LncRNA BCAN-AS1 stabilizes c-Myc via N⁶-methyladenosine-mediated binding with SNIP1 to promote pancreatic cancer

Abstract

C-Myc overexpression contributes to multiple hallmarks of human cancer but directly targeting c-Myc is challenging. Identification of key factors involved in c-Myc dysregulation is of great significance to develop potential indirect targets for c-Myc. Herein, a collection of long non-coding RNAs (lncRNAs) interacted with c-Myc is detected in pancreatic ductal adenocarcinoma (PDAC) cells. Among them, lncRNA BCAN-AS1 is identified as the one with highest c-Myc binding enrichment. BCAN-AS1 was abnormally elevated in PDAC tumors and high BCAN-AS1 level was significantly associated with poor prognosis. Mechanistically, Smad nuclear-interacting protein 1 (SNIP1) was characterized as a new N⁶-methyladenosine (m⁶A) mediator binding to BCAN-AS1 via recognizing its m⁶A modification. m⁶A-modified BCAN-AS1 acts as a scaffold to facilitate the formation of a ternary complex together with c-Myc and SNIP1, thereby blocking S phase kinase-associated protein 2 (SKP2)-mediated c-Myc ubiquitination and degradation. Biologically, BCAN-AS1 promotes malignant phenotypes of PDAC in vitro and in vivo. Treatment of metastasis xenograft and patient-derived xenograft mouse models with in vivo-optimized antisense oligonucleotide of BCAN-AS1 effectively represses tumor growth and metastasis. These findings shed light on the pro-tumorigenic role of BCAN-AS1 and provide an innovant insight into c-Myc-interacted lncRNA in PDAC.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most malignant cancers with a 5-year survival rate of less than 10% [1, 2]. Many PDAC patients are initially diagnosed at advanced stages and eventually die of tumor progression for lack of effective treatments. Therefore, it is imperative to elucidate the biological characteristics and molecular mechanism underlying PDAC development and progression in order to develop more effective diagnostic and therapeutic interventions for PDAC.

Oncoprotein c-Myc, a master transcription factor participating in vital cellular processes such as proliferation, differentiation, metabolism, apoptosis, self-renewal, protein and ribosomal biogenesis, has been reported to play a prominent role in various human malignancies [3–5]. Genomic amplification and overexpression of c-Myc increase the expression of immune checkpoint proteins on the tumor cell surface [6, 7]. Moreover, c-Myc can also instruct remodeling of tumor stroma and angiogenesis [8, 9], and extensively program an immune suppressive stroma that is obligatory for tumor progression [10]. Aberrant expression of c-Myc is an important molecular hallmark of human cancers [11], including PDAC [12, 13]. Acute activation of c-Myc in indolent pancreatic intraepithelial neoplasm (PanIN) epithelial cells in vivo is sufficient to lead to changes in multiple stromal and immune-cell types and drive transition to PDAC [14]. C-Myc amplification is reported to be a driver of metastatic spread and is enriched in metastatic lesion of some cancers [15–17]. Besides mutations in the gene itself, it is more common that the c-Myc expression is induced via upstream oncogenic pathways or controlled by the stability of the c-Myc protein [18]. However, until now, the underlying mechanism of c-Myc upregulation in PDAC has not been fully understood. Notably, accumulating evidence has suggested that decreasing c-Myc expression elicits tumor regression in multiple tumor models [19, 20], and targeting c-Myc is considered to be an attractive therapeutic option for many human cancers [12, 13, 21]. Therapies targeting the c-Myc pathway will be key to reversing cancerous growth and restoring anti-tumor immune responses [22]. However, direct targeting of c-Myc protein is challenging and technically impractical due to the potentially toxic to normal tissues and the lack of druggable binding pockets in c-Myc itself [21, 23]. Thus, great efforts are undertaken to identify key targets involved in c-Myc deregulation in order to explore potential indirect targets for c-Myc.

Long noncoding RNAs (lncRNAs) are an important class of transcripts which can function as master gene regulators through various mechanisms, such as signals, decoys, guides or scaffolds [24–26]. Epigenetic regulatory mechanism, such as m⁶A modification, is found to exist on most lncRNAs which can affect RNA-protein interaction [27]. Increasing evidences have shown that dysregulation of lncRNA-protein interaction contributes to various cancer hallmarks [28, 29], such as aberrant c-Myc expression [30, 31]. However, the exact mechanisms remain scarce. Recent studies have shown that some lncRNAs can interact with c-Myc directly or indirectly. LncRNA PVT1 physically interacts directly or indirectly with and stabilizes c-Myc by preventing its phosphorylation and degradation [32]. Besides, binding to c-Myc contributes to the function of lncRNA EPIC1 in regulating the transcriptional activity of c-Myc protein [33]. Other lncRNAs such as PCGEM1, LINC01638, MALAT1 and PD-L1-lnc can also directly interact with c-Myc protein and play a considerable role in cancers [34–37]. Additionally, some circRNAs are also reported to have the ability to interact with c-Myc [38, 39]. All these studies indicate that c-Myc-interacted RNAs may play a vital role in regulating c-Myc function. Given that lncRNAs have drew widespread attention due to their potential role in cancer therapy [40], targeting c-Myc-interacted lncRNAs might be an ideal strategy to indirectly target c-Myc for cancer treatment. Thus, better understandings of the precise mechanism of c-Myc/lncRNAs interaction in cancer are urged to explore potential therapeutic strategies in cancer treatment.

In the present study, we have characterized a c-Myc-interacted lncRNA named BCAN-AS1, as an oncogenic molecule in PDAC. We have found that Smad nuclear-interacting protein 1 (SNIP1) can recognize and bind to m⁶A-modified BCAN-AS1. BCAN-AS1 serves as a scaffold for c-Myc and SNIP1 and strengthens the c-Myc/SNIP1 interaction, which in turn blocks S phase kinase-associated protein 2 (SKP2)-mediated c-Myc ubiquitination. Treating mouse models with BCAN-AS1 inhibitors results in decreased tumor burdens in mice. Our findings highlight BCAN-AS1 as an important mediator in the regulation of c-Myc in PDAC and imply the potential of targeting BCAN-AS1 as an adjuvant therapeutic method in PDAC.

Materials and methods

Tissue specimens

Surgically removed PDAC samples and their matched adjacent normal tissues used in this study were obtained from Sun Yat-sen University Sun Yat-sen Memorial Hospital (Guangzhou, China, n = 158) and Chinese Academy of Medical Sciences Cancer Hospital (Beijing, China, n = 73) between 2010 and 2016 (Supplementary Table 1). PDAC diagnosis was histopathologically confirmed by three independent pathologists. Tumor stages were defined according to the 7th edition of the AJCC Cancer Staging System [41]. The tumor and stromal contents were evaluated from the continuous tissue section slides stained with H&E by three board-certified pathologists who were blinded to the patients’ clinicopathological status and only the samples contain ≥60% tumor cells were used. The characteristics and clinical data of individuals were obtained from medical records. All individuals received no preoperative radiotherapy or chemotherapy. Patient’s overall survival time was defined as the interval between the date of cancer diagnosis and the date of last follow-up or death. Follow-up information was obtained from inpatient and outpatient records, the patient’s family or follow-up telephone calls. Informed consent was obtained from each patient, and this study was approved by the Institutional Review Board of the Sun Yat-sen Memorial Hospital and Chinese Academy of Medical Sciences Cancer Hospital.

Cell lines and cell culture

Human PDAC cell lines (PANC-1 and SW1990) and human embryonic kidney cell line 293T (HEK293T) were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences Shanghai Institute of Biochemistry and Cell Biology. Human PDAC cell lines (Panc 08.13 and Panc 10.05) were obtained from ATCC. Cells were cultured in DMEM (PANC-1 and HEK293T) or RPMI 1640 (SW1990) medium supplemented with 10% fetal bovine serum (FBS, Invitrogen) or RPMI 1640 (Panc 08.13 and Panc 10.05) medium supplemented with 15% FBS and 10 µg/mL insulin (GIBCO) at 37 °C in an atmosphere containing 99% relative humidity and 5% CO₂. All cell lines were authenticated by DNA finger printing analysis and detected without mycoplasma infection.

Northern blot assays

Total RNA extracted from PDAC cells was applied to formaldehyde gel electrophoresis and transferred to Biodyne Nylon Membrane (Pall). The digoxigenin-labeled BCAN-AS1 probes were synthesized by Bersinbio (Supplementary Table 2). After 30 min of pre-hybridization in DIG Easy Hybrid buffer (Roche), the membrane was hybridized in buffer containing the denatured probes for 12 h at 68 °C. Then, the membrane was washed and incubated with anti-digoxigenin-AP. Signal on the membrane was detected utilizing an Odyssey infrared scanner (Li-Cor, Lincoln).

RNA extraction and quantitative real-time PCR analysis (qRT–PCR)

Total RNA was isolated from cell lines or pancreatic tissue specimens using TRIzol reagent (Invitrogen) and was treated with DNase I (Thermo Fisher). The cDNAs were synthesized using the RevertAid First-Strand cDNA Synthesis Kit (Thermo Fisher) and were used for qRT-PCR analysis on a Roche Light Cycler 480 II using the SYBR-green method. β-ACTIN was used as an internal control for normalization of the RNA levels of indicated genes. The relative levels of RNAs were calculated using the comparative Ct method. Three replicates were performed in each experiment. The gene specific primer sequences are shown in Supplementary Table 3.

Quantification western blot of c-Myc Protein

For the quantification of c-Myc molecules in PANC-1 and SW1990 cells, quantitative western blot was performed. Lysates with a certain number of PANC-1 and SW1990 cells were blotted next to a 1:2 dilution series of recombinant human c-Myc protein (ab169901) ranging from 250 pg to 31.3 pg and from 500 pg to 62.5 pg respectively. Afterward, bands were quantified by Image J software and linear regression was performed respectively, showing satisfactory linearity. The c-Myc protein masses of lysates with a certain number of PDAC cells were predicted using the linear standard curves. The number of c-Myc molecules per cell was determined based on the predicted c-Myc protein masses of lysates.

Protein immunoprecipitation

Protein immunoprecipitation was performed with the Pierce™ Crosslink Magnetic IP/Co-IP Kit (Thermo Fisher). Cells were lysed with immunoprecipitation lysis buffer supplemented with the Protease/Phosphatase Inhibitor Cocktail (Pierce). After centrifugation, supernatants of the lysates were incubated with the indicated antibodies crosslinked to Protein A/G magnetic beads for 1 h at room temperature. The bead-bound complexes were subsequently eluted after washing with immunoprecipitation lysis buffer.

Western blot assays

Protein extracts from cells, tissue samples, RNA pull-down or immunoprecipitation samples were prepared using 1× RIPA buffer supplemented with a Protease/Phosphatase Inhibitor Cocktail (Pierce). Total protein (20 µg) was subjected to SDS-PAGE and transferred to PVDF membranes (Millipore). Membranes were incubated with specific primary antibody overnight at 4 °C. The immunoreactive bands were detected using a Phototope Horseradish Peroxidase Western Blot Detection kit (Thermo Fisher). Antibody against METTL3 (ab195352), MYL6B (ab243737), p-S62 c-Myc (ab185656) and p-T58 c-Myc (ab185655) were from Abcam. Antibodies against CDC45 (11881), CDC6 (3387), Cyclin A2 (4656) or Ubiquitin (3933) were from Cell Signaling Technology. Antibodies against c-Myc (10828-1-AP), SNIP1 (14950-1-AP), SKP2 (15010-1-AP), PAICS (12967-1-AP), MCM7 (11225-1-AP), HADHB (29091-1-AP), MYO18A (14611-1-AP), KIF5B (21632-1-AP) or β-ACTIN (66009-1-Ig) were from Proteintech.

RNA sequencing

Total RNA extracted from PANC-1 cells with or without BCAN-AS1 silenced and with or without BCAN-AS1 overexpression was subjected to RNA sequencing (Novogene). Sequencing libraries were generated using NEBNext® UltraTM RNA Library Prep Kit for Illumina® (NEB, USA). The clustering of the index-coded samples was conducted on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina). Then the library preparations were sequenced on an Illumina Hiseq platform.

Analysis of RNA sequencing data and Gene Set Enrichment analysis (GSEA)

For RNA sequencing, reads were examined by FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Adaptor and low-quality bases were trimmed using fastp [42]. Qualified sequencing reads were aligned to hg38 human genome using STAR [43]. Count and FPKM of each gene were calculated by RSEM [44]. Differential gene expression was analyzed using DESeq2 [45]. Pathway analysis was performed by GSEA [46]. For GSEA analysis of RNA sequencing data of PDAC tumors from TCGA, the median of BCAN-AS1 levels was used as cutoff value to define ‘high BCAN-AS1 expression’ group and ‘low BCAN-AS1 expression’ group.

RNA immunoprecipitation (RIP) assay

RIP assays were performed using the Magna RIP RNA-Binding Protein Immunoprecipitation kit (Millipore). For c-Myc RIP-seq assays, PANC-1 cells were lysed on ice and centrifuged. One tenth of the supernatants were used as input controls and the others were incubated with pre-conjugated Dynabeads protein A/G and anti-c-Myc antibody. Then the RNAs from input controls and c-Myc IP samples were subsequently treated with proteinase K and purified by acidic phenol/chloroform extraction and ethanol precipitation. The purified RNA was converted to cDNA for sequencing library construction and sequenced on Illumina NovaSeq 6000 (Genedenovo Biotechnology Co., Ltd). For RIP-qPCR assays, the co-precipitated RNAs were detected by qRT-PCR assays. To determine the m⁶A levels of BCAN-AS1, total RNA was extracted and fragmented using the RNA fragmentation reagents (Ambion), followed by incubated with the pre-conjugated anti-m⁶A antibody (Synaptic Systems, 202003) and magnetic Dynabeads mixture. After proteinase K (10 mg/mL) treatment, the bound RNAs were extracted with phenol/chloroform/isoamyl alcohol and subjected to qRT-PCR using the specific primers for BCAN-AS1 around its m⁶A site (Supplementary Table 3), which was normalized to input.

RIP sequencing analysis

Preprocessing steps were the same as described above for RNA sequencing data. Qualified sequencing reads were aligned to hg38 human genome using STAR [43]. RSEM [44] was used to calculate FPKM of each gene. Genes with input FPKM value >0.1 were further analyzed. The enrichment of target genes was calculated as IP fold-enrichment (IP/Input). The c-Myc-interacted lncRNAs were shown in heatmap ranked according to their fold-enrichment.

Measurement of absolute BCAN-AS1 copy number per cell

The accurate copy number of BCAN-AS1 per cell was evaluated using an absolute qRT-PCR method. Briefly, in vitro transcribed full-length BCAN-AS1 at different dilutions were synthesized cDNAs using the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher) and then performed qPCR. The CT values and the corresponding copy number of the BCAN-AS1 template were used to make the standard curve. For sample test, total RNAs from a certain number of PDAC cells were extracted using TRIzol reagent (Invitrogen) and treated with DNase I (Thermo Fisher). First-strand cDNA was synthesized and then perform qPCR assay. Then the CT values were used to calculate RNA concentration of each sample by the standard curve. Basing on molecular weight and cell count, we calculated the exact BCAN-AS1 copy number per cell.

Cytoplasmic and nuclear fractionation

The NE-PER Nuclear and Cytoplasmic Extraction Reagent (Thermo Fisher) was applied to extract cytoplasmic and nuclear fractions from PANC-1 and SW1990 cells. The levels of BCAN-AS1 were detected in each fraction using qRT-PCR assays. RNA samples used were treated with DNase I (Thermo Fisher). GAPDH and U6 were respectively used as cytoplasmic and nuclear control.

Single-molecule fluorescence in situ hybridization (smFISH) and immunofluorescence

SmFISH was conducted using a set of 48 custom Stellaris FISH probes targeted BCAN-AS1 RNA (Biosearch). SmFISH and immunofluorescence staining of c-Myc and SNIP1 were performed to analyze the co-localization of BCAN-AS1, c-Myc and SNIP1. Briefly, PDAC cells were fixed with 3.7% formamide at room temperature for 10 min and permeabilized with 0.1% Triton X-100. Then the cells were incubated with primary antibodies against c-Myc (Proteintech, 10828-1-AP) and SNIP1 (Abcam, ab169577), followed by secondary antibodies staining. After fixation again, cells were washed with smFISH wash buffer and hybridized with smFISH probes overnight in a humidified chamber at 37 °C in the dark. Cell nuclei were counterstained with DAPI. Images were captured with Olympus FV1000 confocal microscope. SmFISH probes are listed in Supplementary Table 2.

Ribosome sedimentation analysis

Cells were treated with cycloheximide (100 μg/mL; Cell Signaling Technology) and lysed on ice in lysis buffer. After centrifugation, the supernatant was collected and loaded onto the top of a 5–50% sucrose gradient. The gradients were centrifuged (Beckman), pumped out and collected by measuring RNA absorbance at 254 nm using an ISCO fractionator (Brandel). Then, RNA in each fraction was purified and subjected to qRT-PCR assays. RNA distributions across the polysome profile are displayed as percentages.

Cell proliferation and colony formation assays

For cell proliferation assays, PANC-1, SW1990, Panc 08.13 and Panc 10.05 cells were seeded in 96-well plates (2,000 cells per well). Cell viability was measured at indicated time points via Cell Counting Kit-8 (CCK-8, Dojindo Laboratories). Each experiment was performed three times, with six replicates each time. For colony formation assays, PANC-1 and SW1990 cells were seeded in 6-well plates (1,000 cells per well). After incubation for 12 days, cell-colonies were fixed with methanol, stained with crystal violet and counted.

In vitro migration and invasion assays

For cell migration assays, cells (1 × 10⁵) in serum-free medium were seeded in the upper chamber of Millicell chamber with 8-μm pores. Medium with 15% FBS was added to the lower chamber. After incubation for 16 h at 37 °C and 5% CO₂, cells migrated through the filters were fixed with methanol, stained with crystal violet and photographed. Cells in three random fields were counted. Cell invasion assays were performed using a similar method but the upper chamber pre-coated with 30 μg of matrigel (BD Biosciences). Each experiment was repeated three times.

Small interfering RNAs and transfection

Small interfering RNA (siRNA) targeting the METTL3, MYC or SKP2 gene and non-targeting siRNA control (Supplementary Table 4) were purchased from GenePharma. Transfections were performed with Lipofectamine 3000 (Life Technologies).

Construction of vectors and transient transfection

To construct expression vector for Flag-tagged c-Myc, SNIP1, BCAN-AS1 and WSPAR, synthesized cDNA encoding full-length of these genes or truncated versions of c-Myc and SNIP1 were subcloned into pcDNA3.1-3× Flag vector (Umine Biotechnology Co.,LTD). Transient transfections of plasmids were performed with Lipofectamine 3000 (Life Technologies).

Plasmids, lentivirus production and transduction

To construct a lentiviral vector expressing wild-type or m⁶A site mutant human BCAN-AS1, synthesized full-length of wild-type BCAN-AS1 sequence or BCAN-AS1 sequence with adenosine (A) to thymine (T) mutant at m⁶A site was inserted into pHS-BVC-LW308-Puromycin lentiviral expression vector (Umine Biotechnology). To achieve depletion of BCAN-AS1 in cells, short hairpin RNA (shRNA) specifically targeting BCAN-AS1 (Supplementary Table 4) was respectively synthesized and inserted into pLKD-U6-MCS-CMV-Puro lentiviral shRNA vector (Umine Biotechnology). This approach was also used for the construction of SNIP1 shRNA. A scramble shRNA served as control.

To achieve depletion of METTL3 in cells, CRISPR-Cas9 system with gRNA (Supplementary Table 4) was used as previously described [47]. Overexpression vectors of wild-type and catalytic mutant (aa395–398, DPPW → APPA) METTL3 were synthesized by Obio Technology.

PspCas13b-ALKBH5 plasmid, guide RNAs (gRNAs) targeting sequences adjacent to the BCAN-AS1 m⁶A site and non-targeting gRNA control (NT-gRNA) were designed and synthesized by Umine Biotechnology. The sequence of gRNAs and NT-gRNA were provided in Supplementary Table 4.

To achieve endogenous shRNA or ASO-BCAN-AS1 binding site mutation of BCAN-AS1 in cells, CRISPR system with gRNA (Supplementary Table 4) were designed and constructed by SYNBIO Technologies. Homology-directed recombination donor template carrying each mutation was co-transfected with the CRISPR system and gRNA.

HEK293T cells were transfected with the vectors described above and the lentiviral vector packaging system (Obio Technology) to produce lentiviruses. PDAC cells were infected with the resultant lentiviruses in the presence of polybrene (Sigma-Aldrich) and selected by corresponding antibiotics for two weeks. Transduction efficiency was confirmed by qRT-PCR or western blot assays. To induce expression of dCas13b-ALKBH5, cells were cultured in medium supplemented with doxycycline (2 µg/mL) and mice were fed with a doxycycline (2 µg/mL) diet.

Ubiquitination assays for c-Myc

Cells were treated with 5 μM proteasome inhibitor MG132 (M1902; Abmole) for 24 h and lysed. The lysates were incubated with c-Myc antibody pre-crosslinked to Protein A/G magnetic beads and the co-precipitated proteins were eluted. Then the ubiquitination levels of c-Myc were determined by western blot assays with antibody against ubiquitin (Cell Signaling Technology; 3933).

RNA pull-down and mass spectrometry analysis

RNA pull-down assays were performed as described. Briefly, the biotin-labeled 50-bp RNA probes containing BCAN-AS1 sequences with or without m⁶A modification (Supplementary Table 2) were synthesized and incubated with protein extracts from PANC-1 cells. Streptavidin beads were then added and the co-precipitated proteins were extracted and subjected to mass spectrometry analysis or western blot assays. Full-length or truncated BCAN-AS1 RNA or its antisense was transcribed in vitro using the MEGAscript® Kit (AM1333, Life Technologies) and biotin-labeled using the PierceTM RNA 3’ End Desthiobiotinylation Kit (20163, Thermo Fisher). Pull-down assays with these RNA probes were similar to the approach mentioned above. In vitro transcription of BCAN-AS1 and its truncated fragments were generated with primers containing the T7 promoter sequence (Supplementary Table 3).

RNA electrophoretic mobility shift assays (REMSA)

REMSA assays were performed using the LightShift Chemiluminescent RNA EMSA Kit (Life Technologies). Biotin-labeled RNA probes with or without m⁶A modification (Supplementary Table 2) were incubated in the binding buffer containing various concentrations of recombinant SNIP1 proteins at room temperature for 30 min. The RNA-protein mixture was separated into the pre-run native polyacrylamide gels at 4 °C for 1 h. Complexes were then transferred to a nylon membrane. The membrane was crosslinked using the UVP cross-linker. Signals were detected by chemiluminescence.

Chromatin isolation by RNA purification (ChIRP) assays

ChIRP assays were performed with the EZ-Magna ChIRP RNA Interactome Kit (Millipore). Antisense biotin-labeled DNA probes targeting the BCAN-AS1 sequence were designed online (https://www.biosearchtech.com) and synthesized (Supplementary Table 2). The 24 probes were used and divided into even and odd sets. Probe sets targeting LacZ RNA were employed as a negative control. Cells were cross-linked, quenched, lysed and sonicated. Cell lysate was hybridized with odd or even sets of BCAN-AS1 probes or LacZ probes and the hybridized mixtures were incubated with magnetic streptavidin beads. 20% of the bead-probe-RNA complex was used for RNA extraction while 80% for protein extraction. BCAN-AS1 was detected by qRT-PCR while the interested proteins were detected by western blot assays.

Establishment of mouse xenograft models

5-week-old female BALB/c nude mice were obtained from Beijing Vital River Laboratory Animal Technology. For subcutaneous xenograft model, mice (n = 5 per group) were subcutaneously injected with 0.1 mL of cell suspension containing 2 × 10⁶ PDAC cells in the back flank. When palpable, the tumor was measured every week and tumor volume was calculated by length × width² × 0.5. For pancreatic xenograft model, 2 × 10⁶ luciferase-labeled PDAC cells were injected into the mouse pancreas. Tumor volume was monitored by bioluminescence imaging in a Living Image® system (Perkin Elmer). The mouse survival time was recorded from the day of tumor implantation to the date of death. For lung metastatic models, 1 × 10⁶ luciferase-labeled PDAC cells were injected into the mouse tail vein. The pulmonary metastasis of athymic nude mice was detected by bioluminescence imaging. For patient-derived xenograft (PDX) models, surgically removed fresh PDAC samples from three patients without any chemotherapy or radiation therapy were propagated as subcutaneous tumors in 4-week-old NSG mice (F1, Beijing Vital River Laboratory Animal Technology). Tumors from F1 mice were cut into small pieces and then implanted into other mice (F2). When tumors attained a size of 1500 mm³, they were excised, cut again into small pieces and transplanted to other mice (F3).

Treatment of xenografts in mice

Experimental treatment in mouse lung metastatic, pancreatic xenograft or PDX models was performed. For ASO-BCAN-AS1 treatment, mice with lung metastases were randomly divided into two groups (n = 5 per group) after PDAC cell inoculation and treated with antisense oligonucleotide (ASO)-control (Ctrl) or ASO-BCAN-AS1 (i.v. injection; 75 mg/kg; RiboBio; Supplementary Table 4) every three days. Lung metastases were monitored every three days by bioluminescence imaging. For PDX treatment, when PDX reached about 200 mm³, mice were randomly divided into two groups (n = 5 per group) and received the same treatment as the mice with lung metastases. The PDX volume (length × width² × 0.5) was monitored every three days. For cholesterol-conjugated SNIP1 siRNA treatment, luciferase-labeled BCAN-AS1 overexpressing PDAC cells and the corresponding cells were injected into the mouse pancreas or the mouse tail vein. Mice inoculated with BCAN-AS1 overexpressing PDAC cells were randomly divided into three groups (n = 5 per group). Two BCAN-AS1 overexpressing groups received either siControl treatment (i.v. injection; 10 nmol in 0.1 mL saline buffer; RiboBio) or SNIP1 siRNAs treatment (i.v. injection; 10 nmol in 0.1 mL saline buffer; RiboBio; Supplementary Table 4) every three days. The control group and another BCAN-AS1 overexpressing group received a comparable volume of saline buffer. Lung metastases and tumor volumes of pancreatic xenograft were monitored every three days by bioluminescence imaging. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Sun Yat-sen University Cancer Center and the animals were handled in accordance with institutional guidelines.

miCLIP-seq

miCLIP-sequencing were performed as the previous study described [48] with some modifications. Briefly, total RNA from PANC-1 cells was extracted and fragmented. The fragmented RNA was then incubated with 15 µg of anti-m⁶A antibody (Synaptic Systems, 202003) in RIP buffer at 4 °C for 60 min with rotating. After cross-linking twice with UV light, the antibody-RNA mixtures were incubated with Dynabeads protein A/G (Millipore) at 4 °C overnight. Bead-bound antibody-RNA mixtures were then washed and treated with T4 PNK (New England BioLabs), followed by proteinase K treatment, acidic phenol/chloroform extraction and ethanol precipitation. Purified RNA was reverse transcribed, size selected and used for library construction with NEBNext small RNA library prep kit (E7330S, NEB). Sequencing was performed on Illumina Hiseq4000.

Identification of m⁶A modification by miCLIP-seq

Read pre-processing was performed as previously reported [49]. After removing adaptors and low quality bases using cutadapt (v1.16) and deleting reads less than 20nt, reads were mapped to hg38 human genome by BWA [50] with parameters: bwa aln -n 0.06 -q 20. Crosslinking induced mutation sites (CIMS) and crosslinking-induced truncation sites (CITS) were detected by CLIP Tool Kit (CTK) [51]. For identification of the m⁶A locus, the mode of mutation calling was conducted as previously reported [52].

SNIP1 CLIP-sequencing

CLIP-sequencing were performed as the previous study described [48] with some modifications. Briefly, PANC-1 cells were washed with ice-cold PBS, cross-linked and lysed on ice. Cell lysate was sonicated and centrifuged. The supernatants were incubated with pre-conjugated Dynabeads protein A/G and anti-SNIP1 antibody (Proteintech; 14950-1-AP) overnight at 4 °C with rotating. Bead bound antibody-RNA complexes were subsequently treated with proteinase K. Then RNA was purified by acidic phenol/chloroform extraction and ethanol precipitation. Purified RNA was reverse transcribed and utilized for library construction. Sequencing was performed on Illumina Hiseq4000. For SNIP1 CLIP-qPCR, the input and co-immunoprecipitated RNA samples were simultaneously reverse transcribed, followed by qRT-PCR assays using the specific primers for BCAN-AS1 around the SNIP1 binding site (Supplementary Table 3).

Identification of SNIP1 binding site

Reads were mapped to hg38 human genome by Bowtie2 [53] with parameters: -p 6 -3 5 --local followed by removing adapters and removing low quality reads by fastp [42]. CLIP-seq peaks were identified using Piranha [54] with parameters: -s -b 20 -d ZeroTruncatedNegativeBinomial -p 1e−5. IGV [55] was used to visualize the binding sites of SNIP1.

Public data processing

We used data from TCGA and Genotype-Tissues Expression (GTEx) by GEPIA to assess the differential expression and prognostic value of BCAN-AS1. A public single-cell sequencing data of PDAC (GSA: CRA001160) were used to analyze BCAN-AS1 expression in neoplastic cells and stromal cells in PDAC. RNA-seq data of PDAC tumors from TCGA were used to perform GSEA to gain insight into the biological pathways regulated by BCAN-AS1.

Statistical analysis

For functional analysis, data are presented as mean ± SD of at least three biological replicates. Two-sided Student’s t test was used to examine the difference of mean between two groups when data showed normal distribution, while two-sided nonparametric test was used to analyze data in abnormal distribution. Kaplan–Meier method was used to compare the survival time by different levels of BCAN-AS1. Two-sided log-rank test was used in univariate survival analyses and the Cox proportional hazards model was used in multivariate survival analyses. The hazard ratio (HR) and 95% confidence interval (CI) were calculated with sex, age, smoking status, drinking status, tumor stage, tumor differentiation, neural invasion, vascular invasion, lymph node metastasis and treatment as covariates. We chose median value as cutoff value to distinguish patients with high (≥median) or low (<median) levels of BCAN-AS1. Pearson correlation analysis was employed to measure the correlation between two continuous variables. Correlations were considered significant when P < 0.05 and |r| >0.25. All statistical analyses were performed using the SPSS software package (version 20.0; IBM SPSS) and GraphPad Prism (version 8.0.0). P < 0.05 was considered significant for all statistical analyses.

Results

Identification of c-Myc-interacted lncRNA BCAN-AS1 in PDAC

We started from identifying functional lncRNAs interacted with c-Myc protein in PANC-1 cells using RIP-Seq experiment. As expected, various lncRNAs including the previously reported c-Myc-interacted lncRNAs PVT1 [32], EPIC1 [33] and MALAT1 [36] were identified in our RIP analysis, confirming the reliability of our sequencing data. We selected the top 10 markedly enriched lncRNAs according to the rank of their fold enrichment values (Fig. 1A) to verify their binding to c-Myc in PANC-1 and SW1990 cell lines. Among these 10 lncRNAs, a modestly conserved intergenic lncRNA BCAN-AS1 showed the highest binding affinity to c-Myc in both two PDAC cells (Supplementary Fig. 1A, B). Therefore, we focused our research on BCAN-AS1. By analyzing the public single-cell sequencing data of PDAC, we found that BCAN-AS1 is mainly distributed in neoplastic cells rather than stromal cells (Supplementary Fig. 1C). We then examined the BCAN-AS1 levels in pancreatic tissue samples collected at two cancer centers (Cohort 1: n = 158; Cohort 2: n = 73; Supplementary Table 1), which found that the BCAN-AS1 levels were significantly higher in tumors than in adjacent normal tissues (Fig. 1B), and higher in stages III/IV PDACs than in stages I/II PDACs (Fig. 1C) in both cohorts. Kaplan–Meier analysis showed that patients with high BCAN-AS1 level in PDAC had shorter overall survival time than those with low BCAN-AS1 level (Fig. 1D–F). Additionally, univariate and multivariate Cox regression analysis for overall survival in our PDAC patients showed that BCAN-AS1 level can serve as an independent prognostic factors for PDAC (Supplementary Table 5). These results indicate an oncogenic role of BCAN-AS1 in PDAC progression.

Fig. 1. c-Myc associated lncRNA BCAN-AS1 is overexpressed in PDAC and associated with clinical outcomes in patients with PDAC.

A Heatmap showing the top 10 c-Myc-enriched lncRNAs in input sample and each of the two biological replicates of c-Myc RIP-Seq in PANC-1 cells. LncRNAs shown in heatmap were ranked according to their fold-enrichment (IP/Input). Red (higher enrichment) or blue (lower enrichment) color represents the relative enrichment of indicated lncRNAs. qRT-PCR assays showing that BCAN-AS1 levels were significantly higher in PDAC tumors than in paired non-tumor tissues (B) and in stage III/IV tumors than in stage I/II tumors (C) in two cohorts and combined samples. Y axis refers to the BCAN-AS1 levels relative to the β-Actin mRNA levels. The relative levels of BCAN-AS1 were calculated using the 2^‒ΔCt method. Data are shown in boxplots; the centerlines of the box are medians and the upper and lower lines indicate 25th and 75th percentiles. *P < 0.05; **P < 0.01 and ***P < 0.001 for two-sided Mann–Whitney tests. D–F Kaplan–Meier estimates of survival time in two PDAC patient cohorts and combined samples by different BCAN-AS1 levels in tumors showing that patients with high BCAN-AS1 levels (≥median) had shorter survival time than those with low BCAN-AS1 levels (<median), with adjusted HR = 2.55 (95% CI = 1.69–3.85) for Cohort 1 (D); HR = 2.85 (95% CI = 1.47–5.53) for Cohort 2 (E) and HR = 2.24 (95% CI = 1.59–3.16) for combined samples (F). P values were calculated by two-sided log-rank test. HR hazard ratio, CI confidence interval.

We subsequently tried to characterize BCAN-AS1 in PDAC cell lines. Northern blotting assays detecting total RNA extracted from PANC-1 and SW1990 cells confirmed that BCAN-AS1 is a transcript with a length of about 3200 nucleotides (nt) (Supplementary Fig. 1D). Quantitative RT-PCR assays showed that BCAN-AS1 presented about 400 copies per cell in PANC-1 and SW1990 cells (Supplementary Fig. 1E). Intracellular location analysis and smFISH assays confirmed that BCAN-AS1 was located in both cytoplasm and nucleus but predominantly in nucleus (Supplementary Fig. 1F, G). The coding probability of BCAN-AS1 assessed by the Coding Potential Assessment Tool was 6.2% (with a score >36.4% indicating a potential coding gene) [56]. Consistently, FLAG-fused tagging of putative protein assays confirmed that the BCAN-AS1-FLAG fusion proteins did not exhibit the predicted relative molecular weight determined by western blot assays (Supplementary Fig. 1H) [57]. Additionally, ribosome sedimentation analysis also revealed that BCAN-AS1 exhibits a low affinity for polysome (Supplementary Fig. 1I), together supporting the non-protein-coding feature of BCAN-AS1.

BCAN-AS1 promotes proliferation and invasiveness of PDAC cells

Next, we sought to investigate the effects of BCAN-AS1 on PDAC cell phenotypes by changing its expression in PDAC cell lines (Supplementary Fig. 2A). Results showed that BCAN-AS1 overexpression substantially enhanced but silence markedly suppressed the abilities of cell proliferation, colony formation, migration and invasion of PDAC cells (Fig. 2A–C; Supplementary Fig. 2B, C). To verify the specificity of BCAN-AS1 shRNA, we used the CRISPR-mediated mutation techniques to generate endogenous BCAN-AS1 shRNA binding site mutation in the BCAN-AS1 gene in PDAC cells. We found that transfection of BCAN-AS1 shRNA resulted in significantly decrease BCAN-AS1 level (Supplementary Fig. 2D) and inhibition of PDAC phenotypes (Supplementary Fig. 2E–G) in cells with BCAN-AS1 wild-type shRNA binding site, while not significant in cells with mutant shRNA binding site. We also found similar inhibitory impacts of BCAN-AS1 silencing on cell proliferation, migration and invasion in c-Myc amplified PDAC cells (Supplementary Fig. 2H–J). Comparable effects were observed in mouse xenograft models. BCAN-AS1 overexpression significantly increased growth rates of subcutaneous xenografts while BCAN-AS1 silencing showed opposite effects (Fig. 2D). By utilizing orthotopic xenograft models, we also found that pancreatic tumors formed by PDAC cells with BCAN-AS1 overexpression exhibited remarkably higher growth rates and shorter survival time compared with control cells, while those with BCAN-AS1 silencing showed opposite phenotypes (Fig. 2E–G). Furthermore, BCAN-AS1 overexpression significantly promoted while BCAN-AS1 silencing obviously suppressed lung metastasis of PDAC cells (Fig. 2H, I). Moreover, exogenous overexpression of BCAN-AS1 could restore the inhibition of cell proliferation, migration and invasion in PDAC cells with BCAN-AS1 silencing (Supplementary Fig. 2K, L). Collectively, these findings indicate that BCAN-AS1 acts as an oncogenic lncRNA stimulating proliferation and invasiveness of PDAC cells.

Fig. 2. BCAN-AS1 promotes malignant phenotypes of PDAC in vitro and in vivo.

Effects of BCAN-AS1 overexpression or knockdown on PDAC cell proliferation (A), colony formation (B) and migration or invasion (C). Data are from at least 3 independent experiments. D Images of xenograft tumors (left panel) and tumor growth curves (right panel) showing effects of BCAN-AS1 expression change on PDAC subcutaneous xenograft growth in mice (n = 5). E Representative bioluminescence images showing effects of BCAN-AS1 overexpression or knockdown on the growth of PDAC xenograft transplanted in mouse pancreas (n = 5). F Quantification of fluorescence intensity of tumor burden in mice with orthotopically transplanted PDAC (n = 5). G Effect of BCAN-AS1 expression change on survival time of mice with orthotopically transplanted PDAC (n = 8). H Representative bioluminescence images showing effects of BCAN-AS1 overexpression or knockdown on the pulmonary metastasis of mice implanted PDAC cells from the tail vein (n = 5). I Statistics of fluorescence intensity of pulmonary metastasis in mice with xenograft metastasis models (n = 5). Data are displayed as mean ± SD in A–D, F and I. P-values were calculated by two-sided Student’s t test in A–D, F and I (*P < 0.05; **P < 0.01 and ***P < 0.001) and two-sided log-rank test in G.

BCAN-AS1 exerts oncogenic role via regulating c-Myc function

To uncover the molecular mechanism underlying the effects of BCAN-AS1 on PDAC cell phenotypes, we firstly examined whether BCAN-AS1 has the cis regulation effect on neighboring genes within 0.2 M-bases centering the gene loci of BCAN-AS1 (Supplementary Fig. 3A). Results showed that BCAN-AS1 silencing did not alter the RNA expression of these genes (Supplementary Fig. 3B), suggesting that BCAN-AS1 may not function via the cis regulation. We then performed RNA-seq analyses in PANC-1 cells with BCAN-AS1 silence or BCAN-AS1 overexpression to investigate the gene expression profiles. GSEA revealed that comparing with each PDAC control cells, cells with BCAN-AS1 depletion showed decreased enrichment of gene sets of c-Myc targets (Fig. 3A, B; Supplementary Fig. 3C) while those with BCAN-AS1 upregulation showed opposite results (Supplementary Fig. 3D, E). Furthermore, we performed the GSEA using the PDAC RNA-Seq data from TCGA based on the median of BCAN-AS1 levels, and also found that the gene signatures of c-Myc targets were enriched in patients with high BCAN-AS1 expression (Supplementary Fig. 3F, G). Notably, most of gene sets positively regulated by BCAN-AS1 including c-Myc targets, E2F target, G2M checkpoint target, DNA repair target, Oxidative Phosphorylation target and Mitotic Spindle target (Supplementary Fig. 3C, E) were positively enriched in c-Myc amplified tumors [15], indicating that BCAN-AS1 regulates most of pathways directly through c-Myc. To further gain insight into the biological mechanism involved in the oncogenic role of BCAN-AS1, we overlapped genes downregulated (>1.5-fold) by BCAN-AS1 silencing in PANC-1 cells with genes positively associated with BCAN-AS1 expression (r > 0.25, P < 0.05) in the PDAC RNA-Seq data from TCGA (Fig. 3C). 202 high-confidence BCAN-AS1-regulated targets were identified, most of which were enriched in canonical c-Myc regulated pathways such as cell cycle, differentiation and metabolism pathways (Fig. 3C, Supplementary Fig. 3H). These findings were further verified in PDAC cells by qRT-PCR assays (Supplementary Fig. 3I). Western blot analysis also showed that BCAN-AS1 knockdown significantly reduced the expression of c-Myc targets such as CDC6, CDC45, Cyclin A2, MCM7 and PAICS (Fig. 3D), mimicking the effects of c-Myc silencing (Fig. 3E). We further observed that knocked down of c-Myc abolished the promoting effects by BCAN-AS1 overexpression on c-Myc targets expression and malignant cell phenotypes (Fig. 3F–I, Supplementary Fig. 4A, B), and overexpression of c-Myc reversed the inhibition effects by BCAN-AS1 knockdown (Supplementary Fig. 4C–E). Moreover, BCAN-AS1 knockdown suppressed the upregulated levels of c-Myc targets (Supplementary Fig. 4F) and the strengthened abilities of cell proliferation, migration and invasion (Supplementary Fig. 4G, H) by forced c-Myc expression, suggesting that BCAN-AS1 and c-Myc cooperate to regulate its targets in a c-Myc-dependent manner. To clarify the contribution of BCAN-AS1 in cells with different c-Myc genomic statuses, we treated PDAC cells with JQ1 upon BCAN-AS1 knockdown. Compared to the control groups, we found that BCAN-AS1 knockdown inhibited proliferation, migration and invasion more significantly when PDAC cells were treated with JQ1 (Supplementary Fig. 4I, J), indicating that BCAN-AS1 may promote the malignant phenotype of PDAC cells without c-Myc amplification more strongly than that of PDAC cells harboring amplified c-Myc. Together, these results demonstrate the association between BCAN-AS1 and c-Myc protein which is required for the pro-tumorigenic function of BCAN-AS1.

Fig. 3. BCAN-AS1 upregulates c-Myc targets to enhance malignant phenotypes of PDAC cells.

A, B Gene set enrichment analyses (GSEA) plots of the c-Myc gene sets showing association between the enrichment of c-Myc targets and BCAN-AS1 expression in BCAN-AS1 knockdown cells. C Overview of BCAN-AS1-regulated gene identification (left panel) and expression level of representative BCAN-AS1-regulated gene by RNA-seq (right panel). Western blot analysis of c-Myc-regulated targets in PDAC cells upon BCAN-AS1 (D) or MYC (E) knockdown. F Western blot analysis of c-Myc-regulated targets in PDAC cells with BCAN-AS1 overexpression and MYC silence. Effects of MYC knockdown on cell proliferation (G), colony formation (H), migration and invasion (I) in BCAN-AS1 overexpression PDAC cells. Data represent mean ± SD in G–I) from 3 independent experiments. *P < 0.05; **P < 0.01 and ***P < 0.001 by two-sided Student’s t test in G–I. β-Actin served as an internal control in D–F.

BCAN-AS1 specifically interacts with c-Myc and upregulates c-Myc expression

We further explore the interaction between BCAN-AS1 and c-Myc protein. RNA pull-down assays showed that c-Myc protein could be specifically co-precipitated by biotin-labeled BCAN-AS1 sense probe, but not by the BCAN-AS1 antisense probe (Fig. 4A). ChIRP assays (Fig. 4B, C) further confirmed the particular interaction between endogenous BCAN-AS1 and c-Myc protein. To further determine the BCAN-AS1 functional motifs corresponding to c-Myc binding, we conducted an in vitro RNA pull-down assay with various truncated BCAN-AS1 fragments. Results revealed that nucleotides 1550–2073 of BCAN-AS1 are required for its interaction with c-Myc protein (Fig. 4D), which was further verified by c-Myc CLIP-qPCR assays (Fig. 4E). Protein domain mapping experiments with Flag-labeled truncated c-Myc and BCAN-AS1 further showed that deletion of the 148 to 220 amino acid (aa) region of c-Myc protein abolished its interaction with BCAN-AS1 (Fig. 4F), which was supported by Flag-CLIP-qPCR assays (Fig. 4G). Together, these findings indicate that BCAN-AS1 directly and specifically binds to c-Myc protein. Additionally, quantitative western blot assay documented an average of 14,846 and 20,327 c-Myc proteins per cell in PANC-1 and SW1990 cell lines respectively (Supplementary Fig. 5A). We further used the RNA-Protein Interaction Prediction (RPISeq) tool [58] to map the c-Myc potential binding region (MBR) in the 1550–2073 fragment of BCAN-AS1 transcript. Strikingly, there are 20 potential binding regions of c-Myc in the 1550–2073 fragment of BCAN-AS1 transcript (Supplementary Fig. 5B), indicating that BCAN-AS1 has the capacity to bind >40% c-Myc protein molecules per cell in PANC-1 and SW1990 cell.

Fig. 4. BCAN-AS1 directly interacts with c-Myc and increases c-Myc protein expression in PDAC cells.

A RNA pull-down coupled western blot analysis showing that c-Myc bound to BCAN-AS1 but not its antisense. B, C ChIRP assays showing proteins retrieved by BCAN-AS1-odd, or -even antisense probe sets. LacZ and β-ACTIN respectively served as negative controls for the probe and retrieved protein. B Shows RNA retrieval rate and C shows western blot assays of c-Myc and β-ACTIN in ChIRP and input. Data are mean ± SD from 3 assays. D Pull-down assays with BCAN-AS1 full-length and truncated fragments showing the binding fragments of BCAN-AS1 to c-Myc protein. Left panel, schematic of BCAN-AS1 full-length and truncated fragments. Right panel, RNA sizes of in vitro transcribed BCAN-AS1 full-length and truncated fragments (Upper panel), and immunoblot analysis of c-Myc pulled down by different BCAN-AS1 fragments (Bottom panel). E c-Myc CLIP-qPCR assays showing direct binding of c-Myc protein towards the 1550–2073 nucleotides of BCAN-AS1. P1 primer 1 targeting 1550–1788 region of BCAN-AS1, P2 primer 2 targeting 1789–2073 region of BCAN-AS1. F Domain mapping of c-Myc binding to BCAN-AS1. Left panel, schematic diagram of Flag-tagged c-Myc and its truncated versions used in BCAN-AS1 pull-down assays. Right panel, western blot analysis of Flag-tagged full-length (FL) c-Myc and its truncated versions pulled down by in vitro transcribed biotinylated BCAN-AS1. G FLAG-CLIP-qPCR assays showing in-vivo binding of BCAN-AS1 to 148-220aa fragment of c-Myc protein. Cells were transfected with Flag-tagged c-Myc and its truncated versions. P1 primer 1 targeting 1550–1788 region of BCAN-AS1, P2 primer 2 targeting 1789–2073 region of BCAN-AS1. Effects of BCAN-AS1 expression change on c-Myc mRNA levels by qRT-PCR assays (H) and on c-Myc protein levels by western blot assays (I) in PDAC cells. β-Actin served as internal control. Data are mean ± SD in B, E, G and H. ***P < 0.001 and ns not significant for two-sided Student’s t tests in G and H.

With the observation that BCAN-AS1 directly interacts with c-Myc, we further examined the effect of BCAN-AS1 on c-Myc expression. We found no change of c-Myc mRNA levels in PDAC cells with BCAN-AS1 overexpression or silencing (Fig. 4H). Consistently, result of TCGA data analysis also showed no significant correlation between BCAN-AS1 levels and c-Myc RNA levels in PDAC (Supplementary Fig. 5C). However, BCAN-AS1 overexpression significantly increased while BCAN-AS1 knockdown decreased the protein level of c-Myc (Fig. 4I, Supplementary Fig. 5D). Similar results were observed in c-Myc amplified PDAC cell lines (Supplementary Fig. 5E, F). Collectively, all these results demonstrate that BCAN-AS1 directly binds to c-Myc protein and enhances c-Myc protein expression.

BCAN-AS1 strengthens interaction of SNIP1 and c-Myc to block SKP2-mediated c-Myc ubiquitination

We next investigated how BCAN-AS1 promotes c-Myc protein expression. Treatment of BCAN-AS1-silenced cells with the protein synthesis inhibitor cycloheximide (CHX) led to apparently decreased protein level and shorter half-life for c-Myc than in control cells (Fig. 5A). Furthermore, the proteasome inhibitor MG132 treatment abolished the decrease in c-Myc protein induced by BCAN-AS1 knockdown (Fig. 5B). These results indicate that the ubiquitin-proteasome pathway might have a significant role in the BCAN-AS1-mediated regulation of c-Myc protein. Indeed, ubiquitination assays showed that BCAN-AS1 overexpression inhibited endogenous ubiquitinated c-Myc in PDAC cells, while BCAN-AS1 knockdown increased the c-Myc ubiquitination (Fig. 5C).

Fig. 5. BCAN-AS1 mediates increased binding of SNIP1 to c-Myc and prevents c-Myc degradation by SKP2.

A Protein levels of c-Myc in BCAN-AS1 knockdown or control PDAC cells with or without the treatment of cycloheximide (CHX; 20 μg/mL). Left panel, immunoblotting images of c-Myc; Right panel, curves showing the half-life of c-Myc protein. B Protein levels of c-Myc in BCAN-AS1 knockdown or control PDAC cells with or without the treatment of MG132 (20 μM) for 24 h. C Effects of BCAN-AS1 expression change on c-Myc ubiquitination level in PDAC cells treated with MG132 as detected by immunoblotting. D Protein levels of c-Myc in BCAN-AS1 knockdown or control PDAC cells without or with SKP2 silencing by western blot assays. E c-Myc ubiquitination level in BCAN-AS1 knockdown or control PDAC cells without or with SKP2 silencing by immunoblotting. F, G Immunoprecipitation and immunoblotting assays showing effects of BCAN-AS1 expression change on the associations between endogenous c-Myc and SNIP1 or SKP2 in PDAC cells. β-Actin served as internal control.

Although c-Myc phosphorylation plays an important role in regulation of c-Myc ubiquitination [59], we found no change of phosphorylation level at the two conserved c-Myc phosphorylation sites, threonine 58 (T58) and serine 62 (S62) (Supplementary Fig. 6A). Since BCAN-AS1 is predominantly distributed in nucleus, and SKP2 is a substrate-recognition subunit of the SCF E3 ubiquitin ligase that has been shown to interact with c-Myc in nucleus and stimulate c-Myc degradation via mediating c-Myc phosphorylation-independent ubiquitination [59–61], we next examined whether BCAN-AS1 regulates SKP2-mediated c-Myc ubiquitination. As expected, the effects of BCAN-AS1 on c-Myc protein level (Fig. 5D) and ubiquitination (Fig. 5E) were attenuated when SKP2 was silenced, suggesting that BCAN-AS1 inhibits SKP2-mediated c-Myc ubiquitination and degradation.

We next investigated how BCAN-AS1 inhibits SKP2-mediated c-Myc ubiquitination. SNIP1 has been found to compete with SKP2 for binding to c-Myc and stabilize c-Myc against SKP2-mediated ubiquitination and proteosomal degradation [62]. We found that BCAN-AS1 expression change had no influence on the protein levels of SNIP1 and SKP2 (Supplementary Fig. 6B). Then we assessed whether BCAN-AS1 inhibits SKP2-mediated c-Myc ubiquitination via regulating the competitive bindings of SNIP1 and SKP2 towards c-Myc. As expected, overexpression of BCAN-AS1 significantly strengthened the binding of SNIP1 towards c-Myc, while silencing of BCAN-AS1 substantially weakened this interaction (Fig. 5F, G). On the contrary, in BCAN-AS1-overexpressed cells, the interaction of c-Myc and SKP2 was markedly reduced, whereas in BCAN-AS1-silenced cells, this interaction was significantly increased (Fig. 5F, G). Taken together, these findings suggested that BCAN-AS1 facilitates the binding of SNIP1 to c-Myc, and then competitively inhibits the SKP2-mediated c-Myc ubiquitination and degradation.

We further explored whether SNIP1 is required for aggressive phenotypes mediated by BCAN-AS1. Results showed that SNIP1 knockdown substantially inhibited growth of pancreatic orthotopic transplanted xenografts and reduced lung metastasis of PDAC cells compared with control cells (Supplementary Fig. 6C–E), which was similar to the effects of BCAN-AS1 silencing. Moreover, the promotion effects of BCAN-AS1 overexpression on the growth of orthotopic pancreatic xenografts and lung metastasis of PDAC cells were attenuated by intravenous injection of cholesterol-conjugated SNIP1 siRNAs (Supplementary Fig. 6F, G). To confirm that SNIP1 was knocked down in vivo by siRNA, we measured SNIP1 expression by western blot and found that it was significantly decreased in SNIP1 siRNA-treated orthotopic pancreatic xenografts (Supplementary Fig. 6H). All these in vivo results indicate a critical role of SNIP1 in the BCAN-AS1 regulation of malignant phenotypes in PDAC cells.

SNIP1 binds to BCAN-AS1 through recognizing the m⁶A modification of BCAN-AS1

Since the above data showed that BCAN-AS1 binds to c-Myc, and strengthens the c-Myc/SNIP1 interaction, we presumed that BCAN-AS1 may act as a scaffold to mediate the binding between c-Myc and SNIP1. We performed ChIRP assays and found that SNIP1 but not SKP2 could be precipitated by BCAN-AS1 (Fig. 6A). Consistently, RIP assays further showed enrichment of BCAN-AS1 in complexes precipitated by antibodies against SNIP1 (Supplementary Fig. 7A), indicating direct interaction between BCAN-AS1 and SNIP1. These results suggest that BCAN-AS1 serves as a scaffold of c-Myc/SNIP1 interaction.

Fig. 6. BCAN-AS1 recruits SNIP1 binding via its m⁶A modification.

A ChIRP assays showing the levels of SNIP1 or SKP2 retrieved by BCAN-AS1-odd, or -even antisense probes in PDAC cells. B IGV snapshots showing the SNIP1 CLIP-seq and miCLIP-seq read distribution on BCAN-AS1. Yellow tracks are unique tag coverage of SNIP1 CLIP-seq and blue tracks are unique tag coverage of miCLIP-seq. Filled red circle denotes the miCLIP-called m⁶A site. C Effects of METTL3 knockdown on the m⁶A levels of BCAN-AS1 in PDAC cells as detected by meRIP-qPCR analysis. D Effects of wild-type METTL3 or its catalytic mutant on m⁶A levels of BCAN-AS1 in cells with METTL3 knockout. E m⁶A levels of BCAN-AS1 in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. F Scatter plot showing proteins in PANC-1 cells that bound to 50 bp m⁶A-modified or m⁶A-unmodified BCAN-AS1 probe. Unique peptide of >6 was selected as filter criterium. The red point represents SNIP1 protein. G RNA pull-down coupled western blot showing specific interaction of m⁶A-modified BCAN-AS1 probe with SNIP1 protein. H Electrophoretic mobility shift assays showing higher binding infinity of recombinant SNIP1 towards m⁶A-modified BCAN-AS1 probes than m⁶A-unmodified BCAN-AS1 probes. I The levels of SNIP1 bound to BCAN-AS1 in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. J Western blot assays showing SNIP1 protein levels in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. K Co-localization of c-Myc and SNIP1 protein in PDAC cells was decreased upon BCAN-AS1 depletion, as indicated by smFISH and immunofluorescence. Scale bars, 10 µm. L ChIRP assays showing the levels of SNIP1 and c-Myc retrieved by BCAN-AS1-odd, or -even antisense probes in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. M Immunoprecipitation and immunoblotting assays showing the associations between endogenous c-Myc and SNIP1 or SKP2 in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. N Immunoprecipitation and immunoblotting assays showing the associations between endogenous c-Myc and SNIP1 or SKP2 in BCAN-AS1 overexpressed PDAC cells without or with co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. c-Myc ubiquitination (O) or protein (P) level in PDAC cells co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. c-Myc ubiquitination (Q) or protein (R) level in BCAN-AS1 overexpressed PDAC cells without or with co-transfected with DOX-inducible dCas13b-ALKBH5 plasmid and NT-gRNA (control) or gRNAs with doxycycline pretreatment. LacZ and β-ACTIN respectively served as negative controls for the probe and retrieved protein in A and L. Data are mean ± SD from 3 independent experiments in C–E and I, and *P < 0.05; **P < 0.01 and ***P < 0.001 by two-sided Student’s t test.

We further explored the precise mechanism underlying the interaction of SNIP1 and BCAN-AS1. We performed CLIP-seq of SNIP1 and the results showed a direct binding of SNIP1 to BCAN-AS1 (Fig. 6B). Interestingly, by analyzing our miCLIP-seq data and SNIP1 CLIP-seq data, we observed an m⁶A site in BCAN-AS1 overlapping with the SNIP1 binding site (Fig. 6B). We then verified the presence of m⁶A site in BCAN-AS1 using m⁶A-specific RNA immunoprecipitation coupled qRT-PCR (meRIP-qPCR) analysis and found that m⁶A modification of BCAN-AS1 were mediated by the core m⁶A methyltransferase METTL3 (Fig. 6C; Supplementary Fig. 7B) in a catalytic activity-dependent manner [63, 64] (Fig. 6D; Supplementary Fig. 7C). Besides, we also applied the dCas13b-ALKBH5 (dm⁶ACRISPR) system, which can site-specifically demethylate m⁶A modification in RNAs [65, 66], to confirm the occurrence of m⁶A on BCAN-AS1 (Fig. 6E; Supplementary Fig. 7D, E). However, neither METTL3 silencing nor the dm⁶ACRISPR system had influence on BCAN-AS1 expression (Supplementary Fig. 7F, G).

m⁶A modification involved in the regulation of binding affinity between RNA binding proteins and target RNAs, hence we next investigated whether the m⁶A modification is crucial for the interaction between SNIP1 and BCAN-AS1. We first performed mass spectrometry analysis of proteins obtained by RNA pull-down using 50-nt biotin-labeled m⁶A methylated or unmethylated BCAN-AS1 probes. The results showed that SNIP1 was preferentially bound towards the BCAN-AS1 probe with m⁶A modification (Fig. 6F; Supplementary Data 1). This finding was further verified via RNA pull-down followed by western blot assays (Fig. 6G; Supplementary Fig. 7H) and RNA EMSA (Fig. 6H). Domain mapping assay found that the amino acid residues 97–274 in SNIP1 were required for its binding to m⁶A-modified BCAN-AS1 (Supplementary Fig. 7I, J). Moreover, we found that decreased m⁶A level on BCAN-AS1, by METTL3 silencing, METTL3 catalytic inactivating or applying the dm⁶ACRISPR system, can significantly reduce the binding of SNIP1 to BCAN-AS1 (Fig. 6I, Supplementary Fig. 7K, L) without altering the protein level of SNIP1 protein (Fig. 6J; Supplementary Fig. 7M). All these findings indicate that m⁶A modification of BCAN-AS1 is essential for the binding of SNIP1 towards BCAN-AS1. We further examined the effects of BCAN-AS1 m⁶A modification on aggressive phenotypes of PDAC cells in vivo, and results showed that demethylation of BCAN-AS1 m⁶A using the dm⁶ACRISPR system significantly reduced pancreatic orthotopic xenografts (Supplementary Fig. 7N, O) and lung metastasis (Supplementary Fig. 7P, Q) of PDAC cells, which were similar to BCAN-AS1 knockdown and SNIP1 knockdown. Orthotopic xenografts with BCAN-AS1 m⁶A demethylation also showed decreased binding of SNIP1 to BCAN-AS1 (Supplementary Fig. 7R).

Moreover, integrated analysis of miCLIP-seq and SNIP1 CLIP-seq data further showed a high binding intensity for SNIP1 centering at m⁶A residues and vice versa for the m⁶A sites (Supplementary Fig. 7S). Meanwhile, a substantial overlap between the m⁶A-modified RNAs and the SNIP1 binding RNAs was observed (Supplementary Fig. 7T), suggesting that SNIP1 may function as an m⁶A mediator recognizing m⁶A sites in RNAs besides BCAN-AS1.

m⁶A-modified BCAN-AS1 bridges the c-Myc/SNIP1 complex

As BCAN-AS1/SNIP1 interaction is m⁶A-dependent, we next characterized the role of m⁶A modification in the c-Myc/BCAN-AS1/SNIP1 complex regulation. SmFISH and immunofluorescence assays showed that BCAN-AS1 was co-located with c-Myc and SNIP1 proteins in PDAC cells (Fig. 6K), indicating the presence of a c-Myc/BCAN-AS1/SNIP1 ternary complex. Line scan graphs analysis and staining density quantification demonstrated that BCAN-AS1 silencing led to a significant decrease of c-Myc protein level and a dramatically reduced co-localization of c-Myc and SNIP1 (Fig. 6K; Supplementary Fig. 8A), whereas BCAN-AS1 overexpression showed opposite results (Supplementary Fig. 8B, C). ChIRP assays showed that reduction of m⁶A modification in BCAN-AS1 by using the dm⁶ACRISPR system (Fig. 6L) or depleting METTL3 (Supplementary Fig. 8D) specifically attenuated the interaction between SNIP1 and BCAN-AS1 but not that between c-Myc and BCAN-AS1. Notably, we found weakened c-Myc/SNIP1 interaction and strengthened c-Myc/SKP2 association in PDAC cells treated with the dm⁶ACRISPR system (Fig. 6M; Supplementary Fig. 8E) where both BCAN-AS1 m⁶A levels and the SNIP1 binding to BCAN-AS1 were substantially reduced. Similarly, exogenous overexpression of wild-type but not m⁶A mutant BCAN-AS1 significantly enhanced the interaction between c-Myc and SNIP1, and blocked that of c-Myc and SKP2 (Supplementary Fig. 8F, G). Moreover, we observed that the BCAN-AS1-enhanced interaction of SNIP1 and c-Myc was significantly impaired in PDAC cells where the m⁶A modification of BCAN-AS1 were demethylated using the dm⁶ACRISPR system, whereas effect on the interaction of SKP2 and c-Myc was reversed (Fig. 6N; Supplementary Fig. 8H). These results indicate that the enhancement of c-Myc/SNIP1 interaction mediated by BCAN-AS1 is m⁶A-dependent. As expected, the increased ubiquitinated c-Myc (Fig. 6O) and the decreased c-Myc protein level (Fig. 6P) were observed in cells with BCAN-AS1 m⁶A demethylation. PDAC cells overexpressing wild-type but not m⁶A mutant BCAN-AS1 showed reduced ubiquitination (Supplementary Fig. 8I) and increased protein level (Supplementary Fig. 8J) of c-Myc. Moreover, forementioned effects induced by BCAN-AS1 overexpression could be restored when BCAN-AS1 m⁶A was demethylated (Fig. 6Q, R). Taken together, these results indicate that m⁶A modification of BCAN-AS1 is indispensable for enhancing the c-Myc/SNIP1 interaction.

As SKP2 interaction with c-Myc is compromised by BCAN-AS1, we speculate that BCAN-AS1 may result in SKP2 being more proficient in targeting those genes blocking progression along the cell cycle. Therefore, we detected the post-translational modification of p27, a classical ubiquitin protein ligase complex SCF^Skp2 target protein that regulates cell cycle progression. Results showed that p27 ubiquitination levels were decreased and its protein levels were increased after BCAN-AS1 knockdown (Supplementary Fig. 8K, L). These results indicate that perturbation of BCAN-AS1 expression affect cell-cycle progression may partially due to the upregulation of SKP2-regulated genes, e.g. p27, which can block the progression along the cell cycle.

BCAN-AS1 is a potential therapeutic target for PDAC

To further confirm the clinical significance of BCAN-AS1 and c-Myc, we first examined the levels of c-Myc and BCAN-AS1 in 60 paired PDAC and adjacent normal samples. The results showed that PDAC tumors had significantly higher levels of c-Myc protein (Fig. 7A; Supplementary Fig. 9A) and BCAN-AS1 (Fig. 7B) than the adjacent normal tissues. Correlation analysis showed that BCAN-AS1 levels were significantly and positively associated with c-Myc protein levels in PDAC tumor and normal tissues (Fig. 7C; Supplementary Fig. 9B). Since BCAN-AS1 had a strong oncogenic effect in PDAC cells, we then perform experimental therapy in mouse xenograft and metastatic models using in vivo-optimized antisense oligonucleotide of BCAN-AS1 (ASO-BCAN-AS1), a BCAN-AS1 inhibitor, to evaluate its therapeutic effects. We first examined the effects of ASO-BCAN-AS1 on in-vitro phenotypes of PDAC cells and found that depletion of BCAN-AS1 in PDAC cells by ASO-BCAN-AS1 (Supplementary Fig. 9C) remarkably inhibited cell proliferation, migration and invasion (Supplementary Fig. 9D, E). To verify the specificity of ASO-BCAN-AS1, we used the CRISPR-mediated mutation techniques to generate ASO-BCAN-AS1 binding site mutation in the BCAN-AS1 gene in PDAC cells. We found that BCAN-AS1 level were significantly reduced by ASO-BCAN-AS1 in PDAC cells knocked in wild-type but not ASO-BCAN-AS1 binding site mutant BCAN-AS1 (Supplementary Fig. 9F). ASO-BCAN-AS1 treatment significantly inhibited the cell proliferation, migration and invasion, and c-Myc protein expression of PDAC cells with wild-type BCAN-AS1 but not of those with ASO-BCAN-AS1 binding site mutant BCAN-AS1 (Supplementary Fig. 9G−I). Exogenous overexpression of BCAN-AS1 could restore the inhibition of cell proliferation, migration and invasion in PDAC cells with BCAN-AS1 silence by ASO-BCAN-AS1 (Supplementary Fig. 9J, K). In-vivo experimental therapy showed that treatment with ASO-BCAN-AS1 in mice with metastasis xenograft derived from BCAN-AS1-overexpressing PDAC cells significantly diminished lung metastases compared with treatment with ASO-Control (Fig. 7D–F). In mouse PDX models, comparing with treatment with ASO-Control, treatment with ASO-BCAN-AS1 significantly reduced PDX growth rates (Fig. 7G−I) without influencing the mouses’ bodyweight (Supplementary Fig. 9L). Besides, treatment with ASO-BCAN-AS1 significantly reduced the levels of BCAN-AS1 (Fig. 7J) and c-Myc protein (Fig. 7K). To observe the long-term outcome of mice after withdrawal of ASO-BCAN-AS1 treatment, we extended the feeding time of mice after treatment in the PDX and metastasis xenograft models. Results showed that the growth of PDXs and metastasis xenograft increased but slowly after ASO-BCAN-AS1 withdrawal (Supplementary Fig. 9M, N), indicating that ASO-BCAN-AS1 treatment might have long-term inhibitory effect after the treatment withdrawal. Taken together, these results indicate that BCAN-AS1 may be a new option for targeting treatment of PDAC.

Fig. 7. BCAN-AS1 is a therapeutic target in mouse PDAC xenografts.

A Quantification statistics of c-Myc protein levels in PDAC and paired non-tumor tissues (n = 60) by western blot analysis. B BCAN-AS1 levels in PDAC tumor and paired normal tissues (n = 60) by qRT-PCR assays. C Pearson’s correlation between c-Myc protein level and BCAN-AS1 levels in PDAC tumor tissues (n = 60). D Timeline schematic for treatment of mice with lung metastases. Colored arrows indicate the times when different events occurred. Representative bioluminescence images (E) and quantitative fluorescent intensities (F; n = 5) of mice with lung metastases with in vivo-optimized ASO-BCAN-AS1 or ASO-Ctrl treatment. G Timeline schematic for treatment of mice carrying PDXs. Colored arrows indicate the times when different events occurred. Images of PDXs from 3 patients in 5 mice (H) and PDX tumor growth curves with or without being treated with ASO-BCAN-AS1 (I). Effects of ASO-BCAN-AS1 treatment on BCAN-AS1 level by qRT-PCR (J) and on c-Myc protein level by western blot analysis (K). Upper panel and Bottom panel in K respectively showing the immunoblotting images and quantification of c-Myc protein. Each protein band was semi-quantified by gray density and the value for each band is relative to density of corresponding band of R. L Proposed acting model for BCAN-AS1 that regulates the development and progression of PDAC. Data in this figure are mean ± SD. **P < 0.01; ***P < 0.001 and ns non-significant by two-sided Student’s t test.

Discussion

In this study, we performed c-Myc RIP-seq analysis on PDAC cells and identified a panel of c-Myc-associated lncRNAs. Among them, we characterized that BCAN-AS1 acts as a pro-tumorigenic lncRNA in PDAC by regulating c-Myc function. Mechanistically, overexpressed BCAN-AS1 may directly interact with c-Myc and recruit SNIP1 via its m⁶A modification to form c-Myc/BCAN-AS1/SNIP1 ternary complex, thereby competitively inhibiting SKP2-mediated c-Myc ubiquitination (Fig. 7L). Importantly, we have demonstrated that inhibiting BCAN-AS1 significantly suppressed the growth and metastasis of mouse xenograft and metastatic models. These results shed light on a molecular mechanism underlying aberrantly expressed BCAN-AS1 in c-Myc regulation and provide a potential druggable target for PDAC treatment.

Overexpressed c-Myc is a frequent event in human cancer including PDAC [12]. Dysregulation of lncRNAs has been reported to participate in various cancer hallmarks [28, 29, 33], including c-Myc regulation [30, 31]. All these previous findings encourage us to conduct c-Myc-RIP-seq to screen the decisive tumor-related transcripts bound by c-Myc in PDAC. As expected, some classical c-Myc interacted lncRNAs like PVT1 [32] were identified in our c-Myc RIP analysis, confirming the validity of our RIP-seq. Further analysis showed that lncRNA RP11‐284F21.9 also named BCAN-AS1 was the most c-Myc-enriched lncRNA of two PDAC cell lines. BCAN-AS1 has been reported to promote oral squamous cell carcinoma development via the miR-383-5p/MAL2 axis [67]. BCAN-AS1 also promotes lung carcinoma proliferation and invasion via the regulation of miR-627-3p/CCAR1 [68]. In cervical carcinoma, BCAN-AS1 has been identified as a tumor suppressor, regulating PPWD1 by competitively binding to miR-769-3p [69]. Up to now, all three researches relevant to BCAN-AS1 focus on the competing endogenous RNAs (ceRNA) mechanism and these studies suggest that the role of BCAN-AS1 in cancer is different and complicate. In our study, we have found that the levels of BCAN-AS1 are significantly increased in PDAC, especially in advanced tumors. More importantly, our in-vitro and in-vivo functional analyses further verified the oncogenic role of BCAN-AS1 in PDAC. To exclude the possibility that BCAN-AS1 exerts its oncogenic function by cis regulation, we performed a series of experiments, and found that BCAN-AS1 knockdown did not influence the expression of its neighboring genes and the phenotype induced by BCAN-AS1 silencing could be restored by BCAN-AS1 itself, indicating that the phenotype of BCAN-AS1 knockdown is caused by BCAN-AS1 itself.

Since BCAN-AS1 binds to c-Myc, we hypothesize that BCAN-AS1 may exert its function via c-Myc. This notion has been verified by the high-throughput sequencing data and functional experiments which suggested a c-Myc-dependent oncogenic role of BCAN-AS1. When we explored the mechanisms by which BCAN-AS1 contributes to c-Myc regulation, we demonstrated that BCAN-AS1 formed a ternary complex together with c-Myc and SNIP1, and enhanced the c-Myc/SNIP1 interaction, thereby inhibiting SKP2-mediated c-Myc ubiquitination. SNIP1, a 396-amino acid nuclear protein, exerts its promotion or inhibition functions for transcription of target genes by binding various transcription factors or histone modification complexes [62, 70–72]. Previous studies have reported that SNIP1 competes with SKP2 for binding to c-Myc and stabilizes c-Myc against SKP2-mediated ubiquitination and proteosomal degradation [62]. Meanwhile, SNIP1 bridges c-Myc/p300 complex and cooperates with p300 to enhance the c-Myc transcriptional activity [62]. As observed in our study, perturbing the expression of BCAN-AS1 could affect the expression of c-Myc targets, which we believe is at least partially due to the increased SNIP1/c-Myc interaction and stabilization of c-Myc protein induced by BCAN-AS1. However, whether BCAN-AS1 could alter the SNIP1/c-Myc/p300 ternary complex to influence c-Myc transcriptional activity needs further investigation.

Accumulating evidence has shown that lncRNA could function as a scaffold to bring different proteins together to form functional complex [73]. As our result shown, BCAN-AS1 could bind to both c-Myc and SNIP1, whereas no interaction was observed between BCAN-AS1 and SKP2, indicating that BCAN-AS1 may act as a scaffold between c-Myc and SNIP1. smRNA-FISH further shows the presence of a c-Myc/BCAN-AS1/SNIP1 ternary complex. These findings indicate a new working mechanism of BCAN-AS1 which was different from the ceRNA mechanism reporting in the previous studies [67–69]. Moreover, our study has also demonstrated for the first time that noncoding RNAs act as scaffolds to enhance the c-Myc and SNIP1 interaction, and this acting model was different from a recent study showing that lncRNA could act as a molecular guide interacting with SNIP1 to guide the SNIP1 binding to c-Myc [74]. All these discoveries extend the functional role of noncoding RNAs in the regulation of c-Myc protein, providing an important supplement for the regulation network of c-Myc.

c-Myc elevated expression could be regulated through different mechanisms. Genomic amplifications of c-Myc have been identified as an important contributor to its elevated expression and accumulating studies have found c-Myc amplifications to be a driver of cancer metastasis [15–17]. In our present study, we have elucidated the mechanism of c-Myc elevated expression from the perspective of post-translational modification. However, whether the acting mechanism of BCAN-AS1 to c-Myc is also working in c-Myc amplified tumor cells remained unclear. As observed in our study, BCAN-AS1 could promote aggressive cell phenotypes and upregulate c-Myc protein levels both in c-Myc amplified and non-amplified PDAC cells, indicating that the post-translational regulation of BCAN-AS1 to c-Myc is similar in PDAC cells with and without c-Myc amplifications. It would be interesting to further explore the extent to which BCAN-AS1 mediated post-translational regulation contributes to c-Myc elevated expression in c-Myc amplified cells.

To explore why BCAN-AS1 is overexpressed in PDAC, we first analyzed genomic alterations including mutations and CNAs of BCAN-AS1 in PDAC tissues derived from cBioPortal database [75, 76] and the results were negative (data not shown). We also analyzed DNA methylation of BCAN-AS1 promoter in PDAC tissues from TCGA database and found no significant difference of DNA methylation levels in the promoter region of BCAN-AS1 between tumor and normal tissues (data not shown). These results indicated that elevated expression of BCAN-AS1 may not be caused by abnormal genomic changes. We further predicated potential transcription factors binding to BCAN-AS1 promoter via ChIPBase [77] database and results showed that a set of transcription factors such as SPI1, ESR1, AR and STAT3 potentially bound to BCAN-AS1 promoter, indicating that upregulation of BCAN-AS1 may be caused by dysregulation of transcription. Notably, c-Myc was among the predicted transcription factors, suggesting a possibility that the elevated c-Myc might drive the increased BCAN-AS1 expression. It would be interesting and worth further exploring whether there is a regulatory loop between c-Myc and BCAN-AS1.

Interestingly, our study has linked m⁶A modification to the functional mechanism of BCAN-AS1 in PDAC and characterized SNIP1 as a novel m⁶A mediator. M⁶A modification is one of the most common RNA modifications occurring on almost all types of RNAs and participating in many critical RNA biological processions, such as RNA processing, nuclear export, RNA stability and translation [78]. However, investigations of the functions of m⁶A modification in lncRNAs are still limited. In this study, by comprehensively analyzing the miCLIP-seq and SNIP1-CLIP-seq data, we identified an m⁶A modification site on BCAN-AS1 transcript which overlaps with the SNIP1 binding sites. Since m⁶A modification exerts its important biological function via specific RNA binding proteins (readers) [78], we suspect m⁶A modification may affect the BCAN-AS1 binding to its interacting proteins such as SNIP1. Indeed, this notion was verified by several experimental settings which indicated that SNIP1 specifically binds to BCAN-AS1 transcript via recognizing its m⁶A modification site. Our findings suggest that SNIP1 might be an m⁶A mediator of BCAN-AS1. Notably, the finding that many m⁶A sites overlap with the SNIP1 binding sites indicates that SNIP1 may be a common reader recognizing the m⁶A sites in part of m⁶A-modified RNAs, which is warranted for further investigation. Furthermore, we showed that the SNIP1/c-Myc interaction was positively regulated by the m⁶A modification of BCAN-AS1, suggesting a central role of m⁶A modification in facilitating the formation of c-Myc/BCAN-AS1/SNIP1 ternary complex. These results have extended our knowledge about the functional mechanism of SNIP1 in the regulation of c-Myc and indicated that m⁶A modification may act as a bridge to recruit SNIP1 binding to BCAN-AS1 transcript. However, this acting mechanism was somewhat different from a recent study which suggests that m⁶A in lncRNA Pvt1 enhances its direct interaction with c-Myc and stabilizes c-Myc protein in epidermal progenitor cells [79]. We believe this difference may be due to different types of c-Myc interacted lncRNAs and different disease models used in our studies. Given that research of the regulatory mechanism of m⁶A modification is still in its infancy, additional discoveries of regulatory patterns mediated by m⁶A on the functions of lncRNA are worth researching in the future.

Accumulating evidence has suggested c-Myc as a promising and attractive therapeutic target for cancer treatment, including PDAC [12, 21], but directly targeting c-Myc is challenging [23]. Therefore, from a therapeutic perspective, we now propose that altered expressions of c-Myc modulators like BCAN-AS1 might provide treatment strategies for indirectly targeting c-Myc. Indeed, we observed that ASO-BCAN-AS1, which inhibits BCAN-AS1, significantly reduced the growth and tumor burden on PDX models of PDAC in mice. Moreover, ASO-BCAN-AS1 treatment significantly reduced c-Myc protein levels in xenografts. These results may suggest that inhibiting BCAN-AS1 may be a potential option to indirectly target c-Myc for PDAC therapy. However, although inhibiting BCAN-AS1 showed no effect on mice bodyweight, further investigations are needed to systematically evaluate the potential toxicity caused by BCAN-AS1 inhibition.

In summary, we have identified BCAN-AS1 as a c-Myc-interacted oncogenic lncRNA in PDAC. M⁶A modification of BCAN-AS1 bridges the c-Myc/BCAN-AS1/SNIP1 ternary complex to stabilize c-Myc protein against SKP2-mediated ubiquitination and degradation. PDAC cell growth and invasiveness in vitro and in vivo can be suppressed by inhibiting BCAN-AS1. These findings shed light on the pro-tumorigenic role of BCAN-AS1 and provide novel insights into c-Myc-interacted lncRNA in PDAC.

Author contributions

J. Zheng and J. Zhang conceptualized and supervised this study. G.W., J.S. and L. Zeng performed most experiments. S.D., Y.Y. and R.L. conducted statistical and bioinformatics analyses. J. Zhang and X.H. performed animal experiments. Q.Z., Y.Z., J.D., S.Z. and R.C. contributed to collection of clinic samples. M.L. contributed to histopathological analyses. R.B. and L. Zhuang provided technique supports. G.W. J.S., J. Zheng and D.L. prepared manuscript. All authors reviewed the manuscript.

Data availability

C-Myc RIP-seq, RNA-seq and SNIP1 CLIP-seq raw data generated in this study are publicly available in the Gene Expression Omnibus at GSE181777. miCLIP sequencing data have been deposited in the NCBI Short Read Archive with the BioProject ID-PRJNA693621. All custom code used to generate the data in this study is available upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval

This study was approved by the Institutional Review Board of the Sun Yat-sen Memorial Hospital and Chinese Academy of Medical Sciences Cancer Hospital. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Sun Yat-sen University Cancer Center and the animals were handled in accordance with institutional guidelines.

Informed consent

Informed consent was obtained from each patient.

Supplementary information

The online version contains supplementary material available at 10.1038/s41418-023-01225-x.