1. Introduction
Papillary thyroid carcinoma (PTC) is the most common type of thyroid cancer (TC), generally characterized by a good prognosis. Only a small percentage of PTCs demonstrates a higher aggressiveness and poor outcomes [
1]. A proper stratification is crucial to select a group of patients at high risk of unfavorable PTC course and simultaneously to avoid overtreatment in low-risk cases. Moreover, the intra-tumor and interpatient heterogeneity observed in PTC, seem to have important implications for therapeutic approaches [
2,
3]. Despite significant progress in our understanding of the molecular background of PTCs, reliable prognostic and predictive molecular markers are still lacking. The Cancer Genome Atlas (TCGA) study, published in 2014, substantially reduced the ‘dark matter’ of PTC genome up to 3.5%, as genetic alterations were defined in about 96.5% PTC cases [
4]. The most common ones, as reported previously by other authors, were related to genes of the RAS/RAF/MAPK pathway, with the
BRAF V600E mutation being present in 61.7% of PTCs,
RAS mutations in 12.9%, and
RET/PTC rearrangements in 6.8% of cases. The mutation density in the analyzed cohort was low (0.41 nonsynonymous mutations per Mb, on average). It positively correlated with age, recurrence risk, and the distant Metastasis, patient Age, Completeness of resection, local Invasion, and tumor Size (MACIS) score. These results may reflect an indolent nature of the majority PTC cases on the one hand, and on the other hand, represent the molecular basis of aggressive PTC. Until recently, the
BRAF V600E mutation was considered a marker of poor prognosis in PTC. Many studies were pointing to a possible relationship between its presence and the worse outcome [
5,
6]. Nevertheless, its role as a prognostic or predictive factor is still controversial [
7]. The frequency of the
BRAF V600E mutation ranges from 23 to 83% in different PTC cohorts. Its presence was demonstrated in both classic PTCs as well as in its aggressive subtypes [
8]. According to the current ATA guidelines for diagnostics and treatment of differentiated thyroid carcinoma (DTC), the
BRAF V600E mutation may be considered in risk stratification, however, in the context of clinical data only [
9].
In 2013 two somatic mutations of telomerase reverse transcriptase (
TERT) gene promoter were discovered in different solid cancers, including TCs [
10,
11,
12]. Two alterations, c.-124C > T (C228T) and c.-146C > T (C250T) were reported in PTC with the prevalence of 11.3% [
13]. Their frequency was higher in more aggressive histopathological PTC variants [
10].
TERT encodes the catalytic subunit of telomerase, responsible for telomeres maintenance at the end of chromosomes. Telomerase, typically not expressed in somatic cells, is reactivated in cancer cells. It protects telomeres from further shortening and is also involved in cancer cell proliferation, resistance to apoptosis and antigrowth signals, increased angiogenesis, and metastatic potential [
14]. It has been demonstrated that C228T and C250T
TERT promoter (
TERTp) mutations were much more prevalent in PTCs harboring
BRAF V600E mutation. This co-occurrence was related to poor prognostic factors, including tumor size, older age at diagnosis, distant metastases, or shorter progression-free survival [
13,
15,
16]. Also, the co-existence of
RAS and
TERTp mutations was associated with a worse prognosis. Shen et al. proposed a six-genotype genetic prognostic model for PTC, dividing patients into the following risk groups regarding the detected mutation:
BRAF V600E/
RAS mutation and
TERTp mutation >>>>
BRAF V600E =
TERTp mutation alone >
RAS mutation alone = wild-type genes [
17].
Since the discovery of
TERTp mutations in thyroid cancers, most studies have focused on their clinical implications. However, the data concerning their molecular consequences and potential differences/similarities to
BRAF V600E-positive TCs are scarce. To our best knowledge, there was only one published study devoted to the transcriptomic analysis of
TERTp-mutated PTCs [
18]. Therefore, in the present study, we focused on the gene expression profile of
TERTp-mutated PTC in BRAF(+) cases. We aimed to analyze the molecular consequences of
TERTp alterations that could explain their impact on PTC outcomes. As our PTC cohort involved mainly patients showing unfavorable clinical course, we tried to find molecular mechanisms, responsible for PTC aggressiveness.
3. Discussion
The presence of the
TERTp mutation resulted in a changed expression of four genes only. It is the most important result of our analysis of the PTC gene expression profile. Moreover, only one of these genes was confirmed in an independent set of PTC samples made available by the TCGA [
4] consortium. However, one should notice that the TCGA cohort consisted mainly of low-risk PTCs.
One of the hallmarks of malignant neoplasms is an evasion of replicative senescence and out-of-control proliferation, leading to immortalization of cancer cells [
19]. The pivotal mechanism underlying this process is the reactivation of telomerase, an enzyme typically not expressed in somatic cells, but activated in about 80–90% of all malignant neoplasms [
20]. Telomerase is a complex protein. However, its core is composed only of catalytic component TERT and internal telomerase RNA template (TERC) [
21]. TERC is ubiquitously expressed in various human cells. On the contrary, TERT is repressed in somatic cells leading to telomerase silencing [
22]. Its induction/telomerase activation not only provides telomeres stabilization but is also related to several oncogenic processes, independent of telomeres lengthening. For instance, it has been shown that TERT directly interacts with β-catenin and, as a consequence, stimulates epithelial-mesenchymal transition (EMT) and stemness of cancer cells, and, by interaction with NF-κB p65, up-regulates the expression of metalloproteinases (MMPs) in cancer cells [
23,
24]. Moreover, TERT contributes to survival signaling, growth signaling, invasion/metastasis, angiogenesis, DNA methylation, genetic aberrations, and even to radio/chemo-resistance, which make TERT an important factor related to a higher aggressiveness of cancer cells [
22]. Also, in thyroid cancer, TERT induction has been linked to a poorer prognosis, higher risk of metastases, recurrence, and even death [
13]. There are several mechanisms underlying
TERT activation with
TERTp somatic mutations being the most widely studied in many cancer types since their discovery in 2013 [
11,
12]. In TCs, their frequency is considered as intermediate. However, it increases dramatically from microcarcinomas (frequency reported at the level of about 5%) [
25] to aggressive poorly differentiated and anaplastic TCs (48.8% and 41.8%, respectively) [
20]. It seems to be more common in TCs harboring
BRAF V600E mutation [
26]. Although a lot of data concerning the clinical significance of
TERTp mutations in TCs has been published, there is still much to discover in the field of their molecular consequences. To our best knowledge, this is the first study that analyses the impact of
TERTp mutations on PTC transcriptome in BRAF(+) samples.
Our previous study showed that in Polish PTC patients,
TERTp mutations are present in 8.5% of cases [
16]. However, in the current analysis, in which almost half of the PTC cases presented distant metastases, the frequency of C228T and C250T mutations increased up to 14.8%. Of 30 non-metastatic cases, two PTCs were TERTp(+) (6.7%), whereas six out of 24 PTCs with distant metastases harbored a
TERTp mutation (25%). Similar data obtained Gandolfi et al., who showed the presence of these alterations in 10% of non-metastatic PTCs and 33% of PTCs with distant metastases [
27]. All PTCs harboring
TERTp mutation in our study were
BRAF V600E-positive. Moreover, in one PTC with
BRAF mutation,
NRAS point mutation was reported. However, none of the studied cases displayed the co-existence of
TERTp and
RAS somatic mutations.
In the current study, we focused mainly on genes and processes that are significantly altered in the presence of
TERTp mutations. As indicated in previous studies,
TERTp mutations were associated with elevated
TERT expression [
28] and, as presented by Fredriksson et al., this association was exceptional in its strength and was highest in copy number-stable cancers such as thyroid carcinoma [
29]. Our results are in agreement with these data. We observed significant up-regulation of
TERT transcript in TERTp(+) PTCs comparing to TERTp(−) ones harboring
BRAF V600E mutation and NM-PTCs. Although there was no significant difference in
TERT expression between TERTp(+) and RAS(+) PTCs, a tendency of
TERTp mutations-dependent up-regulation of
TERT mRNA was visible. However, a limited number of RAS(+) PTCs might impair the results and statistics.
The next step was a closer look at the whole gene expression profiles of our PTC set with a particular focus on cases with
TERTp mutation. There was no surprise that all PTCs harboring
BRAF mutation grouped close to each other in an unsupervised PCA analysis. It was the most potent differentiating factor within the first and second components. RAS(+) PTCs grouped with NM-PTC cases. PCA did not show differences between BRAF(+)TERTp(+) and BRAF(+)TERTp(−) samples. Moreover, according to the BRS, BRAF(+) PTCs, including all TERTp(+) cases, were BRAF-like and most RAS(+) and NM-PTCs were RAS-like. However, one RAS(+) and several NM-PTCs were more similar to BRAF-like PTCs. It might result from the presence of some additional alterations, and, in consequence, their gene expression profile was more similar to BRAF(+) PTCs. Some of the NM-PTCs may harbor
RET fusions. In the TCGA study,
RET rearrangements were present in 6.8% PTCs (33/484), and nearly all of them were weakly BRAF-like. However, we focused mainly on
TERTp mutations and genetic alterations that most commonly co-exist with them. The presence of
TERTp mutations, in turn, may impair the expression of thyroid-specific genes by down-regulating them. Although there were no differences in TDS among BRAF(+) PTCs with and without
TERTp mutations, this association was significant in comparison of BRAF(+)TERTp(+) cases to RAS(+) and NM-PTCs. These data suggest that
TERTp mutations may be crucial in PTC dedifferentiation and aggressiveness. It is supported by the association of
TERTp mutations with poor prognostic factors, observed in our PTC cohort. Larger tumor size, invasion of the surrounding tissues, the locally persistent disease after the first surgery, pN1b, and distant metastases were significantly associated with mutated
TERTp, which is in concordance with previous studies [
13].
Taking into consideration significant diversity of BRAF-like PTCs, noticed in TCGA study [
4] and a higher frequency of
TERTp mutations in PTCs harboring
BRAF V600E alteration [
30,
31], we focused on possible gene expression differences within BRAF(+) PTCs resulting from the presence of
TERTp mutations. The observed difference was subtle. However, we did not expect any differences since all TERTp(+) cases were simultaneously BRAF(+). We are aware that not all BRAF(+) PTCs show poor outcomes, so there must be some additional factors responsible for disease aggressiveness.
TERTp mutations certainly participate in this process. We obtained four genes differentiating BRAF(+)TERTp(+) PTCs and BRAF(+)TERTp(−) ones (
CRABP2,
ECM1, KRT17, and
MTMR3). One out of these four genes, the
CRABP2 gene, was positively validated on an independent PTC cohort, although with a lower number of high-risk cases. This result confirms that the TERTp(+)-dependent difference exists. The
CRABP2 gene encodes the cellular retinoic acid-binding protein 2 that is responsible for retinoic acid (RA) transport to retinoic acid receptors (RARs) within the nucleus, inhibiting cell growth and proliferation [
32]. That is why
CRABP2 is considered as an antitumor agent. However, there are contradicting data suggesting the necessity of further studies on the
CRABP2 role in tumorigenesis. For instance, CRABP2 protein was identified as a subtype-specific biomarker of ovarian cancer, since its expression positively correlated with tumor grade and cancer stage [
33]. Similarly, higher CRABP2 expression corresponded to invasive retinoblastomas [
34] and poorer prognosis in estrogen receptor-negative breast cancer [
35]. Moreover, its overexpression was suggested to be a late event of pancreatic carcinogenesis that could be used as a marker to distinguish pancreatic ductal adenocarcinomas from other benign pancreatic conditions [
36]. Kim et al. proposed plasma CRABP2 as a novel diagnostic and prognostic marker in non-small cell lung cancer [
37]. In our study, the expression of
CRABP2 was three times higher in the presence of
TERTp mutation (in TCGA data FC was 2.2). Together with previous studies, it may support the hypothesis that an elevated
CRABP2 expression is a late event, characteristic for more advanced and aggressive tumors. These data do not confirm the recognized main role of CRABP2 as an antitumor agent. An alternative mechanism of CRABP2 activity has been proposed. It has been proven that CRABP2 mediated proliferative activity not by interacting with RAR, but through PPARbeta/delta receptor in the presence of fatty acid binding protein 5 [
32].
Two genes,
KRT17 and
ECM1, which expression was elevated in our TERTp(+) PTCs, demonstrated higher expression in a variety of cancer types. Keratins are components of the cytoskeleton and play a major role in cell protection and structural support. Keratin KRT17, belonging to type I keratin, was regenerated and highly expressed in many cancers [
38], including gastric cancer [
39], cervical cancer [
40], oral squamous cell carcinoma [
41], and breast cancer [
42]. Moreover, KRT17 expression was proposed as a prognostic marker that can discriminate postoperative stage II patients with colorectal cancer with a high probability of disease recurrence, as support to available risk stratification factors [
43]. Extracellular matrix protein 1 (ECM1), in turn, is a glycoprotein involved in cell proliferation, angiogenesis, migration, and metastases. Its elevated expression was observed in several malignancies, including thyroid, gastric, colorectal, lung carcinoma, invasive ductal breast carcinomas, hepatocellular cancer, and others [
44,
45]. A significantly high increase in
ECM1 expression was observed in malignant epithelial tumors, especially in these tumors with distant metastases [
44]. This observation would explain higher
ECM1 expression in our TERTp(+) PTCs when compared to BRAF(+)TERTp(-) and RAS(+) PTCs. However,
ECM1 occurred to be down-regulated in mutated PTCs in comparison to NM-PTCs. So, its role as a potential marker of TERTp(+) PTCs seems questionable.
Regarding the 4-gene signature of TERTp(+) PTCs in our study, the
MTMR3 gene (encodes myotubularin-related protein 3) was the only down-regulated gene in TERTp(+) PTCs comparing to TERTp(−) ones. MTMR3 belongs to phosphoinositide 3-phosphatases with affinity to Ptdlns5P [
46]. It may impair or enhance tumor growth in different types of cancer. For example, the lack of MTMR3 was shown to repress the proliferative and invasive potential of oral cancer cells [
47]. Its exogenous expression inhibited the growth of transfected lung carcinoma cells [
48]. On the contrary, its higher expression had negative effects on overall survival and relapse-free survival in patients with breast cancer [
49]. Nevertheless, the role of
MTMR3 in TERTp(+) PTCs requires further studies.
Our signaling pathways analysis revealed 39 enriched KEGG pathways differentially changed between BRAF(+)TERTp(+) and BRAF(+)TERTp(−) PTCs. The majority of pathways with the highest absolute NES (normalized enrichment score) value were enriched by down-regulated genes, among them Inositol phosphate metabolism, Phosphatidylinositol signaling system, and Ubiquitin mediated proteolysis pathways. The critical feature of cancer cells differentiating them from healthy ones is related to the reprogramming of fundamental pathways determining distinct processes, including proliferation, differentiation, and motility. The Ubiquitin pathway plays an essential role in the regulation of cell growth and cell proliferation through the control of key cell cycle proteins. Because of its involvement in crucial biochemical processes, it is a potential target for cancer-related deregulation. Impaired proteolysis of cell cycle regulators was reported in many human cancers as being contributed to tumorigenesis [
50,
51]. Phosphatidylinositol signaling system is also known as associated with cancer, mainly by its cooperation with the PI3K-Akt pathway that mediates cell proliferation, survival, and metabolism. Mainly, PI3K and PTEN play a key role in cancer, but also other members of this pathway seem to be implicated in the progression of tumors [
52].
Among pathways with the highest NES value, enriched by up-regulated genes, we found the Neuroactive ligand-receptor interaction pathway, which includes receptors and ligands associated with intracellular and extracellular signaling pathways involved in the progression of the bladder, renal cell, and prostate cancer [
53,
54,
55]. Up-regulated genes also enriched the Olfactory transduction pathway. Olfactory receptors (OR), playing a crucial role in healthy tissues, were also involved in multiple pathological processes, including hepatocarcinoma, non-small cell lung cancer, colorectal cancer, melanoma, and breast cancer [
56]. Some of these ORs were proposed as markers in different cancer types to discriminate between cancer and healthy tissues. Moreover, comparing BRAF(+)TERTp(+) and BRAF(+)TERTp(−) PTCs, we found six significantly enriched gene groups related to telomeres. All of these gene groups had a negative NES value (i.e., they are enriched in genes downregulated in BRAF(+)TERTp(+) cases), which may suggest that the presence of
TERTp mutations impairs processes involved in telomere maintenance.
Difference in gene expression profile found in our BRAF(+) PTCs as dependent on the mutated TERT promoter was subtle. However, positive verification of one out of four differentiating genes on an independent PTC cohort allows to hypothesize that obtained data will still be valid on a much larger tumor set. Especially since our and validation PTC sets differed in two important features- number of high-risk PTCs, that was incomparably larger in our cohort, and presence of TERTp mutations only as co-existing with BRAF V600E mutation in our PTC set. Despite these differences expression of CRABP2 gene was significantly up-regulated in TERTp(+) cases in both cohorts.
So far, there was only one published study that analyzed the impact of the presence of
TERTp mutations on PTC transcriptome [
18]. Chien et al., who used the data obtained from the TCGA study [
4], found
TERTp mutations to be associated with proliferative and metabolic alterations in PTC. Pathways related to DNA damage response and cell cycle regulation were enriched among up-regulated genes, whereas transporter and metabolic activities were overrepresented among down-regulated genes [
18]. Although they analyzed a much larger group, they found no association between
TERTp mutations and
BRAF genetic alterations. No difference regarding thyroid differentiation genes was observed. Our PTC cohort was smaller. All TERTp(+) PTCs were BRAF(+). The main limitation of this study is related to the coexistence of the
BRAF mutation in all TERTp(+) PTC samples. However, the presence of
TERTp mutations, without co-existing
BRAF or
RAS genetic alterations is rare in PTC. In our previous study, 3 out of 189 PTCs (1.6%) were positive only for
TERTp mutation (no
BRAF and no
RAS mutations detected). We found significantly lower TDS value in TERTp(+) PTCs comparing to RAS(+) and NM-PTCs. However, the limited number of samples and the presence of other than
TERTp mutations in our analysis impaired the obtained results. In addition, because of the selection bias, almost half of our PTC patients had distant metastases. It increased the number of TERTp(+) PTCs in our cohort. It is also the crucial difference between our and TCGA PTC cohorts since the latter one included mainly low-risk PTCs (only 8 PTCs had distant metastases) [
4]. Moreover, the use of a larger set of high-risk PTCs confined our cohort to cases with co-existing
BRAF and
TERTp mutations and obligated us to search for differences mainly within the BRAF(+) PTC set. On the one hand, it is a limitation of our analysis. On the other hand, it is important to find mechanisms responsible for PTC aggressiveness, since the
BRAF V600E mutation as a prognostic marker in PTC is still controversial and is considered in treatment or risk stratification only regarding clinical features.
Despite mentioned limitations of the study, there are clear differences in gene expression profile between BRAF(+) PTC tumors carrying TERTp mutation and PTC tumors with only BRAF mutation. This suggests potential role of TERTp mutations in down-regulation of thyroid specific genes. Additionally, we confirmed that mutated TERT promoter is associated with poor prognosis of PTC and it might have a potential value as a prognostic marker.