Long non-coding RNA implicated in the invasion and metastasis of head and neck cancer: possible function and mechanisms

Head and neck cancer (HNC), which ranks sixth among the frequent malignant neoplasms worldwide, has an estimate of over 500,000 new cases detected annually [1, 2]. The broad definition of HNC includes not only the mucosal epithelial carcinomas of the head and neck area, presenting as and head and neck squamous cell carcinoma (HNSCC) along with nasopharyngeal carcinoma (NPC), but also thyroid carcinoma [3]. HNSCC, serving as the main subset of HNC, predominantly includes squamous cell carcinoma involving the oral cavity, pharynx and larynx [3]. Despite advances in surgery and chemoradiotherapy, the outcome improvement remains modest, with a 5-year survival rate lower than 50% [4]. Tumor invasiveness and metastasis, typical hallmarks of HNC, are largely responsible for the poor response to treatments [4, 5].

Metastasis is a common characteristic of cancer progression with multiple sequential steps, presenting as enhancement of the invasive ability of tumor cells and its spread to secondary sites of the body [6]. Epithelial-mesenchymal transition (EMT) represents a transcriptional program underpinning invasive and metastatic phenotype of cancers, which could be interpreted as a state of de-differentiation. Thus, EMT endows cancer cells with a high-grade phenotype to drive their migration, invasion and metastasis [7, 8]. Next, a subsequent step, the mesenchymal-epithelial transition, enables these spreading cells to colonize at a second location [9].

Long noncoding RNAs (lncRNAs) are one subtype of RNA transcripts which contain more than 200 nucleotides, lacking in capability of encoding protein or exhibiting limited potential [10]. LncRNA was previously considered as a genetic byproduct because of the absence of biological function [11]. Based on the upgraded DNA sequencing technologies, although the whole human genome is generally transcribed, more than 98% are non-encoding genes. Thus, it causes our conceptual shift regarding the possible role of the non-coding genes [12]. Besides, the importance of lncRNA in cancer biology is becoming appreciated by scholars. In recent years, based on high-throughput sequencing and biological techniques, increasing numbers of lncRNAs are being uncovered and their critical roles in regulating cancer development and progression are being extensively investigated. To date, the dysregulated expression and involvement of lncRNAs have been reported in diverse cancers, including HNC [13], lung cancer [14], breast cancer [15], colorectal cancer (CRC) [16], and esophageal squamous cell carcinoma (ESCC) [17]. Among these studies, lncRNAs seem to be implicated in each process of cancer progression, such as tumor development and metastasis. During metastasis, various lncRNAs are reported to function similarly or differently via diverse molecular mechanisms. Metastasis Associated Lung Adenocarcinoma Transcript 1 (MALAT1), as one lncRNA, was first detected and highly expressed within non-small cell lung cancer (NSCLC). Guo et al. demonstrated that MALAT1 exacerbates cell migration and invasion of NSCLC via binding to its downstream C-X-C motif chemokine ligand 5 (CXCL5). Additionally, the low methylated forms of the MALAT1 promoter in NSCLC accounts for its high expression [14]. Furthermore, homeobox transcript antisense RNA (HOTAIR) was indicated to potentiate the metastatic ability of colon cancer by inducing EMT [18]. In addition, increasing evidence has demonstrated that many lncRNAs could exert their biological function by working as pairs together with their adjacent mRNAs. Pan et al. indicated that the lncRNA Fork head box C1 upstream transcript (FOXCUT) could function together with Fork head box C1 (FOXC1) to potentiate the invasive and migrating capability of ESCC [17]. More importantly, the function and molecular mechanisms of these lncRNAs—i.e. MALAT1 [2], HOTAIR [19] and FOXCUT [20]—have been widely studied regarding their correlation with HNC metastasis. Given the important regulatory role of lncRNA in HNC metastasis, it is promising to be exploited as a prognosticator and therapeutic target for HNC.

In the present paper, we review these aberrantly expressing lncRNAs in the mediation of EMT, migration, invasion and metastasis of HNC (Table 1), and provide insight into their regulatory mechanism.

In general, lncRNAs perform their function based on their functional domains in the secondary or tertiary structure, and these mature structures originate from alternative splicing. Specifically, the domains facilitate the interaction of lncRNAs with chromatin, RNA and proteins; thus lncRNAs could function in HNC metastasis via chromatin remodeling, transcriptional control and post-transcriptional regulation [21, 22]. In general, lncRNAs could exist in the nucleus or cytoplasm, and different subcellular locations might determine the various biological functions of lncRNAs (Fig. 1) [23]. Therefore, lncRNAs are categorized into nuclear or cytoplasmic lncRNAs.

Several studies have suggested that most lncRNAs in the nucleus could function as a guide to lead chromatin modulating complexes into certain genomic loci and then initiate chromatin modification to activate or silence gene expression [24]. Additionally, nuclear lncRNAs could regulate gene transcription by their assembly with transcription factors into complexes. Also, they could be involved in the mRNA/miRNA processing; despite the absence of examples in HNC, some lncRNAs have been suggested to be implicated in mRNA/miRNA processing among other cancers [25, 26]. For instance, lncRNA colon cancer-associated transcript-2 (CCAT2), elevated in CRC, might repress the processing and maturation process of miR-145 in the nucleus, thereby restraining the proliferation and differentiation of CRC stem cells [25]; Additionally, as another lncRNA implied to interact directly with the serine/arginine splicing factor 1 in HeLa cells and influence the distribution and phosphorylation of the splicing factor, MALAT1 could potentially modulate the alternative splicing of pre-mRNAs, thus being involved in mRNA processing [26].

Regarding the great number of cytoplasmic lncRNAs, they typically modulate gene expression by base pairing with specific genes, improving or attenuating the mRNA stability and acting as miRNA sponges; in addition, they could influence mRNA translation by binding to different elements. Moreover, cytoplasmic lncRNAs are also responsible for protein stability control [24, 27].

Based on microarray analysis, NONHSAT037832 was detected as a novel lncRNA remarkably downregulated in PTC, next implied to function as an inhibitor in the LNM of PTC [82]. Besides, another study indicated that the expression level of lncRNA Growth Arrest-specific Transcript 5 (GAS5) was negatively related to the LNM of thyroid carcinoma [83]. Moreover, further studies are desired to determine their detailed mechanisms in attenuating tumor metastasis.

The tumor microenvironment (TME) of HNC is characterized by immunosuppression [111]. CD8+ T cells among the tumor-infiltrating lymphocytes (TILs) express significantly higher levels of immune checkpoint receptors such as programmed cell death 1 (PD-1) than those in the peripheral blood [112, 113]. Interestingly, lncRNAs can directly regulate the expression levels of the checkpoint receptors and their ligands (Fig. 2a) [114]. For example, the lncRNA AFAP1-AS1 upregulates PD-1 expression levels in the TME of NPC, possibly leading to T-cell exhaustion. [114]. TME exhibits strong M2-like skewing, which dampens antigen-presenting cell function and subsequent tumor-specific effector activation. Some TIL subsets such as mast cells appear to contribute to immunosuppression and enhance cancer invasion. Notably, mast cells could directly increase the expression of HOTAIR and lead the HOTAIR-PRC2 complexes to suppress the expression of androgen receptor, thereby contributing to tumor invasion (Fig. 2b) [115]. As another lncRNA exacerbating the M2-like phenotype in the TME, tumor-associated macrophages could potentiate the invasion of breast cancer in vitro by upregulating UCA1 (Fig. 2c) [116]. By contrast, lncRNA advanced glycosylation end-product specific receptor (lncAGER) could strengthen the effect of human monocytes against lung cancer, thereby inhibiting tumor migration and growth (Fig. 2d). Although downregulated in lung cancer due to the hypermethylation of its promoter, lncAGER exhibits the anti-tumor function by first targeting miR-185, thereby increasing the advanced glycosylation end-product specific receptor (AGER) level in lung cancer cells, which is an important innate immune pattern-recognition receptor, and ultimately promoting the anti-tumor effect of human monocytes [117]. Overall, how lncRNAs in cancer cells regulate the immune microenvironment of HNC remains insufficiently characterized. However, emerging evidence suggests that targeting a subset of lncRNAs not only alleviates the cancer invasion phenotype but also potentially contributes to improved immune detection of cancer.

Based on previous findings, there are three main mechanisms that are responsible for the aberrant expression of lncRNA in HNC, which are miRNAs, functional proteins such as RNA binding proteins (RBPs), NF-κB and TP53, and genetic changes such as genomic mutation and epigenetic alteration [47, 81, 118‐125]. Primarily, there are two miRNAs that modulate lncRNA expression in HNC. Specifically, lncRNA papillary thyroid carcinoma susceptibility candidate 3 (PTCSC3) could be remarkably decreased by the overexpression of miR-574-5p in thyroid carcinoma [118]; additionally, in TSCC, miR-26a could upregulate the lncRNA maternally expressed gene 3 (MEG3) by binding to the DNA methyltransferase 3B transcript [119]. Second, two RBPs are involved in regulating lncRNA expression in HNC. HuR, as an RBP, could form a regulatory circuit with HOTAIR and contribute actively to the stability and expression of HOTAIR in HNC [47]; similarly, HuR could regulate the stability of lnc-Sox5 and lead to TSCC progression [120]; in addition, another RBP, RNA-binding protein 24 could degrade MALAT1 by directly upregulating miR-25 expression, revealing the synergistic effect of RBP and miRNA on regulating lncRNA in NPC [121]. NF-κB is another functional protein—i.e. in TSCC, NF-κB could significantly improve the expression level of lncRNA NKILA [81]; BamHI-A rightward transcripts (BARTs), including lncRNAs, are produced by Epstein-Barr virus (EBV) in NPC, and NF-κB functions positively to activate BART promoters and modulate the expression of these lncRNAs [122]. As another representative of functional protein, TP53 could directly upregulate lncRNA LOC401317 in NPC cells, thereby suppressing tumor growth [123].Third, genetic changes such as single-nucleotide polymorphisms (SNPs) or epigenetic alterations within the non-coding genome could markedly affect the transcription of lncRNA; for instance, in PTC, the polymorphism of rs944289 in 14q13.3 could reduce the level of lncRNA PTCSC3 by abolishing the binding domain of CCAAT/enhancer binding proteins α and β and subsequently inhibiting the activation of the PTCSC3 promoter [124]. Additionally, the silencing of lncRNA H19 in well-differentiated NPC cells is attributed to the epigenetic alteration, namely, hypermethylation of the H19 promoter region [125]. Overall, the aberrant expression of lncRNAs in HNC is mainly controlled by the above three upstream regulators, and more upstream modulators are required to be uncovered.

In the past decade, lncRNAs, previously regarded as non-functional [11], have attracted considerable and increasing attention from investigators, primarily attributed to their frequently aberrant expression in cancers and their potential implication in tumor development and progression, particularly tumor metastasis [16], which constitutes the main threat for cancer-related death. LncRNAs, located in the nucleus or cytoplasm, might interact with mRNA, miRNA, DNA or protein to exert their diverse functions [21, 22], acting as a versatile player in regulating neoplasm metastasis.

In this review, we introduced the lncRNAs implicated in modulating the metastatic potential of HNC and attempted to illuminate the mechanisms; meanwhile, the involvement of these lncRNAs in the metastasis of other common malignancies are also briefly summarized to facilitate their study in HNC. Additionally, the upstream regulation of lncRNA that underlies its abnormal expression in HNC was discussed. Based on the screening result of lncRNAs, most of these are suggested to promote the EMT, migration, invasion or metastasis of HNC, while a small proportion of these have opposite effects of inhibiting HNC metastasis in vitro and/or in vivo. In theory, these lncRNAs highly expressed in HNC tissue or cells, driving its metastasis, have the potential to be utilized as therapeutic targets and predictors of poor outcome. However, the prerequisite of its application is the illumination of mechanisms validated in vitro and in vivo. Moreover, those lncRNAs downregulated in HNC and exhibiting the capability of suppressing its metastasis also hold the promise to serve as prognostic predictors.

In terms of clinical translation of lncRNAs, it still has a long way to go since there exist multiple unmet challenges for us to overcome. In order to realize the goal, we need to seek solutions for the listed issues: 1. exploring more lncRNAs affecting HNC metastasis; 2. investigating the detailed mechanisms of these lncRNAs; 3. figuring out the key lncRNAs pathways; 4. constructing a lncRNA interacting and regulating network since TME is complex and not dependent on a single lncRNA; 5. since HNC includes a range of cancers with various genetic background, individualized key lncRNAs and its involved pathways in a specific cancer might be studied; 6. intervening tumor metastasis-promoting lncRNAs to inhibit metastasis by targeting lncRNAs.

To target lncRNAs in HNC, diverse techniques have been developed, namely, RNA interference (RNAi) [126], antisense oligonucleotides(ASO) [127], clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 [128], RNA blocking oligonucleotides [129] and small-molecule modulators [130]. The RNAi technique represents selective silencing of lncRNAs, which is supposed to be more efficient in depleting cytoplasmic lncRNAs, while the successful rate of the deletion of nuclear lncRNAs remains unsatisfied [126, 131]. ASO is another effective approach to target lncRNAs regardless of the cellular location, which applies single-stranded DNA or RNA molecules to direct the RNase H to target lncRNAs, leading to its degradation [127]. One good example of its preclinical study proposed that ASO of MALAT1 could inhibit lung cancer metastasis in vivo significantly [132]. CRISPR/Cas9, as a relatively novel and efficient genome editing tool, has shed light on a novel way to edit lncRNA expression, while its specificity and efficacy remain to be further evaluated [27, 128]. As RNAi and ASO-mediated lncRNA degradation are dependent on enzymatic degradation, limiting the capability of improving their pharmacological qualities, RNA-blocking oligonucleotides is another method that modulates lncRNA by blocking the access of the cellular machinery to the RNA rather than contributing to the degradation of the lncRNA, which do not employ enzymes for the activity, exhibiting the advantage to receive more chemical modifications that enhance their drug-like characteristics [129]. In addition, small-molecule modulators involved in interrupting the lncRNA–protein interaction display high potential to target the lncRNA specifically, reducing its off-target effects [130]. Further validation of these above techniques and more technical innovations are required to select the optimal drug for lncRNA-targeted therapeutics, considering their advantages and disadvantages.

Recently, a novel approach, namely, clustered regularly interspaced short palindromic repeats (CRISPR)-Display, has emerged as a potentially transformative tool to probe into the function and mechanisms of lncRNAs [133]. Specifically, Shechner et al. established the technology by employing a nuclease-deficient Sp. dCas9 mutant, namely, “dCas9”, which displays certain RNA domains on the dCas9 cargo and is delivered to the predetermined DNA loci. Attempting to validate the applicability of the approach for lncRNAs, Shechner et al. fused the RepA domain of lncRNA Xist into the complexes; expectedly, they showed that the domain modestly repressed the expression of reporter gene, which initially revealed the plausibility of CRISPR-Display (CRISP-Disp) for studying lncRNAs [133]. In light of its characteristics, CRISP-Disp presents its potential advantages in lncRNA research. First, it could be utilized to address the question regarding whether lncRNA-regulated gene expression is due to the transcriptional effect or lncRNA domain itself [133]. Second, CRISP-Disp and immunoprecipitation of dCas9 followed by mass spectrometry could assist in identifying interacting proteins with the lncRNA domain [134]. Third, the approach might efficiently serve as a platform for selecting the novel functional domain of lncRNA. Fourth, the method could be applied to study lncRNA up to 4.8 kb and simultaneously explore multiple RNA domains by sharing the same dCas9 [133]. However, more validations are warranted to ensure its application in the future.

In summary, lncRNAs contribute to HNC invasion and metastasis by distinct mechanisms based on their subcellular localization, as well as modulating or regulated by the cancer immune microenvironment. Emerging evidence has identified lncRNAs as a novel set of potential prognosticators and therapeutic targets. The characterization of an expanded repertoire of lncRNAs and their interaction with the cancer immune microenvironment are promising to further improve combinatorial treatment protocols for HNC.