Fibroblast hepatocyte growth factor promotes invasion of human mammary ductal carcinoma in situ

Abstract

Stromal-derived hepatocyte growth factor (HGF) acting through its specific proto-oncogene receptor c-Met has been suggested to play a paracrine role in the regulation of tumor cell migration and invasion. The transition from preinvasive ductal carcinoma in situ (DCIS) to invasive breast carcinoma is marked by infiltration of stromal fibroblasts and the loss of basement membrane. We hypothesized that HGF produced by the infiltrating fibroblasts may alter proteolytic pathways in DCIS cells, and, to study this hypothesis, established three-dimensional reconstituted basement membrane overlay cocultures with two human DCIS cell lines, MCF10.DCIS and SUM102. Both cell lines formed large dysplastic structures in three-dimensional cultures that resembled DCIS in vivo and occasionally developed invasive outgrowths. In coculture with HGF-secreting mammary fibroblasts, the percentage of DCIS structures with invasive outgrowths was increased. Activation of c-Met with conditioned medium from HGF-secreting fibroblasts or with recombinant HGF increased the percentage of DCIS structures with invasive outgrowths, their degradation of collagen IV, and their secretion of urokinase-type plasminogen activator and its receptor. In agreement with the in vitro findings, coinjection with HGF-secreting fibroblasts increased invasiveness of MCF10.DCIS xenografts in severe combined immunodeficient mice. Our study shows that paracrine HGF/c-Met signaling between fibroblasts and preinvasive DCIS cells enhances the transition to invasive carcinomas and suggests that three-dimensional cocultures are appropriate models for testing therapeutics that target tumor microenvironment-enhanced invasiveness.

Figure 1. HGF-secreting mammary fibroblasts increase invasive outgrowths from DCIS 3D structures and their invasion through rBM

MCF-10A and MCF10.DCIS cells were grown in 3D rBM overlay culture in the absence (Control) or presence of mammary fibroblasts (MF) or HGF-secreting mammary fibroblasts (MF:HGF). A, Representative DIC images are illustrated in the 1^st and 2^nd column (bar, 50 μm) and representative merged images of green, red and DIC channels are shown in the 3^rd column (bar, 100 μm). MCF10.DCIS cells were prelabeled with CellTracker Orange (red); fibroblasts were prelabeled with CellTracker Green (green) and embedded within the rBM. B, Images of MCF10.DCIS cells alone and in co-culture with fibroblasts were scored for the number of total structures and those with invasive outgrowths. Data in graphs are pooled from 14 fields of view in three independent experiments and are presented as the mean percentage of 3D structures with invasive outgrowths ± SD. *, p<0.02, (Student's t-test). C, Transwell invasion assay without fibroblasts (Control) or with fibroblasts grown in the well below the rBM-coated Transwell filter. Data in graphs are average number of cells that invaded through rBM coated Transwell filters, pooled from three independent experiments, and presented as mean ± SD. *, p<0.02, (Student's t-test).

Figure 2. Conditioned media from HGF-secreting mammary fibroblasts stimulates invasive outgrowths of MCF10.DCIS cells, their invasion through rBM and degradation of DQ-collagen IV

A, B and C: MCF10.DCIS cells were grown in 3D rBM culture in the absence of fibroblast-conditioned media (Control) or the presence of media conditioned by mammary fibroblasts (MF CM) or HGF-secreting mammary fibroblasts (MF:HGF CM). A, MF:HGF CM stimulates invasive outgrowths (arrow) of cells from 3D structures; bar, 20 μm. B, Images were scored for the percentage of structures with invasive outgrowths. Graphs represent data pooled from 14 fields of view from each of three separate experiments and are presented as mean percentage of 3D structures with invasive outgrowths ± SD. *, p<0.02, significant increase as compared to control and MF CM values (Student's t-test). C, MCF10.DCIS cells were grown in 3D rBM culture containing DQ-collagen IV without (Control) or with MF CM or MF:HGF CM for 4 days before imaging. Graphs represent average integrated fluorescence intensity per nuclei in relative fluorescence units (RFU) and are pooled from three separate experiments and presented as mean ± SEM (n=18). *, p<0.02, significant increase as compared to control and MF CM values (Student's t-test). D, MCF10.DCIS cells were grown in 3D rBM overlay culture on coverslips for 4 days, labeled with CellTracker Orange and the coverslips inverted onto glass bottom dishes coated with rBM, containing DQ-collagen IV, that had been diluted with either MF CM or MF:HGF CM. Cultures were imaged live following 18 hours. Representative DIC images and corresponding merged images of DIC, CellTracker Orange (red) and degraded DQ-collagen IV (green) channels are shown; bar, 100 μm. Each merged image represents a focal plane observed in x/y, y/z and x/z axes.

Figure 3. Phosphorylation of c-Met by MF:HGF CM is correlated with increased expression and secretion of uPA and uPAR by DCIS cells

MCF10.DCIS and SUM102 cells were grown in 3D rBM cultures with either 2 μM SU11274 (INHIBITOR), MF:HGF conditioned media (MF:HGF CM) or MF:HGF conditioned media plus 2 μM SU11274 (MF:HGF CM + I). A, 3D rBM cultures were harvested and lysed in sample buffer, separated by 12% SDS-PAGE, transferred to nitrocellulose and analyzed with antibodies against phospho Y1234/Y1235 c-Met. Membranes were subsequently stripped and probed with antibodies against total c-Met and GAPDH as a loading control. B, Cell lysates and concentrated conditioned media from 3D rBM cultures were separated by 12% SDS-PAGE with loading based on protein concentration and relative volume, respectively. Proteins were transferred to nitrocellulose and analyzed with antibodies against uPA and uPAR.

Figure 4. Recombinant HGF acting through c-Met signaling stimulates invasive outgrowths and increased secretion of uPA and uPAR from DCIS cells

MCF10.DCIS and SUM102 cells were grown in 3D rBM culture with either 100 ng/ml HGF (rHGF), 100 ng/ml HGF + 2 μM SU11274 (rHGF + I), 2 μM SU11274 (I) or DMSO (Control). A, Representative DIC images depicting condition before harvesting cell lysates and conditioned media; bar, 200 μm. B, Images from MCF10.DCIS and SUM102 cell cultures were scored for the number of total structures and those with invasive outgrowths. C, Cell lysates and concentrated CM were separated by 12% SDS-PAGE under non-reducing conditions with lysates loaded based on protein concentration and CM loaded based on the protein concentration of the corresponding cell lysates. Proteins were transferred to nitrocellulose membranes and analyzed with antibodies against uPA or uPAR.

Figure 5. HGF increased degradation of DQ-collagen IV by MCF10.DCIS and SUM102 3D cultures

MCF10.DCIS or SUM102 cells grown in 3D rBM cultures containing DQ-collagen IV were treated with 100 ng/ml HGF (HGF), 100 ng/ml HGF plus aprotinin (HGF+A) or left untreated (Control). Integrated fluorescence due to proteolysis (green) was normalized to the number of cells (nuclei; blue). Representative fluorescence micrographs of one confocal plane of both the degraded DQ-collagen IV channel (degraded DQ-IV) and the corresponding DIC channel are illustrated for MCF10.DCIS (A) and SUM102 (B) 3D rBM cultures. Bar, 20 μm. C, Quantification of proteolysis in 3D structure volumes. Data were pooled from 3 representative experiments and are presented as mean ± SEM (n=18). Student's t-test, p<0.02.

Figure 6. Co-injection with HGF-secreting fibroblasts increases invasiveness of MCF10.DCIS xenografts

Representative 10× images of H&E stained MCF10.DCIS xenografts. MCF10.DCIS xenografts exhibited either atypical hyperplasias (A) or comedo DCIS lesions (B). MCF10.DCIS/MF xenografts exhibited primarily atypical hyperplasias (C) with a few progressing to an invasive phenotype (D). All of the MCF10.DCIS/MF:HGF xenografts had progressed to invasive ductal carcinoma (E and F).