1. Volume 144, Issue 1, Pages 122-133.e9 (January 2013)
Helicobacter pylori Causes Epigenetic Dysregulation of FOXD3 to Promote Gastric Carcinogenesis 
Alfred S.L. Cheng, May S. Li, Wei Kang, Victoria Y. Cheng, Jian–Liang Chou, Suki S. Lau, Minnie Y. Go, Ching C. Lee, Thomas K. Ling, Enders K. Ng, Jun Yu, Tim H. Huang, Ka F. To, Michael W. Chan, Joseph J.Y. Sung, Francis K.L. Chan 
Gastroenterology 
Volume 144, Issue 1, Pages e9 (January 2013)
DOI: /j.gastro
Copyright © 2013 AGA Institute Terms and Conditions

2. Figure 1
Integrative epigenomics analysis reveals FOXD3 as a link between H pylori infection and gastric cancer progression. (A) Overview of the cross-species epigenomics approach used to nominate causal hypermethylated genes with pathological relevance. DNA methylation profiling in a mouse model of H pylori infection and human gastric cancer specimens were performed using MethylCap-microarray and microarray-based methylation assessment of single samples, respectively. The direct target genes of the common hypermethylated target, FOXD3, were identified by ChIP-chip. (B) MethylCap-microarray landscaping map by plotting the enrichment ratio of each probe of the murine Foxd3 gene. Pooled H pylori–induced gastritis samples displayed Foxd3 hypermethylation compared with uninfected controls (n = 3). The methylation region located within the CpG island (green bar) near the TSS (arrow), which was validated by MethylCap-qPCR using specific primers (upper right inset). The locations of qPCR and bisulfite genomic sequencing (BGS) are shown in red arrows and black bar, respectively. (C) Foxd3 promoter methylation was confirmed by BGS from individual murine samples (n = 5). The methylation levels of each of the 8 CpG sites are depicted in a blue color scale. The overall percentages of methylation are shown. (D) Concurrent MethylCap-qPCR (left panel) and qRT-PCR (right panel) analyses of murine samples. Red lines show median methylation and mean expression levels. (E) Upper panel: quantitative methylation-specific PCR analysis of FOXD3 promoter methylation levels in human gastric mucosal tissues (N = 139). Red lines show median methylation levels. Lower panel: qMSP analysis of human gastric cancers vs preneoplastic adjacent tissues (65 pairs). (F) Kaplan–Meier survival curve shows that gastric cancer patients with high FOXD3 methylation (red line) had poorer survival than those with low methylation. **P < 0.01; ***P <
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

3. Figure 2
Reactivation of FOXD3 after genome demethylation. (A) qRT-PCR analysis of FOXD3 mRNA expression in gastric cancer cell lines (white bars) and normal stomach tissues (black bars). Histogram shows mean fold-change of cell lines relative to normal tissues. (B) Quantitative methylation-specific PCR analysis of FOXD3 methylation levels in gastric cancer cell lines. (C) Bisulfite pyrosequencing (upper panel) and semi-quantitative RT-PCR (lower panel) analyses of FOXD3 methylation and mRNA expression in gastric cancer cell lines treated with or without 5-aza-2′-deoxycytidine. Histogram shows mean methylation levels (9 CpG sites) relative to mock-treated cells. (D) Western blot analysis of FOXD3 protein expression in gastric cancer cell lines and normal stomach tissues (upper panel) and in gastric tumor (T) and paired non-tumor (NT) tissues (lower panel). Histogram shows mean fold-change of cell lines relative to normal tissues. (E) Representative pictures of FOXD3 Immunohistochemistry in normal human stomach, H pylori–positive gastritis (magnification 200×), gastric cancer, and adjacent normal-looking mucosa (magnification 400×). Insets indicate uniformity and loss of FOXD3 expression in epithelium of normal and gastritis tissues, respectively. (F) Histogram shows the proportion of gastric cancers (N = 113) and normal stomachs (N = 5) with different levels of FOXD3 nuclear expression. ***P < .005.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

4. Figure 3
FOXD3 exhibits tumor-suppressive functions in gastric cancer cells. (A) Western blot analysis of FOXD3 and cleaved PARP protein in control and FOXD3-overexpressing gastric cancer cells. (B) Cellular proliferation of control and FOXD3-overexpressing gastric cancer cells were determined by MTS and (C) colony-formation assays. Representative images of foci formation are shown. (D) Cell invasion of control and FOXD3-overexpressing gastric cancer cells was determined using Matrigel chambers. Representative images of Gentian violet–stained invaded cells are shown. (E) FOXD3 inhibits tumor growth in vivo. (E1) Western blot analysis and TUNEL assay of control and FOXD3-overexpressing MKN45 cells. (E2) Upper panel: Mean xenograft tumor volumes were plotted against days after injection (n = 10). Lower panel: Weights of the excised xenograft tumors (as shown in the image). Black lines show mean tumor weights. (E3) TUNEL assays of xenograft tumors. Data are expressed in TUNEL-positive cell per high-power fields (HPF). *P < .05; **P < .01; ***P < .005.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

5. Figure 4
Identification of novel FOXD3 direct targets in gastric cancer cells. (A) Immunohistochemical localization of FOXD3 in the nuclei of some MKN1 and N87 gastric cancer cells (original magnification 400×). No staining was observed in the negative control with the primary antibody omitted. (B) Knockdown of FOXD3 by RNA interference (RNAi) in gastric cancer cells increases cell proliferation and decreases apoptosis as determined by MTS assay and Western blot analysis of cleaved PARP, respectively. ***P < (C) FOXD3-binding maps of FOXD3 target genes in N87 cells. The X-axis and Y-axis represent the probe location relative to TSS and the FOXD3 enrichment ratio (IP/input), respectively. (D) Gene ontology analysis of the novel FOXD3 target genes identified by ChIP-chip. The P value and gene number of each enriched gene ontology category are listed.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

6. Figure 5
Direct transcriptional regulation of cell death regulators by FOXD3 in gastric cancer cells. (A) FOXD3-binding maps of CYFIP2 and RARB as described in Figure 4C. ChIP-PCR showed strong enrichment by FOXD3 antibody relative to IgG control in both MKN1 and N87 cells. Input: 2% of total chromatin. (B) Maps of CYFIP2 and RARB promoter-luciferase (LUC) reporter constructs. The FOXD3-bound promoter region of human CYFIP2 or RARB containing the putative FOXD3 consensus sequence22 (stripped box) was cloned to a LUC reporter. The numbers indicate the base-pairs relative to TSS. (C) The elevation of CYFIP2 and RARB promoter activity by FOXD3 overexpression in gastric cancer cells. (D) qRT-PCR analysis of CYFIP2 and RARB in gastric cancer cells after overexpression or (E) RNAi-mediated knockdown of FOXD3. *P < .05; **P < .01; ***P < .005.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

7. Figure 6
Concordant down-regulation of FOXD3, CYFIP2, and RARB in human gastric carcinogenesis. (A) qRT-PCR analysis of FOXD3, CYFIP2, and RARB mRNA levels in human normal stomach, IM, and gastric cancer tissues. (B) Correlation among FOXD3, CYFIP2, and RARB mRNA levels in 65 human gastric mucosal tissues denoted with Spearman correlation coefficients. (C) qRT-PCR analysis of Cyfip2 (upper panel) and Rarb (lower panel) mRNA expression in control and H pylori–induced gastritis samples (n = 10). *P < .05; ***P < (D) A model of epigenetic disruption of tumor-suppressive cascade in gastric cancer. Transcriptional control of tumor suppressors by FOXD3 is epigenetically deregulated by H pylori infection–induced promoter methylation, which can cause inappropriate cell survival leading to gastric cancer development.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

8. Supplementary Figure 1
MethylCap-microarray analysis using a mouse model of H pylori infection. (A) Verification of H pylori infection by bacterial-specific PCR and histology. At 40-week post-infection, the stomach tissues of H pylori–infected and uninfected control mice (n = 3) were obtained. DNA was extracted for PCR analysis using specific primers against CagA, a major H pylori virulence protein. Histological examination of H pylori–infected stomach revealed chronic gastritis as characterized by a variable inflammatory cell infiltrate (arrow). Rapid urease test also demonstrated viable H pylori from the infected group (data not shown). Pooled DNA was used for MethylCap-microarray. (B) Gene ontology analysis of the novel H pylori infection–hypermethylated genes identified by MethylCap-microarray. Histogram shows the number of genes in the significantly enriched cancer-related categories. P values are listed above the corresponding categories. (C) MethylCap-microarray landscaping map by plotting the enrichment ratio of each probe of the Pcdh10 gene. The H pylori–induced gastritis samples displayed an enriched methylation region overlapping with the CpG-island (green), located in proximity of the TSS. Methylation status of Pcdh10 promoter region was confirmed by qPCR using the affinity-purified methylated DNA. (D) A list of cell differentiation-related hypermethylated genes that were reported to be silenced by promoter methylation in human gastric cancer and other malignancies. (E) Effect of H pylori coculture on FOXD3 methylation in gastric epithelial cells. MKN45 was cocultured with H pylori SS1 (bacteria-to-cell ratio of 100:1) for 48 hours, followed by FOXD3 methylation assessment using MethylCap-qPCR. H pylori infection significantly increased FOXD3 methylation level in MKN45 cells. As control, GAPDH promoter region showed low methylation levels with or without H pylori infection, thus demonstrating the specificity of methylation enrichment. ***P < Note: Our findings indicated that methylation in a discrete promoter region may modulate Foxd3 gene expression (Figure 1B to D) and Pcdh10 (Supplementary Figure 1C) during H pylori infection. These results are supported by recent studies showing that promoter methylation in only a single or cluster of CpG sites can be functionally important for transcriptional regulation.19–21 For example, methylation of 3 CpG sites, rather than the whole CpG-island, was shown to cause gene silencing of a proapoptotic regulator XAF1 during gastric cancer progression.19 MKN45 was widely used as a gastric epithelial cell model to investigate the effects of H pylori.22–24 Our in vitro finding indicates that H pylori may directly induce FOXD3 promoter methylation in gastric epithelial cells.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

9. Supplementary Figure 2
Microarray-based methylation assessment of single samples (MMASS) analysis of human gastric cancer specimens. (A) Mean survival years of the 2 groups of gastric cancer patients used for MMASS analysis. The long and short survivors were those who survived 5 years or more and died of disease before 5 years, respectively. ***P < (B) PAM analysis of the differentially methylated genes in gastric cancer specimens. Bar chart shows the relative levels of the significant methylated genes in gastric cancers from patients with short (red)- and long (blue)-term survival. (C) A list of differentially methylated genes that were reported to be silenced by promoter methylation in human gastric cancer and other malignancies.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

10. Supplementary Figure 3
FOXD3 promoter methylation analyses in gastric cancer cell lines. (A) Determination of FOXD3 differential methylated promoter region for qMSP analysis. Gene map of human FOXD3 depicting the locations of the FOXD3 CpG probe (blue bar) of the human CpG 12.1K arrays, qMSP (red bar), pyrosequencing (blue bar), MethylCap-qPCR (purple bar) and BGS/combined bisulfite restriction analysis (COBRA) (black bar) regions. The CpG-island (green), TSS (arrow) and exon (white box) of the FOXD3 gene are also shown. BGS results showed dense methylation in AGS and KATOIII, but not MKN45 gastric cancer cells in the proximal FOXD3 promoter region (−500 to −300 bp relative to TSS). The methylated and unmethylated CpG sites are depicted as black and white circles, respectively. Histogram shows the FOXD3 mRNA levels of the gastric cancer cell lines as measured by qRT-PCR. (B) Gel electrophoresis of COBRA assays using DNA from 5 gastric cancer cell lines and IVD digested by methylation sensitive enzymes AciI (A), BstUI (B) or mock-digested (uncut [UC]). Concordant with the BGS results, COBRA demonstrated that MKN45 but not the other cell lines exhibited unmethylation in the examined FOXD3 promoter region. The promoter region for qMSP analysis was chosen based on the BGS/COBRA and qRT-PCR results, where FOXD3 promoter hypermethylation in AGS and KATOIII cells was associated with decreased mRNA expression when compared with MKN45 cells. (C) Verification of FOXD3 methylation assessment by bisulfite pyrosequencing. FOXD3 methylation levels of 3 gastric cancer cell lines (AGS: red; MKN28: blue; MKN45: black) measured by pyrosequencing and quantitative methylation-specific PCR were expressed as percentages of IVD. The results from these 2 methods were highly concordant as denoted by the Pearson correlation coefficient. (D) qRT-PCR analysis of FOXD3 mRNA expression in gastric cancer cell lines treated with or without 5-aza-2′-deoxycytidine. No detectable FOXD3 re-expression in SNU16 cells upon DNA demethylation was observed (data not shown).
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

11. Supplementary Figure 4
Bisulfite pyrosequencing analysis of human gastric mucosal tissues. To validate the FOXD3 methylation assessment by quantitative methylation-specific PCR analysis in Figure 1E, 5 samples from each category of normal, H pylori, IM, and gastric cancer were selected for pyrosequencing. The methylation levels (% of IVD) of each of the 9 CpG sites at −488 to −390 base-pair relative to the FOXD3 TSS are depicted in a blue color scale. Scatter plot shows the overall percentages of methylation of the individual samples in different groups. Red lines show median methylation levels. **P < Note: Pyrosequencing demonstrated a progressive increase in FOXD3 methylation levels in human gastric carcinogenesis in accordance with the quantitative methylation-specific PCR findings. Increased FOXD3 methylation level was detected in H pylori–positive gastritis tissues compared with uninfected normal controls (P < 0.01). FOXD3 promoter methylation was significantly increased to a similar degree in IM tissues (P < 0.01), but further elevated in gastric cancers (P < 0.01).
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

12. Supplementary Figure 5
FOXD3 target genes in gastric cancer cell lines. (A) Western blot analysis of FOXD3 protein expression in gastric cancer cell lines and normal stomach tissues. Higher exposure time demonstrated previously undetectable FOXD3 expression in some low-FOXD3–expressing cell lines. (B) ChIP-PCR analysis of FOXD3 occupancy in the promoter of target genes in MKN1 and N87 gastric cancer cells. Strong enrichment by FOXD3 antibody relative to the control antibody against irrelevant IgG was observed. Input: 2% of total chromatin was PCR-amplified with the same primers. (C) Potential role of GATA3 in RARB gene regulation. Left panel: Schematic diagram showing close proximity of FOXD3 and GATA3 binding sites in the proximal promoter of RARB. The numbers indicate the base-pairs relative to TSS. Right panel: qRT-PCR analysis of GATA3 mRNA expression in gastric cancer cell lines. *P < Note: Precise gene expression depends on the combinatorial interplay of transcription factors with cis-regulatory DNA elements.7 Major transcription factor families such as FOX and GATA have been reported to regulate gene expression within the intestinal epithelium.29 A recent study depicted the potential involvement of GATA3 in the organogenesis of stomach,30 where interaction between GATA3 and FOX factors might occur. Interestingly, an inverse relationship between GATA3 and RARB protein expression seems to appear in human gastric cancer according to the Human Protein Atlas database: GATA3 is shown to be highly expressed in nearly all examined gastric cancer tissues (strong nuclear staining in >75% of tumor cells) ( while RARB is rarely or sparsely expressed (<25% of tumor cells, Given GATA3 functions as a transcriptional repressor during T-cell development,31 GATA3 may repress RARB expression in gastric cancer cells. By transcription factor binding site analysis using TRANSFAC database, we found that FOXD3 and GATA3 binding sites locate within close proximity (16-bp) to each other in the RARB proximal promoter (∼800-bp relative to TSS). We speculate that the relatively high expression of GATA3 in N87 cells (compared with MKN1 cells) may inhibit RARB transcriptional activation by FOXD3; thus resulting in promoter inactivity (Figure 5C and D). The potential molecular interaction between FOXD3 and GATA3 in gastric cancer cells awaits further investigation.
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions

13. Supplementary Figure 6
Correlation between FOXD3 promoter methylation and expression in human gastric mucosal tissues. Concurrent pyrosequencing and qRT-PCR analyses of human normal stomach, IM, and gastric cancer tissues (n = 5). Significant correlation between FOXD3 methylation and mRNA levels in these clinical samples is denoted by the Spearman correlation coefficient (P = 0.013).
Gastroenterology  , e9DOI: ( /j.gastro )
Copyright © 2013 AGA Institute Terms and Conditions