In mouse embryos, by E9.5 the thyroid anlage expands and then forms a bud that moves caudally in the surrounding mesenchyme. In the early phase of morphogenesis, the proliferation rate of pTFC is lower than that of the other endodermal cells [13]. Hence, enlargement of the developing thyroid probably depends on the recruitment of other endodermal cells of the primitive pharynx into the thyroid anlage.

The migration of pTFC is a notable feature of thyroid organogenesis. Morphogenesis of the neck encompasses a number of events that can contribute to the translocation of pTFC in the direction of the trachea. However, translocation mostly involves active migration of the precursor cells, which probably move through a process of “collective cell migration” by which cells move in concert without completely disrupting their cell-cell contacts [17]. The genetic basis of the thyroid bud migration has been, at least in part, elucidated; the presence of transcription factor Foxe1 in pTFC is required for thyroid bud migration [11, 18].

At E11.5, after the thyroid primordium is no longer connected to the pharynx, the process of lobulation begins, attested by a lateral enlargement of the developing thyroid; one day later, thyroid appears as an elongated structure close to third pharyngeal arch arteries that will participate in the formation of the definitive carotid vessels. These reports suggest that inductive signals originating from adjacent vessels could be relevant in the lobulation of the gland. This hypothesis was validated using genetically modified mice deprived of sonic hedgehog [19] (a morphogen relevant in vertebrate organogenesis) or Tbx1 [16] (a transcription factor regulated by sonic hedgehog itself). In these models, the development of vessels is impaired, and the thyroid bud does not stay in contact with them. At the same time, the lobulation process is disturbed, and the gland does not separate into two distinct lobes but appears as a single mass. However, in addition to extrinsic signals, pTFC specific mechanisms are required for correct lobulation. In fact, in mice partially deficient for both the transcription factors Nkx2-1 and Pax8, thyroid hemiagenesis is frequently observed [20].

In the postnatal thyroid, mechanisms triggered by TSH upon its interaction with Tshr are extremely relevant in the regulation of both proliferation and function of TFCs. The finding that in the mouse Tshr is detectable in thyroid cells by E14 and TSH by E15 could suggest a role for TSH/Tshr signaling in the control of functional differentiation and proliferation of TFC in the developing thyroid. However, in mouse embryos lacking a functional Tshr, the size of the fetal thyroid is quite similar to that in wild-type mice [21]. In addition, the number of proliferating cells in the thyroid is comparable in mutants and wild-type embryos. On the contrary, adult mice lacking TSH signaling have a clearly hypoplastic thyroid. These data indicate that in the mouse, TSH/Tshr signaling does not have a relevant effect on the proliferation of thyroid cells during embryonic life [22]. This scenario could be different in humans. In fact, in the mouse, thyroid organogenesis is complete just at the end of gestation, and the functions of the hypothalamic-pituitary-thyroid axis are accomplished only after birth. In humans,, the thyroid is realized by 12–13 weeks; during the rest of gestation, the thyroid continues to grow and the hypothalamic-pituitary-thyroid axis is functioning. While the role of Tshr/TSH signals in controlling the growth of fetal thyroid remain rather obscure, molecular analysis reveals that both TPO and NIS are almost absent in mouse embryos lacking a functional Tshr. Thus, TSH induced signals are required to express key molecules for hormone biosynthesis and to complete the differentiation program of the thyroid follicular cells [21].

Finally, TSH signaling is not required for folliculogenesis because normal thyroid follicles are present in mouse embryos lacking a functional Tshr. Signals from endothelial cells are relevant in forming follicles. In fact, in animal models characterized by a thyroid-specific severe reduction of vascular density, the gland appears as a multilayer mass of nonpolarized cells [23].

In placental mammals, the ultimobranchial bodies are embryonic transient structures derived from the fourth pharyngeal pouch, which will merge with the developing thyroid.

In mice by E10, the fourth pharyngeal pouches are detectable as lateral extroflexions of the primitive foregut expressing the transcription factor Islet1 [24]. One day later, the caudal portion of the pouches grows forming UBB primordia identified for the expression of the factors Hes-1 [25], Isl1, and Nkx2-1 [26]. UBBs begin to migrate and reach the primordium of the thyroid at E13. By E14.5, UBB cells begin to disperse within the thyroid parenchyma, and only remnants of UBBs can be distinguished in the thyroid gland.

The expression of Nkx2-1 is required for the survival of UBBs [26], while Hes-1 plays a role in the merging of these structures into the thyroid [25]. UBB defects have been also described in animal models with an impaired expression of Hox3 genes, encoding factors involved in the morphogenesis of several structures. In these models, UBBs do not merge into the thyroid but remain as bilateral vesicles composed exclusively of calcitonin-producing cells (persistent ultimobranchial bodies) [27].

The generation of an increasing number of mouse models carrying mutations on demand of specific loci has been offering biomedical research an invaluable tool for the identification of the functions of proteins in vivo and the dissection of the pathways underlying developmental processes. Also in the case of the thyroid, a large part of our knowledge of molecular mechanisms of its morphogenesis is due to studies of phenotype mice, in which genes encoding proteins expressed during thyroid development have been inactivated by gene targeting (knockout mice).

Molecular genetics of thyroid development was inaugurated by the discovery of a number of transcription factors expressed in the thyroid anlage, Nkx2-1 (formerly called TTF-1), Foxe1 (formerly called TTF-2), Pax8, and Hhex. These genes, which can be considered thyroid “toolkit genes,” remain expressed in both precursor and differentiated thyroid follicular cells. Although these factors are also expressed in other tissues, the coexpression of all four is a peculiar hallmark of thyroid cells [11].

In the next paragraphs, the focus will be on the role of these genes as deduced by their distribution and by the phenotype of knockout animals; the consequences of mutations in these genes in the pathogenesis of thyroid diseases will be exhaustively treated in later chapters.