RBM15 is overexpressed in LSCC and associated with poor prognosis
To investigate the roles of the m6A-associated genes in LSCC development, we first analyzed the methylation level in 34 LSCC and paired standard samples. We found that the m6A RNA modification in LSCC tissues was increased (Fig. 1a). The hierarchical clustering results showed that there were systematic differences in mRNA expression levels between the 3 pairs of laryngeal carcinoma and adjacent tissues. The screening conditions were restricted to the threshold of absolute fold change greater than 1 and p-value less than 0.05. The results showed that compared with adjacent non-tumor tissues, there were 841 differentially expressed mRNAs in LSCC (Fig. 1b). Protein Cluster Analysis was performed on 5 pairs of LSCC tissues and adjacent non-tumour tissues. The results showed that compared with adjacent non-tumor tissues, there were 1826 proteins differentially expressed in LSCC, of which 885 were up-regulated, and 941 were down-regulated (Additional file 2: Figure S1). Combining with the proteomics results of laryngeal cancer tissue, the RBM15 that were differentially expressed in m6A methyltransferase “writers” was screened out as the most significant (Fig. 1c). Our data revealed a significantly higher RBM15 mRNA level in the LSCC compared to the non-tumorous tissues (Fig. 1d). Moreover, the qRT-PCR results in 34 pairs of LSCC tissues suggested that the expression level of RBM15 was associated with the clinicopathological characteristics of LSCC. T3 + T4 tumours, III + IV tumours, and N1 tumours had higher expression levels of RBM15 (Fig. 1e). The expression of RBM15 was confirmed in 3 LSCC cell lines, namely TU-212, TU-177, AMC-HN-8, and normal human bronchial epithelial cells. The results suggested that in these measured laryngeal cancer cell lines, the expression level of RBM15 was more significant than in NHBEC cells (Fig. 1f). Analysis of TCGA survival data indicated that among HNSC patients, patients with low RBM15 expression had a longer median survival time (Fig. 1g). Furthermore, IHC assays were performed to determine the localization of RBM15 in LSCC tissue and its expression in 122 pairs of matched cancer tissues and adjacent non-cancerous tissues. As shown in Fig. 1i, RBM15 was mainly located in the nucleus in LSCC tissues, and the expression level in cancer tissues was significantly higher than that in adjacent non-tumor tissues. These results implied that RBM15 was overexpressed in LSCC tissues. Besides, the 60-month follow-up results showed that the overexpression of RBM15 was closely related to the poor prognosis of LSCC patients (Fig. 1h).
RBM15 promotes LSCC cells migration and invasion in vitro and in vivo
In order to explore the potential function of RBM15 in LSCC, we synthesized RBM15 specific shRNA to obstruct the expression of endogenous RBM15 in TU-212 and AMC-HN-8 cell lines. RBM15 expression was reduced after shRBM15 transfection. After transfection with overexpressing RBM15 lentivirus, the expression of RBM15 increased (Fig. 2a, b). Similarly, qRT-PCR results were confirmed at the protein level (Fig. 2c). CCK-8 assays showed that RBM15 knockdown significantly repressed cell proliferation of LSCC cells, while RBM15 overexpression markedly promotes cell proliferation (Additional file 9: Figure S7). After knocking down RBM15, the migration ability of TU-212 and AMC-HN-8 cells obviously decreased (Fig. 2d, e). RBM15 overexpression apparently increased the migration speed of TU-212 and AMC-HN-8 cells (Fig. 2f, g). The transwell assays demonstrated that the migration and invasion ability of LSCC cells decreased when transfected with shRNA against RBM15 compared with the control group (Fig. 2h, i). On the contrary, when RBM15 was overexpressed, the migration and invasion ability of LSCC cells significantly increased (Fig. 2j, k).
To further evaluate the effects of RBM15 on the regulation of LSCC cells progression in vivo, LSCC cells were subcutaneously injected into 30 mice. However, as expected, when AMC-HN-8 cells with down-regulated RBM15 expression were injected, the xenograft tumour volume was significantly reduced compared to the control group (Fig. 3a). The tumor volumes in the RBM15 overexpression group were increased compared to those in the vector group (Fig. 3b). Additionally, we performed TUNEL staining and transmission electron microscopy to evaluate cell apoptosis. The typical apoptotic morphology was observed in the RBM15 gene knockout group, while the control group, vector, and RBM15 overexpression group showed a complete cell membrane and complete organelle morphology (Fig. 3c, d). In summary, the above results emphasized the crucial role of RBM15 in the progression of LSCC in vitro and in vivo.
RBM15 participates in the methylation of TMBIM6 in the form of m6A modification
In order to investigate the molecular mechanism of RBM15 regulating LSCC m6A modification, LSCC cells were transfected with shRBM15. As shown in the results of the m6A RNA Methylation Quantification Kit (ab185912, Abcam, UK), the m6A methylation level in LSCC cells significantly decreased after inhibiting RBM15 (Fig. 4a). First, m6A RNA immunoprecipitation (RIP) microarray was performed to identify candidate genes with high m6A methylation modification and increased mRNA expression (Fig. 4b, c). The top 6 mRNAs were screened based on fold change and p-value (Additional file 3: Table S2). Moreover, LSCC cells were transfected with shRBM15 lentivirus for 48 h, and the qRT-PCR data indicated that the mRNA levels of CPNE5, TMBIM6, and ATAD3A decreased after RBM15 knockdown. In contrast, the mRNA levels of CPNE5 and TMBIM6 in LSCC cells in the RBM15 overexpression group showed an increasing trend (Fig. 4d, e).
Following the analysis of the above results, TMBIM6, which had the most remarkable effect of methylation mediated by RBM15, was selected for further research. MeRIP assay was performed to investigate the m6A modification status of TMBIM6 further. Our results suggest that the TMBIM6 sequence is enriched in m6A modifications (Fig. 4f). Next, we performed anti-m6A immunoprecipitation followed by qRT-PCR analysis of the m6A immunoprecipitated RNAs. The results showed that the down-regulation of RBM15 caused a decrease in the m6A modification of the TMBIM6 mRNA. By contrast, RBM15 overexpression increased m6A modification of the TMBIM6 mRNA (Fig. 4g, h). From the perspective of TMBIM6 mRNA, when RBM15 was knocked down or overexpressed, the mRNA level of TMBIM6 was down-regulated or up-regulated, respectively (Fig. 4i, j). These results fully illustrate the regulation of TMBIM6 expression by RBM15.
Next, the bacterial single-stranded RNase MazF assay was used to identify the sites of m6A methylation. This RNase MazF could digest ACA sites of cellular RNAs without m6A methylation, while sites with m6A methylation remained uncleaved (Fig. 4k). A potential m6A motif was identified as the m6A deposition site of TMBIM6 downstream, located at base 2222 of the 3’UTR, which was consistent with the prediction of the SRAMP website. We mutated ACA in the base sequence to CCA and then performed a luciferase assay (Fig. 4l). Dual-luciferase reporter assay was further used to determine the combinations of RBM15 and TMBIM6. The results indicated that the luciferase activity in the shRBM15 group was significantly lower compared to shCtrl group, while the activity of the TMBIM6-Mut group did not significantly change (Fig. 4m). Similarly, in TU-212 cells, the luciferase activity of the RBM15 overexpression group was significantly increased compared to the vector group, while the activity of the TMBIM6-Mut group did not significantly change (Fig. 4n). These results suggested a direct interaction between TMBIM6 and RBM15.
TMBIM6 was determined as a direct target of RBM15, and qRT-PCR was performed in the 34 pairs of LSCC tissues. TMBIM6 was remarkably upregulated in LSCC tissues compared to adjacent nontumor tissues. Compared with the NHBEC cell line, the TMBIM6 mRNA level in the LSCC cell line also increased (Fig. 4o, p). IHC was performed to detect the expression of TMBIM6 in 122 pairs of LSCC samples. As shown in Additional file 4: Figure S2, high expression of TMBIM6 was detected in cancerous tissues. Kaplan-Meier survival curves indicated that patients with high TMBIM6 expression had poorer overall survival than those with low TMBIM6 expression (Additional file 5: Figure S3). By analyzing the information of 34 specimens, it was revealed that RBM15 mRNA expression in LSCC tissues was positively associated with TMBIM6 levels (Fig. 4q), which was consistent with the TCGA database (Fig. 4r). By analyzing the TCGA database, the Kaplan–Meier survival curve showed that the overall survival rate of HNSC patients with high TMBIM6 expression was worse than the prognosis of low expression (Fig. 4s). To further facilitate research, we constructed a lentivirus that knockdown and overexpressed TMBIM6 (Fig. 4t, u).
RBM15 accelerates LSCC malignant progression by upregulating TMBIM6
The ablation of RBM15 could partially neutralize the effects of increased TMBIM6 on the expression of TMBIM6 in AMC-HN-8 cells (Fig. 5a). In TU-212 cells, when RBM15 was overexpressed, the protein expression of RBM15 and TMBIM6 increased. The overexpression of RBM15 could partially neutralize the effect of inhibiting TMBIM6 on TMBIM6 protein expression (Fig. 5b). Moreover, transwell assays confirmed that the invasion ability of AMC-HN-8 cells was significantly reduced after inhibiting the expression of RBM15. The reduction of invasive ability of LSCC cells exerted by shRBM15–2 was reversed by co-transfection with overexpression TMBIM6. The inhibition of RBM15 resulted in obviously increased cell apoptosis compared with those in the control group, characterized by brightly stained nuclei, nuclear condensation, and fragmentation. Yet, the overexpression of TMBIM6 partially neutralized the ablation effect of RBM15 (Fig. 5c). Transwell assays confirmed that the invasive ability of TU-212 cells was significantly reduced after inhibiting the expression of TMBIM6. The reduction of the invasion ability of LSCC cells exerted by shTMBIM6–2 was reversed by co-transfection with overexpression of RBM15. Besides, the inhibition of TMBIM6 resulted in obviously increased cell apoptosis compared with the control group; however, the overexpression of RBM15 partially neutralized the ablation effect of TMBIM6 (Fig. 5d). Consistently, nude mice xenografts injected with the AMC-HN-8 cells revealed that knockdown of RBM15 counteracted the effects of increased TMBIM6 on tumor volumes (Fig. 5e). Nude mice xenografts injected with the TU-212 cells revealed that knockdown of TMBIM6 counteracted the effects of increased RBM15 on tumor volumes (Fig. 5f).
In brief, the m6A modification on TMBIM6 that includes RBM15 is essential for the up-regulation of TMBIM6 to promote the progression and invasion of LSCC.
RBM15-mediated m6A modification of TMBIM6 mRNA enhances TMBIM6 stability through IGF2BP3-dependent
Given that the m6A methylation modification level of TMBIM6 and its mRNA expression level are simultaneously increased, we speculate that m6A modification enhances the stability of TMBIM6 through the IGF2BP family dependence. Combined with our data, the mRNA-seq results showed prominent differential expression of IGF2BP3, and the proteomics data revealed that the expression of IGF2BP2 and IGF2BP3 was increased, and the differential expression of IGF2BP3 was particularly significant. Therefore, we verified whether IGF2BP2 or IGF2BP3 were m6A-dependent in a manner that enhances the stability of TMBIM6. Firstly, we detected the enrichment of IGF2BP3 binding to TMBIM6 m6A modification sites by RNA RIP-qPCR assay (Fig. 6a), after which designed and synthesized knockdown and overexpression lentiviral vectors for IGF2BP2 and IGF2BP3, respectively (Fig. 6b, c and Additional file 6: Figure S4). Interestingly, the anti-m6A immunoprecipitation followed by qRT-PCR analysis results showed that the inhibition of IGF2BP3 caused a decrease in the m6A modification of TMBIM6 mRNA. By contrast, IGF2BP3 overexpression increased m6A modification of the TMBIM6 mRNA (Fig. 6d, e).
From the perspective of TMBIM6 mRNA, when IGF2BP3 was knocked down or overexpressed, the mRNA level of TMBIM6 was down-regulated or up-regulated, respectively (Fig. 6f, g). Nevertheless, when IGF2BP2 was knocked-down or overexpressed, it had a trifling impact on TMBIM6 expression (Additional file 7: Figure S5). Furthermore, RIP-qPCR revealed that the ablation of RBM15 could weaken the direct interaction within IGF2BP3 and TMBIM6 mRNA in LSCC cells. This result indicated that RBM15, IGF2BP3, and TMBIM6 mRNA were strongly interrelated (Fig. 6h).
A luciferase reporter assay was performed to further reveal weather TMBIM6 levels were regulated by IGF2BP3 involved in m6A modification. The results indicated that the luciferase activity of the shIGF2BP3 group was significantly lower than that of the shCtrl group, while the activity of the TMBIM6-Mut group did not significantly change. Mutually, the luciferase activity of the IGF2BP3 overexpression group was significantly increased compared to the vector group, while the activity of the TMBIM6-Mut group did not significantly change (Fig. 6i, j). In summary, these data revealed that IGF2BP3 affected the stability of TMBIM6 by participating in the TMBIM6 m6A modification.
Subsequently, qRT-PCR was applied to investigate the IGF2BP3 expression in LSCC, and the detection results were generally consistent with the TCGA database, i.e., IGF2BP3 in LSCC was significantly upregulated compared to healthy tissues adjacent to cancer (Fig. 6k, l). Moreover, immunohistochemical analysis was carried out to detect the expression of IGF2BP3 in LSCC. As shown in Fig. 6m, stronger IGF2BP3 staining was detected in 122 LSCC compared with adjacent non-tumor tissues. The 60-month follow-up results showed that the overexpression of IGF2BP3 was closely related to the poor prognosis of LSCC patients (Fig. 6n). Furthermore, TCGA data indicated that HNSC patients with high expression of IGF2BP3 had a poor prognosis (p < 0.05; Fig. 6o, p). Besides, a prominent positive correlation between TMBIM6 and IGF2BP3 was calculated using RT-PCR quantitation. Additional data obtained from the TCGA database were shown to be consistent with our qRT-PCR results (Fig. 6q, r). Also, the qRT-PCR results in 34 pairs of LSCC tissues suggested that the IGF2BP3 expression was associated with the clinicopathological characteristics of LSCC. T3 + T4 tumours, III + IV tumours, and N1 tumours had higher expression levels of IGF2BP3 (Additional file 8: Figure S6).
To determine whether IGF2BP3 is an important influence factor of TMBIM6 in LSCC tumorigenesis, a rescue experiment was performed. As shown in Fig. 7a, the ablation of TMBIM6 could partially neutralize the effects of increased IGF2BP3 on the expression of TMBIM6 in AMC-HN-8 cells. In TU-212 cells, overexpression of TMBIM6 had a trifling effect on the expression level of IGF2BP3 protein, and the expression of TMBIM6 protein also decreased after IGF2BP3 was knocked down. Nevertheless, the overexpression of TMBIM6 could partially neutralize the effect of inhibiting IGF2BP3 on TMBIM6 protein expression (Fig. 7b). Moreover, transwell assays confirmed that the invasion ability of AMC-HN-8 cells was significantly reduced after inhibiting the expression of TMBIM6. Reduced invasive ability of LSCC cells exerted by shTMBIM6–2 was reversed by co-transfection with overexpression of IGF2BP3. The ablation of TMBIM6 resulted in obviously increased cell apoptosis compared with the control group and was characterized by brightly stained nuclei, nuclear condensation, and fragmentation. However, the overexpression of IGF2BP3 partially neutralized the ablation effect of TMBIM6 (Fig. 7c). Transwell assays confirmed that the invasion ability of TU-212 cells was significantly reduced after inhibiting the expression of IGF2BP3. Reduced invasion ability of LSCC cells exerted by shIGF2BP3–1 was reversed by co-transfection with overexpression of TMBIM6. In addition, the inhibition of IGF2BP3 resulted in obviously increased cell apoptosis compared with the control group; however, the overexpression of TMBIM6 partially neutralized the ablation effect of IGF2BP3 (Fig. 7d). RNA stability assays suggested that the ablation of RBM15 shortened the half-life of TMBIM6 mRNA, while overexpression of RBM15 prolonged the half-life of TMBIM6 mRNA (Fig. 7e, f).
Furthermore, the ablation of IGF2BP3 shortened the half-life of TMBIM6 mRNA. In contrast, overexpression of IGF2BP3 prolonged the half-life of TMBIM6 mRNA (Fig. 7g, h). To further confirm whether RBM15 and IGF2BP3 were involved in regulating the expression of TMBIM6, another rescue experiment was conducted. In AMC-HN-8 cells with simultaneous inhibition of RBM15 and IGF2BP3, the expression of TMBIM6 was significantly reduced (Compared lane 1 to lane 5, Fig. 8). In contrast, while the simultaneous overexpression of RBM15 and IGF2BP3 increased TMBIM6 expression, downregulation of RBM15 or IGF2BP3 could impact TMBIM6 overexpression (compared lane 2, 3 to lane 4, Fig. 8).
Taken together, the above results reconfirmed that RBM15 and IGF2BP3 had an essential role in promoting the stability of TMBIM6.