The Year in Basic Thyroid Cancer Research

Thyrotropin levels are inversely associated with thyroid cancer risk

It has become clear that the levels of thyrotropin (TSH) are, at least in part, genetically determined and heritable. In fact, several loci associated with TSH variance have already been identified. The relationship between TSH levels and risk of developing thyroid cancer, on the contrary, is still a matter of debate. Five thyroid cancer risk alleles, identified using genome-wide association studies, have previously been found to associate with low TSH levels.

Now, a new study by Zhou et al. (1) has identified novel loci associated with TSH levels through a meta-analysis of over 100,000 individuals from three large genetic studies. Seventy-four loci were identified, 28 of which were completely novel, and the 5 loci mentioned above were included in the list, confirming the validity of the approach. Given the modest effect size of each variant, the authors looked at the cumulative effect of multiple alleles. Thus, polygenic scores (PGS) were obtained for the 95 independent non-human leukocyte antigen top variants, and their association with 1238 diseases was examined.

The most relevant finding for the purpose of this review is that the TSH PGS was inversely associated with the risk of thyroid cancer: the odds ratio for thyroid cancer of subjects having a PGS in the highest quintile was between 0.11 and 0.89, depending on the cohort analyzed. Two-sample Mendelian randomization confirmed this association, with one SD increase in TSH levels (still within normal range), corresponding to a 45% decrease in thyroid cancer risk. These findings strengthen the notion that low TSH is associated with increased thyroid cancer risk, and represent a solid framework to fully elucidate the role of TSH in thyroid cancer.

Radiation exposure-associated papillary thyroid cancer is functionally similar to sporadic papillary thyroid cancer

Morton et al. (2) reported the large-scale genomic, epigenetic, and transcriptomic profiling of 359 papillary thyroid cancers (PTCs) and matching normal tissues from individuals with measurable childhood exposure to ¹³¹I as a consequence of the Chernobyl nuclear accident, compared with 81 PTCs from unexposed children. The goals of the study were to identify biomarkers of radiation-induced cancers and to characterize the genomic landscape of such tumors.

The results presented in the article show that while the frequency of clonal small deletions and balanced translocations increases with radiation dose, the frequency of single-nucleotide changes does not. Small deletions and balanced translocations were found to specifically originate from nonhomologous end-joining rather than from alternative end-joining repair mechanisms. Similar to sporadic PTCs, most radiation-induced tumors carried a single genetic driver. However, while most tumors from patients with low exposure carried point mutations in BRAF and in RAS genes, those exposed to higher radiation doses were enriched for gene fusions, often involving RET.

Strikingly, transcriptomic and epigenetic profiling failed to identify molecular changes specific for radiation-induced PTCs, and, as in sporadic tumors, the transcriptomic and epigenetic profiles were strictly dictated by the nature of the driver mutation. Thus, while radiation-induced PTCs cannot be recognized using specific biomarkers, the overlap in driver mutations and expression profiles between sporadic and radiation-associated PTCs indicates that the clinical approaches to sporadic PTCs are also valid for radiation-induced PTCs.

Novel gene for familial PTC

Familial nonmedullary thyroid cancer accounts for up to 15% of all thyroid cancer cases. While a few germ line mutations conferring susceptibility to hereditary thyroid cancer have been identified, the etiology of most cases remains obscure. Zhao et al. (3) reported the discovery of germ line loss-of-function mutations in the WDR77 gene in two unrelated families with hereditary PTC. WDR77 encodes a protein that recruits substrates to and activates the arginine methyltransferase, PRMT5.

Accordingly, affected patients displayed significantly reduced histone H4 methylation (a PRMT5 target) and altered gene expression profiles compared with noncarriers, including dysregulated expression of cell cycle- and apoptosis-related genes. In vitro experiments showed that WDR77 knockdown increases cell proliferation not only in thyroid cancer cells but also in the HEK293 cell line, suggesting a more general effect of WDR77 reduction. In fact, Wdr77 heterozygous mice were previously shown to have increased proliferation in the prostate and testis (4,5).

Furthermore, mining several cancer and noncancer databases revealed germ line as well as somatic WDR77 mutations, some of which were associated with different tumor types. Additional studies are now required not only to clarify the overall prevalence of WDR77 mutations in familial thyroid cancer but also to unravel the molecular mechanisms leading to neoplastic transformation specifically of thyroid epithelial cells in these individuals. In summary, this report has the merit to add one novel target to the list of predisposition genes for thyroid cancer and possibly other tumor types.

Early anaplastic thyroid cancer evolution from a subset of PTC cells

Using single-cell RNA sequencing, Luo et al. (6) have provided new compelling evidence that anaplastic thyroid cancer (ATC) often derives from a subset of cells within preexisting PTC, and have identified CREB3L1 as a master transcription factor potentially driving the dedifferentiation process associated with ATC. Three PTCs, three ATCs, and one normal thyroid tissue were profiled at the single-cell level and the different cell populations thus identified were clustered using canonical markers. The results offer a detailed view of the microenvironmental changes associated with tumor progression and of the evolutionary trajectory of ATC. ATC cells display a molecular evolution that appears to be different from that of pure PTCs, while overlapping with that of some of the PTC cells derived from ATC samples, indicating their common early origin and coevolution.

Copy number alterations were, in large part, shared between ATCs and ATC-associated PTCs, further supporting the evolution of ATC from PTC. Notably, ATC cells showed not only the specific overexpression of genes in the MAPK and PI3K/mTOR pathways but also a significant enrichment in the expression of CREB3L1 and of its target genes, many of which control epithelial-to-mesenchymal transition and mTOR signaling. In fact, high expression of CREB3L1 was found to be associated with increased likelihood of recurrence and with worse survival in PTC patient data sets. Thus, this study has identified a possible master mechanism of dedifferentiation associated with progression of a subset of PTC cells to ATC.

Aberrant RNA editing contributes to thyroid cancer progression

The ADAR1 adenosine deaminase acts on double-stranded RNA to convert adenosine into inosine (a process called editing). Inosine, in turn, is read by the splicing and translation machinery as guanosine, de facto creating a point mutation. Thyroid cancer RNAs appear to have an increased level of RNA editing, compared with normal thyroids, and one of the highest editing levels among different cancer types (from TCGA data). Ramirez-Moya et al. (7) recently reported that ADAR1 silencing in different PTC and ATC cell lines reduced cell viability, invasion, and in vivo growth, indicating a possible contribution of this mechanism to the oncogenic process. In a follow-up article (8), the same group used RNA-seq to investigate the effect of ADAR1 silencing on gene expression in an ATC cell line.

The results show a strong effect not only on global gene expression but also on splicing patterns of several genes and on the editing of specific transcripts. Notably, only one mRNA, encoding the splicing-related CDK13 kinase, was edited in a way that introduced a missense mutation in the coding region. This mutation was shown to affect cell proliferation and invasion properties, suggesting that this might be a novel putative oncogenic process contributing to thyroid cancer development. This notion warrants further in-depth studies to validate this novel mechanism and understand the extent of its involvement in thyroid cancer development and progression.

Mitochondrial stress upregulates GDF15 and increases invasiveness

Kang et al. (9) have shown that several thyroid cancer cell lines exhibit reduced mitochondrial function and show molecular markers of the “mitochondrial stress response,” a process that leads to the upregulation of the expression of mitokines that not only help these cells to maintain their functions, but may also contribute to tumor behavior. In fact, one of these mitokines, GDF15, was found to be significantly upregulated in thyroid cancer cells, and its levels further increased upon doxycycline-induced additional mitochondrial stress.

The authors also found that GDF15 expression was elevated in patients with more aggressive PTC, and that siRNA-mediated GDF15 knockdown reduced cell proliferation and motility. From a mechanistic standpoint, it appears that GDF15 activates STAT3, and that STAT3 regulates GDF15 expression in a positive feedback loop. Combined GDF15 and STAT3 targeting reduced ATC cell proliferation in a xenograft model. Thus, this study identified a novel regulatory loop contributing to thyroid cancer aggressiveness and sets the stage for novel studies aimed at targeting this mechanism in the preclinical setting.

Isoform-specific role of AKTs in thyroid cancer

AKT activation is frequently observed in thyroid cancer, in particular, in its most aggressive variants. However, the relative contribution of the three different AKT genes to the tumorigenic process is not known. Saji et al. (10) described a systematic genetic approach to address this issue in vivo. Mice carrying deletion of Akt1, Akt2, or Akt3 were crossed with mice carrying the TRβ PV mutation, which develop follicular thyroid carcinoma, and the resulting compound mutants were analyzed to assess the impact of each AKT mutation on several tumor-related phenotypes. Analysis at one year of age revealed that loss of any of the three Akt isoforms reduced the overall thyroid size, and that loss of Akt1 specifically impaired both tumor initiation (by impacting cell survival rather than proliferation) and local invasion.

Both Akt1 and Akt3 were found to control vascular invasion, while deletion of any Akt isoform severely reduced lung metastasis, suggesting nonredundant roles in controlling the metastatic process. Expression analysis of Akt-depleted thyroid tumors revealed the overexpression of the antigen-presenting cell marker CD209a exclusively in the Akt1 KO model, suggesting that dendritic cells were specifically depleted in thyroid cancer in an Akt1-dependent manner. These studies point at AKT1 as a major player in thyroid cancer development through the control of cell survival and by inducing an immune-suppressive microenvironment.

Chromatin remodeling locks ATC in a dedifferentiated state

Mutations in genes encoding subunits of the switch/sucrose non-fermentable (SWI/SNF) chromatin remodeling complexes are relatively common in poorly differentiated thyroid carcinoma (PDTC) and ATC. Saqcena et al. (11) recently described a sophisticated in vivo approach to analyze the role of these mutations in thyroid cancer biology, control of differentiation, and response to therapy. To this end, mice carrying a conditional Braf^V600E allele were crossed with mice carrying floxed alleles of Arid1a, Arid2, and Smarcb1, which are among the most commonly mutated SWI/SNF subunits. While single Braf mutants developed PTC by five weeks and single SWI/SNF mutants were histologically normal, compound mutants developed PDTC and ATC with high penetrance and short latency.

Cells from these tumors showed that the loss of SWI/SNF genes significantly reduced chromatin accessibility, in particular at sites with DNA binding motifs for the thyroid specification factors FOXE1, NKX2-1, PAX8, and at genes involved in thyroid hormone synthesis. Notably, despite a reduction in the MAPK transcriptional output upon MEK inhibition, the deletion of the SWI/SNF genes impaired the ability of MEK inhibitors to restore the expression of thyroid differentiation markers, and significantly impaired the restoration of radioactive iodine (RAI) uptake. Accordingly, a phase II redifferentiation clinical trial using trametinib showed no RAI uptake in patients carrying SWI/SNF subunit mutations. Thus, loss of epigenetic modifier genes locks thyroid cancer cells in an irreversible dedifferentiated state leading to RAI therapy failure.

Autoimmune thyroiditis affects PTC natural history

Despite having been reported in numerous clinical studies, the significance of the association between autoimmune thyroiditis and PTC remains controversial. Pani et al. (12) have used a genetic approach to induce in a temporally controlled and independent manner both thyroiditis and PTC. Tamoxifen-inducible BRAF^V600E mice were crossed with NOD.H2^h4 mice, which develop Hashimoto-like features that are accelerated by sodium iodide treatment.

Three cohorts were studied: mice in which tamoxifen induced PTC without concomitant thyroiditis, simultaneous thyroiditis and PTC induction, and preexisting thyroiditis followed by PTC induction. Interestingly, the results showed that preexisting thyroiditis was associated with significantly increased overall survival of the mice. Although all the mice eventually developed PTC, analysis of mice at a younger age, before the death endpoint, showed that PTCs in the preexisting thyroiditis cohort were less frequent and smaller in size, suggesting that autoimmune thyroiditis delays the onset and decreases aggressiveness of BRAF-driven PTC. Mice with preexisting thyroiditis displayed an increase in infiltration of CD8-positive T cells and CD19-positive B cells, suggesting a role for these cells in the observed delay in tumor progression.

In addition, TSH elevation was not observed in the preexisting thyroiditis cohort, indicating that high TSH is a consequence of tumor progression rather than of autoimmunity, and that it might contribute to tumor growth. Although further studies will be necessary to validate and extend these results, this report provides a novel experimental model to address the interaction between autoimmunity and thyroid cancer development and behavior.

Innovative intravital imaging approaches

The anatomical location of the thyroid gland prevents the in vivo analysis, at the single-cell level, of the interactions between tumor cells and the microenvironment. In addition, its position does not allow the study in real time of the response of tumor cells to drug treatments. Shanja-Grabarz et al. (13) have developed an innovative approach to imaging thyroid cancer in vivo, based on the implantation of a permanent, minimally invasive optical window that permits high-resolution, intravital imaging of thyroid tumors in the mouse over several days.

This technology allows visualization of the behavior and dynamics of fluorescently tagged tumor cells both in immunocompetent, genetically engineered mouse models of ATC and in immunocompromised mice carrying orthotopic implants of human or mouse ATC cells. Recipient mice in which endothelial cells and macrophages are fluorescently tagged allow the detailed analysis of the spatial and functional relationships between tumor cells and their microenvironment. Finally, Shanja-Grabarz et al. showed that the use of cells carrying a fluorescent biosensor for caspase 3 activity allows the real-time analysis of the efficacy and kinetics of action of novel small-molecule therapeutics.

Inhibition of a novel sodium/iodide symporter interactor increases RAI uptake

Correct expression and trafficking of sodium/iodide symporter (NIS) are critical to its role as a therapeutic and imaging target. Fletcher et al. (14) have used mass spectrometry to identify novel proteins involved in the regulation of NIS trafficking. Two of these proteins were analyzed in this study: ARF4, which increases the vesicular transport of NIS to the plasma membrane, and valosin-containing protein (VCP), which induces NIS unfolding and proteasomal degradation. Interestingly, VCP was found to be highly expressed in more aggressive thyroid tumors, such as PDTC and ATC, and its elevated expression was associated with reduced disease-free survival in patients treated with RAI therapy.

Two FDA-approved VCP inhibitors, ebastine and clotrimazole, increased cell surface NIS expression and RAI uptake in an NIS-positive PTC cell line as well as in mouse and human primary thyrocytes, underlining the potential clinical application of these findings. Although additional studies are warranted to determine whether these inhibitors can increase RAI uptake in vivo, in a more physiological setting, and whether such an increase is sufficient to alter the clinical course of the disease in actual patients, these data establish a novel viewpoint for NIS biology that might result in improved theranostic applications.

Targeting FOS to impair TERT expression

TERT promoter mutations are a key driver in thyroid cancer: they create novel binding sites for the GABPA/GABPB transcription factor complexes, which significantly increase TERT expression and thus its telomerase-dependent and telomerase-independent oncogenic activities. Since TERT cannot be directly inhibited, and GABPB is a direct FOS target, Liu et al. (15) used a novel FOS inhibitor, T5224, to suppress GABPB expression. As a result, TERT expression was specifically inhibited only in TERT-mutant cell lines, including PTC and ATC lines. This inhibition resulted in reduced cell viability in vitro and tumor growth in vivo.

Interestingly, TERT suppression upon FOS inhibition led to increased apoptosis, which was associated with decreased survivin and increased TNFRSF10B (TRAIL-R2/DR5) expression. Mechanistically, the authors showed that TERT directly binds to and represses the survivin promoter activity, while the mechanism leading to DR5 repression is still unclear and likely to involve intermediate players. In summary, this study provides a proof of principle that a pharmacologically improved FOS inhibitor may be a novel and important addition to the therapeutic arsenal not only for thyroid cancer but also for other TERT-mutant tumors.