Study overview
As shown in Figure 1A, AS-SNPs with annotation information of “splice-3” or “splice-5” were screened from dbSNP database, and previous Illumina chips were used for the genotyping of AS-SNPs in bladder cancer cases and healthy controls. Besides, the effect of AS-SNPs on bladder cancer risk were calculated to identify risk SNPs, and the association was further validated in another independent population. Finally, molecular experiments were conducted to explore the biological function of risk SNPs involved in bladder cancer.
Identification of bladder cancer risk associated AS-SNPs
A total of 10,810 AS-SNPs were extracted from dbSNP database (Figure S1). After genotyping using Illumina chips, 1,518 AS-SNPs were remained (Figure S2). We conducted strict quality control for SNPs including a call rate > 95%, minor allele frequency (MAF) > 0.05, and P > 0.05 for Hardy-Weinberg Equilibrium (HWE). Eventually, a total of 580 bladder cancer cases and 1,101 controls with 206 AS-SNPs were retained for further analysis (Table S1). Among them, there were 9 AS-SNPs passing a P value threshold of 0.05 (Figure 1B and Table S2). According to bioinformatics tools, we further predicted and scored the function of AS-SNPs, which showed that rs558814 with the highest functional score and strongest effect was served as a candidate SNP for further investigation [odds ratio (OR) = 0.78, 95% confidence interval (CI) = 0.67-0.91, P = 1.91×10-3] (Figure 1C and Figure 1D).
SNP rs558814 was subsequently genotyped in validation study including 1,050 bladder cancer cases and 1,403 controls, and characteristics of subjects have been previous reported(22) (Table S3). The results showed that the SNP rs558814 A>G was associated with the decreased bladder cancer risk with P value of 0.033 in the additive model (OR = 0.88, 95% CI = 0.78-0.99) and 0.037 in the dominant model (OR = 0.84, 95% CI = 0.72-0.99; Table S4). The stratified analysis results showed that the protective effect was also observed in subgroup of age > 65 (OR = 0.78, 95% CI = 0.62-0.97, P = 0.027), males (OR = 0.77, 95% CI = 0.65-0.93, P = 0.005), and muscle invasive bladder cancer (OR = 0.73, 95% CI = 0.58-0.92, P = 0.008) (Table S5-6 and Figure S3). The combined analysis showed that SNP rs558814 displayed protective effect of the G allele on bladder cancer risk (OR = 0.84, 95% CI = 0.76-0.92, P = 3.26×10-4) with no heterogeneity (I2 = 21.5%, Phet = 0.259) (Table 1).
Table 1
Association of rs558814 with bladder cancer risk
Stages
|
Cases
/Controls
|
Genotypes (AA/AG/GG)
|
MAF
(Cases/Controls)
|
OR (95%CI)
|
P a
|
Cases
|
Controls
|
Discovery
|
580/1,101
|
292/202/52
|
476/451/135
|
0.28/0.34
|
0.78(0.67-0.91)
|
1.91×10-3
|
Validation
|
1,050/1,403
|
502/436/102
|
617/625/159
|
0.31/0.34
|
0.88(0.78-0.99)
|
3.29×10-2
|
Combined
|
1,630/2,504
|
794/638/154
|
1093/1076/294
|
0.30/0.33
|
0.84(0.76-0.92)
|
3.26×10-4
|
SNP, single nucleotide polymorphism; MAF, minor allele frequency; OR, odds ratio; CI, confidence interval.
aP values were calculated from logistic regression analysis adjusted for age and sex.
|
Effect of rs558814 on mediating lncRNA BCLET expression in bladder cancer
According to the Ensembl (GRCh37 version), rs558814 is located on the intron of the ENSG00000245498 (RP11-677M14.7), which is annotated as MSANTD2 (Myb/SANT-like DNA-binding domain containing 2) antisense RNA and generally categorized as an ncRNA. For transcripts, the coding probability prediction revealed extremely low coding predicted values from Coding Potential Assessing Tool (CPAT) (probabilities of 0.0073 for ENST00000499143 and 0.0026 for and ENST00000529392; Figure 2A). The public and in-house dataset showed that the expression of RP11-677M14.7 was significantly lower in bladder cancer tissues than in normal tissues (Figure 2B-C), so we named this novel long ncRNA (lncRNA) as BCLET (Bladder Cancer Low-Expressed Transcript), and there were two transcripts as BCLET-long and BCLET-short, respectively. According to The Genotype-Tissue Expression (GTEx) database, the expression level of BCLET-long was more abundant than that of BCLET-short in bladder tissue (Figure 2D). In addition, the differential expression pattern of BCLET was also found in bladder cell lines (Figure 2E). As detected by quantitative analyses, the distribution of BCLET in the nucleus accounted for more than 90% (Figure 2F).
We further explored the expression quantitative trait loci (eQTL) effect of the SNP rs558814 to BCLET transcripts expression in bladder cancer. We found that with the increase of rs558814 G allele, the expression of BCLET increased significantly in The Cancer Genome Atlas (TCGA) bladder cancer tissues and in-house bladder adjacent tissues (Figure 2G). This effect was also detected in the two transcripts of BCLET, but there was no significant statistical difference in the BCLET-long transcript (Figure 2H-II-J). In addition, the luciferase reporter experiment further revealed that compared with the A risk allele, the G allele showed a significant increase in transcriptional activity (Figure 2K). These results suggested that rs558814 may participate in bladder cancer by affecting the expression of BCLET through transcriptional activity regulation mechanism (Figure 2L).
Biological function of lncRNA BCLET in bladder cancer cellular phenotypes
We further explored the biological effect of BCLET transcripts on bladder cancer cell phenotypes. Overexpression of BCLET-long or BCLET-short significantly suppressed the abilities of proliferation, clone formation, invasion, and migration of bladder cancer cells (Figure S4 and Figure 3A-D). Besides, BCLET transcripts overexpression remarkably increased the proportion of cell apoptosis (Figure 3E). These results indicated the potential of BCLET to serve as tumor suppressor gene involved in bladder cancer risk.
Effect of rs558814 on regulating AS of MSANTD2 in bladder cancer
We further explored the potential mechanism of rs558814 involved in AS events. According to the CancerSplicingQTL database, rs558814 had splicing quantitative trait loci (sQTL) effect and was significantly related to the splicing of exon 1 and exon 2 for MSANTD2, which is located at the 5'end of BCLET and has 6 transcripts (Figure 4A-B). According to GTEx database, among MSANTD2 transcripts, MSANTD2-004 was found to have the highest expression level and lacked exon 1 (Figure 4C). Considering that rs558814 may have a splicing regulation effect on the first exon of MSANTD2, and the expression level of other transcripts except MSANTD2-004 was relatively low, we further focused on the splicing regulation of rs558814 on MSANTD2-004. The eQTL analysis indicated that with the increase of the G allele, the expression of MSANTD2-004 in bladder adjacent tissues had a increasing trend with no statistical difference (Figure 4D). The ratio of the expression of MSANTD2-004 to the total expression of MSANTD2 was used to assess the splicing ratio of MSANTD2-004 transcripts. We found that compare with the A allele, the G allele of SNP rs558814showed a higher splicing ratio for MSANTD2-004 (P = 0.004; Figure 4E). These results indicated that rs558814 may be involved in the splicing regulation of MSANTD2 in bladder cancer.
LncRNA BCLET mediating AS of MSANTD2 in bladder cancer
Studies have shown that antisense RNA located in the nucleus can bind to its corresponding sense RNA to participate in the splicing regulation of sense RNA. Therefore, we speculated whether rs558814 mediated the splicing of MSANTD2 through influencing the expression of lncRNA BCLET. Co-expression analysis showed that BCLET and MSANTD2-004 showed a significant positive correlation (r = 0.828 for bladder cancer tissues and 0.783 for bladder normal tissues, respectively, P < 0.0001; Figure 4F). Besides, the novel positive correlation was also found between the expression of BCLET and the splicing ratio of MSANTD2-004 (r =0.849 for bladder cancer tissues and 0.761 for bladder normal tissues, respectively, P < 0.0001; Figure 4G).
We further verified the splicing regulation effect of BCLET on MSANTD2-004 in bladder cancer cells (Figure 4H-I). The results showed that the overexpression of BCLET-long or BCLET-short significantly increased the expression level and splicing ratio of MSANTD2-004, and low expression of MSANTD2-004 and splicing ratio can be observed with decreasing the expression of BCLET. We proposed a speculation that the BCLET would bind the first exon of MSANTD2 mRNA to regulate the splicing of MSANTD2 mRNA. RIP experiments were conducted to investigate the relationship between BCLET and MSANTD2 mRNA (Figure 4J). Compared with the negative control, MSANTD2 enriched by GFP antibody significantly increased , suggesting a binding between lncRNA BCLET and MSANTD2 mRNA (Figure 4K). These results suggested that as the antisense RNA of MSANTD2, lncRNA BCLET may bind to MSANTD2 mRNA to promote the expression of MSANTD2-004 (Figure 4L).
Effect of MSANTD2 in bladder cancer cell malignant phenotypes
We further explored the biological function of the splicing isoform MSANTD2-004 in bladder cancer. After overexpression of MSANTD2-004, cell proliferation, plate cloning, invasion and migration were inhibited, and the proportion of apoptotic cells increased (Figure S5 and Figure 5A-E). In addition, the abnormally low expression of MSANTD2 was closely related to the poor prognosis of patients with bladder cancer (Figure S6). Taken together, the above results suggested that MSANTD2-004 may play a tumor suppressor gene like role in bladder cancer.
Effect of BCLET overexpression on tumor in vivo
To further investigate the role of BCLET overexpression in bladder cancer in vivo, T24 cells was stably transfected with NC/BCLET-long/BCLET-short lentiviral vector. As shown in Figure 6A-D, stable overexpression of BCLET-long/BCLET-short significantly suppressed the abilities of proliferation, clone formation, invasion, and migration in T24 cell. We injected T24 cells carrying NC/BCLET-long/BCLET-short lentiviral vector into nude mice. In line with in vitro cell phenotype, BCLET-long/BCLET-short overexpression dramatically decreased the mean tumor weight and average tumor volume (Figure 6E-H). As the presented in Figure 6I, the tumor tissues injected with BCLET-long/BCLET-short lentiviral vector showed significantly reduced expression of proliferation marker Ki67 and notably increased expression of MSANTD2.