circMSH3 is a potential biomarker for the diagnosis of colorectal cancer and affects the distant metastasis of colorectal cancer

Specimen source

Tumor and paracancerous tissues were collected from 110 patients with colon cancer who underwent surgical resection in the Central Hospital of Panyu District, Guangzhou. Prior to this, none of the patients had received neoadjuvant chemotherapy, radiotherapy, or other antitumor therapy, and all tissue samples were pathologically confirmed. All patients or their families signed an informed consent form to participate. The study was approved by the Ethics Committee of the Guangzhou Panyu District Central Hospital, approval number PYRC-2023-017.

Cell culture

CRC cell lines (HCoEpiC,HCT15,SW116,RKO,Colo205,HCT116,Caco2,Lovo, HT-29,SW480,SW620 and DLD-1) were purchased from the Shanghai Institute of Cell Biology (Shanghai, China). The cell lines were grown in Dulbecco’s modified Eagle’s medium/F12 medium containing 10% fetal bovine serum and maintained at 37 °C in a humidified incubator with 5% CO₂. The medium was changed every other day. The cells were digested with 0.25% trypsin and subcultured, and cells with logarithmic growth and 95% viability were selected for further experiments (Itoh et al., 2002; Bachmayr-Heyda et al., 2015; Lu et al., 2017; Xu et al., 2020; Li & Li, 2021).

circRNA sequencing (circRNA-seq)

Selected 10 pairs of colorectal cancer patients’ cancer tissue and paired paracancerous tissue samples, among which five patients had tumor metastasis, and the other 5 patients had no tumor metastasis. Total RNA was isolated using a HiPure Total RNA Mini Kit (Magen Biotechnology, Guangzhou, China) following the manufacturer’s protocol. RNase R (Epicentre Biotechnologies, Madison, WI, USA) was used to digest the total RNA for 1 h after DNase I processing at 37 °C. After quantifying the sample RNA, the RNA library was prepared using a KAPA RNA Hyperprep Kit with Riborase (HMR) (Illumina, San Diego, CA, USA). Sequencing was performed using a Hiseq X10 system (Illumina) in PE150 mode. For analysis of circRNA-seq, sequencing readings were mapped to the Hg19 reference genome using STAR software (Dobin et al., 2013), and circRNA was identified using DCC tools (Cheng, Metge & Dieterich, 2016). The statistical power of this experimental design, calculated in RNASeq Power is 0.89. circRNAs that were differentially expressed in CRC were identified using the edgeR tools for screening of differentially expressed circRNA (Robinson, McCarthy & Smyth, 2010). Criteria for screening were a P-value of 0.05 and a fold change of ≥—2—. Differentially expressed circRNAs were visualized using differential distribution volcano plots and differential circRNA cluster analysis.

Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was extracted from tissues and cells using TRIzol. RNA was reverse transcribed into cDNA using an EasyScript First-Strand cDNA Synthesis SuperMix Kit. PCR was performed with a SYBR Green qPCR SuperMix Kit using the primers shown in Table 1 under the following conditions: 95 °C for 5 min; 40 cycles of 95 °C for 15 s and 60 °C for 32 s for plate reading; melting curve analysis temperature was 60 °C-−95 °C. The relative expression of genes was calculated according to the 2^−△△CT formula.

RNA-protein pull-down

An Pierce™ Magnetic RNA-Protein Pull-Down Kit (20164; Thermo Fisher Scientific, Waltham, MA, USA) was used to detect interactions between circMSH3 and proteins. The circMSH3 probe solution (1 µg) was denatured at 90 °C for 2 min and incubated with pre-cooled RNA structure buffer to form RNA secondary structures. Next, streptavidin magnetic beads were incubated with the mixture at 25 °C for 30 min. The cells were lysed at 4 °C and centrifuged at 15,000× g for 15 min, and the supernatant was collected and mixed with the probe-bead mixture. An RNase inhibitor (5 µL) and poly(dI⋅ dC) (5 µL) were added, followed by incubation at 25 °C for 2 h and elution at 37 °C for 2 h. The supernatant was transferred into a new tube and used for protein mass spectrometry and western blotting. circMSH3 probe sequence were designed and generated by Ribo Bio (Guangzhou, China).

circMSH3: 5′-Biotin-UCCUUUCGACUCGAAUUCUG-3′;

Negative Control (NC): 5′-Biotin-UUCUCCGAACGUGUCACGUTT-3′.

Dual-luciferase reporter assay

Wild-type or mutant circMSH3 fragments were constructed and inserted downstream of the luciferase reporter gene in the pmirGLO plasmid (Promega, Madison, WI, USA). We transfected reporter plasmids into 293T cells using Lipofectamine 3000, followed by hsa-miR-409-3p, hsa-miR-942-5p, hsa-miR-1276, hsa-miR-513a-5p, and hsa-miR. The -7151-5p mimic was co-transfected with the reporter gene into 293T cells and the luciferase activity was measured 48 h later using a dual-luciferase reporter kit (Promega, Madison, WI, USA). The relative luciferase activity was normalized to the Renilla luciferase internal control.

Statistical analysis

IBM SPSS statistics 21.0 software (Chicago, IL, USA) was used for statistical analysis, GraphPad Prism 8.0 software (GraphPad Software, La Jolla, CA, USA) was used for graphing, and values were expressed as the mean ±standard deviation. The expression levels of circMSH3 in CRC and adjacent tissues were compared by a paired t-test. The relationship between the expression levels of circMSH3 in CRC tissue and the clinicopathological characteristics of the patients was compared between two groups using an independent samples t-test, and the comparison between multiple groups was performed using one-way analysis of variance. P-values <0.05 were considered statistically significant. The diagnostic performance of circMSH3 was analyzed using receiver operating characteristic (ROC) curves.

Screening for differentially expressed circRNAs

In order to explore the expression characteristics of circRNAs in colorectal cancer tissues, RNA-seq results showed that most circRNAs originated from the exon region, chromosome 1-3 formed a large number of circRNAs, and the length of circRNA sequences was mainly distributed in 200 bp–500 bp (Figs. 1A–1C). circRNAs that were differentially expressed between teen pairs of metastatic and non-metastatic CRC groups were screened by high-throughput sequencing technology (Supplementary Materials 1). The results showed that compared with the non-metastatic CRC group, there were 87 upregulated circRNAs and 100 downregulated circRNAs in the metastatic CRC group (Fig. 1D). In the adjacent group, 176 circRNAs were up-regulated and 174 circRNAs were downregulated in the metastatic CRC group (Fig. 1E). Compared with the non-metastatic CRC adjacent group, more circRNAs were differentially expressed in the non-metastatic CRC group: the expression of 420 circRNAs was upregulated, and the expression of 616 circRNAs was downregulated (Fig. 1F). Taking the intersection of each group, there were 12 circRNAs that were differentially expressed in the tumor group, the paracancer group (N), and the non-metastatic and metastatic CRC groups (hsa_circ_0057123, hsa_circ_0000707, hsa_circ_0005704, hsa_circ_0004456, hsa_circ_0005379, hsa_circ_0130312, hsa_circ_0006672, hsa_circ_0001917, hsa_circ_0087641, hsa_circ_0003855, hsa_circ_0018992, and hsa_circ_0008230). At the same time, two circRNAs were differentially expressed between the metastatic CRC, non-metastatic CRC, and CRC adjacent groups (hsa_circ_0003761 [circMSH3] and hsa_circ_0133953) (Fig. 2A). hsa_circ_0133953 was discarded because its sequence was too long, and finally circMSH3 was selected for further research.

Verification of circMSH3 cyclability

circMSH3 was derived from exons 9, 10, 11, and 12 of the MSH3 gene with a length of 423 bp. To verify the circularity of circMSH3, we performed reverse Sanger sequencing of the PCR product. The sequencing results were consistent with the sequence of the circMSH3 circularization site in the CircBase database (Figs. 2B, 2C).

circMSH3 is a potential biomarker for the diagnosis of CRC

In 110 pairs of CRC and adjacent tissues, the relative expression of circMSH3 was significantly higher in CRC tissues (P < 0.001) (Fig. 3A). circMSH3 expression was significantly upregulated in the HCoEpiC epithelial cell line and colon cancer cell lines (HCT15,SW116,RKO,COLO205,HCT116,Caco2,LOVO,HT-29,SW480,SW620 and DLD-1), with the most significant upregulation in the SW480 cell line (Fig. 3B). circMSH3 and MSH3 gene expression was positively correlated in 30 CRC tissues, with a correlation coefficient of 0.64 (Fig. 3E). MSH3 gene and circMSH3 expression was upregulated in 30 CRC tissues (Figs. 3C–3D). Receiver Operating Characteristic curve (ROC) analysis showed that when circMSH3 was used for the diagnosis of CRC, its area under the curve (AUC) was 0.75, with a sensitivity and specificity of 70.9% and 66.4%, respectively (Fig. 3F). These results show that circMSH3 is a potential biomarker for the diagnosis of CRC.

circMSH3 expression affects the distant metastasis of CRC

Using the clinical data of 110 patients, we analyzed the relationships between the expression levels of circMSH3 and the patients’ sex, age, tumor size, tumor infiltration degree (T), lymph node metastasis (N), distant metastasis (M), TNM stage, and differentiation. The results showed that circMSH3 expression was significantly correlated with distant metastasis in CRC patients (P = 0.043) but not with sex (P = 0.105), age (P = 0.526), tumor size (P = 0.953), degree of invasion (P = 0.509), lymph node metastasis (P = 0.596), TMN stage (P = 0.416), or degree of differentiation (P = 0.492) (Table 2). The results showed that circMSH3 expression affects the distant metastasis of CRC.

circMSH3-binding proteins and microRNAs (miRNAs)

Many studies have shown that circRNAs can act by binding to proteins or miRNAs (Hansen et al., 2013; Zang, Lu & Xu, 2020; Okholm et al., 2020). To explore the proteins and miRNA molecules that circMSH3 can bind to, we performed an RNA–protein pull-down assay combined with mass spectrometry and bioinformatics prediction combined with a dual-luciferase reporter assay. Biotin-labeled circMSH3 probe-captured proteins were identified by high performance liquid spectrometry (Figs. 4A, 4B). The Metascape website (Zhou et al., 2019) was used to perform function enrichment analysis of proteins, which showed that circMSH3 could bind to a variety of proteins involved in RNA transport and biological processes related to the cell cycle and cell junctions (Fig. 4D). These results suggest that RNA-splicing proteins may be involved in the formation of circMSH3, and circMSH3 may also regulate cell fate by binding directly to proteins involved in the cell cycle and cell junctions. The miRNAs that circMSH3 might bind to were predicted according to the starBase database, and the predictions were verified by a dual-luciferase reporter gene experiment. The results showed that circMSH3 could bind directly to hsa-miR-409-3p, hsa-miR-942-5p, and hsa-miR-1276, whereas no binding signal was detected for hsa-miR-513a-5p and hsa-miR-7151-5p (Fig. 5).