An attempt to isolate cells having stem/progenitor characteristics was first carried out using side population (SP) cells of mouse thyroid (34). SP cells were originally identified in the adult mouse bone marrow as a small subset of cells able to efflux the vital dye Hoechst 33342 in flow cytometric analysis (35). It was later demonstrated that this was based on the expression of the ATP binding cassette (ABC) family of transporter proteins in stem/progenitor cells, which are able to pump out the dye, thus rendering the dye efflux of SP cells sensitive to inhibitors such as verapamil and fumitremorgin (36, 37). Various non-hematopoietic adult tissues were subsequently demonstrated to possess SP cells, including the liver, skeletal muscle, lung, kidney, and mammary gland (38–45). SP cells isolated from adult muscle or liver contributed to tissue regeneration (40, 44), demonstrating that SP cells contain cells characteristics of stem cells.

Mouse thyroid SP cells were obtained, ranging from 0.3 to 1.4% of the total population of cells. They were composed of two groups of cells characterized as CD45(−)/c-kit (−)/SCA1(+) and CD45(−)/c-kit(−)/SCA1(−), each representing approximately a half of SP cells (34). SCA1 is known to be expressed in SP cells and/or enriched progenitor cell population from various tissues such as bone marrow, mammary gland, and muscle (35, 42, 44, 45). Thyroid SP cells expressed stem cell marker genes such as nucleostemin and Oct4 in addition to Abcg2 through which the SP fraction of cells were isolated, while thyroid differentiation marker genes such as Tg, Tpo, Tshr, and two transcription factors critical for thyroid development and thyroid-specific gene expression, Nkx2-1 and Pax8 were expressed only in non-SP fraction of cells as determined by qRT-PCR (34). When SP cells were cultured in three-dimensional (3D) collagen gel in media containing fetal calf serum (FBS) in vitro, they remained at similar numbers for 9 weeks without significant changes in morphology, while non-SP cells rapidly expanded and spherical structures began to form at around 3 weeks, which further increased in number and size by 5 weeks (34). The fact that SP cells remained viable for 9 weeks without any sign of cell death suggested that they may have stem/progenitor ability, but require correct signal to start proliferation. Further, histological analysis and immunohistochemistry of cultured non-SP cells carried out at 9 weeks demonstrated the presence of follicle-like structures positive for NKX2-1 and TG. These results revealed that mouse thyroid SP cells exhibit stem/progenitor cell-like characteristics.

Side population cells were also found in various human thyroid cancer cell lines (18). They showed clonogenic ability higher than non-SP cells in vitro, however in in vivo tumorigenesis assays using nude mice, tumors were formed in most cases regardless of cell types. These results suggested that cancer stem cells may not be identical with cells contained in SP cells obtained from wild-type mouse thyroids.

Monolayer cultures derived from human nodular goiters in media containing FBS expressed mRNAs encoding stem cell marker OCT4 and endoderm markers GATA4 and HNF4α for at least seven passages (46). Addition of TSH to the media did not change the percentage of OCT4-positive cells, and no cells were double-positive for OCT4 and GATA4 or HNF4α as judged by flow cytometric analysis. By use of immunostaining of monolayer cultures and human thyroid goiter tissues, a very small number of single isolated cells were present that were positive for OCT4, GATA4, or HNF4α protein. However, qRT-PCR analysis of FACS-isolated GATA4 or HNF4α-positive fraction of cells, or OCT4-positive cells demonstrated the expression of mRNA for OCT4, or the endodermal markers, respectively, suggesting heterogeneous population of OCT4-positive cells; some cells may have started commitment toward endoderm lineage. The results thus provided the evidence for the presence of adult stem/progenitor cells of endodermal origin in human thyroid gland (46).

Side population cells were isolated from human goiters that express OCT4 and ABCG2, but not other genes HNF4α, GATA4, PAX8, and TG, TPO, TSHR, and NIS as thyroid differentiation markers (47). When maintained in monolayer culture or in Matrigel under serum- and TSH-containing medium for up to 14 days, neither cell attachment nor growth was observed even under intense growth stimulation. This situation was somewhat similar to mouse SP cell cultures as described above, in which stem/progenitor-like cells remained non-proliferative in monolayer or 3D cultures for 9 weeks (34).

Single cell suspension of primary human thyrocytes from nodular goiters was cultured in a special serum or TSH free-medium that allowed formation of floating spherical colonies (47). These spheres grew in size during first 3–4 days in culture, and expressed stem cell markers OCT4 and ABCG2, and endodermal markers GATA4 and HNF4α. The sphere-derived cells contained SP cells enriched about 50-fold as compared with those before sphere formation (5 vs. 0.1%, respectively). After 3 days of differentiation initiation with serum, sphere-derived cells began to express TSHR gene, which further intensified in the presence of both serum and TSH. These sphere-derived cells differentiated into thyrocytes by day 21, which expressed thyroid differentiation markers, PAX8, TG, NIS, TSHR, and TPO mRNA, but not stem cells or endodermal markers. Collagen embedded sphere-derived differentiated cells formed thyroid follicle-like structures, which displayed TSH-dependent ¹²⁵iodide uptake, a hallmark of differentiated thyroid cells.

Thyroid cell aggregates called “thyrospheres” were generated by culturing fresh surgical human thyroid fragments residual to collagenase digestion in defined media containing EGF and bFGF (48). Normal thyroid, normal perinodular tissue of the thyroid adenomas as well as all the different pathological thyroids were used to generate thyrospheres, from which 23 lines were established after 2 months of culture. Twelve such cell lines examined contained a subpopulation of CD34(+)/CD45(−) cells. When these spheroid cells were seeded in collagen gels in the presence of “differentiation medium” containing serum, follicle-like structures were formed. In contrast, spheroids cultured in “spheroid culture medium,” follicle-like structures did not form (48). Cells within the follicles were all positive for TG and the levels of T4 progressively increased in the supernatant of the spheroids cultured in “differentiation medium.” Further, when most freshly isolated spheroids were co-cultured with a neuroblastoma cell line or in adipogenic medium, the produced progeny had the expression of the neuronal marker β-tubulin III or positive staining for oil red O. These experiments demonstrated that stem/progenitor cells exist within the thyroid with an intrinsic ability to generate thyroidal cells and the potential to produce non-thyroidal cells.

All the studies as described provide clear evidence that adult thyroid indeed possesses stem/progenitor cells although the real origin of these cells has yet to be determined. Further, spheres/spheroids formation in serum-free media seems to be critical as the first step to have isolated stem/progenitor cells differentiate into functional thyrocytes in vitro. Indeed, when mouse and human embryonic stem cells were subjected to differentiation into thyrocytes in serum-free or growth factor-reduced medium, they went through embryoid bodies, similar in shape as spheres/spheroids (1, 2, 7, 49). Further, stem cells from other tissue sources are also often grown in suspension cultures to maintain stem cell properties (50, 51). Taken together, spheres/spheroids formation may provide stem cells the condition that mimics the thyroid microenvironment so-called “niche” that regulates activities of thyroid stem cells in culture (52, 53).

In the past decades, partial thyroidectomy was mainly used to study the effect of decreased levels of endogenous thyroid hormone or exogenously administered thyroid hormone on liver regeneration, enzymatic activities/functions, or the levels of thyroid hormone-regulated molecules in the brain, hypothalamus, pituitary, and liver (56–61). Other studies used partial thyroidectomy to carry out quantitative analysis of the thyroid function after inoculating thyroid cells into partial/total thyroidectomized rodents and their relation to neoplasms (31–33).

A mouse model for thyroid regeneration was established using partial thyroidectomy to study the effect of partial thyroidectomy on the morphology and gene expression of the remnant tissues/cells (62). In this study, partial thyroidectomy was carried out to remove one whole thyroid lobe and ~2/5 caudal segment of the other lobe. During this study, it was noted that the central areas of both thyroid lobes (without partial thyroidectomy) serve as the proliferative centers where many immature microfollicles, and bromodeoxyuridine (BrdU)-positive and/or C cells are located. One week after partial thyroidectomy, serum TSH levels were drastically elevated, resulting in a goitrogenesis condition, which was almost completely resolved by 2 weeks post-partial thyroidectomy. At that time, a marked increase of the number of BrdU-positive cells and cells with clear or faintly eosinophilic cytoplasm were observed in the central area and the area continuous to the cut edge (Figure 1A) (62). Some clear cells were BrdU-positive, suggestive of active proliferation, and expressed Foxa2, the definitive endoderm lineage marker (Figure 1B). Microarray followed by pathway analysis revealed that the expression of genes involved in embryonic development and cancer was affected by partial thyroidectomy. Electron microscopy demonstrated that clear cells were scant in cytoplasm, and retained features reminiscent of C cells having dense neuroendocrine granules or follicular cells juxtaposed to a lumen with microvilli. Taken together, clear cells may be immature cells that were previously C cells or follicular cells that participate in repair, regeneration of the thyroid gland, and/or goitrogenesis after partial thyroidectomy.

The first evidence of participation of stem cell antigen-1 positive (SCA-1+) cells in development of thyroid follicles after partial thyroidectomy was provided by the use of β-galactosidase (β-gal) reporter mice in conjunction with partial thyroidectomy as a model for thyroid regeneration, and BrdU long label-retaining cell analysis (63). The β-gal reporter mice express β-gal in a thyroid follicular cell-specific fashion upon expression of Cre recombinase that is controlled by the human TPO gene promoter. The TPO promoter is regulated by NKX2-1 (10, 11), and becomes active around embryonic day (E) 14.5–15.5 of mouse gestation (64), around the time when thyroid hormone production commences (65). This indicates that cells expressing β-gal have gone through thyroid differentiation, thus β-gal mice being able to use in a follicular cell lineage tracing experiment.

In this study, the main focus was SCA1(+) cells based on the previous observation that SP cells exhibited stem/progenitor cell-like characteristics, and a half of them were SCA1(+) (34). Partial thyroidectomy was carried out by removing caudal one-third of both thyroid lobes as opposed to removing one whole thyroid lobe and ~2/5 caudal segment of the other lobe used in the study in which the presence of clear cells were observed (62). In non-partial thyroidectomized thyroid, SCA1(+) cells were occasionally found only in vascular endothelial cells. One to two weeks after partial thyroidectomy, SCA1(+)/BrdU(+), but β-gal(−)/NKX2-1(−) cells were found in the non-follicular mesenchymal areas (Figure 2A). They were negative for CD34, suggestive of non-hematopoietic origin, and also negative for OCT3/4, GATA4, and SOX10, a marker for stem cells, early endoderm cells, and neural crest-derived cells, respectively. However, lack of expression of these genes may have simply been due to missed timing for detection. Interestingly, they temporarily expressed cytokeratin 14 (KRT14), suggesting that they had started acquiring epithelial features. Some SCA1(+)/BrdU(+)/KRT14(+) cells were also found in part of follicles at 14 days post-partial thyroidectomy. After partial thyroidectomy, follicles in the area close to the cut edge became irregular in shape with low or no colloid inside, frequently having ciliated intrafollicular cells. The number of ciliated intrafollicular cells peaked at day 35 through 75 post-partial thyroidectomy. In the affected irregular follicles at day 35 post-partial thyroidectomy, most intrafollicular cells were SCA1(+)/BrdU(+)/β-gal(−)/NKX2-1(−), suggesting that these newly proliferated cells did not originate from previously differentiated follicular cells and have not yet differentiated to express NXK2-1 because NKX2-1 regulates TPO gene expression, on which β-gal expression depends (Figure 2B). However, at day 120 post-partial thyroidectomy thyroid, some intrafollicular cells were found to be SCA1(+)/BrdU(+)/β-gal(+)/NKX2-1(+) (Figure 2C). This suggests that after partial thyroidectomy, the irregular shaped follicles may become functional follicles expressing TPO. It was estimated that only ≤0.1% of intrafollicular cells expressed SCA1 at day 35 post-partial thyroidectomy that may have participated in the regeneration and/or differentiation of thyroid follicular cells after partial thyroidectomy. In this study, however, it was not possible to demonstrate the origin of SCA1(+) cells, whether SCA1(+) cells actually migrate from mesenchymal areas into follicles, and whether SCA1 can be used as a marker for thyroid stem/progenitor cells. The origin of SCA1(+) cells may be resident adult thyroid stem/progenitor cells, bone marrow-derived cells (see below), or cells derived from yet to be determined origins. That no CD34 expression were found in SCA1(+) cells suggests against that SCA1(+) cells are bone marrow-derived cells. However, the possibility cannot be excluded that bone marrow-derived cells have changed their characteristics after settling in the thyroid microenvironment that includes loss of CD34 expression. These questions remain to be answered.

It is known that bone marrow-derived mesenchymal stem cells are pluripotent progenitors that have self-renewal proliferation and multipotent differentiation capacity into many lineages of tissues from mesoderm, ectoderm, and endoderm origins, and they home in response to injury signals (66–68). The mesenchymal stem cells, however, exhibit high heterogeneity without expression of specific surface markers (66–68). GFP-expressing bone marrow cells derived from C57BL/6 mice were grafted to irradiated C57BL/6 mice to study whether mesenchymal stem cells participate in the thyroid gland regeneration after X-ray exposure (69). GFP-positive signals in cells which co-expressed TG were found in thyroid follicles at the rate of ~6% when examined at 40 weeks post-X-ray irradiation at 5 Gy. This was the first demonstration that bone marrow mesenchymal stem cells indeed participate in the repair of thyroid after irradiation.

In the experimental autoimmune thyroiditis model mice established by immunization with TG, their thyroids were virtually completely destroyed after a month (70). However, the thyroid showed remarkable regeneration when observed up to 100 days, suggesting the presence and activation of adult thyroid stem cells (70). On day 0 of immunization, the expression of Oct4 was observed by RT-PCR, the level of which decreased in the thyroid of day 35 post-immunization, in keeping with the notion that OCT4 expression decreases when stem cells begin differentiation (71). Regeneration was faster in the absence of CD24, possibly due to the effect of CD24 on the infiltrating lymphocytes. These mice may provide an alternative model to study the mechanisms of thyroid regeneration.

Ultimobranchial body is composed of two types of cells: one expressing NKX2-1 and the other expressing p63 (91). Most cells express NKX2-1 while only a few cells express p63 when examined at E13.5, the time before UBB fuses with the thyroid primordium (Figures 3A–C) (91, 92). The NKX2-1-expressing cells do not overlap with the p63-expressing cells. Using Nkx2-1-null mice (8), it was demonstrated that NKX2-1-expressing cells were essential for survival of the UBB cells during its migration to meet with thyroid primordium, while p63-positive cells could proliferate in the absence of NKX2-1 expression (91). At E18.5, the latter cells gave rise to a mass of p63-positive cells in a nested pattern and/or a cystic vesicular structure that was lined by a monolayer of p63-negative cells of unknown phenotype, surrounded by a cluster and/or single layer of p63-positive cells, displaying the basal/stem cell appearance (Figures 3D,E). Occasionally, ciliated cells were found in the lining epithelium of the vesicle (Figures 3D,F). This structure demonstrated a striking similarity in appearance to SCN. Note that the thyroid primordium cells of Nkx2-1-null mice disintegrated through apoptosis (93) and only this cystic structure remained in Nkx2-1-null mice around the area where the thyroid was supposed to be located (91, 92). On the other hand, p63 was not required for the normal development of thyroid gland (92).

The nature of the cystic structure found in Nkx2-1-null mice was further studied using Nkx2-1; p63-double null mice (92). In these mice, the p63-positive outer layer of cells surrounding the cystic vesicular structure or those in nested patterns continued to the vesicular structure were no longer found. Instead, only p63-negative single layer cells in tubular or cystic patterns remained, which appeared dilated (Figures 3F,G) (92). Occasionally, ciliated cells were observed in the cysts (Figures 3F,G) (92). Cells of the Nkx2-1; p63-double null UBB cysts were negative for all the following markers; neuron-specific enolase and S100 as neuron-specific markers, OCT4 as a stem cell marker, nestin and SOX10 as a neural crest marker, and vimentin as a mesenchymal marker. Electron microscopic examination revealed that the Nkx2-1; p63-double null UBB were composed of immature cells with scarce cytoplasm and no clear intracellular organelles or adhesion structures (92). While the presence of these immature cells were observed only in Nkx2-1; p63-double null mice, they may always be present within the thyroid of wild-type mice; whether they are dispersed as individual cells throughout the thyroid, or stay clustered within thyroid remain unknown.

In the partial thyroidectomy model mice, ciliated intrafollicular cells were mainly found in the area where many irregular shaped follicles were present, which drastically increased in number after partial thyroidectomy; 1 in 100 intrafollicular cells had cilia (63), which otherwise were only sporadically found in fully developed adult thyroids (78–80). Further, in these irregular follicles, more number of newly differentiated intrafollicular cells was found after partial thyroidectomy (63). It is possible that the immature cells found in SCN may represent irregular shaped follicles with ciliated intrafollicular cells that become functional follicles after partial thyroidectomy. However, these cells do not express OCT4, suggesting that they may be different from OCT4-expressing adult-resident thyroid stem/progenitor cells.