Preview

PDF, English Download (8MB) | Terms of use

Abstract

Colorectal cancer (CRC) is the third most common cancer worldwide and a leading cause of cancer-related mortality. Despite advances in molecular stratification and the integration of genomic and transcriptomic data into clinical decision-making, treatment options for patients with metastatic CRC remain limited, and therapeutic resistance is common. A key challenge in precision oncology is the inability of DNA and RNA profiling alone to capture the functional state of cancer cells, particularly the activity of signaling pathways that govern cell behavior. To address this gap, this thesis presents a phosphoproteomic characterization of CRC within the framework of the DKFZ/NCT/DKTK-MASTER precision oncology program. I developed and optimized workflows for protein extraction, clean-up, digestion, and phosphopeptide enrichment from fresh-frozen clinical tissue samples using a high-throughput liquid handling robotic platform. These protocols enabled consistent and scalable preparation of patient samples with minimal hands-on time, while maintaining high proteome and phosphoproteome coverage. In parallel, I fine-tuned mass spectrometry acquisition and data analysis strategies to control false discovery rates and improve quantification accuracy, even in samples with low phosphosite abundance. The resulting dataset revealed substantial heterogeneity in kinase signaling activity among CRC patients. By inferring kinase activation states from phosphosite profiles, I identified three kinase-based patient subgroups with distinct biological and clinical features, including differences in immune infiltration. These subtypes did not overlap with established transcriptomic classifications, highlighting the unique value of phosphoproteomic data in uncovering functional tumor phenotypes. Importantly, I demonstrate that genomic alterations such as gene amplifications (e.g., ERBB2, CDH17) do not consistently translate to elevated protein expression or activity. This disconnect suggests that relying solely on DNA or RNA data for therapeutic stratification may lead to suboptimal treatment choices. Moreover, comparison of tumor tissues with patient-derived organoids (PDOs) revealed marked divergence in proteomic signatures, calling into question the fidelity of PDOs as preclinical models, particularly in the absence of the tumor microenvironment. Overall, this work provides both technical and conceptual advances for integrating phosphoproteomics into precision oncology workflows. It underscores the need to consider protein-level information when making clinical decisions and sets the stage for future studies evaluating proteome-informed treatment strategies in CRC and beyond.