The Emerging Landscapes of Long Noncoding RNA in Thyroid Carcinoma: Biological Functions and Clinical Significance

Thyroid carcinoma (TC) is one of the most prevalent primary endocrine tumors, and its incidence is steadily and gradually increasing worldwide. Accumulating evidence has revealed the critical functions of long noncoding RNAs (lncRNAs) in the tumorigenesis and development of TC. Many TC-associated lncRNAs have been documented to be implicated in TC malignant behaviors, including abnormal cell proliferation, enhanced stem cell properties and aggressiveness, and resistance to therapeutics, through interaction with proteins, DNA, or RNA or encoding small peptides. Therefore, further elucidating the lncRNA dysregulation sheds additional insights into TC tumorigenesis and progression and opens new avenues for the early diagnosis and clinical therapy of TC. In this review, we summarize the abnormal expression of lncRNA in TC and the fundamental characteristics in TC tumorigenesis and development. Additionally, we introduce the potential prognostic and therapeutic significance of lncRNAs in TC.


INTRODUCTION
Thyroid carcinoma (TC) is the most prevalent malignancy of the endocrine system and accounts for about 1% of all malignancies. The yearly incidence of TC in 2019 was 15.94 per 100,000 population (1). Currently, TC ranks the fourth most common cancer for women (2,3). Early TC-associated investigation mainly focused on the biological functions of protein-coding genes (PCGs) for their fundamental functions in the regulation of signaling transduction and various biological activities. Therefore, elucidating the functional roles and molecular mechanisms of noncoding RNAs (ncRNAs) in TC tumorigenesis is of critical significance. It has been well accepted that less than 2% of genome sequence encodes proteins, whereas the remaining 98% genome sequences do not have coding potential and their transcriptional products (ncRNAs) have not been functionally characterized. Moreover, in the past, this was once considered to be "transcriptional noise" (4)(5)(6). Based on the length of transcripts, ncRNAs can be subdivided into small ncRNAs (<200 base pairs) or long ncRNAs (>200 base pairs, lncRNAs). The small ncRNAs that play crucial roles in tumorigenesis could be further classified as miRNAs, tsRNAs, and piRNAs. LncRNAs are transcripts like mRNA in length ranging from 200 nt tõ 100 kilobases (kb) lacking significant open reading frames and have no coding potential. LncRNAs include subclasses such as pseudogenes and circRNAs (7)(8)(9). At present, human ncRNAs include more than 10,000 small ncRNA genes, approximately 14,500 pseudogenes, and almost 16,000 lncRNA genes according to the GENCODE Release 24 annotation (10). LncRNAs are characterized by extremely weak expression, worse conservation in different species, and specifically expressed in different tissues and developmental stages (11). Recently, a growing body of evidence suggests that ncRNAs, particularly lncRNA, have emerged as important regulators of gene expression during signaling transduction and diverse physiological and pathological processes through interacting with RNA, DNA, and protein and forming RNA-RNA, RNA-DNA, and RNA-protein complexes, thereby regulating gene expression via various molecular mechanisms, including modulation of transcription, mRNA stability, and translation (12,13).
The first lncRNA was identified in fetal liver tissue in 1990 (14). The following year, X chromosome inactivation mediated by one of the most famous lncRNA XIST was found (15). To date, more than 50,000 genes have been found to transcribe lncRNA (16), and this number is still steadily and rapidly increasing. Emerging evidence has been well accepted that lncRNAs are involved in gene regulation at the transcriptional or posttranscriptional level and play critical roles in signaling transduction and multiple physiological and pathological disease initiation and processes. LncRNAs were characterized by differentially expressing different organ types and cancers including TC (17)(18)(19). Currently, the functional roles of lncRNAs in TC tumorigenesis and development remain largely unknown, and only a small part of lncRNAs has been extensively studied (3,20). Interestingly, many lncRNAs were identified in body fluids, which were easily detected and analyzed, making them hold the potential to act as attractive biomarkers in liquid biopsy of TC. For example, lncRNA HOTAIR is overexpressed in the serum and is an independent prognostic marker to predict lymph node metastasis of papillary thyroid carcinoma (PTC) (21). The unique and pivotal characteristics of lncRNA reveal their attractive clinical significance in the TC diagnosis and targeted therapy. LncRNAs bear a tremendous potential to improve our understanding of the steps involved in TC tumorigenesis. In this review, we summarize the aberrant expression of lncRNA in TC and generally introduce the functional roles of lncRNA-mediated pathogenesis in TC tumorigenesis by functioning as tumor-suppressor gene or oncogene. Additionally, we discuss the prognostic, diagnostic, and therapeutic potentials of lncRNA in TC.

CLASSIFICATION OF HUMAN TC
TC includes multiple histological types, such as PTC, follicular thyroid carcinoma (FTC), Hurthle cell carcinoma, poorly differentiated thyroid carcinoma, anaplastic thyroid carcinoma, squamous cell carcinoma, and medullary thyroid carcinoma ( Figure 1). TC derived from follicular cells could be further divided into three classes based on the diverse histological and pathological elements: well-differentiated TC, poorly differentiated TC, and highly aggressively dedifferentiated TC or ATCs. Well-differentiated TC is composed of PTC and FTC subtypes. Approximately 80% of TCs are diagnosed as PTCs and 10% as FTCs in clinical trials worldwide (22)(23)(24)(25). According to surveillance epidemiology and end results (SEER) data from the USA, 5%-10% of TCs originate from calcitonin-producing parafollicular cells (also known as C cells) within the thyroid gland and are known as medullary TC (MTCs) (26)(27)(28)(29).

Papillary Thyroid Carcinoma
Having a papillary architecture and containing carcinoma cells with classic nuclear changes is the unique feature of PTCs (30). PTC incidence is considered to three to six times higher in women than in men among the regions in the world. Exposure to ionizing radiation is considered as an awfully risk element for PTC, as revealed by the higher incidence of PTC among the survivors of the Hiroshima and Nagasaki atomic bombing and among children in Belarus and Ukraine after the Chernobyl accident (31,32).

Follicular Thyroid Carcinoma
FTCs are routinely well-differentiated, unifocal, and encapsulated, but they trend toward aggressiveness and extend to blood vessels and invade systemically through the bloodstream. The incidence of FTC is significantly superior in endemic goiter regions, indicating a close correlation between FTC tumorigenesis and iodine deficiency (33)(34)(35).

Anaplastic Thyroid Carcinoma
ATCs are highly undifferentiated and singularly invasive. ATCs are uncommon carcinoma and approximately account for 2%-5% of all TCs (36). ATC cells lack the ability to seize iodine and form thyroglobulin. A large proportion of ATCs originate from pre-existing PTCs and FTCs, but the others seem to derive from de novo. The median survival time of ATC patients is only 6 months.

Medullary Thyroid Carcinoma
MTCs display well-defined carcinoma, which are characterized by lacking a well-formed capsule. High migration ability is the feature of MTCs. MTCs is frequently found to invade outside the thyroid, migrate to lymph nodes and vasculature. Nevertheless, most MTCs are sporadic, and 25% of the incidence of MTCs is hereditary. This familial disease is inherited in an autosomal dominant manner and composed of three subtypes: MEN2A, MEN2B, and familial MTC.

GENETIC LESIONS IN THYROID CARCINOMA
Like other tumor types, TC tumorigenesis and development require increasing accumulation of diverse genetic and/or epigenetic transformation. Many genetic factors are implicated in TC tumorigenesis. Genetic mutations regulating the biological functions of oncogenic or tumor-suppressive genes have been fully established in thyroid neoplasias. More than one-quarter of MTC cases are hereditary. Approximately 3.5%-6.2% of TCs are idiopathic forms of familial non-MTC (36,37). In addition, TC develops in patients with cancer syndromes derive from germline mutations (38). Fusion protein generated from rearrangements including NTRK1 has been revealed in 10% of cases of PTCs (39). 1799T>A mutation in BRAF, which results in the switch of V600E at the protein level, was observed in about 40% of PTCs and 20% of ATCs ( Figure 1) (40). As shown in Figure 1, more than 70% of PTCs contain crucial mutations in genes encoding core components of the MAPK signaling pathway, namely, the proto-oncogene RET, which is also known as TRK, RAS, and BRAF (41)(42)(43). The RET gene encodes a transmembrane receptor protein for neurotrophic growth factor binding and frequent rearrangement between the conserved domain of tyrosine kinase of RET and the 5′ region of their partner genes leading to formulate the RET/PTC oncogene in about 30% of PTCs (44).
RAS proto-oncogene mutations have been considered to be closely correlated with FTC and have been suggested in about 30% of tumors of the follicular variant of PTC (PTC-FV) (45). Genetic alteration of the RAS gene is also the primary molecular incident indicated in poorly differentiated TC (46). Human RAS genes, which can be further classified into HRAS, KRAS, and NRAS, encode highly evolutionarily conserved G proteins that reside at the intracellular and transduce signaling transmitted from receptor tyrosine kinases and the downstream signaling along the MAPK and PI3K-AKT ( Figure 2). Point mutations that activate these signaling pathways typically are codons 12, 13, and 61 of the RAS genes (47)(48)(49). In TC, codon 61 mutations in NRAS and HRAS are most the frequent. RAS mutations are commonly found in various TCs, including 10%-20% of PTCs, 40%-50% of FTCs, and 20%-40% of ATCs (50). Conversely, mutations in MAPK signaling effector genes are correlated with early stages of TC tumorigenesis. Interestingly, mutations in AKT1 and BRAF, but not in PIK3CA, have been revealed in poorly differentiated TC (51). Mutations that repress p53 effects are commonly validated in ATCs ( Figure 1) (52,53). Abnormal activation of Wnt/b-catenin (54)(55)(56) or STAT3 (55,57,58) signaling has been commonly observed in TC initiation and progression and predicts poor survival with TC patients. However, other signaling pathways involved in TC initiation and progression are poorly recognized. Thus, a better understanding of the molecular basis and signaling events associated with TC development not only elucidates the mechanism of the TC process but also offers a new window for diagnostic and prognostic prediction and a profound effect on the targeted therapy for patients with TC.
These genetic alterations open new avenues for rational principles of TC tumorigenesis and have been conventionally considered as biomarkers for early diagnosis and prognosis or therapeutic targets for patients with TC. However, increasing studies on the aberrant expression of lncRNA in thyroid neoplasias might contribute to deep elucidation of the genetic basis of TC and shed additional insights into TC diagnosis, prognosis, and therapy.

FUNCTIONAL MECHANISM OF LNCRNA
LncRNAs are defined as transcripts without coding potential longer than 200 base pairs in length. Like PCGs, lncRNA genomic regions are characterized by the transcriptional start site (TSS) of H3K4 trimethylated abundantly. LncRNA transcripts are composed of several exons that are spliced through canonical mechanisms into a mature transcript and frequently include 5′ caps and 3′ poly(A) tails. However, lncRNAs commonly have fewer exons and are expressed at very weaker levels generally compared with PCGs. Similar to PCGs, lncRNAs are transcribed by RNA polymerase II from independent promoters, but unlike mRNAs, lncRNAs preferentially localize in the nucleus (59,60), which exert diverse functions, such as regulation of gene expression in cis or in trans, regulation of splicing, and nucleation of subnuclear domain (61,62). LncRNAs exert their regulatory effects in regulating signaling transduction and cell fate determination through diverse molecular ways, including interaction with DNA, RNA, and proteins, as well as encoding small peptides (6,63). Firstly, binding to DNA enables lncRNAs to alter chromatin structure and be involved in the modulation of epigenetic modifications, thereby affecting the expression of target genes. Secondly, lncRNAs function as a molecular sponge to bind to mRNAs or miRNAs, resulting in regulating the stability and translation efficiency of mRNAs or the binding of miRNAs with their own targets.

MiRNA Sponging
LncRNA-miRNA interaction displays a common molecular mechanism involved in regulating gene expression and signaling transduction through complementary base pairing ( Figure 3A). It has been well accepted that miRNAs interact with mRNAs to mediate mRNA turnover, thereby regulating bound mRNA expression (64)(65)(66). LncRNAs can interact with both miRNAs and mRNAs, leading to RNA expression regulation being more complicated. In general, lncRNA-mRNA physical interaction affects the degradation and intracellular localization of the target mRNAs, whereas lncRNA-miRNA binding suppresses the competitive interaction of miRNAs binding to their own targets. To date, there are more than 16,000 lncRNAs in the human genome, and some of them were found to exert their functions through acting as miRNA sponge, therefore also known as competitive endogenous RNA (ceRNA) (67)(68)(69). LncRNA-miRNA interaction results in the re-expression of target genes inhibited by miRNA. For example, Lei et al. revealed that TUG1 promotes the development of TC cells via acting as a ceRNA to sponge miR-145 (70). It is worthy to note that multiple miRNA species might simultaneously bind to a single lncRNA, indicating that the interaction of lncRNAs with other RNA molecules could be more effective and multifunctional (71,72).

Protein Binding
The binding of lncRNAs with proteins plays crucial roles in affecting protein stability or subcellular localization, as well as protein complex formation or the release of proteins from their own bound partners, resulting in effecting their biological function. Apart from miRNAs, many lncRNAs have been reported to bind to proteins, thereby regulating their turnover. These binding proteins primarily cover RNA-binding proteins (RBPs), which are routinely accepted as interacting with mRNA (73,74). Due to the existence of remarkable similarities between lncRNAs and mRNAs, several RBPs have been reported to orchestrate lncRNA functional roles by physical interaction. The binding between lncRNAs and proteins affects protein intracellular localization or degradation ( Figure 3B). For example, Yuan et al. reported that SLC26A4-AS1 simultaneously binds to DDX5 and the E3 ligase TRIM25, thereby accelerating DDX5 degradation through the ubiquitinproteasome pathway, resulting in the inhibition of TC cell migration and metastasis (75). Meng   KHSRP, which decreases KHSRP expression and promotes TC cell proliferation and migration (77).

DNA Binding and Transcriptional Regulation
Nuclear lncRNAs are diffusely accepted to bind with DNA, including the noncoding or protein-coding region through complementary base pairing. This RNA-DNA interaction resulted in lncRNAs regulating target gene expression in cis (lncRNA gene sequence and target gene sequence are in the same chromosome) or in trans (lncRNA gene sequence and target gene sequence are in different chromosomes) ( Figure 3C). A fully studied DNA-interacting lncRNA is XIST, which is involved in regulating the inactivation of X chromosome during early embryonic development (15). To date, a growing body of evidence has reported that lncRNA is documented to bind to DNA. Regrettably, only a subfraction has been functionally determined. FTNRC6C-AS1 was revealed to specifically bind with the promoter of STK4, which significantly promotes STK4 methylation and results in inhibiting STK4 expression in TC (78)

Encoding Small Peptides
LncRNAs are defined as long transcripts without coding potential. Interestingly, shreds of evidence represented that diverse lncRNAs contain small open reading frames (sORFs) indeed and execute their functions by coding functional peptides (10, 83, 84) ( Figure 3D). These novel findings broaden the complexity of lncRNA function. The first lncRNA that was identified to have coding potential is ENOD40, which is derived from plants and was found to encode two small peptides that interact with sucrose synthase (85). Subsequently, Rohrig et al. further revealed that the peptide encoded by ENOD40 binds sucrose synthase and elevates the sucrose cleavage activity of sucrose synthase (86). Currently, no lncRNAs that translate into small peptides were identified in TC tumorigenesis and development. The lack of related research may be attributed to several causes. Firstly, the small peptides encoded by lncRNAs are too insecure to monitor. Secondly, a fair portion of these small peptides has no biological function. Therefore, a better understanding is needed to fully recognize the biological function of small peptides encoded by lncRNA in TC initiation and development.

THE BIOLOGICAL FUNCTION OF LNCRNAS IN TC HALLMARKS
Different types of cancer share similar hallmarks (87,88). The crucial roles of lncRNAs in TC initiation and development are evident by the fact that lncRNAs are closely related to TC malignant phenotypes, including sustaining uncontrolled cell growth, having resistance to cellular apoptosis, enhancing invasive ability, and acquiring cancer stem cell-like features. TC-associated lncRNAs facilitate or suppress these malignant behaviors by interacting with their binding macromolecules, indicating rational therapeutic treatment for the therapy of TC.

Maintaining Proliferative Signaling
Tumor cell uncontrolled proliferation is activated by sustaining proliferative signaling (87,88). Normal cells restrict cell growth to precisely regulate an appropriate architecture of organisms, whereas transforming cells lead to tumorigenesis by constitutively activating RTK signaling or bypassing cell cycle checkpoints. For example, excessive activation of RTKs has been documented to play essential roles in maintaining TC growth ( Figure 2

Therapeutic Resistance
To date, the therapeutic strategies routinely applied for TC clinical therapy are surgery, radiotherapy, and chemotherapy (109). However, TC cells can gradually develop resistance to therapeutics, particularly chemotherapy, during clinical treatment which can diffusely be attributed to but not limited to the following: 1) considerably enhanced certain transport protein expression that reduces the intracellular concentration of therapeutic drugs, resulting in decreasing chemotherapy efficacy; 2) genetic or epigenetic alterations that enable TC cells to become resistant to cellular death; and 3) augmentation in DNA repair ability (110,111 (Figure 4B). Xia et al. proved that CCND2-AS1 was significantly highly expressed in PTC and targeting CCND2-AS1 suppressed cell proliferation, migration, and invasion (116). Esposito et al. proved that silencing COMET significantly potentiated the sensitivity of TC cells to vemurafenib, which is an ordinary inhibitor of mutated Braf (117).

Activating Migration and Invasion
Tumor metastasis is the real cause of the vast majority of cancerrelated deaths (118)(119)(120). In brief, the nature of metastasis of TC shares remarkable similarities with other cancer types, whereas epithelial-mesenchymal transition program and increased angiogenesis capability have been considered as critical incidents driving TC metastasis. Many lncRNAs are well established to implicate in the regulation of TC cell invasion and migration, leading to affect TC metastasis. beta1 (TGF-b1), was involved in facilitating PTC cell migration and invasion (123). Shen et al. characterized that PROX1-AS1 was overexpressed in PTC and played an important role in PTC proliferation, invasion, and migration (124). NEAT1 was found to be overexpressed in both ATC tissues and cells and was induced in the hypoxia niche. NEAT1 depletion inhibits ATC cell invasion and glycolysis by exerting as a ceRNA sponge of miR-206 and miR-599 under hypoxia (125). HOTAIR was found to promote PTC cell proliferation and invasion via implicating the miR-488-5p/NUP205 signaling axis (21,126). RGMB-AS1 is transcriptionally promoted by E2F1 and activated PTC cell proliferation and invasion (127) ( Figure 4C). CCAT1 has been reported to promote PTC cell proliferation and migration through inhibiting miR-143 expression (128).

Sustaining Stem Cell-Like Characteristics
Cancer stem cells (CSCs) are a small subpopulation of cells that reside in tumors, which are well investigated being accounted for tumor initiation, therapeutic resistance, and tumor relapse (129)(130)(131). Emerging evidence suggested that Wnt/b-catenin, JAK/STAT, and Hippo/YAP have emerged as crucial players in affecting the stemness of TC CSCs (129,132). A growing body of evidence validated that lncRNAs participate in CSC formation and maintenance in human cancers, including TC (133,134). In the past several decades, the biological function of proteincoding genes in TC tumorigenesis and CSC self-renewal maintenance was well established, whereas the roles of lncRNAs have not been fully elucidated. Several pieces of evidence have demonstrated that lncRNA plays a crucial role in sustaining TC CSC self-renewal. For example, Dai et al. proved that lncRNA DOCK9-AS2 interacted with SP1 and activated CTNNB1 expression at the transcriptional level promoting PTC CSC maintenance (135) ( Figure 4D). LncRNA-H19 accelerates cancer stem-like properties in PTC by activating estrogen receptor b expression (136). PTCSC3 was found to play critical roles in inhibiting PTC stem cell selfrenewal (137). In addition, another group further validated that PTCSC3 suppresses PTC cell proliferation and migration by sponging miR-574-5p and represses b-catenin signaling (138).

CLINICAL RELEVANCE OF LNCRNA IN TC
Several lncRNAs were identified to be dysregulated in TC and lncRNAs have been proved to hold the potential to function as biomarkers in TC diagnosis and prognosis. Table 1 summarizes that lncRNAs are dysregulated in TC and have oncogenic and tumor-suppressive roles in TC tumorigenesis and development, given the various regulatory molecular mechanisms and the diverse downstream signaling cascades affected. It is not surprising that several lncRNAs, including many of those emphasized above, have been observed through in vivo experiments to be essential players in TC tumorigenesis and progression and display attractive targeted therapeutics ( Table 1). Given the poor conservation in different species, most of the lncRNA-related studies relied on regulating

LncRNA as Cancer Biomarkers
Currently, TC diagnosis mainly depends on ultrasonography imaging. In addition, very limited therapeutic alternatives have been applied to TC-related deaths for lacking prognostic biomarkers or therapeutic targets. Thus, therapeutic targets and novel biomarkers are urgently required for developing the clinical therapies of TC. LncRNA can be easily detected from the serum and urine, and many lncRNAs have been reported to secrete from tumor cells into the circulatory system (144).
Recent reports have documented that many lncRNAs are stable enough to be detected in the serum or urine of patients with cancer (53). Therefore, examination of the existence of circulating lncRNAs in serum or urine can be employed as diagnostic biomarkers for thyroid cancer diagnosis and prognosis in a noninvasive manner. LncRNA is a useful diagnostic biomarker because dysregulated lncRNA expression is observed across diverse cancers including TC and can be easily detected. Ge

CONCLUSIONS AND PERSPECTIVES
Numerous studies have demonstrated that lncRNAs can function as regulatory molecules involved in a variety of biological processes. This review highlights the differentially expressed lncRNAs in TC initiation and progression, showing the great potential of lncRNAs as a regulatory molecule, as well as its diagnostic, prognostic, and therapeutic potential in TC studies. LncRNAs have been reported to be involved in various complex functions and molecular mechanisms; however, only a small fraction of lncRNAs and their functions in TC initiation and development have been investigated. Significant improvement has been achieved in uncovering lncRNAs and in understanding their functional roles and molecular mechanism in TC tumorigenesis and development. However, several fundamental issues remain to be addressed so as to obtain a deeper understanding of the biological function and clinical significance of lncRNAs in TC tumorigenesis and development. The functions of the few known lncRNAs still need to be characterized in detail and additional molecular mechanisms of lncRNAs need to be defined. Since many lncRNAs exert diverse functions in affecting multiple cellular processes in many tissues, the elaborate functional principles of lncRNA biological functions require to be understood on a case-by-case basis. In addition, it is indispensable to define whether the sensitivity and reliability of lncRNAs are enough for their substantial clinical application as biomarkers. Whether lncRNAs have advantages compared with other biomarkers used for TC diagnosis needs to be further elucidated. We are optimistic that the application of the new sequencing strategies will decisively determine the functional mechanism of lncRNAs implicated in TC tumorigenesis and development, which will eventually expedite the clinical application of lncRNAs for their use in early diagnosis, clinical therapy, and prognosis evaluation. The vast majority of PTC patients can be managed effectively; however, mortality associated with advanced and iodine-refractory TC remains high. Meanwhile, ATC displays an extraordinary worse prognosis. Whether lncRNAs hold great potential in better understanding the pathogenesis or serve as biomarkers for these advanced TC early diagnosis and effective treatment still needs to be ulteriorly investigated. Epigenetic regulation of lncRNAs bears an exceedingly superb potential in mediating TC initiation and development and should focus on investigating not only through achieving a better understanding of TC but also acquiring therapeutic successes. Overall, a growing body of evidence demonstrates that specifically targeting lncRNAs might be a novel strategy for the clinical therapy of patients with TC. However, whether lncRNAs could be applied in deeper understanding of the pathogenesis or if it can be potentially used for the biomarkers for TC patients' early diagnosis and advanced research is needed to further characterize. Additionally, it is of significance to decipher the biological functions of supernumerary lncRNAs correlated with TC tumorigenesis and translate them into clinical biomarkers applied for TC early diagnosis and therapeutic target.

AUTHOR CONTRIBUTIONS
QH wrote the manuscript. PL searched the literature and made the figures and tables. JZ was the major contributor in preparing and writing the manuscript. CL, DW, XC, SW, YL, and JW provided help in editing and writing the manuscript. All authors contributed to the article and approved the submitted version.