Integrated Analysis of Mouse and Human Gastric Neoplasms Identifies Conserved microRNA Networks in Gastric Carcinogenesis

Abstract

Background & aims: microRNAs (miRNAs) are small noncoding RNAs that bind to the 3' untranslated regions of mRNAs to promote their degradation or block their translation. Mice with disruption of the trefoil factor 1 gene (Tff1) develop gastric neoplasms. We studied these mice to identify conserved miRNA networks involved in gastric carcinogenesis.

Methods: We performed next-generation miRNA sequencing analysis of normal gastric tissues (based on histology) from patients without evidence of gastric neoplasm (n = 64) and from TFF1-knockout mice (n = 22). We validated our findings using 270 normal gastric tissues (including 61 samples from patients without evidence of neoplastic lesions) and 234 gastric tumor tissues from 3 separate cohorts of patients and from mice. We performed molecular and functional assays using cell lines (MKN28, MKN45, STKM2, and AGS cells), gastric organoids, and mice with xenograft tumors.

Results: We identified 117 miRNAs that were significantly deregulated in mouse and human gastric tumor tissues compared with nontumor tissues. We validated changes in levels of 6 miRNAs by quantitative real-time polymerase chain reaction analyses of neoplastic gastric tissues from mice (n = 39) and 3 independent patient cohorts (n = 332 patients total). We found levels of MIR135B-5p, MIR196B-5p, and MIR92A-5p to be increased in tumor tissues, whereas levels of MIR143-3p, MIR204-5p, and MIR133-3p were decreased in tumor tissues. Levels of MIR143-3p were reduced not only in gastric cancer tissues but also in normal tissues adjacent to tumors in humans and low-grade dysplasia in mice. Transgenic expression of MIR143-3p in gastric cancer cell lines reduced their proliferation and restored their sensitivity to cisplatin. AGS cells with stable transgenic expression of MIR143-3p grew more slowly as xenograft tumors in mice than control AGS cells; tumor growth from AGS cells that expressed MIR143-3p, but not control cells, was sensitive to cisplatin. We identified and validated bromodomain containing 2 (BRD2) as a direct target of MIR143-3p; increased levels of BRD2 in gastric tumors was associated with shorter survival times for patients.

Conclusions: In an analysis of miRNA profiles of gastric tumors from mice and human patients, we identified a conserved signature associated with the early stages of gastric tumorigenesis. Strategies to restore MIR143-3p or inhibit BRD2 might be developed for treatment of gastric cancer.

Keywords: Progression; Stomach; Transcription Factor; Tumor Suppressor.

Figure 1.. miRNA signature in mouse and human gastric cancer

A) The number of significantly de-regulated miRNAs in Stage I or II human gastric cancer (GC), as compared with normal gastric tissues from non-cancer patients (NG) (P<0.05). B) The number of significantly de-regulated miRNAs in human tumor-adjacent normal gastric tissue (TANG) samples, as compared with NG (P<0.05). C) A Venn diagram analysis of panels A and B depicts 676 miRNAs that are similarly up-regulated in GC and TANG, as compared to NG (P<0.05). D) A Venn diagram analysis of panels A and B depicts 46 miRNAs that are similarly down-regulated in GC and TANG, as compared to NG (P<0.05). E) Hierarchical cluster analysis of miRNA expression in GC (stages I and II), normal gastric (NG), and tumor-adjacent normal gastric (TANG) tissues. F) The number of significantly de-regulated miRNAs in mouse gastric cancers (mGC) from the TFF1-KO mice 0 model, as compared with mouse normal gastric (mNG) tissues (P<0.05). G) Hierarchical cluster analysis of miRNA expression in mouse tissues (mGC and mNG). H) A Venn diagram analysis of significantly de-regulated miRNA in human gastric cancer and mouse gastric cancer samples demonstrates consistent and conserved up-regulation of 58 miRNAs and down-regulation of 5 miRNAs in both mouse and human (P<0.05). Upper panel for up-regulated miRNAs. Lower panel for down-regulated miRNAs. I) Circular plot from Ingenuity pathway analysis of sequencing data identified 18 significantly de-regulated signaling pathways (inner circle) in human (middle circle) and mouse (outer circle) gastric cancer samples, the number of miRNAs regulating each pathway is denoted.

Figure 2.. Quantitative real-time RT-PCR (qRT-PCR) validation of expression levels of up-regulated miRNAs in human and mouse gastric tumors.

A) Expression analysis of MIR135B-5p analysis in TFF1-KO gastric low-grade dysplasia (LGD) and cancer, as compared to normal gastric tissues. (B) Expression analysis of MIR135B-5p in human gastric cancer (GC) as compared to tumor-adjacent normal gastric (TANG) tissues, cohort 1 (United States). (C) Expression analysis of MIR135B-5p in GC as compared to normal gastric tissues from non-cancer patients (NG), cohort 2 (Chile). (D) Expression analysis of MIR135B-5p in NG, TANG, and GC tissues, cohort 3 (China). E-H) Expression analysis of MIR196B-5p (E-H) and MIR92A-5p (I-L) in the same samples as in A-D. *, P<0.05, **P<0.01, ***P<0.001, Mann Whitney Test.

Figure 3.. Quantitative real-time RT-PCR validation of expression levels of down-regulated miRNAs in human and mouse gastric tumors.

A) Expression analysis of MIR143-3p analysis in TFF1-KO gastric low-grade dysplasia (LGD) and cancer, as compared to normal gastric tissues. (B) Expression analysis of MIR143-3p in human gastric cancer (GC) as compared to tumor-adjacent normal gastric (TANG) tissues, cohort 1 (United States). (C) Expression analysis of MIR143-3p in GC as compared to normal gastric tissues from non-cancer patients (NG), cohort 2 (Chile). (D) Expression analysis of MIR143-3p in NG, TANG, and GC tissues, cohort 3 (China). E-H) Expression analysis of MIR204-5p (E-H) and MIR133–3p (I-L) in the same samples as in A-D. *, P<0.05, **P<0.01, ***P<0.001, Mann Whitney Test.

Figure 4.. MIR143-3p suppresses cellular proliferation and gastric organoids’ growth.

Reconstitution of MIR143-3p was established by using lenti-virus infection followed by puromycin selection (A-H). A) qRT-PCR analysis of MIR143-3p expression level following its stable reconstitution in STKM2 cells, as compared to control (left panel). ATP-glo cell viability assay analysis of MIR143-3p stably reconstituted STKM2 cells and control cells (right panel). Similar experiments, as in A, were performed in MKN45 (B), AGS (C), and MKN28 (D). E) EdU immunofluorescence staining in MIR143-3p stably reconstituted STKM2 cells and controls (left), quantification of data is shown on the right panel. Similar experiments, as in E, were performed in MKN45 (F), AGS (G), and MKN28 (H). Reconstitution of MIR143-3p in gastric organoids from mouse low-grade dysplasia (LGD) lesions in TFF1-KO mouse (I-N). I) Light field images of mouse gastric organoids (Day 1, 3, 5, and 7). J) Light field images of mouse gastric organoids, 3 days after MIR143-3p reconstitution using lenti-virus infection, or control lentivirus. K) Quantification data of organoids’ diameter from J, Mann Whitney Test. L) qRT-PCR analysis of MIR143-3p expression levels, following reconstitution of MIR143-3p in gastric organoids and controls. M) Ki-67 immunofluorescence staining in gastric organoids, 3 days after MIR143-3p reconstitution using lenti-virus infection, or control lentivirus. N) Quantification data of Ki67 positive cells in M. *, P<0.05, **P<0.01, ***P<0.001, Student t test.

Figure 5.. BRD2 is a direct downstream target of MIR143-3p.

A) BRD2 is a predicted downstream target of MIR143-3p, by using three online databases. B) Two predicted MIR143-3p binding sites on human BRD2 3′UTR are shown. C) Western blot analysis of BRD2, MYC and β-actin protein expression levels in 4 normal human gastric tissues (NG) and AGS, MKN28, MKN45, and SNU-1 gastric cancer cell lines. D) Western blot analysis of BRD4, BRD2 and β-actin protein expression levels in AGS cell, following transient reconstitution of MIR143-3p using a mimic (2.5–40 pmol). BRD2, not BRD4, is downregulated following reconstitution. Western blot analysis following reconstitution of MIR143-3p in AGS (E), STKM2 (F) MKN45 (G) and mouse LGD gastric organoids (H). Wild-type or mutant (missing both MIR143-3p bindings sites) BRD2 3′UTR luciferase reporter analysis in AGS cells (I) or MKN45 (J) following stable reconstitution of MIR143-3p or control. K) c-MYC promoter (4xTBE1 wt) luciferase reporter analysis in AGS cells following reconstitution of MIR143-3p or control, with and without BRD2 transient transfection.K) qRT-PCR analysis of c-MYC gene expression level in mouse LGD gastric organoids following reconstitution of MIR143-3p or control. *, P<0.05, **P<0.01, ***P<0.001, Student t test or one-way ANOVA.

Figure 6.. MIR143-3p sensitizes gastric cancer cells to cisplatin treatment.

A) IC50 analysis using ATP-glo assay in STKM2 cells with stable reconstitution of MIR143-3p, or control, treated with CDDP (0, 0.75, 1.55, 3.1, 6.25, 12.5, 25, 50 μmol/L). B-D) Similar experiments in MKN45 (B), AGS (C), and MKN28 (D) cells. IC50s were calculated using the Prism 5 software. E) Representative clonogenic survival assay in AGS cells with stable reconstitution of MIR143-3p, or controls, with or without 2.5μM of CDDP treatment, quantification of surviving colonies is showing in F. G) IC50 analysis based on clonogenic assay similar to E (CDDP: 10, 5, 2.5, 1.25, 0.625 μmol/L) for 24h and measurements at day 7. H) Western blot analysis of BRD2, total PARP (PARP), cleaved PARP (c-PARP), c-MYC, and β-actin protein expression level in MKN45 cells with stable reconstitution MIR143-3p, or control, following treatment with CDDP or JQ-1 (BRD2 inhibitor) alone or in combination. ***P<0.001, Student t test.

Figure 7.. MIR143-3p inhibits gastric cancer tumor xenograft growth and sensitizes to cisplatin treatment in vivo.

A) Representative images of gastric cancer cells (AGS) tumor xenografts using stable reconstitution of MIR143-3p or control with and without treatment with CDDP. B) qRT-PCR analysis of MIR143-3p expression in tumor xenografts. C) Xenograft tumor growth rates. D) Western blot analysis of BRD2, total PARP (PARP), cleaved PARP (c-PARP), c-MYC, and β-actin protein expression levels tumor xenografts. E) Immunohistochemistry staining of Ki-67 or cleaved-caspase 3 in tumor xenografts (left panel) with quantification of immunohistochemistry data (right panel). *, P<0.05, **P<0.01, ***P<0.001, Student t test or one-way ANOVA.