permissive impact on TSH action. To better understand the importance of the IGF-I receptor in the thyroid in vivo, we inactivated the Igf1r with a Tg promoter-driven Cre-lox system in mice. We studied male and female mice with thyroidal wild-type, Igf1r^+/−, and Igf1r^−/− genotypes. Targeted Igf1r inactivation did transiently reduce thyroid hormone levels and significantly increased TSH levels in both heterozygous and homozygous mice without affecting thyroid weight. Histological analysis of thyroid tissue with Igf1r inactivation revealed hyperplasia and heterogeneous follicle structure. From 4 months of age, we detected papillary thyroid architecture in heterozygous and homozygous mice. We also noted increased body weight of male mice with a homozygous thyroidal null mutation in the Igf1r locus, compared with wild-type mice, respectively. A decrease of mRNA and protein for thyroid peroxidase and increased mRNA and protein for IGF-II receptor but no significant mRNA changes for the insulin receptor, the TSH receptor, and the sodium-iodide-symporter in both Igf1r^+/− and Igf1r^−/− mice were detected. Our results suggest that the strong increase of TSH benefits papillary thyroid hyperplasia and completely compensates the loss of IGF-I receptor signaling at the level of thyroid hormones without significant increase in thyroid weight. This could indicate that the IGF-I receptor signaling is less essential for thyroid hormone synthesis but maintains homeostasis and normal thyroid morphogenesis.

Due to the strong negative symptoms associated with thyroid hormone imbalance, the activity of the thyroid gland is tightly controlled, mainly by the hypophyseal hormone TSH. However, other signaling cascades such as the IGF-I receptor (IGF-IR) are suspected to be very important as well, being part of a reciprocal permissive effect with TSH (1, 2).

IGF-IR activation by IGF-I or high concentration of insulin stimulates cell growth in particular by the activation of the rat sarcoma/rat fibrosarcoma/proto-oncogene B-RAF (BRAF) MAPK cascade and phosphatidylinositol 3-kinase (PI3K)/AKT pathway (3). Thereby it permits the mitogenic effect of TSH and cAMP. It was shown that mainly the type I IGFR is expressed in the thyroid gland (4). Overexpression of the IGF-IR together with IGF-I in the thyroid leads to an increase of gland weight and follicular lumen together with decreased TSH levels and increased serum T₄ concentrations, which suggest that IGF-I and IGF-IR stimulate thyroid function (5). Moreover, a population-based study suggests that with higher serum IGF-I levels the odds for developing thyroid goiter increase (6). In thyroid carcinogenesis the IGF-I axis is a major regulator (7, 8). Follicular thyroid carcinomas often show deregulation in the PI3K pathway, which is a component of IGF-IR signaling. The PI3K/Akt/mammalian target of rapamycin pathway, which controls essential processes such as proliferation and survival, has recently emerged as a pivotal signaling cascade in follicular thyroid carcinoma development (9). Not only follicular thyroid carcinoma but also papillary thyroid carcinoma pathology is characterized by deregulation in MAPK signaling or mutations in the PI3K gene (10).

However, the physiological role of IGF-IR signaling in thyroid tissue in vivo has not been systematically studied. To investigate the role of the IGF-IR in the development and metabolism of the thyroid, we generated mice lacking the Igf1r in thyroid tissue using a conditional gene-targeting approach based on the Cre recombinase (Igf1r^TgCre). Subsequently, we characterized the consequences of Igf1r deletion in thyroid tissue on morphological and metabolic parameters of Igf1r^TgCre mice.

The IGF and the structurally related insulin are essential for the control of embryonic and postnatal development. Through binding to its cell surface receptor, IGF stimulates the autophosphorylation of the receptor and activation of its intrinsic tyrosine kinase activity. In turn, this leads to the tyrosine phosphorylation of various intracellular substrates (11, 12). Conventional gene-targeting strategies used to abrogate IGF-IR signaling reveal severe phenotypes of Igf1r-deficient mice. Mice with a complete germline inactivation of the Igf1r gene die shortly after birth (13). Furthermore, heterozygous Igf1r-knockout mice have increased longevity, most likely due to greater resistance to oxidative stress (14). These data point to a role of IGF-IR signaling in the regulation of lifespan. Total ablation of the Igf1r or members of the Igf family in mice, shortly after birth, results in reduced body weight (11), brain growth retardation (15), muscle hypoplasia (16), and prolonged lifespan (14). To determine the relevance of IGF-IR signaling in individual tissues, mice with tissue-specific inactivation of the Igf1r have been created. Tissue-specific knockout of the Igf1r gene displayed several defects in specific organs and secondary effects on the organism (17–21). Tissue-specific knockout of the Igf1r gene in the white adipose tissue leads to increased IGF-I serum concentrations and Igf1 mRNA levels in liver and adipose tissue combined with increased liver weight (22). However, despite the evidence for an important role of IGF-I in the thyroid, loss of IGF-IR signaling has not been investigated in this tissue to date.

Results

Conditional Igf1r knockout in the thyroid

Mice lacking one or two alleles of the Igf1r in thyrocytes (Igf1r^+/− or Igf1r^−/−) were generated by crossing mice carrying the loxP-flanked Igf1r allele with transgenic mice expressing the Cre recombinase under control of the thyroid-specific thyroglobulin (Tg) promoter. The targeting strategy is shown in Fig. 1A. Mice with the Igf1r^flox/wt or Igf1r^flox/flox geneotype expressing Cre transgene were obtained close to the expected Mendelian frequency (data not shown). Because Cre expression was targeted to thyrocytes, detection of an Igf1r^flox allele together with the Cre transgene in PCR analysis of the genomic DNA from tail biopsies results in loss of the allele in the thyroid. Hence, Igf1r^flox/flox and Igf1r^flox/wt mice that inherited the Cre transgene display Igf1r^−/− and Igf1r^+/− genotypes in thyrocytes, respectively. In fact, quantitative PCR analyses of reverse transcribed total RNA from thyroid showed significantly decreased Igf1r mRNA expression in knockout mice and strongly decreased mRNA expression in heterozygous mice (Fig. 1C). Furthermore, immunohistochemistry with an anti-IGF-IR antibody detects IGF-IR protein in Igf1r^−/− only in cells other than thyrocytes (Fig. 1D). Western blot analysis of thyroid tissue lysates clearly indicated that IGF-IR protein was reduced in the thyroid of Igf1r^−/− mice (Fig. 2A). No reduction of IGF-IR protein expression was found in the liver (Fig. 2A), brain, and skeletal muscle (data not shown).

Fig. 1.

Targeting strategy, assessment of Igf1r recombination, and Igf1r expression. A, Schematic representation of the WT (upper panel) and loxP-flanked Igf1r allele before (middle panel) and after (lower panel) recombination. Cre expression deletes exon 3 (numbered box) by recombination of the loxP sites flanking the Igf1r gene. F and R mark the location of the primers used in genotyping of tail biopsies. EcoRV, Restriction site; filled triangle, loxP site; ellipse, FRT site. B, Results from PCR analysis of DNA prepared from tail biopsies. DNA of WT mice produced a 300-bp band (lane 1), whereas a single 380-bp band was detected for loxP-flanked Igf1r allele (lane 3). Heterozygous expression of the transgene was detected by both a 300-bp and a 380-bp band (lane 2). C, Gene expression analysis of the Igf1r in the thyroid of WT, Igf1r^+/−, and Igf1r^−/− mice (n = 10). Significant down-regulation of the Igf1r mRNA expression in heterozygous and knockout mice compared with the WT is evident in the 2^ddCT values of Igf1r normalized to the housekeeping gene Hmbs. **, P < 0.01; ***, P < 0.001. D, Immunohistochemistry with anti-IGF-IR-antibody in thyroid tissue of female WT (Igf1r^+/+) and Igf1r^−/− mice. The black arrowhead marks a stained blood vessel among unstained thyrocytes in Igf1r^−/− mice.

Fig. 2.

Western blot analysis of proteins from WT (Igf1r^+/+), Igf1r^+/−, and Igf1r^−/− mice. A, Thyroid-specific Igf1r knockout strongly reduces IGF-IR expression in thyroid but not in liver tissue of Igf1r^−/− mice. In contrast, IGF-IIR protein is increased in thyroid tissue but not in liver of Igf1r^+/− and Igf1r^−/− mice compared with WT. β-Actin (ACTB) has been used as loading control. B, Activation of the AKT protein, a downstream signaling target of IGF-IR, is studied using a phosphorylation specific antibody (pAKT). Whereas expression of AKT is unchanged in thyroid and liver tissue of mice with different genotypes, phosphorylation of AKT is reduced in thyroids but not in liver of Igf1r^+/− and Igf1r^−/− mice. C, Activation of p42–44, a signaling target for several cell surface receptors that might play a role in the development of cancer. However, we do not detect differences of the expression of p42–44 as well as phosphorylated (phospho)-p42–44 in thyroids of Igf1r^+/− and Igf1r^−/− mice compared with WT.

Phenotype of Igf1r-knockout mice

Thyroid weight and follicle size

Thyroid weight of both female (Igf1r^−/−: 2.0 mg ± 0.5 vs. Igf1r^+/+: 1.7 mg ± 0.6) and male mice (Igf1r^−/−: 1.7 mg ± 0.5 vs. Igf1r^+/+: 2.1 mg ± 0.4), were not significantly different compared with wild-type (WT) littermates at the age of 4 months (Fig. 3A). As expected, thyroids of mice at the age of 1 y had a higher weight compared with the 4-month-old animals. However, we found no significant differences between knockouts and WT animals at the age of 1 yr (male Igf1r^−/−: 3.1 mg ± 0.6 vs. male Igf1r^+/+: 3.0 mg ± 0.3; female Igf1r^−/−: 3.2 mg ± 0.5 vs. female Igf1r^+/+: 2.8 mg ± 0.4).

Fig. 3.

Thyroid-specific Igf1r knockout only affects thyroid architecture. A, No significant changes in thyroid weight were detected in 4-month- and 1-yr-old male and female Igf1r^−/− mice compared with WT mice. B, The histogram shows the distribution of the thyroid follicle size of the three genotypes. Knockout (filled squares) and heterozygous (gray circles) mice show a high number of very large follicles that are not present in WT mice (open circles). Follicle sizes were measured for three distinct thyroid sections of three knockout and three heterozygous as well as four WT mice of both genders.

Figure 2B illustrates the size distribution of thyroid follicles as a histogram. In knockout and heterozygous mice we detect a high number of very large follicles that exceeds the range of sizes in WT mice. Moreover, the medium size of thyroid follicles is significantly increased in knockout mice (+140%) as well as in mice with a deletion of one Igf1r allele (+130%) compared with Igf1r^+/+ (1547 μm² ± 1397).

Hormone level

Measurement of the thyroid hormones (Fig. 4) showed no significant change of serum T₃ levels at the age of 4 months and a slight increase of 41% of T₃ in 1-yr-old male mice, respectively. A significant decrease was detected for serum T₄ at 8 wk in male (−37%) and female (−36%) Igf1r^−/− mice compared with WT. At 4 months and 1 yr T₄ levels of knockout animals are similar to those of WT mice, except for female knockout mice at 1 yr. TSH levels of Igf1r^−/− and Igf1r^+/− male and female mice (Fig. 5A) were significantly increased (male Igf1r^−/−: +754%, male Igf1r^+/−: +606% vs. male Igf1r^+/+: 131 mU/liter ± 53; female Igf1r^−/−: +933%, female Igf1r^+/−: +920% vs. female Igf1r^+/+: 30 mU/liter ± 15).

Fig. 4.

Serum T₃ and T₄ concentrations of mice after Igf1r loss. A, Serum concentrations of T₃ of male (upper graph) and female (lower graph) mice at 4 months and 1 yr (n = 5). B, Serum T₄ concentrations of male (upper graph) and female (lower graph) mice at 8 wk, 4 months, and 1 yr (n = 5). A significant reduction of the T₄ level was detected in male and female knockout mice at 8 wk. *, P < 0.05.

Fig. 5.

Serum TSH concentration and activity of serum TSH of mice after Igf1r loss. A, Strong increase of serum TSH in male and female mice with homozygous and heterozygous loss of Igf1r in thyroid tissue at 4 months and 1 yr. B, Increased cAMP concentrations in HEK hTSHR cells stimulated with Igf1r^+/− and Igfr1^−/− sera compared with Igf1r^+/+ indicate biologically active TSH. Sera of three mice per genotype were measured in triplicates in two independent experiments. C, Serum TSH reaches levels below the detection limit after injection of a T₃/T₄ mixture in WT and Igf1r^−/− mice compared with mock injection. n.d., Not detectable; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Bioactivity of TSH

Furthermore, cAMP accumulation was determined to analyze the bioactivity of serum TSH elevated in transgenic mice compared with WT mice (Fig. 5B). We used a cell model [human embryonic kidney (HEK) grip tide] with stable overexpression of the TSH receptor (TSHR) to test the sera of Igf1r^−/−, Igf1r^+/−, and WT mice for their ability to stimulate cAMP accumulation. Analyzing three sera of each genotype and sex, we found both sera of Igf1r^+/− and Igf1r^−/− mice to be able to induce cAMP accumulation in the cultured cells (male Igf1r^−/−: +256%, male Igf1r^+/−: +152% vs. Igf1r^+/+: 14.1 nm ± 3.2; female: Igf1r^−/−: +189%, female Igf1r^+/−: +190% vs. Igf1r^+/+: 8.8 nm ± 1.7), suggesting that the circulating TSH is biologically active.

Thyroid hormone feedback

After ip injection of thyroid hormone in 6-month-old male and female Igf1r^−/− and Igf1r^+/+ mice serum TSH fell below the detection level of our assay (Fig. 5C), which demonstrates an intact thyroid hormone feedback loop.

Insulin, IGF-I, and IGF-II

Measurement of the serum insulin and IGF-I level showed no significant change between WT and Igf1r^−/− mice at the age of 4 months and 1 yr [insulin: 237 pm ± 33 (Igf1r^+/+; 4 month) vs. 261 pm ± 38 (Igf1r^−/−; 4 month) and 265 pm ± 34 (Igf1r^+/+; 1 yr) vs. 245 pm ± 34 (Igf1r^−/−; 1 yr) and IGF-I: 436 pm ± 36 (Igf1r^+/+; 4 month) vs. 486 pm ± 27 (Igf1r^−/−; 4 month) and 564 pm ± 53 (Igf1r^+/+; 1 yr) vs. 579 pm ± 33 (Igf1r^−/−; 1 yr)].

For serum IGF-II we detect a slight trend (not significant) for increased levels of knockout vs. WT mice at 4 months and 1 yr [1.10 nm ± 0.14 (Igf1r^+/+; 4 month) vs. 1.37 nm ± 0.27 (Igf1r^−/−; 4 month) and 0.93 nm ± 0.10 (Igf1r^+/+; 1 yr) vs. 1.07 nm ± 0.07 (Igf1r^−/−; 1 yr)].

mRNA expression

In addition to hormone levels we studied other functional markers of thyroid physiology that could be affected by loss of IGF-IR signaling (Table 1). mRNA expression of the respective transcripts were analyzed after normalization to the housekeeping gene Hmbs, which demonstrates very stable and ubiquitous expression in different tissues, is less sensitive to retardation of tissue asservation, and is not affected by experimental manipulation in thyrocytes (23, 24). Among transcripts relevant for thyroid physiology we found a significantly decreased (−66%) mRNA expression of thyroid peroxidase (Tpo) in female mice, whereas male mice showed only a slight decrease (−35%) of Tpo mRNA expression. We did not detect a significant alteration of sodium-iodide-symporter (Nis) and Tg mRNA expression (Table 1), underlining that the expression of the Cre transgene under the Tg promotor does not interfere with normal Tg expression.

Table 1.

Gene expression analyses in WT, Igf1r^+/− and Igf1r^−/− thyroids of one year old male and female mice

	Tg			Nis			Tpo			Igf2r			Tshr			Insr
	Mean	±sd	n	Mean	±sd	n	Mean	±sd	n	Mean	±sd	n	Mean	±sd	n	Mean	±sd	n
Igf1r +/+, m	1.00	0.32	4	1.00	0.28	5	1.00	0.22	4	1.00	0.15	5	1.00	0.31	5	1.00	0.20	5
Igf1r +/−, m	0.81	0.48	5	1.07	0.30	5	0.65	0.07	5	1.28	0.20	5	1.75	0.24	5	0.75	0.14	5
Igf1r −/−, m	0.74	0.33	5	1.41	0.13	5	0.80	0.10	5	1.78 a	0.14	5	1.97	0.84	5	1.19	0.12	5
Igf1r +/+, f	1.00	0.44	4	1.00	0.25	7	1.00	0.07	5	1.00	0.08	5	1.00	0.08	5	1.00	0.06	5
Igf1r +/−, f	1.36	0.72	5	0.83	0.12	5	0.42a	0.08	5	1.93b	0.43	5	1.20	0.26	5	1.03	0.27	5
Igf1r −/−, f	1.02	0.29	4	0.64	0.27	6	0.33c	0.05	4	1.58c	0.18	4	0.96	0.38	4	1.00	0.13	4

Analyzed are the Tg, Nis, Tpo, Igf2r, Tshr, and Insr mRNA expression levels with Taqman probe kits from Applied Biosystems. Shown are the means of 2^ddCT values of the respective gene (no bold) normalized to the housekeeping gene Hmbs compared to the WT mice. Student's t test was performed per genotype and sex (P < 0.05 was considered significant).

^a P < 0.01;

^b P < 0.01;

^c P < 0.001. f, Female; m, male.

Reduced or lost IGF-IR signaling could induce the expression of related receptors like IGF-IIR and insulin receptor (INSR). We observed an increased expression (up to +90%) of the Igf2r mRNA that is significant in female heterozygous and homozygous mice as well as in homozygous male mice (Table 1). In contrast, no significant change of expression was detected for the mRNA of the Insr. With an up to 10-fold increase of serum TSH in mice with loss of Igf1r we expected compensatory down-regulation of the Tshr mRNA. However, quantitative RT-PCR showed no change or slightly increased Tshr mRNA expression (Table 1).

IGF-IIR and signaling proteins

In addition to the expression of Igf2r mRNA, we used Western blot analysis to study the expression of the IGF-IIR protein (Fig. 2A). IGF-IIR protein is increased in the thyroid of both Igf1r^+/− and Igf1r^−/− compared with WT mice.

Furthermore, we focused on the expression and activation of AKT protein, which is a part of the IGF-IR signaling cascade. In Western blot analysis (Fig. 2B) we did not detect changes in AKT expression for the different genotypes. However, we demonstrate a down-regulation of activated/phosphorylated AKT protein in thyroid tissue with partial (Igf1r^+/−) or total loss of Igf1r expression in thyrocytes, which could be a direct effect of a loss of IGF-IR signaling. In contrast, we do not detect changes in the expression or activation of p42–44 ERK (Fig. 2C), which acts downstream of AKT but also mediates the signaling of other receptors (e.g. epidermal growth factor receptor).

BRAF sequencing

To look for a possible genetic alteration in the genome of cells from papillary structures we studied the BRAF gene in microdissected tissue of Igf1r-deficient mice and screened for mutations in the region equivalent to the human V600E mutation. However, in mRNA from eight dissected papillary structures from six mice we only detected WT BRAF sequences.

Histology

Thyroids of Igf1r^−/− and Igf1r^+/− are characterized by a profound alteration of thyroid architecture that is defined by an increase in heterogeneity. We detect a significant increase of medium follicle size as well as a number of very small follicles (Fig. 6). The most prominent findings are papillary structures both in Igf1r^−/− and Igf1r^+/− mice that strongly resemble papillary thyroid hyperplasia. They were detectable in 18/21 Igf1r^−/− mice and three of three Igf1r^+/− mice. One of the Igf1r^+/− mice with papillary structures was studied at the age of 4 months, and the other two were studied at 1 yr. Igf1r^−/− mice developed papillary thyroid architecture in 13/15 cases at the age of 1 yr, four of five cases at the age of 8 months, and one of one mice at 4 months since birth. None of the WT animals showed similar changes in thyroid morphology.

Fig. 6.

Hematoxylin and eosin-stained thyroid slices of 1-yr-old mice at low (right panel) and high magnifications (left panel). The upper panels show a male WT mouse. Igf1r knockout (lower panel, male) and heterozygous (middle panel, female) mice show a heterogeneous follicle structure with very large follicles (stars). Igf1r^+/− and Igf1r^−/− mice developed papillary thyroid architecture (arrowheads) resembling papillary hyperplasia.

A thyroid marker protein involved in thyroid hormone synthesis is the TPO, which is identified with decreased levels in malignant thyroid tissue, e.g. papillary thyroid carcinomas (25). We detect a reduced expression of the Tpo mRNA in mice with targeted Igf1r loss (Table 1). Immunohistological staining for the TPO protein in Igf1r^−/− and Igf1r^+/− mice showed a reduction in the papillary regions compared with the WT and the normal thyroid tissue (Fig. 7A). An alteration in the normal thyroid tissue of Igf1r^−/− and Igf1r^+/− mice compared with the WT was not detectable.

Fig. 7.

Immunohistological staining of 1-yr-old male Igf1r^+/+, female Igf1r^+/−, and male Igf1r^−/− mice at low and high magnifications. Thyroids were stained with antibodies against TPO (A), NIS (B), TG (C), and T₄-TG (D). A, There is no difference in the TPO staining of normal thyroid tissue between Igf1r^+/+ and Igf1r^−/− mice (arrowheads, upper and middle right panel), whereas the papillary hyperplasia in Igf1r^−/− (lower right panel) shows only slight TPO staining. B, NIS staining is detected not only in the follicular structures of small and large follicles (star, lower right panel) but also in the papillary structures (arrowhead, middle right panel). C, Homogenous TG-stained thyrocytes in Igf1r^+/+, Igf1r^−/− mice and magnification of a stained large follicle (right panel). D, Thyroids of both Igf1r^+/+ and Igf1r^−/− mice contained tetraiodinated TG in their follicles.

For other thyroid markers such as NIS, we found homogenous stained thyrocytes in Igf1r^−/−, Igf1r^+/−, and Igf1r^+/+ mice (Fig. 7B). The same is seen for TG staining (Fig. 7C). There is no visible difference in the intensity of the stained follicle with normal architecture between the transgenic and WT mice. To detect changes in the iodination status of follicular TG, thyroids were stained with a T₄-TG antibody. As shown in Fig. 7D. thyroids of both Igf1r^−/− and Igf1r^+/− mice contained tetra-iodinated thyroxyl residues in the TG protein in their follicles, which indicates normal thyroid hormone synthesis.

Caspase-3 staining

To study the consequences of targeted Igf1r inactivation on apoptosis we stained thyrocytes in Igf1r^−/− and Igf1r^+/+ mice with an antibody against activated caspase-3 (Asp175). Activation of caspase-3 is an important marker of cellular apoptosis induced by a wide variety of apoptotic signals (26). An index of stained to all follicular thyrocytes was calculated in three slices of three WT and three Igf1r^−/− mice. In thyroid tissue of Igf1r-deficient mice we detect a reduced index (−56%; 4.5% ± 5.3) compared with WT mice (10.2% ± 3.7).

Hypothalamus-pituitary-thyroid axis

To study the status of the hypothalamus feedback regulation, we determined the transcript levels of TRH (Trh) a target gene of thyroid hormone action in the paraventricular hypothalamic nucleus (PVN) of the hypothalamus. Within the negative-feedback loop of the hypothalamus-pituitary-thyroid (HPT) axis, Trh mRNA expression in the PVN is inversely correlated to thyroid hormone levels and thus represents a suitable sensor of intracellular T₃ content. Compared with WT animals, Trh transcript levels were 18% up-regulated in PVN neurons of Igf1r^−/− mice (Igf1r^−/−: 13.4±1.9 vs. Igf1r^+/+: 11.3 ± 1.0, P = 0.13) (Fig. 8). The slightly elevated Trh mRNA levels in the PVN of Igf1r^−/− mice prompted us to analyze other components of the HPT axis. However, D2 transcript levels in hypothalamic tanycytes of Igf1r^−/− mice were not affected by the Igf1r knockout (Igf1r^−/−: 16.8 ± 1.4 vs. Igf1r^+/+: 18.2 ± 5.3, P = 0.35) (Fig. 8).

Fig. 8.

In situ hybridization of the Trh and Dio2 gene in the hypothalamic region of female WT and Igf1^−/− mice.

Body weight

At the age of 4 months, Igf1r^−/− male mice exhibit a 10% increase (Igf1r^−/−: 32.5 g ± 6.2 vs. Igf1r^+/+: 27.8 g ± 1.1) in body weight compared with Igf1r^+/+ mice (Fig. 9A). These differences were amplified at the age of 1 yr, when Igf1r^−/− male mice exhibited a 22% body weight increase (Igf1r^−/−: 43.8 g ± 5.6 vs. Igf1r^+/+: 35.7 g ± 2.2). Although these gains in body weight in male mice are consistent with a latent hypothyroidism, we unexpectedly saw a body weight decrease in female Igf1r^−/− mice at the age of 4 months (Igf1r^−/−: 20.9 g ± 0.6 vs. Igf1r^+/+: 25.2 g ± 3.3) and 1 yr (Igf1r^−/−: 25.3 g ± 3.7 vs. Igf1r^+/+: 35.9 g ± 6.8).

Fig. 9.

Consequences of thyroid-specific Igf1r knockout on body composition of male mice. A, Body weight of 4-month- and 1-yr-old male Igf1^−/− mice are significantly increased compared with WT (Igf1^+/+) mice. B, EAT weight in Igf1^+/− and Igf1^−/− mice showed no difference compared with WT mice. *, P < 0.05. BM, Body mass.

Epigonadal adipose tissue (EAT) weight

A similar distribution is seen in the relative EAT weight (calculated as percentage of the whole body weight, Fig. 9B). At the age of 4 months Igf1r^−/− male and female mice exhibit 81% increased (Igf1r^−/−: 2.0 g ± 0.8 vs. Igf1r^+/+: 1.1 g±0.3) and 13% decreased (Igf1r^−/−: 0.7 g ± 0.2 vs. Igf1r^+/+: 0.8 g ± 0.2) EAT weight, respectively (Fig. 9B). However, the heterozygote littermates show no differences compared with the WT. Those differences were not observed in male mice at the age of 1 yr, where Igf1r^−/− male and female mice exhibited 8% decreased (Igf1r^−/−: 3.9 g ± 1.4 vs. Igf1r^+/+: 4.2 g ± 1.1) and 55% decreased (Igf1r^−/−: 1.8 g ± 1.2 vs. Igf1r^+/+: 4.0 g ± 2.2) EAT weight, respectively (Fig. 9B).

Metabolic effects of the thyroid-specific Igf1r knockout

Igf1r-specific knockout in adipose tissue revealed alterations in insulin tolerance tests (ITT) and glucose tolerance tests (GTT) (22). To determine the physiological consequences of reduced thyroid specific Igf1r expression on metabolic parameter glucose, we also performed ITT and GTT at the age of 3 months. Intraperitoneal ITT showed low insulin resistance in male and no changes in female Igf1r^−/− mice compared with control mice (data not shown). Intraperitoneal GTT demonstrated normal glucose tolerance in male and female Igf1r^−/− and control mice (data not shown).

Neurological effects of the Igf1r knockout

Alterations in the thyroid hormone status often cause behavioral abnormalities, e.g. anxiety and depressive and cognitive disorders (27, 28). To test for locomotor deficiencies or anxiety-like behavior, we performed the open-field test, which is a basic test of behavioral phenotyping for motional and emotional behavior (29). We found reduced entered field numbers in Igf1r^−/− mice (Igf1r^−/− +30% vs. Igf1r^+/+: 18.5 fields/min ± 4.9, n = 6) with no difference regarding which fields were entered (corner, bolder, or middle field). Therefore, Igf1r^−/− mice show no abnormal anxiety but reduced activity compared with the WT mice. To further analyze the reduced motional attendance of Igf1r^−/− mice, we placed them at the age of 15 wk for 7 d in wheel cages. Male Igf1r^−/− mice exhibit significantly reduced rotations over a period of 7 d (Igf1r^−/− −47% vs. Igf1r^+/+: 3074 ± 829 rotations per day, n = 7). The female knockout mice showed just an akin motional attendance compared with control animals.

Discussion

Our study, for the first time, describes the consequences of thyroid-specific loss of IGF-IR signaling. Our genetic mouse model disrupts the expression of the Igf1r gene in the thyroid and thereby allows the dissection of normal extrinsic thyroid stimulation and gives important insights into thyroid growth regulation. Although thyroid weight appeared normal, we detected a transient decrease of serum T₄ at 8 wk, which returned to a normal level from 4 month on together with a tremendous increase in serum TSH. Moreover, the strong increase of serum TSH can be abolished by exogenous thyroid hormone treatment, which suggests that the thyroid hormone feedback is functioning normally.

Reduced serum T₄ at 8 wk indicates early transient hypothyroidism, which later classifies as subclinical hypothyroidism characterized by normal thyroid hormone and high TSH together with the symptoms of weight gain and reduced activity. Increased body and EAT weight, as well as reduced motional attendance, would be the typical results of such a classification which perfectly applies to male mice with Igf1r loss. Unexpectedly, we detected a reversed metabolic phenotype in females. This is not easily explained, and at present we cannot exclude for instance that a subtle phenotype of the pregnant Tg Cre mothers interferes with the fetal programming of the metabolic setpoint, an epigenetic mechanism known to lead to a sexual dimorphism in offspring metabolism.

However, the thyroid histology of all mice with loss of Igf1r is very similar but not completely in line with human latent hypothyroidism, which is histologically dominated by signs of autoimmune disease. In fact, thyroid weight is normal, and histology reveals great heterogeneity with foci of papillary structures. These structures present as papillary thyroid hyperplasia because we did not detect evidence of capsular or vascular invasion and did not detect BRAF mutations frequently seen in human papillary carcinoma. Papillary structures are already detectable at 4 months and reach a frequency of about 86% in mice with complete loss of the Igf1r alleles. Importantly, we see papillary thyroid architecture with a similar frequency also in mice with loss of one Igf1r allele. This strongly suggests that this pathological histology is the result of a gene dosage effect of the IGF-IR and could also be related to the strong increase in serum TSH. However, nontransgenic mice treated with antithyroid drugs to cause hypothyroidism along with high TSH levels rarely develop thyroid cancer (30). In contrast, transgenic mice with chronic cAMP stimulation develop papillary growth but show signs of papillary cancer only very late (31). Moreover, a correlation between TSH and risk of papillary thyroid cancer in patients with nodular thyroid disease has been reported (32). The underlying mechanism could involve activation of the MAPK pathway stimulated by strongly elevated serum TSH levels as discussed by Nikiforov (33). On the level of TSHR signaling such positive cross talk between the TSH/cAMP and MAPK/ERK pathways could result from activation of G protein subunits βγ and involve the PI3K pathway (34). Moreover, in our mice, elevation of TSH is very likely high enough to also stimulate the cAMP-independent Gq signaling of the TSHR (35). However, we do not detect changes in the expression or activation of p42–44 ERK protein, which is central to the signaling cross talk mentioned above. Therefore, the very early development of papillary thyroid lesion might result from TSH stimulation in an unexpected way. However, high TSH might rather support the progression of a preexisting lesion (36). If and how disruption of IGF-I signaling causes the primary lesion is still unknown. But another hypothesis for the development of papillary structures comes from in vitro studies with embryonic stem cells. As shown by Arufe et al. (37) the differentiation of murine stem cells into thyrocytes depends on IGF-I and IGF-IR. Therefore, aberrant differentiation caused by loss of Igf1r could be the cause of the papillary structures. However, the coexistence of both normal and papillary thyroid architecture in our model suggests that there is at least partial compensation by other growth factors that might not be present in an in vitro setting or the papillary growth is caused by an additional lesion.

Furthermore, our model produces an unexpected outcome concerning the HPT axis. Normally TSH production in the thyrotroph of the hypophysis is regulated by feedback inhibition through circulating thyroid hormone and via stimulation by hypothalamic TRH (38). Because our mice with loss of Igf1r expression in the thyroid have normal or even slightly elevated thyroid hormone levels and in situ hybridization in the PVN shows normal TRH expression, a normal serum TSH would be expected. However, we detected an increase of serum TSH in both the Igf1r^−/− and the Igf1r^+/− mice in the range of 6- to 9-fold. One plausible explanation could be reduced bioactivity of the TSH (39) produced in transgenic animals. However, our sensitive cell model-based assay with stable overexpression of the TSHR indicated very strong cAMP accumulation in response to sera of Igf1r^−/− and Igf1r^+/− compared with WT mice. Although the mode of regulation is unknown, we hypothesize that elevated functionally active TSH compensates for the loss of IGF-I signaling in respect to thyroid hormone production.

The loss of IGF-I signaling is conditionally targeted to the thyroid with a Tg Cre expression system. As a result, one or two alleles of Igf1r are lost in differentiated thyrocytes. Hence, we detect a decrease of Igf1r mRNA in thyroid tissue of heterozygous mice and a further loss of mRNA in mice with loss of both Igf1r alleles. The latter show remaining Igf1r mRNA expression from cells other than thyrocytes (e.g. c-cells, fibroblasts, endothelial cells) that do not activate the Tg promoter. Furthermore we see a concurrent down-regulation of activated AKT protein, which is a major regulator of the downstream insulin cascade and has been shown in vitro by Zhang et al. (40). The Tg-targeted Cre system has been used for thyroid-selective gene expression in many transgenic studies (41). Recently it was used to impair the Gq/G11-signaling cascade of the TSHR and to cause IGF-IR overexpression (5, 42). Furthermore, in a mouse model of Zeiger et al. (31) the Tg promoter gene has been used for the thyroid-specific overexpression of the choleratoxin A gene.

Although increased TSH stimulation might keep thyroid hormone levels in a normal range, the loss of Igf1r signaling could lead to other systemic or local compensation that would maintain homeostasis. However, we did not detect changes of serum insulin and IGF-I level. At the level of mRNA expression we detected increased expression of the Igf2r in thyroid tissue of Igf1r-deficient mice whereas the insulin receptor and the Tshr mRNA are unchanged compared with WT mice. Unchanged expression of the Tshr mRNA is an especially unexpected outcome because high levels of bioactive serum TSH mean a very strong stimulus for thyrocytes that could lead to down-regulation of Tshr mRNA expression. However, gene expression analysis of primary thyrocytes did not detect TSHR as differentially down-regulated by TSH treatment (23). Moreover, recent work by Moeller et al. (43) implies that the TSHR expression level is set by the thyrocyte and not in response to settings of the HPT loop. Moreover, we detect a reduced mRNA expression of Tpo that is more difficult to interpret considering normal serum level of thyroid hormones. However, this reduction very likely reflects the observation that papillary structures hardly express TPO (Fig. 7A) and thereby dilute TPO expression of normal thyroid tissue due to RNA preparation from whole thyroid glands.

In addition to the increase of Igf2r mRNA the expression of the IGF-IIR protein is also increased. Moreover there is also a slight but not significant increase of serum IGF-II. Together this suggests that in addition to TSH part of the compensatory effect could be attributed to IGF-II signaling.

Our model fits perfectly with results from the reverse mice model: overexpression of the IGF-IR together with IGF-I in thyroid tissue (5). The latter mice develop goiter and show reduced serum TSH level and slightly increased T₄. This could indicate that IGF-IR signaling is less essential for thyroid hormone synthesis but maintains homeostasis and normal thyroid morphogenesis. Our results further suggest that the strong increase of TSH benefits papillary growth and, together with a moderate increase of the IGF-IIR, completely compensates the loss of IGF-IR signaling at the level of thyroid hormones without a significant increase in thyroid weight. However, a direct transfer of our data to human pathology, especially papillary tumorigenesis, would not be appropriate until we better understand the mechanism that leads to papillary structures and compensates hormone synthesis.

Materials and Methods

Generation of Igf1r^TgCre mice

Igf1r^flox/wt mice (22) were crossed with mice on a mixed (C57BL/6 × 129/Sv) genetic background carrying the mouse Tg Cre transgene provided by Professor S. Offermanns (Institut for Pharmacology, University of Heidelberg, Germany) (42).

Mice lacking one or two WT alleles of the Igf1r in the thyroid (Igf1r^+/− or Igf1r^−/−) were derived by crossing male Igf1r^flox/wt mice with female Igf1r^flox/wt mice expressing the Cre recombinase under the control of the Tg promoter/enhancer on one allele (Fig. 1A). All mice were housed in pathogen-free facilities in groups of three to five animals at 22 ± 2 C on a 12-h light, 12-h dark cycle. Animals were fed a standard chow diet (Altromin, Lage, Germany) and had also ad libitum access to water. Food was withdrawn only if required for an experiment. All experiments were performed in accordance with the rules for animal care of the local government authorities (Landesdirektion Leipzig, Germany) and were approved by the institutional animal care and use committee.

Molecular characterization and genotyping of mice

Genotyping was performed by PCR using genomic DNA isolated from the tail tip. In brief, genomic DNA was prepared by using the DNeasy kit (QIAGEN, Hilden, Germany). The following two primer pairs were used to genotype Igf1r loxP sites: 5′-TCC CTC AGG CTT CAT CCG CAA-3′ (forward) and 5′-CTT CAG CTT TGC AGG TGC ACG-3′ (reverse), as well as the Tg Cre recombinase 5′-AGT CCC TCA CAT CCT CAG GTT-3′ (forward) and 5′-ATG CCA ACC TCA CAT TTC TTG-3′ (reverse). PCR was performed for 30 cycles (loxP sites), 35 cycles (Tg Cre) of 95 C, 59 C (loxP sites), or 61 C (Tg Cre) and 72 C (30 sec each) using the QIAGEN Taq Polymerase. DNA from mice with the Igf1r^wt allele produces a 300-bp band. A 380-bp band is the PCR product of DNA with the Igf1r^flox allele (Fig. 1B). All mice that were genotyped Igf1r^flox/flox or Igf1r^flox/wt and were positive for the Tg Cre transgene were considered Igf1r^−/− or Igf1r^−/+ in the thyroid, respectively. In the following we only refer to the genotypes of the thyroid. For Western blot analysis, tissues were removed and homogenized in homogenization buffer with a mixer mill MM400 (Retsch, Haan, Germany), proteins were isolated using standard techniques, and Western blot analysis was performed with antibodies raised against the IGF-IR β-subunit (C20; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), phospho-AKT (Ser473) (Cell Signaling Technology, Danvers, MA), AKT (pan) (Cell Signaling), p42–44 (Cell Signaling), phosphor p42–44 (Cell Signaling), IGF-IIR/M6PR (Antibodies Online, Aachen, Germany), and β-actin (Sigma, Steinheim, Germany) as loading control.

Phenotypic characterization

Five mice of each genotype [Igf1r^−/−, Igf1r^+/− or WT (either Igf1r^wt/wt and Igf1r^flox/flox without or Igf1r^wt/wt with the Tg Cre transgene)] of both sexes were studied from an age of 1 month up to 1 yr of life. Body weight was recorded weekly. ip glucose GTT and ITT were performed at the age of 3 months. GTT was performed after an overnight fast for 16 h by injecting 2 g/kg body wt glucose and measuring the blood glucose levels after tail vein incision at 0 (baseline), 10, 30, 60, and 120 min after injection. ITT was performed in random-fed animals by injecting 0.75 U/kg body wt human regular insulin (40 U Actrapid; Novo Nordisk, Copenhagen, Denmark). Glucose levels were determined in blood collected from the tail tip immediately before and 15, 30, and 60 min after the ip injection.

Mice were killed at the age of 4 months and 1 yr by an overdose of carbon dioxide. Liver, heart, brain, lung, kidney, spleen, skeletal muscle, sc, brown, and EAT were immediately removed. Serum was collected at 8 wk, 4 months, and 1 yr, respectively, and concentrations of TSH, T₃, and T₄ were measured at the University of Chicago by Dr. R. Weiss (44). ELISA or RIA were used to detect serum levels of insulin (ELISA from DRG, Marburg, Germany), IGF-I, and IGF-II (ELISA and RIA from Mediagnost, Reutlingen, Germany) according to the instructions of the manufacturer.

To examine the function of the thyroid hormone feedback, we studied 6-month-old male and female mice with Igf1r^+/+ and Igf1r^−/− genotype for their response to exogenous T₃/T₄ treatment as described by Flamant et al. (45). Groups of three male and female mice with WT and Igf1r^−/− genotype were daily injected ip with a mixture of T₄ and T₃ (2.5 mg/kg T₄ and 0.25 mg/kg T₃ in PBS) or vehicle (PBS) for 5 d. Mice were killed and serum was collected as described above.

Animals were placed for 7 d in single cages with running wheels to record their physical activity as turns per day. To test for neurological alterations of the Igf1r knockout mice a standardized open field test was performed in which the spontaneous locomotor activities of the mice were manually quantified. For that, mice were placed for 5 min in a dark test area with an area of 1 m (2) and 25 white framed quarters. Every entered field was counted and calculated to the whole entered field number.

cAMP accumulation assays

For cAMP assays 0.7 × 10⁴ HEK cells stably expressing the human TSHR were platted in 96-well plates for 24 h, washed once in serum-free DMEM, followed by a incubation with the same medium containing 1 mm 3-isobutyl-1-methylxanthine (Sigma) in a humidified 5% CO₂ incubator. At the same time, cells were stimulated with appropriate concentrations of bovine TSH and 10 μl of the mice sera for 1 h in 37 C. Reactions were terminated by aspiration of the medium and addition of 150 μl 0.1 n HCl. Supernatants were collected and dried. cAMP content of the cell extracts was determined with a commercial α-screen kit (PerkinElmer, Rodgau, Germany) according to the manufacturer's instructions (46, 47).

In situ hybridization

After the animals were decapitated, brains were removed rapidly, embedded in 2-methylbutane (Sigma) and frozen on dry ice. Sections (20 μm) were cut on a cryostat (Leica, Bentheim, Germany), thaw mounted on silane-treated slides, and stored at −80 C until further processing. In situ hybridization was carried out as described previously (48, 49). Briefly, frozen sections were fixed in a 4% phosphate-buffered paraformaldehyde solution (pH 7.4) for 1 h at room temperature, rinsed with PBS, and treated with 0.4% phosphate-buffered Triton X-100 solution for 10 min. After washing with PBS and water, tissue sections were incubated in 0.1 m triethanolamine (pH 8) containing 0.25% (vol/vol) acetic anhydride for 10 min. After acetylation, sections were rinsed several times with PBS, dehydrated by successive washing with increasing ethanol concentrations, and air dried.

After application of the labeled cRNA probes (final concentration: 5 ng/μl for digoxigenin labeling; 25.000 cpm/μl for ³⁵S labeling), sections were coverslipped and incubated in a humid chamber at 58 C for 16 h. Thereafter, coverslips were removed in 2× saline-sodium citrate (SSC) (0.3 m NaCl; 0.03 m sodium citrate, pH 7.0). The sections were then treated with ribonuclease A (20 μg/ml) and ribonuclease T₁ (1 U/ml) at 37 C for 30 min. Successive washes followed at room temperature in 1×, 0.5×, and 0.2× SSC for 20 min each and in 0.2× SSC at 65 C for 1 h. For digoxigenin-labeled probes, sections were rinsed with P1 (100 mm Tris-HCl; 150 mm NaCl, pH 7.5) and then incubated for 2 h in blocking solution provided by the manufacturer of the kit. After incubation overnight with an antidigoxigenin antibody conjugated with alkaline phosphatase (1:1000 dilution; Roche, Indianapolis, IN), the tissue sections were washed with P1. Staining proceeded for 2–6 h in substrate solution containing nitroblue tetrazolium (NBT) chloride (340 μg/ml NBT; Biomol, Hamburg, Germany), X-phosphate (175 μg/ml 5-bromo-4-chloro-3-indolyl phosphate; Biomol), 100 mm Tris-HCl, 100 mm NaCl, and 50 mm MgCl₂ (pH 9.5). For radioactive probes, the tissue was dehydrated and exposed to Kodak Biomax MR Film (Sigma) for 48 h. For microscopic analysis, the sections were dipped in Kodak NTB2 (Integra Biosciences, Zizers, Switzerland) nuclear emulsion and stored at 4 C. After exposure for 14 d, autoradiograms were developed in Kodak D19 (Sigma) for 4 min and fixed in Kodak Rapid Fix (Sigma) for 4 min. The sections were photographed under dark- or bright-field illumination, and the expression was quantified using the NIH ImageJ software. Sense probes were used to confirm the specificity of the hybridization reaction and did not show any signal.

Immunhistochemistry

Tissue was fixed in 4% buffered formaldehyde and embedded in paraffin. Sections at 0.3 μm were cut with an ultramicrotome (Reichert, Vienna, Austria). Multiple sections were obtained from thyroid and analyzed systematically with respect to follicle size and number. For each genotype, at least five (100× magnifications) per slide were analyzed. Paraffin-embedded sections (3 μm) were dewaxed and rehydrated. Subsequently, sections were pretreated in a microwave oven in 0.1 m citrate buffer (pH 6) for one cycle at 750 W for 3 min and for four cycles of 3 min at 350 W. The LSAB+ System (DAKO Cytomation, Hamburg, Germany) was used for immunodetection. Briefly, slides were incubated with 3% H₂O₂ for 30 min followed by three wash steps with PBS/1% BSA. Unspecific binding was blocked for 30 min, and slides were incubated with monoclonal antibodies against NIS (C-term; Acris Antibodies, San Diego, CA), TG (open biosystems), T₄ (Mybiosource) and TPO (50). Antibodies for cleaved caspase-3 (Asp 175) were obtained from Cell Signaling Technology and used at a concentration of 1:100. Immunoreactivity was demonstrated using a biotinylated secondary antirabbit antibody, streptavidin-conjugated peroxidase, and diamino-benzidine as substrate. Sections were counterstained with hemalaun and mounted in Aquatex (Merck & Co., Inc., Whitehouse Station, NJ).

Quantitative real-time PCR

mRNA expression was measured by quantitative real-time PCR in an ABI PRISM 7000 sequence detector (Applied Biosystems, Darmstadt, Germany) using the TaqMan gene expression assay based on 5′-nuclease chemistry. Total RNA was extracted from organs using TRIzol reagent (Invitrogen GmbH, Karlsruhe, Germany), and 0.5 or 1 μg RNA was reverse transcribed with standard reagents (Life Technologies, Darmstadt, Germany). From each cDNA, 1 μl was amplified in a 10-μl PCR according to manufacturer's instructions (Applied Biosystems). The following TaqMan probe kits were used: Tg (Mm00447525_m1), Igf1r (Mm00802831_m1), Igf2r (Mm00439576_m1), Tshr (Mm00442027_m1), Insr (Mm01211875_m1), Nis (Mm00475074_m1), Tpo (Mm00456355_m1), and Trh (Mm01182425_g1).

BRAF sequencing

To test for genetic alterations that could be the cause of papillary structures in mice with targeted Igf1r inactivation, we microdissected eight papillary structures from six male and female mice with both Igf1r^+/− and Igf1r^−/− genotypes (51). We first prepared total RNA using the miRNeasy FFPE kit from QIAGEN according to the instructions of the manufacturer. We then amplified using RT-PCR (see above) part of the BRAF exon that contained the mouse equivalent of human codon 600 frequently mutated in human papillary thyroid carcinoma (52). Primer sequences were as follows: 5′-TCC AGA CAA CTG TTC AAA CTG-3′ and 5′-ATA TAT TTC TTC ATG AAG ACC-3′. The resulting PCR products were directly sequenced using the BigDye Terminator Kit (Life Technologies) on a 3100 Genetic Analyzer (Life Technologies).

Data analysis and statistics

Data are given as means ± sd. Datasets were analyzed for statistical significance using one-way ANOVA corrected by Bonferroni-Holm and a two-tailed unpaired Student's t test using the GraphPad Prism 5.02 (Graph Pad Software Inc., La Jolla, CA). P < 0.05 was considered significant.