Abstract
Tumor-stroma interactions are critical in pancreatic ductal adenocarcinoma (PDAC) progression and therapeutics. Patient-derived xenograft (PDX) models recapitulate tumor-stroma interactions, but the conventional antibody-based immunoassay is inadequate to discriminate tumor and stromal proteins. Here, we describe a species-deconvolved proteomics approach embedded in IonStar that can unambiguously quantify the tumor (human-derived) and stromal (mouse-derived) proteins in PDX samples, enabling unbiased investigation of tumor and stromal proteomes with excellent quantitative reproducibility. With this strategy, we studied tumor-stroma interactions in PDAC PDXs that responded differently to Gemcitabine combined with nab-Paclitaxel (GEM+PTX) treatment. By analyzing 48 PDX animals 24 h/192 h after treatment with/without GEM+PTX, we quantified 7262 species-specific proteins under stringent cutoff criteria, with high reproducibility. For the PDX sensitive to GEM+PTX, the drug-dysregulated proteins in tumor cells were involved in suppressed oxidative phosphorylation and the TCA cycle, and in the stroma, inhibition of glycolytic activity was predominant, suggesting a relieved reverse Warburg effect by the treatment. In GEM+PTX-resistant PDXs, protein changes suggested extracellular matrix deposition and activation of tumor cell proliferation. Key findings were validated by immunohistochemistry (IHC). Overall, this approach provides a species-deconvolved proteomic platform that could advance cancer therapeutic studies by enabling unbiased exploration of tumor-stroma interactions in the large number of PDX samples required for such investigations.