Spatially resolved proteogenomics on human melanoma samples reveals evolutionary trajectories in tumor progression

Although melanoma is less common than other types of skin cancer, its aggressiveness and strong ability to metastasize render it a particularly dangerous neoplastic disease. The molecular evolution from benign precursor lesions to malignant melanoma is a complex and dynamic process, driven by the acquisition of genetic alterations and leading to increased progressivity and invasiveness. While the underlying genetic changes of this transformation have been investigated extensively over the past decades, pathophysiological details of tumor progression and metastasis on a phenotypic level are less well understood. To this end, we set out to apply our recently developed workflow for spatially resolved cell-type specific proteogenomics based on Deep Visual Proteomics (DVP) on a cohort of microscopic formalin-fixed paraffin-embedded (FFPE) sections of melanoma patients. Our aim was to relate genomic mutations to protein levels, providing novel insights to metastatic cancer evolution.

We retrospectively collected FFPE biopsies from melanoma patients displaying at least two different disease progression stages and performed automated immunofluorescence staining for SOX10 and PRAME using an Agilent Dako Omnis platform. AI-driven image classification identified distinct pathological disease stages. Based on this, we excised shapes of similar phenotypic identity using a Leica LMD7 laser microscope and separated DNA and peptides on Evotips, aiming to acquire fully matched genomic and proteomic data sets. We also conducted Next Generation Sequencing (NGS) on an Illumina NovaSeq 6000 and mass spectrometry (MS)-based proteomics on a Thermo Fisher Orbitrap Astral.

Our workflow yielded high-quality whole exome sequences, permitting robust variant calling, and enabled the identification of more than 5000 proteins per single cell shape type on average. We predicted the consequences at the protein level by creating patient-specific FASTA files containing the genetic mutations, increasing the identification of proteoforms by more than 10% compared to the standard UniProt search file. This strategy allowed us to quantify the differential expression of mutated alleles on protein level and molecularly distinguish developmental stages of melanoma. We further conducted gene set enrichment analysis on the proteomic data and related genetic changes to the activation of pathways involved in cell proliferation and metastasis.

Applying our workflow on a clinical melanoma cohort serves as first demonstration of the benefits of integrating deep spatially resolved genomics and proteomics data directly in single cell types in vivo. Specifically, spatial proteogenomics will enable us to investigate trajectories of clonal cancer cell evolution. More broadly, our interdisciplinary technology promises to reveal unique insights into the mechanisms of tumor pathogenesis and beyond.