Y.C. and X.W. contributed equally to this work
Cancer Cell Biology
Runx3 suppresses gastric cancer metastasis through inactivation of MMP9 by upregulation of TIMP-1†
Article first published online: 21 JUN 2011
Copyright © 2011 UICC
International Journal of Cancer
Volume 129, Issue 7, pages 1586–1598, 1 October 2011
How to Cite
Chen, Y., Wei, X., Guo, C., Jin, H., Han, Z., Han, Y., Qiao, T., Wu, K. and Fan, D. (2011), Runx3 suppresses gastric cancer metastasis through inactivation of MMP9 by upregulation of TIMP-1. Int. J. Cancer, 129: 1586–1598. doi: 10.1002/ijc.25831
- Issue published online: 26 JUL 2011
- Article first published online: 21 JUN 2011
- Accepted manuscript online: 2 DEC 2010 09:53AM EST
- Manuscript Accepted: 4 NOV 2010
- Manuscript Received: 18 DEC 2009
- National Hi-Tech Research and Development Program of China (863). Grant Number: 2006AA02Z103
- National Basic Research Program of China (973). Grant Number: 2004CB518702
- National Natural Science Foundation of China. Grant Number: 30500597
- gastric cancer
Recent studies have suggested that loss of RUNX3 expression is involved with gastric tumor metastasis. However, the precise mechanism of RUNX3-mediated suppression of tumor metastasis remains elusive. We aimed to clarify the effect of RUNX3 on tumor metastasis in gastric cancer cell lines and tumors. Immunohistochemistry revealed that RUNX3 was significantly decreased in metastatic gastric cancer. Gelatin zymography and Western blot showed that instead of regulating matrix metalloproteinase 9 (MMP9) expression, RUNX3 expression inhibited MMP9 enzyme activity, and this was consistent with the upregulation of tissue inhibitor of metalloproteinases 1 (TIMP1) by RUNX3. TIMP1 siRNA treatment impaired RUNX3-mediated suppression of gastric cancer cell invasion. Reporter assays demonstrated regulation of TIMP-1 by RUNX3. Two RUNX3 binding sites were identified in the TIMP-1 promoter and direct interaction of RUNX3 with the TIMP-1 promoter was confirmed in vitro and in vivo. These findings provide evidence for RUNX3-mediated suppression of gastric cancer invasion and metastasis and define a novel molecular mechanism that for the metastasis-inhibiting activity of RUNX3. These data may be applied in the development of RUNX3 for gastric cancer metastasis diagnostics and therapeutics.
Gastic cancer is second only to lung cancer as a cause of cancer-related deaths worldwide, and poor prognosis and high mortality continue to be problems with this disease.1–3 Approximately 20% of patients survive to 5 years,4 and metastasis is a major cause of mortality in gastric cancer patients. Clearly, improvements in treatment depend on understanding the molecular mechanisms of gastric cancer metastasis. In general, cancer metastasis consists of a series of sequential, interrelated events involving growth factors, cell-adhesion molecules, matrix-degradation enzymes and motility factors. These molecules induce not only cell growth but also the extracellular matrix (ECM) degradation and angiogenesis that are required for tumor invasion and proliferation.5 Previous research on the molecular pathology of gastric cancer metastasis identified a variety of key oncogenes and tumor suppressor genes in this process, such as c-erbB2, c-met, p27 and pRb.6–11
Recently, deficiency in the Runt-related gene RUNX, which is a candidate tumor suppressor, was found to be causally related with gastric cancer.12 The RUNX family members, RUNX1, RUNX2 and RUNX3, encode DNA-binding α subunits that bind a common β subunit, CBFβ, to generate heterodimeric transcription regulators.13 All three RUNX family members play important roles in normal developmental processes and carcinogenesis.14–18 Recent studies showed that RUNX3 expression levels are downregulated in gastric cancer, either from hemizygous deletion or from promoter methylation of the RUNX3 gene. Furthermore, a decrease in RUNX3 protein expression is significantly associated with decreased survival of gastric cancer patients. Li et al.12 showed that RUNX3 expression was decreased in 60% of primary human gastric tumors; with nearly 90% of late stage tumors that represent highly metastatic tumors showing reduced RUNX3 levels. An analysis of clinical tissue samples from peritoneal metastases arising from gastric cancers showed that RUNX3 expression decreased significantly in the metastatic tissue, compared to normal gastric mucosa or primary main tumors.19, 20 Therefore, RUNX3 deficiency is closely associated with gastric cancer metastasis. In contrast, RUNX3 was recently reported to reduce gastric cancer angiogenesis by inhibiting expression of vascular endothelial growth factor (VEGF), which would be expected to suppress gastric cancer metastasis.21 Clearly, the function of RUNX3 in the pathogenesis of gastric cancer metastasis is not yet fully understood.
Tissue invasion is an essential step in metastasis that requires breakdown of the ECM around the cancer cells. Matrix metalloproteinases (MMPs) play a critical role in tumor invasion and angiogenesis by cleaving the ECM components.22, 23 In addition, MMPs, particularly MMP2 and MMP9, are postulated to promote invasion and lymph node metastasis of gastrointestinal cancer cells. Suppression of MMP activity impairs cancer cell migration and angiogenesis.24–26
No studies have reported on the effect of RUNX3 on MMP activity, so we investigated whether RUNX3 regulated the activity of MMP2 and MMP9 in human gastric cancer cell lines. MMP activity is controlled by specific, endogenous tissue inhibitors of metalloproteinases (TIMPs). TIMPs bind noncovalently with 1:1 stoichiometry to the active form of MMPs to inhibit their proteolytic activity,27 and four TIMPs have been identified. TIMP1 selectively and specifically forms complexes with pro-MMP9 and is a major regulator of MMP9 activity. TIMP-2 and TIMP-3 effectively inhibit MT-MMPs (membrane-type MMPs), but pro-MMP2 is activated by MT1-MMP after interaction with TIMP-2.28, 29
In our study, we examined whether the activities of MMPs and TIMP1 were regulated by RUNX3 in gastric cancer cell lines, by examining the effect of increased RUNX3 expression on the invasive potential of human gastric cancer cells in vitro and in animal models, and on MMP9 activity and TIMP1 expression at the transcriptional level. Our data provide a novel mechanism for the RUNX3-mediated suppression of gastric cancer invasion and metastasis.
Material and Methods
Tissue collection and immunohistochemistry
Our study was approved by the Hospital's Protection of Human Subjects Committee. All gastric cancer cases were clinically confirmed. Specimens were collected from 83 patients with nonmetastatic gastric carcinoma and 40 patients with metastatic gastric carcinoma who underwent gastrectomy at the Department of General Surgery in Xijing hospital, Xi'an, China between January 2004 and July 2005. Gastroscope biopsies were collected for 20 normal gastric mucosa specimens and 5 μm sections of all formalin-fixed paraffin-embedded specimens were made. Slides were dewaxed, rehydrated and incubated in methanol containing 0.3% H2O2 for 30 min to inhibit endogenous peroxidase activity, before blocking with 10% normal goat serum (Boster, Wuhan, China) for 20 min and incubating overnight at 4°C with anti-RUNX3 (1:200, Active Motif, Carlsbad, CA). Slides were washed in phosphate-buffered saline (PBS), then incubated with biotinylated anti-rabbit IgG (1:200, Boster) for 30 min at room temperature, and rinsed with PBS. After incubation with avidin-biotin-peroxidase complex for 15 min, sections were stained with 3,3-diaminobenzidine. Normal rabbit serum was used as a negative control. Two observers with no knowledge of clinical outcomes examined all sections independently, and an average value of intensity score was determined based on the following scale of immunoreactivity (IR): 0, no IR; 1, weak IR; 2, moderate IR; and 3, strong IR. The proportion of positive cells (PROP) was classified as: 0, no tumor cells exhibiting IR; 0.33, 0–33% of the tumor cells exhibiting IR; 0.67, 33–67% of the tumor cells exhibiting IR and 1.0, 67–100% of the tumor cells exhibiting IR. Overall staining scores of IR × PROP, and average intensity scores were calculated.
Cell lines and culture conditions
Human gastric cancer cell lines SGC7901 and MKN28 were from our institute, and were cultured in RPMI1640 medium (Life Technologies, Carlsbad, CA) supplemented with 10% fetal calf serum (FCS) in a 37°C humidified incubator with 95% air, 5% CO2. SGC7901 expresses RUNX3 at a moderate level and MKN28 shows no expression of RUNX3.30
pBK-RUNX3 was kindly provided by professor Paul J. Farrell, Department of Virology, Ludwig Institute for Cancer Research, UK. Human RUNX3-specific siRNA31 fragment pairs containing Bam HI and Hind III sites were synthesized, annealed in annealing buffer (0.1 M NaCl, 10 mM Tris, pH 7.4) and cloned into the pSilencer 3.1-H1 neo vector (Ambion, Austin, TX, USA) to construct a RUNX3 siRNA vector. The RUNX3 target sequences were TGACGAGAACTACTCCGCT31 using the oligonucleotides RUNX3-2F; 5′-GATCCTGACGAGAACTACTCCGCTTTCAAGAGAAGCGGAGTAGTTCTCGTCATTTTTTGGAAA-3′ and RUNX3-2R; 5′-AGCTTTTCCAAAAAATGACGAGAACTACTCCGCTTCTCTTGAA AGCGGAGTAGTTCTCGTCA G-3′.
Promoter sequences of the TIMP1 promoter containing two RUNX binding sites (−362 to +86) were amplified from periphery blood mononuclear cell genomic DNA using primers in Table 1. Promoters were cloned into the pGL3 enhancer vector (Promega, Madison, WI) to construct reporter vector pGL-TIMP1. TIMP1 siRNA(h) and negative control siRNA were from Santa Cruz Biotechnology.
MKN28 and SGC7901 cells were stably transfected with pBK-RUNX3 using Lipofectamine 2000 (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer's protocol. Empty pBK-CMV was transfected as a control. pSilencer-RUNX3 was stably transfected into SGC7901 cells. As a negative control, we used a pSilencer vector that expresses a siRNA with limited homology the human, mouse and rat genomes (Ambion). To clarify the specific effects of RUNX3 siRNA, MKN28 cells (RUNX3 negative) were transiently transfected with pSilencer-RUNX3 or pSilencer. To explore the effects of TIMP1 expression on invasion, the stable transfected cell lines were transcently transfected with TIMP1 siRNA(h) or the negative control siRNA.
For stable transfection, G418 (400 μg/ml for SGC7901cells and 300 μg/ml for MKN28 cells) was added to cultures 24 hr after transfection, mixed clones were screened and expanded for 6 weeks, and RUNX3 expression confirmed by Western blot (data not shown). For transient transfection, cells were harvested within 48 hr of transfection.
In vitro invasion assay
In vitro invasive ability was tested using Transwells (Corning Costar, Cambridge, MA), 8-μm pores with polycarbonate membrane chamber inserts, coated with Matrigel (BD Biosciences, CA). Invasion chambers were rehydrated with serum-free RPMI-1640 for 2 hr in a humidified incubator at 37°C with 5% CO2. The lower compartment of the 24-well chambers was filled with 0.6 ml NIH3T3 culture supernatant and the upper chamber with 0.2 ml of cell suspension (5 × 104 cells/ml). After 24 hr in 5% CO2 at 37°C, noninvasive cells and gel were removed from the upper membrane, fixed in 4% formaldehyde, and invasive cells stained with hematoxylin and counted at ×200 magnification for 10 random fields per well. Each experiment was performed in triplicate.
RNA isolation and reverse transcriptase-polymerase chain reaction
For the six stable transfacted cell lines, total RNA was isolated from 60-mm culture dishes with ∼1 × 106 cells, using Trizol (Life Technologies, CA) and DNase according to the manufacturer's protocol. RNA quantity and purity were determined by spectroscopy. Reverse transcription used a First-Strand cDNA Synthesis Kit (Fermentas, Hanover, MD) in 20 μl. After 60 min at 42°C, reactions were terminated at 70°C for 10 min. cDNA was amplified with primers from Table 1. Products were visualized on 1% agarose with ethidium bromide. Gel were photographed and quantitated using β-actin as control. The experiment was done three times.
Western blot analysis
Whole-cell lysates were prepared from gastric cancer cell lines. Samples were loaded on 10% SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA) and incubated with primary antibody overnight at 4°C. Membranes were incubated with horseradish peroxidase-conjugated secondary antibody (Santa Cruz, CA) for 1 hr at 37°C and immunoreactive protein bands visualized with a chemiluminescent detection kit (ECL; Amersham Pharmacia Biotech, Piscataway, NJ). Signal intensities were quantified and analyzed as described below. Each experiment was repeated three times.
Rabbit polyclonal antibody against human RUNX3 was from Oncogene (San Diego, CA). Mouse monoclonal antibodies against MMP2 and MMP9 were from Santa Cruz Biotechnology (CA). Mouse IgG against TIMP1 was from Santa Cruz (CA). Mouse monoclonal antibody against β-actin was from Sigma (MO).
Enzyme-linked immunosorbent Assay
TIMP1 in culture supernatants were determined using the Hu TIMP-1 ELISA kit (BioSource, CA) following the manufacturer's instructions, from 2.5 × 105 cells cultured in 12-well plates for 24 hr in RPMI1640 medium with 10% FCS (v/v). After 24 h, medium was replaced with 500 μl of 0.5% FCS (v/v) RPMI1640 medium for an additional 24 h before measuring TIMP1 in conditioned medium. Samples and standards of known Hu TIMP1 content were used for a standard curve. A biotinylated monoclonal second antibody was detected with streptavidin-peroxidase at 450 nm absorbance in a microtiter plate reader.
MMP2 and MMP9 activity in conditioned medium was determined by gelatin zymography. Equal numbers of gastric cancer cells (2 × 106 cells per six-well culture plate) were incubated in serum-free RPMI-1640 media, at 37°C in 5% CO2. Medium was obtained after 1 day, vacuum freeze-dried and dissolved in 200 μl PBS. To further determine if TIMP-1 is responsible for controlling MMP activities, cells were transiently transfected with TIMP1 siRNA(h) or the negative control siRNA. Medium was collected within 48 hr of tansfection, vacuum freeze-dried and dissolved in 200 μl PBS. Samples were mixed (3:1) with 40% (v/v) glycerol, 0.25 M Tris-HCl, pH 6.8, and 0.1% bromophenol blue and loaded without boiling onto 10% SDS-polyacrylamide gels containing type I gelatin (1 mg/ml; Sigma, St. Louis). After electrophoresis, gels were soaked in 2.5% Triton X-100 for 45 min with single change of detergent solution. Gels were incubated for 18 hr at 37°C in substrate buffer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl2 and 0.02% NaN3), stained with 0.05% Coomassie brilliant blue G-250 (Sigma, St. Louis) and destained in 10% acetic acid and 20% methanol. Gels were photographed and then quantitatively measured by scanning densitometry. Each experiment was repeated three times.
Luciferase reporter assay
SGC7901 cells were cultured in 24-well plates from 5 × 105 cells/well to 90% confluence. pBK-RUNX3 or empty pBK-CMV plasmid was cotransfected with pGL-TIMP1 using the Lipofectamine™ 2000 (Invitrogen). pRL-TK was used as internal control. Luciferase reporter assays were performed using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions. Each experiment was performed in triplicate.
Chromatin immunoprecipitation (ChIP) was done with a ChIP assay kit (Upstate Cell Signaling Solutions, Lake Placid, NY) according to the manufacturer's instructions. SGC7901 cells were fixed with 1% formaldehyde at 37°C for 10 min, washed twice with ice-cold PBS containing 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin and 1 μg/ml pepstatin A, and collected. Cells were resuspended in SDS lysis buffer and sonicated to shear DNA to 200–500 bp. Supernatants were collected and diluted in ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 2 mM EDTA, 167 mM NaCl, 16.7 mM Tris-HCl, pH 8.1) containing protease inhibitors. Chromatin was precleared with salmon sperm DNA-coated protein A agarose, followed by immunoprecipitation using 5 μg of anti-RUNX3 (Santa Cruz) overnight at 4°C with rotation. Complexes were collected with salmon sperm DNA-coated protein A agarose and extensively washed with ChIP dilution buffer. Crosslinks were reversed at 65°C for 4 hr. Samples were treated with proteinase K for 2 hr at 42°C, phenol/chloroform extraction and ethanol precipitation. Precipitated DNA was amplified using primers in Table 1. Preimmune serum was used as a negative control.
Electrophoretic mobility shift assay
SGC7901 cells were collected and nuclear extracts prepared with NE-PER Nuclear and Cytoplasmic Extraction Reagents Kit (Pierce Biotechnology, Rockford, IL). Probes containing the RUNX3 binding site were obtained by annealing complementary oligomers and biotin labeling at the 3′ end using terminal deoxynucleotidyl transferase (Promega, Madison, WI). Probes were: 5′-AGGCCTGTGGTTTCCGCACC-3′ (Site1,wild-type probe); 5′-AGGCCGCTAGCTTCCGCACC-3′ (Site1, mutant probe); 5′-GGGGATGTGGGTGATTGGAT-3′ (Site2, wild-type probe); 5′-GGGGAGCTAGCGATTGGAT-3′ (Site2, mutant probe). Nuclear proteins (5 μg) were incubated with 20 fmol of the labeled probe for 20 min at room temperature before separation on 4% nondenaturing polyacrylamide gels in 0.5X TBE (45 mM Tris base, 45 mM boric acid, 1 mM disodium EDTA·2H2O) with 2.5% glycerol. Samples were transferred to nylon membranes (Pierce), crosslinked for 10–15 min in a transilluminator with 312 nm bulbs and developed with a Lightshift Chemiluminescent EMSA Kit (Pierce). For competition and antibody supershift, unlabeled or mutated probes or 2 μg of anti-RUNX3 antibody (Active Motif) was added for 1 hr before addition of the probe.
Experimental metastasis assay
Female athymic BALB/c nude mice were used to analyze in vivo metastatic potency. Gastric cancer cells (5 × 106 per mouse) were injected intraveneously into the tail vein and mice housed in laminar flow cabinets under pathogen-free conditions and used at 6 weeks. Before inoculation, exponential phase cells were harvested by digestion with 0.25% trypsin (w/v). Cell viability was determined by trypan blue exclusion and only single-cell suspensions >95% viable were used. Animals were killed after 4 weeks and visible tumors on liver and lung surface counted. Livers and lungs were fixed in Bouin's solution, and hematoxylin and eosin (H&E) stained for metastatic nodules. Each group contained six mice.
Immunohistochemistry of human tumor xenograft specimens
Gastric cancer cells for inoculation were prepared and analyzed as described for experimental metastasis assay before 1 × 106 cells per nude mouse were injected into the stomach wall. Each experimental group contained 10 mice, animals were killed when they became moribund and survival was recorded. For RUNX3 and TIMP1 staining, slides with 5-μm sections were prepared as described above with anti-RUNX3 or anti-TIMP1 (1:100; Santa Cruz, CA) and biotinylated anti-rabbit IgG or anti-mouse IgG (1:200, Boster). After incubation with avidin-biotin-peroxidase complex for 15 min, sections were stained with 3,3-diaminobenzidine. Normal rabbit serum or mouse serum was used as a negative control.
Each experiment was repeated at least three times. Bands from Gelatin zymography, Western blot or RT-PCR were quantified by Molecular Analyst software (Bio-Rad, Hercules, CA). Relative protein or mRNA levels were calculated by reference to β-actin. Numerical data are presented as the mean ± standard error of the mean (SEM). The differences between means were analyzed by ANOVA and post hoc test. All statistical analyses were performed using SPSS11.0 software (Chicago, IL). p < 0.05 was deemed statistically significant.
RUNX3 expression is reduced in metastatic gastric cancer
To explore the relationship between loss of RUNX3 expression and gastric cancer metastasis, we compared RUNX3 expression in primary sites from 83 patients with nonmetastatic gastric cancer with 40 patients with metastatic gastric cancer (Fig. 1a). As shown in Table 2, RUNX3 expression in nonmetastatic gastric cancer samples was 74.7% (62/83) and 42.5% (17/40) in metastatic gastric cancer samples (p < 0.05). The average staining score for nonmetastatic gastric cancers was significantly higher than for metastatic gastric cancers (1.25 ± 0.48 vs. 0.57 ± 0.26, p < 0.05), suggesting that RUNX3 expression was reduced in metastatic gastric cancer at both primary and metastatic sites. We also analyzed RUNX3 expression at primary sites and corresponding metastatic sites in 40 cases and found no significant difference between the two groups in either rate of positives (42.5% vs. 37.5%, p > 0.05) or staining score (0.57 ± 0.26 vs. 0.45 ± 0.33, p > 0.05; Fig. 1b). We also presented evidence that RUNX3 negative primary tumors make metastasis to nodes, as shown in Fig. 1c.
Reduced RUNX3 expression is associated with increased metastatic potential of gastric cancer cells
To examine the effect of RUNX3 expression on the metastatic ability of gastric cancer cells, we stably transfected pBK-RUNX3 plasmids into SGC7901 and MKN28 gastric cancer cell lines with an empty vector as a negative control. pSilencer-RUNX3 or control pSilencer vector was stably transfected into SGC7901 cells. As shown in Figure 2a, transfection with pBK-RUNX3 increased RUNX3 expression compared to the empty pBK-CMV vector control, and transfection of pSilencer-RUNX3 inhibited RUNX3 expression in SGC7901 cells.
In vitro invasion assays were performed to evaluate the effects of RUNX3 on the invasive ability of tumor cells. As shown in Figure 2b, RUNX3-transfected cells exhibited significantly lower invasive potency than the control, with invasive cells by 60.1 and 50.4% in SGC7901 cells and MKN28 cells, respectively. The difference in invasiveness was found to be statistically significant (p < 0.05, respectively). Inhibition of RUNX3 expression in SGC7901 cells by RUNX3 siRNA significantly increased cell invasiveness compared to the control (Fig. 2; p < 0.05). However, no significant difference in cell invasiveness was observed in MKN28 cells transiently transfected with RUNX3 siRNA and the control (p > 0.05), because MKN28 cells are RUNX3 negative, thus confirming the specific effect of RUNX3 siRNA. These results indicated that enhanced expression of RUNX3 in the SGC7901 cells and MKN28 cells was associated with reduced invasive ability.
A tail vein metastatic assay was used to analyze the in vivo effects of RUNX3 expression on the metastatic potency of gastric cancer cells. Compared to control cells transfected with empty vector, tail vein injection of cells stably transfected with pBK-RUNX3 into athymic nude mice led to significantly fewer visible tumors on the liver and lung surface. Compared to the injection of control cells, the injection of the cells transfected with pSilencer-RUNX3 led to more visible tumors on the liver and lung surface of the mice (Figs. 2c–2e). This suggested that RUNX3 suppressed the invasive and metastatic potential of gastric cancer cells.
MMP9 and TIMP1 and regulation of gastric cancer cells invasiveness by RUNX3
To study the possible role of MMPs in RUNX3-induced inhibition of cell invasion, we analyzed the activities of MMP2 and MMP9 by gelatin zymography. As shown in Figure 3a, MMP9 with collagen-degrading activity was detected in the culture supernatant of SGC7901 and MKN28 cells. Compared to the control, MMP9 enzyme activity in pBK-RUNX3 stable transfectants was significantly lower in both SGC7901 and MKN28 lines, and MMP9 activity in pSilencer-RUNX3 transfectants was significantly higher than in control cells. The activity of MMP2 was relatively low compared to MMP9 and showed no significant difference between controls and cells in which RUNX3 was upregulated or downregulated.
Regulation of MMP2, MMP9 and TIMP-1 expression by RUNX3 was determined by Western blot and RT-PCR. We found no difference in MMP2 or MMP9 expression between controls and cells with upregulated or downregulated RUNX3. However, transfection of pBK-RUNX3 induced stronger expression of TIMP1 in SGC7901/pBK-RUNX3 and MKN28/pBK-RUNX3 cells, whereas weaker expression of TIMP1 was observed in cells transfected with pSilencer-RUNX3 compared to controls (Fig. 3b).
ELISA was performed to detect TIMP1 secreted into conditioned culture medium of gastric cancer cells. TIMP1 production in gastric cells transfected with pBK-RUNX3 was significantly higher than in the controls (18.2 vs. 11.3 ng/ml in SGC7901; p < 0.05 and 12.5 vs. 6.4 ng/ml in MKN28; p < 0.05). Consistently, TIMP1 levels in the medium of SGC7901-RUNX3si cells decreased compared to control SGC7901-pSilencer cells (4.7 vs. 12.2 ng/ml; p < 0.05; Fig. 3d).
To study the possible role of TIMP1 in RUNX3-regulated invasion, SGC7901 and MKN28 cells were transiently transfected with TIMP1 siRNA before performing invasion assays. Treatment with TIMP1 siRNA promoted the activity of SGC7901 and MKN28 cells. Increasing rate caused by TIMP1 siRNA in SGC7901-RUNX3 was 63.3%, which was significantly higher than in SGC7901-RUNX3si (15.5%). Increasing rate caused by TIMP1 siRNA in MKN28-RUNX3 was 55.7%, which was significantly higher than in MKN28-p-BK-CMV (19.7%; Fig. 3c).
To further determine if TIMP-1 is responsible for regulating MMP activities, SGC7901 and MKN28 cells were transiently transfected with TIMP1 siRNA before performing gelatin zymography. As shown in Figure 3e, treatment with TIMP1 siRNA increased the MMP9 activity of SGC7901 and MKN28 cells. MMP9 activity in MNK28-RUNX3 cells was significantly increased after TIMP1 siRNA treatment, whereas MMP9 activity changed much less in MKN28-pBK-CMV cells. Similarly, MMP9 activity in SGC7901-RUNX3 cells was significantly increased after TIMP1 siRNA treatment, whereas MMP9 activity changed very little in SGC-7901-RUNX3si cells. The result suggested that transfection of TIMP1 siRNA impaired the inhibitive effect of RUNX3 on MMP9 activity.
TIMP1 expression is upregulated by RUNX3 at the transcriptional level
To investigate the possibility that RUNX3 increased TIMP1 expression by directly enhancing TIMP1 promoter activity, a dual luciferase reporter assay was performed. Transfection of increasing doses of pBK-RUNX3 plasmid led to significant increases in TIMP1 promoter activity compared to transformation with empty vector. Western blot showed that RUNX3 protein expression increased in cells transfected with different doses of pBK-RUNX3 plasmid (Fig. 4a). These results indicated that RUNX3 caused transactivation of TIMP1 promoter, upregulating the mRNA and protein expression of TIMP1.
Further, computer-based analysis of the TIMP1 gene revealed that two putative RUNX binding sites were contained in the TIMP1promoter (−327 to −321 site 1; −66 to −61 site 2). Therefore, we determined whether RUNX3 binds to these regions in the TIMP-1 promoter in vivo using a ChIP assay. To determine whether endogenous RUNX3 binds to the TIMP-1 promoter, chromatin fragments from SGC7901 cells were immunoprecipitated with specific anti-RUNX3 antibody or with normal rabbit serum as a negative control. Two set of PCR forward and reverse primers flanking the putative RUNX3-binding sites were used to perform PCR with the chromatin fragments as template. Specific PCR product show up when templates were the chromatin fragments from anti-RUNX3 antibody immunoprecipataion and from the input, whereas no band appear when templates were the chromatin fragments from normal rabbit serum immunoprecipataion. Therefore both sites were found to bind with RUNX3. To determine whether endogenous RUNX3 by gene transfer also binds to the TIMP-1 promoter, chromatin fragments from MKN28-RUNX3 cells were immunoprecipitated with specific anti-RUNX3 antibody or with normal rabbit serum as a negative control. Similarly, exogenous RUNX3 is recruited to the TIMP-1 promoter (Fig. 4b).
To further confirm the binding of RUNX3 to TIMP-1 promoter, EMSA was conducted to determine if RUNX3 could bind to the two putative binding sites. As shown in Figure 4c, nuclear extracts from SGC7901 cells shifted bands of labeled probes with both sites. These bands were inhibited from forming when competitive cold probes were included, whereas mutated cold probes did not inhibit binding between labeled probes and nuclear protein. Shifted bands were supershifted by addition of anti-RUNX3 antibody. These observations suggest that the bands were specific complexes formed by RUNX3 and the probes with RUNX-binding sites.
TIMP1 expression enhanced by RUNX3 in human gastric cancers in nude mice
The in vivo effects of RUNX3 on TIMP1 expression were determined by generating tumors in nude mice using gastric cancer cell lines. Mice receiving MKN28-RUNX3 cells showed significantly prolonged survival compared to control groups of MKN28 or MKN28-pBK-CMV cells (Fig. 5a). The control group showed 0 survival at 70 days and the pBK-CMV group showed 0 at 80 days. The survival of the MKN28-RUNX3 was 80% at the end-point and was significantly prolonged by log-rank (Mantel-Cox) test (p < 0.0001). RUNX3 also significantly enhanced TIMP1 expression in the primary tumors compared to controls (Fig. 5b).
RUNX3 was first proposed as a tumor suppressor in gastric cancer when RUNX3 knockout mice would found to develop gastric epithelial hyperplasia.12 Numerous studies have reported on gene methylation and silencing of RUNX3 in tumors of liver, colon, gallbladder and lung,32–39 and RUNX3 has been found to suppress gastric cancer by controlling growth and inducing apoptosis. Although loss of RUNX3 expression was recently implicated in gastric and breast cancer metastasis, the effect of RUNX3 on cancer metastasis has not been fully examined.40
We provide experimental evidence that the expression level of RUNX3 is related to gastric cancer cell invasion and metastasis, both in vitro and in vivo, and reduced RUNX3 protein expression is associated with increased MMP9 enzyme activity in gastric cancer cell medium. Ectopic expression of RUNX3 significantly inhibited gastric cancer cell invasion in vitro, and metastasis in animal models, which appears to be related to MMP9 inactivation through enhanced expression of TIMP1, a critical regulator of MMP9 activity. We confirmed that RUNX3 binds to the TIMP1 promoter and enhancing its activity in gastric cancer cells.
In response to exogenous factors or environmental signals, RUNX family members regulate the transcription of the downstream target genes, and we applied this model to understand the role of RUNX3 loss in cancer development. As an integral part of the TGF-β signaling pathway, RUNX3 suppresses gastric epithelial cell growth, at least in part, by inducing p21 expression. In cooperation, with TGF-β-activated SMAD, RUNX3 synergistically activates the p21 promoter to induce transcription.41 RUNX3 also physically interacts with FoxO3a that is expressed in gastric cancer cell lines to activate Bim transcription and induce cell apoptosis.42 VEGF, a main promoter of angiogenesis, is directly regulated by RUNX3 in gastric cancer cells through suppression of VEGF promoter activity.21 We previously demonstrated that overexpression of RUNX3 sensitizes gastric cancer cells to chemotherapeutic drugs by downregulating Bcl-2, MDR-1 and MRP-1 and confirmed that RUNX3 inhibits MDR-1 and MRP-1 expression in gastric cancer cells by repressing their promoters.30
MMPs facilitate the disruption of basement membranes during the initial stages of metastasis and activate growth factors and cytokines. MMP9 cleaves the proangiogenic cytokine IL8, increasing its activity tenfold. MMP9 also degrades and inactivates the angiogenesis inhibitor platelet factor-4.43–45 MMP2 and MMP9 are associated with lymphatic invasion and metastasis.
In our study, MMP9 enzyme activity was reduced by increased RUNX3 expression. We postulate that RUNX3 regulates MMP9 activity instead of transcription. TIMP1 is a main negative regulator of MMP9 enzyme activity and is involved in several tumor metastasis processes, including gastric cancer. Signal transducers and activators of transcription-3 (STAT-3) upregulate TIMP1 expression and decrease the invasiveness of breast cancer.46 In primary prostate cancer cell lines, TIMP1 expression induced by IL10 significantly blocked tumor cell invasion.47 In the orthotopic transplantation model in nude mice, metastatic ability of the highly metastatic human gastric cell line KKLS is significantly decreased by exogenous TIMP-1 gene transfection.48 Western blot, semiquantitative RT-PCR and ELISA showed that TIMP1 expression was elevated by RUNX3 in gastric cancer cells both at the mRNA and protein level, and this may contribute to the reduced MMP9 enzyme activity seen when RUNX3 is exogenously expressed. In vitro assay results showed that TIMP1 siRNA abolished RUNX3-mediated suppression of SGC7901 and MKN28 cell line invasion. Reporter assays showed that the TIMP-1 promoter is activated by RUNX3. ChIP assays demonstrated that both exogenously expressed RUNX3 in MKN28/pBK-RUNX3 cells, and endogenous RUNX3 in SGC7901 cells could be recruited to putative binding sites in the TIMP1 promoter. Specific binding was confirmed by EMSA. Therefore, RUNX3 appears to upregulate TIMP1 expression by interacting directly with the promoter, and enhancing its transcription.
In conclusion, our study provides evidence that RUNX3 enhances TIMP1 transcription, ultimately decreasing MMP9 activity in gastric cancer cells. This may, at least in part, contribute to RUNX3-mediated inhibition of gastric cancer invasion and metastasis. We are currently investigating if other genes involved in gastric cancer metastasis are also regulated by RUNX3.
The authors are grateful to Prof. Paul J. Farrell (Department of Virology, Ludwig Institute for Cancer Research, UK) for the pBK-RUNX3 vector. The authors are also grateful to Zhang Ping (Department of Medical Genetics and Developmental Biology, the Forth Military Medical University, P. R. China) for experimental help, and Chris Tachibana, science writer, for assisting with manuscript preparation.