Downregulated miR-7 is correlated with high RelA/p65 in GC
To investigate the expression profile of miR-7 and RelA/p65 in GC, we evaluate expression of miR-7, RelA/p65 and NF-kB downstream metastatic targets from TCGA STAD and NCBI GEO database. We found low miR-7 and high RelA/p65 expressions in the primary gastric tumors compared with non-tumorous tissues (Fig. 1a). Meanwhile, mRNA levels of NF-kB downstream metastasis-related genes including Vimentin, MMP-2, MMP-9, VEGF, ICAM-1 and VCAM-1 were significantly enhanced in the primary gastric tumors (Additional file 3: Figure S1). To elucidate correlation of miR-7 and NF-κB subunits, we performed gene correlation analysis from NCBI GEO data sets and showed that low miR-7 was significantly correlated with RelA/p65 mRNA expression (Fig. 1b) instead of NFKB1/p50 and NFKB2/p52 (Fig. 1c). Moreover, we further investigated the correlation of miR-7 and NF-kB downstream metastatic targets. Consistently, miR-7 expression displayed a negative correlation with mRNA expression of Vimentin, MMP-2, MMP-9, VEGF, VCAM-1 and ICAM-1 (Fig. 1d-g). Thus, these results indicated that downregulated miR-7 might be associated with aberrant RelA/p65-mediated NF-kB.
Low miR-7 and high RelA/p65 expressions predict poor prognosis of GC patients
Improved survival rate of GC patients has been considered the most important therapeutic objective in advanced GC cancer. To investigate the prognostic and potentially predictive value of miR-7 and RelA/p65 in GC, we conducted a serial of survival analysis for OS, FPS, PPS, PFS and DMFS through TCGA STAD database. Survival rate analysis showed that miR-7 expression was positively whereas RelA/p65 was negatively correlated with OS, FPS, PPS, PFS and DMFS (Fig. 2a-e) among GC patients in TCGA cohort. These results suggest that miR-7 and RelA/p65 levels are potential prognostic biomarkers for GC patients.
Downregulation of miR-7 with a mechanism involving in an impaired mature processing of miR-7 biogenesis in GC cells
Mature miR-7 is individually transcribed and processed from 3 different gene locus in human genome. In these three miR-7 isoforms, miR-7-1 and miR-7-3 are located within introns of the HNRPK and PGSF1 genes, respectively, whereas miR-7-2 is intergenic [7]. Thus, we analyzed approximate host genes mRNA expression of the miR-7-1 and miR-7-3 in TCGA STAD database. HNRPK, the miR-7-1 host gene showed higher expression (Fig. 3a) whereas PGSF1, the host gene for miR-7-3, showed lower expression in GC versus normal tissues (Fig. 3b). These data showed that decreased transcription of the host gene was not responsible for the decreased miR-7 expression, at least for miR-7-1, implying that other mechanisms were responsible.
To further considerate mechanisms for the dysregulated miR-7 in human GC, we firstly quantitated mature miR-7 expression in five human GC cell lines (SGC-7901, BGC-823, MKN-45, HGC-27 and MKN-28) as well as human normal GES-1 and HEK-293 cells. Compared with GES-1 and HEK-293 cells, HGC-27 and MKN-28 cells exhibited a marked decrease of mature miR-7 whereas SGC-7901, BGC-823 and MKN-45 GC cell lines displayed no significant mature miR-7 difference (Fig. 3c), indicating mature miR-7 is dysregulated in HGC-27 and MKN-28 GC cell lines.
miR-7 biogenesis is usually controlled by three key processing: transcriptional processing of pri-miRNAs from genomic DNA, the processing of pri-miRNAs to pre-miRNAs and the mature processing from pre-miRNAs to miR-7. Since the expression profiles of pri-miR-7 s and pre-miR-7 s are absent in TCGA STAD database, we quantitated pri- and pre- forms of all three miR-7 isoforms in GC cells using modified real-time PCR. However, we did not observe significant alterations of pri-miR-7 s (pri-miR-7-1,2,3) and pre-miR-7 s (pre-miR-7-1,2,3) in HGC-27 and MKN-28 cells compared with GES-1 and HEK-293 cells (Fig. 3d-e). Meanwhile, we also observed relative higher or normal HNRPK levels but lower PGSF1 mRNA expressions in GC cells (Additional file 4: Figure S2A), whose expressive trends are similar with the results from TCGA STAD, further indicating that the host gene transcriptions were not responsible for the decreased miR-7 expression.
To obtain more mechanistic insights into the dysregulated miR-7 expression, we analyzed the mRNA levels of Drosha and Dicer1 enzymes, which are essential to pri-miRNA and pre-miRNA processing respectively. We found relative higher or normal Drosha levels but significant lower Dicer1 expressions in GC cells (Additional file 4: Figure S2B), suggested that dysregulated Dicer1 expression might contribute to the downregulation of miR-7 in GC cells. Therefore, we next conducted pre-miRNA-7 processing and binding assays to test Dicer1 enzyme activity using pre-miR-7-1. To this end, we generated 110-nt single stranded (ss) pre-miR-7-1 RNA as substrate by DNA cloning and T7 polymerase in vitro transcription and labeling (Additional file 5: Figure S3) and further performed these analysis using cell extracts or Dicer containing IP complex respectively. In cell extracts, we found that HGC-27 and MKN-28 cells displayed decreased processing abilities to product 21-23 nt mature miR-7 from 110 nt pre-miR-7-1 RNA and reduced in vitro binding activities to pre-miR-7-1 RNA compared with other cells (Fig. 3f-g, upper panel). Interestingly, using Dicer1 containing IP complex, we found that equal amount of Dicer1 IP complex from different cells extracts have similar processing and binding activities of pre-miR-7-1 RNA (Fig. 3f-g, lower panel). Therefore, these data strongly indicate that dysregulated Dicer1 expression impaired pre-miR-7 processing thereby contributed to downregulation of miR-7 expression in GC cells.
Restoration of miR-7 decreases viability and invasiveness of GC cells
Given that miR-7 is downregulated in GC, we next investigated whether delivering mature miR-7 has therapeutic potential in GC. We generated lentiviral vectors harboring mature miR-7 (LV-miR-7) for the restoration of miR-7 in HGC-27 and MKN-28 cells (Additional file 6: Figure S4A-B). QRT-PCR showed that lentivirus-mediated miR-7 transfection stably restored exogenous miR-7 levels up to 2 folds upregulation compared with controls (Additional file 6: Figure S4C). CCK-8 assay showed a significant decrease of cell viability from day 3 following mature miR-7 transfection in both HGC-27 and MKN-28 cells (Fig. 4a). Cell proliferative activity was also confirmed by the Ki67 expression with FACS analysis (Fig. 4b) and IF staining (Additional file 7: Figure S5). Meanwhile, colony formation assay showed that miR-7 transfection significantly decreased colony numbers and diminished colony sizes compared with control groups (Fig. 4c). In parallel, cell cycle assay showed that miR-7 transfection induced a marked increase in G0/G1 fraction and a significant decrease in G2-M fraction but not obvious sub-G0/G1 population (Fig. 4d), indicating a typical G0/G1 cell cycle arrest. In addition, cell apoptosis analysis by Annexin V-PE/7-AAD staining showed no significant cell apoptosis or cell death (Fig. 4e). These results suggest that miR-7 suppressed GC cells viability by inducing cell cycle arrest but not cell apoptosis.
To further elucidate the role of miR-7 in metastasis in vitro, we assessed the impact of miR-7 on cell migration and cell invasion in vitro. Cell migration assay showed that miR-7 inhibited the ability to migrate across the transwell membrane in HGC-27 and MKN-28 cells (Fig. 4f). While, cell invasion assay showed that miR-7 expression markedly abrogated the ability of these cells to invade into the Matrigel-coated member (Fig. 4g). Collectively, these in vitro data demonstrate the capacity of miR-7 in suppressing GC metastasis in vitro.
Delivering miR-7 inhibited GC distant metastasis and improved OS in vivo
Distant liver and lung metastasis are the most frequent forms of advanced GC metastasis [2]. We next performed complementary in vivo studies to explore anti- metastasis effects of miR-7 using experimental metastasis mice models by adoptive transfer of HGC-27 and MKN-28 cells expressing miR-7 or control. We found that body weight from mice injected with HGC-27 and MKN-28 cells expressing miR-7 was markedly decreased starting at day 33 and day 34 compared with corresponding controls (Fig. 5a-b). Subsequently, tumor metastasis was evaluated by analyzing metastatic nodes in lung and liver tissues at the end of these experiments. In HGC-27 injected mice, miR-7 restoration significantly decreased lung but not liver metastasis in vivo (Fig. 5c). Whereas a significantly decreased number of metastatic nodes was observed in both lung and liver tissues from MKN-28-injected mice models (Fig. 5d). Furthermore, H & E staining for embedded lung and liver tissues showed that high amount of tumor cells infiltrated into lungs and livers obtained from control groups but not from the miR-7 transfection group (Fig. 5e-f), suggesting that delivering miR-7 inhibited GC distant lung and liver metastasis in vivo.
Given that miR-7 is correlated with better OS in GC patients, we also assessed the effect of miR-7 on OS in metastatic mice models. As shown in Fig. 5g, delivery of miR-7 significantly prolonged OS of metastatic models from HGC-27 and MKN-28 cells, indicating that delivering miR-7 displays an ability to improve OS in metastasis mice models. These in vivo observations indicated that miR-7 may serve as a potential therapeutic tool for anti-metastasis of GC.
VEGF driven-hemangiogenesis and lymphangiogenesis are involved in anti-metastasis activation of miR-7 in GC
Hemangiogenesis and lymphangiogenesis, two important initial steps, play crucial roles in tumor metastasis [18]. Therefore, we evaluated hemangiogenesis in metastatic lung and liver tissues by IHC staining using CD34 as an hemendothelial marker. As shown in Fig. 6a, a significant decreased MVD in metastatic lung and liver tissues of miR-7 transfected group as indicated by CD34 staining. Like hemangiogenesis, lymphangiogenesis has gained much attention for the increase of the risk for metastasis as an important initial step both in human tumors and animal models [18]. We found decreased LYVE-1 staining densities in metastatic lung and liver tissues of miR-7 groups compared with control groups (Fig. 6b). These results indicate a robust inhibition of hemangiogenesis and lymphangiogenesis by miR-7 in metastatic tissues.
To explore possible mechanisms, we explored the difference expression of VEGF, a critically major driver of hemangiogenesis and lymphangiogenesis during the tumor metastasis [18]. We found miR-7 transfection markedly reduced VEGF production as indicated by flow cytometry analysis (Fig. 6c) and IHC staining for metastatic tissues (Fig. 6d). Furthermore, we also distinguished the expression of VEGF isoforms by Real-time PCR for VEGF-A and VEGF-C and found that both VEGF-A and VEGF-C expressions were significantly downregulated in miR-7-transfected cells compared with control groups (Fig. 6e). Overall, these data show that miR-7 inhibits GC metastasis in vivo by suppressing hemangiogenesis and lymphangiogenesis via reducing VEGF-A and VEGF-C secretion.
miR-7 improves inflammation cells infiltration in vivo
Inflammation can fuel both primary tumor growth and metastasis and has been long recognized as a key aspect of cancer development [19, 20, 23]. To determine whether inflammatory cells are involved in miR-7-mediated metastatic inhibition, using T and B lymphocytes-deficient nude mice model, we analyzed the expression of inflammatory cells marker such as the pan-leukocyte marker CD45, the neutrophil marker MPO, the macrophage markers CD11b, F4/80 and other markers GR1, CD11c in metastatic tissues via IHC staining. We found that CD45 and MPO positive cells in metastatic tissues were significantly decreased in miR-7-transfected group compared with control group (Fig. 7a-b). Subsequently, we also found a significant decreased intensity of CD11c, F4/80, CD11b and Gr-1-positive cells in metastatic lung and liver tissues of miR-7 transfected group compared with control group (Fig. 7c-f). These results indicated that miR-7 could improve infiltrating immune cells-mediated inflammation in GC.
p65-mediated NF-κB activation is critical to anti-metastasis activation of miR-7 in GC
Considering the fact that multiple target genes of miR-7 and diverse regulatory miRNAs for RelA/p65, we want to know whether NF-κB activity is critical to miR-7-counteracted metastasis in GC cells. Using NF-κB luciferase reporter assay, we found that miR-7 remarkably attenuated NF-κB transcriptional activity while these inhibitions were significantly reverted upon NF-κB activator LPS stimulation in HGC-27 and MKN-28 cells (Fig. 8a), indicating miR-7 negatively regulates NF-κB transcriptional activity. To demonstrate the essential role of NF-κB activation in anti-metastasis activation of miR-7 in vitro, we further test whether LPS-induced restoration of NF-κB activity could reverse miR-7-inhibited metastasis in vitro. As we expected, LPS treatment markedly restored the abilities of cell migration and invasion with dose dependent manner in miR-7-transfected HGC-27 and MKN-28 cells (Fig. 8b-c). Moreover, we also examined NF-κB downstream metastasis-related targets levels including Vimentin, ICAM-1, VCAM-1, MMP-2 and MMP-9 through flow cytometry analysis and IHC staining. Our finding showed that miR-7-transfection significantly decreased these downstream targets expression in GC cells (Additional file 8: Figure S6A-B) and metastatic tissues (Additional file 8: FigureS6C-E), respectively. These results suggested that NF-κB transcriptional activation is critical to anti-metastasis activation of miR-7.
To determine the details for the regulation of miR-7 to NF-κB activation, we further analyzed the levels of NF-κB subunits by FACS analysis in vitro. Interestingly, miR-7 transfection did not alter p50 and p52 expressions (data not shown) but significantly decreased levels of total p65 and p-p65(ser536) (a surrogate marker for NF-κB activity) in HGC-27 and MKN-28 cells (Fig. 8d-e). Moreover, IF analysis showed that miR-7 obviously decreased both cytoplasmic and nuclear distribution of p65 and p-p65(ser536) in GC cells (Fig. 8f-g), Meanwhile, we did not observe typical nuclear export of p65 in these cells. In addition, decreased p65 and p-p65(ser536) expressions were also confirmed by IHC staining in metastatic tissues (Fig. 8h-i). indicating that decreased NF-κB/p65 expression should be responsible for the miR-7-inhibited NF-κB activation. Together, these data suggest that miR-7 negatively controls NF-κB transcriptional activity and its downstream metastasis-associated targets expression by decreasing NF-κB/p65 activation and thereby inhibits GC metastasis.