ADAMTS9 was associated with LN metastases and survival outcome in patients with GC
We examined the mRNA levels of ADAMTS9 in the 16 pairs of tumor and adjacent non-tumor tissues. A total of 9/16 of tumor tissues showed lower mRNA expression level compared with that of adjacent non-tumor tissues (Figure 1A), which were classified into low expression group. The correlation between the clinicopathological characteristics and mRNA expression of ADAMTS9 was analyzed. ADAMTS9 mRNA expression was significantly associated with pN stage (P=0.044) and lauren type (P=0.019) (Supplementary Table S1). Patients with low ADAMTS9 mRNA expression were at high risk with increased metastatic LNs (12.89±3.49 vs 2.571±1.020, P=0.024, Figure 1B).
We further examined the protein expression of ADAMTS9 in GC tissues. With the IHC staining, ADAMTS9 protein expression was found in the cytoplasm and nucleus of the gastric tissue cells (Figures 1C and 3A). H-score was used to evaluate the ADAMTS9 protein expression. A total of 111 pairs of GC tissues and adjacent non-tumor tissues and additional 18 GC tissues were completely saved and well stained. Compared with that of adjacent non-tumor tissue, the H-Score of GC tissues was significantly decreased (P=0.008, Figure 1D). Exactly 129 GC tissues were then divided into the low and high expression groups according to the median of H-score as the cut-off value. The clinicopathological characteristics are shown inTable 1. Univariate analysis suggested that patients with low ADAMTS9 expression was associated with advanced pN stage (P=0.018). We further examined the number of LNs involvement between the two groups, and the number of metastatic LNs of high expression group was significantly decreased (11.38±11.64 vs 6.73±7.63, P=0.007; Figure 1E). ADAMTS9 expression was closely correlated to the LN metastasis.
To evaluate the prognostic value of ADAMTS9 expression in GC, we employed the Kaplan–Meier analysis.Figure 1F shows that patients with GC with low ADAMTS9 expression had a significantly poorer 5-year survival rate than those with high expression (P=0.007). To remove the various interference factors, we also acquired the multivariate Cox regression analysis. In multivariate Cox regression analysis, ADAMTS9 expression level (HR 1.602, 95%CI 1.069–2.399, P=0.022) was an independent predictor of the prognosis of patients with GC with the lowest Akaike information criterion (AIC, AIC=71.555) and Bayesian information criterion (BIC, BIC=88.714, Table 2). We demonstrated the crucial effect on the survival outcomes through stratification by pN stage. In the subgroup of patients with pN3 stage, ADAMTS9 expression was significantly associated with survival outcomes (P=0.011,Figure 1F).
Low-expression of ADAMTS9 in GC cell lines
The ADAMTS9 mRNA expression in 10 types of human GC cell lines and normal GES-1 was determined. Excluding SNU-1 and HGC-27, the mRNA expression of ADAMTS9 in 8/10 GC cell lines was lower than that in GES-1 (Figure 2A). To confirm whether the protein expression follows its mRNA expression, we tested seven GC cell lines and GES-1. Western blot analysis results showed that ADAMTS9 protein was silenced or down-regulated in 6/7 GC cell lines (Figure 2B), which was consistent with their mRNA expression. The low ADAMTS9 expression in most GC cell lines might be a tumor suppressor.
ADAMTS9 suppressed GC cell proliferation and viability
To explore the biological function of ADAMTS9 in GC cells, we overexpressed ADAMTS9 in AGS, BGC-823, and SGC-7901 cells through the pCNDA3.1-ADAMTS9 plasmid. The mRNA and protein expression of ADAMTS9 were proven by RT-PCR and Western blot analysis (Figure 2c and Supplementary Figure S1A). CCK8 assay and cell growth curve results then showed that ADAMTS9 suppressed cell proliferation and viability (Figure 2D and Supplementary Figure S1B). The results of colony formation assays were consistent with the CCK8 assay results (Figure 2E and Supplementary Figure S1C). Thus, ADAMTS9 inhibited GC cell proliferation and viability.
ADAMTS9 inhibited tumor growth in nude mouse
We investigated the effects of ADAMTS9 in vivo. The size of subcutaneous tumor transfected with ADAMTS9 was significantly decreased than that in tumor transfected with empty vector. The growth curve and tumor weight figure of the subcutaneous tumor are shown in Figure 2F. ADAMTS9 inhibited tumor growth in vivo.
ADAMTS9 inhibited GC cells migration and invasion
The effects of ADAMTS9 expression on AGS, BGC823, and SGC-7901 cells on migration and invasion were also assessed. Wound healing assay result showed that the migration distances of over-expressed ADAMTS9 group were significantly reduced compared with that of the control group (AGS, 48H, P<0.001; BGC-823, 48H, P=0.009; and SGC-7901 36H P<0.001, Figure 2G and Supplementary Figure S1D). To confirm the effect on migration and invasion, we further examined motile ability by using Transwell chambers. Similar results were obtained, thereby indicating that the number of migratory and invaded AGS and BGC-823 cells was significantly decreased in the over-expressed ADAMTS9 group (AGS, BGC-823, and SGC-7901; migration: P=0.003, P=0.010, and P=0.010; invasion: P=0.035, P=0.010, and P=0.004; Figure 2H and Supplementary Figure S1E). ADAMTS9 inhibited GC cell migration and invasion.
ADAMTS9 inhibited GC metastasis through CCL5 and CXCL11 pathways
With IHC results for ADAMTS9, we found ADAMTS9 was partly located in the nucleus of the gastric tissue cells (Figure 3A). To elucidate the downstream molecular mechanism of ADAMTS9, we performed the mRNA sequencing analysis of BGC-823 cells transfected with pCDNA3.1-ADAMTS9 plasmid or pCDNA3.1 empty plasmid. Among the 20031 genes screened by the mRNA sequencing, 183 genes showed significant decrease or increase greater than 1.5-fold change (Figure 3B). Several DEGs were significantly up- or down-regulated in AGS and BGC-823 cells by real-time PCR analysis (Figure 3F). The gene information and potential biological functions of these DEGs are shown in the Figure 3D and Figure 3E. To further examine the key roles of ADAMTS9 on GC metastasis, we performed the KEGG pathway analysis (Figure 3C). The chemokine signaling pathway and the cytokine-cytokine receptor interaction pathway were included in the top 20 enrichment pathways (Supplementary Figure S2). Two of the significantly down-regulated genes, namely, CCL5 and CXCL11, played a crucial role in metastasis. To further validate this finding, we detected the down-regulated mRNA and protein expression of CCL5 and CXCL11 in the four GC cell lines (AGS, BGC-823, SGC-7901, and NCI-N87; Figures 3G and 3H). The relationships between ADAMTS9 and CCL5 and between ADAMTS9 and CXCL11 in human GC tissues were further explored. We performed immunohistochemical staining to examine the correlation between the protein levels of CCL5, CXCL11, and ADAMTS9 in the GC tissue specimens on TMAs (Supplementary Figure S3).Figure 3I shows that the protein expressions of CCL5 and CXCL11 were both negatively associated with ADAMTS9 protein expression (CCL5, N=122, Pearson r=−0.308, P=0.001; CXCL11, N=125, Pearson r=−0.501, P<0.001). ADAMTS9 inhibited the GC cell metastasis by down-regulating CCL5 and CXCL11 expressions.
DNMT3A mainly regulated the promotor methylation of ADAMTS9 gene
According to the MethHC database(http://methhc.mbc.nctu.edu.tw/php), the ADAMTS9 promotor is hypermethylated in the GC, thereby leading to the down-regulation or silencing of the expression of ADAMTS9 (Figure 4A). We also validated the hypermethylation of ADAMTS9 promotor by using the 5-Aza (2 µM) treatment in the AGS and BGC-823 cell lines. Figure 4B shows that the mRNA expression of ADAMTS9 significantly increased after 5-Aza (2 µM) treatment. MassARRAY analysis was adopted to examine the methylation of ADAMTS9 DNA promotor. Figure 4C and Supplementary Figure S4A show that aberrant methylation was detected in BGC823 cells treated with 5-Aza (2 µM). In human tissues, DNMT1, DNMT3A, and DNMT3B are the most common methyltransferases that play key roles in various biological functions. We then explored the mechanism underlying the regulation of promotor methylation of ADAMTS9 DNA. AGS and BGC-823 cell lines were transfected with shDNMT1, shDNMT3A, or shDNMT3B (Supplementary Figure S5). PCR and Western blot analysis results showed that the DNMT3A should be the main methyltransferase that regulate ADAMTS9 expression (Figure 4D). DNA methylation analysis through MassARRAY platform also confirmed that downregulated DNMT1 and DNMT3A expressions significantly increased the methylation level of ADAMTS9 gene in the BGC823 cell, especially for DNMT3A (Figure 4Eand Supplementary Figures S4B and S4C).
RNF180 restores ADAMTS9 expression by promoting DNMT3A ubiquitination and degradation
Our previous studies reported RNF180 plays the tumor suppressive roles on patients with GC. In this study, the IHC results of TMAs showed a significant positive correlation between RNF180 and ADAMTS9 expression.(N=126, Pearson r=0.454, P<0.001, Figure 5A and Supplementary Figure S6). In addition, with the correlation analysis from the GEPIA database (http://gepia.cancer-pku.cn/), we found that the mRNA expression level of RNF180 might be positively correlated with that ADAMTS9 (Spearman r=0.28, P=5e-09, Pearson r=0.18, P=0.00024, Figure 5B)Therefore, we hypothesized that the RNF180 restores ADAMTS9expression by decreasing ADAMTS9 methylation. Figure 5C shows the results of PCR and Western blot analysis, which indicated that RNF180 upregulated ADAMTS9 expression through the epigenetic pathway. We then performed the methylation analysis of ADAMTS9 in the BGC823 cells transfected with RNF180, and the significant downregulated methylation of ADAMTS9 DNA was detected (Figure 5Dand Supplementary Figure S4D). These results confirmed our previous hypothesis.
Considering that RNF180 restores the expression of ADAMTS9 by decreasing the methylation of ADAMTS9 promotor and that the methylation of ADAMTS9 promotor is mainly regulated by DNMT3A, we hypothesized that RNF180 ubiquitinated DNMT3A, which promoted their degradation via proteasome pathway. The CHX pulse-chase assay was performed to examine the function of RNF180 on DNMT3A stability. We observed that RNF180 promoted the degradation of DNMT3A protein and reduced their half-life. (Figure 5E). AGS and BGC823 cells were then incubated with MG132 (10 µM) to inhibit protein degradation via proteasome pathway. Figure 5F shows that RNF180 promoted the accumulation of DNMT3A after treatment with MG132. Hence, we speculated that DNMT3A might be the substrate of RNF180. To explore the interaction between RNF180 and DNMT3A, we performed the co-immunoprecipitation assay. The reciprocal co-immunoprecipitation confirmed the interaction between DNMT3A and RNF180 (Figure 5G). Ubiquitination assay was also performed to validate that the ubiquitination level of DNMT3A was greatly promoted by RNF180 (Figure 5H). In conclusion, ADAMTS9 was restored by RNF180 via promoting DNMT3A ubiquitination and degradation.