Stromal fibroblasts mediate extracellular matrix remodeling and invasion of scirrhous gastric carcinoma cells

Abstract

Scirrhous gastric carcinoma (SGC) has the worst prognosis of all gastric cancers, owing to its rapid expansion by invasion and frequent peritoneal dissemination. Due to the increased proliferation of stromal fibroblasts (SFs) that occurs within SGC lesions and the peritoneal metastatic sites, SFs have been proposed to support the progression of this disease. However, the biological and molecular basis and the pathological role of the intercellular interaction between SGC cells and SFs remain largely unknown. In this study, we investigated the role of SFs in the invasion of the extracellular matrix (ECM) by SGC cells. When SGC cells were cocultured with SFs derived from SGC tissue on three-dimensional (3D) Matrigel, they were attracted together to form large cellular aggregates that invaded within the Matrigel. Time-lapse imaging revealed that this process was associated with extensive contraction and remodeling of the ECM. Immunofluorescence and biochemical analysis showed that SGC cells stimulate phosphorylation of myosin light chain and actomyosin-mediated mechanical remodeling of the ECM by SFs. By utilizing this assay system for inhibitor library screening, we have identified several inhibitors that potently suppress the cooperation between SGC cells and SFs to form the invasive structures. Among them, a Src inhibitor dasatinib impaired the interaction between SGC cells and SFs both in vitro and in vivo and effectively blocked peritoneal dissemination of SGC cells. These results indicate that SFs mediate mechanical remodeling of the ECM by SGC cells, thereby promoting invasion and peritoneal dissemination of SGC.

Figure 1. Formation of invasive foci by SGC cells and SFs on 3D Matrigel.

A and B, 44As3 and CaF37 cells were labeled with CellTracker and cultured individually or together on either 2D (A) or 3D (B) Matrigel for 2 days. C, Non-SGC gastric cancer cell lines (MKN7, MKN1, and MKN74) and SGC cell lines (58As9, HSC59, HSC44PE, and 44As3) were cocultured with CaF37 cells on 3D Matrigel for 2 days. D, The relative number of invasive foci formed by CaF37 and indicated gastric cancer cell lines. Bars show mean ± SEM (n = 4). *, p<0.005 vs SGC cell lines, by ANOVA with Tukey's test.

Figure 2. Imaging of the formation of invasive foci and concomitant remodeling of ECM.

A, Time-lapse imaging of 44As3 and CaF37 cells cultured on 3D Matrigel. Yellow arrowheads denote chain-like structures formed by 44As3 cells. Lower panels are magnified image sequences of the boxed region. White arrowheads denote cell protrusions of CaF37 cells. B, Movement of the microbeads that were embedded in 3D Matrigel was tracked for 9 h and trajectories of each bead were shown as colored lines. C, Quantification of the movement of microbeads. Bars show mean ± SEM (n = 48 beads analyzed). *, p<0.0001 by Student's t-test. Similar results were obtained in three independent experiments. D, 44As3 and CaF37 cells were cultured on fluorescent gelatin-coated cover slips for 16 h and stained for F-actin and the nucleus (DAPI). Arrowheads denote the black areas where gelatin matrices were removed from cover slips. E, Quantification of the areas of gelatin detachment. Bars show mean ± SEM (n = 6). *, p<0.0001 by Student's t-test.

Figure 3. Formation of invasive foci and associated remodeling of ECM require actomyosin contraction.

A, 44As3 and CaF37 cells cultured on gelatin-coated cover slips were stained with the antibody against phospho-myosin light chain 2 (p-MLC2) and DAPI. B, Immunoblot analysis of cell lysates prepared from individually cultured or cocultured (coculture) 44As3 and CaF37 cells. A mixture of cell lysates of individually cultured cells was used as a control (mixture). C, 44As3 and CaF37 cells were cultured on fluorescent gelatin in the absence or presence of blebbistatin (10 µM) for 16 h. Arrowheads denote the areas where gelatin matrices were disrupted. D, Quantification of the areas of gelatin disruption. Bars show mean ± SEM (n = 4). *, p<0.000001 by Student's t-test. E, 44As3 and CaF37 cells were cocultured on 3D Matrigel in the absence or presence of blebbistatin (10 µM) for 2 days. F, The relative number of invasive foci was determined. Bars show mean ± SEM (n = 4). *, p<0.005 by Student's t-test.

Figure 4. Rock and Src regulate the formation of invasive foci.

A, 44As3 and CaF37 cells were plated onto 3D Matrigel in the presence or absence of H1152 (10 µM) or dasatinib (10 µM) for 2 days. B, Quantification of the number of invasive foci. Bars show mean ± SEM (n = 3). *, p<0.05 by Student's t-test. C, Dose response effects of H1152 and dasatinib on cell growth of 44As3 and CaF37 cells and the formation of invasive foci. Bars show mean ± SD (n = 8 for cell growth and 3 for invasive foci). *, p<0.05 by Student's t-test. D, Immunoblot analysis of 44As3 and CaF37 cells that were cocultured and treated with H1152 or dasatinib. E, The effect of H1152 and dasatinib on remodeling of the gelatin matrix. F, Quantification of the areas of gelatin disruption. Bars show mean ± SEM (n = 3). *, p<0.0005 by Student's t-test.

Figure 5. Dasatinib inhibits cell aggregation, migration, and invasion in vitro.

A, Fluorescent images of individually cultured and cocultured 44As3 and CaF37 cells on 3D Matrigel in the presence or absence of dasatinib (1 or 10 µM) for 2 days. B, Quantification of the areas and number of cell clusters. Bars show mean ± SEM (n = 48–502 for the cluster area, 6 for the number of clusters). *, p<0.05; **, p<0.001 by Student's t-test. C, Total and net migration distances of cells per hour were measured from time-lapse movies. Bars show mean ± SEM (n = 40). *, p<0.0001 by Student's t-test. D, Invasion depth of the invasive foci or cell clusters. Bars show mean ± SD (n = 20). *, p<0.0001 by Student's t-test.

Figure 6. Dasatinib suppresses peritoneal dissemination of SGC cells and their association with stromal fibroblasts in vivo.

A, 44As3 cells were intraperitoneally injected into nude mice and DMSO or dasatinib was administered via intraperitoneal injection. The number of mesentery nodules was calculated as described in the Materials and Methods. Bars show mean ± SEM (n = 10). *, p<0.005 by Mann-Whitney test. B, Representative macroscopic views of metastatic tumor nodules (arrowheads) formed in the mesentery. C, Immunofluorescence analysis of the mouse mesenteries bearing tdTomato-labeled 44As3 tumor nodules. Arrowheads denote the regions where FSP1 positive stromal fibroblasts were accumulated around tumor nodules. D, Mesentery nodules were stained with hematoxylin and eosin and anti-αSMA antibody for histological examination.

Figure 7. Schematic diagram of the SF-mediated invasion of SGC cells.

SGC cells and SFs physically interact with each other through active migration and cell-cell contact, which is dependent on Src activity and therefore blocked by dasatinib treatment. This interaction promotes Rock-dependent phosphorylation of MLC, resulting in actomyosin contraction and mechanical ECM remodeling during invasion of SGC. In peritoneal dissemination, SGC cells may also associate with SFs to invade mesentery and form tumor nodules.