Background
Pancreatic cancer (PC), the mortality of which ranks third among all cancers globally, is extremely fatal [
1]. Pancreatic carcinoma, which accounts for approximately 95% of various pancreatic cancer types, is mainly divided into pancreatic ductal adenocarcinoma and other types of pancreatic adenocarcinoma [
2]. So far, many studies on pancreatic cancer have focused on drug therapy for cancer-related genes. For example, pro-oxidant drugs can treat pancreatic cancer with p53 mutation [
3], and troglitazone can enhance the anticancer effect of IFN-β through the interaction of STAT-3 dependent survival pathway and directly induce the increase of p21 and p27 expression [
4]. Currently, the primary treatment for pancreatic cancer includes a combined strategy of surgery, radiotherapy, chemotherapy and targeted therapies [
5]. Despite great progress in timely monitoring, diagnosis and early intervention, the mean survival time has remained unchanged in the last few decades, while an estimation of death in the following decades marks pancreatic cancer as an important cancer-related cause of death globally [
6]. In addition, patients with metastases and advanced stages have a worse prognosis [
7]. All in all, further exploration of the fundamental mechanism underlying pancreatic cancer and recognition of more treatment target points are of great importance.
Since they were first reported in nematodes, microRNAs (miRNAs) have received significant attention as essential regulators of organic processes in living organisms [
8]. miRNAs, which are endogenous, approximately 22 nt RNAs, mostly target mRNAs and inhibit biological translation in animals, plants and bacteria [
9]. Consequently, a variety of miRNAs can form comprehensive networks that regulate cell functions, such as growth, proliferation, and apoptosis [
10]. miRNA dysregulation is conspicuously associated with a significant number of human diseases, especially cancer [
11]. Many disease-specific miRNAs display a capacity to become new cancer biomarkers in diagnosis, intervention, and outcome determinations [
12]. Lately, great attention has been focused on the effects of miRNAs in tumour generation and development.
MicroRNA-221-3p (miR-221-3p), one of the cancer-related miRNAs, possesses various functions in numerous cancer types. miR-221-3p serves as a cancer-promoting factor in liver carcinoma [
13], colorectal tumours [
14], breast neoplasms [
15], lung tumours [
16], and prostate carcinoma [
17] but as a suppressive factor in gastrointestinal stromal tumours [
18] and cholangiocarcinoma [
19]. Additionally, miR-221-3p might act as an underlying biomarker and target [
20]. Despite all this, the clinicopathological value and integrated mechanism behind the role of miR-221-3p in pancreatic cancer pathogenesis, proliferation capacity, invasion ability, drug resistance, and apoptosis remain mostly unknown.
Therefore, our research aims to explore how miR-221-3p expression relates to clinicopathological features and how miR-221-3p affects pancreatic cancer. Importantly, to explain potential downstream networks, the mechanism underlying miR-221-3p activity in pancreatic cancer was analysed via target gene prediction, bioinformatics analyses and related protein experiments.
Materials and methods
Acquisition of data
Bioinformatics data associated with miR-221-3p in pancreatic cancer were derived from the NCBI-GEO database (
http://www.ncbi.nlm.nih.gov/geo/) according to the following terms: (“small RNA” OR “nc RNA” OR “non-coding RNA” OR miRNA OR “micro RNA”) AND pancreatic AND (cancer OR tumour OR neoplasm OR malignancy OR carcinoma OR adenocarcinoma OR PC OR PaC OR PDAC). The labels “series” and “Homo sapiens” were checked.
Other qualifications were as follows: (1) patients diagnosed with pancreatic cancer along with confirmed histological subtypes; (2) inclusion of both cancer and non-cancer groups; (3) cancer and non-cancer samples contained a minimum of different sources, including tissue, plasma, or blood; (4) the miR-221-3p content in cancer or normal samples was available. Data from relevant studies were derived from PubMed, Web of Knowledge, CNKI, and OVID. Additional file
1: Fig. S1 demonstrates the working wireframes in this study.
miR-221-3p content data derived from TCGA
Informative data regarding miR-221-3p expression in pancreatic cancer samples were acquired from TCGA (
https://cancergenome.nih.gov/) with the following keywords: (Primary Site is pancreas) and (Experimental Strategy is miRNA-Seq). The differences in miR-221-3p content between pancreatic cancer and associated controls were calculated using GraphPad Prism 8.0.2 software.
RNA isolation and real-time quantitative PCR
Briefly, 16 paired samples taken from patients at the Department of Gastroenterology of the Ruijin Hospital of Shanghai Jiao Tong University were selected. This research was approved by the ethics committee of Ruijin Hospital. Total RNA was isolated from tissues with TRIzol Reagent (Invitrogen, Carlsbad, CA). RT-qPCR analysis for determination of miR-221-3p content was employed with LightCycler
® 96 SW 1.1 Real-Time PCR System software and TaqMan miRNA probes (Applied Biosystems, Foster City, CA) according to the manufacturer’s protocol. Expression profiles were normalized to a miRNA RT-qPCR Standard, U6 snRNA or GAPDH primer. The sequences of probes are presented in Additional file
1: Table S6.
Statistical process and systematic meta-analysis
After log2-transformation and calculation via GraphPad Prism 8.0.2 software, the miR-221-3p content profiles of each pancreatic cancer and control dataset were described as the mean (M) ± standard deviation (SD). An integrated meta-analysis was conducted according to the different data origins (GEO, TCGA, literature, and RT-qPCR) using StataSE 12.0 software. To analyse miR-221-3p content in pancreatic cancer and non-cancerous samples, forest plots were utilized to depict standardized mean difference (SMD) with a 95% confidence interval (CI). To determine sample heterogeneity and estimate the effectiveness of the pooling method, a Chi squared test was applied to test the Q and I2 statistic values. A funnel plot was constructed for assessment of publication bias. All significance was defined as p < 0.05 and is displayed using asterisks (*p < 0.05, **p < 0.01, ***p < 0.001 for cell experiments.
Cells
The pancreatic cancer cell lines BxPC-3 and MIA PaCa-2 were purchased from Shanghai Institute of Cell Biology (Shanghai, China) and incubated at 37 °C with humidified 5% CO2. BxPC-3 cells were grown in RPMI-1640 (Gibco; Life Technologies, Carlsbad, CA, USA) medium supplemented with 10% FBS (Genial Biological Inc; Colorado, USA), while the MIA PaCa-2 cells were cultured in DMEM (Gibco; Life Technologies, Carlsbad, CA, USA) supplemented with 10% FBS.
Cell transfection
PC cells were seeded on plates, allowing approximately 20 h of growth. Transfection was implemented with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Cell treatments were divided into 5 groups: treatment with mimic or inhibitor of mir-221-3p and the respective controls and treatment with only Lipofectamine 2000. For genetic manipulation, cells were transfected with mimic at 100 pmol to induce miR-221-3p overexpression or with inhibitor at 100 pmol to suppress miR-221-3p. At 48 h post-transfection, the cells were harvested for the following measurements.
Cell proliferation assay
PC cells cultured onto 6-well plates were counted using a Celigo® Image Cytometer (Nexcelom, USA) after 6 h of transfection and then were seeded at 6000 cells/well into 96-well plates. After 24 h, 10 μl CCK-8 reagent (Dojindo, Tokyo Japan) was added to each well in the 96-well plates and allowed to react for 2 h. After 2 h of incubation, the absorbance at 450 nm (OD value) was measured. A CCK8 assay was conducted every 24 h for 96 h. All experiments were performed independently five times. During data analysis, the average value of each well at every time point was recorded, normalized by the value at the 0th h, and then analysed.
Cell wound-healing assay
PC cells were cultured in 6‐well plates and grown in medium with 2% FBS to limit the effect of wounding-related cell proliferation. After cells were grown to confluence, scratches were created in the confluent cell single layers by employing a sterile 200 μL pipette tip. At time points of 0, 24, and 48 h, the linear wound was monitored and photographed under a microscope (Olympus, Japan). The statistical assessment of the wound-healing rate was made using ImageJ software.
Cell migration and invasion assays
The ability of PC cells to migrate or invade was verified in Transwell Permeable Supports (6.5 mm Insert; Costar, Cambridge, Mass). After 24 h of reaction, cells were resuspended in DMEM without FBS at 5 × 104 cells/mL and seeded in the upper chamber at 5 × 103 cells per well, while the lower well was filled with 500 μl of DMEM supplemented with 20% FBS. Then, the plate was placed at 37 °C with humidified 5% CO2. After approximately 1 day of culture, the cells that migrated through the filter membrane were fixed with 4% polyformaldehyde for 20 min at room temperature, following by rinsing with PBS. Cells were stained for 20 min with 0.1% crystal violet at room temperature. Cotton buds were utilized to wipe away residual cells left on the lower cover of the chamber. Images were obtained with Image-Pro Plus (IPP) 6.0 after the above steps. For each treatment group, 3 individual fields of cells were counted to evaluate cell transferability. Cell invasion assays were performed following the same steps used to conduct the cell migration assay, with the only difference being that the bottom of the chamber was precoated with 10 μg/mL fibronectin gel.
In vitro cytotoxicity tests
Gemcitabine was acquired from ApexBio Technology Company (Boston, Indiana State, USA). PC cells were plated in 96-well plates at 8000 cells/well, with five replicates. Following 6 h of incubation, gemcitabine at different concentrations was added to the two cell lines: BxPC-3 cells from 0 to 20 μM and MIA PaCa-2 cells from 0 to 50 μM, and incubation was continued for 48 h. Cell viability was evaluated with CCK8 assays, and the results were determined by measurement of absorbance of 450 nm. Then, 50% inhibition (IC50) was computed with GraphPad Prism 8.0.2 software.
Flow cytometry
Apoptosis was assessed with an Annexin V-FITC/Propidium Iodide (PI) Apoptosis Detection Kit (eBioscience, 88-8005, USA) following the standard protocol. PC cells were resuspended in 300 μl binding buffer and mixed with 5 μl fluorochrome-conjugated Annexin V-FITC. Following 15 min of room temperature reaction in the dark, PC cells were suspended in 200 μl binding buffer with 5 μl propidium iodide. The results were obtained using FlowJo software.
Underlying targets of miR-221-3p in pancreatic cancer
Data from MiRWALK2.0, a public database that includes targets of miRNAs [
21], were extracted for predicting underlying miR-221-3p targets. The sum of results from 7 online tools, namely, miRWalk, GEPIA, Mirtarbase, TargetScan, PICTAR2, miRanda, and miRDB, was used. Target genes were defined as genes that occurred in the results of no less than 6 tools. The downregulated genes in pancreatic cancer, pancreatic adenocarcinoma, pancreatic ductal adenocarcinoma and other types of pancreatic adenocarcinoma were obtained from Gene Expression Profiling Interactive Analysis (GEPIA). Potential miR-221-3p targets in pancreatic cancer were genes that shared cross target gene sets forecasted by online tools and downregulated gene sets in pancreatic cancer. Articles were searched to identify additional reported possible miR-221-3p targets in the specific field of pancreatic cancer to prepare a further analysis. Therefore, underlying targets, which were selected for functional analysis, were defined as a combination of target genes revealed by previous studies and forecasted by online tools.
Functional analysis of underlying targets
Metascape (
http://metascape.org/gp/) [
22] includes several items from the GO project, including biological processes (BPs), cellular components (CCs), and molecular functions (MFs). For deciphering potential target functions, KEGG pathway analysis was conducted with the Metascape tool. Moreover, based on STRING, a website aiming to elucidate the integrated function of multiple genes [
23], a PPI network was generated for identifying important targets.
Luciferase
A luciferase reporter gene experiment was performed to verify whether the combination of miR-221-3p and mRNAs of these hub genes was achieved through matching of the miR-221-3p seed sequence and the mRNA binding site. We inserted a fragment containing the miR-221-3p binding site or mutated sequences of the binding site on the 3′-UTR of KIT, CDKN1B, RUNX2 or BCL2L11 into the pMIR-REPORT Luciferase vector (Nanjing Kingsley). PC cells were transfected with luciferase and miR-221-3p mimic, inhibitor or controls. Promega’s luciferase activity test kit was used for the luciferase experiment. In addition, we used a β-galactosidase reporter plasmid (β-gal) as a control and co-transfected cells with the luciferase plasmid.
The culture medium in the cell well plate was aspirated, PBS was added and then removed to rinse the cells, and 500 μL of trypsin was added for digestion. After digestion, an equal volume of medium with 10% FBS was used to stop the digestion. The cells were collected and centrifuged at 300 g for 4 min, the supernatant was removed, and PBS was added to resuspend the cells. The cells were centrifuged at 300g for 4 min, and the supernatant was removed. Next, 80 μL of premixed protease inhibitors (PMSF and PI) was added into RIPA lysate. The cells were mixed by pipetting, placed on ice for 40 min, and centrifuged at 12,000g for 10 min at 4 °C. The supernatant was aspirated into a new EP tube, and 20 μL of 5 × SDS loading buffer was added to the protein, which was then heated at 99 °C for 9 min and stored at − 80 °C.
Western blotting
Constant voltage mode was used for electrophoresis. When the protein was concentrated, the voltage was constant at 80 V. When the sample ran into the separation gel, the voltage was adjusted to 120 V. The separated proteins were transferred to a PVDF membrane, which was then blocked with 5% skim milk in TBST. The target protein and internal control bands were incubated separately in antibody at 4 °C overnight (Anti-KIT Antibody, Abcam, 32363, 1:1000 dilution; Anti-CDKN1B Antibody, Abcam, 32034, 1:1000 dilution; RUNX2 Polyclonal Antibody, SAB, 41746, 1:500 dilution; Anti-BCL2L11 Antibody, Abcam, 32158, 1:1000 dilution; Anti-GAPDH antibody, Abcam, ab181602, 1:1000 dilution). The bands were washed on a shaker with TBST solution and then incubated with the appropriate HRP-conjugated antibody (Abcam, ab205718, 1:2000 dilution) on the shaker for 1 h at room temperature. Finally, the bands were washed on the shaker with TBST solution strips; and using Supersignal West Pico chemiluminescent substrate from Thermo, the strips were exposed in an exposure machine.
Discussion
Non-coding RNA is increasingly being studied extensively. In different types of tumors such as hematological malignancies, the role of non-coding RNA in the development of the disease has received extensive attention [
24]. In the early 1990s, the discovery of miRNAs enriched research in the field of non-coding RNAs, and miRNAs can be used as cancer markers and therapeutic targets [
25]. For miR-221-3p, a few studies have previously demonstrated the effect of it in different cancer types. In cancers such as liver cancer, colorectal cancer, and breast cancer, miR-221-3p is considered to have a cancer-promoting effect, while in gastrointestinal stromal tumours and cholangiocarcinoma, miR-221-3p is considered to have a cancer-suppressing effect. Our research revealed that pancreatic cancer expresses a high level of miR-221-3p, indicating a potential miR-221-3p role as a prognosis predictor in pancreatic cancer. Moreover, miR-221-3p promotes proliferation capacity, migration ability, invasion ability, and drug resistance but inhibits apoptosis in pancreatic cancer. In addition, we believe that miR-221-3p may work by acting on KIT, CDKN1B, RUNX2 and BCL2L11. Our experimental results are similar to those of Yang W, Sarkar S, Zhao S, and others. Yang W believes that miR-221-3p can promote the proliferation of pancreatic cancer cell Capan2 through PTEN-Akt [
20]. Sarkar S demonstrated that down-regulation of miR-221-3p inhibits the proliferation of pancreatic cancer cells by up-regulating the expression of PUMA and other proteins [
26]. Zhao L showed that miR-221-3p promotes cancer development by inhibiting RB1 to increase the resistance of PC cells to gemcitabine [
27]. However, we have adopted more and more comprehensive cell function tests and explored the relationship between the potencial targets KIT, CDKN1B, RUNX2, BCL2L11 and miR-221-3p. In the experimental design, we considered that one cell type was not sufficiently persuasive, and thus, we chose two cell lines. The reason for choosing these two cell lines is that they are in situ tumour cell lines. MIA PaCa-2 is derived from pancreatic tumours with aortic invasion, and BxPC-3 is derived from pancreatic tumours without metastasis. These cell lines have different invasion capabilities, and the experimental results are more representative. Meanwhile, little effort has been devoted to investigation of the association between the clinical characteristics of pancreatic cancer and miR-221-3p. To this end, biological informatics and cell experiments were employed to verify the influence of miR-221-3p on pancreatic cancer, with the aim of seeking out and confirming relative targets and providing a comprehensive study.
In sum, 13 microarrays selected from the GEO database conformed to the above standard for analysis, and miR-221-3p expression was found to be higher in pancreatic cancer tissues and blood than in para-carcinoma samples. Other data selected included TCGA, literature, and RT-qPCR. Based on an integrated meta-analysis of the employed datasets collected from different origins, it was verified that pancreatic cancer has higher miR-221-3p content, which is consistent with findings of earlier studies [
20]. Therefore, a conclusion could be made that pancreatic cancer has higher miR-221-3p content.
Moreover, through an analysis of TCGA expression information, miR-221-3p in pancreatic ductal adenocarcinoma and other types of pancreatic adenocarcinoma was found to be affected by age, tissue and prognosis results. Considering the RT-qPCR results, factors including age, diabetes and tumour location can affect miR-221-3p content. To be specific, patients with pancreatic cancer who had a lower age, diabetes or metastatic tumours presented with higher miR-221-3p content. In subgroup analyses, the miR-221-3p content in the pancreatic ductal adenocarcinoma group was shown to vary significantly after adjustment for age and the presence of diabetes; the other pancreatic adenocarcinoma types group indicated that miR-221-3p was dramatically affected by patient age. After analyses of TCGA and RT-qPCR data, we found that survival times differed between the high miR-221-3p subgroup and the low miR221-3p subgroup. In conclusion, miR-221-3p has the potential to become a biomarker for cancer monitoring.
To demonstrate the molecular influence of miR-221-3p, a series of cell experiments were performed. The results revealed that miR-221-3p could enhance the proliferation ability, migration ability, invasion ability, and drug resistance in pancreatic cancer while suppressing apoptosis. In other words, we also demonstrated a relevance between miR-221-3p and pancreatic cancer at the cellular level. Our research revealed that miR-221-3p may be related to pancreatic cancer tumourigenesis and progression, while increased miR-221-3p could become a predictor of pancreatic cancer acceleration.
To date, the molecular pathogenesis of pancreatic cancer has not been entirely elucidated. Accordingly, we employed biological informatics tools to explore the inner genetic interactions of miR-221-3p and regulation of pancreatic cancer progression and selected 30 underlying target genes of miR-221-3p according to data from miRWALK2.0, miRANDA and other programmes. For advanced functional exploration of these targets, GO annotation and KEGG pathway analyses were employed. The GO results showed that possible miR-221-3p targets may be crucially related to pancreatic cancer growth by exerting effects on diverse cellular biological processes, such as regulation of lipid kinase activity; for MF, the target genes were significantly enriched in the function of insulin receptor binding. Moreover, the roles of candidate miR-221-3p targets in pancreatic cancer were explained according to KEGG results, in which microRNAs in cancer and viral carcinogenesis pathways were ranked as the top two pathways. The above results demonstrate that underlying targets of miR-221-3p are likely to be concerned with these previously mentioned pathways, affecting pancreatic cancer origin and progression.
MicroRNAs in cancer pathways are recognized as a crucial pathway regulators in cancer, which was verified by Shi et al., who showed that microRNAs may regulate gastroesophageal cancer development, discrimination, and therapy [
28]. Furthermore, microRNAs might be connected with the low content of several essential proteins in oesophageal squamous cancer [
29], as well as decreased gene expression in colorectal cancer [
30]. In addition to gastrointestinal cancer, several other studies have explored microRNA functions in cancer. The microRNAs in cancer pathways are primarily associated with regulation of gene expression and initiation and progression in breast cancer [
31]. Additionally, bladder cancer is relevant to the microRNAs in cancer pathways [
32]. However, few studies have been conducted on the microRNAs in cancer pathways in association with pancreatic cancer. As a result, more studies are needed to further explore the latent pathological mechanism in pancreatic cancer.
To this end, the current study focused on validation of the increase in miR-221-3p expression and its effect in pancreatic cancer. The result of the PPI network projected four genes (KIT, CDKN1B, RUNX2, and BCL2L11) as hub genes in pancreatic cancer, which may be possible targets of miR-221-3p. Among the four hub genes, we found that KIT protein can phosphorylate a variety of intracellular proteins, which play a role in the proliferation, differentiation, migration and apoptosis of a variety of cells [
33]. CDKN1B can regulate the cell cycle [
34]. RUNX2 is a transcription factor in the RUNX family and plays an important role in bone growth and development [
35]. BCL2L11 is related to apoptosis [
36]. These hub genes are all related to cell proliferation, apoptosis and other biological behaviour. The inner connection between hub genes and miR-221-3p in other cancer types has been demonstrated elsewhere. Studies have also found that miR-221-3p might enhance the apoptosis rate through KIT/AKT signalling in gastroenteric carcinomas [
37]. Through regulation of CDKN1B, miR-221-3p suppressed proliferation, migration ability, and invasion ability in osteosarcoma cells [
38]. RUNX2 is closely related to the occurrence and development of various tumours, such as leukaemia [
39] and breast cancer [
40]. These hub genes may also play a role in the development of pancreatic cancer. Therefore, we further studied the connection between miR-221-3p and these targets in PC cells. Through our research, we found that miR-221-3p can bind to KIT, CDKN1B, RUNX2, and BCL2L11 in PC cells and that the expression of miR-221-3p can affect KIT, CDKN1B, RUNX2, and BCL2L11 expression. Thus, miR-221-3p may affect the incidence and development of pancreatic cancer by affecting KIT, CDKN1B, RUNX2, and BCL2L11 expression.
Our research mainly featured the employment of biological informatics and cell experiments to comprehensively verify the influence of miR-221-3p on pancreatic cancer and predict relative targets. The experimental highlights of this study include the use of multiple databases for a variety of studies, including both blood and tissue samples. In addition, this study also analysed pancreatic cancer subtypes and demonstrated the effect of miR-221-3p from different perspectives. Moreover, a variety of cellular and molecular experiments were selected to explore the clinical features of miR-221-3p upregulation at the cellular and molecular level, and finally, the pathways through which miR-221-3p acts were predicted using a variety of biological information function analyses. The PPI network was projected to find hub genes with the aim of investigating potential targets. In addition, we examined the hub genes KIT, CDKN1B, RUNX2, and BCL2L11, which may act in pancreatic cancer, and confirmed that the expression of KIT, CDKN1B, RUNX2, and BCL2L11 is indeed affected by miR-221-3p. In brief, the present study verified that miR-221-3p is highly expressed in pancreatic cancer, promotes proliferation ability, migration ability, invasion ability, and resistance to gemcitabine and inhibits apoptosis in pancreatic cells; predicted relative hub genes; and validated a target gene. Taken together, the results show that miR-221-3p may act as an underlying tumour marker for prognosis prediction in pancreatic cancer and may work by acting on KIT, CDKN1B, RUNX2, and BCL2L11 expression. The findings of the bioinformatics analyses might shed new light on the tumourigenesis of pancreatic cancer.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.