MicroRNA Expression Profiles in Gastric Carcinogenesis

Intestinal-type gastric carcinoma exhibits a multistep carcinogenic sequence from adenoma to carcinoma with a gradual increase in genomic alterations. But the roles of microRNAs (miRNA) in this multistage cascade are not fully explored. To identify differentially expressed miRNA (DEM) during early gastric carcinogenesis, we performed miRNA microarray profiling with 24 gastric cancers and precursor lesions (7 early gastric cancer [EGC], 3 adenomas with high-grade dysplasia, 4 adenomas with low-grade dysplasia, and 10 adjacent normal tissues). Alterations in the expression of 132 miRNA were detected; these were categorized into three groups based on their expression patterns. Of these, 42 miRNAs were aberrantly expressed in EGC. Five miRNA (miR-26a, miR-375, miR-574-3p, miR-145, and miR-15b) showed decreased expression since adenoma. Expression of two miRNA, miR-200C and miR-29a, was down-regulated in EGCs compared to normal mucosa or adenomas. Six miRNA (miR-601, miR-107, miR-18a, miR-370, miR-300, and miR-96) showed increased expression in gastric cancer compared to normal or adenoma samples. Five representative miRNAs were further validated with RT-qPCR in independent 77 samples. Taken together, these results suggest that the dysregulated miRNA show alterations at the early stages of gastric tumorigenesis and may be used as a candidate biomarker.

In addition, we used RNA sequencing data from our previous study to investigate the correlation between the expression of miRNAs and their target mRNAs.

Results
Altered miRNA expression in normal gastric mucosa, adenoma, and EGC. We performed miRNA expression arrays by using NanoString to identify DEM patterns in normal gastric mucosa, adenoma with LGD and HGD, and EGC (Table 1). We identified 132 DEMs among three groups (p-value < 0.01, average expression >5, and fold change > = 2.5 or < = 0.4). DEMs were categorized into three groups based on the changing patterns of miRNA expression. The DEM-1 group exhibited a decrease in the miRNA expression from adenoma lesion (DEM-1a, n = 16) or EGC (DEM-1b, n = 7) compared to normal gastric mucosa. DEM-2 group comprised cases with highest miRNA expression levels in adenomas (n = 3). DEM-3 group showed an upregulated miRNA expression in EGC as compared with normal mucosa or adenomas (n = 98) (Supplementary Table 1).
Principal component analysis (PCA) showed that the cancer and non-cancer samples were separated along the PCA1 axis, whereas the normal gastric mucosa and adenoma samples were differentially located along the PC2 axis (Fig. 1A). Unsupervised PCA results showed that normal and adenoma samples showed similar expression patterns, but EGC samples tended to cluster to a separate group (Fig. 1B).
Additionally, we validated two novel miRNAs which have been established as tumor suppressor or suppressor of epithelial-to-mesenchymal transition (miR-655) in hepatocellular 36 and esophageal squamous cell carcinoma 37,38 , and reported as oncogenic (miR-938) miRNA in colorectal cancers 39 , but their roles have not been studied in gastric cancer. Unexpectedly, we found that expressions of miR-655 decreased from normal to adenoma and carcinoma, suggesting tumor suppressive role in gastric cancer. Meanwhile, miR-938 was up-regulated in EGCs compared to normal gastric mucosa and adenomas, suggesting oncogenic function in gastric cancer ( Supplementary Fig. 1).
Validation of results and correlation between miRNA and target genes. For further experimental validation, we measured the expression of five miRNAs (miR-107, miR-300, and miR-370 from DEM-3; miR-26a and miR-375 from DEM-1a) in 77 independent samples using RT-qPCR, owing to the obvious difference (fold change >4) in their expression between normal and EGC group in the miRNA microarray analysis. We observed a progressive increase in the expression of miR-107 and miR-300 from normal to adenoma and EGC in microarray analysis; these results were subsequently validated in an independent sample set with RT-qPCR. In comparison to the normal gastric mucosa, adenomas and EGCs showed a decrease in the expression of miR-26a and miR-375 in both microarray and qPCR assays ( Fig. 2A). However, we failed to confirm the gradual increase in the expression of miR-370 in the validation set. From TCGA datasets, we found that 4 miRNAs (miR-107, miR-370, miR-26a and miR-375) were dysregulated in gastric cancer compared to normal gastric mucosa and also showed differential gene expression levels between tubular adenocarcinoma (well-and moderately differentiated, intestinal-type by Lauren) group (n = 76) and other types of gastric cancer group (n = 313) (Fig. 2B).
To determine the impact of these miRNAs on target genes, we calculated the correlation coefficient between the expression of miRNAs and their target genes using our previous RNA sequencing data. A negative correlation was observed between miR-26a and EZH2 (r = −0.689), miR-375 and YWHAZ (r = −0.629), and miR-375 and RUNX1 (r = −0.432) (Fig. 3A). We also found weak negative correlation between miR-26a and EZH2 (r = −0.101), miR-375 and RUNX1 (r = −0.218) from TCGA dataset (Fig. 3B). Three target genes (EZH2, YWHAZ and RUNX1) were upregulated in gastric cancer compared to normal gastric mucosa. Moreover, their normalized mRNA levels also showed slight difference between well-and moderately differentiated tubular, intestinal-type gastric cancer group compared to other types of cancers (Fig. 3C). We also performed correlation analyses with miR107, miR300, and miR370 with their target genes and only miR375 and miR26a showed high correlations with their target genes ( Supplementary Fig. 2).

Discussion
The pathogenesis of gastric cancer involves multistep genetic and epigenetic alterations, which predispose cells to neoplastic transformation [1][2][3] . Given the importance of miRNAs in the regulation of cell growth and viability, miRNA dysregulation is believed to be closely correlated with the development and progression of gastric cancer 4 . In the stomach, previous research on miRNA dysregulation was focused on gastric cancer its elf 6,7,18,20,22,23,26,27,29,33,40 and only one study has been conducted to explore the role of miRNA during the histologic progression from gastric adenoma to carcinoma without non-tumorous controls, warranting more research 16 . To our best knowledge, this is the first study to evaluate and compare the expression profile of miRNAs in non-tumor tissue, LGD, HGD, and EGC, using a high-throughput screening array. The aim of our study was to screen alterations in miRNA expressions during the stepwise gastric carcinogenesis and we demonstrated the role of miRNAs during the stepwise gastric carcinogenesis and confirmed their expressions with RT-qPCR. Although previous reports have revealed several molecular alterations in gastric cancer such as DNA mutations, copy number variations, mRNA expression changes, and miRNA alterations, most of these studies were conducted by comparing gastric cancer tissues or cell lines with normal gastric samples, without adenoma samples. Zhu et al. 16 investigated the miR-106a expression in gastric dysplasia and gastric samples using in situ hybridization. The frequency and extent of miR-106a expression was found to show a gradual increase along the histologic progression from mild, moderate, and severe dysplasia to EGC. However, they failed to evaluate the expression of other miRNAs. In the present study, we identified alterations in miRNA expressions using normal, LGD, HGD, and EGC samples. Overall, the expression of miRNAs showed distinct patterns as compared with the expression pattern of mRNAs. In comparison to the EGC samples, samples from the non-cancerous group (normal and adenoma samples), showed similar miRNA expression pattern. Within the non-cancerous group, the normal and adenoma samples were separated based on their expression patterns.
The expression of miR-107 has been reported to increase in the serum and tissues of gastric cancer patients and gastric cancer cell lines. miR-107 acts as an oncogene and regulates gastric cancer development and progression by targeting NF1, DICER1, FOXO1, and CDK8 25,27,28,41 . We found a gradual increase in the expression of miR-107 with the histologic progression from LGD to EGC. Microarray and RT-qPCR results revealed that the expression of miR-107 was significantly upregulated in HGD and EGC as compared with the normal mucosa (**p = 0.006 and *p = 0.031, respectively).
Overexpression of miR-300 promotes cell cycle progression, cell proliferation, and invasion in several cancers, including gastric cancer 34 , liver cancer 40 , and osteosarcoma 42 . miR-300 displays the potential to be used in the treatment, diagnosis, and prognosis of gastric cancer, as its expression decreased in gastric cancer following chemotherapy or exposure to radiation 43 .
In gastric cancer, miR-370 showed altered expression and acted as either an oncogene or a tumor suppressor. Previous studies have shown that the up-regulation of miR-370 resulted in the progression of gastric carcinoma via suppression of transforming growth factor beta receptor II (TGFβRII) or FOXO1 32,33 . Although we observed a relative increase in the expression of miR-370 in LGD, it showed a weak correlation with the expression of its target gene, FOXO3.
The expression of miR-26a is frequently aberrant in many tumors such as gastric cancer, bladder tumor, breast cancer, oral squamous cell carcinoma, and Burkitt lymphoma 44 . miR-26a is known to be significantly down-regulated in gastric cancer and suppresses tumor growth and metastasis by targeting FGF9 gene 17 . Furthermore, miR-26a improves the sensitivity of gastric cancer cells to cisplatin-based chemotherapies by targeting NRAS and E2F2 45 . We found a decrease in the expression of miR-26a from LGD and a strong negative correlation between the expression of miR-26a and its known target EZH2 (r = −0.689).
Previous studies have demonstrated that miR-375 inhibits cell proliferation of gastric cancer cells by repressing JAK2, ERBB2, and YWHAZ 18,46-48 . We found a significant decrease in miR-375 expression in LGD, HGD, and EGC, as evident from the microarray and RT-qPCR results. Moreover, miR-375 expression was inversely correlated (r = −0.629) with the expression of YWHAZ (14-3-3ζ), a member of the 14-3-3 family of proteins, which has been implicated in the initiation and progression of cancers and is a potential biomarker for gastric cancer. Additionally, we validated two novel miRNAs which have been known as tumor suppressive (miR-655) and as oncogenic (miR-938) miRNA in gastrointestinal cancers other than gastric [37][38][39]49 . Although they did not show any stepwise elevation during multistep carcinogenesis, we could prove that they worked as tumor suppressive and oncogenic in gastric cancers. These results also prove that those novel miRNAs are important in gastric cancers, but do not work in multistep carcinogenesis, so they were not detected in our DEM miRNA groups.
Our study has several limitations. First, the sample size in miRNA array experiments is quite small (n = 24). To overcome this limitation, we replicated our microarray results in 77 FFPE samples and TCGA datasets. Secondly, the effects of altered miRNA on their target mRNA were not investigated. It is widely accepted that miRNAs have multiple -sometimes hundreds-of targets and the main approach to explore connections between miRNAs and their targets has been focused on the most significant target for each miRNA 50 . In the present study, we selected target genes that have been known to interact with miRNA in the previous studies by experimental investigation to confirm the relationship between miRNA and target mRNA. Further in vitro experiments using cell lines and larger scale studies are recommended to confirm our results and to analyze the potential effects of those deregulated miRNA.
In conclusion, we found that the upregulated (miR-107 and miR-300) or down-regulated (miR-26a and miR-375) miRNAs show alteration at early stages of gastric tumorigenesis and may be used as candidate biomarkers. Further investigation is needed to elucidate the exact role of these miRNAs in early gastric carcinogenesis.

Materials and Methods
Ethics statement. Fresh tumor and non-tumor samples were obtained by forceps biopsy at Samsung Comprehensive Cancer Center. Informed consent was obtained from all individuals who participated in this study. The study protocol was approved by the institutional review board of Samsung Medical Center (IRB 2010-09-020-008) and all experiments were performed in accordance with the approved guidelines and regulations.
Samples. Fresh tumor samples from 14 patients were obtained during endoscopy by forceps biopsy and were used as a test set (seven well to moderately differentiated intestinal-type EGCs, three HGDs, and four LGDs) ( Table 1). No patient had prior chemo or radiation therapy. All lesions were completely removed further with an endoscopic submucosal dissection technique after tissue acquisition with forceps biopsy, and the resected specimens were reviewed by two pathologists. The pathologic diagnosis of histological grade (LGD, HGD, or EGC) was made based on the review of both forceps biopsy and endoscopic submucosal dissection specimens to ensure histological homogeneity as previously described 3 . Tumors mixed with components of different histological grades were excluded from the study. Non-tumor gastric samples more than 2 cm apart from the tumor were used as a reference. All non-tumor samples showed intestinal metaplasia. All fresh tissue samples were snap frozen in liquid nitrogen immediately after collection and stored at −80 °C until use. Helicobacter pylori infection status was determined using both urea breath test and histology, while Epstein-Barr virus status was determined by the Epstein-Barr encoding region in situ hybridization.
In addition, a total of 77 formalin-fixed paraffin-embedded (FFPE) samples were collected and used as a validation set (21 well/moderately differentiated intestinal-type EGCs, 22 HGDs, 24 LGDs, and 10 non-tumor tissues). Informed consent was obtained from all individuals who participated in the study, and the study protocol was approved by the institutional review board. miRNA microarray by NanoString. For the nCounter analysis, 10 consecutive tissue sections (4-μm thick) from archival FFPE tissues of precursor lesions (LGD and HGD) and carcinoma were used. Total miRNA was isolated using the Qiagen miRNeasy Kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. Total miRNA samples were analyzed using the nCounter Human miRNA Expression Assay kit (NanoString, Seattle, WA) according to manufacturer's instructions. Briefly, 100 ng RNA was incubated in the presence of miRNA-specific capture and reporter probes and non-hybridized probes were removed, followed by immobilization of the purified hybridized complexes. The abundance of specific target molecules was subsequently quantified on the nCounter Digital Analyzer by counting the individual fluorescent barcodes and assessing the target molecules, as previously described 51 . Differentially expressed miRNA analysis and correlation with mRNA. To identify the differentially expressed miRNAs (DEMs) among LGDs, HGDs, and EGCs, we calculated analysis of variance (ANOVA) p-value, average of miRNA expression within the same group, and fold change by comparing expression levels among the three groups. DEMs were defined by a p-value < 0.01, average expression >5 in at least one group, and |log2(fold change)| >1. The adjusted p-value was calculated using Benjamini and Hochberg algorithm in R. We next performed a k-means clustering to categorize DEMs into groups using expression patterns. To quantify the mRNA expression level, we counted the number of aligned fragments for each gene using HTSeq-0.6.1 with parameters (−s no, −r pos, −f bam, −m intersection-non-empty, and −t exon) according to the Ensembl transcript annotation (GRCh37 version) and calculated the fragments per kilobase per million mapped read (FPKM) values of each gene. Correlation between miRNA and mRNA expression was calculated by Spearman's correlation using R. miRNA and mRNA expression data from The Cancer Genome Atlas (TCGA) datasets 52 were also used to validate the patterns of DEMs and correlation between expression of miRNA and target mRNA expression.
Real-time quantitative PCR analysis with FFPE samples. Ten serial paraffin cuts obtained in an Eppendorf tube were deparaffinized in xylene. Total RNA was isolated using RNeasy Micro Kit (Qiagen, Germany) according to the manufacturer's instructions. RNA concentrations were measured using NanoDrop (Thermo Scientific, USA). Total RNA from each sample was reverse transcribed with the TaqMan MicroRNA Reverse Transcription kit (Thermo Fisher, USA). Reverse transcription was performed with the following thermal cycling parameters: 30 minutes at 16 °C, 30 minutes at 42 °C, and 5 minutes at 85 °C (Bio-Rad).