Background
Gastric cancer (GC) is one of the leading causes of cancer morbidity and mortality worldwide, especially in Eastern Asian and Eastern European countries [
1‐
3]. Patients in advanced stages benefit less than expected from palliative chemotherapies, mainly due to tumor metastasis and chemoresistance [
4,
5], although the underlying molecular mechanism remains largely unknown. In our previous studies, the polycomb-group protein B cell-specific Moloney murine leukemia virus integration site 1 (Bmi-1) was reported to be associated with tumor size, clinical stage and prognosis of GC [
6,
7]. Furthermore, we demonstrated that Bmi-1 is inversely associated with Raf kinase inhibitory protein (RKIP), a regulator of apoptosis induced by chemotherapeutic agents and a clinically relevant cancer metastasis suppressor gene [
8,
9]. Based on previously reported results, the negative correlation between Bmi-1 and RKIP was deemed valuable in predicting patient survival and therapeutic response in GC [
10]. The inverse expression pattern of Bmi-1 and RKIP was confirmed in GC cell lines by utilizing in vitro gene overexpression and silencing methods, suggesting the likelihood of RKIP being regulated by Bmi-1. However, the regulatory mechanism between Bmi-1 and RKIP needs further elucidation.
A variety of factors may be involved in the mechanism regulating the expression of Bmi-1 and RKIP. Among all potential regulatory mechanisms, microRNAs (miRNAs) emerged as a top research target due to their significance in various biological activities. miRNAs are an endogenous group of small noncoding RNAs that regulate up to 60% of human protein-coding genes at the posttranscriptional level by binding to the 3′ untranslated region (3’UTR) of a target mRNA [
11‐
13]. miRNAs are able to affect tumor cell proliferation, invasion, metastasis and chemoresistance by regulating gene expression. For instance, Bmi-1 upregulates miR-21 and miR-34a in addition to regulating GC stem cell-like properties via the activation of the AKT-NF-κB pathway [
14]. RKIP is also reported to be directly regulated by miR-543, which controls cell proliferation and metastasis in human prostate cancer cells [
15]. Given the importance of miRNAs, the dysregulation of miRNAs is thought to be closely related to metastasis and chemoresistance in GC. We hypothesized that miRNAs may account for the negative correlation between Bmi-1 and RKIP. Therefore, this study aimed to explore Bmi-1-induced miRNAs that regulate RKIP. The candidate miRNAs were identified and predicted and were then validated by in vitro and in vivo experiments.
Overall, a novel microRNA regulatory mechanism of the Bmi-1-RKIP signaling axis was found. The inhibition of the RKIP tumor suppressor and a mechanism underlying tumor metastasis and resistance to chemotherapy in GC were also elucidated in this study. These findings may be significant in terms of screening, diagnosis, prognosis, disease monitoring and therapeutic value in GC.
Methods
Tissue specimens
Fifteen pathologically confirmed GC specimens with tumor and adjacent normal paired fresh-frozen tissues were obtained at Sun Yat-sen memorial hospital, Sun Yat-sen University (Guangdong, China). No preoperative treatments were received. Total RNA and protein of frozen tissues were extracted for quantitative real-time PCR (qRT-PCR) and Western blotting assays. The clinical specimens were obtained with patients’ informed consent and permission from the Institutional Ethics Committee of Sun Yat-sen memorial hospital. RNA-seq data, clinicopathological information and follow-up data of GC were downloaded from The Cancer Genome Atlas (TCGA) database (
https://www.cancer.gov). Patients with censored overall survival time, gene expression profiles or certain pathological classification were excluded. The cut-off values of Bmi-1, miR-27a-3p (miR-27a), miR-155-5p (miR-155) and RKIP expression in Kaplan-Meier analysis were determined by X-tile software [
16,
17].
Cell lines
The human GC cell lines (BGC823 and SGC7901) were obtained from the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). The human gastric epithelial immortalized GES-1 cell line was purchased from Beijing Institute for Cancer Research. The human leukemia monocytic U937 cell line was purchased from American Type Culture Collection (ATCC, Washington, USA). All these cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, MA, USA) supplemented with 10% fetal bovine serum (FBS, Biological Industries, Beit Haemek, Israel) and incubated at 37 °C in a humidified atmosphere of 5% CO2.
Plasmids, siRNAs and stable cell lines
pLNCX2-Bmi-1 was constructed as previously described [
7]. Stable cells overexpressing Bmi-1, SGC7901-Bmi-1, BGC823-Bmi-1, GES-1-Bmi-1#1 and GES-1-Bmi-1#2 were generated by retroviral transfection as described in our previous study [
7,
10]. The Bmi-1 overexpressing cells were also named Vector-Bmi-1, and the negative control was named Vector-Ctrl. Lentiviral shRNA virus specific to human Bmi-1 and Bmi-1-specific siRNAs were purchased from GenePharma (Shanghai, China). The shRNA sequences (the same as the specific siRNAs) for targeting Bmi-1 are shown in Supplementary Table S
1. To construct cell lines for constitutive miRNA expression, lentiviruses containing GFP-miR-27a, GFP-miR-155, or GFP-negative control miRNA vector were purchased from GenePharma, Inc. The lentiviral vectors were used to infect targeted cells. BGC823 and SGC7901 cells were pre-seeded in a 6-well plate overnight and infected with 10 μl of virus. Infected cells were selected by adding 400 ng/ml puromycin for 5 days and then transferred to cultured flasks for proliferation. Stable cell lines were verified by qRT-PCR.
Microarray data processing
The Bmi-1-overexpressing GES-1 cell line GES-1-Bmi-1#1 and its vector control cells were harvested with TRIzol, and miRNAs were extracted using a miRNeasy mini kit (QIAGEN, Beijing, China). Then, samples were sent to KangChen Bio-tech Inc. (Shanghai, China), quantified and analyzed the differential expression of miRNAs using a miRNA microarray. Microarray hybridization, data generation, and normalization were performed following standard protocols. Finally, differentially expressed miRNAs were identified through fold-change filtering. The cut-off fold change was 2.0.
Luciferase assay
The RKIP 3’UTR was amplified by PCR separately using the primers described in Supplementary Table S
1 from cDNA of SGC7901 cells. The PCR product was ligated into the multiple cloning region of the pGL3 luciferase reporter plasmid (Promega, Wisconsin, USA) according to the manufacturer’s recommendations. 293T cells plated in 96-well plates at a density of 4 × 10
4 cells per well were cotransfected with 100 ng of the constructed luciferase plasmid or the control luciferase plasmid and 15 ng of the pRL-TK Renilla plasmid (Promega) using the Lipofectamine 2000 reagent (Thermo Fisher Scientific, MA, USA). miR-27a mimic/mut and miR-155 mimic/mut (50 nmol/L) were then cotransfected with the luciferase plasmid containing the RKIP 3’UTR for microRNA detection. After 24 h, cells were lysed and detected for Renilla and firefly luciferase activity using the Dual Luciferase Reporter Assay Kit (Promega). Three independent cotransfection experiments were carried out in triplicate.
RNA electrophoretic mobility shift assays
All oligonucleotides and primers described as follows were obtained from GenePharma (Shanghai, China). The oligonucleotides hsa-miR-27a: 5′-UUCACAGUGGCUAAGUUCCGC-3′ and hsa-miR-155: 5′- UUAAUGCUAAUCGUGAUAGGGGU-3′ were synthesized and 5′-labeled with Cy5 dye. The 2′-O-methyl-modified RNA oligonucleotides RKIP-miR-27a: 5′-GGGGUAUUUUGGUACUGUGAU-3′ and RKIP-miR-155: 5′-AGUUGCUGAAUGUUGCAUUAAU-3′, which correspond to the hsa-miR-27a and hsa-miR-155 targeting sequences present in the 3’UTR of RKIP, were 5′-labeled with FAM dye. Cytoplasmic extracts of SGC7901 cells were prepared using NEPER Nuclear and Cytoplasmic extraction reagents (Thermo Scientific, MA, USA). The RNA EMSA was carried out according to the manufacturer’s instructions (Thermo Fisher Scientific) and our previous studies [
18,
19].
Animal experiments
All mouse experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals by the National Institutes of Health and were approved by the Animal Care and Use Committee of Sun Yat-sen University. To study the effect of the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP axes on GC tumorigenesis, chemoresistance and metastasis, we conducted tumor xenograft and tail vein injection studies in mice.
For the tumorigenesis experiments, BGC823-shcon, BGC823-shBmi-1, BGC823-NC mimic, BGC823-miR-27a mimic, BGC823-miR-155 mimic, BGC823-shBmi-1 + NC mimic, BGC823-shBmi-1 + miR-27a mimic and BGC823-shBmi-1 + miR-155 mimic cells (1.5 × 106 cells in 0.15 ml of PBS) were subcutaneously inoculated into the right flanks of female BALB/c nude mice (5 weeks old) to establish tumor xenografts. Tumor size was monitored and calculated every 3 days. The tumor volume was estimated using the equation: tumor volume (mm3) = (length in mm) × (width in mm)2/2. In the chemoresistance experiments, tumor-bearing xenografts were established the same as in the tumorigenesis experiment. Once the subcutaneous tumors grew big enough to be palpable, intraperitoneal (IP) injection of either 5-Fu (15 mg/kg) or vehicle control (PBS) was performed every other day for 14 days. On the 21st day after implantation, mice in the tumorigenesis and chemoresistance groups were sacrificed. For the tumor metastasis experiment, 1 × 106 cells were injected into mice via the tail vein. The mice were sacrificed 30 days after cell administration. The tumors, lungs or livers dissected in all animal experiments were fixed and paraffin-embedded for histopathological analysis.
Immunohistochemistry
Paraffin-embedded tissue sections that were 4 μm thick were used in immunohistochemistry. Following deparaffinization, hydration, antigen retrieval and blocking, the slides were successively incubated with primary antibodies at 4 °C overnight and secondary antibodies conjugated with horseradish peroxidase (HRP) at 37 °C for 30 min. The slides were then incubated in DAB solution for 10 min and counterstained with hematoxylin for 1 min. Images of the stained slides were obtained with a microscope (Nikon, Tokyo, Japan).
Statistical analysis
All statistical analyses were performed with Statistical Product and Service Solutions version 22.0 (SPSS, Illinois, USA) and GraphPad Prism 5.0 software (GraphPad, Inc., CA, USA). Data are presented as the mean ± standard error (−x ± SE). The statistical significance of the differences was determined by t-test and one-way ANOVA test according to the homogeneity of variances. P < 0.05 was considered to be statistically significant.
The detailed methodology of qRT-PCR, Western bloting and functional experiments can be found in the
Supplementary Materials.
Discussion
Recently, therapies targeting metastasis and chemoresistance of GC have attracted the attention of researchers [
22,
23]. However, insufficient understanding of the underlying molecular mechanisms involved creates a barrier for the development of effective strategies in clinical practice. We previously reported the distinctive biological activity of Bmi-1/RKIP and their inverse relationship in GC, but the detailed mechanisms needed further investigation. In this study, we screened for Bmi-1-induced miRNAs and successfully identified the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP signaling axes. For the first time, we elucidated the molecular mechanism of how Bmi-1 regulates RKIP in GC, which ultimately affects the metastasis and chemoresistance of GC. Our results suggest that therapies targeting the Bmi-1/miR-27a/RKIP and Bmi-1/miR-155/RKIP signaling axes would be effective in GC patients, especially those with a high risk of metastasis and chemoresistance.
RKIP is a proven metastasis suppressor protein in various human cancer types and is linked to an invasive phenotype [
24,
25]. Low levels or the absence of RKIP expression in GC is associated with poor patient prognosis [
10,
26]. However, the molecular mechanisms of how RKIP expression is downregulated in GC remain unknown. Polycomb group (PcG) proteins are epigenetic gene-silencing proteins that have been shown to play important roles in human cancer occurrence and progression [
27,
28]. Bmi-1, the first functionally identified PcG member, is frequently dysregulated in various cancers and strongly correlates with tumor aggressiveness; thus, its presence predicts a poor prognosis [
10,
29,
30], but little attention has been paid to the downstream regulatory mechanism of Bmi-1 in GC promotion. The opposite biological functions and negative correlation between Bmi-1 and RKIP suggested a potential regulatory mechanism in GC. The results from our previous study led to the likely hypothesis that RKIP is regulated by Bmi-1 [
10].
To directly examine the role of Bmi-1 in RKIP expression, we introduced a Bmi-1 expression plasmid by retroviral infection into the nonmalignant human gastric epithelial GES-1 cell line. RKIP protein expression was suppressed by ectopic expression of Bmi-1 in GES-1 cells [
10]. However, forced expression of Bmi-1 barely changed the mRNA level of RKIP (Fig. S
2A). Therefore, RKIP protein expression is likely to be regulated at the posttranscriptional level in GC cells. This result is in line with previous reports that have suggested the critical role of posttranscriptional mechanisms in the regulation of RKIP expression in human hepatocellular carcinoma [
31]. Mechanisms such as protein degradation and microRNA-mediated suppression may also play a role in the posttranscriptional regulation of the expression of specific genes. Moreover, protein half-life analysis demonstrated that forced expression of Bmi-1 did not induce RKIP protein degradation (Fig. S
2B). On the other hand, microRNAs are RNAs with sizes between 20 and 25 nucleotides, and they contribute to gastric carcinogenesis by regulating the expression of oncogenes and tumor suppressors at the posttranscriptional level to affect cell proliferation, apoptosis, motility and invasion [
32‐
34]. microRNAs have been shown to be involved in regulating RKIP expression in human cancer [
24,
35]. To our knowledge, this study is the first to report a difference in miRNA expression based on the overexpression of Bmi-1 in GES-1 cells. Potential miRNAs targeting RKIP were selected using miRNA microarray and publicly available algorithms (TargetScan). Upon combining the outcomes of the literature review, miR-27a and miR-155 were identified as candidate microRNAs that are involved in the Bmi-1 repression of RKIP.
In this study, we demonstrated that Bmi-1 functions as a repressor of RKIP, which is mediated by miR-27a and miR-155. miR-27a and miR-155 were also able to directly repress RKIP expression, which consequently affected the expression of genes necessary for regulating metastasis and chemoresistance, including Vimentin, Cadherin 1 (E-cadherin), Bax and Bcl-2. Through in vitro and in vivo experiments, it was observed that Bmi-1-downregulated cells showed slower growth, poorer migration and invasion capacities, and a clear sensitivity to in vitro drug treatment. Further in vivo experiments in mice demonstrated that miR-27a and miR-155 downregulated RKIP and promoted metastasis to the lungs and livers. The level of inhibition to oxaliplatin and 5-Fu sensitivity was also greater than that of the control. Furthermore, the miR-27a mimic and miR-155 mimic could significantly counteract the metastasis and chemoresistance effects in Bmi-1-downregulated cells. Similarly, the miR-27a inhibitor and miR-155 inhibitor weakened the effects of Bmi-1 overexpression on migration, invasion and drug resistance in gastric cancer cells. The results were consistent with previous results that highlight the significant role of RKIP in altering tumor cell metastasis and chemoresistance both in vitro and in vivo. In addition, we observed differences in miR-27a mimic- and miR-155 mimic-induced cell proliferation and the size of the tumors when compared to control cells. This implies that there may be other pathways influencing cancer growth simultaneously. Collectively, these findings strongly indicate the pivotal role of RKIP in Bmi-1-mediated promotion of metastasis and suppression of chemosensitivity.
Some researchers have reported that miR-27a is highly expressed in GC tissues and that the overexpression of miR-27a promotes the tumorigenicity, metastasis and chemoresistance of GC [
36‐
38], which is in accordance with this study. In addition, previous studies illustrated that increased expression of miR-155 was closely related to tumor invasion and metastasis in advanced GC, implying poor prognosis [
39]. Qu Y et al. showed that miR-155 downregulated the expression of TGFβR2 to promote the proliferation and migration of GC cells, which is consistent with our study [
40]. Moreover, it was reported that
Helicobacter pylori infection was closely linked to miR-155, possibly upregulating miR-155 expression to inhibit the DNA mismatch repair (MMR) gene and induce a mutant phenotype that is conducive to error-prone translation synthesis and thus promotes GC progression [
41‐
43]. However, some studies have reported that miR-155 is expressed at low levels in GC tissue and acts as a tumor suppressor gene [
44‐
46]. Combined with the prognostic analysis of miR-155 shown in Fig.
1f, we propose that miR-155 plays different roles in GC of different histological types. Therefore, the occurrence and development of GC is complex. Although the carcinogenic role of miR-27a and miR-155 in GC has been reported, our study demonstrates that miRNA as a key junction plays a posttranscriptional regulatory role in the Bmi-1/RKIP pathway, further revealing the specific molecular mechanism of GC metastasis and chemoresistance.
Previous published literature illustrates that GC is histologically complex and can be characterized by the expression profile of microRNAs. It was reported that miR-105, miR-145, and miR-133a were upregulated in diffuse-type lesions, while miR-498 and miR-494 were upregulated in intestinal-type GC [
47,
48]. We analyzed the clinical significance of miR-27a and miR-155 from TCGA and found that these two indicators were not identical in different histological types, suggesting that these two indicators could be signatures linked to the tumorigenesis and development of GC. Therefore, we need to include a larger patient population and collect follow-up information to clarify the correlation between miR-27a, miR-155 and clinical prognosis in further studies. Moreover, we will verify the expression of miR-27a and miR-155 and its clinical significance in different histological types.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.