Introduction

Colorectal cancer (CRC) accounts for nearly one-tenth of all cancers and is the fourth leading cause of cancer-related death worldwide [1]. Deregulated DNA methylation is one of the hallmark events in colorectal carcinogenesis, characterized by global hypomethylation of the genome and paradoxical hypermethylation of CpG islands [2]. The former is thought to influence CRC development by inducing chromosomal instability, while the latter could result in transcriptional silencing of tumor suppressor genes [2]. Hypermethylation of the CpG islands around promoters of specific suppressor genes, predominantly maintained by the DNA methyltransferase 1 (DNMT1), can be stably inherited for multiple generations in tumor cells and is involved in the process of carcinogenesis and progression of CRC [3, 4]. Besides, approaches for convenient and reliable detection of methylation with small amounts of DNA are available now [5]. These characteristics make DNA methylation as favorable CRC molecular marker with significant clinical value. Accordingly, extensive efforts have been made for comprehensive assessment of aberrantly methylated genes in CRC, which may not only improve our understanding of the epigenetic regulation of the disease but also identify CRC-related suppressor genes that might influence clinical management of patients [6].

UNC-5 family members, including four homologs (UNC5A-D), were identified as receptors of netrin-1 [7]. Due to their ability to trigger apoptosis in the absence of netrin-1, UNC5 receptors have been reported to function as dependence receptors [8]. For instance, apoptosis induced by UNC5 receptors in normal colonic epithelium is precisely regulated by varied expression of netrin-1, with highest expression in the crypts and lowest expression in the upper portion of the villi [9]. This indicated that the system may play essential roles in maintaining the normal intestinal epithelial microenvironment, and its de-regulation might promote the formation of hyperplasia, adenoma, or adenocarcinoma [10]. Therefore, UNC5 receptors have been proposed as putative tumr suppressor genes, with down-regulated expression in a variety of human malignancies due to genetic and epigenetic alterations [8]. We previously showed that loss of heterozygosis and DNA methylation contributed to the inactivation of UNC5C [11] and UNC5D [12] in renal cell carcinoma. Moreover, down-regulated expression of UNC5D in prostate cancer due to the hypermethylation of promoter was involved in the distant metastasis of the disease [13]. And increasing attention has been focused on the repression of UNC5 receptors and their biological function in human malignancies.

However, apart from UNC5C, few studies have focused on methylation-mediated inhibition of UNC5 receptors and the associated clinical significance in human colorectal cancer [14,15,16,17]. Although loss of expression of UNC5C due to epigenetic alterations has been observed in human colorectal cancer [14,15,16,17], its clinical significance in CRC diagnosis or prognosis remains to be evaluated. Considering their importance in the maintenance of intestinal epithelial homeostasis, adequate assessment of the methylation-mediated repression of UNC5 receptors in CRC may have important clinical applications. In the present study, a comprehensive assessment of methylation-mediated repression of UNC5 receptors was performed. The methylation alterations of receptors with clinical significance were assessed quantitatively, and their potential clinical values in diagnosis and prognosis of CRC were also evaluated.

Materials and methods

CRC clinical specimens

Tumor and corresponding non-cancerous tissues were obtained from 59 CRC patients, who underwent surgery at Tianjin Medical University Cancer Institute and Hospital between January 2016 and April 2020. All procedures performed in studies involving human participants were approved by the Research Ethics Committee of Tianjin Medical University Cancer Institute and Hospital, and in accordance with the 1964 Helsinki Declaration ethical standards, and all specimens were collected following written informed consent. Inclusion criteria for study participants including CRC as primary tumor confirmed by pathologic examination, complete follow-up of patients, and deaths caused by tumors and related complications. And participants who do not meet any of the above criteria were excluded from this study. Disease status was assessed by serial CT scans and another diagnostic testing as needed. The follow-up time ranged from 6 to 50 months, and the cutoff date follow-up was 10 April 2020.

Cell lines and DNA methyltransferase inhibitor treatment

The human colon cancer cell lines SW480 and SW620 involved in this study were obtained from the Cell Bank of Chinese Academy of Medical Sciences (Beijing, China). For demethylation assays, cell lines were treated with 10 μM of 5-aza-2-deoxycytidine (5-Aza-dC, Sigma-Aldrich) for 3 days. DMEM medium containing 10% FBS (Gibco-BRL, Gaithersburg, MD, USA) and 1% penicillin/streptomycin, as well as 5-Aza-dC or 0.1% DMSO alone was changed every 24 h.

Reverse transcriptase-polymerase chain reaction (RT-PCR) and real-time PCR

Quantitative real-time PCR was performed as previously described [11]. All primers used in this study are shown in Additional file 1: Table S1.

Methylation analysis

Genomic DNA (500 ng) was bisulfite converted by EZ DNA-Methylation Gold kit [18] (Zymo Research, Irvine, CA, USA). Bisulfite genomic sequencing (BGS) analyses were conducted as described previously [19]. PCR for Methylight assays and the calculation for the percentage of methylated reference (PMR) was performed as described previously [20]. Pyrosequencing assays were performed by Saigon Biotech Co. Ltd., Shanghai (China). Sequence information for the primers and probes used is listed in Additional file 1: Table S1.

Bio information analysis

The CpG islands around promoter were predicted by Methprimer 2.0 (http://www.urogene.org/methprimer2/) [21]. TPM expression data of CRC and paired normal tissues were downloaded from UCSC Xena hubs (http://xena.ucsc.edu/) [22], and the expression and the methylation level of genes in CRC from TCGA database (http://cancergenome.nih.gov/) [23] were analyzed using R (version 4.0.1). Whole genes methylation were generated using the Methylation Plotter web tool (http://maplab.imppc.org/methylation_plotter/) [24].

Statistical analysis

Statistical significance was determined with the nonparametric Mann–Whitney U-test (differences between groups), paired Student's t-test (differences between CRC and paired noncancerous tissues). Nonparametric Spearman's correlation coefficients method was used to evaluate the association between methylation or expression levels. Receiver operating characteristic (ROC) curves were generated to assess diagnostic efficiency. Overall and progress survival was analyzed using the Kaplan–Meier product-limit method and log-rank test. All of these statistical analyses were performed with SPSS 23.0 software (SPSS Inc., Chicago, IL, USA). Values of P ≤ 0.05 were considered statistically significant.

Result

Whole gene methylation analysis of UNC5 receptors in TCGA dataset of CRC

To gain insight into the overall methylation alterations of UNC5 receptors in CRC, we performed whole gene methylation analysis using the web tools Methylation Plotter. Based on the 450 K array data in TCGA database, the mean methylation levels (beta values) of CpG sites across each member in CRC and paired normal tissues were calculated. As shown in Fig. 1, UNC5A exhibited some degree of changes, but with low methylation level around TSS (Fig. 1A). As for UNC5B, nearly no methylation alteration was observed, and CpG sites around TSS remained hypomethylated in both normal and tumor tissues (Fig. 1B). However, typical methylation changes of tumor suppressor genes in CRC were observed in UNC5C and UNC5D, that was, global hypomethylation of the whole gene and specific hypermethylation of CpG islands around the TSS (Fig. 1C, D). These results indicated that methylation may exert different regulatory effects in the expression of UNC5 receptors.

Fig. 1
figure 1

Whole gene methylation analysis of UNC5 receptors in TCGA dataset of CRC. Whole gene methylation analysis of UNC5 receptors in TCGA dataset of CRC using Methylation Plotter. Based on the 450 K array data in TCGA database, the mean methylation levels (beta values) of CpG sites across each gene in CRC (red) and paired normal tissues (blue) were calculated. Representation of individual receptor, A UNC5A, B UNC5B, C UNC5C, D UNC5D. Chromosome location of each receptor was as indicated. TSS, transcription start site

Full size image

Promoter methylation and expression of UNC5 receptors in TCGA dataset of CRC

Next, we focused our analysis on the methylation alterations in the CpG islands around the promoter regions of UNC5 receptors. The mean beta values of corresponding CpG sites in promoter region for every patient from TCGA database were shown in Fig. 2A, B. Although the beta values of UNC5A were elevated in CRC tissues, the mean value in CRC tissues was only about 0.22. And the mean beta values of UNC5B were at a very low level, about 0.10, in both normal and tumor tissues (Fig. 2A, B). On the contrary, the methylation levels of UNC5C and UNC5D were elevated in CRC, with mean beta values of 0.48 and 0.68, respectively (Fig. 2A, B). It is widely recognized that probes with beta values below or equal to 0.2 were classified as unmethylated, between 0.2 and 0.8 as intermediate, and higher or equal to 0.8 as methylated [25, 26]. Thus, among the members of this family, promoter regions UNC5C and UNC5D were hypermethylated.

Fig. 2
figure 2

Promoter methylation and expression of UNC5 receptors in TCGA dataset of CRC. A promoter methylation analysis of UNC5 receptors in TCGA dataset of CRC using Methylation Plotter. B scatter plots of the mean promoter methylation levels of UNC5 receptors in TCGA dataset of CRC. Each point represented the mean methylation levels (beta values) of promoter CpG sites in CRC (red) or paired normal tissues (blue). C Comparison of TPM expression of each receptor between normal and tumor tissues of CRC. D The association between expression and methylation of each receptor

Full size image

Differential expression analysis of UNC5 receptors between CRC and paired adjacent noncancerous tissues in the TCGA database were further performed. As shown in Fig. 2C, UNC5A was observed with some degree of elevated expression, and expression of UNC5B was not apparently altered in CRC. Conversely, expression of UNC5C and UNC5D was down-regulated nearly tenfold in CRC, respectively. Further correlation analysis between promoter methylation and gene expression levels indicated that expression of UNC5C and UNC5D, but not UNC5CA and UNC5B, were highly correlated with the methylation levels (Fig. 2D). These observations with data from TCGA database implicated UNC5C and UNC5D as methylation-driven genes, with expression levels significantly affected by DNA methylation events.

Verification of methylation-mediated repression of UNC5 receptors in CRC cells

Next, quantitative detection methods were established to verify the promoter methylation-mediated repression of UNC5 receptors in CRC cell lines. As shown in Fig. 3A, bioinformatics analysis revealed that the UNC5 receptor genes harbor typical DNA sequence fulfilling the criteria for CpG islands around the promoter region. And oligonucleotides of four receptors for BSP and methylight detection were designed accordingly. To demonstrate that CpG methylation was functionally associated with expression of UNC5 receptors, CRC cell lines SW480 and SW620 were treated with the DNA demethylation reagent 5-azadC. The conversion effect of 5-azadC was confirmed by BGS analysis (Fig. 3B). As shown in Fig. 3C, such treatment exerted little effects on the expression of UNC5A and UNC5B but restored the expression UNC5C and UNC5D. UNC5B, with little alterations in methylation and expression in CRC and was left out of subsequent analyses.

Fig. 3
figure 3

Verification of methylation-mediated repression of UNC5 receptors in CRC cells. A detection methods established for methylation evaluation. CpG islands spanning the promoter region of UNC5 receptors. Horizontal bars, CpG sites; detection area for BGS and methylight are indicated. TSS, transcription start site; BSP, bisulfite sequencing primers. B BGS results of UNC5 promoters in CRC cell lines treated with 5-aza-2′-deoxycytidine; Circles, CpG sites analyzed; row of circles, individual promoter allele cloned and sequenced; filled circle, methylated CpG sites; open circle, unmethylated CpG site. C, methylight results of UNC5 promoters in CRC cell lines treated with 5-aza-2′-deoxycytidine. Real-time PCR results were displayed correspondingly

Full size image

Methylation-mediated repression of UNC5 receptors in clinical samples of CRC

The study was extended to primary tumors of CRC. To assess the performance of the methylight assay in clinical samples, pyrosequencing assay for UNC5 receptors was also performed in randomly selected paired specimens. Elevated methylation of UNC5A, UNC5C, and UNC5D in CRC was confirmed by both pyrosequencing and assay methylight assays (Fig. 4A, B). Differential methylation status of the three UNC5 receptors in paired CRC samples was verified in both assays (Fig. 4C). The results of methylight assays in 59 paired samples showed markedly increased methylation of the UNC5 receptors in tumor in contrast to that in the adjacent non-cancerous tissues (Fig. 4D). Among those samples, mRNA of 21 pairs was available (Fig. 4E). Expression of UNC5C and UNC5D, but not UNC5A, were significantly associated with methylation levels of promoter (Fig. 4F), which was consistent with the above results.

Fig. 4
figure 4

Methylation-mediated repression of UNC5 receptors in clinical samples of CRC. A Pyrosequencing results of UNC5A, UNC5C and UNC5D in randomly selected paired CRC specimens. B The corresponding methylight results of UNC5A, UNC5C and UNC5D in the selected CRC specimens. C Comparison of results from pyrosequencing and methylight assay. D Scatterplots of the methylation levels of UNC5A, UNC5C and UNC5D detected by methylight in 59 paired CRC specimens. E Scatterplots of the relative expression levels of UNC5A, UNC5C and UNC5D detected by real-time PCR in 21 paired CRC specimens. F The correlation between the expression of and the promoter methylation levels of UNC5A, UNC5C and UNC5D in 21 paired CRC specimens

Full size image

The clinical significance of methylation alterations of UNC5 receptors in CRC

SEPT9 is known to be highly methylated in CRC tissues and methylated SEPT9 in blood has diagnostic value in CRC patients [27], and was included in this study as a hyper-methylated gene control in CRC tissues. As was shown in Fig. 5A, the methylation of SEPT9 in CRC tissues was significantly higher than the adjacent non-cancerous tissues. Moreover, methylation alterations of UNC5C and UNC5D, but not UNC5A, were correlated with SEPT9 to some extent (Fig. 5B). The receiver operating curve (ROC) analysis was performed to evaluate the related discrimination efficiency. To discriminate adjacent non-cancerous tissues from CRC tissues, the performances of UNC5A, UNC5C and UNC5D were comparable, slightly lower than that of SEPT9 (Fig. 5C).

Fig. 5
figure 5

The clinical significance of methylation alterations of UNC5 receptors in CRC. A scatterplots of the methylation levels of SEPT9, a well-known highly methylated gene in CRC tissues. B The correlation between the methylation levels of UNC5A, UNC5C, UNC5D and SEPT9 in CRC specimens. C ROC analysis was performed to evaluate the performances of methylation alterations of UNC5A, UNC5C, UNC5D and SEPT9 in discriminating adjacent non-cancerous tissues from CRC tissues. Kaplan–Meier analysis were performed for progression-free survival (D) and overall survival (E) in CRC patients with different methylation levels of UNC5A, UNC5C, UNC5D and SEPT9

Full size image

Finally, we evaluated the prognostic value of methylation changes of UNC5 receptors. Kaplan–Meier curve survival analysis showed that higher promoter methylation levels of UNC5A and UNC5C were significantly associated with poor progression-free survival (log-rank, P = 0.031 and = 0.0075, respectively, Fig. 5D). For UNC5D and SEPT9, a trend toward poorer progression-free survival was seen, although failed to reach statistical significance (log-rank, P = 0.074 and = 0.14, respectively; Fig. 5D). In addition to SEPT9, promoter methylation of UNC5C and UNC5D showed an association with poor overall survival (log-rank, P = 0.016, 0.016, and 0.0049, respectively; Fig. 5E). In Cox proportional hazards analyses of progression-free survival and overall survival, promoter methylation of UNC5C and UNC5D were associated with unfavorable patient outcomes (Table 1). And HRs did not substantially change after adjustment for TNM stage. These results indicated that the promoter methylation of UNC5C and UNC5D were independent predictors of CRC survival.

Table 1 Univariate and multivariate of Cox proportional hazard modeling of factors associated with survival of CRC patients
Full size table

Discussion

Aberrant methylation in colorectal cancer, particularly on the CpG islands of tumor suppressor genes, is thought to be involved in disease development and progression [3, 4]. In colorectal carcinogenesis, some gene-specific hypermethylation events have been reported in precancerous lesions from CRC patients [28]. Moreover, the suppressive effect of aberrant DNA methylation on suppressor genes involved in adenoma to cancer progression has also been demonstrated [6]. These findings suggest that methylation alterations may be promising biomarkers in the etiological study of colorectal cancer. A better understanding of when these epigenetic changes occur and how they are involved in colorectal progression may be of great value in the risk assessment and clinical management of this disease. In this study, we focused on UNC5 receptors, which play vital roles in the maintenance of normal intestinal epithelial microenvironmental homeostasis.

Netrin-1-UNC5 receptors signaling axis has been reported to play a key role in neuronal navigation during the development of nervous system [29]. Recently, it has also been shown to regulate diverse processes in a number of non-neuronal tissues, especially in the regulation of tumorigenesis [29]. Due to remarkable structural homology and functional similarities, coupled with the down-regulated expression in various human malignancies, the boundaries between UNC5 receptors appear to be blurred as if they share exactly the same tumor biological significance. However, this proved not to be the case. In the present study, a comprehensive assessment of methylation-mediated repression of UNC5 receptors was performed in CRC. Significant variations were observed among the four receptors. Significantly elevated methylation in CRC was observed in UNC5A, UNC5C, and UNC5D, but not in UNC5B. UNC5B was hypomethylated in both CRC and adjacent normal tissues, which was consistent with the previous study [30]. Moreover, although with elevated methylation in CRC tissues, methylation-mediated inhibition did not occur in UNC5A, but only in UNC5C and UNC5D.

Numerous studies have provided ample evidence that UNC5C, a tumor suppressor, is down-regulated in a large fraction of colorectal malignancies, primarily through promoter methylation [14,15,16,17]. Recently, TCGA datasets analysis involved methylation of UNC5C has been performed in two studies [31, 32]. However, neither of them validated the methylation alterations of UNC5C by cost-effective quantitative approaches with possible clinical translation values. In contrast, quantitative results from the current study with our own specimen cohort indicated that hypermethylation of UNC5C was an independent predictor of CRC survival. Different from UNC5C, UNC5D is the most recently identified member of UNC5 family [33]. Consistent with other members of this family, UNC5D has also been characterized as a tumor suppressor gene for several cancers, such as neuroblastoma [34, 35], renal cell carcinoma [12], prostate cancer [13], and bladder cancer [36]. Recently, Uhan et al. evaluated the methylation status of probe cg13561879 located in the promoter region of UNC5D by methylation-specific high-resolution melting analysis (MS-HRM), and found that hypermethylation of this probe could be served as a potential novel diagnostic biomarker for colorectal cancer, but with no significant prognostic value [37]. Methylight, the quantitative method adopted in this study, relies on methylation-specific primers and probes [38]. As for UNC5D, 9 CpG sites in total were overlapped by the probe (4 sites) and primers (5 sites). In contrast, because more sites are covered, the methylight assay can reflect the methylation status of CpG islands much more accurately than the MS-HRM assay, which may partially explain the difference in Cox proportional hazards regression results for overall survival.

Some limitations of this study should also be noted. The conclusion that hypermethylation of UNC5C and UNC5D has diagnostic and prognostic values in CRC was drawn from a relatively small CRC cohort. A large cohort should be used to verify the clinical significance of methylated UNC5 receptors for patients with CRC. Another limitation of this study is the lack of control group with benign colorectal polyps or adenomas. Implementation of this work with an enriched control group will further elucidate the applicability of the methylation assay for UNC5 receptors. In addition, recent work has provided ample evidence that genome-wide DNA methylation is altered in aging and age-related diseases [39]. However, due to the limited number of cases included in this study, the potential age-related differences in the methylation of UNC5 receptors were not considered. Finally, UNC5C, UNC5D, and SEPT9 were observed with similar methylation patterns in CRC, which suggests that elevated methylation of UNC5C and UNC5D may have the same clinical applications as SEPT9. As is known that the methylated SEPT9 (mSEPT9) assay was the first blood-based methylation test approved by the United States Food and Drug Administration for colorectal screening. Quantitative mSEPT9 levels have been successfully applied for the screening, diagnosis, monitoring, and prognosis of CRC [40,41,42]. The potential of methylated UNC5C or UNC5D in circulating tumor DNA as biomarker for CRC should also be taken into consideration in future study.

Conclusions

In summary, this is the first study that comprehensively evaluates the aberrantly methylation of UNC5 receptors and the associated clinical significance in human colorectal cancer. Except for UNC5B, UNC5A, UNC5C and UNC5D were observed with significantly increased methylation levels in the promoter region. Methylation-mediated repression was observed in UNC5C and UNC5D, but not in UNC5A and UNC5B in CRC. A cost-effective and PCR-based quantitative DNA methylation detection approach with high sensitivity, methylight, was employed for quantitative methylation evaluation. And it was explored that hypermethylation of UNC5A, UNC5C, and UNC5D was associated with poor progression-free and overall survival. And methylation levels of UNC5C and UNC5D were independent predictors of CRC survival. Our findings proposed that hypermethylation of UNC5 receptors could be potential diagnostic and prognostic biomarkers for CRC.