The promoter region of Discoidin, CUB and LCCL domain containing 2 (DCBLD2) was found to be aberrantly methylated in gastric cancer cell lines and in primary gastric cancers, as determined by restriction landmark genomic scanning. DCBLD2 expression was inversely correlated with DCBLD2 methylation in gastric cancer cell lines. Treatment with 5-aza-2′-deoxycytidine and trichostatin A partially reversed DCBLD2 methylation and restored gene expression in DCBLD2-silenced cell lines. In an independent series of 82 paired gastric cancers and adjacent normal tissues, DCBLD2 expression was down-regulated in 79% of gastric cancers as compared with normal tissues as measured by real-time reverse transcription-PCR. Pyrosequencing analysis of the DCBLD2 promoter region revealed abnormal hypermethylation in gastric cancers, and this hypermethylation was significantly correlated with down-regulation of DCBLD2 expression. Furthermore, ectopic expression of DCBLD2 in gastric cancer cell lines inhibited colony formation in both anchorage-dependent and anchorage-independent cultures and also inhibited invasion through the collagen matrix. These data suggest that down-regulation of DCBLD2, often associated with promoter hypermethylation, is a frequent event that may be related to the development of gastric cancer. (Mol Cancer Res 2008;6(2):222–30)
Advances in diagnostic and treatment technologies for gastric cancer have resulted in excellent long-term survival, but gastric cancer remains the second most common cause of cancer-related death worldwide (1). The molecular mechanisms underlying gastric cancer development and progression remain poorly understood. DNA methylation, associated with histone modification, is a key mechanism to inhibit the expression of tumor suppressor genes, metastasis suppressor genes, angiogenesis inhibitors, and other genes involved in tumor progression (2, 3). DNA methylation markers can be applied in cancer risk assessment and in the early detection, prognosis, and prediction of response to cancer therapy (4-6). To identify novel epigenetic targets in gastric cancer, we have studied global DNA methylation patterns in gastric cancer cell lines and gastric cancer tissues using restriction landmark genomic scanning (RLGS). RLGS is a highly reproducible two-dimensional gel electrophoresis of genomic DNA that allows the simultaneous assessment of more than 2,000 loci when the methylation-sensitive enzyme NotI is used as the landmark enzyme (7-10).
In this study, we used RLGS to identify DCBLD2 as a novel epigenetic target in gastric cancer. DCBLD2 (also known as ESDN or CLCP1) is a transmembrane protein first isolated from coronary artery cells, and its domain structure is similar to that of neuropilins (11, 12). The neuropilins were identified as isoforms of a specific vascular endothelial growth factor receptor (13), and soluble neurophilin-1 functions as an antagonist to vascular endothelial growth factor-165 and inhibits tumor angiogenesis and progression (14). The neuropilins are also receptors for axon guidance factors called semaphorins (15, 16), and some semaphorins, such as SEMA3B and SEMA3F, have been classified as tumor suppressors (17, 18). DCBLD2 is considered to play a role in regulation of cell proliferation. DCBLD2 overexpression leads to a decrease in cell proliferation in 293T cells (11) and vascular smooth muscle cells (19). Although little is known about the function of DCBLD2, previous data suggest that DCBLD2 may play an important role in cancer cell proliferation and metastasis. Here we report that DCBLD2 is frequently silenced by epigenetic mechanisms in gastric cancer and show its suppressive roles in gastric cancer cell proliferation and invasion.
Decreased DCBLD2 in Gastric Cancer by RLGS Analysis
We found a DNA spot that was decreased in 5 (SNU-016, SNU-520, SNU-601, SNU-620, and SNU-638) of 11 gastric cancer cell lines tested and in 4 of 15 primary gastric cancers compared with normal tissues profiled by RLGS. Figure 1A shows representative changes in the spot intensity in two gastric cancer cell lines (SNU-601 and SNU-620) and three primary gastric cancers (GC#73, GC#18, and GC#11), but no changes in two gastric cancer cell lines (SNU-484 and SNU-668). The decreased DNA spots corresponded to spot number 5B13 (GenBank accession no. CG465101), which we previously identified by mixing RLGS gels with NotI-linked clones (20). The sequence of clone 5B13 was matched to the NotI/EcoRV fragment covering the 5′ upstream region and exon 1 of DCBLD2 on human chromosome 3q12.1. The sequence also overlapped with a CpG island at the promoter region of DCBLD2 (Fig. 1B).8
The master RLGS profile that addresses spot 5B13 can be found at http://21cgenome.kribb.re.kr/html/2004_new/RLGS_master/image01.html (20).
DCBLD2 Promoter Hypermethylation Is Inversely Correlated with Its mRNA Expression in Gastric Cancer Cell Lines
To investigate a potential relationship between promoter methylation and down-regulation of DCBLD2 expression in gastric cancer, we carried out real-time reverse transcription-PCR (RT-PCR) for DCBLD2 using 11 gastric cancer cell lines. The level of DCBLD2 expression was relatively low in five cell lines (SNU-016, SNU-520, SNU-601, SNU-620, and SNU-638) with decreased 5B13 spot DNA in RLGS profiles (Fig. 2D, left). We next assessed the methylation status of each CpG site around the DCBLD2 CpG island (regions 1-3) by bisulfite sequencing. SNU-216 and SNU-484 cells that expressed normal levels of DCBLD2 preserved hypomethylated CpG dinucleotides. On the other hand, SNU-601 and SNU-620 cells that expressed DCBLD2 at low levels showed heavily methylated CpG sites. In addition, these CpG sites were hypomethylated in normal tissues but moderately methylated in paired tumor tissues (Fig. 2A). The methylation pattern of these tumor tissues suggests normal cell contamination, which commonly occurs in dissected tumors. H&E staining confirmed that contaminating normal cells comprised ∼60% of the cells in GC#011 and 40% in GC#410 (data not shown). Methylation of region 2 was inversely correlated with DCBLD2 expression, suggesting that region 2 may be crucial to regulate the transcription level of DCBLD2. We tested the three regions around the DCBLD2 CpG island for promoter activity by luciferase assay in SNU-216 or HeLa cells. Region 2 showed a remarkable increase in transcriptional activity, whereas region 1 and region 3 showed weak activity in both cell lines (Fig. 2B), suggesting that region 2 may contain a sequence critical for DCBLD2 silencing. Pyrosequencing analysis was also done to quantitate the methylation of seven DCBLD2 CpG sites at region 2 in 11 gastric cancer cell lines and selected gastric cancer tissues (Fig. 2C). The methylation status measured by pyrosequencing correlated with down-regulation of DCBLD2 in gastric cancer cell lines (Fig. 2D).
Restoration of DCBLD2 Expression by Treatment with 5-Aza-2′-Deoxycytidine and Trichostatin A
We next tested if the expression of DCBLD2 could be restored by treatment with the DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine (5-aza-dC; ref. 21) or the histone deacetylase inhibitor trichostatin A (TSA; ref. 22). SNU-601, SNU-620, and SNU-638 cells (in which DCBLD2 mRNA expression is down-regulated and DCBLD2 promoter is hypermethylated) were treated with 5-aza-dC and/or TSA. After treatment, cells were harvested and analyzed for changes in DCBLD2 methylation and expression compared with the untreated cells. Pyrosequencing revealed that the CpG sites of DCBLD2 were partially demethylated by 5-aza-dC and/or TSA in all three cell lines (Fig. 3A). In addition, real-time RT-PCR (Fig. 3B) and Western blotting (Fig. 3C) revealed an increase in DCBLD2 mRNA and protein expression. TSA was more effective than 5-aza-dC in SNU-620 and SNU-638. Combining 5-aza-dC with TSA had a synergistic effect in all cell lines tested (Figs. 3B and C). These results suggest that DCBLD2 expression in these gastric cancer cells was regulated by an epigenetic mechanism that includes both DNA methylation and histone deacetylation. Western blotting of primary gastric cancers (which were moderately methylated, as shown in Fig. 2A) and adjacent normal tissues (which were hypomethylated, as shown in Fig. 2A) revealed that methylation correlated with DCBLD2 protein expression in primary gastric cancer tissues (Fig. 3D).
DCBLD2 Is Frequently Silenced by CpG Methylation in Primary Gastric Cancers
We next analyzed DCBLD2 expression in a series of 82 paired gastric cancers and adjacent normal tissues by real-time RT-PCR. DCBLD2 expression was significantly reduced in gastric cancers compared with normal tissues (P = 0.00011; Fig. 4A). DCBLD2 expression was decreased in 79% (65 of 82) of gastric cancers compared with normal tissue and was at least 2-fold underrepresented in 51% (42 of 82) of gastric cancers. When the down-regulation of DCBLD2 was compared within each clinicopathologic category such as tumor depth (early versus advanced gastric cancer), tumor-node-metastasis staging, or Lauren's classification, no significant difference was observed (data not shown).
We next quantitated the methylation status of the seven CpG sites at region 2 by pyrosequencing analysis. The methylation levels in normal tissues were in the range of 0% to 8.6% (mean, 0.7%), whereas those in gastric cancers were in the range of 0% to 73.1% (mean, 12.2%), indicating that overall methylation was higher in gastric cancers (P < 0.0001; Fig. 4B). To see if the methylation of DCBLD2 CpG dinucleotides was associated with DCBLD2 expression in the clinical samples, we calculated a Pearson's correlation coefficient between relative expression and methylation and inferred its significance by a t test. For each pair of tumor and normal samples, we used log 2–transformed ratios between tumor and normal samples as the relative expression and defined methylation change as the difference in methylation between tumor and normal samples. We found a significant negative correlation between relative expression and methylation change (r = −0.2728, P = 0.0143; Fig. 4C), indicating that decreased DCBLD2 expression was associated with increased methylation in the clinical samples. When the abnormal methylation was compared within each clinicopathologic category, no difference was observed in any of the chosen variables such as gender, histology, or tumor stage (data not shown). In addition, we found no correlation between methylation and age in either the normal or gastric cancer samples (r = −0.0272, P = 0.5478 for normal tissue; r = −0.0404, P = 0.8039 for gastric cancers; Fig. 4D). This result indicates that abnormal methylation of DCBLD2 is a typical cancer-specific (type C), but not age-related (type A), occurrence (23).
Ectopic Expression of DCBLD2 Suppresses Colony Formation in Gastric Cancer Cell Lines
Frequent down-regulation of DCBLD2 in gastric cancers suggests that its dysfunction may play an important role during gastric carcinogenesis. To test if ectopic expression of DCBLD2 suppresses colony formation in gastric cancer cell lines, we transfected a full-length cDNA for DCBLD2 (pcDNA3-DCBLD2-HA) or empty vector (pcDNA3.1) into SNU-216, which endogenously expresses DCBLD2, or SNU-601, with reduced DCBLD2 expression. The expression of DCBLD2 in transfected cells was confirmed by Western blotting with anti-DCBLD2 (Fig. 5A). We selected drug-resistant cells for 2 weeks and carried out anchorage-dependent colony formation assays on monolayer cultures. The DCBLD2-transfected cells formed fewer colonies than did empty vector–transfected cells in both cell lines (Fig. 5B). Moreover, DCBLD2-transfected cells formed fewer colonies than did control cells in an anchorage-independent colony formation assay in soft agar (Fig. 5C). These results suggest that DCBLD2 suppresses cell proliferation signals in gastric cancer cells.
DCBLD2 Inhibits Gastric Cancer Cell Invasiveness In vitro
DCBLD2 is reportedly up-regulated in a highly metastatic lung cancer cell line (12). In this study, by contrast, in cell lines derived from metastatic gastric cancers, such as SNU-016, SNU-601, SNU-620, and SNU-638 (24, 25), DCBLD2 expression was down-regulated. To test if DCBLD2 affects gastric cancer cell invasiveness, we analyzed the pattern of invasion of SNU-601 cells through fibrillar collagen after transfection with pcDNA3-DCBLD2-HA or empty vector (Fig. 6A). DCBLD2-transfected cells were less invasive through the collagen matrix than control cells (P = 0.0016; Fig. 6B), suggesting that DCBLD2 inhibits the invasiveness of gastric cancer cells.
We report for the first time that DCBLD2 is a newly defined epigenetic target in gastric cancer related to tumor cell proliferation and invasiveness. We identified decreased representation of DCBLD2 in 26.7% (4 of 15) of primary gastric cancers as well as in 45.5% (5 of 11) of gastric cancer cell lines by RLGS analysis. Down-regulation of DCBLD2 expression was associated with promoter hypermethylation in gastric cancer cell lines, and its expression was restored in cells treated with inhibitors of DNA methyltransferase and histone deacetylase. In 51% of primary gastric cancers, DCBLD2 expression was decreased >2-fold. Furthermore, we frequently detected hypermethylation of the DCBLD2 promoter region in primary gastric cancers, and this hypermethylation was significantly correlated with DCBLD2 down-regulation. These results show that DCBLD2 is frequently down-regulated due to an epigenetic mechanism in gastric cancer, suggesting that this gene may play an important role in gastric carcinogenesis.
A novel molecular phenotype based on promoter CpG hypermethylation in colorectal cancers has been proposed by Toyota et al. (26). Genes with hypermethylated CpG dinucleotides were classified into two types of methylation: age related (type A) and cancer specific (type C). They also found frequent hypermethylation of type C genes in a subset of cancers and designated this as the CpG island methylator phenotype (CIMP). The CIMP+ phenotype was also identified in 24% to 47% of gastric cancers (23, 27), indicating that multiple promoter regions are methylated in many gastric cancers. We did not observe a correlation between DCBLD2 methylation and age, indicating that abnormal methylation of DCBLD2 is a typical type C response according to the previous classification (26).
DCBLD2 is a single-pass type I transmembrane protein that is highly conserved in mammals (11). DCBLD2 is expressed in various tissues including skeletal muscle, placenta, heart, colon, ovary, and prostate (12). It is particularly highly expressed in cultured vascular smooth muscle cells and is up-regulated by serum- and platelet-derived growth factor (11). Epidermal growth factor signaling also increases DCBLD2 and induces tyrosine phosphorylation of DCBLD2 in A431 human cancer cells (28). It has been suggested that DCBLD2 up-regulation leads to a decrease in growth rate by reducing the response to growth factors in a negative feedback loop (11, 19). Overexpression of DCBLD2 inhibits cell proliferation in 293T cells (11) and vascular smooth muscle cells (19), suggesting that the cellular level of DCBLD2 may play a role in the generation of tumor proliferation signals in gastric epithelial cells. To further characterize DCBLD2 function in vitro, we examined the effect of DCBLD2 overexpression in gastric cancer cells. Our results revealed that overexpression of DCBLD2 strongly inhibited both anchorage-dependent and anchorage-independent cell growth in vitro. Furthermore, overexpression of DCBLD2 significantly inhibited cell invasion through a collagen matrix, suggesting that DCBLD2 may suppress cell proliferation and invasion signals in gastric cancer.
Taken together, we propose that DCBLD2 is frequently hypermethylated in gastric cancer and that DCBLD2 is also related to at least two aspects of gastric carcinogenesis: tumor cell proliferation and invasion. Further studies are needed to evaluate the potential clinical application of hypermethylated DCBLD2 as a biomarker in gastric cancer. Our results also illustrate the need to study signaling mechanisms involved in DCBLD2 silencing during tumor progression and metastasis.
Materials and Methods
Cell Lines and Tissue Samples
Genomic DNA was extracted with the phenol-chloroform method (29), and RLGS was done as described (30). Briefly, genomic DNA (5 μg) was incubated in the presence of DNA polymerase I (Takara), ddTTP, ddATP, dGTP(αS), and dCTP(αS) to fill in randomly broken ends. The treated DNA was then digested with a methylation-sensitive enzyme, NotI (New England BioLabs), end-labeled with [α-32P]dGTP and [α-32P]dCTP using Sequenase (U.S. Biochemical), and subsequently digested with EcoRV (New England BioLabs). The labeled DNA (1.5 μg) was separated by size on a 0.8% agarose gel at 8 V/cm for 12 h for first-dimensional separation. Thereafter, DNA was digested in the gel with HinfI (New England BioLabs) to further fragment the DNA. The gel was fused with a 5% polyacrylamide gel by adding melted agarose to fill the gap. Second-dimension electrophoresis was carried out at 8 V/cm for 7 h. The gel was dried and exposed to X-ray film. RLGS gels were run for paired samples of primary gastric cancer and adjacent normal tissue. For cell line DNA, RLGS gels were also run in pairs consisting of cell line DNA alone and DNA of the cell line mixed with DNA from normal tissue. The differences between the two profiles of tumor and normal DNA or of cell line and normal DNA were detected as described (31). Once a difference in spot intensity was detected between paired normal and tumor samples or between normal and cell line samples, we compared the spots with the previous master RLGS profile (7) or our RLGS profile (20) to identify their sequences.
Real-time RT-PCR Analysis
Total RNA was prepared using the RNeasy kit (Qiagen) according to the manufacturer's guidelines and treated with DNase I (Promega). DNase-treated RNA (5 μg) was reverse transcribed with Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's guidelines. Real-time PCR was done in an Exicycler Quantitative Thermal Block (Bioneer). The reverse transcribed product (100 ng) was amplified in a 15-μL reaction with 2× SYBR Premix EX Taq (Takara) using the DCBLD2 primer set, 5′-CTCAGCCACTGGTAGGAGGA-3′ and 5′-GGCACCTGGTACACCAATTC-3′. The product size was 394 bp. Samples were heated to 95°C for 1 min and then amplified for 45 cycles consisting of 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. All reactions were then incubated at 72°C for 10 min and cooled to 4°C. The β-actin gene was amplified as a control. Relative quantification of DCBLD2 mRNA was analyzed by comparative threshold cycle (CT) methods (32).
Bisulfite Sequencing Analysis
Genomic DNA (1 μg) was modified by sodium bisulfite using the EZ DNA Methylation kit (ZYMO Research) according to the manufacturer's instructions. Bisulfite-modified DNA was amplified using three primer sets designed to amplify three regions of interest. The primer sequences are 5′-AAAAAGAGTAGAGAAAATAGGGAAT-3′ and 5′-CCTCAACCCAAACTAATCC-3′ (for region 1; 342 bp), 5′-GGGATTAGTTTGGGTTGAG-3′ and 5′-CTTTCACCTACTCCTCCTTC-3′ (for region 2; 531 bp), and 5′-GAAGGAGGAGTAGGTGAAAG-3′ and 5′-CACTCACCTTACTAAACTCCAA-3′ (for region 3; 388 bp). Samples were heated to 95°C for 12 min and then amplified for 35 cycles consisting of 95°C for 45 s, 55°C for 35 s, and 72°C for 60 s. All reactions were then incubated at 72°C for 10 min and cooled to 4°C. The PCR products were visualized on a 1.5% agarose gel by ethidium bromide staining, purified from the gel using the Qiagen Gel Extraction kit, and cloned using the pGEM-T Easy Vector (Promega). Twelve clones were randomly chosen for sequencing. Complete bisulfite conversion was verified by the fact that <0.01% of the cytosines in non-CG dinucleotides was unconverted in the final sequence.
Promoter Reporter Assay
DNA fragments around the DCBLD2 CpG island were obtained by PCR using primer sets designed to amplify three regions of interest. The primer sequences are 5′-CCGCCATAAAAAGAGCAGAG-3′ and 5′-AGCCAGCTCGTCAGTGACTT-3′ (for region 1; 318 bp), 5′-AAGCTTGCTGAGGGCCCAACAATAAC-3′ and 5′-CGCCATGGCGGAGCTAAGGAACGTG-3′ (for region 2; 401 bp), and 5′-gastric cancerTCGAGGACTCCGCTTC-3′ and 5′-GGAGAAGGAGGAGGAGTTGG-3′ (for region 3; 394 bp). The DNA fragments were inserted into the pGEM-T easy vector (Promega) and confirmed by sequencing. Then, the SacI/NcoI (for regions 1 and 3) or HindIII/NcoI (for region 2) fragment was subcloned into the pGL3-Basic vector (Promega). Each construct or a control empty vector and an internal control pRL-CMV vector (Promega) were transfected into SNU-216 or HeLa cells using Lipofectamine Plus reagent (Invitrogen) according to the manufacturer's protocol. Firefly and Renilla luciferase activities were measured 48 h after transfection. Relative luciferase activities were calculated after normalization of the transfection efficiency by Renilla luciferase activity.
The promoter region of DCBLD2 was amplified using the forward primer 5′-AGTAAGGAGTAAGTTGGGTTTGA-3′ and the biotinylated reverse primer 5′-TCCCCTTCCACCTTTAAAA-3′, designed by PSQ Assay Design (Biotage AB). The product size was 218 bp. Bisulfite-modified DNA was amplified in a 25-μL reaction with the primer set and f-Taq polymerase (Solgent). Samples were heated to 95°C for 5 min and then amplified for 50 cycles consisting of 95°C for 30 s, 58°C for 40 s, and 72°C for 30 s, followed by a final extension step at 72°C for 5 min. Pyrosequencing reactions were done with a sequencing primer (5′-GAGGGTTTAATAATAATAGG-3′) on the PSQ HS 96A System (Biotage AB) according to the manufacturer's specifications.
5-Aza-dC and TSA Treatment
Gastric cancer cells (SNU-601, SNU-620, and SNU-638) were seeded in 10-cm dishes at a density of 1 × 106 per dish 1 day before the drug treatment. The cells were treated with 1 μmol/L 5-aza-dC (Sigma) every 24 h for 3 days and then harvested. Another culture of cells was treated with 500 nmol/L TSA (Sigma) for 1 day. To test the combined effect of 5-aza-dC and TSA, cells were treated with 1 μmol/L 5-aza-dC for 3 days followed by treatment with 250 nmol/L TSA for 1 day. DNA was prepared and tested for reversion of DCBLD2 methylation by pyrosequencing. Total RNA and protein were prepared and tested for restoration of DCBLD2 expression by real-time RT-PCR and Western blotting.
Transfection and Colony Formation Assay
pcDNA3-DCBLD2-HA is a COOH-terminal hemagglutinin (HA)-tagged full-length DCBLD2 expression vector that was constructed in the laboratory of Dr. T. Takahashi (12). For colony formation assays in a monolayer culture, equimolar amounts of pcDNA3-DCBLD2-HA or empty control vector (pcDNA 3.1) were transfected into SNU-216 or SNU-601 cells in six-well plates using Lipofectamine Plus reagent (Invitrogen) according to the manufacturer's protocol. The cells were selected with G418 (300 μg/mL) for 2 weeks and then plated into fresh six-well plates at 1 × 104 per well. After 2 weeks of incubation with G418, colonies were stained with crystal violet and counted. To investigate colony formation in soft agar, cells were transfected and selected as above. The cells were suspended in RPMI containing 0.3% agarose (Sigma, A-9045), 10% fetal bovine serum, and 300 μg/mL G418 and layered on RPMI containing 0.6% agarose, 10% fetal bovine serum, and G418 in a six-well plate. Colonies were photographed and counted after 2 weeks of incubation at 37°C. The expression of DCBLD2 was confirmed by Western blotting with anti-DCBLD2 (Sigma). α-Tubulin was analyzed as a control in the same samples.
Invasion assays were done as described (33). Fibrillar collagen type I gels (1 mg/mL) were prepared by neutralizing a solution of collagen type I (Vitrogen 100, Cohesion) in 8-μm-pore, 6.5-mm polycarbonate transwells (Corning) for 1 h at 37°C. SNU-601 cells (4 × 104) transfected with pcDNA3.1 or pcDNA3-DCBLD2-HA in RPMI containing 0.2% fetal bovine serum were seeded on top of the collagen gel. Transwell inserts were placed in RPMI supplemented with 10% fetal bovine serum and 30 ng/mL EGF to provide a chemotactic gradient and were incubated at 37°C for 5 days. Whole cells were stained with 4 μmol/L calcein-acetoxymethyl ester (Invitrogen C1430) in serum-free RPMI and visualized by confocal microscopy using a 10× objective. Optical sections (Z sections) were scanned at 15-μm intervals moving down from the top of the collagen gel. The quantification of invading cells at each section was evaluated using ImageJ software from the Research Service Branch website of the NIH.10
The Student unpaired t test was used to test differences in DCBLD2 expression or DCBLD2 promoter methylation between primary gastric cancers and adjacent normal tissues. Results for which P < 0.05 were considered significant. Correlation between the level of DCBLD2 expression and DCBLD2 CpG methylation or between methylation and aging was determined using the Pearson's correlation coefficient (r), and its significance was inferred from a t test using a t value calculated from the following formula: t = r[(n − 2) / (1 − r2)]1/2 with n − 2 degrees of freedom. The clinicopathologic factors in various groups of patients with positive or negative DCBLD2 expression and positive or negative DCBLD2 promoter methylation were compared using the χ2 test.
Grant support: FG06-11-01 of the 21C Frontier Functional Human Genome Project from the Ministry of Science and Technology of Korea.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. T. Takahashi (Nagoya University Graduate School of Medicine, Nagoya, Japan) for the HA-tagged DCBLD2 expression vector.