Fig 2: Omental adipocytes facilitate GC growth through the CXCL2–VEGFA axis in a humanised omental adipose tissue model using NSG mice.RFP-labelled shNT-OmPrAd or shCXCL2-OmPrAd cells was subcutaneously injected into the left flank of NSG mice. One week later, AGS cells were subcutaneously injected in the vicinity. shNT-OmAd group (n = 7); shCXCL2-OmAd group (n = 7). Each graph represents the mean ± SE from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. a Red fluorescent protein expression in omental preadipocytes. RFP was detected in OmPrAd transfected with shNT and shCXCL2. b Efficiency of shRNA. Omental preadipocytes were transfected with shRNA of non-targeting genes and CXCL2. Bar graph represents the relative ratio of shNT and shCXCL2 to control using 2-??Ct by qPCR. Mean, 1.0 (control), 0.95 (shNT-OmPrAd), 0.59 (shCXCL2-OmPrAd). c Humanised omental adipose tissues. RFP was detected in adipose tissues where shNT-OmPrAd or shCXCL2-OmPrAd was subcutaneously injected. d Body weight curves. e Representative macroscopic images of GC tumours. Excised tumours were shown at 11 weeks after transplantation of GC cells. f Tumour growth curve. Tumour size and volume were measured twice per week. g Representative bioluminescence images. h Quantification of bioluminescence signals. Total luminescence flux from the tumour region was quantified using MetaMorph-MIIS software. Mean, 6.9 × 108 (shNT-OmAd), 1.0 × 108 (shCXCL2-OmAd). i VEGFA expression in tumour tissues. VEGFA expression levels were analysed using qPCR in tumour tissues. Each graph bar represents the relative ratio of 2-??Ct of the shCXCL2-OmAd group to that of shNT-OmAd. Mean, 1.0 (shNT-OmAd), 0.55 (shCXCL2-OmAd). j Representative images of Ki67 immunohistochemistry (×200). Representative images of Ki67 immunostaining in tumour tissues. k Ki67 index. Each bar represents the average rate of Ki67-positive cells in GC tumour tissues (×400). Mean, 40.2 (shNT-OmAd), 6.5 (shCXCL2-OmAd). l Representative images of tumour angiogenesis (×200). m Quantification of tumour angiogenesis. Each bar represents the average number of CD31-positive microvessels in GC tumour tissues (×400). Mean, 63.4 (shNT-OmAd), 13.2 (shCXCL2-OmAd). n Serum CXCL2 level in NSG mice. Each bar represents the average serum CXCL2 level in the shNT-OmAd and shCXCL2-OmAd groups. Mean, 20.8 (shNT-OmAd), 10.8 (shCXCL2-OmAd). o Urinary level of CXCL2 in human GC patients. Each bar represents the average urinary level of CXCL2 in GC patients with or without peritoneal metastasis. Mean, 0.55 (non-metastasis), 15.2 (peritoneal metastasis). p Urinary level of VEGFA in human GC patients. Each bar represents the average urinary level of VEGFA in GC patients with or without peritoneal metastasis. Mean, 344 (non-metastasis), 457 (peritoneal metastasis).

Fig 3: Gene expression and immunohistochemical analysis of SOX9. (A) Timeline of experimental protocols of AML12 for quantitative reverse transcription PCR (qRT-PCR) and immunofluorescence labeling of cells. AML12 cells were seeded at 5.0 × 104 cells·mL-1 into 6-well plates (1.5 × 105 cells/well). After incubation for 24 h, medium with 10 ng·mL-1 of KC or MIP-2 was changed at 24 and 72 h. (B) Sox9 mRNA levels in AML12 cells collected after 120 h of incubation was determined using qRT-PCR. mRNA levels were normalized using 18s rRNA as a housekeeping gene. Student's unpaired two-tailed t-test was performed. Results are represented as mean ± SEM (n = 3). (C) Immunofluorescent staining of SOX9 in AML12 cells treated with or without KC and MIP-2 after 120 h of incubation. The number of SOX9-positive cells was quantified per 50 cells. Fisher's exact test was performed to compare treated and control samples. Scale bar, 100 µm. (D) Timeline of the experimental protocols of primary hepatocytes for qRT-PCR and immunofluorescence labeling of cells. Primary hepatocytes were seeded at 1.0 × 105 cells·mL-1 into 6-well plates (2.0 × 105 cells/well). After incubation for 24 h, medium with 10 ng·mL-1 of KC or MIP-2 was changed at 24 and 48 h. (E) Sox9 mRNA levels in primary hepatocytes collected after 72 h of incubation was determined using qRT-PCR. mRNA levels were normalized using 18s rRNA as a housekeeping gene. Student's unpaired two-tailed t-test was performed. Results are represented as mean ± SEM (n = 3). (F) Immunofluorescent staining of SOX9 in primary hepatocytes treated with or without KC and MIP-2 after 72 h of incubation. Primary hepatocytes were cultured in the absence or presence of 10 ng·mL-1 of KC or MIP-2. The number of SOX9-positive cells was quantified per 50 cells. Fisher's exact test was performed to compare treated and control samples. Scale bar, 100 µm. PH, primary hepatocyte.

Fig 4: CXCL2 in omental adipocytes is critical for GC cell growth/migration and in vitro angiogenesis.Each graph represents the mean ± SE from three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001. a Gene expression of GRO family in omental adipocyte. Bar graph represents the relative ratio of each RNA expression to CXCL2 expression level, using 2-??Ct, where ?Ct indicates the difference in Ct values between each gene and ß-actin [?Ct = Ct (target gene) - Ct (ß-actin)]. Mean, 0.41 (CXCL1), 1.0 (CXCL2), 0.18 (CXCL3). b Western blotting. c Efficiency of siRNA. Omental preadipocytes were transfected with non-targeting siRNA and CXCL2. The left graph represents RNA gene expression of CXCL2 in the OmAd cells. Bar graph represents the relative ratio of siNT and siCXCL2 to control, using 2-??Ct, where ?Ct indicates the difference in Ct values between each gene and ß-actin [?Ct = Ct (target gene) - Ct (ß-actin)]. The right graph represents protein concentration of CXCL2 in the OmAd-CM. Mean, RNA level: 1.0 (control), 1.0 (siNT), 0.27 (siCXCL2); protein level: 11.0 (control), 11.0 (siNT), 1.3 (siCXCL2). d Cell growth. Shown are the relative ratios of absorbance under each condition of OmAd-CM to those under control media (n = 5). Mean, AGS: 1.0 (control), 1.2 (control OmAd-CM), 1.3 (siNT OmAd-CM), 1.1 (siCXCL2 OmAd-CM); IM95: 1.0 (control), 1.3 (control OmAd-CM), 1.2 (siNT OmAd-CM), 1.0 (siCXCL2 OmAd-CM). e Representative images of migration assay (×100). f Quantification of migration assay. Migrated GC cells were counted from averages at four microscopic fields, and each result was presented as the mean of at least three independent experiments. Each value represents the mean relative ratio of migrated GC cells under each type of OmAd-CM to those under control media. Mean, AGS: 1.0 (control), 3.8 (control OmAd-CM), 4.3 (siNT OmAd-CM), 1.7 (siCXCL2 OmAd-CM); IM95: 1.0 (control), 2.5 (control OmAd-CM), 2.9 (siNT OmAd-CM), 0.7 (siCXCL2 OmAd-CM). g Representative images of EC recruitment assay (×100). h Quantification of EC recruitment assay. Migrated HMVECs were counted from averages at four microscopic fields, and each result was presented as the mean of at least three independent experiments. Each value represents the mean relative ratio of migrated HMVECs co-cultured with GC cells treated with each type of OmAd-CM to those co-cultured with GC cells treated with control media. Mean, AGS: 1.0 (control), 1.4 (control OmAd-CM), 1.6 (siNT OmAd-CM), 0.7 (siCXCL2 OmAd-CM); IM95: 1.0 (control), 1.8 (control OmAd-CM), 2.2 (siNT OmAd-CM), 1.0 (siCXCL2 OmAd-CM). i Representative images of tube-formation assay (×100). j Quantification of tube-formation assay. Each value represents the mean number of branched tubes under each condition. Mean, AGS: 4.0 (control), 9.0 (siNT OmAd-CM), 3.3 (siCXCL2 OmAd-CM); IM95: 2.0 (control), 7.0 (siNT OmAd-CM), 2.8 (siCXCL2 OmAd-CM). k Semi-comprehensive analysis for angiogenic factors. RNA was extracted from both AGS and IM95 cells before and after OmAd-stimulation. Each bar represents the relative ratios of each gene expression in OmAd-treated GC cells to those in non-treated GC cells. l VEGFA expression in GC cells. Relative ratios are expressed with 2-??Ct, where ?Ct indicates the difference in Ct values between each gene and ß-actin. Mean, AGS: 1.0 (control), 1.8 (control OmAd-CM), 2.1 (siNT OmAd-CM), 1.4 (siCXCL2 OmAd-CM); IM95: 1.0 (control), 2.1 (control OmAd-CM), 2.4 (siNT OmAd-CM), 1.9 (siCXCL2 OmAd-CM). m AKT phosphorylation and HIF1a expression in GC cells. Each protein extracted from GC cells incubated with control media, OmAd-CM, siNT OmAd-CM or siCXCL2 OmAd-CM for 24 h was immunoblotted with anti-phospho-AKT, anti-AKT and anti-HIFa antibodies. Each band density was quantified with Image J. ß-actin is shown as a loading control.

Fig 5: Proposed model of interaction between omental adipocytes and GC growth.CXCL2 secreted from omental adipocytes activates AKT phosphorylation of gastric cancer cells, which directly promotes gastric cancer growth and motility. In addition, VEGFA in gastric cancer cells is also upregulated through HIF1a upregulation, resulting in angiogenesis. Consequently, gastric cancer cells were transformed to a more aggressive phenotype, which induces peritoneal metastasis thorough recruitment to the omentum itself.