3.1 CTNNBIP1-CLSTN1 is widely expressed among human tissues and cell types.
Previously, we identified 291 recurrent chimeric RNAs by analyzing RN-Seq data from around 300 non-cancer tissues and cells (7). CTNNBIP1-CLSTN1 is one of them, detected by RT-PCR in liver, lung, kidney etc. We recently expanded our chimeric RNA search to the whole Genotype-Tissue Expression (GTEx), which contains 9,495 non-diseased human tissue samples from 53 different tissues (14, 15). From this study, we noticed that CTNNBIP1-CLSTN1 was detected in almost all samples (Fig. 1A and Fig. S1).
We then performed RT-PCR to detect the fusion and its parental genes in 15 non-cancer cell lines belonging to 12 different tissue of origins, including cell types of fibrocytes, amniocytes, epithelial cells, stem cells, and vascular endothelial cells, followed by gel-electrophoresis (Fig. 1B). CTNNBIP1-CLSTN1 was discovered in all the samples. Using qRT-PCR, we quantified its expression in 15 non-cancer cells, and 15 cancer cell lines of esophageal and prostate cancer. There was no statistically significant difference in the expression level between cancer and normal lines (Fig. 1C, and Fig. S2-S3). The junction sequence was shown exact the same among different cancer cell lines, LNCaP and KYSE-30, and non-cancer lines, RWPE-1 and HEEC by Sanger sequencing (Fig. 1D). We also monitored the wild type parental transcripts using primers covering at least one exon not shared with the fusion transcript. Even though the wild type CTNNBIP1 transcript was readily detectable, the wild type CLSTN1 was not. We also found that the relative expressions of the chimeric fusion and CTNNBIP1 are positively correlated (Pearson’s correlation R = 0.37, P < 0.001) (Fig. 1E, and Fig. S2-S3), consistent with the chimeric RNA being a product of 5’ gene read-through, and suggesting that the great majority of CLSTN1 transcripts is used to form chimeric RNA.
This chimeric RNA is predicted to encode an in-frame chimeric protein. Here, we decided to take advantage of the fact that regular protein-coding mRNAs are mainly present in the cytoplasm, while long-non-coding RNAs often reside in the nucleus to regulate transcription (16–18). We chose human embryonic kidney cell HEK-293T, human umbilical vein endothelial cell HUVEC, and hepatocyte line LO2 cells for fractionation assay and extracted RNA to detect the chimeric RNA by qRT-PCR. Classic protein-coding gene GAPDH and long non-coding RNA MALAT1 were used as controls. Not surprisingly, MALAT1 was found mostly in the nuclear fraction. On the contrary, CTNNBIP1-CLSTN1, wild type CTNNBIP1 and GAPDH were all enriched in the cytoplasmic part, (Fig. 1F). This observation is consistent with the previous experiment of Western blot (7), further supporting that the fusion has a protein-coding function.
3.2. Silencing CTNNBIP1-CLSTN1 reduced cell proliferation and cell motility
Based on its ubiquitous expression pattern, we hypothesized that the chimeric RNA may belong to the group of chimeric RNAs, which we call “housekeeping chimeric RNAs”. Indeed, previous report has documented its indispensible role in immortalized astrocytes (7). To further investigate the function of CTNNBIP1-CLSTN1, and support its basic role of principle cell maintenance, we transfected siRNAs targeting the chimera in multiple human non-cancer cell lines, HEK-293T, HUVEC and LO2 cell lines. The siRNA, siCTNNBIP1-CLSTN1 targeting the fusion junction sequence resulted in significant reduction of the fusion transcript, but not the wild-type CTNNBIP1 in all three cell lines (Fig. S4 and S5). Since, we could not design a siRNA that specifically silences the wild type CTNNBIP1, as a control, we designed a siRNA, siCTNNBIP1 that targets the common region sequence in the fusion and wild type CTNNBIP1. As predicted, this siRNA lead to the silencing of both the fusion and wild type CTNNBIP1 (Fig. S5). Both siCTNNBIP1-CLSTN1 and siCTNNBIP1 significantly decreased cell proliferation based on CCK8 measurement (Fig. 2A). Cell migration was monitored by wound healing assay with live-cell-imaging microscopy after scratching. The percent relative wound density was calculated by measuring the density of cells that migrated into the original wound. Both siCTNNBIP1-CLSTN1 and siCTNNBIP1 significantly inhibited cell migration compared to the negative control (Fig. 2B). Figure 2C shows representative microscopic images of cells across a wound at zero and 12 hours in HUVEC and LO2 cells, and zero and 24 hours in HEK-293T cells. These results suggest that CTNNBIP1-CLSTN1 plays a significant role in general cell growth and movement, regardless of cell types.
To rule out the off-target effect of siRNA, we performed rescue experiments by transfecting in the construct of CTNNBIP1-CLSTN1 expressing plasmid and construct encoding the wild type CTNNBIP1. The reduced cell proliferation was indeed rescued by the CTNNBIP1-CLSTN1 construct in both siCTNNBIP1-CLSTN1 and siCTNNBIP1 groups. In contrast, the wild type CTNNBIP1 failed to rescue in either groups (Fig. 2D), suggesting the effect of reduced cell proliferation was caused by the fusion silencing, but the wild type CTNNBIP1.
3.3. Silencing CTNNBIP1-CLSTN1 resulted in cell cycle arrest and apoptosis
To investigate the mechanism of the reduced cell proliferation caused by silencing the fusion, we performed flow cytometry to evaluate cell cycle with propidium iodide staining after siRNA transfection. Cell numbers in G2/M phase were notably increased in both siRNA groups that silenced the fusion, compared to that in the NC group (Fig. 3A). This G2/M arrest was observed in all three cell lines (Fig. 3B).
We then evaluated cell apoptosis using Annexin V-FITC and propidium iodide staining. The results from the flow cytometry assay showed that siCTNNBIP1 and siCTNNBIP1-CLSTN1 transfection noticeably increased apoptosis in all three cell lines, compared to the control (Fig. 3C and Fig. 3D).
3.4. Overexpression of CTNNBIP1-CLSTN1, but not the wild type CTNNBIP1, promotes cell growth and cell migration
In this experiment, HEK-293T, HUVEC, and LO2 cells were transfected with either pCDNA3.1 plasmid that encodes the full-length CDS of CTNNBIP1-CLSTN1, or CTNNBIP1, or the empty vector respectively. In contrast to the loss-of-function experiments, CCK8 assay showed that the expression of the chimera, but not the wild type CTNNBIP1, promoted cell proliferation in all three cell lines (Fig. 4A). Consistently, wound healing assay also demonstrated that the overexpression of CTNNBIP1-CLSTN1, but not CTNNBIP1 enhanced the cell migration ability in all three cell lines (Fig. 4B).
3.5. CTNNBIP1-CLSTN1 influences cell proliferation by regulating autocrine signaling factors SERPINE2
To investigate the downstream pathway mediating the effect of CTNNBIP1-CLSTN1 on cell proliferation, we performed transcriptome sequencing analysis of HEK-293T, HUVEC and LO2 cells transfected with siCTNNBIP1-CLSTN1 and negative control siRNA. Differential expression analysis between the two groups in three cell lines was conducted, obtaining 84, 605, and 393 differentially expressed genes (DEGs) (Supplementary Table 1), respectively. Volcano plot of DEGs in three cell lines are shown in Fig. 5A (padj < 0.05). Gene Ontology (GO) terms for DEGs were also examined. 135 enriched GO terms in HEK-293T, 286 enriched GO terms in HUVEC, and 137 enriched GO terms in LO2 were found (Supplementary Table 1). Several terms showed unconformity in different cells, and were consistent with the general role the fusion plays. However, there are also specific GO terms unique to individual cell lines, suggesting that the fusion may have additional roles that are cell type specific. The top 13 for HEK-293T, the top 23 for HUVEC, and the top 14 for LO2 in the GO term list are displayed (Fig. S6).
Out of the DEGs, Fibronectin 1 (FN1) and Serine protease inhibitor E2 (SERPINE2) stand out, owing to the fact that they both were down-regulated in siCTNNBIP1-CLSTN1 group of all three cell lines. Furthermore, they were included in several GO terms related to cell growth (GO:0001558, GO:0040008, and GO:0016049). Fibronectin 1(FN1)(19, 20), is a member of the glycoprotein family that is widely expressed by multiple cell types (GTEx data, Fig. S7A). Serine protease inhibitor E member 2 (SERPINE2) (21, 22), also called Protease Nexin-1(PN-1) belongs to the Serpin gene super family, is also ubiquitously expressed in human tissues and cells (GTEx, Fig. S7B). We first confirmed their expression by qRT-PCR (Fig. 5B). Both were reduced upon the silencing of the fusion. In contrast, overexpression of CTNNBIP1-CLSTN1 up-regulated both genes (Fig. 5C). We then investigated whether the effect of the fusion is mediated by these two genes. To do so, we transfected in expression plasmids encoding FN1 or SERPINE2 in cells with siCTNNBIP1-CLSTN1 or siCT. As indicated in Fig. 5D, SERPINE2 can restore the cell proliferation rate to a normal level. In contrast, FN1 failed to rescue the reduced cellular growth induced by siCTNNBIP1-CLSTN1 (Fig. 5E), suggesting that the effect of the chimeric RNA on cell proliferation is mostly mediated by SERPINE2.