Methylomics analysis identifies a putative STAT3 target, SPG20, as a noninvasive epigenetic biomarker for early detection of gastric cancer

Gastric cancer is a leading cause of cancer worldwide. Our previous studies showed that aberrant activation of JAK/STAT3 signaling confer epigenetically silences STAT3 target genes in gastric cancer. To further investigate the clinical significance of this phenomenon, we performed Illumina 850K methylation microarray analysis in AGS gastric cancer cells, and cells depleted of STAT3. Integrative computational analysis identified SPG20 as a putative STAT3 epigenetic target, showing promoter hypomethylation in STAT3-depleted AGS cells. Bisulphite pyrosequencing and qRT-PCR confirmed that SPG20 is epigenetically silenced by promoter hypermethylation in a panel of gastric cancer cell lines including AGS cells, but not in immortalized gastric epithelial GES cells. Expression of SPG20 could be restored by the treatment with a DNMT inhibitor, further suggesting that SPG20 is epigenetically silenced by promoter methylation. Clinically, a progressive increase in SPG20 methylation was observed in tissues samples from gastritis (n = 34), to intestinal metaplasia (IM, n = 33), to gastric cancer (n = 53). Importantly, SPG20 methylation could be detected in cell-free DNA isolated from serum samples of gastritis, IM and gastric cancer patients, having a progressive similar to tissues. Taken together, SPG20, a potential STAT3 target, is frequently methylated in gastric cancer, representing a novel noninvasive biomarker for early detection of this deadly disease.


Introduction
Gastric cancer is a leading cause of cancer deaths worldwide [1]. Despite advances in cancer therapy, gastric cancer patients still have a poor 5-year survival of less than 15% [2], likely due to a lack of biomarkers for early detection. Although endoscopic screening for early gastric cancer may improve patient's survival, noninvasive, sensitive, and specific population screening assays are currently not available and are therefore urgently needed.
Epigenetic modifications, including DNA methylation, play an important role in transcriptional regulation, and embryonic and disease development [3,4]. Due to its stability, methylation at the cytosine of CG dinucleotides has been found both in tissues and bodily fluids such as plasma and serum. These attributes make DNA methylation an attractive target for noninvasive cancer detection in cell-free DNA (cfDNA, also known as liquid biopsy) [5,6].
Our laboratory has long been dedicated to the development of DNA methylation as a biomarker for disease monitoring [7][8][9]. Additionally, we previously demonstrated that DNA methylation could be detected in bodily fluids, in several human diseases, including cancers [10,11]. However, promising DNA methylation biomarkers, for noninvasive early detection of gastric cancer, remain lacking.
Our previous studies demonstrated that aberrant activation of JAK/STAT signaling, triggered by H. pylori infection, can lead to epigenetic silencing of tumor suppressors in gastric cancer [12,13]. In this study, we examine additional STAT3 targets that are epigenetically silenced by DNA methylation in gastric cancer, representing sensitive biomarkers for noninvasive detection of gastric cancer.
PCR reactions were performed on an ABI Step-One real-time PCR system (Applied Biosystems, Foster City, CA) with specific primers (Table 2). Relative gene expression was determined by comparing the threshold cycle (Ct) of the test gene against the Ct value of GAPDH, in a given sample. (i.e., the comparative Ct method).

Bisulphite conversion and pyrosequencing
0.5μg genomic DNA was bisulphite modified using an EZ DNA Methylation Kit (Zymo Research, Orange, CA) according to the manufacturer's protocol. The bisulphite modified DNA was then subjected to PCR amplification, using a tailed reverse primer, including a biotin-labeled universal primer, PCR, and sequencing primers were designed using PyroMark assay Design 2.0 software (Qiagen). The SPG20 transcription start site (+1094 to +1387) was PCR amplified using specific primer (Table 3), and pyrosequencing was performed using PyroMark Q24 (Qiagen) Pyro Gold Reagents (Qiagen), according to the manufacturer's protocol. Methylation percentage of six CpG sites located from +1209 to +1229 bp was determined by the fluorescence intensity of cytosines and thymines at each CpG site. In vitro methylated DNA (IVD, Merck Millipore, MA) was included as positive control for bisulphite pyrosequencing.

Methylation-specific PCR (MSP)
Bisulphite-modified DNA was subjected to MSP for SPG20 methylation analysis using specific primers ( Table 2). 4μl of bisulphite-converted DNA were amplified in a total volume of 20μl containing 10x PCR Buffer, 0.25mM dNTPs, 2mM MgCl 2 , 0.2μM of each primer and 1.25U of Platinum Taq DNA polymerase (Invitrogen) at 95˚C for 2 min, followed by 40 cycles of denaturing at 95˚C for 30 sec, annealing at 60˚C for 30 sec, and extension at 72˚C for 30 sec, followed by followed by a final extension step of 72˚C for 10 min. In vitro methylated DNA (IVD, Merck Millipore) was included as positive control and normal blood (NB) was included as a negative control of MSP. 10μl of PCR products were loaded onto 10% polyacrylamide gels, which were then stained with ethidium bromide, and visualized under UV illumination.

Infinium microarray DNA methylation analysis
Bisulphite-modified DNA from AGS gastric cancer cells and cells depleted of STAT3 was subject to methylation analysis, using an Illumina 850K methylation microarray. The methylation level of each probe (β-value) was defined by the intensity of the methylated allele (M) / (intensity of the unmethylated allele (U) + the intensity of the methylated allele (M) + 100). The microarray data has been deposited in the Gene Expression Omnibus database (accession number: GSE109541). Table 2. Primer sequence used in this study.

Statistical analysis
Unpaired t-tests were used to compare parameters of the various groups. All statistical calculations were performed using the SPSS statistical package (version 18.0) for Windows (IBM, Chicago, IL, USA). In this study, P values <0.05 were considered statistically significant.

Results
Our previous studies found that aberrant activation of JAK/STAT signaling could lead to epigenetic silencing of STAT3 targets in gastric cancer [12,13]. We therefore hypothesized that binding of STAT3 to promoter-proximal CpG islands may affect their methylation status. In this regard, we performed Illumina 850K methylation microarray analysis in bisulphite-treated genomic DNA from AGS gastric cancer cells, and cells depleted of STAT3. Computational predictions were also performed to identify all STAT3-binding sites located in open chromatin regions (as demarcated by H3K4me1 and H3K27Ac) in close proximity to promoter CpG islands ( Fig 1A). One probe (cg25179758, Fig 1B and 1C) within the promoter region of SPG20, showing differential hypomethylation in STAT3-depleted AGS cells, was identified.
To further examine the role of DNA methylation in regulating expression of SPG20, we assessed its expression in an immortalized gastric epithelial cell (GES), and a panel of gastric cancer cell lines. Except for GES cells, downregulation of SPG20 was observed in all cancer cell lines (Fig 2A). Treatment with the DNMT inhibitor (5aza-DC) resulted in robust re-expression of SPG20 in those cells (Fig 2B), due to DNA demethylation of its promoter as confirmed by bisulphite pyrosequencing (Fig 2C).
We then examined the clinical significance of SPG20 methylation in a cohort of gastritis, intestinal metaplasia (IM), and paired gastric cancer patient samples (Table 1). Although no clinical parameters significantly associated with DNA hypermethylation (Table 3), results from bisulphite pyrosequencing showed higher SPG20 methylation in gastric tumors, compared to tumor-adjacent normal and gastritis tissues (Fig 2D). Interestingly, IM tissues also showed a higher SPG20 methylation, compared to adjacent normal and gastritis tissues (Fig 2D).
We also compared SPG20 methylation in datasets (GSE103186 and TCGA) from normal gastric epithelial, IM, and gastric cancer tissues obtained from two publicly available databases. Consistently, cancer tissues showed higher SPG20 methylation than normal and IM samples ( Fig 2E). As demonstrated in our cohort, samples from IM also showed higher SPG20 methylation than those from normal tissues. It is interesting to note that SPG20 methylation, in cancer tissues from Asian populations, were higher than those from non-Asian populations, suggesting population-specific underlying epigenotypes.
To further examine the feasibility of SPG20 methylation, as a noninvasive methylation biomarker for the early detection of gastric cancer, we performed conventional methylation-specific PCR to amplify short fragment of methylated DNA from cell-free DNA (cfDNA) obtained from serum samples of non-cancer individuals, and patients with IM or gastric cancer. In agreement with our bisulphite pyrosequencing results, SPG20 methylation could be detected in all gastric cancer cell lines by MSP (Fig 3). Overall, the sensitivity and specificity of gastric cancer detection using SPG20 methylation, were 88.6% and 75%, respectively (Table 4). Importantly, the sensitivity of cancer or IM detection could remain as high as 87.5% (Table 4).

Discussion
DNA methylation, an "epigenetic" mode of transcriptional regulation, is altered in numerous pathologies. Consequently, due to its chemical uniqueness and stability (e.g., vs. RNA), disease-associated methylated DNA sequences represent promising tissue and liquid biomarkers. In the current study, by methylation microarray, we identified that a potential STAT3 In addition, bioinformatics analysis was performed to identify potential STAT3 targets by filtering genes with at least one STAT3-binding site in potential enhancer region (as enriched with H3K4me1 and H3K27Ac) of the promoter CpG island. One hypomethylated probe and a potential STAT3 target, SPG20, was selected for further analysis. (B) Scatter plot showing the average beta-value of each probe in AGS cells depleted of STAT3 or GFP (as control). One of the probes (cg25179758, red dot) showing hypomethylation, located within the SPG20 promoter region, was selected for further analysis. Methylation% of SPG20 at the promoter region as determined by (C) methylation microarray (beta value) or (D) bisulphite pyrosequencing.
https://doi.org/10.1371/journal.pone.0218338.g001 SPG20 as noninvasive epigenetic marker  target, SPG20, is differentially methylated in gastric cancer. Cell line studies further confirmed that SPG20 is epigenetically silenced, by DNA methylation, in gastric cancer cell lines. Clinical studies also demonstrated SPG20 hypermethylation in tissues and serum samples from IM and gastric cancer patients. Importantly, a progressive increase in SPG20 methylation, from gastritis to gastric cancer, was observed. SPG20, encoding the multifunctional protein Spartin, has been shown to be involved in several cellular processes. Several studies have found that SPG20, containing a MIT (microtubule interacting and trafficking) domain, is involved in cytokinesis [14]. Cells depleted of SPG20 showed cytokinesis arrest and convoluted midbodies. More recently, SPG20 has been found to be involved in EGFR trafficking and MAPK signaling pathway [15,16]. Specifically, gastric cancer cells depleted of SPG20 showed increased EGFR expression and phosphorylation of kinase involved in the MAPK signaling. Taken together, these studies suggest that SPG20 may act in early carcinogenesis and proliferation of human cancer. In agreement with those observations, SPG20 methylation was observed in intestinal metaplasia (IM), an early lesion of gastric cancer, from our own cohort and published data [17]. Although gene silencing of SPG20 methylation has been demonstrated previously in several cancers [18,19], including gastric cancer [20], we believe this is the first study to demonstrate SPG20 methylation in IM, suggesting a role in early gastric carcinogenesis.
Infection by H. pylori is considered as a major risk factor for gastric cancer [21], capable of activating multiple signaling pathways including the JAK/STAT signaling [22,23]. In this regard, activation of JAK/STAT signaling plays an important role in gastric carcinogenesis [24,25]. However, the role of STAT3 in the epigenetic silencing of its targets is not fully elucidated. In this study, we found that the putative STAT3 target, SPG20, is epigenetically silenced in gastric cancer, similar to our previous findings of two other STAT3-epigenetically silenced targets, NR4A3 [13] and GATA3 [12]. Although a putative STAT3-binding site, as determined by ENCODE data, is few kb away from the promoter CpG island, the presence of the enhancer histone mark, H3K4me1, suggest that STAT3 might affect promoter methylation by higher-retrieved from GSE103186 (normal and IM) and TCGA (cancer). Significant differences between groups are indicated by � P<0.05, �� P<0.01, ��� P<0.005, as determined by Mann-Whitney U-test.
https://doi.org/10.1371/journal.pone.0218338.g002 order chromatin structure. Such postulate is supported by the observation that Asian gastric cancer patients with more virulent and inflammatory-inducing strain of H. pylori strains [26][27][28] have higher SPG20 methylation than non-Asian populations with less virulent strains ( Fig  2E). However, further experiments are required to demonstrate the role of STAT3 in the epigenetic silencing of SPG20. In this study, we also demonstrated that SPG20 methylation could be detected in cell-free DNA isolated from serum samples of intestinal metaplasia (IM) and gastric cancer patients (Illustrated in Fig 4). Importantly, a progressive increase of SPG20 methylation detection was observed from gastritis, to IM, and to cancer. However, more samples are required to determine the sensitivity and specificity of SPG20 methylation in cancer diagnosis especially in the detection of early gastric cancer lesions.
In conclusion, we herein demonstrate that the putative STAT3 target, SPG20, is epigenetically silenced by promoter methylation in gastric cancer. and that such methylation is detectable in tissues and cfDNA from patients with gastric cancer as well as intestinal metaplasia. Consequently, SPG20 methylation may be able to serve as a non-invasive biomarker for the early detection of gastric cancer.