Pancreatic ductal adenocarcinoma (PDAC) presents a formidable clinical challenge due to its intricate microenvironment characterized by desmoplasia and complex tumor-stroma interactions. Conventional models hinder studying cellular crosstalk for therapeutic development. To recapitulate key features of PDAC masses, this study creates a novel sea-and-island PDAC tumor construct (s&i PTC). The s&i PTC consists of 3D-printed islands of human PDAC cells positioned within an interstitial extracellular matrix (ECM) populated by human cancer-associated fibroblasts (CAFs). This design closely mimics the in vivo desmoplastic architecture and nutrient-poor conditions. The model enables studying dynamic tumor-stroma crosstalk and signaling reciprocity, revealing both known and yet-to-be-discovered multicellular metabolic adaptations. Using the model, we discovered the orchestrated dynamic alterations of CAFs under nutrient stress, resembling critical in vivo human tumor niches, such as the secretion of pro-tumoral inflammatory factors. Additionally, nutrient scarcity induces dynamic alterations in the ECM composition and exacerbates poor cancer cell differentiation-features well-established in PDAC progression. Proteomic analysis unveiled the enrichment of proteins associated with aggressive tumor behavior and ECM remodeling in response to poor nutritional conditions, mimicking the metabolic stresses experienced by avascular pancreatic tumor cores. Importantly, the model's relevance to patient outcomes is evident through an inverse correlation between biomarker expression patterns in the s&i PTCs and PDAC patient survival rates. Key findings include upregulated MMPs and key ECM proteins (such as collagen 11 and TGFβ) under nutrient-avid conditions, known to be regulated by CAFs, alongside the concomitant reduction in E-cadherin expression associated with a poorly differentiated PDAC state under nutrient deprivation. Furthermore, elevated levels of hyaluronic acid (HA) and integrins in response to nutrient deprivation underscore the model's fidelity to the PDAC microenvironment. We also observed increased IL-6 and reduced α-SMA expression under poor nutritional conditions, suggesting a transition of CAFs from myofibroblastic to inflammatory phenotypes under a nutrient stress akin to in vivo niches. In conclusion, the s&i PTC represents a significant advancement in engineering clinically relevant 3D models of PDAC masses. It offers a promising platform for elucidating tumor-stroma interactions and guiding future therapeutic strategies to improve patient outcomes.
