1. IGF2BP1 is highly expressed in ESCC tissues and is associated with the depth of tumor invasion
We examined the protein expression level of IGF2BP1 in 311 ESCC tissues and 9 adjacent normal tissues by IHC. The results showed that IGF2BP1 was highly expressed in ESCC tissues (155/311, 49.8%) but was not expressed or only weakly expressed in normal esophageal epithelia (Fig. 1A, Table 1). Positive staining was predominant in the cytoplasm of ESCC cells. A higher IGF2BP1 expression level was positively correlated with the depth of tumor invasion (T1-2 versus T3-4), but no significant differences were found in other clinicopathological features, such as sex, age, histologic grade, lymph node metastasis and clinical stage (Table 1).
Furthermore, an analysis of RNA-seq data obtained from the TCGA database revealed that the mRNA expression level of IGF2BP1 was elevated in ESCC specimens compared with normal tissues (Fig. 1B), which was consistent with the IHC results.
2. IGF2BP1 promotes ESCC cell migration, invasion and metastasis
To identify the role of IGF2BP1 in ESCC, the expression of IGF2BP1 in 11 ESCC cell lines was detected by Western blotting. We found that IGF2BP1 had higher expression in KYSE30, KYSE450, KYSE140 and TE1 cells but was almost undetectable in KYSE70, KYSE150 and TE10 cells (Fig. 2A). KYSE30 and TE1 were applied for functional study as cell models. After transient knockdown of IGF2BP1 mediated by small interfering RNA (siRNA), there was no significant change in cell growth within six days compared with the negative control group (Fig. 2B). However, the migration and invasion ability of both cell lines were substantially repressed after IGF2BP1 silencing in the Transwell assays (Fig. 2C) and wound healing assays (Fig. 2D).
Next, KYSE30 cells stably expressing shRNA of IGF2BP1 (shIGF2BP1) or nonsilencing shRNA (shNS) conducted by lentivirus infection were injected into nude mice via the tail vein. The formation of lung metastatic tumors was observed in both groups six weeks later, but the number of lung metastases in the shIGF2BP1 group was significantly reduced, and the metastatic nodules were noticeably smaller (Fig. 2E).
3. IGF2BP1 increases INHBA mRNA stability, promoting ESCC cell migration and invasion
Given that IGF2BP1 is an RNA binding protein, we performed RIP-seq in KYSE30 cells to identify its potential RNA targets. RIP-seq profiling revealed that most of the IGF2BP1-binding sites were highly enriched in coding regions (CDSs) of its mRNA targets, indicating that IGF2BP1 mainly regulates gene expression by binding to the CDS (Additional file 2: Figure S1). Through GO analysis of the identified results (log FC >1 and FDR < 0.01) by RIP-seq (Additional file 3: Table S7), we selected 30 genes related to cell migration and invasion as candidates (Additional file 4: Table S8). Then, quantitative real-time PCR was employed to detect the mRNA expression of the candidate genes in IGF2BP1 stable knockdown KYSE30 and TE1 cell lines. Notably, the reduction in INHBA mRNA abundance after IGF2BP1 knockdown was most significant among the 30 candidate genes in KYSE30 and TE1 cells (Fig. 3A, Additional file 5: Table S9). RIP-PCR and RNA pulldown using biotin-labeled DNA probes against INHBA mRNA further confirmed the interaction between IGF2BP1 protein and INHBA mRNA (Fig. 3B-C). Furthermore, silencing IGF2BP1 downregulated the protein expression of INHBA and Smad2/3 in KYSE30 and TE1 cells (Fig. 3D).
Previous studies have shown that IGF2BP1 can recognize N6-methyladenosine (m6A) and enhance mRNA stability and translation in an m6A-dependent manner [14, 15]. Therefore, it is highly possible that IGF2BP1 binds and stabilizes the mRNA of INHBA. RNA stability assays showed that the half-life of INHBA mRNA was significantly shortened after knockdown of IGF2BP1 in KYSE30 and TE1 cells (Fig. 3E). As the m6A writers METTL3 or METTL14 was repressed by siRNAs, the INHBA protein levels were decreased in both cell lines without affecting IGF2BP1 (Fig. 3F). Moreover, gene-specific m6A qPCR further confirmed that INHBA was regulated by m6A modification (Fig. 3G).
INHBA is a member of the transforming growth factor β (TGF-β) superfamily, which is closely associated with tumor invasion and metastasis. We employed a Transwell assay to assess the effects of INHBA on invasive and migratory phenotypes in ESCC cells. We observed that cell invasion and migration were inhibited after knockdown of INHBA mediated by siRNA (Fig. 3H), while transient overexpression of INHBA in cells with stable knockdown of IGF2BP1 partially overcame this inhibition (Fig. 3I). Western blotting results showed that Smad2/3 expression was repressed by IGF2BP1 depletion in KYSE30 and TE1 cells, while the IGF2BP1 knockdown-induced Smad2/3 decrease was reversed by INHBA overexpression (Fig. 3J).
4. IGF2BP1 activates INHBA-Smad2/3 signaling by recruiting G3BP1
It has been reported that IGF2BPs interact with other RNA-binding proteins (RBPs) to regulate mRNA targets [21, 22]. To further elucidate the functional mechanism of IGF2BP1 in ESCC cells, we adopted Co-IP-MS to investigate interactive partners of IGF2BP1 (Additional file 6: Figure S2). A total of 227 potential proteins were identified (Additional file 7: Table S10), and 46% of them participate in RNA regulation (Additional file 8: Table S11). GO and pathway enrichment analysis revealed that most of them were involved in RNA processing, localization, stability, metabolism, transport and RNA splicing (Fig. 4A-B).
We selected RNA-binding proteins as candidates for validation according to the following criteria: 1) By GO analysis, the candidates are related to migration and invasion. 2) According to GeneCards database, the sub-cellular localization of candidates are in the cytoplasm, which is the same as IGF2BP1 (https://www.genecards.org/). Based on the above criteria, we got eight candidate proteins in which G3BP1 had the top Coverage and PSM values in the IP group (Additional file 9: Table S12). Thus, we selected G3BP1 for further study. Endogenous G3BP1 was immunoprecipitated by IGF2BP1 antibody using a Co-IP assay, and vice versa (Fig. 4C). Immunofluorescence images captured using confocal microscopy confirmed the cytoplasmic colocalization of G3BP1 and IGF2BP1 in KYSE30 and TE1 cells (Fig. 4D). Moreover, Western blotting results showed that silencing G3BP1 by siRNA led to a decrease in INHBA and Smad2/3 and a slight reduction in IGF2BP1 (Fig. 4E).
5. INHBA is significantly upregulated in ESCC, HNSC and invasive breast cancer
We then analyzed TCGA transcriptome sequencing datasets and observed that the mRNA of INHBA was also increased in ESCC tissues (Fig. 5A). RISH analysis of tissue microarrays further confirmed the RNA-seq results. More importantly, INHBA mRNA was mainly distributed in the peripheral tumor cells of cancer nests and the stroma of ESCC tissues but was negative in the normal esophageal epithelia and stroma (Fig. 5B). In addition, in HNSC and invasive breast cancer, INHBA mRNA was significantly elevated based on TCGA dataset analysis (http://maplab.imppc.org/wanderer/) (Fig. 5C).
6. The small molecule inhibitor BTYNB significantly inhibits the invasion, migration and proliferation of ESCC cells in vitro
Currently, there are no inhibitors in clinical trials that directly target IGF2BP1. Previous studies reported that a small molecule drug, BTYNB, could inhibit the binding of IGF2BP1 to c-Myc mRNA [17, 23]. BTYNB also impairs cell proliferation in vitro by blocking β-TRCP1, E2F and other transcripts [17, 18]. Therefore, it is reasonable to speculate that BTYNB might interfere with the IGF2BP1-driven malignant phenotypes in ESCC cells. We tested BTYNB with a Transwell assay by adding the drug into the lower chamber and found that the migration and invasion of KYSE30 and TE1 cells were significantly inhibited at 36 h and 24 h, respectively (Fig. 6A). Meanwhile, cell viability and colony formation were impaired in a dose-dependent manner (Fig. 6B-C). In addition, 48 h of exposure to BTYNB increased the number of apoptotic cells (Fig. 6D). Western blotting analysis showed that INHBA and Smad2/3 expression was decreased after BTYNB treatment (Fig. 6E).
7. Elevated expression of IGF2BP1 in ESCC cells is associated with hypomethylation of its first intron
We further explored possible dysregulation of IGF2BP1 in other types of SCCs and found higher IGF2BP1 mRNA in HNSCC, LUSC, and CESC than in the corresponding normal tissues (http://maplab.imppc.org/wanderer/) (Fig. 7A). Moreover, we observed a high degree of consistency between the mRNA and protein levels of IGF2BP1 in 10 ESCC cell lines (Fig. 7B). Interestingly, according to the HPA (Human Protein Atlas) database, IGF2BP1 was almost absent in normal esophageal tissues and other normal tissues except in the embryo and reproductive system (Fig. 7C). To uncover the mechanism of IGF2BP1 mRNA upregulation in ESCC cells, we examined the methylation status of this gene. Three pairs of primers (methylated and unmethylated primers) were designed for MSP-PCR to detect distinct CG sites in the first intron of IGF2BP1 in both high and low IGF2BP1 expression cell lines. The results showed that the IGF2BP1 gene was hypomethylated in cell lines with high IGF2BP1 expression but hypermethylated in cell lines with low IGF2BP1 expression, except for TE10 (Fig. 7D).