Human gastric carcinoma cells were obtained from the Korean Cell Line Bank (KCLB; Seoul, Korea). All cell lines were authenticated by short tandem repeat (STR) analysis at the characterized cell line core facility at Abion, Inc., (Seoul, Korea) during the study. The gastric carcinoma cells were maintained in RPMI-1640 medium (HyClone Laboratories, Inc., Logan, UT, USA), supplemented with 10% fetal bovine serum (FBS; HyClone Laboratories) and 1% penicillin/streptomycin (HyClone Laboratories). Cells were grown in a humidified atmosphere with 5% CO₂ at 37°C and routinely tested for mycoplasma infection using Myco VALiD Mycoplasma PCR detection kit (Intron Biotechnology, Gyeonggi-do, Korea).

Cells were transfected with MZF1 expression plasmid using FuGENE HD transfection reagent (Promega, Madison, WI, USA) and then incubated for 24 h. si-SMAD4 duplexes were synthesized by Invitrogen using the following sequence: 5′-GGU CAG CCA GCU ACU UAC CAU CAU A-3′. si-MZF1 duplexes were synthesized by Cosmo Genetech Co., Ltd., (Seoul, Korea) using the following sequence: 5′-CUA CUG UAG GUG UCC AAU A-3′) (35). Co-transfection of siRNA and plasmid was performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instruction.

Luciferase reporter constructs were cloned using the restriction map of the BAC729G3 bacterial artificial chromosome (BAC) clone from the RPCI-11 human BAC library (Invitrogen), which covers the alternative promoter region of SMAD4. The −1752 bp upstream region of the SMAD4 transcription start site and two intra-gene region (+20 to +427 and +15422 to +16746) were subcloned into the pGL3 basic vector (Promega). Deletion and mutant constructs of putative SMAD4 promoters were generated by PCR. For generation of point mutation, Pfu-DNA polymerase and DpnI were used. All constructs were confirmed by sequencing.

After 24 h of transfection, cells were lysed with luciferase assay buffer. Then, the luciferase activity was measured using the Dual-luciferase reporter assay system according to the manufacturer's instructions (Promega) and was followed by luminescence measurement in a GENios Pro microplate reader (Tecan Trading AG, Mannedorf, Switzerland).

Nuclear extract of MKN74 cells were prepared by Qproteome Nuclear Protein kit (Qiagen, Hilden, Germany) and quantified by Pierce BCA assay kit (Thermo Fisher Scientific, Waltham, MA, USA). DNA mobility shift assays were performed with the following double-stranded oligonucleotides: MZF1 5′-CTC GGA GCG GGA GGC GGG GGC AGC CGG GAG AAA GG-3′. Two complementary oligonucleotides (1000 pmol of each) were annealed and 5 pmol of the annealed oligonucleotides were 5′-end-labeled with [γ-³²P]-ATP (Amersham Biosciences, Uppsala, Sweden) using T₄ polynucleotide kinase (New England Biolabs, Inc., Ipswich, MA, USA). Labeled products were purified on a Sephadex G-25 column (Amersham Biosciences). For antibody supershift analysis, nuclear extracts were incubated for 30 min in binding buffer containing 7.5% glycerol, 15 mM Tris-HCl, pH 7.5, 75 mM NaCl, 1.5 mM EDTA, pH 8.0, 1.5 mM dithiothreitol, 0.3% Nonidet P-40 and 1 µg of poly (dI-dC), and then with the probe for 40 min at 37°C, and with 2 µg of antibodies overnight at 4°C or −20°C. The primary antibodies used were as follows: His-probe (sc-8036) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and MZF1 (ab64866) was purchased from Abcam (Cambridge, UK). For the competition assays, 100-fold excess amount of unlabeled competitor was premixed with the radiolabeled probe before addition of the binding mixture. DNA-protein complexes were resolved on 6% non-denaturing PAGE gel at 250 V for 2.5 h. After separation, the gels were dried and exposed to the phosphor screen. The relevant protein-DNA probe complexes were analyzed by a BAS-1500 Image Analyzer (Fujifilm, Tokyo, Japan). ChIP assays were performed using the EZ-ChIP Kit (Millipore, Bedford, MA, USA) according to the manufacturer's instructions. Immunoprecipitation was performed with MZF1 antibody or rabbit IgG antibody (Santa Cruz Biotechnology). DNA was analyzed by conventional PCR directed to specific regions of the SMAD4 promoter and were amplified using the respective forward and reverse primers: ChIP I (forward) 5′-CTCCCTCAAACAGGCCTTCGC-3′ and (reverse) 5′-CAG CTT TCC TTT CTC CCG GCT-3′; ChIP II (forward) 5′-AGC CGG GAG AAA GGA AAG CTG-3′ and (reverse) 5′-CCA AAC CGC TCC GTT ACC GCA-3′. PCR was performed for 30 cycles at 94°C (30 sec), 60°C (30 sec) and 72°C (30 sec).

Total RNA was extracted by TRIzol (Invitrogen) and reversely transcribed to cDNA using the SuperScript II First-Strand Synthesis system (Invitrogen). Following cDNA synthesis, qRT-PCR was performed as described in a dual system LightCycler (Roche Diagnostics) and the expression levels of target genes relative to HPRT (control) were determined by a SYBR-Green-based comparative CT method (relative fold change = 2^−ΔΔCT). Primers used are as follows: SMAD4 5′-TGG CCC AGG ATC AGT AGG T-3′ (forward) and 5′-CAT CAA CAC CAA TTC CAG CA-3′ (reverse); CTBP1 5′-ACT GCG TGA CCC TGC ACT-3′ (forward) and 5′-GCC CCT TGT CTC ATC TGC-3′ (reverse). All primers were purchased from Cosmo Genetech Co., Ltd. (Seoul, Korea).

Whole cell lysates were prepared in RIPA buffer supplemented with protease inhibitor cocktail (Roche Diagnostics). Lysates were centrifuged at 4°C, 15,000 rpm for 20 min. Equal amounts of protein samples were electrophoretically separated by SDS-PAGE and transferred to nitrocellulose membrane. Membranes were blocked in TBS-T (Tris-buffered saline with 0.05% Tween-20) containing 5% non-fat dry milk and then incubated overnight at 4°C with the primary antibodies [SMAD4 (sc-7966), MZF1, GAPDH (FL-335)] diluted in the same buffer. GAPDH was used as loading control. Membranes were washed with TBS-T and incubated with horseradish peroxidase-conjugated secondary antibodies (Pierce, Rockford, IL, USA). The results were visualized with ECL reaction.

MatInspector (Genomatix Software GmbH, Munich, Germany; http://www.genomatix.de) was used to locate regulatory elements within the aforementioned core promoter region, and the internet-based TFSEARCH: Searching TFBS program (http://www.cbrc.jp/research/db/TFSEARCH.html) was used to localize the putative transcription factor binding sites within the 5′-flanking region of SMAD4. Alignment of human and mouse promoter sequences was performed with NCBI's Ensemble interface. Furthermore, mouse and chimpanzee SMAD4 promoter sequences were compared with human genomic sequences for conservation of the MZF1 binding motif.

Cell proliferation assays were performed using the EZ-Cytox kit (Daeil Lab Service, Co., Ltd., Seoul, Korea) according to the manufacturer's instructions. WST assays were performed as previously described (37). Cell were seeded at the density of 1×10⁵ cells and transfected with the expression vector.

Cell migration was analyzed using 24-well Transwell plates with polycarbonate membranes (Corning Costar, Corning, Ny, USA). For Transwell migration assay, cells were transfected with expression vectors encoding wild-type (WT)-MZF1 or si-MZF1. Cells were prepared 24 h post-transfection and then loaded into the upper compartment. After incubation for 24 h at 37°C, the cell number was detected with a GENios Pro microplate reader (Tecan Trading AG) using 485/535 nm filter set as previously described. The migration assay was performed in at least three independent experiments. Values are expressed as percentages compared to the control. The in vitro wound-healing assay was performed to examine the migration on gastric cancer cells transfected with WT-MZF1 vectors or si-MZF1. Transfected cells were grown on 96-well plates with their respective culture media. After the growing cell layers had reached confluence, wounds were prepared by a single scratch on the monolayer using a wound maker and the wounded layers were washed with phosphate-buffered saline (PBS) to remove the cell debris. Cell plates were applied to the IncuCyte ZOOM (Essen BioScience, Inc., Ann Arbor, MI, USA) and scanned every 1 h for 24 or 48 h. The wound-healing assay was performed in triplicate in at least three independent experiments.

The results were compared using one-way ANOVA analysis followed by the Turkey's test for multiple comparisons. Means were considered significant, at P<0.05. Statistical analysis was performed using a GraphPad Prism package for personal computers (GraphPad Software, Inc., San Diego, CA, USA). Results were considered significant at P<0.05, P<0.01 or P<0.001. All the data with error bars are presented as mean ± SD for at least three independent experiments.

To explore the transcriptional regulator candidates of the SMAD4 promoter, we investigated the region from −1752 to +84 (1836 bp) to the transcription start site using bioinformatic tools, such as MatInspector professional software and the internet-based TFSEARCH database. In our previous study, we identified the transcriptional start site of SMAD4 by reverse transcription-PCR and nucleotide sequencing, and we reported that hypermethylation of CpG site within this region might be related to the transcriptional silencing of SMAD4 (37). Our previous results on the identity of the transcription start site were consistent with the European Molecular Biology Laboratory database (http://www.ensembl.org), despite of the presence of several alternative transcripts encoding exon 1 and exon 2 of SMAD4. In this analysis, we screened three transcription factors, HSF1, RUNX1 (AML-1) and MZF1, which have a higher likelihood of interaction than other transcription factor candidates (Fig. 1A). According to previous reports, we first performed a comparison analysis of reported smad4 promoter regions in HEK293T and SNU638. HEK293T cell was used as positive control (25). The basal promoter activity of these three different SMAD4 promoter regions was investigated using a vector constructs: Luc-1752 (−1752 to +84), Roth et al (38) (+20 to +427) and Minami et al (39) (+15422 to 16746). As shown in Fig. 1B, luciferase activity of the −1752 to +84 (Luc-1752) construct was nearly 80 to 150-fold increased compared with the pGL3 control, whereas the other constructs did not have effect on SMAD4 promoter activity. Our data suggested that the −1752 to +84 region is essential for basal SMAD4 promoter activity, we further concentrated on the role of this region. The transcriptional activity of these three transcription factors on the SMAD4 promoter region was investigated using a vector construct 'Luc-1752', a luciferase-conjugated SMAD4 promoter region. As shown in Fig. 1C, transient co-transfection of Luc-1752 and the MZF1 expression vector exhibited over a 2.5-fold increase in the induction on transcriptional activity of SMAD4 promoter. However, co-transfection of Luc-1752 with HSF1 or AML-1 (RUNX1) did not reveal an increase in the SMAD4 promoter activity. Next, we analyzed SMAD4 and MZF1 expression in 13 gastric cancer cell lines. We found that SMAD4 mRNA (r= 0.28, P= 0.002) and SMAD4 protein (r=0.35, P=0.38) levels showed a close correlation with MZF1 protein level (Fig. 1D). In particular, their expression was substantially decreased in KATOIII, MKN28, MKN74, NCI-N87 and SNU5 cells, whereas MKN1, SNU484, SNU620 and SNU668 had relatively high levels of their expression. In addition, ectopic expression of MZF1 increased SMAD4 levels in MKN74 cells (low expresser), whereas si-MZF1 transfection decreased SMAD4 levels in MKN1 cell (high expresser). MZF1 affects SMAD4 expression at both the mRNA and protein levels in gastric cancer cell lines (Fig. 1E). These results suggest that SMAD4 is a novel target gene of the transcription factor MZF1.

To analyze the core region enabling MZF1-mediated transcriptional activation on the SMAD4 promoter, we generated four constructs, each containing a partial deletion mutant in the SMAD4 promoter, luc-1011, -371, -216 and -41 for luciferase assaying (Fig. 2A). As shown in Fig. 2A, transient co-transfection of the MZF1-expressing vector and each partial deletion mutant construct exhibited induction in SMAD4 promoter activity from 1.7- to 3-fold. Among them, the Luc-216 construct showed the highest SMAD4 promoter activity under MZF1 overexpression. In further bioinformatic analysis, we identified four putative MZF1-binding elements (MBE) on the Luc-216-containing partially deleted SMAD4 promoter region (nucleotides −216 to +84): MBE 1 (−102 to −99), MBE 2 (−80 to −77), MBE 3 (−52 to −49) and MBE 4 (−12 to −9). To demonstrate essential MZF1-binding element, we generated mutant reporter constructs containing mutated MBE sequences (5′-AAAAG-3′) that disturb MZF1 binding. According to the results (Fig. 2B, left panel), mutation of MBE2 showed loss of promoter activity, whereas other mutated MBEs exhibited over 1.5-fold increase in induction compared to the negative control (Fig. 2B, right panel). These results show that MZF1 could positively regulate SMAD4 promoter activity and its critical binding region might be MBE2 (−80 to −77) located from −216 to +84 in the SMAD4 promoter. To examine the direct binding of MZF1 to the SMAD4 promoter region, we performed ChIP assays and EMSAs. For ChIP analysis, we produced ChIP I and ChIP II sequences containing two independent MBE sites and we also designed, an EMSA probe (Fig. 3A). In ChIP analysis, MZF1 directly bound to the ChIP I sequence containing MBE1 and MBE2 sites (Fig. 3B). Moreover, EMSA revealed that MZF1 specifically interacted with the EMSA probe containing the MBE2 site (Fig. 3C) and addition of an MZF1 antibody to the reaction mixture for EMSA resulted in a supershifted band. These results indicate that MZF1 directly bind to MEB2 (−80 to −77) of SMAD4 promoter region.

To elucidate the molecular function of MZF1 in gastric cancer cells, we performed proliferation assay using WST reagent and cell cycle analysis by flow cytometry after overexpression or knockdown of MZF1. As shown in Fig. 4A, transient transfection of WT-MZF1 did not affect MKN28 and MKN74 cell proliferation and siRNA-mediated knockdown of endogenous MZF1 neither affected MKN1 and AGS cell proliferation, nor did it affect the cell cycle change of the 4 cell lines (Fig. 4B). Next, we investigated whether MZF1 influences migration of gastric cancer cells. We performed migration assay using the Transwell migration assay and wound-healing assay. As a result, cell migration was substantially decreased and increased by transfection of WT-MZF1 and si-MZF1 (Fig. 4C and D). These results support the idea that MZF1 overexpression might negatively regulate migration, but does not have an effect on cell proliferation and growth of gastric cancer cells.

According to previous reports, SMAD4 plays an important role in the regulation of cancer cell migration, and therefore, we analyzed whether MZF1 suppresses cancer cell migration through regulation of SMAD4 expression. As shown in the Fig. 5A, co-transfection of MZF1 and SMAD4 into MKN28 cells significantly exhibited suppressive effect on cellular migration in the wound-healing assay. Consistent with this, the inhibitory effect on migration by overexpression of MZF1 and SMAD4 is reinforced due to MZF1-mediated increase in the expression of SMAD4. We next examined whether silencing of SMAD4 expression inhibits MZF1-mediated cellular migration in human gastric cancer MKN74 cells. Transient overexpression of MZF1 revealed over 30% reduction of MKN74 migration. In addition, co-transfection of si-SMAD4 restored this MZF1-induced reduction of migration in a dose-dependent manner (Fig. 5B). Likewise, the data from immunoblotting analysis showed that a knockdown of SMAD4 expression increased phosphorylation of AKT and expression of β-catenin, which are key regulatory molecules in migration. As would be expected, downregulation of AKT phosphorylation and suppression of β-catenin expression was observed when MZF1 was overexpressed (Fig. 5C). In summary, our findings indicated that MZF1 transcriptionally upregulates expression of SMAD4, and this MZF1-mediated expression of SMAD4 act as a suppressor of cancer migration in gastric cancer cells.