Nuclear MYH9-induced CTNNB1 transcription, targeted by staurosporin, promotes gastric cancer cell anoikis resistance and metastasis

Abstract

Rationale: Peritoneal metastasis predicts poor prognosis of gastric cancer (GC) patients, and the underlying mechanisms are poorly understood. Methods: The 2-DIGE, MALDI-TOF/TOF MS and single-cell transcriptome were used to detect differentially expressed proteins among normal gastric mucosa, primary GC and peritoneal metastatic tissues. Lentiviruses carrying shRNA and transcription activator-like effector nuclease technology were used to knock down myosin heavy chain 9 (MYH9) expression in GC cell lines. Immunofluorescence, immune transmission electron microscopy, chromatin fractionation, co-immunoprecipitation, and assays for chromatin immunoprecipitation, dual luciferase reporter, agarose-oligonucleotide pull-down, flow cytometry and cell anoikis were performed to uncover nuclear MYH9-induced β-catenin (CTNNB1) transcription in vitro. Nude mice and conditional transgenic mice were used to investigate the findings in vivo. Results: We observed that MYH9 was upregulated in metastatic GC tissues and was associated with a poor prognosis of GC patients. Mechanistically, we confirmed that MYH9 was mainly localized in the GC cell nuclei by four potential nuclear localization signals. Nuclear MYH9 bound to the CTNNB1 promoter through its DNA-binding domain, and interacted with myosin light chain 9, β-actin and RNA polymerase II to promote CTNNB1 transcription, which conferred resistance to anoikis in GC cells in vitro and in vivo. Staurosporine reduced nuclear MYH9 S1943 phosphorylation to inhibit CTNNB1 transcription, Wnt/β-catenin signaling activation and GC progression in both orthotropic xenograft GC nude mouse and transgenic GC mouse models. Conclusion: This study identified that nuclear MYH9-induced CTNNB1 expression promotes GC metastasis, which could be inhibited by staurosporine, indicating a novel therapy for GC peritoneal metastasis.

Keywords: CTNNB1; MYH9; anoikis resistance; gastric cancer; metastasis.

Figure 1

MYH9 was upregulated in metastatic GC tissues and associated with poor survival of GC patients. (A) Illustration of 2D-DIGE and MALDI-TOF/TOF MS analyses for GC tissues. N, normal gastric mucosae; T, primary GC tissues; M, peritoneal metastasis tissues. (B) t-distributed stochastic neighbor embedding (t-SNE) plot of 10,189 single cells from two advanced GC patients. The tissues included normal gastric epithelium (N), primary tumor (PT) and peritoneal metastasis (MT). Clusters were assigned to indicated cell types by differentially expressed genes (see also Figure S3 and Table S7). (C) The level of MYH9 mRNA in epithelium-derived cells (Cluster 6, 7 and 8) was analyzed using the single-cell transcriptome data (Kruskal-Wallis, p < 2.2e-16). (D) The Kaplan-Meier survival analysis of overall survival in TCGA GC data based on MYH9 expression. The level of MYH9 mRNA was divided into low (<12th percentile) and high (>12th percentile) groups for analysis.

Figure 2

MYH9 promotes CTNNB1 expression in GC cells. (A) Morphology of MYH9 shRNA3-transfected or NC-transfected MGC 80-3 cells (green) were observed under a fluorescence microscope. The nude cells (no green) served as internal controls. (B) Schematic illustration of a pair of TALENs (L2/R3) binding to MYH9 exon 2. (C) Monoclonal MGC 80-3 and AGS sublines were generated by transfection of L2 and R3 plasmids and then subjected to qPCR analysis of MYH9 (a) and CTNNB1 (b) mRNA. WT, wild type monoclonal cells; Type 1, monoclonal cells whose single chain of MYH9 DNA was edited; Type 2, monoclonal cells whose double chains of MYH9 DNA were edited. (D) Three MGC 80-3 monoclonal cells with MYH9 knocked down were subjected to western blot analysis. Three wild-type monoclonal cells were used as controls. (E) Monoclonal MGC-80-3 cells (MU1) was transfected with MYH9 plasmid and subjected to western blot analysis. (F) The correlation between MYH9 and CTNNB1 mRNA expression in gastric epithelium-derived cells (including both gastric epithelial and cancer cells) was analyzed using the single-cell transcriptome data. (G) The correlation between MYH9 and CTNNB1 mRNA expression in GC tissues and normal gastric mucosae was analyzed using the GEPIA dataset.

Figure 3

MYH9 interacts with MYL9, β-actin, RNAPII and TFIIB in the nucleus of GC cells. (A) AGS, MGC 80-3 and HGC-27 cells were grown and subjected to immunofluorescence staining (MYH9, red; DAPI, blue) and confocal microscopy. (B) AGS cells were subjected to immune transmission electron microscopic analysis of MYH9 protein (black dots). (C) MGC 80-3 cells were subjected to chromatin fractioning and western blot. 1, 0.4 M chromatin fraction; 2, 0.8 M chromatin fraction; 3, chromatin residual pellet. 4, control. (D) MGC 80-3 cells were subjected to nuclear protein extraction, IP and western blot. Rabbit IgG was used as the negative control. (E) AGS cells were subjected to immunofluorescence staining (MYH9, MYL9 and β-actin proteins) and confocal microscopy.

Figure 4

Nuclear MYH9 binds to the CTNNB1 promoter with special sequences to promote CTNNB1 transcription in vitro. AGS cells were co-transfected with a Firefly luciferase reporter plasmid, a Renilla luciferase reporter plasmid and a gene overexpression plasmid for 24 h and then subjected to luciferase reporter assay. The Firefly luciferase reporter plasmid contains different lengths of the CTNNB1 promoter cDNA (C1-7). The gene overexpression plasmids contain MYH9, MYL9 or β-actin cDNA. (B) MGC 80-3 and AGS cells were subjected to ChIP analysis. MYH9, MYL9, β-actin, RNAPII and TFIIB antibodies were used. GAPDH primers served as a positive control, while IgG served as a negative control. Nuclear extracts of AGS cells were prepared and pulled down with biotinylated oligonucleotides (C, O1-5; D, O3L, O3M, O3R and O3S) targeting the CTNNB1 promoter (the top graph), which were then subjected to western blot analysis of MYH9 protein. Input served as a positive control. O3S was a scramble biotinylated oligonucleotide that served as a negative control. (E) Schematic representation of mutated CTNNB1 promoters that were cloned into the upstream of the luciferase reporter (M1-7). (F) MYH9-overexpressed (LV-MYH9) AGS cells were co-transfected with a Firefly luciferase reporter plasmid (M1-7) and a Renilla luciferase reporter plasmid for 24 h, and then subjected to luciferase reporter assay. WT, the C3 plasmid, served as a positive control.

Figure 5

STS inhibits nuclear MYH9 phosphorylation at S1943 to decrease CTNNB1 expression. (A) Structural illustration of CTNNB1 promoter (CTNNB1-P) and MYH9 protein. a-c. Ribbon diagram of three potential interactions between CTNNB1-P and MYH9 DBDs using Pymol software. d. Surface representation of the interaction (grey, MYH9; yellow, CTNNB1-P) confirmed by our research. (B) MGC 80-3 and AGS cells were transiently transfected with three different mutated DNA binding sites (DBS1M, DBS2M or DBS3M) after knockdown of MYH9 expression using MYH9 shRNA3, and then subjected to qPCR analysis of CTNNB1 and MYH9 mRNAs. GAPDH mRNA served as an internal control. (C) MGC 80-3 cells with MYH9 knockdown (MYH9-KD) were transiently transfected with different NLS mutation plasmids (M1-15) and then subjected to qPCR analysis of CTNNB1 mRNA. (D) AGS and MGC 80-3 cells with MYH9 knockdown (MYH9-KD) were transfected with different MYH9 mutation plasmids (G1916 or G1943) for 12 h, then treated with STS (100 nmol/L) for 24 h. The cells were treated with Wnt3a (20 ng/mL) for 8 h before being harvested for RNA cleavage. The expression of CTNNB1 mRNA was detected by qPCR. (E) MGC 80-3 cells with MYH9 knockdown (TALEN-MYH9-KD) were transfected with MYH9 mutation plasmids (G1916 or G1943), treated with STS (100 nmol/L) and Wnt3a (20 ng/mL), and then subjected to nuclear protein extraction and western blot analysis of β-catenin expression. (F) MGC 80-3 cells were treated with STS (25, 50, 75, 100 and 125 nmol/L) for 24 h and then subjected to western blot analysis of p-MYH9 (S1943) and β-catenin levels.

Figure 6

MYH9-mediated CTNNB1 transcription promotes anoikis resistance in GC cells by increasing the activation of Wnt/β-catenin signaling. AGS cells were co-transfected with a MYH9 plasmid (wild type or mutated), a luciferase reporter plasmid (TOP-flash or FOP-flash) and a Renilla control plasmid for 16 h, and then subjected to luciferase reporter assay. The cells also treated with Wnt3a (20 ng/ml) for 8 h before the assay. (B) MGC 80-3 cells transfected with MYH9 shRNA3 were treated with Wnt3a (20 ng/mL) for 8 h, and then subjected to qPCR analysis of β-catenin-induced target genes. (C) MGC 80-3 cells were subjected to annexin-V and 7-AAD staining for flow cytometry analysis. (D) MGC 80-3 cells transfected with different plasmids (NC+vector, shRNA3+vector, shRNA3+MYH9 or shRNA3+CTNNB1) were seeded into 12-well plates and then subjected to a soft agar colony formation assay for 21 days. The colony formation was observed under naked eyes, the bright-field microscope (200x) and fluorescence microscope (200x; see also Figure S20A and Figure S20B). (E) AGS cells transfected with different plasmids (NC+vector, shRNA3+vector, shRNA3+MYH9 or shRNA3+CTNNB1) were seeded into a poly-hema pre-coated plate (or a control cell culture plate), grown for 24 h, then subjected to the MTT assay. (F) MGC 80-3 cells were injected into the mouse tail vein. Eight weeks after tumor cell injection, the mice were sacrificed and subjected to a multi-functional in vivo imaging system. (G) The number of lung and liver metastatic nodules per mouse was counted and analyzed.

Figure 7

MYH9-mediated CTNNB1 expression promotes gastric cancer progression and metastasis in vivo. (A) Survival of LSL-hMYH9; Atp4b-cre; Tff1^-/- mice (n = 20) compared with Tff1^-/- mice (n = 20; p = 0.0099), Atp4b-cre mice (n = 20; p < 0.0001), or LSL-hMYH9; Atp4b-cre mice (n = 20; p< 0.0001) were analyzed using the log-rank test. (B) Survival of Myh9^fl/fl; Atp4b-cre; Tff1^-/- mice (n = 20) compared with Tff1^-/- mice (n = 20; p = 0.027), Atp4b-cre mice (n = 18; p = 0.048) or Myh9^fl/fl mice (n = 19; p = 0.042) were analyzed using the log rank test. (C) Representative stomach tumors in Myh9^fl/fl; Atp4b-cre; Tff1^-/- mice (5/20) at 450 days old and LSL-hMYH9; Atp4b-cre; Tff1^-/- mice (16/20) at 600 days old. Myh9^fl/fl; Tff1^-/- mice (8/20) and LSL-hMYH9; Tff1^-/- mice (9/20) were shown as negative controls, respectively. (D) Representative stomach tumors in LSL-hMYH9; Atp4b-cre; Tff1^-/- mice treated with STS (9/10) or DMSO (8/10). (E) GC tissues from LSL-hMYH9; Atp4b-cre; Tff1^-/- mice treated with STS (9/10) or DMSO (8/10) were resected and subjected to qPCR analysis of β-catenin-induced mRNA. (F) Whole-body fluorescence images of metastatic tumors in an orthotropic xenograft GC nude mouse model implanted with AGS cells and treated with STS. (G) The metastatic nodules (IM, intestinal metastasis; PM, peritoneal metastasis) were detected using a hematoxylin-eosin (H&E) stain.