The identification and validation of EphA7 hypermethylation, a novel biomarker, in cervical cancer

Aberrant methylation of EphA7 has been reported in the process of carcinogenesis but not in cervical cancer. Therefore, an integration study was performed to explore the association between EphA7 hypermethylation and cervical cancer and validate the potential value of EphA7 hypermethylation in the diagnosis of cervical cancer. We performed an integration study to identify and validate the association between EphA7 methylation and cervical cancer. First, data on EphA7 methylation and expression in cervical cancer were extracted and analyzed via bioinformatics tools. Subsequently, CRISPR-based methylation perturbation tools (dCas9-Tet1/DNMT3a) were constructed to further demonstrate the association between DNA methylation and EphA7 expression. Ultimately, the clinical value of EphA7 methylation in cervical cancer was validated in cervical tissues and Thinprep cytologic test (TCT) samples by methylation-specific PCR (MSP) and quantitative methylation-specific PCR (QMSP), respectively. Pooled analysis showed that EphA7 promoter methylation levels were significantly increased in cervical cancer compared to normal tissues (P < 0.001) and negatively correlated with EphA7 expression. These prediction results were subsequently confirmed in cell lines; moreover, CRISPR-based methylation perturbation tools (dCas9-Tet1/DNMT3a) demonstrated that DNA methylation participates in the regulation of EphA7 expression directly. Consistent with these findings, the methylation level and the positive rate of EphA7 gradually increased with severity from normal to cancer stages in TCT samples (P < 0.01). EphA7 hypermethylation is present in cervical cancer and is a potential biomarker for the diagnosis of cervical cancer.

Eph receptors represent the largest family of receptor tyrosine kinases. In addition to its physiological roles, recent studies revealed that some Eph genes are associated with human malignancies [6]. For example, EphA1 and EphA4 are upregulated in gastric cancer [7]; EphA2 overexpression in mammary epithelial cells induces tumorigenesis [8] and EphA8 stimulates the proliferation, invasion, and migration of gastric cancer cells [9].
As a member of the Eph receptor group, EphA7 is associated with carcinogenesis [10] but plays a contradictory role in different cancers [11][12][13][14]. EphA7 is overexpressed in hepatocellular carcinoma [15], glioblastoma multiform [11], gallbladder adenocarcinoma [16] and lung carcinoma [17] and contributes to malignant transformation, aggressive progression, and poor prognosis. However, EphA7 may also act as a tumor suppressor since EphA7 downregulation is induced by promoter hypermethylation in prostate cancer patients [18]. Similar results have also been observed in colorectal cancer [20], and oral squamous cell carcinoma [21] etc. Wang J [19,20]confirmed hypermethylation of the promoter of EphA7 in gastric carcinoma since EphA7 expression was restored after demethylation treatment with 5-aza-2-deoxycytidine (5-aza-dc). However, the link between EphA7 and CC remains unclear. Therefore, this study aims to reveal the association of EphA7 methylation with CC and to validate the potential value of EphA7 methylation in the clinical diagnosis of CC.

Cell lines and clinical sample collection Cell culture
Human CC cell lines, including CaSki and SiHa, and a human embryonic kidney cell line were purchased from Zhong Qiao Xin Zhou (Shanghai, China).The cell lines were tested for mycoplasma by PCR and were authenticated using STR profiling. The above 3 cell lines were cultured in RPMI 1640, MEM and DMEM (HyClone, USA) supplemented with 10% fetal bovine serum (BI, USA) and 1% penicillin-streptomycin (Solarbio, China) at 37 °C in 5% CO 2 and saturated humidity.

Clinical specimens
Cervical frozen tissues and Thinprep cytologic test (TCT) specimens were obtained from the Department of Gynecology and Obstetrics, Tianjin Medical University General Hospital, from January 2016 to June 2019. This study was approved by the Medical Ethics Committee of Tianjin Medical University, and followed the Declatation of Helsinki on biology for human trials. All patients gave informed consent.None of the patients received radiotherapy or chemotherapy prior to surgery. According to the results of the pathological diagnosis, patients in the study group were divided into normal, cervical intraepithelial neoplasia II (CINII), CINIII and cancer groups.
A total of 57 frozen tissue specimens were collected for testing. The frozen tissue samples consisted of 25 normal, 24 CIN II/III and 8 cancer samples and were stored at -80 °C. In addition, 114 TCT specimens were collected, including 28 normal cervical samples, 24 CINII grade specimens, 45 CINIII grade specimens and 17 CC specimens. The specimens were obtained using a disposable cervical specimen collection brush and stored in TCT preservation solution (BD Surepath, USA) at 4 °C.

Transient transfection
dCas9-Tet1 or dCas9-DNMT3a plasmids with U6-sgRNA cassettes were transfected into cells, including CaSki, SiHa and HEK293T, using OMNIfect Transfection Reagent (Transomic Technologies, USA) according to the manufacturer's instructions. The ratio of dCas9-Tet1/DNMT3a to an individual sgRNA was 1:1. For the experiments in which sgRNAs were transfected together, the amount of each sgRNA was equal to aliquoted parts of the total amount of sgRNA. The mutated plasmids of dCas9-Tet1CD-IM (Tet1 m ) or dCas9-DNMT3a-IM (DNMT3a m ) were considered as negative controls.

RNA extraction and quantitative real-time PCR (qRT-PCR)
The harvested cells were dissolved in TRIzol ® reagent (Invitrogen, USA), and total mRNA was then extracted following the manufacturer's protocol. qRT-PCR was performed using SYBR Green PCR Mix (Tiangen, China) and a Stratagene Mx3005P sequence detection system (StrataGene, Agilent, USA). The amplification parameters were as follows: 95 °C for 15 min, followed by 40 cycles at 95 °C for 10 s, 60 °C for 20 s, and 72 °C for 20 s. The 2 −∆∆Ct method was performed with the GAPDH gene as an internal control, and the relative quantification procedure was selected. The primer sequences are listed in Table s1.

DNA extraction and bisulfite treatment
Genomic DNA from frozen tissues was extracted using a TIANamp Genomic DNA Kit (Tiangen Biotech, Beijing, China) according to the manufacturer's instructions. DNA from cervical liquid-based cell specimens was isolated by phenol/chloroform extraction. One microgram of genomic DNA per sample was modified using the EZ DNA methylation kit (Zymo Research Corp, Irvine, US) according to the manufacturer's instructions. Leukocyte DNA from healthy women was used as a negative control for methylation, while in vitro methylated leukocyte DNA produced using M. SssI methyltransferase (New England Biolabs, Ipswitch, USA) was used as a positive control.

MSP (Methylation-specific PCR)
Primers were designed via Methy Primer Express v1.0 (Applied Biosystems, USA) and synthesized in Sangon Biotech (Shanghai, China).The primers are shown in Table s1. In total, 1.5 µl bisulfite-treated DNA were amplified in a 30 µl reaction mixture consisting of 1 × PCR Buffer with 0.5U AmpliTaq Gold DNA polymerase (Applied Biosystems, USA), 0.2 mM dNTP mix, and 0.3 µM of each primer. The PCR conditions were as follows: 95 °C for 10 min, then 40 cycles of 95 °C for1 min, 55 °C for 1 min, 72 °C for 1 min and finally an elongation step of 7 min at 72 °C. The PCR products were separated on a 2% agarose gel, prestained with Gelred (Shanghai Life iLab Bio, China) and visualized by UV transillumination. Leukocyte (leu) DNA from healthy women was used as a negative control, and in vitro methylated (iv) leukocyte DNA was used as a positive control.

QMSP (Quantitative methylation-specific PCR)
Quantitative methylation-specific PCR was performed with a double-quenched (FAM/IBHQ)-labelled hybridization probe. The methylated primers for QMSP were the same as those used for MSP. The probes were designed using Clone Manager 9.0 software and synthesized by Sangon Biotech (Table s1). Twenty-five nanograms bisulfite-converted DNA was utilized for PCR with 0.3 µM primers, 0.2 µM probed and 1 × QuantiTect Probe PCR Master Mix (Qiagen, Germany) in 10 µl reaction for 50 cycles in a 7900HT Fast RealTime PCR System (Applied Biosystems). The ACTB gene was used as a methylation-independent internal reference gene. The criteria for the interpretation of positive methylation results were as follows: Ct value < 50 (at least 2 of 3 multiple wells) with sufficient methylated DNA (200 pg DNA). The relative level of EphA7 methylation was analyzed as previously described [31].

Statistical analysis
Statistical analysis was performed using IBM SPSS Statistics 22.0 (IBM Corporation, New York, USA) and Graphpad Prism8.0 (GraphPad Software, USA).Students' t test was used to compare differences in median methylation or expression levels between two groups conforming to normal distribution,whereas a Mann-Whitney U test was used for data not conforming to normal distribution. The correlation between the expression and methylation level was investigated via Pearson correlation coefficients. All transfections were done in triplicate, and for each biological replicate, at least three technical replicates of the qRT-PCR assay were performed. The Kaplan-Meier method was used for survival analysis, and a log-rank test was applied for comparations between groups.A P value below 0.05 was considered significant.

EphA7 is significantly hypermethylated in CC tissues
According to the UALCAN website, the EphA7 promoter was hypermethylated in 15 of 23 categories of cancer (P < 0.05) compared with adjacent normal tissues (Fig. 1a). Furthermore, among cancers, the difference of EphA7 hypermethylation in cervical squamous cell carcinoma & endocervical adenocarcinoma (CESC) and normal tissues was the most prominent (△beta value = 0.464, P < 0.05); the median beta value was listed in Table. s2. However, there was no significant difference in EphA7 methylation between cervical squamous cell carcinoma (SCC) and cervical adenocarcinoma (ADC) (Fig. 1b). Therefore, we further analyzed the methylation status of the EphA7 gene from the TCGA database using Wanderer, an online tool. As shown in Fig. 1c, there were a total of 21 HumanMethylation 450 probes located in the EphA7 gene and 11 probes in the promoter region that exhibited significant differences (P < 0.05) between tumor tissues and adjacent normal specimens. The differences in in global EphA7 methylation are summarized in Table. s3.
Subsequently, pyrosequencing was performed to quantify the methylation level of EphA7 at each CpG site along the sequence positions of the promoter. Compared with normal cervical tissues, the average methylation level of EphA7 the average methylation level of EphA7 within a total of 11 CpG sites was 4.09-fold higher in tumors (Fig. 1d, e).

Promoter methylation of EphA7 is inversely correlated with gene expression in CC
By MEXPRESS web tool, it was observed that 19 out of 21 HumanMethylation 450 probes located in the EphA7 gene showed significantly negative expression (Pearson correlation coefficients from -0.211 to -0.522), and the correlation was more significantly pronounced in the highlighted region, which focused on the EphA7 promoter (Fig. 2a).
This inverse correlation was further confirmed by the UCLCAN and GEPIA web tools. As presented in Fig. 2b and 2c, EphA7 was notably hypermethylated and downregulated in cervical cancer compared with normal tissues (P < 0.05).
Consistent with the bioinformatic findings, EphA7 methylation bands that were observed in CaSki and SiHa cells, but not present in HEK293T cells via MSP (Fig. 2d). Full-length gel pictures are presented in Supplementary Figure. s2. However, the expression level of EphA7 was significantly decreased in the CaSki and SiHa cells compared with the HEK293T cells (P < 0.001) (Fig. 2e).

DNA methylation plays a direct role in EphA7 gene regulation
To confirm that DNA methylation affects EphA7 expression directly, a set of CRISPR-based DNA methylation regulation tools (dCas9-Tet1/DNMT3a) was developed to target the EphA7 promoter region.
Since EphA7 was hypermethylated in CC cells (CaSki and SiHa), we applied 3 sgRNAs targeted for activation and the dCas9-Tet1 plasmid to upregulate EphA7 expression via demethylation. In CaSki cells, the sgRNA1 (6.20fold) and sgRNA 3 (2.04-fold) groups had significantly increased EphA7 gene expression compared with the control group (Tet1 m ) (Fig. 3a). However, EphA7 gene expression did not increase further in the group with all three sgRNAs (sgRNA1 + 2 + 3) (1.49-fold). Then, we performed pyrosequencing to detect the methylation level in the promoter region of EphA7 in the sgRNA1 group in CaSki cells. As expected, compared with the control (Tet1 m ) group, the global methylation level There was no significant difference in EphA7 methylation between SCC and ADC. c The levels of mean EphA7 methylation at a total of 21 probes sites. Among them, all the 11 probes in the promoter region (blue) exhibited significant differences between tumor and adjacent normal specimens (P < 0.05). * represents adjusted P value < 0.05. Plot and P value were produced via Wanderer. d Pyrosequencing showed that EphA7 promoter methylation was higher in the tumor tissues compared with normal. e Compared with the normal cervical tissues (n = 5), the average promoter methylation level of EphA7 within a total of 11 CpG sites was 4.09-fold higher in tumors (n = 5) according to pyrosequencing decreased, and the average methylation level decreased by 19.10% in CaSki cells (Fig. 3b, c, Table. s4).

Validation of the potential clinical value of EphA7 methylation in the diagnosis of CC
MSP was performed in 57 cervical tissues, and the methylation positive rate in tissues was 0% (0/25) in normal tissues, 45.83% (11/24) in CINII/III, and 100% (8/8) in cancer tissues (Fig. 5a). Full-length gel pictures are presented in Supplementary Figure. s3.
Furthermore, Kaplan-Meier curves demonstrated that lower methylation of EphA7 was correlated with an increased patient survival period via LinkedOmics (Figure. s4a). The Human Protein Atlas also confirmed that high expression of EphA7 was associated with improved survival (Figure. s4b).

Discussion
Aberrant methylation of EphA7 has been reported in the process of carcinogenesis but not in CC. Our study is the first to systematically investigate the hypermethylation of EphA7 in CC via an effective protocol that combines dry-lab data and wet-lab experiments. Furthermore, CRISPR-dCas9 tools were employed to precisely modify the CpG island in the promoter, which successfully provided direct evidence that EphA7 methylation is involved in its expression.
Rapid developments in bioinformation have inspired suggestions that the field of precision oncology will greatly benefit from a multiomics analytical approach [32][33][34]. Genomic datasets and web analysis tools have indeed opened the door to deeper and wider exploration of novel biomarkers. However, this novel opportunity also requires creative and skillful solutions to address noisy, unstructured information to offer valuable biological insights. Hence, we developed a logistically pipeline comprising both dry-lab data and wet-lab experiments to uncover EphA7 methylation in CC. First, we investigated EphA7 methylation by applying multiweb tools to data from TCGA and GTEx. TCGA is a public database that includes 33 cancer types and matched clinical data. GTEx is a normal tissue expression dataset that contains RNA-seq profiles of multiple tissues and cell types from hundreds of demographically diverse healthy individuals. However, because the raw data are difficult to interpret, we explored EphA7 hypermethylation in CC tissues using UALCAN and Wanderer web tools. Using MEXPRESS, the methylation level in the promotor region was negatively correlated with the expression, which suggests that EphA7 is a suppressor gene in cervical carcinogenesis. LinkedOmics analysis demonstrated that lower methylation notably increased the patient survival period. After verifying the bioinformatic results in cell lines, we further moved to real tissues and clinical TCT samples, and both confirmed EphA7 hypermethylation in CC.
Additionally, although decades of research have focused on the biological function of DNA methylation, the role of DNA methylation in targeted gene expression remains unclear due to a lack of molecular tools to precisely modify DNA methylation in the genome [35]. The advent of the CRISPR system has enabled precise and stable editing of the epigenome at targeted sgR-NAs by fusing catalytically dead Cas9 (dCas9) to DNA methylation-modifying enzymes such as DNA methyltransferases (DNMTs) and Ten-eleven translocation methylcytosine dioxygenases (Tets) [36]. DNMTs target  3). b Pyrosequencing showed that the methylation level in the dCas9-DNMT3a group were higher than the control (n = 3). c The average CpG sites of the promoter increased nearly 23.67% (n = 3) in the dCas9-DNMT3a group cytosines and facilitate methylation of CpG islands, which contributes to the regulation of gene expression [37]. DNA methylation is maintained by the methyltransferase DNMT1, while DNMT3a and DNMT3b provide de novo CpG methylation [38]. By contrast, Tet promotes the process of DNA demethylation. Both DNMTs and Tets activities play important roles in the dynamic changes in DNA methylation [39,40]. X.S. Liu et al. [30] constructed the dCas9-Tet1/DNMT3a vector by fusing Tet1 or DNMT3a with the dCas9 protein for re-editing DNA methylation in mice. Moreover, they determined that the CRISPR vector system has better methylation editing efficiency than the TALE vector [30] and that the off-target effect of dCas9-Tet1 on the editing DNA methylation is minimal [41]. Therefore, to establish that DNA methylation plays a direct role in EphA7 expression, both CRISPR-dCas9-Tet1, which targets demethylation, and CRISPR-dCas9-DNMT3a, which targets methylation, were applied in our study. As expected, with the dCas9-Tet1 and sgRNA tools, EphA7 was activated with loss of methylation in the promoters of CaSki (19.10%) and SiHa (17.23%), respectively. Simultaneously, we induced a 23.67% increase in methylation in the first exon of EphA7 via dCas9-DNMT3a, and the expression was indeed downregulated. These results offer strong evidence that EphA7 DNA methylation participates in CC through perturbation of gene regulation. Furthermore, the efficiency of the CRISPR tools in our study is comparable with the results of other studies. Nozomi Hanzawa et al. [42] reduced DNA methylation by 31.5% and 27.6% with different single sgRNAs. X.S. Liu and J.Wu et al. induced gene hypermethylation of approximately 20-30% using dCas9-DNMT3a [30,43]. However, Nozomi Hanzawa et al. [42] showed that cotransfected sgRNAs result in additively DNA demethylation levels (approximately 40%) [42], and the efficiency is equal to that of individual sgRNAs when multiple sgR-NAs bind in the same region but on the opposite strand, thus serving as competing sgRNAs [44]. Consistent with these results, EphA7 expression was not enhanced more when all the sgRNAs were combined in our study.
Additionally, DNA methylation may disrupt the binding of transcription factors to regulate the regions of target genes [45], or conversely, the binding of transcription factors to these regions may prevent their methylation [46]. In our study, one of the CpG sites of EphA7 presented much lower methylation (Fig. 1d, 3e), with almost no difference between the demethylation group and control. Hence, we investigated the possible disrupted transcription factor via JASPAR [26]. The results show that Yin Yang 1 (YY1) and transcription factor AP-2 alpha (TFAP2A) are closely related to DNA methylation ( Figure. s5). The YY1/polycomb group (PcG) protein/ DNMT complex maintains gene methylation and contributes to gene inactivation [47]. DNA methylation can increase the binding of the transcription factor TFAP2A to the target site, leading to suppressed gene expression [48]. These could be the reasons for the low methylation level at the 8th CpG site (Fig. 1d, 3e),but the practical implications require evaluation in a separate study.
Last but not least, SCC and ADC are two major subtypes of CC. Compared with SCC, ADC is mainly diagnosed in more advanced stages and has a worse prognosis. In the last decade, the incidence of ADC has increased rapidly [31]. One reason for the upward trend of ADC is the relatively low effectiveness of cytomorphological detection in screening programs; therefore, novel biomarkers for CC are required that ideally can identify both subtypes. Fortunately, EphA7 is hypermethylated in both SCC and ADC in TCGA datasets and in SiHa (SCC) and CaSki (ADC) cells.

Conclusions
EphA7 hypermethylation is a potential biomarker for the diagnosis and screening of CC. CRISPR provides a set of powerful tools to investigate the functional significance of DNA methylation in a locus-specific manner. Subsequently, based on the achievements from this study, it's reasonable to move to the next two directions, one is to further the fundamental research on the detailed mechanism of EphA7 methylation with other cofactors, the other one is to perform a large cohort evaluation that includes all stages from normal to cervical neoplasia, including CINI-CINIII, and cancer to advance the translational research, which both are currently being studied in our laboratory and will be presented in the coming future.
Additional file 1: Supplemental Table1. Primer and probe sequence used in this study. Supplemental Table2. The median of beta value of EphA7 methylation in normal vs. 23 types of tumor. Supplemental Table 3. Each CpG sites of EphA7 methylation in normal vs. CSEC. Supplemental Table 4.The average methylation (%) of each CpG site of EphA7 in actsgRNA1 group (dCas9-Tet1 vs. control). Supplemental Table 5. The average methylation (%) of each CpG site of EphA7 in repsgRNA2 group in HEK293T cells (dCas9-DNMT3a vs. control). Figure.s1 The site of sgRNAs with CRISPR-dCas9 systems targeted in EphA7 promoter and exon 1th. Three active sgRNAs (act-sgRNAs) for demethylation, and four repressive sgRNAs (rep-sgRNAs) for increasing methylation were designed. Figure. s2 Full length gel pictures for the results of EphA7 methylation in cell lines using MSP. The marked part was presented in Figure. 2d. Figure.s3 Full length gel pictures for the verification results of EphA7 methylation in total 57 cervical tissues using MSP, consisted of 25 normal, 24 CINII/III grade and 8 cancer samples. The marked part was presented in Figure. 5. Figure.s4 Kaplan-Meier curve of the survival of patients with high and low levels of EphA7 methylation or expression. a Kaplan-Meier curves demonstrated that lower methylation was correlated with an increased patient survival period via LinkedOmics. b The high expression of EphA7 was associated with improved survival confirming by Human Protein Atlas. Figure.s5 The transcription factors was preditected located in the promoter of EphA7 CpG site. Red "cg" was the target CpG site, which has the possiblity to bind with YY1, TFAP2A via JASPAR.