Fig 2: Identification of CADM3 as a function-related target of FABP4. (A) Volcano plot (left panel) and lollipop chart (right panel) showing the correlations between FABP4 the differentially expressed genes identified by RNA-seq. (B) The results of real-time PCR verifying that the mRNA levels of FABP4 and CADM3 were significantly positively correlated in 36 GC samples. (C) Representative images of FABP4 and CADM3 staining in tissue microarrays are presented (left panel), and CADM3 expression in various FABP4 groups of the internal cohort (right panel) was calculated. (D) Scatter plots showing the correlations between FABP4 and CADM3 expression levels in the TCGA-STAD dataset. (E) Peak signals indicating that FABP4 did not directly bind to the CADM3 promoter region. (F) Gene set enrichment analysis of RNA-seq data and the PPAR signalling pathway is presented. (G) MGC-803 cells were pretreated with or without fenofibrate (50 μM), GW0742 (10 μM), and rosiglitazone (20 μM) for 24 h. The cells were used for real-time PCR to detect the effects of various PPAR subtype agonists on CADM3 transcription. The data are presented as the mean ± SD (**P < 0.01).

Fig 3: Overexpression of FABP4 inhibited tumor development in vivo. (A) Pictures of tumors generated by controls cells and overexpression of FABP4 of YY‐8103 cells. (B) The growth curve of tumors (cm3). (C) The Weights of tumors (g). Each bar represented the mean ± SD. *P < 0.05; **P < 0.01; ***, P < 0.001.

Fig 4: Rosiglitazone can reverse the metastatic ability of FABP4-deficient GC cells in vitro and in vivo. (A-B) Stably transfected MGC-803 cells were treated with siPPAR-γ and rosiglitazone (20 μM). The protein levels of FABP4, PPAR-γ and CADM3 were determined by western blotting. The migration and invasion of GC cells were determined by the Transwell assays. (C) Scheme of experimental design for generation of a lung metastasis model. FABP4-deficient BGC-823 cells were transplanted into the tail veins of BALB/c nude mice, followed by daily treatment with DMSO or rosiglitazone (10 mg/kg or 20 mg/kg) (5 mice per group). (D-E) Representative images of mice and quantified results of bioluminescence imaging are presented. (F-G) The lungs were collected for H&E staining, and the number of metastatic nodules was determined. Scale bars, 1 mm. (H) Scheme of experimental design for generation of a liver metastasis model. FABP4-deficient BGC-823 cells were transplanted into the spleens of BALB/c nude mice, followed by daily treatment with DMSO or rosiglitazone (10 mg/kg or 20 mg/kg) (8 mice per group). (I-J) The livers were collected for H&E staining, and the number of metastatic nodules was determined. Scale bars, 1 mm. (K) The survival of BALB/c nude mice was monitored. The data are presented as the mean ± SD (*P < 0.05; **P < 0.01; ***P < 0.001). Rosi, rosiglitazone.

Fig 5: FABP4 promotes PPAR-γ-mediated CADM3 transcription. (A) A coimmunoprecipitation assay was performed to analyse the interaction between FABP4 and PPAR-γ in GC cells. (B) The results of immunofluorescence staining indicating that FABP4 facilitates the translocation of PPAR-γ from the cytoplasm to the nucleus. Scale bars, 20 μm. (C) The protein levels of FABP4 and PPAR-γ in GC cells measured by western blotting. (D) MGC-803 cells with or without altered FABP4 expression were treated with 25 μg/mL cycloheximide at indicated time points. (E) Stably transfected BGC-823 and MGC-803 cells were treated with or without MG132 (10 μM) for 6 h. The protein levels of PPAR-γ were analysed by western blotting. (F-G) MGC-803 cells were collected for real-time PCR or western blotting detection of time- and dose-dependent effects of rosiglitazone on CADM3 expression. (H) The PPRE region of the CADM3 promoter sequence is presented. (I-J) Dual-luciferase reporter assays were performed to investigate the effect of FABP4 and PPAR-γ on CADM3 transcription. The data are presented as the mean ± SD (*P < 0.05; **P < 0.01; ***P < 0.001). CHX, cycloheximide. Rosi, rosiglitazone. TSS, transcription start site. PPRE, peroxisome proliferator responsive element.