SH3BP5-AS1 was upregulated in GEM-treated PDX pancreatic cancer
To explore the mechanism of GEM resistance in PC, we first established a PDX model, which was treated with either saline or GEM (Fig. 1A). We performed genomic sequencing of p4-pdx treated with GEM or the negative control to screen for hub genes. The top 10 differentially expressed lncRNAs are shown in Fig. 1B. The role of SH3BP5-AS1, one of the most strongly upregulated lncRNAs, in PC tumorigenesis and chemosensitivity was further investigated. The TCGA database suggested that SH3BP5-AS1 is closely related to the tumor grade of PC (Fig. 1C). Analysis of the correlation between SH3BP5 expression and clinicopathological data revealed that high SH3BP5-AS1 expression was closely related to tumor TNM stage, tumor size, and degree of differentiation in our center (Table 1). SH3BP5-AS1 expression was measured in 87 pair-matched PC tumor and adjacent normal tissues by RT-qPCR, and the results suggested that SH3BP5-AS1 was significantly upregulated in tumor tissue compared to normal tissue (Fig. 1D). ROC curve analysis revealed that it was an independent risk factor for the prognosis of PC (Fig. 1E). The survival of patients with low SH3BP5-AS1 expression was significantly better than that of patients with high SH3BP5-AS1 expression (Fig. 1F). At the same time, the expression levels of SH3BP5-AS1 in five PC cell lines and normal pancreatic cells were detected in vitro. SH3BP5-AS1 expression was significantly higher in PC cells than in normal pancreatic cells (Fig. 1G). Collectively, the above results demonstrated that SH3BP5-AS1 plays a key role in PC tumorigenicity and might serve as a key regulator of GEM resistance.
Table 1
Relationship between SH3BP5-AS1 expression and clinicopathological parameters in 87 PC patients.
Variables | All cases | SH3BP5-AS1 expression | P |
Low (n = 31) | High (n = 56) |
Age (years) | | | | |
<50 | 40 | 14 | 26 | 0.910 |
≥50 | 47 | 17 | 30 | |
Gender | | | | |
Male | 59 | 22 | 37 | 0.640 |
Female | 28 | 9 | 19 | |
Alcohol history | | | | |
No | 41 | 14 | 27 | 0.785 |
Yes | 46 | 17 | 29 | |
Smoking | | | | |
No | 37 | 12 | 25 | 0.592 |
Yes | 50 | 19 | 31 | |
CEA | | | | |
normal | 17 | 9 | 8 | 0.097 |
High | 70 | 22 | 48 | |
CA199 | | | | |
normal | 26 | 12 | 14 | 0.181 |
High | 61 | 19 | 42 | |
Tumor size (cm) | | | | |
<4 | 52 | 23 | 29 | 0.041* |
≥4 | 35 | 8 | 27 | |
TNM stage | | | | |
I-II | 35 | 16 | 19 | 0.02 |
III-IV | 45 | 32 | 13 | |
Lymph node invasion | | | | |
Absent | 36 | 17 | 19 | 0.035* |
Present | 44 | 31 | 13 | |
Metastasis | | | | |
No | 63 | 34 | 29 | 0.034 |
Yes | 17 | 14 | 3 | |
*P < 0.05. |
Abnormal expression of SH3BP5-AS1 was involved in the invasion, migration, and stemness of pancreatic cancer cells
To further explore the effects of SH3BP5-AS1 on the biological behavior of PC cells, PC cells (BxPC3, MIA Paca-2, and PANC-1) were transfected with lentiviral vectors encoding SH3BP5-AS1 inserts or short hairpin RNA (shRNA) and Transwell and colony formation assays were conducted. After silencing SH3BP5-AS1, the invasion, migration, and colony formation abilities of PC cells (BxPC3, MIA Paca-2) were significantly reduced (Fig. 2A–D and F–I). On the contrary, overexpression of SH3BP5-AS1 enhanced the above cellular behaviors (Fig. S2A–E).
Epithelial-to-mesenchymal transition (EMT) has been proved to be closely related to cell invasion and migration[11]. Therefore, we also detected EMT-related proteins. The results showed that the epithelial markers E-cadherin and α-catenin were upregulated, whereas the levels of the mesenchymal markers N-cadherin, fibronectin, and vimentin were decreased when SH3BP5-AS1 was silenced (Fig. 2E). The opposite results were obtained when SH3BP5-AS1 was overexpressed in PC cells (Fig. S3C). Stem cell-like cell markers Lin28, OCT4, NANOG, and SOX-2 have been found to be related to cell stemness [12], and these proteins were significantly downregulated when SH3BP5-AS1 was silenced and upregulated when SH3BP5-AS1 was overexpressed (Fig. 2J and K, Fig. S3F and G). These findings suggested that SH3BP5-AS1 stimulates PC cell migration, invasiveness, and stemness.
SH3BP5-AS1 promotes pancreatic cancer cell resistance to gemcitabine in vitro and in vivo
These above results strongly supported the notion that SH3BP5-AS1 is involved in PC carcinogenicity; however, whether SH3BP5-AS1 affects GEM chemoresistance in PC still remained to be elucidated. First, lentiviral vectors encoding SH3BP5-AS1 inserts or shRNA were transfected into BxPC3 and MIA Paca-2 cells. Then the cells were treated with GEM or DMSO for 72 h. We found that the PC cell proliferation ability was significantly reduced when SH3BP5-AS1 was silenced compared to the negative control group (Fig. 3A and B, Fig. S3A and B). Moreover, SH3BP5-AS1 silencing enhanced the inhibitory effect of GEM on PC cells (Fig. S3E and F). To corroborate the above results, cell viability was tested after exposing the cells to different GEM concentrations for 48 h. The results showed that silencing of SH3BP5-AS1 enhanced the sensitivity of PC cells to GEM. Compared with the control group, the IC50 value of GEM was significantly decreased in the SH3BP5-AS1 silencing group (Fig. 3C, Fig. S3C). On the contrary, after overexpression of SH3BP5-AS1, the sensitivity of the PC cell line PANC-1 to GEM was significantly reduced (Fig. S3E–H). Furthermore, we found that silencing of SH3BP5-AS1 reduced the viability of GEM-treated PC cells in a time-dependent manner (Fig. 3D, Fig. S3D).
We further explored the effects of SH3BP5-AS1 on the sensitivity of PC to GEM through in vivo experiments. Stably silenced SH3BP5-AS1 PC cell lines and control cell lines were injected subcutaneously and via the tail vein, and the mice received 50 mg/kg GEM or saline. After 4 weeks, a significantly lower tumor volume and weight were observed in the SH3BP5-AS1 silencing group, after treatment with either GEM or saline, especially in the GEM treatment group (Fig. 3E–G). The Ki67 immunostaining results suggested that silencing of SH3BP5-AS1 inhibits proliferation and enhances GEM sensitivity of PC cells (Fig. 3H). In the lung metastasis model, we found that silencing of SH3BP5-AS1 significantly reduced the distant metastasis of the tumor (7/16) compared with the control group (13/16). Interestingly, compared with the GEM treatment group, silencing of SH3BP5-AS1 can significantly improve the efficacy of GEM, which reduced the lung metastasis rate and the number of lung metastases (Fig. 3I and J). The above results suggested that SH3BP5-AS1 silencing enhanced the sensitivity of PC cells to GEM both in vitro and in vivo.
ALKBH5/IGF2BP1-mediated m6A modification contributed to SH3BP5-AS1 overexpression
m6A modification has been confirmed to be involved in the regulation of RNA expression[13]. m6A modifications are placed and removed by methylases and demethylases, respectively. ALKBH5, a demethylase, is well known to be involved in the occurrence and development of various cancers [14-16]. In this study, we found that ALKBH5 was significantly downregulated in tumor tissues compared to the adjacent tissues (Fig. 4A), and its expression was negatively correlated with SH3BP5-AS1 levels in tumor tissues (Fig. 4B). m6A RIP results showed that m6A levels in PC cell lines (BxPC-3 and PANC-1) were significantly higher than in normal pancreatic cells (HPDE6-C) (Fig. 4C). The in vitro results also revealed that ALKBH5 was significantly downregulated while SH3BP5-AS1 was overexpressed; on the contrary, ALKBH5 upregulation was observed after SH3BP5-AS1 was silenced (Fig. 4D and E). The dual luciferase report assay indicated that ALKBH5 could attenuate luciferase activity in the wild-type SH3BP5-AS1 group; however, luciferase activity was not affected in the mutant SH3BP5-AS1 group (Fig. 4F). MeRIP assays showed that ALKBH5 silencing increased the m6A modification level of SH3BP5-AS1 (Fig. 4G), whereas overexpression of ALKBH5 decreased the m6A modification level of SH3BP5-AS1 in PC cells (Fig. 4H). The actinomycin D assay (a drug that blocks de novo RNA synthesis) showed that ALKBH5 promoted SH3BPH5-AS1 degradation, while silencing of ALKBH5 contributed to the stability of SH3BP5-AS1 (Fig. 4I). Collectively, ALKBH5 acted as an important “eraser” in regulating SH3BP5-AS1. However, an “eraser” cannot regulate SH3BP5-AS1 expression alone; it also requires a “reader.” Studies have found that ALKBH5-mediated m6A demethylation inhibited mRNA transcription by the m6A effector IGF2BP1 [15, 17]. Our correlation analysis revealed that IGF2BP1 expression was positively correlated with SH3BP5-AS1 levels (Fig. 4J). We hypothesized that IGF2BP1 acted as a “reader” for the regulation of SH3BP5-AS1. To confirm the above hypothesis, the RIP assay was performed to affirm the direct binding between IGF2BP1 and SH3BP5-AS1 (Fig. 4K). SH3BP5-AS1 expression was significantly decreased after IGF2BP1 silencing and increased after IGF2BP1 overexpression (Fig. 4L and M). The results of our actinomycin D assay confirmed the above conclusion (Fig. 4N). Then an RIP assay was conducted to explore the correlation between the “eraser” ALKBH5 and the “reader” IGF2BP1. The results suggested that IGF2BP1 expression was negatively correlated with ALKBH5 levels (Fig. 4O). In conclusion, low expression of the “eraser” ALKBH5 maintains a high m6A modification level of SH3BP5-AS1, which further promotes the recognition of m6A-SH3BP5-AS1 by the “reader” IGF2BP1.
miR-139-5p is involved in the oncogenic roles of SH3BP5-AS1 in PC
It has been reported that lncRNAs exert carcinogenic effects through acting as ceRNAs [18-20]. Here, we explored whether this is also the case for SH3BP5-AS1. First, Combined with the public database ENCORI, previously known as starBase v2.0, and the DIANA Tool, miRNA-194-5p, miRNA-589-5p, miRNA-26a-5p and miRNA-139-5p were predicted to be a target miRNA of SH3BP5-AS1 (Fig. 5A). Furthermore, RNA pulldown assay was performed to prove miRNA-139-5p was the most enriched in the precipitate of SH3BP5-AS1 than other miRNA candidates (Fig. 5B). Therefore, miRNA-139-5p was selected for follow-up study. RT-PCR was performed to further prove the interaction between miRNA-139-5p and SH3BP5-AS1. The results suggested that SH3BP5-AS1 expression is negatively correlated with miRNA-139-5p levels (Fig. 5C). The results of our luciferase assays confirmed that in both H293T and PC cells, miR139-5p significantly decreased the luciferase activity of wild-type SH3BP5-AS1 but did not affect that of mutant SH3BP5-AS1 (Fig. 5D–F). Furthermore, silencing of SH3BP5-AS1 can significantly upregulate the expression of miR139-5p, whereas the opposite results were obtained after SH3BP5-AS1 overexpression (Fig. 5G and K). The above results indicated that SH3BP5-AS1 might sponge miR139-5p, thereby affecting downstream target genes. The relationship was verified by RIP assay. We found that both SH3BP5-AS1 and miR139-5p were enriched in AGO2 precipitate (Fig. 5H–J, L–N). The results of our RNA pull-down assays confirmed the above results. miR-139-5p and AGO2 were more strongly enriched in the biotin-labeled SH3BP5-AS1 group than in the control group (Fig. 5O–R). These results indicated that SH3BP5-AS1/miR-139-5P participate in the carcinogenic mechanism of PC.
To further explore the effect of SH3BP5-AS1/miRNA-139-5p on the biological behavior of PC and GEM resistance, first of all, Transwell and colony formation assays proved that silencing of SH3BP5-AS1 could reduce the invasion, migration, and spheroidizing abilities of PC cells, while miRNA-139-5p inhibition could reverse the effect of shSH3BP5-AS1 on PC cells. On the contrary, overexpression of miRNA-139-5p could reverse the carcinogenic effects of SH3BP5-AS1 (Fig. S4A–I). In the drug resistance intervention experiment, we found that miRNA-139-5p inhibition could reverse the effect of shSH3BP5-AS1 on GEM sensitivity of PC cells. Overexpression of miRNA-139-5p increased the sensitivity of PC cells to GEM. This further illustrates that SH3BP5-AS1 can affect the biological behavior and GEM resistance of PC cells through miRNA-139-5p (Fig. S4J–O).
miR-139-5p sponging activity decreases CTBP1 levels, affecting the biological behavior of PC cells
As shown above, miR-139-5p could inhibit the expression of downstream target genes by forming a RISC complex with AGO2. Transcriptome sequencing was used to further screen downstream target genes. As shown in Fig. 5A and B, we found that CTBP1 was significantly downregulated when miR-139-5P was overexpressed. We conducted functional enrichment and signal pathway analyses of differentially expressed genes and found that genes, including CTBP1, are mainly enriched in Wnt signaling (Fig. 5C and D). Based on the miR-TARGET database to predict miRNA target genes, it was found that there were a total of 218 genes that could be regulated by miR-139-5p. Among the overlapping genes, the difference in CTBP1 expression is particularly obvious (Fig. 5C–E), indicating that CTBP1 might be a hub gene downstream of miR-139-5p. To prove this hypothesis, first, correlation analysis was conducted, and it was found that the expression of miR-139-5p was negatively correlated with CTBP1 levels (Fig. 6A). Then a luciferase assay confirmed that miRNA-139-5p could inhibit the fluorescence activity of wild-type CTBP1 without affecting the fluorescence activity of mutant CTBP1, confirming the regulatory relationship between miRNA-139-5p and CTBP1 (Fig. 6B and C). Overexpression of miR-139-5p downregulated CTBP1 mRNA and protein expression levels, while silencing of miR-139-5p upregulated CTBP1 mRNA and protein expression levels (Fig. 6D–F). All these results indicated that miR-139-5p has a targeted regulatory relationship with CTBP1. In the drug resistance study, we co-transfected miR-139-5p mimics and CTBP1 overexpression plasmid into PC cells, which were subsequently treated with GEM. We found that CTBP1 could reverse the effects of miR-139-5p on PC cell invasion, migration, and GEM resistance (Fig. 7A–I). On the contrary, when miR-139-5p and CTBP1 were silenced at the same time, si-CTBP1 could significantly inhibit PC cell invasion and migration, stemness, and GEM resistance (Fig. 7J–O). Therefore, CTBP1 is an important molecule downstream of miR-139-5p, and miR-139-5p exerts tumorigenic effects by regulating CTBP1.
SH3BP5-AS1 localization contributes to the effects of CTBP1 and increases tumor malignancy via the Wnt/β-catenin pathway
SH3BP5-AS1 participates in the GEM resistance of PC cells through a ceRNA mechanism, as confirmed in the above described experiments. We detected the expression and location of SH3BP5-AS1 by RT-PCR and immunofluorescence, and found that SH3BP5-AS1 is expressed in both the nucleus and the cytoplasm (Fig. 6). We hypothesized that SH3BP5-AS1 might be transferred to the cytoplasm after regulation in the nucleus. When we further analyzed the regulatory relationship between SH3BP5-AS1 and CTBP1, we found that SH3BP5-AS1 silencing reduced cell invasion, migration, and stemness and enhanced GEM sensitivity in PC cells, whereas CTBP1 could reverse the abovementioned effects of sh-SH3BP5-AS1 (Fig. 8A and B, Fig. 7A and C). Conversely, silencing of CTBP1 could significantly inhibit the biological effects of SH3BP5-AS1 overexpression on PC cells (Fig. 7B and D). Similarly, CTBP1 restored the effects of sh-SH3BP5-AS1 on GEM sensitivity in PC cells, as observed in the drug sensitivity assay (Fig. 8D and E). Further correlation analysis revealed that the expression of SH3BP5-AS1 was negatively correlated with CTBP1 levels (Fig. 8F).
According to our transcriptome sequencing results, the Wnt pathway might be an important signaling pathway for SH3BP5-AS1/miRNA-139-5p/CTBP1 to exert its effect. We further detected Wnt pathway-related proteins through western blot and IHC and found that after SH3BP5-AS1 overexpression, the Wnt pathway was significantly activated (Fig. 8G and H). When CTBP1 was silenced at the same time, the Wnt signaling pathway was blocked. These results indicated that SH3BP5-AS1/miRNA-139-5p/CTBP1 might exert their biological functions via the Wnt signaling pathway.