GAS6-AS1 is closely associated with 5-FU resistance and might serve as a sensitive indicator for predicting patient prognosis in CRC
To investigate the key regulators in chemotherapy resistance of CRC, RNA-seq analysis (TRG 0–1 vs TRG 3) was performed to reveal the differentially expressed lncRNAs (Fig. 1a) and mRNAs (Fig. 1b). Then, we performed WGCNA to explore key regulators closely associated with TRG stage (Fig. 1c, Supplementary Fig. S1a). 18 modules were identified to be correlated with TRG trait, of which the MEblue module is the most closely associated module (Fig. 1d, Supplementary Fig. S1b). To screen the hub genes in the MEblue module, we drew the profile of module membership vs gene significance, and highlighted the most significant part of hub genes (Fig. 1e). Then, we constructed comprehensive network analysis with these hub genes (Fig. 1f). We focused on the four lncRNAs (Fig. 1g), and found that both GAS6-AS1 and CATIP-AS1 were highly expressed in CRC tissues from TCGA database (Fig. 1h, Supplementary Fig. S1c and e). In addition, patients with high GAS6-AS1 showed poor overall survival though there was no significant statistical difference (Supplementary Fig. S1d), while CATIP-AS1 showed opposite effect (Supplementary Fig. S1f). We didn’t find another two lncRNAs in TCGA database. Thereby, we focused on GAS6-AS1 in following study. Then, we retrieved the CRC data from GEO database. GAS6-AS1 was highly expressed in CRC tissues (GSE39582) and positively correlated with poor disease-free survival (GSE106584) (Supplementary Fig. S1g-h).
GAS6-AS1 is located in 13q34 with the length of 902 bp (Supplementary Fig. S2a). Along with the thorough research, it has been discovered that a small number of lncRNAs can be translated into peptides. Therefore, we next analyzed the protein-coding potential of GAS6-AS1with four independent mathematical methods, including Coding Potential Assessment Tool[21], Coding Potential Calculator[22], length and guanine-cytosine (LGC) algorithms[23] and ORF finder software from NCBI (Supplementary Fig. S2b-e). All these methods showed that GAS6-AS1 didn’t have protein-coding potential. The Ensemble database indicated that GAS6-AS1 shared no homology with other genomic regions. The secondary structure of GAS6-AS1 was listed in Supplementary Fig. S2f.
Next, the clinical significance of GAS6-AS1 were analyzed. GAS6-AS1 level was up-regulated in 55.38% (175/316) CRC tissues (Fig. 1i), and positively correlated with differentiation grade, T stage, N stage, M stage and TNM stage (Fig. 1j-k, Supplementary Fig. S3a-g, Supplementary Table S1). To analyze the correlations of GAS6-AS1 expression with objective response rate (ORR) of 5-FU based chemotherapy regimen, the chemotherapy response was evaluated in 138 cases with distant metastasis of 316 patients. The ORR reached 29.84% and the level of GAS6-AS1 was higher in cases with stable disease (SD) and progressive disease (PD) than these with partial response (PR) (Fig. 1l). Kaplan-Meier analysis showed that the high level of GAS6-AS1 correlated with poorer overall survival and disease-free survival in all patients (Fig. 1m-n). Then, patients were stratified based on clinicalpathological characteristics. Patients with high level of GAS6-AS1 in stage III + IV, T4, T1-3, M1 and M0 all showed poorer overall survival and disease-free survival (Supplementary Fig. S3h-m). More importantly, the Hazard Ratio (HR) value in patients with stage III + IV, T4 and M1 were higher than these with stage I + II, T1-3 and M0, respectively. Univariate and multivariate regression analysis indicated that GAS6-AS1 level is an independent prognostic factor for CRC (Supplementary Table S2). Also, receiver operating characteristic curve (ROC) analysis confirmed that GAS6-AS1 level was a sensitive indicator for predicting patient prognosis, especially for 5 years overall survival (Fig. 1o).
GAS6-AS1 promotes 5-FU resistance and malignant behaviors in CRC cells in vitro
To further elucidate the roles of GAS6-AS1 in 5-FU resistance, we firstly constructed CRC cell lines with 5-FU resistance. The results showed that IC50 value was significantly higher in HCT15-Re (Fold change = 5.26) and HCT8-Re (Fold change = 3.74) cells compared with HCT15 and HCT8 cells (Fig. 2a-b). Secondly, RT-PCR results indicated that the expression of GAS6-AS1 was significantly higher in both HCT15-Re and HCT8-Re cells than that in HCT15 and HCT8 cells (Fig. 2c). Thirdly, we generated stable overexpression cell lines of GAS6-AS1 with HCT15 and HCT8 cells, and stable knockdown cell lines of GAS6-AS1 with HCT15-Re and HCT8-Re cells (Supplementary Fig. S4a-b). Then, the effects of GAS6-AS1 on 5-FU resistance was explored. GAS6-AS1 significantly increased the IC50 of 5-FU, while knockdown of GAS6-AS1 decreased it, in CRC cells (Fig. 2d-g). Next, CCK8, colony formation, edu and flow cytometry for cell cycle assays confirmed that overexpression of GAS6-AS1 promoted cell growth and cell cycle G1/S transition with or without 5-FU, while knockdown of GAS6-AS1 exerted the opposite effects, in HCT15 and HCT8 cells (Fig. 2h-o, Supplementary Fig. S4c-j). Collectively, these data implied that GAS6-AS1 might enhanced 5-FU resistance and cell growth in CRC cells through promoting cell cycle G1/S progression.
GAS6-AS1 promotes 5-FU resistance in vivo
Furthermore, the effects of GAS6-AS1 on 5-FU resistance were observed in xenograft mouse models. Overexpression of GAS6-AS1 increased the volume and weight of tumors under the treatment of 5-FU, while knockdown of GAS6-AS1 decreased those under the treatment of 5-FU, in HCT15 and HCT8 cells (Fig. 3a-l). Then, we detected the expression of MCM3, KI67 (biomarkers of cell proliferation), cyclin D1 and CDK4 (biomarkers of cell cycle) in continuous tumor tissue slice. Overexpression of GAS6-AS1 increased the expression of these genes under the treatment of 5-FU, while knockdown of GAS6-AS1 decreased those under the treatment of 5-FU, in HCT15 and HCT8 cells (Fig. 3m). Consequently, GAS6-AS1 promotes 5-FU resistance of CRC cells in vivo.
GAS6-AS1 physically interacts with PCBP1 in CRC cells
Studies have revealed that the molecular mechanism of lncRNAs diversifies according to its subcellular distribution. Both FISH and nuclear cytoplasmic separation assays indicated that most of GAS6-AS1 located in cytoplasm (Fig. 4a-b). In addition, most of the roles of lncRNAs require interactions with one or more RNA-binding proteins (RBPs). Next, we dedicated to seeking for the RBPs of GAS6-AS1. RNA pulldown and mass spectrometry were performed in HCT15 and HCT8 cells (Fig. 4c). With the criterions of molecular weight between 20-55kDa, more than 3 peptides and more than 3 unique peptides, five proteins existed between the intersections of the results of HCT15 and HCT8 (Fig. 4d). After checking the features of those proteins, hnRNPK and PCBP1 were selected for further research. Only PCBP1 protein was pulled down by GAS6-AS1 sense probe (Fig. 4e). Moreover, RIP assays confirmed the association between GAS6-AS1 with PCBP1 (Fig. 4f). As expected, there was no enrichment of MALAT1 in each group. According to the binding motif of PCBP1 (Fig. 4g), three potential binding sites in GAS6-AS1 was found. Threes truncated RNA separately containing three binding sites and full length RNA with biotin labeling were constructed according to the secondary structure of GAS6-AS1 (Fig. 4h-i). RNA pulldown assays confirmed that the binding site 2 in GAS6-AS1 was required for the interaction with PCBP1, not the binding site 1 and 3 (Fig. 4j). Furthermore, three Flag-tagged truncated PCBP1 and full length were generated based on the functional domains (Fig. 4k). RIP assays showed that deletion mutation of KH3 abolished its interaction with GAS6-AS1 (Fig. 4l). In a word, we demonstrated that GAS6-AS1 physically interacted with PCBP1 in CRC cells through binding site 2 and KH3.
PCBP1 is required for GAS6-AS1 in the role of 5-FU resistance
In order to explore the role of PCBP1 in the regulation of GAS6-AS1 on 5-FU resistance, we generated overexpression plasmid and threes shRNAs targeted PCBP1 (Fig. 5a). CCK8 results showed that knockdown of PCBP1 rescued the enhance of GAS6-AS1 on the IC50 of 5-FU, while overexpression of PCBP1 increased the IC50 of 5-FU decreased by downregulation of GAS6-AS1 (Fig. 5b-c). Furthermore, CCK8, colony formation, edu and flow cytometry for cell cycle assays elucidated that PCBP1 could rescue the regulation of GAS6-AS1 on the cell growth and cell cycle G1/S transition under the treatment of 5-FU (Fig. 5d-g). In brief, PCBP1 is required for GAS6-AS1 in the promotion of 5-FU resistance, cell growth and cell cycle G1/S progression in CRC.
MCM3 is a functional downstream mediator for GAS6-AS1
To search for the functional downstream of GAS6-AS1, RNA-seq on HCT15 cells with overexpression of GAS6-AS1 was performed (Fig. 6a). In addition, two GEO datasets were selected: GSE16236 (RNA-seq on HCT116 cells with knockdown of PCBP1) and GSE131210 (CLIP-seq on HCT116 cells with anti-PCBP1) (Fig. 6a). There were 17 genes in the intersect of the three datasets (Fig. 6b). GO analysis indicated that those genes mainly enriched in the biological processes related with DNA replication (Fig. 6c). Furthermore, the 22 RNA sequencing samples mentioned above were classified into high GAS6-AS1group and low GAS6-AS1 group according to the mean of GAS6-AS1 level. GSEA analysis showed that high GAS6-AS1 participated in DNA replication, DNA repair, DNA damage repair signal transduction, DNA biosynthetic process and DNA binding (Supplementary Fig. S5). MCM3, MCM4, PARD6B, BRIP1 and RAD54B were enriched in the above biological process. RT-PCR results elucidated that only MCM3 mRNA level upregulated after overexpression of GAS6-AS1 or PCBP1, and downregulated after knockdown of GAS6-AS1 or PCBP1 (Fig. 6d, f). West blot also confirmed the results (Fig. 6e, g). Meanwhile, both GAS6-AS1 and PCBP1 promoted the expression of cyclin D1 and CDK4 (Fig. 6e, g).
To explore whether MCM3 could act as the downstream mediator of GAS6-AS1, rescue experiments were employed. Overexpression plasmid and threes shRNAs targeted MCM3 were constructed (Supplementary Fig. S6a). CCK8, colony formation, edu and flow cytometry for cell cycle assays confirmed that MCM3 could rescue the regulation of GAS6-AS1 on the 5-FU resistance, cell growth and cell cycle G1/S progression (Supplementary Fig. S6b-g). In summary, MCM3 is required for GAS6-AS1 in the role of 5-FU resistance, cell growth and cell cycle G1/S progression in CRC.
GAS6-AS1 enhances the stability of MCM3 mRNA through increasing its binding to PCBP1
To analyze the interaction between PCBP1 and MCM3, RNA pulldown and RIP assay were performed. PCBP1 protein was pulled down by MCM3 mRNA sense probe but not antisense (Fig. 7a). Moreover, RIP assays confirmed the binding between MCM3 mRNA with PCBP1 (Fig. 7b-c). According to the binding motif of PCBP1, three potential binding sites in the 3’-UTR of MCM3 mRNA was marked. Threes truncated RNA and full length RNA with biotin labeling were constructed (Fig. 7d). RNA pulldown assays confirmed that the binding site 3 in the 3’-UTR of MCM3 mRNA was required for the interaction with PCBP1, not the binding site 1 and 2 (Fig. 7e). Furthermore, RIP assays with the above three Flag-tagged truncated PCBP1 and full length indicated that deletion mutation of KH1 failed to bind to the 3’-UTR of MCM3 mRNA (Fig. 7f).
As a member of heterogeneous nuclear ribonucleoproteins (hnRNPs), the main roles of PCBP1 include regulation of transcription, control of mRNA stability, translation regulation and alternative splicing. Given that GAS6-AS1 located in cytoplasm and regulated the expression of mRNA and protein of MCM3, we speculated that PCPB1 might regulated the stability of MCM3 mRNA through binding to the 3’-UTR of mRNA. HCT15 cells were treated with Actinomycin D to block de novo transcription. RT-PCR results revealed that knockdown of GAS6-AS1 or PCBP1 both decreased the stability of MCM3 mRNA (Fig. 7g-h). Moreover, PCBP1 rescued the regulation of GAS6-AS1 on the stability of MCM3 mRNA (Fig. 7i).
To dissect the role of GAS6-AS1 on the binding of PCBP1 on MCM3, further RIP assays were conducted. The enrichment of MCM3 mRNA by PCBP1 was remarkably increased by overexpression of GAS6-AS1, and decreased by knockdown of GAS6-AS1 (Fig. 7b-c), suggesting that GAS6-AS1 promoted the interaction between MCM3 mRNA and PCBP1. Meanwhile, we generated luciferase plasmids containing wild type of 3’-UTR of MCM3 mRNA or mutant type in binding stie 3 of 3’-UTR of MCM3 mRNA. GAS6-AS1 significantly enhanced the luciferase activity of wild type, while knockdown of GAS6-AS1 decreased the luciferase activity of wild type, but not the mutant type (Fig. 7j). Interestingly, knockdown of PCBP1 decreased the luciferase activity of wild type enhanced by overexpression of GAS6-AS1, while overexpression of PCBP1 increased it which was reduced by knockdown of GAS6-AS1 (Fig. 7j). Furthermore, we confirmed that the promotion of MCM3 expression by GAS6-AS1 is dependent on PCBP1 by western blotting (Fig. 7k). We observed that knockdown of PCBP1 decreased MCM3 expression enhanced by overexpression of GAS6-AS1, while overexpression of PCBP1 increased it downregulated by knockdown of GAS6-AS1. To sum up, GAS6-AS1 can serve as a ‘guider’ to promote the interaction of PCBP1 and MCM3 mRNA 3’-UTR, thereby enhancing the stability and expression of MCM3 mRNA.
GAS6-AS1 might serve as a powerful biomarker and potential therapeutic target for combination drug therapy in CRC
To further dissect the clinical significance of GAS6-AS1 in CRC, we detected its expression in our TMA with RNA-FISH. GAS6-AS1 level was up-regulated in 65.82% (208/316) CRC tissues (Fig. 8a, d, Supplementary Fig. S7a). The level of GAS6-AS1 was positively correlated with differentiation grade, T stage, N stage, M stage and TNM stage (Fig. 8b, c, e, f, Supplementary Fig. S7b-h, Supplementary Table S3). Kaplan-Meier analysis showed that patients with high level of GAS6-AS1 showed poorer overall survival and disease-free survival in all patients (Fig. 8g-h). Then, patients were stratified based on clinicalpathological characteristics. The results showed that HR in patients of different stages was different (Supplementary Fig. S7i-p). In addition, univariate and multivariate regression analysis indicated that GAS6-AS1 level is an independent prognostic factor for CRC (Supplementary Table S4). Also, ROC analysis confirmed that GAS6-AS1 level was a specific indicator for predicting patient prognosis (Fig. 8i).
To test the practicality of GAS6-AS1 as combination treatment in CRC, ASO targeted GAS6-AS1 (Supplementary Fig. S8) and PDX model (Fig. 8j) were constructed. Clinical characterization of the donator patients is listed (Fig. 8k). ASO of GAS6-AS1 combined with 5-FU decreased the volume and weight of tumor from two patients (Fig. 8l-o). IHC results also indicated that ASO of GAS6-AS1 combined with 5-FU significantly decreased the expression of MCM3, KI67, cyclin D1 and CDK4, compared with that with only 5-FU (Fig. 8p). Taken together, GAS6-AS1 might serve as a powerful biomarker and potential therapeutic target for combination drug therapy in CRC.