MALDI imaging-guided microproteomic analyses of heterogeneous tumors – A proof of concept with breast adenocarcinoma

Background: Intratumor heterogeneity (ITH) is a well-known fact in histopathology that may greatly influence the evolution of cancers and the clinical outcome of patients. Recently, matrix-assisted laser desorption/ionization (MALDI) imaging was demonstrated to be adequate to study proteomic ITH and its implication in prognostic stratification of patients. However, there are some drawbacks concerning protein identification. On the other hand, laser microdissection (LMD)-based microproteomics allows retrieving thousands of protein identifications from small tissue pieces. As a proof of concept, we combined these two complementary approaches to analyze heterogeneous regions in breast tumors. Methods: Invasive ductal breast cancer FFPE tissue sections from five patients were analyzed by MALDI imaging using an ultrafleXtreme MALDI TOF mass spectrometer (MS) (Bruker Daltonics, Bremen, DE). The resulting dataset was processed by segmentation using SCiLS Lab (SCiLS GmbH, Bremen, DE). Regions that were found heterogeneous within tumors were collected by microdissection and further processed for micro-proteomics, in duplicates. An ultra performance liquid chromatography (UPLC) 2D nanoAcquity (Waters, Milford, USA) was used in combination with a Q Exactive MS(Thermo Fischer Scientific, Waltham, USA). Max-quant and Perseus were used for data normalization and statistical analyses. Peptide identifications from LC-MS/MS data were compared to m/z detected by MALDI imaging. Results: Liquid chromatography-tandem mass spectrometry data were classified by hierarchical clustering. Heterogeneous tissue regions were discriminated on the basis of their actual molecular heterogeneity. The dataset was correlated with MALDI imaging to identify m/z values discriminating heterogeneous regions. It was possible to identify Keratin, type II cytoskeletal 8 (KRT8) and Keratin, type I cytoskeletal 19 (KRT19), corresponding to m/z 911.5 and 850.5 respectively in the MALDI image as heterogeneous proteins within a tumor. Conclusion: The molecular characterization of cell clones in tumors relat- ed to bad patient outcome could have great impact for pathology. In the future, this combined workflow could be used for biomarker discovery assays associated to ITH, using large cohorts of patients. Tissue clusters determined by MALDI imaging could be correlated to prognostic values. Regions corresponding to good and bad prognosis could be compared by microproteomics to find markers of ITH correlated to bad survival.