Integrative analysis and validation of dysregulated long non‐coding RNAs in colon cancer

Abstract It is an increasing evidence that long non‐coding RNAs (lncRNAs) are involved in tumour initiation and progression. Here, we analysed RNA‐sequencing data from the Cancer Genome Atlas (TCGA) datasets. Totally, 1176lncRNAs, 245miRNAs and 2081mRNAs were identified to be differentially expressed (DE) in colon cancer tissues compared with normal tissues. CASC21, a novel lncRNA located in 8q24.21 locus, was significantly overexpressed in 30 colon cancer tissues compared with matched normal tissues by qRT‐PCR assay. CASC21 tended to higher expression as the increase of the tumour‐node‐metastasis (TNM) classification. Functionally, CASC21 promoted cell proliferation by regulating cell cycle and enhanced tumour metastasis by epithelial‐mesenchymal transition (EMT) in colon cancer. Mechanism study indicated that CASC21 might be involved in activating WNT/β‐catenin pathway in colon cancer. In addition, we also built a competing endogenous RNA (ceRNNA) network by bioinformatic analysis using TCGA datasets. Together, our results not only provide novel lncRNAs as potential candidates for further study but also prove that CASC21 is an oncogenic regulator through activating WNT/β‐catenin signalling in colon cancer.

metastasis including colon cancer. [3][4][5][6] Thus, increasing dysregulated lncRNAs have been demonstrated to be associated with prognosis of cancer patients and even predict clinical responses to cancer therapy.
In our previous reports, we proved that lncRNA KCNQ1OT1 is up-regulated in and serves as an oncogene while PGM5-AS1 is down-regulated and serves as a tumour suppressor in colorectal cancer.
It is reported that lncRNAs exert its function by diverse mechanisms depending on its cellular location. [7][8][9][10] The 8q24.21 locus harbours several cancer-related lncRNAs and few protein-coding genes. 11 The MYC oncogene, located in this region, contributes to the tumorigenesis in many human cancers including colon cancer. 12,13 Previous reports demonstrated that lncRNAs mapped to 8q24.21 locus paly potential roles by directly or indirectly interacting with MYC. For instance, CCAT1, CCAT2 and CASC11 are identified to be closely associated with MYC, 14,15 emphasizing the clinical significance of dysregulated lncRNAs in this region.
In this study, we first identified DElncRNAs and constructed a reliable ceRNA network for further research in colon cancer based on TCGA datasets using bioinformatic analysis. CASC21 is located on chromosome 8q24.21, and it is significantly up-regulated in colon cancer. However, the biological function of CASC21 in colon cancer remains unknown. 1. Screening of DE lncRNAs: We downloaded RNA-sequencing data from TCGA colon cancer datasets including 480 colon cancer and 41 normal tissues. The following exclusion criteria were applied for the selection process: (a) Histological diagnosis was not colon cancer, and (b) data were incomplete. Clinical survival data were also obtained from TCGA. OS was the time from tumour resection to death, loss-to-follow-up or study conclusion. R software and packages were used to analyse the data, and a gene with an absolutely log fold change (FC) ≥2 and false discovery rate (FDR) <0.01 was considered to be differentially expressed. We used the Kaplan-Meier method to analyse OS, and the log-rank test was used to analyse differences in OS P < .05 was regarded as statistical significance (*P < .05; **P < .01; ***P < .001).

| TCGA analysis
2. Gene set enrichment analysis (GSEA): RNA-sequencing data from TCGA datasets were analysed, and we divided CASC21 into two groups based on its expression. Then, GSEA was carried out by the GSEA software. The enrichment scores were compared with F I G U R E 1 Transcriptome landscape of colon cancer. A, Volcano plot of DElncRNAs identified from TCGA colon cancer datasets. (log fold change ≥2 and P < .01); B, The 50 most up-regulated and down-regulated DElncRNAs in colon cancer. C, Volcano plot of DEmiRNAs identified from TCGA colon cancer datasets (colon cancer tissues vs normal tissues, log fold change ≥2 and P < .01); D, Volcano plot of DEmRNAs identified from TCGA colon cancer datasets (colon cancer tissues vs normal tissues, log fold change ≥2 and P < .01) the enrichment results of 1000 random sequences of the gene set to assess statistical significance.
3. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed to analyse functions of DEmRNAs. Construction of ceRNA network: Firstly, DElncRNAs in colon cancer were identified as described above.

| Clinical samples and Cell lines
Five colon cancer cell lines (HT29, HCT116, SW480, SW620 and LOVO) were cultured according to the instructions recommended by American Type Culture Collection (ATCC). The colon epithelial cell line NCM460 was purchased from the Chinese Academy of Sciences (China) and cultured according to the manufacture's protocol. 30 paired of clinical tissues (colon cancer and corresponding normal tissues) and 20 colon cancer tissues from 50 patients were obtained from Nanjing Drum Hospital (Nanjing, China). All patients signed informed consent forms.

| RNA sequencing
RNA-sequencing assay was carried out to detect the expression profiles of lncRNAs and mRNAs using 20 colon cancer tissues at F I G U R E 2 CASC21 is significantly up-regulated in colon cancer. A, Relative expression of CASC21 in colon cancer tissues compared with normal tissues in TCGA datasets; B, The relative expression of CASC21 was detected by qRT-PCR using 30 paired of colon cancer and corresponding normal tissues, and GAPDH was used as internal control. Data were presented as -ΔCt values. (2 −ΔΔCt was used to calculate the relative expression of CASC21. ΔCt = Ct CASC21-Ct GAPDH, ΔΔCt = ΔCt(cancer tissues)-ΔCt(normal tissues)). C, CASC21 expression was analysed in 30 paired of colon cancer tissues and corresponding normal tissue, and data were presented as the fold change in tumour tissues compared with the matched normal tissues. D, Relative CASC21 expression was analysed in five colon cancer cell lines and one normal colon epithelial cell line, and normalized to GAPDH expression in colon epithelial cell line. (*P < .05, **P < .01) Sangon biotech. RNA labelling and hybridization array assays were performed according to the manufacturer's protocol. Libraries were controlled for quality and quantified using Hiseq 2500 system (Illumina).
For CASC21 stable knockdown, sh-CASC21 construct was generated by Shanghai GenePharma Co, and then, we transfected HCT116 and SW480 cells with sh-CASC21. Puromycin with a concentration of 1 μg/mL was used to select cells with CASC21 stable knockdown. After 7 days of filtration with puromycin, qRT-PCR was used to validate cells with CASC21 stable knockdown, and then, cells were cultured in RPMI 1640 (Hyclone) with 10% foetal bovine serum (FBS) and 0.5 μg/mL puromycin. The CASC21 shRNA sequences are as follows: 5′-GGTTGTTGCTTCCTAGTCT -3′.

| Cell proliferation assay
In CCK-8 assay, 3 × 10 3 cells were cultured in a 96-well plate for 24, 48, 72 and 96 hours. Then, each well added 10 μL CCK-8 solution and incubated for 2 hours. The absorbance was measured at 450 nm for each well.
In colony formation assay, cells were counted and 1 × 10 3 cells were cultured each well in a 6-well plate for 14 days. Then, colonies were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. Finally, the mean colony numbers were calculated.

| Transwell assay
Transwell assay (Corning) was used to detect the capacity of colon cancer cell invasion and migration. For cell invasion assay, cells

| Flow cytometric assay
For cell cycle assay, 1 × 10 6 cells were washed by normal saline (NS) and stained with propidium iodide and then incubated in the dark for 30 minutes. For cell apoptosis assay, 1 × 10 6 cells were stained with annexin V, fluorescein isothiocyanate (FITC) and incubated in the dark for 5 minutes. Samples were finally analysed by flow cytometry.  Note: P-value <.05 was considered statistically significant.
P-value <.05 or a borderline P-value was marked in bold.
Signals were visualized by ECL reagent.

| Immunofluorescence (IF) and Fluorescence in situ hybridization (FISH)
Cells were harvested and washed by NS and then fixed with 4% formaldehyde for 15 minutes. 0.3% Triton X-100 was used to permeate cells.
Cells were incubated with primer antibodies and then corresponding secondary antibodies. The primer antibodies were E-cadherin (CST) and N-cadherin (CST). Finally, DAPI was used to stain cellular nuclei.
For fluorescence in situ hybridization, HCT116 and SW480 cells were fixed and washed. Cells were permeated by 0.3% Triton X-100.
Subsequently, cells were incubated with 20 uM FISH Probe in hybridization buffer at 37°C overnight. Cells were washed by 4 × saline sodium citrate (SSC), 2 × SSC and 1 × SSC. Then, cellular nuclei were stained with DAPI. Finally, fluorescence was visualized with a microscope.

| Immunohistochemistry (IHC) assay
Tissue sections were incubated with the primary antibody targeting ki67 (Abcam). Then, they were incubated with corresponding secondary antibody.

| Animal experiments
We performed animal experiments according to the experimental animal use guidelines of the National Institutes of Health. Male 5-to 6-week-old BABL/c nude mice were purchased from the Experimental Animal Center of Nanjing Drum Tower Hospital. For the tumorigenicity assay, 1 × 10 6 cells were stably transfected with Sh-CASC21 or negative controls, and they were inoculated subcutaneously into the left groin region of mice. The tumour growth was measured every 3 days.
All mice were killed after 15 days, and tumour tissues were used to perform haematoxylin and eosin staining, IHC and TUNEL assays.
Tumour volumes = 0.5 × D (the longest diameter of the tumour) × d 2 (the longest diameter of the tumour). For the in vivo metastasis assay, 3 × 10 6 HCT116 cells stably expressing shCASC21 or negative controls were injected into nude mouse tail vein. These mice were killed after 3 weeks, and the number of metastatic foci was counted.

| Statistical analysis
We used R software and packages to analyse the RNA-sequencing data. SPSS 17.0 was used to data analysis. All data were presented as mean ± SD. Image-Pro Plus (IPP) software was used to perform quantitative analysis of photographs obtained from immunochem-

| Identification of dysregulated lncRNAs using RNA-sequencing data from TCGA colon cancer datasets
We identified 1176 DElncRNAs in colon cancer tissues compared with normal tissues using TCGA datasets ( Figure 1A, Table   S1). The 100 most dysregulated lncRNAs including 50 most upregulated and down-regulated lncRNAs were listed in Figure 1B.

| CASC21 is significantly up-regulated in colon cancer
Bioinformatic analysis of TCGA datasets showed that CASC21 expression was significantly up-regulated in colon cancer ( Table 1). However, no significant differences between CASC21 expression and other clinical pathological parameters were observed in this study due to the limited sample size. Interestingly, we observed CASC21 expression was slightly higher in tumours with mismatch-proficient (pMMR) than tumours with mismatch-deficient (dMMR), although the difference did not achieve statistical significance (P = .068, Table 1).
Meanwhile, in comparison with NCM460 cell line, CASC21 expression was obviously increased in five human colon cancer cell lines ( Figure 2D). Of them, HCT116 and SW480 cell lines, with relative higher expression of CASC21, were selected for further functional assays.

| CASC21 knockdown suppresses cell proliferation and induces cell apoptosis in colon cancer
As expected, CASC21 expression was down-regulated after CASC21 knockdown by transfecting SW480 and HCT116 cells with CASC21 siRNA ( Figure 3A,B). The colony formation assay indicated that CASC21 knockdown inhibited the ability of colony formation in both SW480 and HCT116 cells ( Figure 3C,D). CCK8 assay also confirmed that knockdown of CASC21 suppressed cell proliferation in SW480 and HCT116 cells ( Figure 3E,F). Cell proliferation assays including CCK8 and colony formation assays emphasized the possibility that CASC21 might be associated with tumour growth in colon cancer.
Apoptosis assays showed that SW480 and HCT116 cells transfected with CASC21siRNA had higher apoptotic rates than negative controls ( Figure 4A). The protein expression levels of cleaved Caspase-3 and bax were increased while bcl-2 was decreased in HCT116 and SW480 cells transfected with CASC21siRNA ( Figure 4B).
Furthermore, CASC21 knockdown significantly decreased the number of cells in S phase while increasing cells in G0/G1 phase compared with negative controls ( Figure 4C). Western blot assays showed that CASC21 knockdown markedly reduced the protein expression levels of CDK4, CDK6 and cyclin D1 ( Figure 4D). To further clarify the impact of CASC21 on cell proliferation, we performed the gene set enrichment analysis (GSEA). We found that CASC21 coexpressed with cell cycle-related genes and cell cycle pathway was significantly activated in CASC21 high expression group ( Figure 4E).
Furthermore, we also proved the positive relationships between the expression levels of CASC21 and co-expressed cell cycle-related genes including CCND1 and CDK4 by RNA sequencing using 20 colon cancer tissues ( Figure 4F). We uploaded the results of RNAsequencing analysis in Table S3.

| CASC21 promotes cell migration and invasion by activating WNT/β-catenin signalling in colon cancer
Since CASC21 expression was increased in patients with III-IV stage, we investigated whether CASC21 affected cell invasion and and HCT116 cells ( Figure 5A,B). Immunofluorescence assay showed that CASC21 silence decreased N-cadherin expression while increasing E-cadherin expression ( Figure 5C). Similar results from Western blot assay were presented in Figure 5D. To investigate the potential mechanism of CASC21, we detected the subcellular location of CASC21. According to FISH assay, CASC21 mainly located in cytoplasm compared with nuclear ( Figure 6A). Thus, we hypothesized CASC21 might exert its function by modulating mRNA. It is reported that a lncRNA can regulate the expression levels of genes located around it. CASC21 was located around MYC. In this study, we found a positive relationship between CASC21 and MYC expres-

| CASC21 knockdown inhibits colon cancer cell tumorigenesis and metastasis in vivo
To elucidate whether CASC21 promotes tumour growth and metastasis in vivo, we established tumorigenicity and metastasis models, respectively. We observed that the average size and weight of tumours were significantly decreased in sh-CASC21 group than negative control group ( Figure 7A,B). IHC assay showed that Ki-67 expression was decreased in sh-CASC21 tumours compared with control tumours (Figure 7C), indicating CASC21 knockdown inhibited tumour growth in vivo. Moreover, TUNEL staining test showed tumours in sh-CASC21 group exhibited more apoptotic cells ( Figure 7D). Besides, we assessed the role of CASC21 in affecting tumour metastasis and found the number of hepatic metastatic nodules was decreased in sh-CASC21 tumours than negative controls ( Figure 7E). Taken together, we suggest that CASC21 promotes colon cancer growth and metastasis in vivo.

| D ISCUSS I ON
Recent studies have revealed that lncRNAs are crucial regulators in tumour initiation and metastasis by multiple mechanisms. 16 WNT/β-catenin signalling abnormal activated is one of the majority causes leading to various human cancers including colon cancer. 19,20 WNT signalling also induces EMT process and finally promotes tumour metastasis. 21 As mentioned previously, CCAT2, located at 8.24.21, was proved to be involved in tumorigenicity through activating WNT/β-catenin signalling. We observed CASC21 located around CCAT2; thus, we speculated that CSAC21 might have the similar biological function to CCAT2 by activating WNT/β-catenin signalling. In this study, we found CASC21 knockdown obviously decreased the protein expression levels of WNT/β-catenin signalling targets. GSK3β plays a crucial role in regulating β-catenin. Our findings indicated CASC21 knockdown increased the GSK3β expression.
These results confirm that CASC21 silence may increase GSK3β expression and then reduce WNT/β-catenin signalling activation.
Up to now, the detailed mechanisms of dysregulated lncRNAs in cancer have not been interpreted completely. Here, we attempted to predict the molecular mechanisms of DElncRNAs using the ceRNA network and provided possible approaches for experimental verification. In our previous work, we have demonstrated lncRNA KCNQ1OT1 promoted EMT in colorectal cancer (CRC) by sponging miRNA217, 22 and the KCNQ1OT1-miRNA217 axis was included in the ceRNA network we built in this study.
In conclusion, our work first proves CASC21 may serve as an oncogene by activating WNT/β-catenin in colon cancer. Furthermore, we provide a great number of DElncRNAs and a ceRNA network for further study in colon cancer. Although CASC21 promotes colon cancer growth and metastasis by activating WNT/β-catenin signalling, there are more experiments required to clarify it clearly.

ACK N OWLED G EM ENTS
We thank TCGA database for sharing the RNA-sequencing datasets.
This work was supported by grants from Jiangsu Provincial Natural Science Foundation (BK20161107), Health and Family Planning Commission development programme of Jiangsu (H2017042) and Health and Family Planning Commission key programme of Nanjing (ZKX17012). We thank Min Lin for her support.

CO N FLI C T S O F I NTE R E S T S
No conflict of interests.

AUTH O R S' CO NTR I B UTI O N S
Xiaoping Qian and Baorui Liu designed the study. Qun Zhang and Yinzhu Bian prepared manuscript drafting and performed the in vitro experiments. Yiping Zhu and Li Wan contributed to the in vivo experiments. All authors reviewed and revised this manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in the supplementary materials section.