Tumors are not merely cancerous cells that undergo mindless proliferation. Rather, they are highly organized and interconnected organ systems. Tumor cells reside in complex microenvironments in which they are subjected to a variety of physical and chemical stimuli that influence cell behavior and ultimately the progression and maintenance of the tumor. As cancer bioengineers, it is our responsibility to create physiologic models that enable accurate understanding of the multi-dimensional structure, organization, and complex relationships in diverse tumor microenvironments. Such models can greatly expedite clinical discovery and translation by closely replicating the physiological conditions while maintaining high tunability and control of extrinsic factors. In this review, we discuss the current models that target key aspects of the tumor microenvironment and their role in cancer progression. In order to address sources of experimental variation and model limitations, we also make recommendations for methods to improve overall physiologic reproducibility, experimental repeatability, and rigor within the field. Improvements can be made through an enhanced emphasis on mathematical modeling, standardized in vitro model characterization, transparent reporting of methodologies, and designing experiments with physiological metrics. Taken together these considerations will enhance the relevance of in vitro tumor models, biological understanding, and accelerate treatment exploration ultimately leading to improved clinical outcomes. Moreover, the development of robust, user-friendly models that integrate important stimuli will allow for the in-depth study of tumors as they undergo progression from non-transformed primary cells to metastatic disease and facilitate translation to a wide variety of biological and clinical studies.

Within the ovarian cancer tumor microenvironment, cancer stem-like cells (CSC) interact with carcinoma associated mesenchymal stem/stromal cells (CA-MSC) through multiple secreted cytokines and growth factors. These paracrine interactions have been revealed to cause enrichment of CSC and their chemoprotection; however, it is still not known if platelet-derived growth factor (PDGF) signaling is involved in facilitating these responses. In order to probe this undiscovered bidirectional communication, we created a model of ovarian malignant ascites in the three-dimensional (3D) hanging drop heterospheroid array, with CSC and CA-MSC. We hypothesized that PDGF secretion by CA-MSC increases self-renewal, migration, epithelial to mesenchymal transition (EMT) and chemoresistance in ovarian CSC. Our results indicate that PDGF signaling in the CSC-MSC heterospheroids significantly increased stemness, metastatic potential and chemoresistance of CSC. Knockdown of PDGFB in MSC resulted in abrogation of these phenotypes in the heterospheroids. Our studies also reveal a cross-talk between PDGF and Hedgehog signaling in ovarian cancer. Overall, our data suggest that when the stromal signaling via PDGF to ovarian CSC is blocked in addition to chemotherapy pressure, the tumor cells are significantly more sensitive to chemotherapy. Our results emphasize the importance of disrupting the signals from the microenvironment to the tumor cells, in order to improve response rates. These findings may lead to the development of combination therapies targeting stromal signaling (such as PDGF and Hedgehog) that can abrogate the tumorigenic, metastatic and platinum resistant phenotypes of ovarian CSC through additional investigations.

In this protocol, we outline the procedure for generation of tumor spheroids within 384-well hanging droplets to allow for high-throughput screening of anti-cancer therapeutics in a physiologically representative microenvironment. We outline the formation of patient derived cancer stem cell spheroids, as well as, the manipulation of these spheroids for thorough analysis following drug treatment. Specifically, we describe collection of spheroid morphology, proliferation, viability, drug toxicity, cell phenotype and cell localization data. This protocol focuses heavily on analysis techniques that are easily implemented using the 384-well hanging drop platform, making it ideal for high throughput drug screening. While we emphasize the importance of this model in ovarian cancer studies and cancer stem cell research, the 384-well platform is amenable to research of other cancer types and disease models, extending the utility of the platform to many fields. By improving the speed of personalized drug screening and the quality of screening results through easily implemented physiologically representative 3D cultures, this platform is predicted to aid in the development of new therapeutics and patient-specific treatment strategies, and thus have wide-reaching clinical impact.