Background The tumor microenvironment is a highly complex and continuously adapting tissue. The crosstalk between the different cell types and non-cellular components is difficult to monitor, specifically with the aim to identify possible novel anticancer drug targets.

Methods We co-cultivated two human NSCLC cell lines, H1437 and Calu-1, human fetal dermal fibroblasts (HDF) or murine lung fibroblasts (muLF) and human monocytes in different combinations of co- and triple cultures. To discriminate between phenotypic events driven by cell-cell contact against those induced by secreted factors only, experiments were conducted either applying complex tissue cultures with cell-cell contact or in trans-well plates or using cultures with conditioned media. 40 human and 23 mouse cytokines were determined at four time points using a bead based multiplex assay. Monocyte differentiation and polarization was analyzed via flow cytometry.

Results CXCL6, IL-1b and CCL20 were detected in the supernatant exclusively when co-cultivating human H1437 with HDF in the presence of human monocytes. The elevated levels of GM-CSF, IL-2, CCL23 and CCL8 in the triple culture were HDF-derived (all >5fold). CXCL6 and IL-1b displayed highest concentrations in the cell-cell contact setting, whereas the expression of CCL20 was similar in all conditions. In contrast, when analyzing the same setting using Calu-1 instead of H1437 only CXCL6 and CCL20 were exclusively determined in the triple culture setting. IL-1b together with GM-CSF and CXCL1 were secreted by the HDF and IL-2, CXCL11, CCL3 and CCL15 were Calu-1 derived.

The use of muLF induced a cell contact dependent upregulation of murine MIP-1a (>5fold) and murine IL-1a (> 5fold). In contrast, the human tumor cell lines secreted different human cytokines in the presence of muLF as compared to HDF. A decrease in human CXCL12 was observed in the presence of muLF but not in co-cultures with HDF.

Monocyte derived dendritic cells were decreased when tumor cells or fibroblasts were added. Calu-1 increased the percentage of CD40+ antigen presenting cells, whereas H1437 had no major impact. The addition of tumor cells shifted the M1/M2 ratio towards M2 which was supported by the presence of HDF or muLF.

Conclusions The current dataset elucidates the interplay between different cellular and secreted factors in the tumor microenvironment. The outcome was influenced by the tumor model as well as the fibroblast source. Further investigations will provide more detail about the influence of the molecular make-up of the tumor model, donor variability of monocytes and fibroblast source.