The Challenge of Classifying Metastatic Cell Properties by Molecular Profiling Exemplified with Cutaneous Melanoma Cells and Their Cerebral Metastasis from Patient Derived Mouse Xenografts

Proteome profiling data of eight xenografted cutaneous and cerebellar metastasis of human melanoma cells were combined with eicosanoid and glutathione measurements as well as multiparametric morphologic analyses and immunofluorescence. The analyses revealed significant differences in the expression of molecules associated with metastatic properties, while lacking any commonalties when comparing the metastatic variants from four different donors. Apparently we lack appropriate meta-analysis strategies making use of the large number of presently identified metastasis-associated molecules for successful classification of cells. [Display omitted] Highlights •Comprehensive molecular profiling of cutaneous and cerebellar metastasis variants.•Identification of differentially regulated metastasis-associated molecules.•Evidence for individually distinct patterns of metastasis-associated molecules.•Highlighting the evident need for establishing meta-analyses strategies. The prediction of metastatic properties from molecular analyses still poses a major challenge. Here we aimed at the classification of metastasis-related cell properties by proteome profiling making use of cutaneous and brain-metastasizing variants from single melanomas sharing the same genetic ancestry. Previous experiments demonstrated that cultured cells derived from these xenografted variants maintain a stable phenotype associated with a differential metastatic behavior: The brain metastasizing variants produce more spontaneous micro-metastases than the corresponding cutaneous variants. Four corresponding pairs of cutaneous and metastatic cells were obtained from four individual patients, resulting in eight cell-lines presently investigated. Label free proteome profiling revealed significant differences between corresponding pairs of cutaneous and cerebellar metastases from the same patient. Indeed, each brain metastasizing variant expressed several apparently metastasis-associated proteomic alterations as compared with the corresponding cutaneous variant. Among the differentially expressed proteins we identified cell adhesion molecules, immune regulators, epithelial to mesenchymal transition markers, stem cell markers, redox regulators and cytokines. Similar results were observed regarding eicosanoids, considered relevant for metastasis, such as PGE2 and 12-HETE. Multiparametric morphological analysis of cells also revealed no characteristic alterations associated with the cutaneous and brain met