Molecular pathogenesis and mechanisms of thyroid cancer

Gene amplifications and copy-number gains

Oncogenic gene amplification or copy-number gain are additional prominent genetic mechanisms in thyroid tumori-genesis (TABLE 3). This is particularly the case for genes encoding receptor tyrosine kinases (RTKs)²⁶. Copy-number gains are also common for the genes encoding PI3K–AKT pathway members, including PIK3CA, PIK3CB, 3-phosphoinositide-dependent protein kinase 1 (PDPK1; also known as PDK1), AKT1 and AKT2 (REFS ^25,26,32). Overall, genetic copy-number gains of these genes are more prevalent in ATC than in DTC, suggesting that these genetic alterations are important for the progression and aggressiveness of thyroid cancer. Copy-number gain of these genes in ATC could occur either through genetic amplification or chromosomal instability and aneuploidy. An important and expected consequence of these copy-number gains is the activation of downstream signalling pathways, as is suggested by the association of copy-number gains of genes encoding RTKs with increased phosphorylation of AKT and ERK in thyroid cancer²⁶.

Many of the genes with copy-number gains are proto-oncogenes. A mechanism for them to contribute to thyroid tumorigenesis is through increased protein expression and consequent aberrant activation of the signalling pathways in which they are involved^25,26,33. It is interesting to note that in DTCs, PIK3CA mutation is mutually exclusive with PIK3CA copy-number gain^31,33, suggesting that either of these genetic alterations is sufficient to promote thyroid tumorigenesis through the PI3K–AKT pathway. However in aggressive undifferentiated ATC, PIK3CA mutations and amplifications often occur simultaneously in the same tumour^26,33, which is a mechanism through which mutant PIK3CA can be amplified to drive the progression and aggressiveness of thyroid cancer.

Gene amplification or copy-number gain of the IQ-motif-containing GTPase-activating protein 1 (IQGAP1) is another important genetic event that has been recently discovered in thyroid cancer⁴⁸. IQGAP1 is a multifunctional scaffold protein that has an important role in human tumorigenesis⁴⁹. IQGAP1 amplification was found to increase protein expression and to be associated with invasiveness of thyroid cancer cells. Interestingly, coexistence of IQGAP1 copy-number gain and BRAF^V600E mutation was particularly associated with a high incidence of recurrent PTC⁴⁸.

Gene translocations

Gene translocation resulting in oncogenic rearrangements in thyroid cancer is best exemplified by RET–PTC. There are more than 10 types of RET–PTC translocation, as determined by the types of partner genes, and the most common types are RET–PTC1 and RET–PTC3^50–52. RET is a proto-oncogene encoding an RTK. RET–PTC occurs as a consequence of genetic recombination between the 3′ tyrosine kinase portion of RET and the 5′ portion of a partner gene, such as the coiled-coil domain-containing gene 6 (CCDC6; also known as H4) in RET–PTC1 and nuclear receptor co-activator 4 (NCOA4; also known as ELE1) in RET–PTC3 (REFS ^53,54). Spatial contiguity of RET and the partner gene in the nucleus is a structural basis for the formation of RET–PTC rearrangements⁵⁵. The rearrangements result in ligand-independent dimerization and constitutive tyrosine kinase activity of RET. RET–PTC can occur in benign FTA and follicular variant PTC (FVPTC), but more commonly in classical PTC⁵⁶. A recent study found a correlation between the presence of RET–PTC and a high growth rate of benign thyroid tumours⁵⁷. However, the role of RET–PTC in early thyroid tumorigenesis is unclear. RET–PTC is a classical oncoprotein that activates the MAPK and PI3K–AKT pathways^58,59. RET–PTC activates both pathways by recruiting signalling adaptors to phosphorylated Tyr1062 on the intracellular domain of the RET fusion protein⁶⁰. It is thus not surprising that RET–PTC occurs in FTA and FVPTC; these are follicular thyroid tumours in which the PI3K–AKT pathway has a primary role in tumorigenesis⁶¹.

The paired box 8 (PAX8)–peroxisome proliferator-activated receptor-γ (PPARG) fusion gene (PAX8–PPARG) is another prominent recombinant oncogene in thyroid cancer, occurring in up to 60% of FTC and FVPTC^62–64. PAX8–PPARG also occurs in FTA⁶⁴, although, like RET–PTC, its prevalence is low in benign thyroid tumours and its oncogenic role is unclear. PAX8–PPARγ exerts a dominant-negative effect on the wild-type tumour suppressor PPARγ and also transactivates certain PAX8-responsive genes⁶⁵. Interestingly, the expression of PPARγ was dramatically reduced in FTC that developed in the TRβPV mouse model — which expresses a knock-in dominant-negative mutant of human thyroid hormone receptor-β (TRβ) — resulting in follicular thyroid tumorigenesis⁶⁶. An AKAP9–BRAF fusion gene resulting in activation of the BRAF kinase also occurs in ionizing-radiation-induced PTC⁶⁷, but not in sporadic PTC⁶⁸. Unlike RET–PTC, the importance of AKAP9–BRAF in thyroid tumorigenesis is probably limited owing to its rarity.

The PI3K–AKT signalling pathway

The PI3K–AKT pathway also has a fundamental role in thyroid tumori-genesis^61,98 (FIG. 2). Early evidence to support the role of this pathway in thyroid tumorigenesis came from the finding that Cowden’s syndrome (which is caused by germline mutations of PTEN) was associated with FTA and FTC⁹⁹. A functional role for the PI3K–AKT pathway in sporadic thyroid tumorigenesis was initially revealed by the finding of increased expression and activation of AKT in thyroid cancers, particularly FTC^100–102. Among the three AKT isoforms, AKT1 and AKT2, but not AKT3, were found to be robustly expressed and activated in thyroid cancer, suggesting a particularly important role for these two isoforms in thyroid tumorigenesis¹⁰¹.

Open in a separate window

Figure 2

The PI3K–AKT and related pathways in thyroid cancer

Extracellular signals activate receptor tyrosine kinases (RTKs) in the cell membrane, leading to the activation of RAS and subsequent activation of PI3K. Activated PI3K catalyses the conversion of phosphatidylinositol (4,5)-bisphosphate (PIP2) to phosphatidylinositol (3,4,5)-trisphosphate (PIP3). PIP3 activates 3-phosphoinositide-dependent protein kinase 1 (PDK1; also known as PDPK1), which consequently associates with AKT and leads to phosphorylation (P) and activation of AKT by PDK1. Phosphorylated AKT, which is an activated form of AKT, enters the nucleus where it induces tumour-promoting genes. In the cytoplasm, phospho-AKT also activates other signalling molecules or pathways, a prominent one being the mTOR pathway, which has an important role in tumorigenesis by promoting translation. Phospho-AKT can also directly phosphorylate glycogen synthase kinase 3β (GSK3β) and consequently inactivates it. GSK3β normally inhibits β-catenin, thus the effect of phospho-AKT is to relieve GSK3β-mediated suppression of β-catenin. Consequently, β-catenin can enter the nucleus where it promotes the expression of tumour-promoting genes. In the nucleus, phospho-AKT can phosphorylate forkhead box O3 (FOXO3) on its AKT-specific motif. Such phosphorylated FOXO3 is translocated out of the nucleus to the cytoplasm where it binds 14-3-3 proteins to be sequestered in the cytoplasm, thus terminating the pro-apoptotic activities of the FOXO3 pathway. The downward arrow for the FOXO activities in the nucleus in the figure indicates this negative effect of AKT on pro-apoptotic genes in the FOXO pathway, which would otherwise be upregulated by the FOXO3 pathway. This unique coupling of phospho-AKT to the three pathways provides a powerful driving force for thyroid tumorigenesis. The major negative regulatory mechanism of the PI3K–AKT pathway is PTEN, which is a phosphatase that converts PIP3 to PIP2, thus terminating the activation of the pathway. The inset shows the self enhancement mechanism of PI3K–AKT signalling in which genetic-alteration-driven activation of the pathway causes PTEN methylation and silencing with consequent loss of termination of the signalling, thus maintaining the pathway in full and constitutive activation.

In the TRβPV transgenic mouse model — in which the PI3K–AKT pathway is activated, which involves the interaction of TRβPV with the p85α regulatory subunit of PI3K — FTC spontaneously developed^103,104, and the activation and nuclear localization of AKT1 was observed¹⁰⁵. Human tumour studies suggested that the invasiveness and metastasis of FTC promoted by the PI3K–AKT pathway particularly involved the activation and nuclear localization of AKT1 (REF. 102). This is consistent with the occurrence of AKT1 mutations in metastatic thyroid cancers³⁵. Furthermore, Akt1 ablation delayed or prevented tumour progression, vascular intravasation and distant metastasis of FTC in TRβPV mice¹⁰⁶. The role of AKT2 in thyroid tumorigenesis has also been demonstrated in vivo, as the occurrence of thyroid tumours was reduced in Pten^+/−Akt2^−/− mice¹⁰⁷. Thus, whereas the MAPK pathway has a central role primarily in PTC, the PI3K–AKT pathway has such a role primarily in FTC and its invasion and metastasis. This is consistent with genetic alterations in the PI3K–AKT pathway occurring most frequently in FTC, whereas those in the MAPK pathway occur most frequently in PTC^{17,18,61,78,98}. The inhibition of thyroid cancer cell proliferation by PI3K–AKT pathway inhibitors was dependent on the presence of genetic alterations in the PI3K–AKT pathway^108,109, suggesting that thyroid cancer cells harbouring such genetic alterations have become dependent on the over-activation of this pathway.

Vascular invasiveness and capsular invasiveness are defining histopathological features of FTC¹¹⁰. Genetic alterations in the PI3K–AKT pathway are far more common in FTC than in its precursor, FTA^31,33. It is therefore conceivable that overactivated PI3K–AKT signalling promotes the conversion of FTA to FTC by conferring tumour cell invasiveness. There are also secondary molecular alterations, albeit limited in number so far, that are robustly induced by PI3K–AKT signalling and that have an important role in the pathogenesis of FTC; these include alterations of the WNT–β-catenin¹¹¹, HIF1α^112,113, FOXO3 (REF. 114) and NF-κB¹¹⁴ pathways.

The NF-κB signalling pathway

The NF-κB pathway has an important role in the regulation of inflammatory responses that are linked to tumorigenesis¹¹⁵ (FIG. 1). Increased NF-κB activation in thyroid cancer cell lines and tissues has long been documented^116–118. Recent studies demonstrate that the NF-κB pathway controls proliferative and anti-apoptotic signalling pathways in thyroid cancer cells^119,120. The NF-κB pathway is involved in upregulating the expression of several oncogenic proteins that are also upregulated by the MAPK pathway. Furthermore, RET–PTC, RAS and BRAF-V600E — members of the MAPK pathway — can cause activation of the NF-κB pathway in thyroid cancers¹¹⁹. Interestingly, NF-κB signalling was upregulated by expression of BRAF-V600E^121,122, but not by expression of a constitutively active mutant of MEK in NIH3T3 cells¹²³, suggesting a direct coupling of BRAF-V600E to the NF-κB pathway. Through a mechanism that is not yet specifically defined, BRAF-V600E was shown to cause IκB degradation (and consequent activation of NF-κB) independently of MEK–ERK signalling in thyroid cancer cells⁸⁴. This dual coupling of BRAF-V600E to the NF-κB pathway and to the MEK–ERK pathway is consistent with the finding that simultaneously suppressing the two pathways using NF-κB and MEK inhibitors synergistically inhibited the proliferation of thyroid cancer cells harbouring a BRAF-V600E mutation¹²⁴.

The RASSF1–MST1–FOXO3 signalling pathway

RASSF1 is a member of the RAS association domain family, which, in response to apoptotic signalling, associates with and activates mammalian STE20-like protein kinase 1 (MST1; also known as STK4)¹²⁵. Activated MST1 phosphorylates Ser207 in the forkhead domain of the forkhead transcription factor FOXO3. This phosphorylation disrupts the interaction of FOXO3 with 14-3-3 proteins in the cytoplasm and promotes FOXO3 translocation to the nucleus, where it promotes the transcription of pro-apoptotic genes¹²⁶ (FIG. 1). Thus, this RASSF1–MST1–FOXO3 pathway has an important tumour-suppressive role by promoting apoptosis.

Hypermethylation of the promoter of RASSF1A is common and is associated with its silencing in thyroid cancers^127,128. This occurs even in benign FTA, albeit to a lesser extent, suggesting that impairment of the RASSF1–MST1–FOXO3 pathway is involved in early thyroid tumorigenesis¹²⁸.

Activation of the RASSF1–MST1–FOXO3 signalling pathway promoted thyroid cancer cell apoptosis²¹. Interestingly, this study demonstrated that BRAF-V600E, but not wild-type BRAF, directly interacted with the carboxyl terminus of MST1 and inhibited its kinase activities, resulting in decreased FOXO3 transactivation. This was independent of MEK–MAPK signalling. This suggests that, in addition to the classical coupling to MEK–MAPK, negative regulation of RASSF1–MST1–FOXO3 is a mechanism that is involved in BRAF-V600E-driven thyroid tumorigenesis. This unique mechanistic model explains the finding that thyroid cancers that developed in BRAF^V600EMst1^−/− transgenic mice were more aggressive than those that developed in BRAF^V600E mice²¹. The direct interaction of BRAF-V600E with MST1 to prevent RASSF1-mediated MST1 activation also provides an explanation for the mutually exclusive occurrence of BRAF^V600E mutation and RASSF1A hypermethylation in thyroid cancer¹²⁸, as either event may be sufficient to inactivate RASSF1–MST1–FOXO3 signalling. Therefore, BRAF-V600E is coupled independently to three major pathways: MEK–MAPK, RASSF1–MST1–FOXO3 and NF-κB, and thus represents a unique and powerful oncogenic mechanism in thyroid tumorigenesis (FIG.1). The activation of the PI3K–AKT pathway, such as that induced by PTEN inactivation, can also downregulate the activities of the RASSF1–MST1–FOXO3 pathway in FTC¹¹⁴. This involves AKT-mediated phosphorylation of FOXO family members, which causes their translocation from the nucleus to the cytoplasm, where they are sequestered by 14-3-3 proteins, thus reducing the transcription of pro-apoptotic genes¹²⁹ (FIG. 2).

The WNT–β-catenin signalling pathway

The WNT–β-catenin pathway has a well established role in the regulation of cell growth and proliferation, as well as in stem cell differentiation, and its constitutive activation is commonly found in human cancers¹³⁰. β-catenin, when upregulated by upstream WNT signalling, is translocated into the nucleus and transcribes various tumour-promoting genes. In thyroid cancer, activation of WNT–β-catenin signalling is often caused by activating mutations of CTNNB1 (which encodes β-catenin), particularly in PDTC and ATC^37,38. Furthermore, the expression of β-catenin was higher in ATC than in DTC¹³¹. Thus, the WNT–β-catenin pathway seems to have a particularly important role in thyroid tumour aggressiveness.

Aberrant activation of WNT–β-catenin signalling often occurs as a consequence of the activation of the PI3K–AKT pathway in thyroid cancers^132,133. This occurs through glycogen synthase kinase 3β (GSK3β), which is directly phosphorylated — and hence inactivated — by AKT¹³⁴. As GSK3β promotes the degradation of β-catenin, its inactivation results in the upregulation of WNT–β-catenin signalling (FIG. 2). Interestingly, RET–PTC can activate the WNT–β-catenin pathway by activating the PI3K–AKT pathway and also by directly phosphorylating β-catenin in thyroid cancer cells⁵⁸.

The HIF1α pathway

It has long been known that hypoxia is a strong stimulus of cancer metabolism, growth and progression. HIF1α is a key mediator of the response to hypoxia, in which it binds to HIF1β (also known as ARNT) to form the HIF1 transcription factor that induces the expression of various genes associated with cell metabolism and tumour angiogenesis¹³⁵. Angiogenesis, which is a key step in the progression of solid tumours, is a common response to intratumoural hypoxia. This process is promoted by VEGFA, the expression of which is strongly upregulated by HIF1. HIF1α is not expressed in normal thyroid tissues but is expressed in thyroid cancers, particularly in aggressive types, such as ATC, and this is consistent with a role in thyroid cancer progression^112,136. The oncogene MET, which is another target of HIF1, is also abundantly expressed in association with upregulated HIF1α in thyroid cancer¹³⁶. Interestingly, HIF1 can be upregulated in thyroid cancer by both the PI3K–AKT^112,113 and the MAPK pathways⁸⁸, thus contributing to the effects of these two pathways on thyroid tumour progression.

This work is supported by the US National Institutes of Health (NIH) R01 grants CA113507 and CA134225 to the author. The author thanks A. K. Murugan, Y. Trink and D. Liu for their help in preparing the figures and references. The author also thanks the numerous colleagues and investigators in this field whose outstanding work made it an extremely enjoyable experience to write this Review. The author wishes to apologize to those whose work is not cited owing to space limitations.