Thyroid-Specific Enhancer-Binding Protein/NKX2.1 Is Required for the Maintenance of Ordered Architecture and Function of the Differentiated Thyroid

Abstract

Thyroid-specific enhancer-binding protein (T/ebp)/Nkx2.1-null mouse thyroids degenerate by embryonic day (E) 12–13 through apoptosis whereas T/ebp/Nkx2.1-heterogyzgous mice exhibit hypothyroidism with elevated TSH levels. To understand the role of T/ebp/Nkx2.1 in the adult thyroid, a thyroid follicular cell-specific conditional knockout (KO) mouse line, T/ebp(fl/fl); TPO-Cre, was established that expresses Cre recombinase under the human thyroid peroxidase (TPO) gene promoter. These mice appeared to be healthy and exhibited loss of T/ebp/Nkx2.1 expression in many, but not all, thyroid follicular cells as determined by immunohistochemistry and real-time PCR, thus presenting a T/ebp-thyroid-conditional hypomorphic mice. Detailed analysis of the thyroids from T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) mice, where the latter mouse line is derived from crosses with the original T/ebp/Nkx2.1-heterozygous mice, revealed that T/ebp(fl/fl); TPO-Cre mice can be classified into two groups with different phenotypes: one having atrophic/degenerative thyroid follicles with frequent presence of adenomas and extremely high serum TSH levels, and the other having an altered thyroid structure with reduced numbers of extraordinary dilated follicles consisting of excessive numbers of follicular cells as compared with those usually found in the normal thyroid. The latter phenotype was also observed in aged T/ebp(fl/ko) mouse thyroids. In vitro three-dimensional thyroid primary cultures using thyroids from T/ebp(fl/fl); TPO-Cre, T/ebp(fl/ko), and T/ebp(fl/fl) mice, and the latter treated with recombinant adenovirus with and without Cre expression, demonstrated that only cells from T/ebp(fl/fl) mice without adeno-Cre treatment formed follicular structures. Taken together, these results suggest that T/ebp/Nkx2.1 is required for maintenance of the normal architecture and function of differentiated thyroids.

The thyroid gland develops from the primitive pharynx as a thickened epithelium, which can be distinguished as the thyroid primordium in mouse embryos at embryonic d 9.5 (E9.5) (1). The thyroid primordium descends during E10–14 to its final location ventral to the caudal aspect of the condensing laryngeal cartilage. It then bifurcates to give rise to the two lobes of the thyroid gland located on either side of the larynx that are the origin of thyroid hormone-producing follicular cells. By E14, the ultimobranchial bodies, derived from the fourth pharyngeal pouches, are fused to the thyroid lobes, and their cellular components disseminate within it, ultimately giving rise to the calcitonin-producing parafollicular or C cells (1–5).

The thyroid gland presents a highly organized architecture characterized by spherical and/or spheroidal structures called follicles that are composed of a single layer of epithelial cells, so-called thyroid follicular cells or thyrocytes, enclosing a space called the follicular lumen where thyroid hormone precursor colloid is stored (3, 6). The follicles have been defined as the functional unit of the thyroid (3, 4), and around E14 –15 begin expressing a specific set of genes encoding proteins that are essential for producing thyroid hormones, including thyroglobulin, thyroid peroxidase (TPO), TSH receptor, and sodium iodide symporter, which directly participate in thyroid hormone synthesis. These genes are regulated by transcription factors such as thyroid-specific enhancer-binding protein (T/ebp) (also called Ttf1, Titf1, or Nkx2.1), Foxe1 (Ttf2), and Pax8 (3, 4, 7).

Thyroid folliculogenesis has been studied mainly using porcine thyroid cultures on dishes (8, 9), in suspension (10), or in collagen gel three-dimensional structures (11, 12). TGFβ1 was shown to have an effect on converting follicular cells to mesenchyme-like cells and inhibiting thyroid function (9, 10, 12). The importance of apical polarization on folliculogenesis was documented by primary culture studies (8, 13) as well as FRT cells transfected with gap junction protein, connexin-32 (14). Despite these studies, the mechanism of folliculogenesis and the identification of genes involved in this process are not fully understood.

T/ebp, Foxe1, and Pax8, each containing a homeobox (15, 16), forkhead (17), and paired-box (18) domain, respectively, are the three transcription factors that are essential for regulating the expression of the aforementioned genes involved in thyroid hormone synthesis (3, 4, 7). When the expression of these three transcription factors are disrupted, mice presented disorganogenesis or agenesis of the thyroid gland, thus demonstrating that these transcription factors are required not only for thyroid-specific gene expression, but are also essential for thyroid organogenesis (19–21).

We previously produced T/ebp-null mice that die at birth due to profoundly hypoplastic lungs (20, 22), a severely defective hypothalamus, and absence of thyroid and pituitary glands (20). The thyroid primordium, however, is present at E10 in these mouse embryos, but is lost by E12–13 through apoptosis (23). Conditional knockout (KO) mice that delete T/ebp gene specifically in the thyroid with the use of the Cre-loxP system would provide a means to understand the role of T/ebp in development of the thyroid beyond E12–13 and in the maintenance of thyroid after the completion of differentiation. On the other hand, T/ebp-heterozygous mice were shown to exhibit hypothyroidism with elevated TSH levels and have a neurological defect (24), although they appeared to be healthy and were fertile (20). This thyroid phenotype was caused by a reduction in expression of the TSH receptor due to T/ebp haploinsufficiency (25). These results suggest that informative phenotypes may be found in the thyroids of T/ebp-heterozygous mice once they are subjected to extensive analysis. However, to date, detailed studies on the phenotypes of the T/ebp-heterozygous mouse thyroids have not been carried out.

In the present study, we have established and characterized thyroid-specific T/ebp-conditional KO mice. In these mice, not all thyroid follicular cells have the T/ebp gene deleted, and thus result in a conditional hypomorph. Detailed analysis of thyroids from the T/ebp-thyroid-conditional hypomorphic and T/ebp-heterozygous mice, together with the results obtained from in vitro three-dimensional thyroid primary culture studies, demonstrate that T/ebp is essential for maintenance of the normal architecture and function of differentiated thyroids.

RESULTS

Generation of a Conditional Allele of the T/ebp Gene

A targeting vector containing three loxP sites in the same orientation was constructed (Fig. 1A). The first loxP site with an artificial BamHI site was inserted at the NsiI site located in intron 1 of the T/ebp gene, and the neomycin-resistant cassette flanked by a loxP site at each end was inserted between two XhoI sites located approximately 2.3 and 2.8 kb downstream of the T/ebp gene. After standard electroporation and culture of embryonic stem (ES) cells, one homologous recombinant was identified out of approximately 500 ES cells screened by Southern blotting using 5′- and 3′-external probes, and BamHI and StuI restriction enzyme digested genomic DNAs, respectively (Fig. 1B). The targeted ES cells were injected into C57BL/6 blastocysts, and highly chimeric male mice were bred with C57BL/6 females to generate heterozygous loxP targeted/wild-type mice (t/+). These mice were further crossed with an EIIa-Cre transgenic line to remove the neomycin cassette by recombination in the embryo to produce mice with a T/ebp-floxed allele (fl/+). Recombination events were analyzed by Southern blotting of tail DNAs with the probes described above and a neo probe (data not shown). Mice heterozygous for the floxed T/ebp allele [T/ebp(fl/+)] were bred with the original T/ebp heterozygous mice [T/ebp (+/ko)] (20) and then TPO-Cre transgenic mice (26). From this stage on, genotyping was carried out using PCR as described in Materials and Methods (Fig. 1C). This breeding scheme eventually generated mice for use in this study with the following genotypes: T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, T/ebp(fl/ko). All lines of mice were fertile and appeared to be healthy based on body weight, death rate, and breeding ability for at least a year.

Fig. 1. Generation of the T/ebp-Floxed Mouse.

A, Diagram of the T/ebp gene (wild-type allele), targeted allele, floxed allele, deleted allele, and the original KO allele. The targeting vector was composed of EcoRI-EcoRI and EcoRI-EcoRV fragments, shown on the top. Shaded and open boxes represent coding and noncoding regions of the gene, respectively, and exons are numbered. The neomycin cassette is shown by an open box marked with pGK-Neo for the targeted allele and the black box marked with Neo for the original KO allele. Solid triangles represent the loxP sites. The probes (5′-probe and 3′-probe) used to assess recombination events and the expected lengths of restriction enzyme fragments are indicated. The locations of PCR primers (5neo/3neo and 5KO/3KO) used for genotyping are also shown. Restriction enzyme sites: B, BamHI; RI, EcoRI; RV, EcoRV; H, HindIII; Ns, NsiI; Sc, SacI; St, StuI; Xh, XhoI. B, Southern blot analysis of homologous recombination in ES cells electroporated with the targeting vector. Genomic DNAs were digested with StuI or BamHI and hybridized with a 3′-probe or 5′-probe, respectively. The StuI-digested DNA analyzed with the 3′-probe detected approximately 14 and 1 kb bands for wild-type allele (+), and 3 and 1 kb bands for targeted allele (t), whereas BamHI-digested DNA analyzed with the 5′-probe produced 11- and 8-kb bands for wild-type and targeted alleles, respectively. C, PCR analysis of mouse tail DNA. Left panel: Floxed allele (fl) was distinguished from wild-type or KO (ko) allele by primer set 5neo/3neo. Floxed allele was represented by an amplified 220-bp band, whereas wild-type or KO allele produced a 540-bp band. Right panel: The KO allele was further distinguished from wild-type or floxed allele by primer set 5KO/3KO, which amplified a 220-bp band only for the KO allele. D, Southern blot analysis for deletion of the floxed T/ebp gene in various organs from 1-month-old T/ebp(fl/fl); TPO-Cre mice. Genomic DNAs (5 μg) from eight different tissues were digested with StuI, subjected to electrophoresis, and hybridized with ³²P-labeled 3′-probe. Ex, Exon.

Cre-Mediated Recombination of Floxed T/ebp Allele

To confirm that Cre-mediated recombination occurred in the floxed T/ebp allele in vivo, resulting in the deletion of exon 2 (Fig. 1A), Southern blot analysis was carried out using various tissues obtained from 1-month-old T/ebp(fl/fl); TPO-Cre mice (Fig. 1D). The result confirmed that recombination had occurred with the T/ebp-floxed allele in the thyroid gland; however, only about 30–40% of floxed allele had undergone recombination, suggesting a partial T/ebp gene disruption in the thyroid at this age. A small amount of floxed T/ebp gene recombination was unexpectedly found in the kidney (Fig. 1D). We previously reported that Cre mRNA was exclusively expressed in the thyroid gland of 1-month-old adult TPO-Cre transgenic mice and no expression was found in kidney (26). The Cre recombinase might have been temporarily and ectopically expressed in the kidney during an early stage of development. Histological analysis of T/ebp(fl/fl); TPO-Cre mouse kidney revealed no gross changes as compared with wild-type mice (data not shown). Together with the fact that kidney is not the organ in which T/ebp is naturally expressed, we concluded that this ectopic recombination would not have any influence on studies of the thyroid gland. Detailed histological analysis of T/ebp(fl/fl) thyroids exhibited no differences as compared with wild-type thyroids (see below).

Previously, in TPO-Cre transgenic mice crossed with the ROSA26 reporter line, recombination was found to begin around E14.5, at the time during thyroid organogenesis when the thyroid has completed differentiation and TPO expression begins (26). To confirm that the recombination mediated by TPO-Cre transgene did not occur at earlier gestational ages, neck regions of E12.5 T/ebp(fl/fl); TPO-Cre embryos were subjected to DNA isolation, followed by PCR analysis using recombined allele-specific primer pair P1/P3 (see Fig. 3A) as described in Materials and Methods. None of DNAs obtained from 16 embryos exhibited recombination of the T/ebp-floxed allele (data not shown).

Fig. 3. Analyses of the Recombination Frequency of the T/ebp-Floxed Allele, T/ebp Expression Levels, and Serum TSH Levels.

A, Schematic diagram of the location of primers P1, P2, and P3 (arrows) that were used for determination of recombination frequency, and primers a and b (shaded arrowheads) that were used for quantitation of T/ebp mRNA levels by real-time PCR. The thick line indicates the sequence in between two XhoI sites that is present only in the floxed allele derived from the targeting vector and differs from that of KO allele. The P2/P3 primer pair generates an amplicon (164 bp) only from the unrecombined floxed allele, but not from the recombined deleted allele because of the loss of the P2 primer binding sequence following Cre/loxP excision of exon2, nor the KO allele because P2 primer sequence does not exist. The P1/P3 primer pair generates an amplicon (165 bp) from the recombined deleted allele, but not from the unrecombined floxed or KO allele because of the large amplicon size. The diagram is not drawn to scale. Restriction enzyme sites: Xh, XhoI. B, Frequency of T/ebp-floxed gene recombination in T/ebp(fl/fl); TPO-Cre mice. The ratio of the recombined allele is shown as a percentage of the total floxed (unrecombined and recombined together) alleles, which was set at 100%. Values are the means ± SE. C, Expression levels of T/ebp mRNA in thyroid glands. All mRNA levels are normalized to 18S rRNA and presented as a relative ratio to the level in T/ebp(fl/fl) mice in each age group set to 1.0. Values are the means ± SE. D, Serum TSH levels in T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) mice. Each dot represents a TSH level for single mouse, and horizontal bars represent the mean values. Asterisks indicate mice that demonstrated histological atrophic/degenerative abnormalities in their thyroids as described in the text. NS, Statistically not significant; mo, month.

Reduced T/ebp Protein Expression in T/ebp(fl/fl); TPO-Cre Mouse Thyroid

To confirm that T/ebp protein expression was lost by TPO-Cre mediated recombination, immunostaining was carried out for detection of T/ebp in situ using 1-month-old adult thyroid glands and E16.5 developing thyroid glands collected from crosses of T/ebp(fl/fl); TPO-Cre and Tebp(fl/fl) mice. Both adult and E16.5 thyroid glands demonstrated a similar morphological appearance between the two genotypes (Fig. 2). In adult, T/ebp expression was detected in most thyrocytes of T/ebp(fl/fl) control mice (Fig. 2A). In contrast, many, but not all, thyrocytes from 1-month-old T/ebp(fl/fl); TPO-Cre mice appeared to have lost T/ebp expression when examined by immunohistochemical analysis (Fig. 2B). Eighty-five and 51% of cells were T/ebp positive when counted using two sections from three each of T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre mice, respectively, which was statistically significantly different (P < 0.05). Note that the frequency of T/ebp-positive cells in T/ebp(fl/fl) mice should be higher than 85% because all cells including nonfollicular cells such as endothelial and mesenchymal cells were included for counting due to the difficulty in distinguishing them from follicular cells. In the developing thyroid gland of E16.5, many cells were positive for T/ebp protein, and occasionally strongly positive cells were found in both T/ebp(fl/fl) control and T/ebp(fl/fl); TPO-Cre embryo thyroids whereas some cells had very weak or no expression of T/ebp (Fig. 2, C and D). Cell numbers were counted in three groups, each having strong, weak, or no expression of immunostained T/ebp using two to seven sections each from five T/ebp(fl/fl) and four T/ebp(fl/fl); TPO-Cre embryo thyroids. The average number of cells expressing strong, weak, or no T/ebp did not show any statistically significant differences between E16.5 T/ebp(fl/fl) control and T/ebp(fl/fl); TPO-Cre thyroids (strongly and weakly T/ebp-positive cells combined are 59% and 58%, respectively), suggesting that Cre-mediated recombination may be low in embryonic T/ebp(fl/fl); TPO-Cre thyroids.

Fig. 2. Partial Loss of T/ebp Protein Expression in T/ebp(fl/fl); TPO-Cre Mouse Thyroids.

Thyroids from 1-month-old or E16.5 embryos of T/ebp(fl/fl) or T/ebp(fl/fl); TPO-Cre mice were subjected to T/ebp immunostaining, TUNEL assay, or hematoxylin and eosin (H&E). T/ebp immunostaining and TUNEL assay of 1-month-old thyroids were performed on serially prepared sections. A and B, In control T/ebp(fl/fl) mice, T/ebp is highly expressed in most follicular cells of 1-month-old mice (panel A: representative is shown by arrow). In T/ebp(fl/fl); TPO-Cre mouse thyroids, the expression of T/ebp is frequently lost (panel B: representatives are shown by arrowheads). C and D, T/ebp is highly expressed (representative is shown by arrow), weakly expressed (representative is shown by arrowhead), or not expressed (representative is shown by red arrow) in both T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre E16.5 thyroids. E and F, TUNEL analysis shows no evidence of apoptotic degeneration in either 1-month-old T/ebp(fl/fl) or T/ebp(fl/fl); TPO-Cre mouse thyroids. G and H, Hematoxylin and eosin staining of 1-month-old adult thyroid glands revealed no gross histological differences between T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre mouse thyroids. Magnification, ×200. mo, Month.

We previously demonstrated that the thyroid primordium undergoes apoptotic degeneration in the T/ebp-null embryos (20). However, no cells were detected in T/ebp(fl/fl); TPO-Cre thyroids having fragmented nuclei that are characteristic of apoptotic cells. To confirm that cells that had lost T/ebp expression did not undergo apoptosis in the 1-month-old adult thyroids, terminal deoxynucleotidyl transferase biotin-deoxyuridine triphosphate nick end labeling (TUNEL) assay was performed. The results revealed no evidence of apoptotic cells in these young adult thyroids (Fig. 2, E and F). The colloid within the follicles exhibited the same quality of hematoxylin-eosin staining between T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre mouse thyroids (Fig. 2, G and H), even though many follicular cells had lost T/ebp expression in the latter mouse thyroids. These results suggest that the partial loss of T/ebp in the thyrocytes did not cause significant changes in morphology or alteration of colloid at least in the 1-month-old thyroid.

Statistical Analysis of T/ebp Gene Deletion and mRNA Expression

Male mice (10 mice per group) of the T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) genotypes, were categorized by three age groups (2–4, 6–8, and 10–12 months old) and subjected to analysis of the extent of floxed gene recombination, levels of T/ebp mRNA expression, and detailed histological analysis of the thyroid as well as serum TSH levels.

First, the recombination frequency was determined by comparing the amount of intact T/ebp-floxed allele to that of deleted allele by means of real-time PCR. To only detect the floxed allele, a unique primer was designed that is present only in the floxed allele and derived from a part of the sequence of a targeting vector, located in between XhoI and LoxP site (Figs. 1A and 3A). Approximately one-half of the floxed alleles underwent recombination in both 6- to 8- and 10-to 12-month-old groups of T/ebp(fl/fl); TPO-Cre mice (Fig. 3B). Recombination frequency was lower with the 2- to 4-month-old mouse group as compared with those of the other age groups. No recombination was detected in any DNA samples from T/ebp(fl/fl) or T/ebp(fl/ko) control group mice (data not shown). These results suggest that the T/ebp-floxed gene recombination was strictly mediated by expression of the TPO-Cre transgene.

Next, the level of T/ebp mRNA expression in the T/ebp(fl/fl); TPO-Cre thyroids was determined using quantitative RT-PCR analysis (Fig. 3C). T/ebp mRNA levels were reduced to approximately 40–60% in these mice as compared with T/ebp(fl/fl) control mice at all age groups although the reduction was less in 2-to 4-month-old group of mice than in the other two age groups. This is in agreement with the finding that the recombination frequency is approximately 50% in T/ebp(fl/fl); TPO-Cre mice as compared with T/ebp(fl/fl), and the 2- to 4-month-old group of mice have a lower recombination frequency as compared with the other mice of other ages (Fig. 3B). For comparison, T/ebp(fl/ko) mice in three different age groups were also subjected to analysis of T/ebp mRNA levels. The levels of T/ebp mRNA in these mice were about one-half of the (fl/fl) control mice for all age groups as expected. These results were in good agreement with the previous report (25). It is important to note that thyroid follicular cells of T/ebp(fl/fl); TPO-Cre mice were highly mosaic from those expressing high levels of T/ebp to those expressing no T/ebp (see Fig. 2B) whereas most T/ebp(fl/ko) thyroid follicular cells expressed T/ebp just like that seen with T/ebp(fl/fl) mouse thyroids (see Figs. 2A and 5, F and J) as judged by immunohistochemistry. Yet mice of both genotypes had approximately 50% of the T/ebp mRNA levels in the whole thyroid when examined by quantitative RT-PCR as compared with T/ebp(fl/fl) mice.

Fig. 5. Three-Dimensional Thyroid Primary Cultures in Collagen Gels.

Seven-day culture of thyroid follicles (A–D), or 3-wk culture of individual thyroid follicular cells (E–J) obtained from T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) mouse thyroids. Paraffin sections were subjected to histological analysis using hematoxylin and eosin or T/ebp immunohistochemistry. A and B and C and D are serially prepared sections. T/ebp(fl/fl) mouse follicles maintained round shape with colloid in the lumen (A) that expressed T/ebp (B). T/ebp(fl/fl); TPO-Cre follicular cultures consisted of a mixture of follicles that were different in size and shape from round (C, representative is shown by red arrowhead) to oval to irregular (C, representatives are shown by black arrowheads), and the levels of colloid in the lumen from normal to none (C, the latter representative is shown by arrow). Some follicles exhibited T/ebp expression (D, arrowhead) whereas others had no T/ebp expression, particularly in the area where most follicles were obliterated, exhibiting solid growth pattern (C and D, shown by thin arrow). Single follicular cell grew to a cluster of spherical follicles in T/ebp(fl/fl) with a plenty of colloid in the lumen (E, representative is shown by arrow) whereas T/ebp(fl/fl); TPO-Cre and T/ebp(fl/ko) follicular cell produced irregular shaped polygonal follicles with no colloid accumulation (G and I, representative is shown by arrow). Most follicular cells from T/ebp(fl/fl) and T/ebp(fl/ko) mice were positive for T/ebp expression (F and J) whereas those of T/ebp(fl/fl); TPO-Cre mice exhibited various degrees of T/ebp expression (H). Experiments were separately carried out at least three times for each culture condition and genotype, and each time the same results were obtained. Magnification: A–D, ×100, E–J, ×200.

Thyroid Function and Histological Alterations Found in T/ebp(fl/fl); TPO-Cre Mice

Thyroid function was analyzed in all 90 mice by measuring serum TSH levels at the time when they were subjected to necropsy (Fig. 3D). Serum TSH level is considered to be a critical indicator of the severity of dysfunction for the pituitary-thyroid axis (6). The mean TSH levels were higher in T/ebp(fl/ko) mice than in T/ebp(fl/fl) mice in all age groups as previously reported (25) although no statistical significance was obtained in this study. This could be due to the mixed genetic backgrounds of mice used in this study as compared with the previous study using 129Sv mice. It was reported that serum TSH concentrations differ among different strains of mice (27). Further, the mean TSH levels were much higher in T/ebp(fl/fl); TPO-Cre mice as compared with those of T/ebp(fl/fl) or T/ebp(fl/ko) mice, although no statistical significance was obtained. The Cre-mediated deletion assay clearly demonstrated, as shown in Fig. 3B, that a partial deletion of the floxed-T/ebp gene occurred in the thyroids of all T/ebp(fl/fl); TPO-Cre mice. Altogether, these results suggest that thyroid function was maintained almost normally for as long as 12 months in T/ebp(fl/fl); TPO-Cre mice with a trend of slight hypothyroidism that was likely to be caused by partial disruption of T/ebp gene. Interestingly, however, extremely high serum TSH levels were sporadically observed only among T/ebp(fl/fl); TPO-Cre mice (indicated by asterisks in Fig. 3D), but never with other genotypes of mice.

Extensive histological analysis, carried out using serial sections of whole thyroids from all mice, demonstrated no histological differences between T/ebp(fl/fl) and wild-type thyroids. These analyses further revealed that among T/ebp(fl/fl); TPO-Cre mice, those that exhibited highly elevated levels of serum TSH commonly presented severely affected thyroid glands. The size of the thyroid glands was markedly smaller as compared with normal control thyroid glands from either wild-type or T/ebp(fl/fl) mice, with approximately half the diameter as estimated at the maximal coronary section (Fig. 4, panel B vs. panel A). The glands consisted mostly of atrophic and/or degenerative follicles, and multiple smaller follicles that were lined by cuboidal to columnar epithelial cells in various ratios. Normal follicles were found in only a small portion of the gland (Fig. 4D). Atrophic/degenerative follicles varied in size and shape and were lined by flattened to cuboidal epithelial cells (Fig. 4E). In the lumen of these cells, colloid was completely or partially depleted and was sometimes accompanied by desquamated epithelium and/or cell debris and partial disappearance of follicular cell linings. T/ebp immunostaining revealed that a majority of cells lining atrophic/degenerative follicles frequently had lost T/ebp expression (Fig. 4G), suggesting that the loss of T/ebp may be the cause of the atrophic/degenerative lesions. These atrophic/degenerative follicular cells did not exhibit any apoptotic appearance with fragmented nuclei. To confirm that the loss of T/ebp expression did not lead to apoptosis, TUNEL analysis was performed (data not shown). Similar to the thyroid of young adulthood, no apoptotic cells were found. This suggests that the loss of T/ebp expression did not cause apoptotic degeneration at least as judged by TUNEL assays. Small follicles that filled the space between normal and atrophic/degenerative follicles resembled hyperplastic thyroid follicles containing no or little colloid in the lumen (Fig. 4D), which occasionally formed nodular lesion that was considered to be a follicular adenoma (Fig. 4F). Surprisingly, immunostaining for T/ebp demonstrated that T/ebp was highly expressed in these small follicles (Fig. 4G). These results suggest that the appearance of hyperplasic lesions and/or follicular adenoma was not a direct consequence of partial T/ebp gene disruption, but rather a secondary event that occurred in intact follicles due to prolonged exposure to high serum TSH levels caused by decreased levels of T/ebp expression in atrophic/degenerative follicles. The mean value for the T/ebp mRNA levels of the thyroid from T/ebp(fl/fl); TPO-Cre mice having both atrophic/degenerative and hyperplastic/adenomatous thyroid follicles did not significantly differ from that of same genotype mice without these histological abnormalities when determined by real-time PCR (relative T/ebp mRNA level is 1 vs. 1.13, respectively). This is because T/ebp mRNA levels were measured using RNAs isolated from a mixture of tissues having atrophic/degenerative lesions that did not express T/ebp and hyperplastic/adenomatous lesions that expressed T/ebp. Thus, the mean T/ebp mRNA levels of the thyroids as a whole may not be directly correlated to lesions.

Fig. 4. Histological Abnormalities of T/ebp(fl/fl); TPO-Cre Conditional Hypomorphic Mouse Thyroids.

Hematoxylin and eosin staining of transverse sections of thyroids from 10-month-old T/ebp(fl/fl) control (A) and T/ebp(fl/fl); TPO-Cre conditional hypomorphic mice (B and C). Tr, Trachea. B, Severely impaired thyroid gland commonly seen among mice that have highly elevated serum TSH levels. Size of the gland is significantly smaller as compared with that of control thyroid gland shown in panel A. C, Cystic phenotype of the thyroid gland, which is frequently observed in T/ebp(fl/fl); TPO-Cre and T/ebp(fl/ko) mouse lines, particularly as they age. Dilated follicles preferentially reside near the periphery of the lobe. D, Higher magnification of severely affected thyroid gland common to mice that have highly elevated serum TSH levels. The lobe consists of intact follicles (representatives are shown by red asterisks), degenerative follicles with complete or partial colloid depletion (representatives are shown by black asterisks), and multiple small follicles (arrows) that intervene between intact and degenerative follicles. E, High-power view of degenerative follicle. Desquamative thyrocyte (arrowhead) and cell debris (arrow) in depleted colloid are seen in the follicular lumen. Follicular cell lining is partly disappeared (representatives are shown by asterisks). F, Follicular adenoma (lower left part of the panel), which is frequently seen in mice with high TSH levels, is well demarcated from adjacent degenerative follicles (upper right). G–I, Immunostaining for T/ebp. Follicular cells in adenoma highly express T/ebp (G, representatives are shown by black asterisks), whereas the follicles in which lining thyrocytes have lost T/ebp expression become degenerative (G, representatives are shown by red asterisks), and sometimes contain cell debris and desquamative cells in the lumen (H, arrows). I, Almost all follicular cells in dilated follicles as seen in panel C including those from T/ebp(fl/fl); TPO-Cre mice clearly express T/ebp. A part of the follicles is enlarged (inset). Magnification: A–C, ×10; D, ×100; F, G, and I, ×200; E and H, ×400.

Histological examination carried out on thyroids from all other mice revealed that the most prominent difference in thyroid morphology found between T/ebp(fl/fl) control group and T/ebp(fl/fl); TPO-Cre mice maintaining apparent normal thyroid function with a trend of slight hypothyroidism, as judged from the serum TSH levels, was the presence of a disordered architecture of the thyroid follicles (Fig. 4C), particularly as mice aged. In this lesion, the number of follicles was reduced to various extents, and some of the remaining follicles were extraordinarily dilated, resulting in a severely disorganized appearance of the thyroid gland. Degenerative follicles and microfollicular structures as seen in Fig. 4D coexisted in some of these T/ebp(fl/fl); TPO-Cre thyroids although not conspicuous. In the dilated thyroids, almost all cells lining dilated follicles were positive for T/ebp immunostaining with colloid in the lumen (Fig. 4, I and C). Of particular interest, the dilated disorganized thyroid structure also appeared in T/ebp(fl/ko) mice as they aged even though they did not carry the Cre transgene. In the thyroids of these mice, unlike T/ebp(fl/fl); TPO-Cre mice, degenerative follicles and microfollicular structures were not found, suggesting that the appearance of the degenerative follicles and/or microfollicular structures may be due to partial T/ebp ablation caused by Cre expression. These results further suggest that T/ebp may play a role in maintaining the ordered differentiated architecture of the thyroid.

T/ebp Is Required for Normal Folliculogenesis in Vitro

Primary isolated thyrocytes can reconstruct thyroid follicular structure in a three-dimensional collagen gel culture system (28). To examine the effect of insufficient expression of T/ebp on folliculogenesis, follicles or individual thyrocytes isolated fromT/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, or T/ebp(fl/ko) mice were embedded in a three-dimensional collagen gel. When a whole follicle was cultured for 7 d, cells already exhibited drastically different patterns of growth between T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre mice (Fig. 5, A and B vs. C and D). All cultured follicles from T/ebp(fl/fl) mice had spherical shapes of similar size that contained colloid in the lumen (Fig. 5A), most of which expressed T/ebp (Fig. 5B). In contrast, the majority of T/ebp(fl/fl); TPO-Cre mouse follicles exhibited irregular-shaped structures of different size, were obliterated to various extents, and had no or less colloid accumulation in the lumen as compared with T/ebp(fl/fl), although a few spherical-shaped follicles containing colloid were observed (Fig. 5C). These spherical or oval-shaped follicles maintained T/ebp expression similar to that seen with T/ebp(fl/fl) whereas no T/ebp expression was detected in the area showing a solid growth pattern, where follicular lumens were completely obliterated (Fig. 5D). To eliminate a possible contribution of contaminated mesenchymal cells surrounding the follicles to development of different shaped follicles, individual follicular cells from T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) thyroids were cultured in collagen gels. After 3 wk, cells from all genotypes formed three-dimensional structures, having a cavity-enclosed monolayer of cells that expressed T/ebp although some variations in expression levels were found (Fig. 5, F, H, and J). This was particularly obvious in T/ebp(fl/fl); TPO-Cre mouse thyroids (Fig. 5H). Interestingly, cells from T/ebp(fl/fl) mice reconstructed mature, spherical follicles, containing colloid material in the lumens (Fig. 5E), whereas cells from T/ebp(fl/fl); TPO-Cre and T/ebp(fl/ko) mice reconstructed irregular-shaped, polygonal structures lacking colloid accumulation in the lumens (Fig. 5, G and I). These findings suggest that insufficient T/ebp expression in thyrocytes affects both structure and function of the thyroid follicle.

To confirm that T/ebp expression is critical for the structure and function of the thyroid, individual thyrocytes isolated from T/ebp(fl/fl) mice were first treated with recombinant adenovirus expressing Cre recombinase or containing vector only as control, and were subjected to three-dimensional collagen culture (Fig. 6). Most thyrocytes treated with control adenovirus formed round follicles with colloid in the lumen that expressed T/ebp (Fig. 6, A and B), whereas thyrocytes treated with Cre-expressing recombinant adenovirus did not form any follicle-like structures (Fig. 6, C and D). In these cells, relatively strong T/ebp expression was observed in only a few cells, and a majority of cells expressed very weak or no T/ebp (Fig. 6D), as determined by immunohistochemical analysis. Genomic PCR demonstrated that only cells treated with Cre-expressing adenovirus had a band derived from a recombined T/ebp allele, but not no-virus control or cells treated with control virus (Fig. 6E). Average recombination frequency from three T/ebp(fl/fl) mouse thyroids that were treated with Cre-expressing adenovirus was 54.3 ± 13.9%, as determined by real-time PCR, suggesting that the situation of T/ebp gene partial ablation in cells treated with Cre-expressing adenovirus resembles that of T/ebp(fl/fl); TPO-Cre mouse thyroids. These results clearly demonstrated that T/ebp is required for normal folliculogenesis of the thyroid at least in vitro.

Fig. 6. Effect of the Loss of T/ebp on Structure of T/ebp(fl/fl) Thyroid in Three-Dimensional Thyroid Primary Cultures.

Individual thyrocytes isolated from T/ebp(fl/fl) thyroids were treated with recombinant adenovirus containing vector only as control (A and B) or those expressing Cre recombinase (C and D), followed by three-dimensional culture in collagen. Two weeks later, the cultures were subjected to histological analysis with hematoxylin and eosin (A and C) or T/ebp immunohistochemistry (B and D). Round follicles started appearing in control cultures with a weak pink color, indicative of colloid in the lumen (A) and strong T/ebp expression in most of cells (B, representatives are shown by arrow) whereas no follicular structures were obtained from Cre-expressing adenovirus-treated thyroid cells (C), most of which had weak (D, representatives are shown by arrowhead) or no expression of T/ebp (D, representatives are shown by red arrowhead), and only a few cells showed strong positive staining (D, shown by arrow). Experiments were carried out using thyroids from three independent T/ebp(fl/fl) mice, and the same results were obtained for all three. Magnification, ×200. E, Genomic PCR analysis of recombined (164 bp) and unrecombined T/ebp allele (165 bp) using DNAs isolated from control cells without any virus treatment (No), cells treated with control virus (Cont), and cells treated with adenovirus expressing Cre recombinase (Cre). Only the latter DNA produced a recombined band. exp, Experiment; H&E, hematoxylin and eosin.

DISCUSSION

We have established a T/ebp conditional gene KO mouse line using the Cre-loxP system and were able to successfully lower T/ebp gene and protein expressions in the thyroid gland of adult mice, resulting in the production of T/ebp-thyroid-conditional hypomorphic mice. T/ebp is involved in both the morphogenesis and differentiation of the thyroid gland (20, 29). Our previous study using the T/ebp-null mouse demonstrated that the thyroid rudiment was present but failed to develop and was eliminated by apoptosis starting around E10.5 (20, 23). Because of the elimination of the thyroid rudiment during early embryogenesis and stillbirth, the T/ebp-null mouse is not of value for studying the role of T/ebp in the thyroid during the late stages of development and/or in postnatal life. Conditional KO mice that disrupt the T/ebp gene specifically in the thyroid with the use of the Cre-loxP system would provide a means to understand the role of T/ebp in the maintenance of the thyroid after the completion of differentiation.

In T/ebp(fl/fl); TPO-Cre thyroid-conditional hypomorphic mice, a partial loss of T/ebp expression was confirmed in the 1-month-old thyroids as judged by immunohistochemistry and Southern blotting. Cells having no or partial disruption of the T/ebp gene and those that appeared to have lost T/ebp expression were found to coexist within a follicle. It is important to note, however, that due to the nature of immunohistochemical technique, “no immunostaining” does not necessarily mean null expression. It is known that each follicle is individually regulated and does not express the same level of T/ebp and thyroglobulin within a thyroid (30). In fact, in our previous study, we found that not all follicles express TPO (26). The presence of follicular cells with various levels of T/ebp expression within a follicle may suggest the possibility of heterogeneity at the follicular cell level as well, which might partly explain apparent incomplete recombination of the floxed T/ebp allele in T/ebp(fl/fl); TPO-Cre mouse thyroids. The fact that TPO expression is positively regulated by T/ebp itself (31, 32) may also contribute to incomplete deletion of the T/ebp gene.

Because no histological and functional abnormalities are found in young T/ebp(fl/fl); TPO-Cre mice thyroids, the partial loss of T/ebp expression does not seem to immediately affect the structure and/or function of the differentiated thyroid. Pax8 and Foxe1 (Ttf2) are critical thyroid-specific transcription factors that have important roles in thyroid organogenesis. These transcription factors, together with T/ebp, regulate the expression of thyrocyte differentiation marker genes such as thyroglobulin (33), TPO (31, 32), TSH receptor (34, 35), and sodium iodide symporter (36). Thyroid functions are maintained through cooperative events involving these transcription factors (7). The cooperativity in the regulation of specific target genes may explain why loss of T/ebp expression does not, at least immediately, affect thyroid function and structure.

Hyperplastic lesions and frequent adenomas in the thyroids, accompanied by extremely high serum TSH levels, were found albeit very low incidence, only among T/ebp(fl/fl); TPO-Cre mice, in which T/ebp expression was partially ablated due to the TPO-Cre transgene expression. However, these thyroid lesions have never previously been found in T/ebp(fl/fl), T/ebp(fl/ko), or original T/ebp heterozygous mice in mixed, C57BL/6, 129Sv, or Black Swiss genetic background, suggesting that this is likely due to the expression of Cre that partially deletes T/ebp gene. Cre expression leads to apparent ablation of T/ebp in many cells within a follicle in as early as 1-month-old adult thyroid, which may result in atrophic/degenerative lesions and cell death, leading to the production of high serum TSH levels. The high TSH levels, in turn, affect the remaining thyroid follicles retaining T/ebp expression, which may become hyperplastic and adenomatous. The hyperplastic and adenomatous lesions may have never developed without the initial presence of atrophic/degenerative lesions that was caused by the ablation of T/ebp. On the other hand, cells having only one T/ebp allele ablated may exist, and an additional ablation of T/ebp may happen only sporadically because of insufficient T/ebp activity, which may result in low expression of the TPO-Cre transgene. This results in a phenotype similar to that observed in the T/ebp(fl/ko) mouse, and thus T/ebp(fl/fl); TPO-Cre mice thyroids do not immediately exhibit defective thyroid function. However, thyroids of both genotype mice eventually become severely altered in structure as they become older, characterized by a significant decrease in the number of follicles and cystic dilatation of the remaining follicles with a disorganized appearance, consisting of excessive numbers of follicular cells as compared with those usually found in the normal thyroid. This could be simply a compensatory reaction of thyroid to its prolonged exposure to slightly higher levels of TSH, due to haploinsufficiency in the case of T/ebp(fl/ko) mice and incomplete deletion of T/ebp in the case of T/ebp(fl/fl); TPO-Cre mice. This may be the reason for the rather mild phenotypes of hypothyroidism found in the latter mice. However, the size of follicles and the number of follicular cells contained in a follicle, as we observed in this study, are quite unusual. It is generally considered that compensation for thyroid dysfunction is mostly achieved by an increase in number of follicles, and not by the enlargement of each follicle (6). This suggests that the thyroid is unable to make new follicles, and each follicle cannot regulate its normal size under partial loss of T/ebp function. It is also interesting that the somatic stages of partial mosaic ablation of T/ebp in the postdifferentiated thyroids appear to recapitulate the germline haploinsufficiency phenotype. Altogether, these results suggest that two intact T/ebp alleles may be required for the maintenance of normal architecture and function of the thyroid.

The exact mechanism by which the follicle becomes either atrophic/degenerative or dilated is not known. However, our results clearly demonstrate that 1) loss of T/ebp expression has occurred in a significant number of follicular cells in 1-month-old T/ebp(fl/fl); TPO-Cre mouse thyroids; 2) hyperplastic and adenomatous lesions were found only among T/ebp(fl/fl) mice carrying Cre transgene that partially deletes T/ebp; 3) majority of T/ebp(fl/fl); TPO-Cre mouse thyroids become dilated where most of follicular cells express T/ebp; and 4) dilated follicles are also found in aged T/ebp(fl/ko) mice. These apparent contrary findings might be explained using the following hypothesis. In T/ebp(fl/fl); TPO-Cre mouse thyroids, complete deletion of T/ebp gene occurs little by little, and cells that have lost T/ebp expression would not survive, resulting in fewer follicular cells and/or follicles remaining in these mice thyroids. In the meantime, a stem/progenitor-like population of cells may regenerate new follicular cells/follicles in the T/ebp-disrupted thyroid gland. Damage and repair of such thyroid glands are two events that collaterally proceed at a delicate balance. The loss of this balance could lead to a severely affected, atrophic/degenerative thyroid that triggers extraordinary high serum TSH level, which, in turn, causes the rest of T/ebp-expressing cells to become hyperplastic and adenomatous. The majority of follicular cells in T/ebp(fl/fl); TPO-Cre mouse thyroids, however, remain within this delicate balance, gradually resembling the situation observed in T/ebp(fl/ko) mice thyroids. Lower T/ebp activity found in the thyroids of T/ebp(fl/fl); TPO-Cre and T/ebp(fl/ko) mice may not be sufficient to maintain proper folliculogenesis, leading to dilated disorganized follicles. In support of this hypothesis, ongoing chemical carcinogenesis studies using T/ebp(fl/fl) and T/ebp(fl/fl); TPO-Cre mice revealed an increased incidence of thyroid adenoma in the latter mice (Hoshi, S., and S. Kimura, unpublished observation). Further support of the hypothesis was obtained from in vitro three-dimensional thyroid primary culture studies using thyroids from T/ebp(fl/fl); TPO-Cre, T/ebp(fl/ko), and T/ebp(fl/fl) mice, and the latter treated with recombinant adenovirus with and without Cre expression. The results demonstrated that only cells from T/ebp(fl/fl) mouse thyroids and those treated with control adenovirus, but not Cre-expressing adenovirus, formed follicular structures. Further studies are required to understand the exact mechanisms to explain the current observed phenotypes and how T/ebp plays a role in the process of developing the phenotypes described herein and in folliculogenesis during postnatal life.

We previously demonstrated that thyroid primordial cells undergo apoptotic degeneration by E12–13 in the absence of T/ebp (23). The current study demonstrates that the lack of T/ebp expression does not seem to lead to apoptotic degeneration of cells in adult thyroids. The discrepancy found between the previous and the current studies may simply be due to the sensitivity of the TUNEL method used. Alternatively, the mechanisms of cell death may be different between early developmental thyroid precursor cells and those that have completed differentiation. Recently, alternative nonapoptotic mechanisms of programmed cell death, autophagy, and/or paraptosis have been proposed. Their definition has not been established and whether or not they represent an identical pathway and/or are caspase dependent has yet to be determined (37–40). Nevertheless, both lack the characteristic features of apoptosis, including nuclear fragmentation, apoptotic body formation, and chromatin condensation (37–40). Nonapoptotic programmed cell death might be a mechanism that takes place in thyroid follicles that have completed differentiation. Whether this is the case and, if so, when and why a nonapoptotic pathway dominates over apoptosis awaits further study.

In conclusion, we have established, for the first time, a conditional T/ebp-KO mouse and, in combination with thyroid-specific Cre-transgene, a T/ebp-thyroid-conditional hypomorphic mouse. By carrying out detailed analysis of thyroids from the T/ebp-conditional hypomorphic and T/ebp-heterozygous mice, and in vitro three-dimensional thyroid primary culture studies, we demonstrated that T/ebp is required for the maintenance of normal architecture and function of differentiated thyroids. Thus, the T/ebp-thyroid-conditional hypomorphic mouse may provide a good model for studying a possible role for T/ebp in relation to thyroid diseases because gradual and/or sporadic, rather than total complete, ablation of the T/ebp gene could resemble the natural incidence of loss of gene expression should T/ebp play a role in the pathogenesis of thyroid diseases.

MATERIALS AND METHODS

Targeting Vector and Generation of T/ebp(fl/fl) Mouse

A bacterial artificial chromosome genomic library derived from mouse 129Sv genomic DNAs (Incyte Genome Systems, St. Louis, MO) was screened with a mouse T/ebp cDNA (16). An EcoRI fragment (~6 kb) and an EcoRI-EcoRV fragment (~5.4 kb) were subcloned into the pET17b vector (Novagen EMD Biosciences, San Diego, CA); the EcoRI fragment contained a sequence covering 3.5 kb upstream of the T/ebp gene, the gene’s coding and a part of the 3′-noncoding regions whereas the EcoRI-EcoRV fragment contained a sequence downstream of the EcoRI site that is present in the 3′-noncoding region of the gene (Fig. 1A). The first loxP site was inserted at the NsiI site of the 6-kb EcoRI fragment, located within intron 1 of the T/ebp gene, assuming that an exon containing a start codon ATG is the first exon (41, 42). A BamHI site was incorporated along with the first loxP site in such way that, upon recombination, this BamHI site is lost. The second and third loxP sites flanking the pGK-Neo cassette that are in the same orientation as that of the first loxP site were inserted in between two XhoI sites within the EcoRI-EcoRV fragment, located approximately 3 and 3.5 kb downstream of the EcoRI site. After loxP sites were individually introduced into EcoRI-EcoRI and EcoRI-EcoRV fragments, they were ligated to each other to produce a targeting vector that contained approximately 4.2 kb of homologous DNA upstream of the first loxP site and 1.9 kb of homologous DNA downstream of the loxP-pGK-Neo cassette.

The ES cells (RW4, Incyte Genome Systems) were propagated and electroporated with the linearized targeting vector DNA. G418-resistant ES clones were selected, expanded, and analyzed by Southern blotting using a 5′- and 3′-probe for detection of homologous recombinants (Fig. 1B). The correctly targeted ES cells were injected into C57BL/6 blastocysts to generate chimeric founder mice as previously described (43, 44). Chimeric founder male mice were bred with C57BL/6 females. The germline-transmitted F1 mice were crossed with EIIa-Cre mice to produce a floxed allele by deleting the pGK-Neo cassette in vivo (45). Neomycin gene deletion was confirmed by both PCR and Southern blotting hybridization with the neo probe. These mice (+/fl) were further crossed with the original T/ebp heterozygous mice (+/ko) (20) and with TPO-Cre transgenic mice (26) to eventually produce T/ebp(fl/fl), T/ebp(fl/fl); TPO-Cre, and T/ebp(fl/ko) mice.

Genotyping

ES cells or isolated mouse tail DNAs were subjected to Southern blot hybridization and/or PCR analysis. Southern blot hybridization was carried out using StuI or BamHI digested genomic DNAs and 3′- or 5′-external probes, respectively (Fig. 1, A and B). The probes were labeled with [α-³²P]dCTP (PerkinElmer Life Sciences, Inc., Boston, MA) using a commercially available kit (Amersham Biosciences, Piscataway, NJ). Hybridization was carried out using Perfecthyb (Sigma, St. Louis, MO) according to the manufacturer’s instruction. Signals were obtained using Storm PhosphorImager and ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA).

PCR genotyping was used for distinguishing wild-type or the original null (KO) allele from the floxed allele using primer pairs with the following sequences: (5neo) 5′-TGCCGTGTA-AACACGAGGAC-3′ and (3neo) 5′-GACTCTCAAGCAAGTCCATCC-3′ (Fig. 1A). Fragments produced were 540 and 220 bp for the wild-type or KO allele, and floxed allele, respectively (Fig. 1C, left panel). The targeted allele with the neo cassette was not amplified due to its large size. The KO allele was further distinguished from the wild-type or floxed allele by using the following primer pairs: (5KO) 5′-TCGCCTTCTATCGCCTTCTTGACGAG-3′ and (3KO) 5′-TCTTGTAGCGGTGGTTCTGGA-3′, where the forward primer is specific to the neo cassette gene inserted at the XhoI site of exon 2 (Fig. 1A) (20). The primer pairs amplified a 220-bp fragment for the KO allele whereas no band was produced for the wild-type or the floxed allele (Fig. 1C, right panel). Both PCRs were carried out at 94 C for 5 min, followed by 94 C for 15 sec, 60 C for 15 sec, and 72 C for 30 sec for 35 cycles, and 72 C for 5 min. Detection of Cre transgene was as previously described (26).

Animal Studies

All animal studies were carried out in accordance with the Using Animals in Intramural Research Guidelines (NIH Animal Research Advisory Committee, NIH, Bethesda, MD) after approval by the NCI Animal Care and Use Committee. For statistical analysis, a total of 90 male mice were categorized by age and genotype. Blood was obtained and centrifuged in serum separator tube (Becton Dickinson, Franklin Lakes, NJ) to obtain serum for TSH measurements. The left lobe of the thyroid gland was fixed in 10% buffered formalin immediately after removal and embedded in paraffin for histological analysis, whereas the right lobe was used for RNA and DNA isolation.

Culture of Primary Mouse Thyroid

Thyroid lobes were aseptically dissected from 1-month-old mice. The lobes were collected in a 1.5-ml microcentrifuge tube containing 1 ml of digestion medium, which consisted of 100 U/ml of type I collagenase (Sigma) and 1 U/ml of dispase (Roche, Indianapolis, IN) dissolved in Ham’s F12/DMEM. Enzymatic digestion was carried out for 30 min in 37 C water bath, with shaking at 150 strokes per min. After digestion, isolated individual follicles were washed three times with culture medium [Ham’s F12/DMEM containing 40% Nu-Serum IV (Collaborative Biomedical, Bedford, MA)] and embedded in a collagen gel as described below. For the culture of individual thyroid follicular cells, isolated follicles were seeded in a 35-mm culture dish and grown to confluence in monolayer for 3–4 d. They were then completely dissociated to single cells by Trypsin (0.25%)-EDTA (0.02%). For preparation of collagen solution, rat tail collagen (Roche) was dissolved in sterile 0.2% acetic acid (vol/vol), pH 3.0, at a final concentration of 3 mg/ml. Eight parts of collagen solution were mixed on ice with one part of 10× concentrated DMEM and one part of reconstitution buffer (32.2 g NaHCO₃ ± 4.77 g HEPES in 100 ml of 0.05 N NaOH), and 0.5 ml of this reconstituted collagen solution was placed on a 12-well culture dish, which was immediately warmed at 37 C for gel formation. This acellular basal layer was overlaid with cell suspension in 1 ml of reconstituted collagen solution. After the cellular layer was solidified, the gel was further covered with culture medium containing antibiotic-antimycotic (Invitrogen). The medium was changed every other day until the three-dimensional structure was formed. At that point, the entire collagen gel was removed from the dish and fixed in 4% paraformaldehyde overnight at 4 C. Fixed collagen gel was embedded in paraffin and then sliced at 5 μm in thickness for histological analysis.

For infection of primary cultured thyroids with recombinant adenovirus expressing Cre recombinase or vector only as control, digested follicles were seeded on a dish, cultured for 2 d, and then treated with recombinant adenovirus at multiplicity of infection of 50 for 3 d, followed by embedding in a collagen gel.

Real-Time PCR

Quantitative RT-PCR analysis was performed with an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA) to determine the level of T/ebp mRNA expression and the extent of T/ebp(fl/fl) gene recombination. For mRNA analysis, total RNAs were isolated from one fourth of the thyroid gland of each mouse using Trizol Reagent (Invitrogen). Isolated RNAs were treated with DNase I by using DNA-free (Ambion, Inc., Austin, TX) to eliminate contaminating genomic DNAs from the prepared RNA samples. Reverse transcription was carried out by using random hexamers and Superscript II reverse transcriptase (Invitrogen). The Taqman assay was adopted for quantitative analysis using a 6-FAM-TAMRA-labeled T/ebp probe. The primers and probe used were as follows: primer a (5′-GCGCCGGGTGCTCTTC-3′) and primer b (5′-CGTGGGTGTCAGGTGAATCA-3′), and (6-FAM)5′-CCGGAGCGCGAGCATCTGG-3′(TAMRA) as a probe. To discriminate transcripts that are derived from the original T/ebp-KO allele, primers were designed that flank the XhoI site in the second exon, where the 1.1-kb neo cassette was inserted for T/ebp gene disruption (see Fig. 3A) (20, 25). The cycle conditions were 50 C for 2 min, 95 C for 10 min, followed by 40 cycles of 95 C for 15 sec, and 60 C for 40 sec. The data were analyzed by the standard curve method and normalized for 18S rRNA measured by using Taqman Ribosomal RNA Control Reagent, VIC Probe (Applied Biosystems). A standard curve was obtained on each 96-well reaction plate using a serially diluted amplicon for T/ebp or 18S rRNA as a standard that was obtained using the primer pair used for actual real-time PCR analysis. Only curves obtained having a high correlation coefficient (r² > 0.99) were used.

To analyze recombination frequency for the T/ebp-floxed allele, two sets of primers were designed to distinguish the unrecombined (no T/ebp gene deleted) floxed allele from the recombined (T/ebp gene deleted) floxed allele. The unrecombined floxed allele was detected using primers P2 (5′-GCCAGTACTAGTGAACCTCTTCGAG-3′) and P3 (5′-GACTCTCAAGCAAGTCCATCC-3′), which amplified a 165-bp fragment (Fig. 3A). This primer pair did not amplify any fragment from the recombined deleted allele or the original KO allele because the P2 sequence is derived from the area in between the XhoI and loxP site that was created during construction of the targeting vector and is found only in the unrecombined Tebp-floxed allele (Fig. 3A). Primers P1 (5′-GAGCCGCCCTGCTGGGAT-3′) and P3 generated an amplicon (164 bp) only from the recombined deleted allele, but not from unrecombined floxed allele or original KO allele because of its large size (Fig. 3A). Reactions with primer pairs P1/P3 and P2/P3 were performed in separate wells of the same 96-well reaction plate using SYBR Green master mix (Applied Biosystems) and the following conditions: 95 C for 5 min, followed by 40 cycles of 95 C for 15 sec, 64 C for 15 sec, and 72 C for 30 sec. Data are presented as a percentage of the total (recombined plus unrecombined) floxed alleles being set to 100%. Statistical analysis was carried out using the non-parametric Mann-Whitney test for the analysis of T/ebp mRNA levels and recombination frequency for T/ebp-floxed allele.

TSH Measurements

Serum TSH concentrations were measured by RIA. Mouse serum or mouse serum TSH (msTSH) (50 μl) standard serially diluted with TSH-deficient mouse serum (a kind gift from Dr. Sheue-Yann Cheng, NCI, Bethesda, MD) were incubated in a 5ml polypropylene tube with 100 μl of 1:600 diluted guinea pig anti-msTSH antibody (Antibodies, Inc., Davis, CA) in RIA buffer (150 mM boric acid, 67.5 mM NaOH, 0.5% BSA, 0.02% NaN₃) for 18 h at room temperature. Nonspecific binding was determined by incubating TSH-deficient mouse serum with RIA buffer instead of guinea pig anti-msTSH antibody (Antibodies Inc.). [¹²⁵I]TSH (ICN Radiochemicals, Costa Mesa, CA) (100 μl containing 20,000 cpm) was added 18 h later. Tubes were vortexed thoroughly and incubated for 4 h at room temperature, followed by the addition of 200 μl goat antiguinea pig IgG diluted at 1:8 in RIA buffer as a secondary antibody and 100 μl guinea pig serum diluted at 1:50 in RIA buffer as carrier. The tubes were vortexed again and incubated for additional 2 h at room temperature. After incubation, the reactions were stopped by adding 1 ml PBS and were centrifuged at 3000 rpm for 30 min. The supernatant was removed by aspiration, and pellets were washed with PBS. The tubes were counted for 2 min with a γ-scintillation counter and data analyzed as previously described (27) using nonparametric Mann-Whitney test statistical analysis.

Immunohistochemistry

Deparaffinized sections (5 μm) were incubated in a solution of 0.3% H₂O₂ (vol/vol) in methanol for 30 min to inactivate endogenous peroxidases, followed by rinsing three times for 10 min each with PBS. Sections were subjected to antigen retrieval in citric acid buffer (pH 6.0) at 95 C for 30 min or by microwave for 15 min. Tissues were blocked in 10% goat serum (Vector Laboratories, Burlingame, CA) for 30 min at room temperature, and then incubated overnight with rabbit anti-TTF-1 (T/EBP) antibody (1:1000 dilution; Biopat, Caserta, Italy) at 4 C in a humidified chamber. After washing three times for 20 min in PBS, the tissues were processed by the avidin biotinylated enzyme complex method using a commercially available kit (Vector Laboratories) according to the manufacturer’s instruction. Immunocomplexes were visualized with 3,3′-diaminobenzidine tetrahydrochloride (DAKO Corp., Carpinteria, CA).

Abbreviations

E9.5

Embryonic d 9.5

ES cell

embryonic stem cell

KO

knockout

msTSH

mouse serum TSH

T/ebp

thyroid-specific enhancer-binding protein

TPO

thyroid peroxidase

TUNEL

terminal deoxynucleotidyl transferase biotin-deoxyuridine triphosphate nick end labeling