Background
Bladder urothelial carcinoma (BLCA) is the tenth most common cancer worldwide with an estimated 550,000 new cases and 200,000 deaths in 2018 [
1]. It is widely accepted that BLCA is a heterogeneous disease, which is classified into two distinct subtypes: non-muscle invasive bladder cancer (NMIBC) and muscle invasive bladder cancer (MIBC) [
2]. MIBC is responsible for most cases involving metastases leading to death, which arise from 10 to 20% of advanced NMIBC cases, and its clinical management remains limited [
3]. Currently, several classifications have proposed sets of molecular classes, including basal and luminal subtypes, which are partially characterized by
KRT5 and
KRT20 gene expression [
4]. In addition, epidermal growth factor receptor (EGFR) is expressed at high level in basal tumors, involved in controlling the basal gene signature [
4]. Although a number of groups have reported molecular classifications of BLCA to evaluate the severity and prognosis of this disease, reliable and effective biomarkers for early diagnosis and prognostic prediction are still lacking [
5,
6]. Thus, in-depth understanding of molecular events and underlying mechanisms involved in the carcinogenesis of MIBC may provide effective therapeutic targets and predictive biomarkers, which are urgently needed.
Long noncoding RNAs (lncRNAs) are a class of RNA transcripts, which are longer than 200 nucleotides and do not have protein-coding capacity [
7]. Recent studies have demonstrated that lncRNAs play essential roles in a wide range of biological processes, such as proliferation, apoptosis, cell cycle arrest, cell migration, and invasion [
8,
9]. Furthermore, it has been shown that several lncRNAs are deregulated in many tumors and may be involved in both carcinogenesis and cancer metastasis [
10]. Recently, the miR-31 host gene (
MIR31HG, also known as
LOC554202) has been identified in several cancers, such as breast, colorectal, gastric cancer, and pancreatic ductal adenocarcinoma [
11‐
14]. It is also reported that
MIR31HG expression was down-regulated in BLCA cell lines and tumor tissues [
15]. However, the functional role of
MIR31HG and its association with molecular classifications in BLCA are as yet unknown.
Despite the continuously growing knowledge on lncRNAs and cancer, a reliable clinical molecular marker has not yet been found. Considering that dysregulation of mRNA splicing can trigger cancer signaling pathways and contribute to almost all hallmarks of cancer [
16], the alternative splicing of lncRNAs may also impact cellular processes, which could open new possibilities for biomarker discovery. Previously, it was reported that splice variants of osteopontin have prognostic value in breast cancer [
17]. In BLCA, CD44 splice variants have been demonstrated to be involved in tumor progression and chemosensitivity [
18]. In addition, the lncRNA
PVT1 and its splice variant are highly expressed in clear cell renal cell carcinoma, and function as oncogenic transcripts [
19]. These findings advocate for the use of lncRNAs and their splice variants as tissue-specific transcripts and promising prognostic biomarkers in certain cancers. However, the precise function of most lncRNAs and the mechanisms of their molecular regulation remains to be elucidated.
In this study, the expression level and clinicopathologic significance of MIR31HG were first evaluated by in silico database analysis and then by qRT-PCR in a cohort from our institution. Upon knockdown of MIR31HG, a series of in vitro experiments was performed to investigate the effects of MIR31HG on proliferation, colony formation, and migration of BLCA cells. Transcript-specific knockdown with consecutive cell functional assays were performed to determine the role of MIR31HG splice variants and to investigate their association with molecular subtypes of MIBC. Further survival analyses were carried out to determine if MIR31HG and its splice variants could be used as prognostic biomarkers.
Discussion
This study aimed to investigate the expression pattern and biological function of MIR31HG in MIBC, as well as its clinical significance and prognostic value in patients. In order to evaluate its role in tumorigenesis and progression, functional in vitro assays were combined with lncRNA expression analysis based on molecular subtypes and relevant clinicopathologic parameters.
In previous studies on multifarious tumors,
MIR31HG showed a tissue-specific expression pattern. In breast cancer and non-small cell lung cancer (NSCLC) cells,
MIR31HG expression was upregulated [
11,
35]. In gastric cancer tissues and cell lines,
MIR31HG was poorly expressed [
13]. Another study showed that MIR31HG level is substantially upregulated in oral carcinoma, significantly associated with poor clinical outcomes and representing an independent prognostic predictor [
36]. In our study, lower
MIR31HG transcript levels were found in luminal-like and mixed-type BLCA cell lines compared with a normal urothelium cell line. Accordingly, down-regulated
MIR31HG expression was found in cancer tissues compared to normal tissues, which supports expression results measured by qPCR from a previous study [
15]. However, the previous results were measured in stage- and type-mixed BLCA tissues, and in this study,
MIR31HG was measured in MIBC and associated with multiple molecular subtype respectively. In contrast,
MIR31HG was found to be highly expressed in cells lines and clinical tumor samples with the basal subtype compared to luminal and other subtypes, indicating that
MIR31HG not only shows tissue specific, but also subtype-specific overexpression in MIBC.
In contrast to a previous study, which reported that
miR-31 and
MIR31HG are down-regulated in triple-negative breast cancer (TNBC) cell lines of basal subtype [
37], the present study shows that
MIR31HG is highly expressed in the BLCA cell line of basal subtype and markedly correlates with the survival of patients with MIBC basal subtype. This might be due to tissue specific expression of
MIR31HG. In this study, two MIBC cohorts with multiple molecular subtypes were involved rather than single cell line, which may also lead to the dissimilar results. Further studies of BLCA preclinical models are needed for validation. It is noteworthy that two transcript variants of
MIR31HG (
MIR31HGΔE1 and
MIR31HGΔE3) were identified and their expression was analyzed in BLCA cells and MIBC tissues. Besides the different expression levels in MIBC patient tissues, the two splice variants showed distinguished expression patterns in basal and luminal subtypes, respectively.
MIR31HGΔE3 showed high expression in the basal subtype, both in the TCGA cohort and the Mannheim cohort, which is also observed for the group with high
KRT5 expression. In contrast,
MIR31HGΔE1 showed high expression in luminal subtype tumors in the Mannheim cohort, corresponding to tumors with a high
KRT20 expression.
Additionally, higher
MIR31HG transcript levels were found to be associated with worse OS and DFS in the basal subtype cohort, but not in the whole TCGA cohort. It is the first time to discover the prognostic value of
MIR31HG in BLCA, or associated with subtypes of tumors. Furthermore, expression of
MIR31HGΔE1 and
MIR31HGΔE3 was significantly associated with OS and DFS in the Mannheim cohort, rather than the full-length transcript of
MIR31HG. In the TCGA cohort, it was demonstrated that
MIR31HGΔE3 expression was significantly associated with OS of the basal subtype group. These results, together with univariable and multivariable Cox regression analysis suggested that the alternative splice variants of
MIR31HG may serve as potential biomarkers for certain molecular subtypes of MIBC, which could contribute to an individualized bladder cancer subclassification and therapy decision making. It is worth noting that lymph node status is not associated with OS based on survival analysis, which is controversial to other studies [
38]. The possible reasons could be limited number of patients, diverse distribution of lymph node status and T stages as well as different scale of lymph node examination. Therefore, validation in larger patient cohorts with long-term follow-up is needed. In the Mannheim cohort, few patients were treated with neoadjuvant or adjuvant chemotherapy. For further validation, the number of positive lymph nodes and chemotherapy status should be also taken into consideration.
Two groups independently recognized the significance of a distinct basal MIBC subtype [
39,
40]. According to classification of the TCGA, basal tumors were divided into two subsets that were largely distinguished by differential expression of biomarkers associated with EMT, which is a reversible developmental process by promoting invasion, metastasis, “stemness”, and drug resistance [
32,
41]. The potential significance of the mesenchymal basal BLCA was identified using a “claudin-low” gene expression signature in breast cancer [
42]. A previous study reported that tumor suppressor microRNA-361 was de-repressed by
MIR31HG in osteosarcoma cells, leading to cell growth and mesenchymal phenotype [
43]. These results may indicate that high expression
MIR31HG could be served as a surrogate marker of poor outcome defined by relative activation of EMT and sponge of tumor suppressor. The discovery that
MIR31HG is highly expressed and significantly outcome-correlated in MIBC with basal subtype, has complemented the selection of MIBC markers.
Due to its potential clinical relevance as a marker, the expression of
MIR31HG and its underlying biological functions could be vital to the tumor. In this study, by knocking down
MIR31HG expression using siRNA, diminished cell proliferation, colony formation, and migration were assessed in BLCA cell lines. This study is the first to highlight the function of
MIR31HG in BLCA cells, which indicates that
MIR31HG might serve as an oncogene in certain types of BLCA. With the in-depth study of MIR31HG, several downstream targets were found in present researches. For example, it is reported that overexpression of
MIR31HG significantly decreased the expression of miR-575, enhanced the suppression of tumorigenicity 7 like (ST7L) in hepatocellular carcinoma (HCC). Thus,
MIR31HG regulated
ST7L expression through sponging miR-575, and acted as tumor suppressor in HCC [
44]. Another published research showed that the level of
MIR31HG in esophageal squamous cell carcinoma (ESCC) tissues was positively correlated with the expression of furin and matrix metalloproteinase 1 (MMP1). When
MIR31HG was silenced, the expressions of furin and
MMP1 in ESCC cells were significantly inhibited. These results suggest that the involvement of
MIR31HG in invasion and migration of ESCC cell may be partly achieved through the furin /
MMP1 pathway [
45]. In a study in oral cancer,
MIR31HG was identified as a hypoxia-inducible lncRNA and forms a complex with hypoxia-inducible factor-1 α (HIF-1α), thus as an adverse prognostic predictor for the cancer progression [
36]. In addition, it was reported that
MIR31HG could function as an oncogene that promotes pancreatic cancer progression, by acting as an endogenous sponge competing for miR-193b [
14]. Similarly, it was shown that silencing of
MIR31HG significantly inhibited NSCLC cell migration, invasion, and metastasis by attenuated sponging of miR-214 [
35]. The various mechanisms of
MIR31HG suggest that downstream pathways could be involved with other non-coding RNAs, which requires further verification. In this study, similar to the full-length transcript, a series of functional experiments validated the corresponding roles in certain types of BLCA cells. The knock-down of
MIR31HGΔE3 resulted in reduced colony formation ability and cell viability solely in SCaBER cells, suggesting that exon-specific or transcript-specific mechanisms could be a new direction in studying basal-like BLCA. Recently, emerging data have demonstrated that RNA splice variants are associated with drug resistance in cancer. The expression of androgen receptor (AR) splicing variants in castration-resistant prostate cancer (CRPC) samples increased significantly. The most common AR splicing variants are AR-V7 and ARv567es [
46,
47]. These variants are important factors for insensitive to AR antagonists of CRPC patients. Furthermore, splice variants in V600E BRAF-mutant-positive malignant melanoma patients were shown to be associated with vemurafenib resistance, indicating that aberrant splicing could be a novel mechanism of acquired resistance [
48]. Furthermore, interference of the pre-mRNA splicing modulators sufficiently inhibited formation of these splice variants, suggesting that splice variant-specific siRNAs may be proposed as a therapeutic strategy to overcome drug resistance [
49].
For basal or squamous-like bladder cancer, molecular signatures were found based on clustering RNA-seq data, including
EGFR [
34].
EGFR is overexpressed in up to 74% of bladder cancer tissue specimens, and is amplified in squamous cell carcinomas (SCC) of the bladder [
50]. To further investigate the potential relationship between
MIR31HG and
EGFR, a computational method called lncPro was applied to predict the associations within. All the nine isoforms of EGFR, as well as PI3K and HER2 protein, were predicted as interactive with
MIR31HG. The positive interaction score suggested that
MIR31HG might be involved in the EGFR/PI3K/AKT signaling pathway. Furthermore, expression of
EGFR was detected in three
MIR31HG-knockdown BLCA cell lines. In SCaBER cells, which shows the basal / squamous signature, expression of
EGFR was reversely correlated with
MIR31HG. According to another study on lung cancer, by knocking down
MIR31HG, reversal of gefitinib resistance was found by regulation of the EGFR/PI3K/AKT signaling pathway [
51]. Taken together, these findings suggested that
MIR31HG might potentially correlate with the EGFR pathway.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.