Cell culture

Human PDAC cell lines, Capan-2, Panc-1, Patu8988T, AsPC-1, SW1990, and BxPC 3, were collected from the Cell Bank of the Chinese Academy of Sciences CAS (Shanghai, China). All cell lines were kept in Dulbecco’s modified Eagle medium (DMEM; Gibco) that contained 10% fetal bovine serum (FBS; Gibco) as well as 1% penicillin-streptomycin. Human normal pancreatic epithelium cell line HPNE, supplied by American Type Culture Collection, was incubated in the DMEM/F12 medium (Gibco) that contained 0.5 µg/ml hydrocortisone, 5% horse serum, 20 ng/ml epidermal growth factor, 2 mM l-glutamine, 10 µg/ml insulin, and 0.1 µg/ml cholera enterotoxin. The cell lines were subjected to authentication through a short tandem repeat (STR) profiling analysis in the Cell Bank of Type Culture Collection of CAS (Shanghai, China). All cells were mycoplasma-free confirmed with the Universal Mycoplasma Detection Kit (ATCC, Manassas, VA, USA). To achieve the hypoxic condition, the cultured cells were flushed with 1% O2, 5% CO2, and 94% N2, or 5% O2, 5% CO2, and 90% N2 mixture gas in a hypoxia chamber.

lncRNA expression microarray analysis

Microarray analysis was conducted as a previous report [14]. In brief, total RNA from Panc-1 cells treated with 1% or 20% O2 was extracted using TRIzol reagent (Invitrogen, Thermo Fisher Scientific). Next, cDNA was synthesized using 0.5 µg total RNA via a GeneChip Transcription Express Kit (Thermo Fisher Scientific Inc., Waltham, MA, USA). The microarray was then scanned using an Agilent G2505B Microarray Scanner (Agilent Technologies).

Plasmids, short hairpin RNAs, and cell transfection

To construct the expression vector, cDNAs encoding full-length of GATA6-AS1, ETS1, FTO, and SNAI1 were subcloned into the pcDNA3.1 (+) vector (Invitrogen, Carlsbad, CA). Short hairpin RNA (shRNA) against GATA6-AS1, ETS1, FTO, METTL3, METTL14, WTAP, ALKBH5, YTHDF2, and SNAI1, as well as shRNA negative control (shNC), were designed and purchased from Genepharma (Shanghai, China). For RNA investigation, at least two independent shRNA sequences were evaluated for every gene. Sequences for shRNAs involved are outlined in Additional file 1: Table S1. All products were validated by DNA sequencing. When reaching 70–80% confluence, the cells were transiently transfected utilizing Lipofectamine 3000 reagent (Invitrogen, Carlsbad, California, USA) as per the manufacturer’s specifications.

Lentivirus synthesis and transduction

A lentiviral vector expressing either a full-length GATA6-AS1 or SNAI1 or shRNA targeting GATA6-AS1 was synthesized and subcloned into the lentiviral vector. Centrifugation was instrumental in concentrating recombinant lentiviruses, and we then dissolved them in DMEM and kept them at − 80 °C for subsequent use. These lentivirus vectors were transfected into PDAC cells (3 × 105/well) with polybrene (5 µg/ml, Sigma-Aldrich) according to the manual. The infected cells were thereafter chosen with 2 mg/l puromycin for 14 days.

Human PDAC tissue specimens

This study employed 2 distinct cohorts of human PDAC tissues. The formalin-fixed and paraffin-embedded (FFPE) PDAC tissue containing 116 pairs of PDAC tissues (T) and adjacent nontumor tissues (N) were utilized to create a tissue microarray (TMA) (cohort A) for in situ hybridization (ISH) and immunohistochemical (IHC) analysis. The fresh PDAC tissue samples containing 72 pairs of T and N tissues were used for the RT-qPCR assay (cohort B). Cohort A was gathered from Ruijin Hospital, Shanghai Jiaotong University School of Medicine (Shanghai, China) from January 2014 to May 2016 and cohort B from the First Zhengzhou University (Zhengzhou, China) from April 2016 to December 2019. Demographic characteristics and clinical information of individuals were collected from the medical records. Regarding these patients, none of them was subjected to chemotherapy or radiotherapy before surgery. The study protocol was performed according to the ethics committee of the chamber of physicians of Shanghai Jiaotong University School of Medicine, and Zhengzhou University, China. Informed consent was available from all patients.

Scoring

ISH assay of GATA6-AS1 was performed using a double digoxigenin (DIG)-tagged mercury locked nucleic acid (LNA) probe (miRCURY LNA™, Exiqon, Denmark). The GATA6-AS1 probe sequence used is shown below: 5′ Dig-ACTCACAGTTACGTGCAGAGGA-Dig 3′. For the IHC assay of E-cad, Vim, and SNAI1, the DAKO Envision system (DAKO, Carpinteria, California) was used as described earlier [15]. After quenching endogenous peroxidase action and preventing nonspecific attachment, the slides were treated overnight at 4 °C with the specified antibody or probes. The staining intensity was graded utilizing a scale of 0–3 (0, negative; 1, weak; 2, moderate; 3, strong) and the staining range was assessed according to the percentage of staining in five random fields (1, 0–10%; 2, 10–50%; 3, 50–75%; 4, > 75%). The eventual score was derived by multiplying both scores and was utilized to categorize the samples into 3 grades: weak staining (score 0–3); medium staining (score 4–6); and strong staining (score 7–12). When conducting survival analyses, we integrated weak and medium staining as a low expression (score 0–6), whereas strong staining was regarded as a high expression (score 7–12). Two pathologists who were blinded to the clinical data graded all sections independently.

RNA isolation and real-time quantitative PCR analysis

Total RNA was extracted from fresh pancreas tissues and cell lines utilizing TRIzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s specifications. We used the SYBR® Premix Ex Taq kit (Takara Bio Inc., Shiga, Japan) to determine the relative RNA level, which was derived utilizing the comparative Ct method. The specific primer sequences were shown Additional file 1: Table S2 and were synthesized and purified by the Shanghai GenePharma Co. (Shanghai, China). β-Actin was served for normalization. The BLAST algorithm (National Center for Biotechnology Information) was instrumental in the verification of the sequence specificity.

Methylation-specific polymerase chain reaction

The genomic DNA of PDAC cells was isolated using a commercial kit (TIANGEN, Beijing, China), and a DNA Methylation-Gold™ kit (D5005, Zymo Research, Irvine, CA) was exploited to assess the methylation level of FTO promoter. Methylated (M) and unmethylated (U) primers were displayed in Additional file 1: Tables S1, S2. The PCR products were then electrophoresed with 3% agarose gel, and the target bands were observed by gel imager. The experiment was repeated three times.

Cytoplasmic and nuclear fractionation

Nuclear and cytoplasmic fractionation from PDAC cells was executed using PARIS Kit (Life Technologies, Carlsbad, CA) and assessed by RT-qPCR detection. U6, as well as β-Actin, were taken as the nuclear and cytoplasmic controls, correspondingly.

RNA stability analysis

On a 12-wells plate, PDAC cells with transfection of the indicated vectors were seeded and kept in an incubator overnight at 37 °C. Stability of RNA was conducted with 5.0 µg/ml of Actinomycin D (Act-D, Sigma, U.S.A) at the stipulated times. Total RNA was obtained with the help of TRIZOL reagent (Invitrogen) and underwent RT-qPCR analysis. β-Actin served as the internal control. The half-life of mRNA was estimated based on previous study [16]. The rate of disappearance of mRNA concentration at a given time (dC/dt) is proportional to both the rate constant for decay (Kdecay) and the cytoplasmic concentration of the mRNA (C). This relation is described by the following equation: dC/dt = − KdecayC. The mRNA decay rate (Kdecay) was estimated as ln(C/C0) = − Kdecay t. Thus, the half-life (t1/2) was estimated by the following equation: t1/2 = ln2/Kdecay.

Protein extraction and Western blot assays

Total protein was prepared from cells and tissues utilizing detergent-containing lysis buffer, and the concentration of protein was ascertained utilizing a bicinchoninic acid assay kit (Bio-Rad Laboratories). Total protein (25 µg) received SDS-polyacrylamide gel separation and was transferred to PVDF membrane (Millipore). Primary antibodies against: ETS1 (ab238645, Abcam), hypoxia-inducible factor 1 alpha (HIF1A, ab51608, Abcam) and 2 alpha (HIF2A, ab243861, Abcam), E-cad (ab40772, Abcam), Vim (ab8069, Abcam), ZEB1 (ab245283, Abcam), TWIST1 (CST, 69366S, Abcam), FTO (ab126605, Abcam), SNAI1 (sc-271977, Santa Cruz), DNMT1 (ab19905, Abcam), DNMT3A (ab188470, Abcam), DNMT3B (ab2851, Abcam), and METTL3 (ab195352, Abcam) were used. The protein loading control was β-actin. Quantifications of Western blots were analyzed using Image J V1.53c (National Institutes of Health).

In vitro proliferation, invasion, and migration assays

Onto a 96-well plate cells were seeded (2000 cells per well). Afterward, upon attaining a predetermined time of culture, cell viability was ascertained utilizing MTT assays. The optical density (OD) was taken at 450 nm in every well using a microplate reader (BioRad, Hercules, CA, USA). For the EdU assay, the Yefluor 594 Edu Imaging Kits (Yeason, China) were adopted following the manufacturer’s protocols. Transfected cells were subjected to culturing with Edu for 2 h. Then, they were fixated with 4% paraformaldehyde, stained with Yefluor 594 Azide Solution, and ultimately mounted with DAPI (Sigma-Aldrich, St. Louis, Missouri, USA). EdU positive cells were reported as the quantity of EdU (red) positively stained cells/the quantity of DAPI (blue) positively stained cells in three randomly picked fields. On the other hand, a pipette tip was utilized in the wound-healing assay to create horizontal streaks in the cells that were grown in the 12 well plates until they attained confluence. After 24 h, photos were retaken, and the distance of migration was measured at 0 and 24 h. About invasion assays, the 8-µm pore inserts were coated with 30 µg of Matrigel (BD Biosciences). PDAC cells (2 × 105) were placed in the top chambers in a medium that was free of serum. As a chemoattractant, DMEM medium having 20% FBS was added to the bottom chambers. Upon incubation for 24 h at a temperature of 37 °C, the invaded cells were fixed, stained utilizing crystal violet, and observed using a microscope at 200× magnification for enumeration in three randomly picked fields. The migration assay was performed in a similar method devoid of coating the filters utilizing Matrigel.

Tumor xenografts

Female athymic BALB/c mice, aged 4–5 weeks, were procured from Slack, Shanghai, China, and were fed under the standard pathogen-free settings. Logarithmic phase PDAC cells (4 × 106/100 µl) were infected using lentiviruses possessing constructs. Then, the nude mice (n = 5 per group) were inoculated into the dorsal flank subcutaneously. The animals under investigation were thoroughly monitored, and the tumor size was measured at an interval of 5 days. The volume of the tumor was ascertained using the equation below: volume (mm3) = length × width2 × 0.5. Mice were killed utilizing CO2 as per the animal welfare specifications. All animal studies were approved by the Institutional Animal Care and Use Committee of Zhengzhou University (Henan, China). For the metastatic lung model, a total of 1 × 106 cells of SW1990 cells stably-transfected with sh-GATA6-AS1#1, #2 or control vector were injected into tail veins of BALB/c nude mice. After all these mice were sacrificed, the tumors and lung were excised and processed for analysis.

LncRNA fluorescence (FISH) and immunofluorescence staining

RNA FISH was conducted on PDAC cells and tissues utilizing RNA Fluorescence In Situ Hybridization Kit (Exonbio Lab, Guangzhou, China). RNA FISH, as well as immunofluorescence staining, were used to ascertain the colocalization of GATA6-AS1 in cells. Fluorescein-labeled RNA probes for GATA6-AS1 were utilized to hybridize the PDAC cells overnight at 37 °C. Slides were washed and thereafter counterstained with 4′-6′diamidino-2-phenylindole (DAPI, Beyotime). For immunofluorescence (IF) staining, the tissue slices were fixed in 4% paraformaldehyde for 30 min, permeabilized, and blocked utilizing 5% bovine serum albumin (BSA) (Sigma, St Louis, MO) for 1 h. After the tissue sections underwent incubation with primary antibody against E-cad and Vim, and secondary fluorescent antibody (Invitrogen, 594 nm), respectively, the tissue sections were subjected to DAPI for 10 min to stain them. Finally, fluorescence images were obtained by confocal microscope (Olympus FV1000, Tokyo, Japan).

Chromatin immunoprecipitation

As described previously [17], chromatin immunoprecipitation (ChIP) assays were carried out utilizing the EZ ChIP Kit (Millipore, MA, USA). The proteins were cross-linked to the DNA by adding 1% formaldehyde at RT for 15 min, and the DNA was split to a mean fragment size of 200–800 bp by sonication. After DNA-protein samples were precleared with protein A/G beads, they were exposed to immunoprecipitation overnight with an anti-ETS1 antibody or IgG. Next, the related genomic DNA was heated to reverse histone-DNA crosslinks and was assessed by qPCR with GATA6-AS1 promoter-specific primers.

RNA immunoprecipitation

RNA immunoprecipitation (RIP) assay was conducted with the utilization of the Magna RIP Immunoprecipitation kit (Millipore, MA) based on the manufacturer’s specifications. After washing with PBS twice, 1 × 107 PDAC cell lines were lysed in a complete RIP lysis buffer. After centrifuging and removing the cell debris, the supernatant was retained as input, and the whole cell extract was co-immunoprecipitated with an antibody against FTO, METTL3, METTL14, WTAP, and ALKBH5 or control IgG at 4 °C overnight. The enrichment of RNA was determined by RT-qPCR.

Luciferase reporter gene assays

The GATA6-AS1 promoter regions having two ETS1 putative binding areas (wild type, WT) or two mutant areas (mutant type: MUT1 or MUT2) were respectively introduced into the pGL3 vector (Promega, Madison, WI), and then was transfected into 293 T cells. To determine the effects of FTO on SNAI1 mRNA containing potential two m6A sites (supported by SRAMP prediction) or mutant type (adenosines in m6A positions were replaced by cytosines) were inserted into luciferase reporter vectors. Relative luciferase activity was calculated by “F-luc/R-luc” utilizing the Dual-Luciferase Reporter Assay System (Promega).

RNA pull-down assay

Biotin-labeled RNAs, including FL (full-length), Antisense and serial truncations of GATA6-AS1, were transcribed in vitro with the Biotin RNA Labeling Mix and T7 RNA polymerase (Roche, Basel, Switzerland). Purified biotin-labeled RNA was heated and annealed to form a secondary structure, mixed with streptavidin agarose beads (Life Technologies, Gaithersburg, MD) at 4 °C for 1 h. Total cell lysates were freshly prepared and added to each binding reaction with Protease/Phosphatase Inhibitor Cocktail and RNase inhibitor. Finally, the RNA-binding proteins were analyzed by Western blot.

Methylated RNA immunoprecipitation (MeRIP) assay

The Magna MeRIP m6A Kit (Millipore, USA) was employed in determining the enrichment of m6A at particular sites on the SNAI1 transcript. After total RNA was extracted from PDAC cells and tissues, total RNA (1 µg/µl) was sheared to fragmentation about 100 nt in length. Thereafter, the fragmented RNA underwent incubation with an anti-m6A antibody (1 µg) or control IgG-conjugated Dynabeads in an IP buffer for 4 h with rotation. Ultimately, the co-precipitated RNA specimens having m6A modification sites were utilized for qPCR determination. Certain primers were designed based on the m6A modification sites on the SNAI1 transcript with greater confidence (anticipated by SRAMP) [18]. Additional file 1: Table S1 lists the primer sequences.

Bioinformatics analysis

RNAseq data were retrieved from two independent PDAC cohorts, GSE15471 and The Cancer Genome Atlas (TCGA)-Pancreatic adenocarcinoma (TCGA-PAAD) databases, correspondingly. GSE15471 has 39 pairs of PDAC malignant tissues as well as normal tissues. TCGA-PAAD contains 178 cancerous samples and 4 normal samples. The potential RNA–protein interactions were anticipated by the machine learning classifier RPISeq [19] utilizing Random Forest (RF) or Support Vector Machine (SVM) classifiers. The prediction of m6A modification sites was performed using the online tools SRAMP [18], depending on epitranscriptome sequencing information as well as machine learning pattern.

Statistical analysis

Data were presented as means ± SD for at least n = 3 independent experiments except otherwise explanation. Regarding comparisons, the student’s t-test, Pearson chi-square test, and the Wilcoxon signed-rank test were conducted when necessary. Spearman correlation analysis analyzed correlations. Kaplan–Meier plots as well as log-rank tests ascertained overall survival analysis. Univariable and multivariable Cox proportional hazards regression models were instrumental in analyzing independent prognostic factors. P < 0.05 was statistically significant. Statistical analyses were carried out by SPSS19.0 (SPSS, Chicago, IL) and GraphPad Prism version 8.0 (GraphPad Inc., La Jolla, CA, USA).