As one of the most common malignant tumors, CRC is characterized by high recurrence rate and poor prognosis, especially in those developed countries. It is the third most common cancer among males and ranks second among females [9, 10]. So far, there have been various methods applied to predict the biomarkers of CRC prognosis [11]. RBPs are capable to regulate mRNA stability and contribute to cancer-associated pathways [12]. In this paper, the RBPs of CRC were analyzed. Through a series of analysis, there were 12 marker genes related to the prognosis of CRC obtained.
Tudor domain-containing (TDRD) refers to a family of evolutionarily conserved proteins. In general, PIWI and TDRD proteins are recognized as the major influencing factors in piRNA biogenesis and germ cell development [13]. In this study, it was found out that methyl lysine-bound TDRDs are primarily involved in histone modification and chromatin remodeling, while methyl arginine-bound TDRDs are usually associated with RNA metabolism, alternative splicing, small RNA pathways, and germ cell development [14, 15]. TDRDs have now been detected in various cancers. TDRD9 is highly expressed in a subset of non-small cell lung carcinomas and derived cell lines by hypomethylation of its CpG island [16]. TDRD1 is closely associated with ERG over-expression in primary prostate cancer [17]. According to the findings of Jiang et al. [18] , 7 TDRD genes (PHF20L1, ARIB4B, SETDB1, LBR, TDRKH, TDRD10, and TDRD5) show high levels of amplification in more than 10% of TCGA breast cancers. In liver cancer, TDRD5 has been found to have a significant prognostic value. An early study revealed that TDRD5 is expressed in normal gastric and colonic mucosal tissues, suggesting the possibility that the TDRD5 gene is modified in CRC [19]. TDRD6 is capable to categorize irradiated prostate cancer patients into early and late relapse groups [20]. In addition, TDRD7 may play a certain role in the migration of tumor cells [21]. By analyzing CRCs, Mo et al. [22] have discovered not only frameshift mutations but also intratumoral heterogeneity of TDRD1, TDRD5 and TDRD9, which in combination might alter TDRD gene functions and make difference to the tumorigenesis of high microsatellite instability CRC. In our study, it was found out that TDRD5, TDRD6 and TDRD7 are differentially expressed in CRC, which makes it necessary to conduct a further study on the role of these three genes in colon cancer.
POP1 is referred to as a component of ribonuclease P, which is a ribonucleoprotein complex that generates mature tRNA molecules by cleaving their 5'-ends [23, 24]. In addition, it is a component of the MRP ribonuclease complex, which cleaves pre-rRNA sequences[25]. In the study, POP1 was found to be enriched in human prostate cancer cell lines [26], suggesting that it may be suitable as a potential marker for the diagnosis and prognosis of prostate cancer. Besides, it was found out that POP1 is up-regulated in CRC and applicable as a prognostic factor for CRC. Nevertheless, there is still no relevant research on the mechanism of POP1 in CRC, for which a further study is deemed necessary.
Kown as PGC1α, PPARGC1A is a transcriptional coactivator of genes encoding proteins responsible for the regulation of mitochondrial biogenesis and function[27]. D'Errico et al. [28] made a discovery that in the presence of Bax, the PGC1α-induced ROS accumulation represents one of the main apoptosis-driving factors in CRC cells. They also found out that PGC1α induces mitochondrial proliferation and activation in human intestinal cancer cells. [29]. Shin et al. [30] demonstrated that PGC-1α over-expression is effective in up-regulating the proliferation of HEK293 and CT26 cells. In addition, this expression exhibits a correlation with the enhancement of tumorigesis. In a case-control study, the heterozygous carriers of rs3774921 in PPARGC1A showed an increased risk of CRC [31]. PPARGC1A plays an essential role in the pathogenesis of colon cancer.
ZNF385A refers to a variety of RNA-binding protein that affects the localization and translation of a subset of mRNA. It binds the 3'-UTR of p53/TP53 mRNAs with ELAVL1 to control their nuclear export induced by CDKN2A. Thus, it has a potential to regulate p53/TP53 expression and mediate in part with the CDKN2A anti-proliferative activity[32]. The study led to a finding that ETV6-ZNF385A may be Acute lymphoblastic leukemia(ALL) new fusion gene [33]. Despite no ZNF385A involved in the pathogenesis of CRC, it may still play a certain role in CRC through the interaction with p53/TP53.
It is speculated that LRRFIP2 functions as activator of the canonical Wnt signaling pathway, which is associated with DVL3, the upstream of CTNNB1/beta-catenin. It regulates Toll-like receptor (TLR) signaling positively in response to agonist probably by competing with the negative FLII regulator for MYD88-binding, which plays a crucial role in the progression of colon cancer [34, 35]. In the study, it was found out that LRRFIP2 can be a candidate gene for the alternative splicing in colon and prostate cancer. There were three splice variants differing in their inclusion or skipping of exons 5 and/or 6 observed. Containing five predicted putative serine phosphorylation sites and one putative O-glycosylation site, these exons could modulate LRRFIP2 protein function [36]. As a familial hereditary disease, hereditary nonpolyposis CRC (Lynch syndrome) is mainly caused by DNA mismatch (mismatch repair). In Lynch syndrome, Morak and colleagues discovered a paracentric inversion on chromosome 3p22.2 between the DNA mismatch repair gene, MLH1, and the downstream LRRFIP2 gene transcribed in antisense direction. This contributes to two new stable fusion transcripts, thus removing MLH1 gene and protein function [37]. Another study was conducted on a Lynch Syndrome family to find out that the MLH1.ITGA9 fusion allele cased loss of heterozygosity(LOH) occurred to five genes, with LRRFIP2 included, which resulted in the loss of mismatch repair capabilities [38]. It can be known from above that LRRFIP2 may play a critical role in the pathogenesis of CRC.
CELF4 is responsible for encoding a sort of protein with three domains to bind RNA-recognition motif and to regulate pre-mRNA alternative splicing. In some research, it was found out that CELF4 was hypermethylated in endometrial cancer. Methylated CELF4 may be suitable for endometrial cancer screening of cervical scrapings [39]. For CELF4, it remains necessary to conduct a further research to figure out its role in tumors.
As a member of the Musashi, MSI2 belongs to the family of Drosophila melanogaster RNA binding proteins. It has been identified as a critical regulator of haematopoietic stem cell (HSC) self-renewal and fate determination[40, 41]. In this study, MSI2 was found to be a central component in an unappreciated oncogenic pathway to promote intestinal transformation via the PDK–AKT–mTORC1 axis[42]. MSI2 is highly expressed in a variety of cancers, including hepatocellular carcinoma and lung cancer[43, 44]. As for colon cancer cell lines, some studies have been carried out recently to suggest that both USP10 and MSI2 proteins are up-regulated. Besides, USP10 could stabilize the oncogenic factor MSI2 through deubiquitination [45]. It was also found out that its expression is up-regulated in CRC, which makes it applicable as a prognostic marker gene for CRC.
NOP14 refers to a stress-responsive gene as required for 18S rRNA maturation and 40S ribosome production[46]. As indicated by Zhou et al. [47] , the NOP14 in pancreatic cancer cells is capable to promote motility, proliferation and metastatic capacity. According to the findings of Du et al. [48] , NOP14 primed tumor invasion and metastasis by improving the stability of mutp53 mRNA. By inhibiting the Wnt/β-catenin pathways, NOP14 suppresses breast cancer[49]. In addition, NOP14 can reduce melanoma cell proliferation and metastasis by regulating the Wnt/b-catenin signaling pathway30484495[50]. It was found out in this study that the expression of NOP14 was upregulated in CRC, which means its pathogenesis requires further research and confirmation.
MRPS23 gene, which is responsible for encoding a 28S subunit protein, was found to be over-expressed in breast cancer[51], uterine cervical cancer [52], hepatocellular carcinoma [53], colorectal[54] and uterine leiomyoma [55]. As revealed by Gao et al. [56] , inhibiting MRPS23 could lead to a significant reduction in breast cancer metastasis by inhibiting EMT phenotype. Though the expression of MRPS23 is increased in CRC, its specific pathogenesis remains unclear.
MAK16 encodes a ribosomal protein and plays an important role in ribosome biogenesis throughout the cell cycle [57]. In this study, it was found out that the mutations in MAK16 can induce the arrest of the cell cycle at G1 phase, during which the cell synthesizes mRNA and proteins as part of the preparation for cell division[58]. At present, there is still no role of MAK16 in the pathogenesis of tumors, which requires a further research to confirm.
LIN28, a heterochronic developmentally regulated RNA-binding protein, was originally identified in mutation studies of the nematode Caenorhabditis elegans [59]. LIN28B is a homologue of LIN28, was overexpressed in many cancer types [60, 61]. King et al. [62] found out that LIN28B over-expression is associated with the reduced survival time and increased probability of tumor recurrence for patients. The constitutive LIN28B expression promotes not only tumorigenesis, but also the induction of LGR5 and PROM1 in colonic epithelial cell[63]. In addition, LIN28B could promote the proliferation, clone formation and tumorigenesis of colon cancer cells by increasing BCL-2 expression [64]. It thus can be known that LIN28B is highly expressed in CRC and plays an important role in its pathogenesis. It was found suitable as a target gene for CRC prognosis.
In this paper, a discussion was conducted about the role of 12 genes in tumors. Though some genes were found irrelevant to the pathogenesis of CRC, their biological functions and changes in their expression in CRC suggest that they may play a role in CRC to some extent, which requires a lot of experiments to be conducted for verification. This is also the limitation that our study is subject to. It is hoped that more research can be done to explore the pathogenesis of CRC.