CircPLCE1 facilitates the malignant progression of colorectal cancer by repressing the SRSF2‐dependent PLCE1 pre‐RNA splicing

Abstract Studies have demonstrated that circular RNAs (circRNAs) play important roles in various types of cancer; however, the mechanisms of circRNAs located in the nucleus have rarely been explored. Here, we report a novel circular RNA circPLCE1 (hsa_circ_0019230) that facilitates the malignant progression of colorectal cancer (CRC) by repressing serine/arginine‐rich splicing factor 2 (SRSF2)‐dependent phospholipase C epsilon 1 (PLCE1) pre‐RNA splicing. Quantitative real‐time polymerase chain reaction was used to determine the expression of circPLCE1 in CRC tissues and cells. Cell Counting Kit‐8, Transwell and flow cytometric assays were used to assess the role of circPLE1 in CRC cell proliferation, migration and apoptosis, respectively. An animal study was conducted to test the role of circPLCE1 in vivo. Furthermore, catRAPID and RPISeq were used to predict the possible binding proteins of circPLCE1. RNA fractionation and RNA immunoprecipitation assays were used to confirm the RNA‐protein interaction. In this study, we found that circPLCE1 was more significantly down‐regulated in CRC tissues compared with that in adjacent normal tissues. However, circPLCE1 knockdown suppressed CRC cell proliferation, migration and invasion and increased apoptosis. Nude mouse experiments showed that ectopic expression of circPLCE1 dramatically increased tumour growth in vivo. Mechanistically, circPLCE1 directly bound to the SRSF2 protein, repressing SRSF2‐dependent PLCE1 pre‐RNA splicing, resulting in the progression of CRC. Individually mutating the binding sites of circPLCE1 abolished the inhibition of PLCE1 mRNA production. Our study revealed a novel molecular mechanism in the regulation of PLCE1 and suggested a new function of circular RNA.


| INTRODUC TI ON
Phospholipase C epsilon 1 (PLCE1) is the most essential member of the phospholipase family. 1 PLCE1 catalyses the hydrolysis of phosphatidylinositol 4,5-bisphosphate to generate two secondary messengers inositol 1,4,5-triphosphate and diacylglycerol. 2 These products subsequently initiate a cascade of intracellular responses that result in cellular proliferation, differentiation and apoptosis. 3 In addition, PLCE1 is a suppressor of P53 by promoting p53 promoter methylation. 4 To date, many studies have reported that PLCE1 may play an essential role in carcinogenesis and progression of various human cancers, including oesophageal cancer, 5 lung cancer, 6 colorectal cancer (CRC) 7 and hepatocellular carcinoma. 8 However, little information has been reported regarding the regulation of PLCE1 protein in tumours.
Most genes in the human genome contain numerous exons interspersed with introns that undergo RNA splicing to form mature mRNA and protein products. 9 Pathological abnormal splicing is a crucial component of cancer development and progression, affecting critical cancer-associated gene expression. 10 Serine/arginine-rich splicing factor 2 (SRSF2) is an SR protein that consists of an RNA recognition motif (RRM) domain and an arginine-serine (RS) domain. 11 SRSF2 frequently binds to specific splicing enhancer sequences and functions as a general splicing activator. 12 Remarkably, SRSF2 regulates transcriptional elongation in a gene sequence-selective manner. 13 To date, the role of SRSF2 in the formation of mature PLCE1 mRNA has not been reported.
Circular RNAs (circRNAs) are generated from back-splicing of pre-RNAs to form covalently closed loops. 14 Most circRNAs arise from constitutive exons, whereas others contain introns. The biogenesis of circRNAs is abundant, stable, conserved and cell type-specific, indicating that circRNAs have potential regulatory roles. 15 Most current circRNA studies focus on circRNA as a sponge of miRNA molecules. 16 Moreover, endonuclear circRNAs have been discovered to compete against classical pre-RNA splicing, thereby reducing parent gene mature mRNA production. 17,18 Here, we provide the first evidence for a molecular mechanism by which circPLCE1 reduces the expression of PLCE1 mRNA by binding to SRSF2 protein in CRC. These results indicate that circPLCE1 is a potential therapeutic target for CRC.

| Colorectal cancer (CRC) tissues and cell lines
Fresh CRC tissues and matched adjacent normal tissues were acquired from 41 patients who underwent surgery between 2018 and 2019 at Beijing Chao-Yang Hospital. All resected fresh tissues were immediately frozen in liquid nitrogen and stored at −80℃ until use.
Written informed consent was signed by the patients or their families before the surgery began. The study was approved by the Ethics

| Small interfering RNAs and transfection
To silence circPLCE1, two antisense oligonucleotides (siRNAs) against the circPLCE1 junction site were designed and synthesized by GenePharma. The CRC cell lines SW480 and HCT116 were chosen for the transfection and subsequent functional tests. All siRNAs were transfected at a final concentration of 75 nM using Lipofectamine 3000 reagent (Invitrogen). All the siRNA sequences used are listed in Supplementary Table 2.

| circPLCE1-overexpressing lentivirus construction and infection
To overexpress cirPLCE1, full-length human circPLCE1 cDNA was synthesized and cloned into the expression vector pHBLV-CMVcirc-EF1-Puro-N by Hanbio. The control vector or circPLCE1overexpression vector was packaged into the lentivirus and infected into CRC cells. Puromycin was used to select stable circPLCE1overexpressing cells for at least 2 weeks after infection.

| Cell proliferation and colony formation assays
Cell proliferation was detected using the Cell Counting Kit-8 (CCK-8) assays (Dojindo Laboratories). The treated cells were allowed to grow in 96-well plates and incubated with 10 μL of CCK-8 solution for 2 h at 37℃. The absorbance was measured at 450 nm using a microplate reader (Thermo Fisher Scientific). The proliferation rate was measured at 0, 24, 48, 72 and 96 h after transfection. All experiments were performed five times independently.
Colony formation assays were used to evaluate the clonogenic ability of the CRC cells. Furthermore, 1000 cells per well were seeded into 6-well plates, and 14 days later, 4% paraformaldehyde was used to fix the cells, and crystal violet staining was used to visualize the clones. We counted the clones that were 2 mm or greater in size under a light microscope. The average number of colonies was determined from three independent experiments. Eventually, the migrated and invaded cells were stained with crystal violet and counted using a Leica DM4000B microscope (Leica).

| Flow cytometric assays of cell cycle and apoptosis
Cell cycle distribution and cell apoptosis rate were measured by flow cytometry. After treatment, CRC cells were collected and mixed with 75% ethanol at 4℃ overnight. The following day, the cells were For the apoptosis assay, the cells were rinsed with phosphatebuffered saline and resuspended in 500-μL binding buffer.
Fluorescein isothiocyanate and propidium iodide (KeyGen Biotech) were added and incubated for 15 min according to the manufacturer's instructions. Subsequently, the flow cytometer FACSCalibur was used to assess early cell apoptosis.

| Tumour formation in vivo
Six-week-old male BALB/c nude mice were purchased from the Charles River Laboratories. For the tumour formation experiment,

| Western blot
After removing the culture supernatants, the cells were lysed using radioimmunoprecipitation assay buffer (Beyotime

TA B L E 2
The correlation between PLCE1 mRNA expression and clinicopathological features in 41 CRC patients was harvested and quantified using a BCA Reagent Kit (Beyotime).

| Bioinformatic analysis
The secondary structure of circPLCE1 was formed by RNAfold, whereas the secondary structure of the SRSF2 protein was acquired by UniProt. We used catRAPID (http://servi ce.tarta glial ab.com/page/catra pid_group) to predict the target proteins of circPLCE1. In addition, the interaction strength was determined using the RPISeq program (http://pridb.gdcb.iasta te.edu/RPISe q/ index.html).
RNA-binding residues in SRSF2 were predicted using RNABindRPlus and Pprint web tools. We used the protein-RNA docking analysis tool NPDock server (http://genes ilico.pl/NPDock) to discover the possible interaction of circPLCE1 with SRSF2 RRM.

| Cytoplasmic and nuclear RNA fractionation
The subcellular localization of circPLCE1 was detected using a PARIS Kit according to the manufacturer's protocol (Ambion, AM1921).
GAPDH mRNA and U6 served as controls.

| Statistical analyses
All the analyses were performed using Statistical Package for the Social Sciences version 23.0 (IBM). The data are presented as the mean ±standard deviation from at least three independent experiments. Statistical significance was determined using the chi-squared test, Student's t test, Wilcoxon signed rank test and one-way analysis of variance, as appropriate. The association between circPLCE1 and PLCE1 mRNA expression in CRC tissues was analysed using the Pearson correlation coefficient. Statistical significance was set at P < .05.

| circPLCE1 was identified as a tumour promoter in CRC progression
In our previous work, we analysed the expression levels of different circRNAs in CRC tissues relative to the paired adjacent mucosal tissues (four pairs). 19,20 We observed that the circular transcript of PLCE1 (circBase ID: hsa_circ_0019230) was decreased in tumour tissues. CircPLCE1 is derived from exons 12 and 13 of the PLCE1.
Sequencing confirmed that the "head-to-tail" splice junction was identical to the circRNA database ( Figure 1A). Next, we examined circPLCE1 levels in different CRC cell lines and found that circPLCE1 levels were highest in SW480 and lovo among other CRC cell lines ( Figure 1B). To further investigate the correlation between circ-PLCE1 and CRC, we collected another 41 pairs of CRC tissues and adjacent normal mucosal tissues and performed qRT-PCR to detect circPLCE1 expression. As shown in Figure 1C, CRC tissues expressed circPLCE1 at a notably lower level than the paired mucosal tissues. Interestingly, we observed a significant positive correlation between the patients' tumour node metastasis (TNM) tumour stage and circPLCE1 expression level (P < .001, Figure 1D). Additionally, we also analysed the correlation between circPLCE1 expression and various clinicopathological characteristics of CRC patients ( Table 1).
The results indicated that the expression level of circPLCE1 was significantly associated with the depth of invasion (P = .03) and lymph node metastasis (P < .001). Taken together, circPLCE1 was identified as a tumour promoter in CRC progression.

| PLCE1 mRNA expression was antagonized by circPLCE1
In the above experiment, we observed an interesting phenomenon in which circPLCE1 was decreased in CRC tissues, but was a tumour promoter. Mechanistically, circPLCE1 was derived from back-splicing of pre-PLCE1. The primers for pre-PLCE1 and PLCE1 mRNA were designed ( Figure 4A), and the accuracy of the conjunction was confirmed by Sanger sequencing. Moreover, we experimentally confirmed that pre-PLCE1 and PLCE1 mRNA expressions were markedly lower in CRC tissues compared with those in paired adjacent normal tissues ( Figure 4B, C). Statistical analysis revealed that low pre-PLCE1 expression levels were not related to clinicopathological features in CRC patients, whereas low PLCE1 mRNA expression was considered a tumour suppressor. As presented in Table 2 and Figure 4D, reduced PLCE1 mRNA expression was related to lymph node metastasis (P = .02), distant metastasis (P = .03), TNM tumour stage (P = .02) and lymphovascular invasion (P = .04). Subsequently, we divided the 41 CRC patients into the low-and high-expression groups using the median expression level of circPLCE1 as a cut-off value ( Figure 4E). The results revealed that CRC patients in the high-expression circPLCE1 group exhibited lower PLCE1 mRNA expression than the CRC patients in the lowexpression circPLCE1 group ( Figure 4F). Similarly, we observed a significant negative correlation between circPLCE1 expression and PLCE1 mRNA expression levels (r = −.3806, P = .0141 for Spearman correlation; Figure 4G).
To further test the association between circPLCE1 and PLCE1 mRNA, we constructed circPLCE1-silence and circPLCE1-overexpression cells. The results indicated that PLCE1 mRNA was remarkably increased in response to circPLCE1 inhibition in both SW480 and HCT116 cell lines ( Figure 4H). Overexpression of circPLCE1 showed the opposite phenomenon ( Figure 4I).
However, there was no significant variation in pre-PLCE1 levels.
Western blot analysis was used to detect the expression of PLCE1 protein in CRC cells. The trends of PLCE1 protein were similar to those of mature mRNA of PLCE1, indicating that the process of translation was not influenced ( Figure 4J, K). Collectively, these findings indicate that circPLCE1 plays a fundamental role in PLCE1 mRNA processing.

| circPLCE1 suppressed RNA splicing of pre-PLCE1 by binding to SRSF2 protein
Next, we deciphered the possible mechanism by which PLCE1 mRNA processing was repressed by circPLCE1. We used the catRAPID website to predict the potential protein interactions with circPLCE1 (Table 3). Among these candidates, SRSF2 was selected as the downstream target of circPLCE1. SRSF2 is the only SR family member that is exclusively localized in the nucleus and is therefore restricted to nuclear processes ( Figure 5A). Next, we analysed the interaction strength of SRSF2 with circPLCE1 or pre-PLCE1 using a computer algorithm (RPISeq). Bioinformatic analysis suggested that circPLCE1 or pre-PLCE1 had promising potential for binding with SRSF2 protein (Table 4).
To further verify the molecular mechanisms of circPLCE1 in CRC cells, we first analysed the subcellular distribution of circPLCE1. The results of qRT-PCR analysis after cell fractionation showed that circPLCE1 was mainly localized in the nucleus ( Figure 5B). Next, we conducted RNA immunoprecipitation (RIP) assays and found that circPLCE1 and pre-PLCE1 were more significantly immunoprecipitated by anti-SRSF2 antibody compared with negative anti-IgG antibody ( Figure 5C, D). To clarify the function of SRSF2 in PLCE1 mRNA processing, we designed siRNAs for SRSF2 mRNA ( Figure 5E). Silencing of SRSF2 decreased the expression of PLCE1 mRNA, and silencing of circPLCE1 could not rescue the SRSF2-silence-mediated decline of PLCE1 mRNA ( Figure 5F).

| Identification of the binding sites of circPLCE1 and SRSF2 protein
To identify the binding sites of circPLCE1 with SRSF2, we constructed a docking model using computational approaches.
RNABindRPlus and Pprint servers predicted that the RNA-binding residues were concentrated in the RRM region of the SRSF2 protein ( Figure 6A). Previous studies have demonstrated that the SRSF2 RRM region strongly binds to the RNA sequence 5'-UGCAGU-3', which was also found in circPLCE1 13 ( Figure 6B) Interaction probabilities generated by RPISeq range from 0 to 1. In performance evaluation experiments, predictions with probabilities >0.5 were considered "positive," that is, indicating that the corresponding RNA and protein are likely to interact.
used to perform the in silico molecular docking of circPLCE1 with SRSF2 ( Figure 6C).
A plasmid containing the circPLCE1-SRSF2 binding site mutation  Figure 6D). In addition, inhibition of PLCE1 mRNA expression by circPLCE1 was abolished by mutating the binding sites of circPLCE1 with SRSF2 ( Figure 6E). The in vitro cell experiments supported the conclusion that circPLCE1 could dock the SRSF2 RRM region.

| D ISCUSS I ON
It has been well-established that PLCE1 plays a critical role in inhibiting CRC progression. 7 Exploring distinctive and effective In the circPLCE1 low-expression CRC cells, SRSF2 is responsible for initiating spliceosome assembly on pre-PLCE1 ①, which promote the production of PLCE1 mRNA ②. Conversely, in the circPLCE1 high-expression CRC cells, circPLCE1 is directly binding to SRSF2 and suppress the process of PLCE1 pre-RNA splicing ③ Recently, evidence has accumulated that circRNAs are frequently dysregulated in human cancer and affect tumorigenesis, invasion and metastasis. 23,24 Many molecular functions are performed by cir-cRNAs to implement their cellular effects, such as interacting with different proteins to form specific RNPs and influence their intrinsic function. [25][26][27] In our study, we present evidence that PLCE1 mRNA production is negatively controlled by its circRNA via suppression of pre-PLCE1 splicing. One of our main findings is that circPLCE1 directly binds to the SRSF2 protein and suppresses the spliceosome assembly on nascent pre-PLCE1 to catalyse intron removal before the completion of transcription ( Figure 6F).
The role of circPLCE1 in the genesis and progression of CRC tumours remains contradictory. In our experiments, we found that the expression of circPLCE1 was higher in normal compared with cancerous tissues and in advanced-stage tumours compared with early-stage tumours. This phenomenon has indicated that the expression of circPLCE1 may be regulated by some upstream molecules. However, the small sample size limits the generalizability of this study. Moreover, investigations are required to understand circPLCE1 functions in CRC cells both in the nucleus and in the cytoplasm. Although we have discovered that circPLCE1 represses PLCE1 pre-RNA splicing in the nucleus, the role of circPLCE1 in the cytoplasm remains unknown. Finally, the theoretical basis of this hypothesis is that pre-PLCE1, SRSF2 protein and circPLCE1 are located in proximity to the nucleus, indicating that the mechanism of action requires visual evidence under a microscope. 28 In summary, our study uncovered a novel mechanism underlying the direct regulation of PLCE1 mRNA expression through SRSF2dependent RNA splicing and provided a molecular basis for a new understanding of the pathophysiological function of circPLCE1. In addition, due to the critical role of PLCE1 in tumour progression, our findings may also pave the way for the pursuit of circPLCE1 as a potential target for CRC treatment.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author on reasonable request.