Poster S4 - 3D MatriSpheres Recapitulate Tumor Morphology and Phenotypic Heterogeneity

Introduction: Precision medicine - utilizing genetic screening to inform treatment decisions - has shown limited success for cancer patients. One reason for this poor response to targeted therapy is the variable and heterogenous tumor microenvironment (TME), which plays a significant role in cancer cell behavior and can lead to drug resistance. The TME is comprised of several components in addition to cancer cells, such as immune cells, fibroblasts, and extracellular matrix (ECM). The influence of the ECM within the TME has been well-documented; however, decoupling its effects on cancer cells using 3D in vitro models, such as spheroids and organoids, remains a significant challenge. In this study, we address this limitation with tumor “MatriSpheres,” a novel method of inducing spontaneous intercellular assembly of ECM stroma utilizing facile spheroid forming conditions. In other words, MatriSpheres are spheroids with intraspheroid ECM that may be used to decouple the role of ECM stroma in tumor function in vitro (e.g. traditional ECM-free spheroids compared to ECM-dense MatriSpheres). In this study, we (1) established MatriSpheres from colorectal cancer (CRC) cell lines with orthotopic decellularized small intestine ECM (SIS ECM), (2) characterized ECM composition and integration mechanism within MatriSpheres, and (3) used MatriSpheres to determine the role of ECM in cancer cell organization and transcriptomic regulation compared to in vivo CRC tumors.

Materials and

Methods: Porcine small intestine submucosa (SIS) was decellularized using physical and chemical treatments, and enzymatically digested to obtain a partially solubilized ECM suspension. Mass spectrometry and proteomics was performed to compare ECM biomaterial (SIS ECM, Commercial Collagen I, Matrigel) protein composition similarity to in vivo MC38 colorectal tumor ECM. 3D MatriSpheres were established by combining mouse (MC38 and CT26) and human (HT-29) CRC cells with sub-gelation concentrations of ECM in ultra-low attachment conditions. Cell viability assays (CellTiter-Glo 3D, Live-Dead) and histological staining (H&E, PicroSirius Red (PSR), Immunofluorescence) were performed to characterize MatriSpheres generated with different ECM biomaterials. Time lapse imaging was used to monitor spheroid formation kinetics and ECM biomaterials were evaluated using rheology and optical absorbance to determine gelation properties at varied ECM concentrations. MatriSpheres were benchmarked for similarity to in vivo tumors by comparing to single cell RNA Seq (in vivo MC38 tumors) to bulk RNA Seq (in vitro MC38 spheroids and MatriSpheres).

Results, Conclusions, and Discussions: Orthotopic SIS ECM shared greater proteomic similarity to in vivo tumors than traditional purified ECM preparations Collagen I and Matrigel. Out of 101 MC38 tumor matrisome proteins detected, SIS ECM shared the highest percentage of proteins (75%) with 42 proteins found exclusively in SIS. In contrast, Matrigel and Collagen I had a substantially lower number of similar proteins with 38% and 4%, respectively providing rationale for using a decellularized ECM for CRC tumor modeling (Figure 1a,b). 3D MatriSpheres incorporated ECM and established a complex and heterogeneous stroma, lacking with cells alone spheroids (Figure 1c). Further, ECM dense regions within SIS ECM MatriSpheres reduce Cadherin expression and indicates cell-ECM dominant adhesions (Figure 1d). ECM concentrations used for establishing MatriSpheres were at a sub-gelation threshold, which indicated that intercellular ECM biomaterial concentration and assembly within MatriSphere was cell-mediated (Figure 1e). MatriSphere phenotype strongly correlated with an in vivo subset of MC38 cells via “Scissor” bioinformatics analysis of single-cell and bulk RNA Seq found that MC38 MatriSpheres showed 35% overlap with in vivo MC38 tumor cells in comparison to 19% for cells alone spheroids (Figure 1f). In addition, MatriSpheres displayed a heterogeneous transcriptomic profile, whereas the traditional spheroids clustered predominantly in a specific cancer cell subpopulation. Ingenuity pathway analysis showed that CRC cancer cells have a robust (|z-score|> 3) and variable/cell-line dependent response to SIS ECM with MC38 cells primarily slowing metabolism pathways and CT26 and HT-29 activating IFN signaling pathways (Figure 1g).

We demonstrate a novel 3D in vitro tumor model with a robust ECM-rich stroma, or MatriSpheres. Intratumoral ECM is self-assembled and guided by cell-ECM interaction to create a unique heterogenous microtissue. Decellularized intestine ECM induces dramatic changes in CRC cell phenotypes, which shares transcriptomic expression patterns with in vivo CRC cells. MatriSpheres are a high throughput, rapid, and facile method of complex ECM assembly in vitro. Thus, this model may provide a high-fidelity tool for in vitro testing to improve predictiveness of drug efficacy.

Acknowledgements (Optional):