Long non-coding RNAs are important players in cellular function
The word epigenetics for most is the synonym for changes in the chromatin structure and DNA. However, it has changed in the last decades, and new elements of cellular control in response to physiological or pathological signals have transformed our knowledge considering epigenetics. Now, in the “modern era of epigenetic research”, we look not only at the chromatin and DNA state but also, which seems more fascinating, the action of various molecular regulators, such as non-coding RNA (ncRNA) [1] (Fig. 1). ncRNAs consist of constitutive RNAs: tRNA, rRNA, snRNA, snoRNA, and a group of regulatory RNAs: siRNA, piRNA, miRNA, NAT, circRNA, and lncRNA [2–4]. Though long non-coding RNAs (lncRNAs) are not translated into proteins, they become over time valuable and significant molecules. The lncRNAs consist of over 200 nucleotides and their activity enables them not only to interact with proteins and other RNAs but also to regulate gene transcription and expression through changes in the structure of chromatin [5, 6]. Specified types of lncRNAs play a role in signaling, decoying, guiding, and building scaffolds for proteins and other RNA molecules [7].
Fragments encoding lncRNAs can occupy intergenic or intragenic positions. Specific intragenic genes can be located in intron, enhancer, promoter, or 3’UTR flanking regions [5, 6]. Here, it is important to mention that the lncRNAs of interest are closely related to genes that are encoded in the vicinity of the paired mRNA, resulting in their interaction with each other. lncRNAs appear to play an important role in the regulation of gene transcription in the nucleus or subsequent post-transcriptional modifications in the cytoplasm. Despite the opportunities afforded by lncRNA research, many difficulties are still encountered in fully understanding them due to their location in the genome. It should be noted that more than 50% of lncRNAs are long intergenic non-coding RNAs (lincRNAs) with no annotated proteins but some short peptides can also be found [8]. The direction of transcription and the distance at which the transcript responsible for encoding protein is located from the lncRNA lead to the division of this group of RNAs into four classes: i) located on the same strand, ii) convergent, iii) divergent, and iv) isolated, which are located at least 50 kb from the closest protein-coding gene [15]. The main challenges in understanding the mechanisms of lncRNAs behavior are caused by different expression levels depending on tissue localization [10], frequent heterogeneity of isoforms, and numerous repeats in transcriptional initiation regions [11]. Moreover, as lncRNAs have been shown to vary with different expression levels, the specificity of their function in different cell types changes [12]. As it emerged, dysfunction of the activity or cellular mechanisms of lncRNAs may appear through pathological states including tumor progression through influence not only on the chromatin structure but also on several transcription factors [13]. According to GENCODE, NCBI Refseq, LNCipedia, and NONCODE number of lncRNAs’ genes is estimated between 56’946 to 17’952 locus which gives 27’381 to 172’216 transcripts [14–17]. It is proven that aberration of both coding and non-coding RNAs play a crucial role in cancer biology [18]. lncRNAs regulate cell growth, cell cycle, cell phenotype, migration and invasion ability, and apoptosis [7]. The functions and activity of lncRNAs are still being investigated. More and more new elements are added, such as the possibility of interaction of lncRNAs with RNAi molecules, for example UCA1, CASC2, GAS5, FER1L4, WDFY3-AS2, TP53TG1, FENDRR or SNHG1 lncRNAs with miR-18a [19]. What is more interesting, lncRNAs, such as MIR31HG, can be host genes for miRNAs and play a dual role as lncRNA and as a primary-miRNA transcript.
MIR31HG is a member of long non-coding microRNA-host-genes
MIR31 host gene (MIR31HG) belongs to a group of long non-coding microRNA-host-genes (lnc-MIRHGs) distinguished from lncRNAs due to coding the microRNA gene and transcript. MIR31HG functions in two specified RNA forms, as a long transcript, lncRNA, and as a host gene, which under processing is changed into a short non-coding RNA molecule, microRNA named miR-31. Dhir et al. estimated the distribution of miRNA between lncRNA and protein-coding genes and it is 82.5% and 17.5%, respectively. These lncRNAs can be divided into lincRNA (57.1%), pseudogene (13.2%), antisense (16.0%), and other (19.0%) types of transcripts. Moreover, these lncRNAs are the source of about 17.5% of miRNAs in humans [20].
It should be noted that lnc-MIRHGs are an under-studied class of lncRNAs in contrast to the well-known microRNAs which are hosted by them [21].
It is worth mentioning that miRNAs are transcribed as pri-miRNAs whose structure includes a terminal loop, stem, and 50 and 30 single-stranded overhangs at the ends. According to miRbase and GENCODE, there are 1’917 human miRNA genes, 1’917 hairpin precursors, and 2’654 mature sequences, and 1’881 miRNA genes [22]. Pri-miRNA does not perform the gene silencing function, so it is post-transcribed through canonical or non-canonical miRNA biogenesis pathways [23]. miRNAs, concerning their relationship with MIRHGs, can be categorized into: i) intronic, ii) exonic, ii) exon-intron junction (SO-miRNAs), and iv) intergenic miRNAs. These localizations in the gene and genome influence their biogenesis [21]. There are two main proposed biogenesis models of MIRHG and its intragenic miRNA: i) synergetic model which includes mirtron processing, cooperative of the splicing machinery and microprocessor, and a splicing-independent manner of miRNA production with the presence of splicing factors, and the second model, ii) competition model where miRNA production is alternative- and non-alternative-splicing-mediated. A detailed description of these two models is presented by Sun et al. [21]. It was shown that there is a correlation in expression between the miRNA and the corresponding MIRHG. Moreover, miRNA molecules, which are usually located within 50 kb, are derived from a single transcript. When miRNAs originate from an intron region, their expression is often correlated with MIRHG expression [26]. It should be noted that the expression of MIRHG can be regulated by tumor cells, more specifically by the promoters. The changes in methylation of the above promoters result in an altered expression of the encoded miRNA [27]. Moreover, in the case of intragenic miRNAs and their host, a negative feedback loop can be created which regulates the level of both transcripts [21]. It is worth noting that the expression of miRNA and MIRHG pairs is not always related in this way, but in some cases, it offers great diagnostic possibilities [27].
MIRHGs function as primary transcripts of miRNAs but some evidence indicates they also have the ability to act the role as lncRNA transcripts. Some authors show the miRNA-independent role of lnc-MIRHGs such as MIR22HG, MIR100HG, MIR205HG/LEADeR, RMST, CYTOR, LINC01138, LINC-PINT, MIR503HG, NEAT1, PVT1, H19 or MIR222HG, which are fully described by Sun et al. [25]. MIRHGs can be categorized as oncogenes or tumor suppressors, but this function can be specific to the type of cancer. MIRHGs can: i) act as ceRNA elements, ii) interact with DNA elements, or iii) with proteins as well as vi) regulate the interaction with proteins [25]. All these capabilities and functions make them an astonishing group of RNA molecules. Therefore, in this review we will continue to focus on one of them, MIR31HG, presenting the current state of knowledge.
MIR31HG — localization in the genome and biogenesis
The MIR31HG molecule has been known in the past by names: LOC554202, hsa-lnc-31, or lncHIFCAR. It is localized on chromosome 9 and consists of 4 exons. The length of the whole MIR31HG is about 150–106 Kb [28, 29]. According to GeneCards The Human Gene Database, the latest assembly of genomic locations shows that MIR31HG Gene is situated on the minus strand orientation on chr9:21’380’073-21’591’766 (GRCh38/hg38) and its size is estimated as 211’694 bases [29]. Based on GeneHancer (GH) data, 3 different enhancers and 2 different promoter/enhancer regions were distinguished for the MIR31HG gene, which creates regulatory elements and has the transcription factor binding sites such as: different TRIM, ZNF, MYC, STAT3, ZEB2, NANOG or EZH2 proteins [29].
MIR31HG is over 200 nucleotides long and it is transcribed by RNA polymerase II and the mature transcript is polyadenylated. Augoff et al. showed that the MIR31HG (LOC554202) transcript is 2’246 bp long and it does not encode any protein products [27]. Moreover, the MIR31HG gene does not encode any short peptides which could be produced by this type of transcript [29]. These features cause MIR31HG to be indisputably classified as a long non-coding RNA (lncRNA). However, MIR31HG belongs to the unique subtype of lncRNAs because, in the first intron of MIR31HG, it harbors a sequence of other type of ncRNA molecule which is classified as a short-noncoding RNA, named miR-31. It should be noted that this first intron contains a CpG island, which is responsible for the transcription regulation of both types of ncRNA molecules [27]. What is interesting, the MIR31HG transcript is not post-translationally modified and 32 different transcriptional variants are distinguished with a length between 287 and 10’980 bp. It should be noted that no orthologs or paralogs for MIR31HG have been identified up to date [29].
The expression level of MIR31HG is tissue-specific and according to the GENE NCBI, the highest expression is observed in normal tissue taken from the urinary bladder and the lowest in the pancreas, liver, and heart [28].
The biological function of MIR31HG
It is known that MIR31HG plays an important role in cellular processes whose disturbance causes cancerogenesis or increases the rate of this process. The most described, where MIR31HG is involved, are cell proliferation, cell cycle, invasiveness, EMT process as well as apoptosis. However, MIR31HG has also been described in the context of angiogenesis, hypoxia, senescence, or inflammation (Fig. 2).
Cell proliferation and cell cycle
One of the most characteristic biological functions linked with MIR31HG is its influence on cellular proliferation. Nie et al. were the first to observe that upregulation of MIR31HG caused lower cell proliferation of gastric cancer cells in vitro and in vivo and its knockdown caused a reversed effect partly by regulating E2F1 and p21 [30]. Similarly, in HNSCC (head and neck squamous cell carcinoma) cells the knockdown of MIR31HG affects proliferation, cell cycle arrest in G1 or S phase, and apoptosis. This effect was caused by decreasing expression of HIF1A and CCND1, and increasing p21 on mRNA and protein levels [31]. However, the opposite effect was observed in the case of breast cancer, where silencing of MIR31HG expression inhibits the proliferative ability of the cells, and its function is linked with the POLDIP2 expression level [32]. Similarly, in lung squamous cell carcinoma (LSCC), inhibition of MIR31HG causes reduced cell proliferation, but the molecular way was not clearly explained [33]. Another study showed that MIR31HG knockdown inhibits not only cancer cell migration but also colony formation and cell proliferation [34].
In addition to the tumors’ model, also in the case of human periodontal ligament stem cells, the influence of of MIR31HG on proliferation was found. Methylation of the MIR31HG promoter induced by mechanical force causes reduced expression of MIR31HG and upregulation of IL-6, DNMT1, and DNMT3B. The changes in the stem cells’ proliferation can be overcome by DNMT1 and DNMT3B knockdown, which interact with the upstream region of the MIR31HG promoter and induce its expression [35].
Invasiveness and EMT process
The migration and invasive ability are also linked with MIR31HG. It caused higher invasiveness of breast cancer cells [32] as well as of lung cancer cells [33]. It was observed that in the case of non-small-cell lung carcinoma (NSCLC), reduction of MIR31HG expression was associated with the EMT process and manifested by reduction of Twist1 and Vimentin expression and upregulation of E-cadherin. Authors stated that MIR31HG causes changes in the Wnt/β-catenin signaling pathway by reduction of GSK3β and β-catenin and also its knockdown was linked with phosphorylation of GSK3β [33]. It should be noted that MIR31HG, depending on the cellular state, does not induce cancer cell invasion but promotes paracrine senescence [36]. The role of MIR31HG in the EMT process was also observed in osteosarcoma, where upregulation of MIR31HG caused down-regulation of miR-361. This effect was manifested by upregulation in the protein levels of miR-361’s target genes, vascular endothelial growth factor (VEGF), forkhead box M1 (FOXM1), and Twist in in vitro model and patients samples, as well as by downregulation of E-cadherin observed in cell lines after upregulation of MIR31HG [37].
Apoptosis
Feng et al. based on nasopharyngeal carcinoma indicated that the knockdown of MIR31HG causes inhibition of apoptosis by negative regulation of the PI3K/AKT signaling pathway [38]. Moreover, based on U2OS and Saos-2 cell lines it was indicated that apoptosis could be regulated by the expression level of miR-361 and its targets, VEGF, FOXM1, and Twist, as well as by changes in anti-apoptosis of B-cell lymphoma 2 (BCL2) and cyclin D (CCND1) proteins levels whose expression was inhibited by artificial up-regulation of MIR31HG [37].
Inflammation
Gao et al.’s studies identified the elevation of MIR31HG in psoriatic skin. The expression of two types of keratin, KRT6 and KTR16, was significantly up-regulated in keratinocytes from patients with psoriasis in comparison to normal samples, and MIR31HG-dependent. As mentioned before, also the cell cycle was changed after MIR31HG silencing in keratinocytes and it was shown that there were more keratinocytes in the G2/M phase relative to cells in the S phase, indicating that MIR31HG knockdown inhibits the proliferation of HaCaT keratinocytes. Moreover, it was indicated that stimulation of proinflammatory interleukin 17A (IL-17A), interleukin 22 (IL-22), tumor necrosis factor alpha (TNF-α), and interleukin 1 alpha (IL-1α) cytokines was able to elevate MIR31HG expression. It should be noted that BAY, NF-κB inhibitor, and p65 regulations additionally showed that NF-κB activation has an impact on MIR31HG [39].
Senescence
Senescence can also be induced in the presence of oncogenes as a response aimed at tumor suppression, such a mechanism is called oncogene-induced senescence. A group of researchers has shown that MIR31HG is also involved in this induction [40]. Montes et al. based on cell models, showed that MIR31HG plays a dual role depending on the localization in cytoplasm and nucleus. MIR31HG is overexpressed and translocates to the cytoplasm during BRAF-induced senescence and by kinase RSK causes phosphorylation of YBX1.Finally, it changes the senescence-associated secretory phenotype (SASP) of cells by IL1A translation activation [41].
MIR31HG can be used as a biomarker
The potential role of MIR31HG as a biomarker was analyzed on both the Cancer Genome Atlas (TCGA) data and collected samples by specified research groups or cell lines. However, independent research about the same cancer type was done in only a few cases [33, 34, 42–53]. A growing number of studies indicate that MIR31HG plays an important role in cancer as an oncogene or as a suppressor. It was shown that MIR31HG is down-regulated in gastric cancer [30], bladder cancer [34, 42], hepatocellular carcinoma [57], pancreatic ductal adenocarcinoma [58], and glioblastoma [65]. Moreover, up-regulation of MIR31HG was indicated in the case of breast cancer [32], head and neck cancer cancers [31, 67, 68], melanoma [54], cervical carcinoma [56], osteosarcoma [57], non-small cell lung cancer [33, 43, 44], lung adenocarcinoma [44–47], colorectal cancer [48–51], thyroid cancer [52, 53], as well as papillary thyroid cancer [64]. It should be noted that two independent studies checking the role of MIR31HG in esophageal squamous cell carcinoma indicated opposite expression levels of this lncRNA [59, 60].
MIR31HG can be used as a diagnostic biomarker and it is detectable in tissue as well as in biological fluids such as plasma [31, 60]. Moreover, MIR31HG could describe the more invasive types of cancer, with advanced tumor–nodules–metastases (TNM) stages, lymph node invasion as well as distant metastasis [31, 32, 43, 54, 57, 59, 60, 62] and its higher expression is associated with worse disease-free and overall survival (OS). However, the opposite meaning of MIR31HG as a prognostic biomarker is observed in the case of bladder cancer [42], hepatocellular carcinoma [57], esophageal squamous cell carcinoma [59], and gastric cancer [30]. For these cancers, a higher expression level of MIR31HG is negatively correlated with more advanced TNM-stage [30, 42, 57], with lower tumor nodule number, vascular invasion [57], metastasis [57, 59] or poor tumor differentiation [59]. Analysis of patients’ survival revealed that the higher expression levels of MIR31HG were associated with longer survival for these types of cancers [30, 57, 59].
In the end, it should be noted that Chang et al. indicated potential features of MIR31HG as a predictive biomarker, and higher levels of MIR31HG were associated with increased cisplatin resistance [55], but there is only one study exploring this problem. All studies and results describing MIR31HG as a biomarker are included in Table 1.
Type of cancer |
Methods of detection |
Sample type |
Type of biomarker |
Description |
Ref. |
Head and neck cancer |
qRT-PCR |
60 paired normal and adjacent cancer samples; plasma; FaDu and Cal-27 cell lines |
Diagnostic, prognostic |
Overexpressed in cancer tissue and plasma of the early and advanced stages of patients Correlated with advanced T-stages and lymph node invasion Lower level was associated with better OS and RFS MIR31HG regulated cell cycle progression and apoptosis by targeted HIF1A and p21 |
|
Melanoma |
qRT-PCR |
55 patients’ tissues and Human Epidermal Melanocytes (cell line) |
Diagnostic, prognostic |
overexpressed in cancer tissue and cell lines Associated with lymph nodes invasion Distal metastasis and higher TNM-stages Lower MIR31HG expression was characteristic with lower malignancy by decreased cell proliferation, and migration and invasion rates |
|
Breast cancer |
qRT-PCR |
50 paired normal and adjacent cancer samples and T47D, BT-474, SUM149-Luc, BT549, and MCF-10A cell lines |
Diagnostic, prognostic |
Overexpressed in cancer tissue and cell lines Correlated with the patient’s tumor diameter Correlated with tumor TNM-stages and lymph node metastasis Lower level associated with better survival MIR31HG influenced on proliferation, migration and invasion abilities by regulation of POLDIP2 |
|
Oral squamous cell carcinoma |
TCGA (RNAseq); qRT-PCR |
520 cancer and 44 normal TCGA samples; Cytobrushed samples and matched mucosa from 28 patients and OC3, OC4, OC5, SAS, OECM1, FaDu, and NOK cell lines |
Diagnostic, prognostic and predictive |
Overexpressed in cancer tissue and cell lines MIR31HG enhanced oncogenic phenotype especially by enrichment of Wnt pathway Lower level associated with better survival Correlated with higher proliferation and wound healing closure rates Higher MIR31HG associated with increased cisplatin resistance |
|
Cervical carcinoma |
TCGA (RNAseq); qRT-PCR |
24 normal and 104 lesions/cancer samples from GEO and 306 cancer and 13 normal TCGA data sets; 46 pairs of cervical cancer tissues and adjacent patients’ tissues and CasKi, SiHa, C33A and HcerEpic cell lines |
Diagnostic, prognostic |
Overexpressed in cancer tissue and cell lines, Knockdown of MIR31HG suppressed cell growth and invasion MIR31HG regulated miR-361-3p and through it modulated epithelial membrane protein 1 (EMP1) mRNA expression level |
|
Osteosarcoma |
qRT-PCR |
40 paired normal and adjacent cancer samples patients’ tissues and 143B, MG63, U2OS, Saos-2 and hFOB1.19 cell lines |
Diagnostic |
Overexpressed in cancer tissue and cell lines Higher MIR31HG expression associated with higher tumor stages and distant metastasis miR-361 was sponged by MIR31HG and down-regulated Knockdown of MIR31HG restored the expression of miR-361 in cell lines miR-361 induced cell apoptosis and G1/S arrest, inhibited proliferation and migration in Saos-2 and U2OS cells, and MIR31HG had reversed effect MIR31HG by regulation of miR-361 targeted VEGF, FOXM1 and Twist, and caused upregulation of BCL2, CCND1 and EMT phenotype Higher level of VEGF, FOXM1 and Twist were positively correlated with MIR31HG in patients’ samples MIR31HG promoted tumor growth by regulation of miR-361 and VEGF, FOXM1 and Twist in vivo |
|
Bladder cancer |
TCGA (RNAseq); qRT-PCR |
102 FFPET patients’ samples and 370 TCGA patients, and SCaBER, UMUC3, T24, RT112, RT4 and UROtsa cell lines |
Diagnostic, prognostic |
decreased in cancer tissue and cell lines and depended on the spliced variants (MIR31HGΔE1 and MIR31HGΔE3) MIR31HGΔE3 highly expressed in the case of the basal subtype Higher expression of MIR31HGΔE1 and MIR31HGΔE3 associated with worse OS and DFS Knockdown of MIR31HG inhibited cell proliferation, colony formation, and migration abilities |
|
Colorectal cancer |
GEO and TCGA (Arrays/RNAseq); qRT-PCR |
nearly 2000 CRC biopsies and preclinical models; patient-derived xenografts; cell lines |
Diagnostic, prognostic |
Strongly correlated with miR-31-5p MIR31HG changed in 12% of patients and associated with depletion of CMS2-canonical subgroup and shorter RFS 5-year RFS for patients (stage II subgroup) with MIR31HG outlier status lower than those with normal-like expression MIR31HG outlier status associated with worse outcome in clinical high risk groups (CMS4-mesenchymal gene expression subtype) Patients with MIR31HG outlier expression had reduced expression of MYC targets, higher expression of epithelial-mesenchymal transition, TNF-α/NFκB, TGF-β, and IFN-α/γ gene expression signatures Prognostic value of MIR31HG outlier status was independent of cytotoxic T lymphocyte and fibroblast infiltration |
|
Non-small cell lung cancer |
qRT-PCR |
88 paired normal and adjacent cancer samples patients’ tissues and A549, H1299, NCIH460 and 16HBE cell lines |
Diagnostic, prognostic |
overexpressed in tumor tissues compared with adjacent normal tissues Higher MIR31HG expression associated with histological differentiation grade, lymph node metastasis and higher TNM-stages Higher MIR31HG expression associated with worse OS MIR31HG knockdown inhibited proliferation and invasion abilities Lower expression suppressed the EMT phenotype (reduced Twist1 and Vimentin, and increased E-cadherin expressions) Inhibition of the Wnt/β-catenin signaling pathway (reduced expression of GSK3β and β-catenin, and increased phosphorylation of (p)-GSK3β) |
|
Lung adenocarcinoma |
qRT-PCR |
132 patients’ tissues and 20 adjacent non-cancerous samples, and A549, H2228, H1975, H1299 and BEAS-2B cell lines |
Diagnostic, prognostic |
Overexpressed in cancer tissues and cell lines Associated with higher TNM-stages and differentiated degree Higher MIR31HG was an independent unfavorable OS factor Knockdown MIR31HG caused inhibition of cells proliferation and blocked cell-cycle and didn’t changed cell apoptosis No correlation between MIR31HG and miR-31 expressions and knockdown of MIR31HG had no effect on the miR-31 level |
|
Non-small cell lung cancer |
qRT-PCR |
50 paired normal and adjacent cancer patients’ tissues and H1299, A549, H1975, H460 and BEAS-2B cell lines |
Diagnostic, prognostic |
Overexpressed in tumor tissues and cell lines SP1, transcription factor, binds to promoter region of MIR31HG and induces its expression Higher MIR31HG was an independent predictor worse OS MIR31HG associated with less differentiation degree and higher TNM-stages Knockdown of MIR31HG inhibited migration, invasion and metastasis abilities, Overexpression of MIR31HG reduced the expression of miR-214 and induced cancer progression |
|
Colorectal cancer |
qRT-PCR |
30 paired normal and adjacent cancer patients’ tissues and RKO, SW480, SW620, LoVo and HCT116 cell lines |
Diagnostic, prognostic |
Overexpressed in tumor tissues and cell lines Associated with worse prognosis Overexpression of MIR31HG induced proliferation, growth, invasion, glycolysis and lung metastasis and angiogenesis observed in vitro and in vivo MIR3HG upregulated higher expression of YY1 (mRNA and protein) Forced overexpression of YY1 induced overexpression of enhanced MIR31HG MIR31HG inhibits miR-361-3p which has and anti-tumor effect by targeting YY1 |
|
Bladder cancer |
qRT-PCR |
55 paired normal and adjacent cancer patients’ tissues and T24, 5637, UM-UC-3, SW780 and SV-HUC-1 cell lines |
Diagnostic |
Downregulated in tumor tissues and cell lines MIR31HG negatively associated with TNM-stages |
|
Thyroid cancer |
qRT-PCR |
29 paired normal and adjacent cancer patients’ tissues and SW579, TPC-1, HTH83 and Nthy-ori 3–1 cell lines |
Diagnostic, prognostic |
Overexpressed in tumor tissues and cell lines Higher MIR30HG associated with worse prognosis Knockdown of MIR30HG reduced proliferation, invasion, migration, promoted cell apoptosis in vitro and tumor growth in vivo MIR30HG regulated the expression of miR-761 which in turn regulates MAPK1 |
|
Hepatocellular carcinoma |
qRT-PCR |
42 paired normal and adjacent cancer patients’ tissues and SMMC7721, HepG2, Huh7, SK-hep1 and 293 T (HEK) cell lines |
Diagnostic, prognostic |
Downregulated in tumor tissues and cell lines Higher expression associated with better OS Higher expression correlated with lower tumor nodule number, lower vascular invasion and lower TNM-stages Overexpression of MIR31HG reduced proliferation and metastasis in vitro and in vivo MIR31HG regulates miR-575 expression, which has oncogenic properties, and influences on its target — ST7L There was a reciprocal inhibition between MIR31HG and miR-575 in the same RISC complex |
|
Pancreatic ductal adenocarcinoma |
GEO (Array); qRT-PCR |
GEO 45 paired normal and adjacent cancer patients’ tissues and AsPC-1, PANC-1, CFPAC-1, Hs 766 T, SW 1990, MIA PaCa-2, BxPC-3 and hTERT-HPNE cell lines |
Diagnostic |
Overexpressed in tumor tissues and cell lines Knockdown of MIR31HG reduced cell growth, induced apoptosis and G1/S arrest, and inhibited invasion in vitro as well as tumor growth in vivo miR-193b targets MIR31HG and they have inverse correlation MIR31HG may act as an endogenous “sponge” by regulation of miR-193b and its’ targets (CCND1, Mcl-1, NT5E, KRAS, uPA, and ETS1) |
|
Esophageal squamous cell carcinoma |
qRT-PCR |
185 paired normal and adjacent cancer patients’ tissues |
Diagnostic, prognostic |
Downregulated in tumor tissues and cell lines Lower expression MIR31HG associated with poor differentiation, advanced lymph node metastasis positive distant metastasis and higher TNM-stages, Higher expression of MIR31HG associated with better OS and it is an independent prognostic marker for survival |
|
Esophageal squamous cell carcinoma |
qRT-PCR |
53 paired normal and adjacent cancer patients’ tissues, 53 plasma samples from patients and 39 from healthy donors, and EC9706, EC1 and Het-1A cell lines |
Diagnostic, prognostic |
Overexpressed in tumor tissues, plasma and cell lines Expression level of MIR31HG in tissue and plasma from the same patient was positively correlated Higher expression observed in tissue and plasma samples of patients with higher TNM-stages and positive lymph node metastases MIR31HG displayed high diagnostic sensitivity and specificity for predicting cancer occurrence Knockdown of MIR31HG reduced proliferation, migration, and invasion abilities Reduction of MIR31HG caused inhibition of Furin and MMP1 |
|
Gastric cancer |
qRT-PCR |
42 paired normal and adjacent cancer patients’ tissues and SGC7901, BGC823, MGC803, MKN45 and GES-1cell lines |
Diagnostic, prognostic |
Downregulated in tumor tissues and cell lines Associated with larger tumor size and advanced pathological stages Lower MIR31HG expression associated with worse PFS and OS Overexpression of MIR31HG inhibited cell proliferation in vitro and tumor growth in vivo Knockdown of MIR31HG promoted cell proliferation partly via regulation of E2F1 and p21 expressions |
|
Oral squamous cell carcinoma |
GEO (array) qRT-PCR |
GEO 22 normal and 57 cancer patients’ tissues, 15 paired normal and adjacent cancer patients’ tissues and SAS cell line |
Diagnostic, prognostic |
Overexpressed in tumor tissues Higher level of MIR31HG associated with worse OS and RFS (independent prognostic predictor) Overexpression of MIR31HG induced pseudo-hypoxic phenotype Knockdown of MIR31HG reduced hypoxia-induced HIF-1α transactivation, sphere-forming ability, metabolic shift and metastatic potential in vitro and in vivo MIR31HG directly bound and facilitated the recruitment of HIF-1α and p300 cofactor to the target promoters |
|
Ovarian cancer |
TCGA (RNAseq), qRT-PCR |
352 TCGA patients’ cancer tissues, and TOV-21G, A2780, SKOV3, and IOSE80 cell lines |
Diagnostic, prognostic |
mIR31HG and other lncRNAs (ACTA2-AS1, CARD8-AS1, HCP5, HHIP-AS1, HOTAIRM1, ITGB2-AS1, LINC00324, LINC00605, LINC01503, LINC01547, MIR155HG, OTUD6B-AS1, PSMG3-AS1, SH3PXD2A-AS1, and ZBED5-AS1) associated with OS Those lncRNAs correlated with patient age at initial pathologic diagnosis, lymphatic invasion, tissues source site, and vascular invasion |
|
Colon Cancer |
TCGA (RNAseq) |
166 TCGA stage II colon cancer patients |
Diagnostic, prognostic |
mIR31HG as well as WASIR2, miR-200a and miR-155 overexpressed in cancer tissue Lower MIR31HG expression associated with better OS 4 lncRNA-miRNA signature can be used as independent prognostic value of OS for stage II colon cancer with high sensitivity and specificity Correlated genes with MIR31HG were associated with the EMT process and the VEGFR3 signaling in lymphatic endothelium pathways |
|
Colorectal Cancer |
TCGA (RNAseq), qRT-PCR |
TCGA 593 tumor and 51 paired normal and adjacent cancer patients’ tissues |
Diagnostic, prognostic |
mIR31HG as well as LINC00461, LINC01055, ELFN1-AS1, LMO7-AS1, CYP4A22-AS1, AC079612.1, LINC01351 associated with OS Risk factors for the prognosis with high sensitivity and specificity The index based on the 7 survival-related IRLs were accurate in the prognosis monitoring IRLs index were correlated with a tumor status and N-stage and immune cell infiltration of CD4+ T cells and dendritic cells |
|
Papillary thyroid cancer |
GEO (array) qRT-PCR |
GEO 136 normal and 157 cancer patients’ tissues, 50 paired normal and adjacent cancer patients’ tissues |
Diagnostic |
five upregulated (ENTPD1, THRSP, KLK10, ADAMTS9, MIR31HG) and five downregulated (SCARA5, EPHB1, CHRDL1, LOC440934, FOXP2) genes For this ten genes the most highly enriched GEO terms were: extracellular exosome, cell adhesion, positive regulation of gene expression, ECM organization, tyrosine metabolism, complement and coagulation cascades, CAMs, transcriptional misregulation and ECM-receptor interaction pathways |
|
Lung adenocarcinoma |
TCGA (RNAseq), qRT-PCR |
TCGA 465 tumor and 43 paired normal and adjacent cancer patients’ tissues |
Diagnostic, prognostic |
MIR31HG , CEBPA-AS1, GVINP1 and RAET1K were selected after Cox analysis and OS prognostic gene signature was developed with high sensitivity and specificity MIR31HG significantly associated with survival rate, Four‐lncRNA signature had prognostic value to predict tumor stage T stage, N stage, neoplasm cancer and primary therapy outcome 494 genes, which were coexpressed with lncRNAs of the risk score model, were associated with signal transduction, blood coagulation, pathways in cancer and chemokine signaling pathways |
|
Glioblastoma |
qRT-PCR |
|
Diagnostic, prognostic |
mIR31HG deleted in over 73% of all GBMs miR-31 status: 30.92% homozygous null, 42.68% heterozygous and 26.40% wildtype In low grade gliomas MIR31HG status: 6.96% homozygous null, 27.04% heterozygous, and 66% wild type Loss of one or both copies of MIR31HG significantly reduced the levels of miR-31 Homozygous MIR31HG deletions predominantly associated with Mes- and C-GBMs MIR31HG deletions associated with shorter MMS (Median Months Survival) for patients with primary GBM and for patients with Mes-GBM CDKN2A deletions associated with diminished DFS times in all GBM, and patients with N-GBM but which lies adjacent to MIR31HG, did not predict shorter MMS in patients with Mes-GBM or primary GBM MIR31HG deletions not associated with diminished DFS miR-31 inhibits TRADD and consequently NF-κB signaling and influencing on MIR31HG promoter containing three putative NF-κB binding sites |
|
Lung squamous cell carcinoma and lung adenocarcinoma |
TCGA (RNAseq) |
504 LUSC and 522 LUAD samples from TCGA |
Diagnostic, prognostic |
mIR31HG is altered as deleted gene in 0.14 frequency in the case of LUAD MIR31HG, CDKN2A-AS1 and LINC01600 predicted poor OS in LUAD MIR31HG and LINC01600 play their roles in female patients, while CDKN2A-AS1 play its role in male patients Important molecular functions for both CDKN2A-AS1 and MIR31HG-coexpressed genes were binding and catalytic activity — the top two enriched biological pathways were cellular process and metabolic process, and the most enriched pathway was the P53 pathway; IFNE, CDKN2A and MTAP cBioPortal analysis results showed that all three coexpressed genes shared very similar alteration patterns with CDKN2A-AS1 and MIR31HG MTAP was the only gene located between CDKN2A-AS1 and MIR31HG. CDKN2A was next to, and partially overlapped with, CDKN2A-AS1, and IFNE was located in the MIR31HG intron |
|
Colon Cancer |
GEO and TCGA (Arrays/RNAseq) |
1089 colon cancer patients from GEO and 391 patients from TCGA |
Diagnostic, prognostic |
mIR31HGincluded in as one of the recurrence-associated six-lncRNAs (LINC0184, AC105243.1, LOC101928168, ILF3-AS1, and AC006329.1) Score model based on the six-lncRNA signature higher in the recurrent patients than non-recurrent patients lncRNA signatures effectively distinguish between high and low risk of cancer recurrence Only combination of six lncRNAs gives the greatest predictive ability (accuracy rate of 72.2% and AUC of 0.724) Six-lncRNA signature was independent of clinicopathological factors, has potential to differentiated patients, with similar clinical stage, into low and high risk subgroups Patients with higher expression level of these six lncRNAs displayed significantly higher recurrence risk status and RFS Functional analysis of the six lncRNA signature indicated implication of them into ATP metabolic processes, cell proliferation and angiogenesis, cell death, leukocyte differentiation |
|
Thyroid Cancer |
GEO and TCGA (Arrays/RNAseq), qRT-PCR |
57 PTC with a reference sample (pool of 9 adjacent normal thyroid tissue) and 4 PTC against 4 matched adjacent normal thyroid tissues from GEO and 391 patients from TCGA, CAL62 and SW579 cell lines |
Diagnostic, prognostic |
Upregulated in patients’ cancer samples Correlated with M and N stages and positive lymph nodes examined status MIR31HG overexpression correlated with high immune infiltrate levels of CD8+ T cells, macrophage, neutrophil, myeloid dendritic cells, and B cells Knockdown of MIR31HG reduced cell proliferation and cycle progression MIR31HG associated with metabolic pathways, vesicle-mediated transport, tricarboxylic acid cycle, Hedgehog signaling pathway, and Hippo signaling pathway including CCND2, CCND3, SDHC, SDHD, SUCLA2, and SUCLG1 |
|
Colorectal cancer |
GEO and TCGA (Arrays/RNAseq) |
TCGA 647 tumor and 51 paired normal and adjacent cancer patients’ tissues and 122 patients’s samples from GEO |
Diagnostic, prognostic |
mIR31HG with other four lncRNAs (H19, HOTAIR, WT1‐AS, and LINC00488) closely related to the OS Five‐lncRNA signature was independent prognostic marker of the high‐risk scores’ patients which had poor survival rates High‐risk score based on five-lncRNA signature associated with more advanced TNM stages and residual tumor In the univariate analysis risk score of the five‐lncRNA model and some of clinical features (age, TNM stages, residual tumor) were associated with the OS In the multi-variate analysis, the five‐lncRNA model displayed an independent prognostic factor Patient’s prognosis separated by risk score and TNM staging were different: patients with lower risk score and tumor grade displayed better prognosis The risk score and clinicopathological features displayed better informative predict the patient’s 1, 3, 5‐year survival Five-lncRNA model was associated with signaling pathway regulating pluripotency of stem cells, WNT, Hippo signaling path-way, basal cell carcinoma and colorectal cancer, negative regulation of translation, extracellular space, transcription from RNA polymerase II promoter, odontogenesis and negative regulation of fibroblast proliferation |
|
Oral squamous cell carcinoma |
GEO (Arrays) |
167 OSCCs and 45 oral mucosa from healthy controls from GEO and validation using 74 oral cavity squamous cell carcinoma and 29 adjacent normal tissue; GSE9844 with 26 tongue squamous cell carcinoma and 12 matched adjacent normal tissue and GSE6791 comprised of 28 cervical cancers, 42 head and neck cancers and 14 site-matched normal oral tissue from GEO |
Diagnostic |
mIR31HG and 13 lncRNAs (LOC441178, C5orf66-AS1, HCG22, FLG-AS1, CCL14/CCL15-CCL14, LOC100506990, TRIP10, PCBP1-AS1, LINC01315, LINC00478, COX10-AS1/LOC100506974, MLLT4-AS1, and DUXAP10/LINC01296) were validated in all three datasets, and were upregulated Its expression differs between HPV-positive OPC and HPV-negative OPC, and is downregulated in HPV positive ones It is significantly differentially expressed between subsites of OSCC (OPC vs. OCC) miR31HG levels between smokers and non-smokers were indicated miR31HG was not validated by qRT-PCR in patients samples |
|
Lung squamous cell carcinoma and lung adenocarcinoma |
TCGA (RNAseq) |
TCGA 504 tumor and 46 paired normal and adjacent cancer patients’ tissues |
Diagnostic, prognostic |
3’366 mRNAs, 79 miRNAs and 151 lncRNAs were identified as involved in development of LUSC Only lncRNA MIR99AHG positively correlated with OS and PLAU, miR-31-5p, miR-455-3p, FAM83A-AS1, and MIR31HG were negatively associated with OS Only PLAU was validated using qRT-PCR and was upregulated in SK-MES-1 cells compared with 16-BBE-T cells Changed genes were associated with signal transduction, cell adhesion, blood coagulation, immune response, cell proliferation, apoptosis, transmembrane transport or small molecule metabolic processes |
|
Laryngeal squamous cell carcinoma |
Arrays and qRT-PCR |
39 pairs of LSCC tissues and adjacent non-neoplastic tissues, microarray results deposit (GSE84957) |
Diagnostic, prognostic |
1459 lncRNAs (846 up-and 613 down-regulated) and 238 mRNAs (1542 up- mRNAs and 839 down-regulated) were differentially expressed, ITGB1, HIF1A, and DDIT4 were core mRNAs involved in matrix organization, cell cycle, adhesion, and metabolic pathway MIR31HG was positively correlated with HIF1A and lncRNA NR_027340 was positively correlated with ITGB1 |
“Small but crazy”: miR-31
miR-31 is a short non-coding RNA, classified as microRNA (miRNA), and is the product of specific cleavage of MIR31HG transcript. miRNA molecules can be responsible for controlling many genes in a differentiated way [69, 70]. miR-31 affects many processes not only in normally functioning cells, but is also important in disease and cancer processes. Based on results obtained from different tissues and cancer cell lines it was demonstrated that miR-31 exhibits a whole spectrum of expression depending on tissue type. It is worth mentioning that the miR-31 precursor was present in all tested cell lines but its mature form was predominantly found in tumor cell lines [71]. miR-31 regulates cell proliferation by affecting the genes responsible for this process, such as LATS1, and CREG, in vascular smooth muscle cells [72, 73]. An excellent example is the mechanism studied in colon cancer cells when miR-31 overexpression targets E2F2, which acts as a tumor suppressor on colon cancer cell proliferation [74]. As for Ewing sarcoma, the effect of miR-31 was identified as a tumor suppressor. In cell lines transfected with miR-31, there was a significant reduction in Ewing sarcoma cell proliferation, as well as increased apoptosis resulting in a decrease in tumor invasiveness [75]. The expression level of miR-31 in cervical cancer, CIN, and normal uterine tissues was also examined. The results proved that the expression in cancer cells was elevated compared to non-tumor cells, and in addition, this phenomenon appeared to be associated with the occurrence of lymph node metastasis (LNM). It was demonstrated that such an effect increases cell proliferation, and cell migration, and thus invasiveness is also significantly increased, indicating that miR-31 acts as an oncogene in this case [76]. In patients with non-small cell lung cancer (NSCLC), a relationship was discovered between the expression levels of MIR31HG and, consequently, miR-31, and the proliferation and clonogenic growth of tumor cells. The study showed that reducing their expression in vitro also reduces the growth of NSCLC cells, supporting the fact that miR-31 had an oncogenic effect [77].
Moreover, miR-31 plays an important role in the induction of senescence in breast cancer cells. Obtained results showed that overexpression of miR-31 affected the repression of the polycomb group (PcG) protein BMI1, accompanied by the induction of cell aging. This offers the prospect of manipulating miR-31 expression to control senescence and oncogenesis in breast cancer [78]. Few studies are based on healthy tissues and the effects on the aging of non-tumorigenic cells. For example, miR-31 influences aging by regulation of dystrophin protein. Increased expression of miR-31 resulted in direct inhibition of the translation of this protein which causes aging and damaged processes in muscle cells [79]. Capri et al. examined the miRNA expression levels to determine the age match between donor and recipient in the case of hepatocytes. It was indicated that miR-31 is hyper-expressed in older donors, at levels up to 4.5 times higher than in younger individuals. Interestingly, these results were only noted in male patients, and female samples showed only a trend of increased expression with age [80]. Moreover, it was indicated in the case of endothelial cells (umbilical cord-derived), that the fold change of miR-31 was one of the most upregulated miRNAs during the aging process [81].
It should be noted that miR-31 regulates cell differentiation and has a role in determining cell fate. Li et al. investigated neural stem cells (NSCs) and the process of their differentiation into motor neurons (MNs) and correlated changes in miR-31 expression with different states. Initial studies showed high levels of expression of this molecule in NSCs derived, for example, from the spinal cord, while much lower levels in MNs. Comparison of the expression profiles led to the conclusion that high levels of miR-31 have a stemness-maintaining effect in NSCs by inhibiting differentiation, especially in MNs [82]. Neuronal precursor cells (NPCs) appeared to be another cell in which miR-31 plays a role in differentiation. It was shown that Lin28, c-Myc, SOX2, and Oct4 act to inhibit the activity and expression of miR-31, consequently, impairing the process of NPC differentiation into astrocytes and astrocyte maturation itself in gliomas. As later analyses showed miR-31 also downregulates selected stem cell factors mentioned above, which may suggest a reciprocal control of these molecules in astrocytogenesis, where miR-31 plays a key role [83].
However, based on the PubMed.org database, only 5 studies took into account both types of transcripts, miR-31 and MIR31HG. Tu et al. observed co-upregulation of miR-31 and MIR31HG in oral squamous cell carcinoma, with a linear correlation between both ncRNAs estimated to R = 0.304 and p = 0.047 in patients’ samples [84]. Another study, done by Chang et al. showed that artificial upregulation of MIR31HG caused the significant elevation of MIR31HG and miR-31 in two cancer cell lines [85]. Surprisingly, Qin et al., in lung adenocarcinoma, indicated no correlation between MIR31HG and miR-31, and the down-regulation of MIR31HG did not cause decreased levels of miR-31 [45]. Moreover, some evidence indicates that MIR31HG does not always regulate the level of miR-31. It was pointed out that MIR31HG regulates miR-361 and downstream targets influencing cellular phenotype of osteosarcoma cell lines [37].
It is surprising that for some studies the authors did not analyze the common correlation between miR-31 and MIR31HG, which makes it difficult to assess the actual interaction of these molecules with each other [86].
Not only miR-31, but also other miRNAs interact with MIR31HG
lncRNAs can not only affect protein-coding genes, but also other RNAs. The “sponge effect”, as it is commonly known, involves the action of lncRNA on individual miRNAs, which directly results in the reduction of its effect on most mRNAs [44]. Recent studies have shown that MIR31HG expression is markedly upregulated in pancreatic ductal adenocarcinoma (PDAC). However, knock-out of this molecule resulted in inhibition of PDAC cell growth, apoptosis, and ultimately reduced invasion. Here, the researchers noted the inverse correlation of MIR31HG and miR-193b in these cells. When miR-193b was overexpressed, MIR31HG levels decreased significantly and vice versa. Such results demonstrate the negative regulation of MIR31HG by miR-193b. Closer examination using a luciferase reporter and RIP assays suggested that miR-193b bound to MIR31HG by blocking the binding sites of this molecule to miRNA. This activity of both molecules speaks to MIR31HG acting as a sponge binding miR-193b to regulate miRNA targets [87].
Conclusions and future directions
Allis and Jenuwein named RNA “one of the master molecules of epigenetic control” [1]. Not so long ago, ncRNAs were perceived as genetic noise and ignored in all analyses and treated as background or experimental errors [5, 88]. The development of new techniques, such as massive RNA sequencing, bioinformatics and even artificial intelligence (AI), caused an unbelievable growth in the number of discoveries of different types of ncRNA molecules [89, 90]. The public release of data from the The Human Cancer Genome (TCGA) project made it possible to analyze the data in terms of searching for, for example, lncRNA, in 33 different cancers [91, 92]. The TCGA is an immeasurable source of knowledge used as the first source of data for selection of genes and later validated in in vivo and in vitro models. Unfortunately, not all studies adopt this model of analysis. The lack of validation based on an in vivo or in vitro model is questionable as an effect of errors accumulation or assumptions of mathematical models. On the other hand, results derived only from in vivo or in vitro studies can be misleading as the consequences of the cell lines used or the set of patients. It is essential to apply a holistic approach when a study is designed. The lack of comprehensive research causes the knowledge about specific types of ncRNAs still not to be fully defined or validated experimentally. Moreover, many different studies indicated that one lncRNA could behave differently depending on the cancer type [7]. In some cases only one study defines the role of lncRNA in a specific biological process or pathway, which should be verified by independent study. The lack of such an approach is also visible in the case of MIR31HG. The second important question concerns the potential use of lncRNA as a biomarker. lncRNAs have all the characteristics of biomarkers, but there are no standardized methods for their measurement and testing [10]. Another issue to consider is which transcript variant should be taken into account. As mentioned earlier, MIR31HG has 32 different transcriptional variants, with a length between 287 and 10’980 bp [29]. In some cases, not all variants are equal and they could be expressed in different ways depending on the specific cancer subtype, and have various diagnostic potencies [34]. Most studies regarding not only lncRNAs, but also other types of transcripts, do not explicitly point which transcript variant or variants are being analyzed. Moreover, this information is not included in the TCGA data either. This causes great difficulties in the adaptation of lncRNAs as biomarkers in clinical practice. However, many studies indicated that lncRNAs may become important tools to predict the development and eventual treatment of cancer in the future [10]. Will they be as useful as classic markers? We assume that the answer to this question will be known within the next decade.
Conflict of interest
Concerning publication of the article titled: Host gene and its guest: short story about relation of long-noncoding MIR31HG transcript and microRNA miR-31, written by Tomasz Kolenda, Anna Paszkowska, Alicja Braska, Joanna Kozłowska-Masłoń, Kacper Guglas, Paulina Poter, Piotr Wojtczak, Renata Bliźniak, Katarzyna Lamperska and Anna Teresiak, the authors declare:
- • the contents of this manuscript have not been copyrighted or published previously;
- • the contents of the manuscript are not now under consideration for publication elsewhere;
- • there is no conflict of interest including financial, personal or other relationships with other people or organizations regarding the publication of this paper;
- • this work was supported by Greater Poland Cancer Center — grant No.: 12/2021 (248), 10/11/2021/PGN/WCO/03 and other institutions listed in the acknowledgment.
Financial disclosure
Article titled: Host gene and its guest: short story about relation of long-noncoding MIR31HG transcript and microRNA miR-31, written by Tomasz Kolenda, Anna Paszkowska, Alicja Braska, Joanna Kozłowska-Masłoń, Kacper Guglas, Paulina Poter, Piotr Wojtczak, Renata Bliźniak, Katarzyna Lamperska and Anna Teresiak, was supported by the Greater Poland Cancer Center — grant No.: 12/2021 (248), 10/11/2021/PGN/WCO/03 and other institutions listed in the acknowledgment.
Funding
This work was supported by Greater Poland Cancer Centre — grant No.: 12/2021 (248), 10/11/2021/PGN/WCO/03
While writing this manuscript, Alicja Braska received a “The best in nature 2.0. The integrated program of the Poznań University of Life Sciences” scholarship from University of Life Sciences in Poznań; Joanna Kozłowska-Masłoń received a PhD program scholarship from Adam Mickiewicz University in Poznan; Kacper Guglas received a scholarship from the European Union POWER PhD program; Piotr Wojtczak is a PhD student at Doctoral Studies, Medical University of Lodz. We would like to thank all the institutions supporting our scientific work.