Background The tumor´s immunogenicity is modulated by the tumor microenvironment (TME), which mostly turns out to be immunosuppressive, thereby promoting disease progression. To better understand the crosstalk between tumor, immune and stromal cells we developed an in vitro system enabling the interaction of the different cell types in a 3D environment.

Methods We cultivated three different colorectal cancer cell lines HCT-116, HT-29 and CXF269, two fibroblast lines, human dermal fibroblasts (HDF) or cancer associated fibroblasts (CAFs) and macrophages from healthy donors as mono-, co- and triple cultures in a matrigel-collagen matrix in a 96-well format. Analysis of the secreted cytokines in the supernatant as well as the differentiation of the macrophages was performed after 10 days of culture. We analyzed 45 cytokines in duplicates by using a multiplex bead-based assay. The macrophage differentiation was determined by flow cytometry (FC) based on the expression of CD45, CD11b, CD86 and CD209.

Results The macrophages in monoculture displayed a M1/M2 ratio of 1.39 after 10 days of culture. The co-culture with HDF did not change the phenotype dramatically whereas the cell-cell contact will CAFs induced a marked increase in M1 leading to a ratio of 3.36. A similar diverse pattern was seen in co-culture with tumor lines. Whereas HT29 did not influence the ratio remarkably, HCT116 reduced the ratio to 1.14 and CXF269 increased the M2 fraction clearly, leading to a M1/M2 ratio of 0.85. The triple culture with CAFs presented an additive effect on the macrophage polarization: CAFs led to an increase in the M1 population which was most pronounced in HT29 and least distinct with CXF269. The triple culture with HDF supported the M2 population most prominent in the CXF269 (M1/M2 ratio: 0.77). The presence of fibroblasts induced an enhanced secretion of IL-6, CCL2 and CCL3 in all settings. All tumor lines secreted high levels of VEGF-A and SDF-1 in mono- and co-culture with fibroblasts. However, the secretion was down regulated in the presence of macrophages. Finally, the addition of macrophages to any co-culture induced an increase of IL-1Rα in the supernatant. A cluster analysis of the cytokine data defined two major clusters mainly separated by the presence or absence of macrophages.

Conclusions The modular nature of the 3D heterotypic spheroid model enables the deconvolution of the different players in the TME. This flexible mid throughput platform will hopefully help to shed more light into the complex role of tumor associated macrophages in cancer progression.