Poster T4 - Tissue-engineered colorectal cancer model for recapitulating tumor heterogeneity and advancing drug screening

Introduction: Colorectal cancer (CRC) ranks as the third leading cause of cancer-related deaths in the United States, necessitating models that accurately recapitulate the tumor microenvironment (TME) and yield translatable research outcomes in the drug discovery process. While patient-derived xenografts (PDXs) of CRC show potential in mimicking tumor heterogeneity and patient-specific characteristics, their in vivo application is limited by high costs and low throughput. This limitation drives the need for corresponding in vitro models that balance physiological relevance (modulation of tumor microenvironment attributes) and high-throughput aspects (scalability and uniformity). Our study aims to bridge this gap by evaluating whether tissue-engineered models using CRC PDX cells can maintain patient- and cancer stage-specific tumor heterogeneity, focusing on the development of a tissue-engineered colorectal cancer model for tumor heterogeneity. Simultaneously, our microfluidic cell encapsulation platform enables the rapid production of microspheroidal tissue-engineered CRC models, CRC VivoSpheres. This ensures a high level of reproducibility between replicates and the adaptability to be used in well-plate-based drug testing platforms.

Materials and

Methods: We utilized three PDX tumor lines from stage II, III-B, and IV CRC tumors to create 3D engineered CRC PDX (3D-eCRC-PDX) tissues. These tissues were formed by expanding CRC-PDX tumors in vivo, dissociating them, and encapsulating the cells in poly(ethylene glycol)-fibrinogen (PF) macroscale hydrogels. We maintained these cells in culture for 29 days, observing their viability and proliferation. To evaluate the preservation of tumor heterogeneity, flow cytometry, mechanical testing, and transcriptomic analysis were conducted. The cell subpopulation, tissue stiffness, and gene expression were compared between the 3D-eCRC-PDX tissues and their corresponding original CRC-PDX tumors. As control, dissociated PDX cells were cultured in 2D monolayers (2D-CRC-PDX).

CRC VivoSpheres were created by encapsulating CRC cells, including CRC PDX cells and HT-29 CRC cells, in PF hydrogel microspheres using our previously developed microfluidic platform. The uniformity of VivoSpheres were assessed using microscopy. In a preliminary drug testing, the HT-29 VivoSpheres were shipped to the oncology team at Southern Research for drug testing. The CRC VivoSpheres were treated with DMSO, GANT61, and SRI-38832, the latter two of which are GLI1 inhibitors. Phase contrast images and western blot were used to assess the response of CRC VivoSpheres to the treatments.

Results, Conclusions, and Discussions: In each stage of CRC PDX, our study successfully showed that the growth patterns, including changes in cell colony sizes and cell numbers within the 3D-eCRC-PDX tissues, accurately reflected those of the original CRC-PDX tumors. Additionally, we closely tracked the changes in cell subpopulations over time, comparing them to those observed in 2D-CRC-PDX cultures. Unlike 2D-CRC-PDX cultures, our 3D-eCRC-PDX tissues managed to preserve all cell subpopulations across different CRC stages. Notably, the time-dependent stiffness of the 3D-eCRC-PDX tissues mimicked both region-dependent and batch-dependent stiffness of the originating CRC-PDX tumors for a given CRC stage. In addition, the average stiffness of all batches of 3D-eCRC-PDX tissues mimicked the stage-dependent average stiffness of the CRC-PDX tumors. The transcriptomic analysis showed stage-specific similarities in cancer-related gene expression.

For the CRC VivoSpheres, they were rapidly produced with high cell densities (20×106 cells/ml) and high uniformity on day 0 with a coefficient of variation (COV) < 7%. This high uniformity was maintained for 15 days (COV ≤ 10%), which is critical for long-term dose studies. For the preliminary drug testing, the response of HT-29 VivoSpheres to GLI1 inhibitors paralleled that observed in in vivo tumor models, highlighting their potential in mimicking real tumor environments. Oncogenic GLI1 transcription activity and NBS1 overexpression have been found to contribute to chemotherapeutic resistance, negating the anti-tumor effects of 5-fluorouracil. While 2D cultured HT-29s responded to treatment with GANT61, HT-29 VivoSpheres continued to express NBS1 following GANT1 treatment, but downregulated NBS1 in response to the GLI1 inhibitor SRI-38832, which is the same response Southern Research has seen in in vivo tumor models.

This study demonstrates the ability to create in vitro CRC models that balance physiological relevance with scalability and uniformity. The 3D-eCRC-PDX models closely mimic the originating in vivo CRC-PDX tumors in terms of tumor heterogeneity and stage-specific characteristics. The unique microspheroidal geometry of the tissue-engineered CRC models makes a valuable tool for early-stage drug discovery and screening processes.

Acknowledgements (Optional): This work has been supported by NIH NCI grant number CA267170-0IAI (E.A.L., M.W.G) and NSF grant number 2141205 (Y.T., E.A.L.).