MC 6460
Candidate
Dorsa Mohammadrezaei | Applied Mathematics, University of Waterloo
Title
Integrating Experimental and Computational Approaches to Optimize 3D Bioprinting of Cancer Cells
Abstract
A key feature distinguishing 3D bioprinting from other 3D cell culture techniques is its precise control over created structures. This property allows for the high-resolution fabrication of biomimetic structures with controlled structural and mechanical properties such as porosity, permeability, and stiffness. However, for bioprinting to be successful, a comprehensive understanding of cell behavior is essential, yet challenging. This includes the survivability of cells throughout the printing process, their interactions with the printed structures, and their responses to environmental cues after printing. There are numerous variables in bioprinting that influence the cell behavior, so bioprinting quality during and after the procedure. Thus, to achieve desirable results, it is necessary to consider and optimize these influential variables. So far, these optimizations have been accomplished primarily through trial and error and replicating several experiments, a procedure that is not only time-consuming but also costly. This issue motivated the development of computational techniques in the bioprinting process to more precisely predict and elucidate cells’ function within 3D printed structures during and after printing.
During printing, we developed predictive machine learning models to determine the effect of different variables such as cell type, bioink formulation, printing settings parameters, and crosslinking condition on cell viability in extrusion-based bioprinting. To do this, we first created a dataset of these parameters for gelatin and alginate-based bioinks and the corresponding cell viability by integrating data obtained in our laboratory and those derived from the literature. Then, we developed regression and classification neural networks to predict cell viability based on these bioprinting variables. Compared to models that have been developed so far, the performance of our models was superior and showed great prediction results. The study further demonstrated that among the variables investigated in bioprinting, cell type, printing pressure, and crosslinker concentration, respectively, had the most significant impact on the survival of cells.
Additionally, we introduced a new optimization strategy that employs the Bayesian optimization model based on the developed regression neural network to determine the optimal combination of the selected bioprinting parameters for maximizing cell viability and eliminating trial-and-error experiments. In our study, this strategy enabled us to identify the optimal crosslinking parameters, within a specified range, including those not previously explored, resulting in optimum cell viability. Finally, we experimentally validated the optimization model's performance.
After printing, we developed a cellular automata model for the first time to predict and elucidate the post-printing cell behavior within the 3D bioprinted construct. To improve our model, we bioprinted a 3D construct using cell-laden hydrogel and evaluated cellular functions, including viability and proliferation in 11 days. The results showed that our model successfully simulated the 3D bioprinted structure and captured in-vitro observations. The proposed model is beneficial for demonstrating complex cellular systems, including cellular proliferation, movement, cell interactions with the environment (e.g., extracellular microenvironment and neighboring cells), and cell aggregation within the scaffold. We also demonstrated that this computational model could predict post-printing biological functions for different initial cell numbers in bioink and different bioink formulations with gelatin and alginate without replicating several in-vitro measurements.
Taken all together, this thesis introduces novel bioprinting process design strategies by presenting mathematical and computational frameworks for both during and after bioprinting. We believe such frameworks will substantially impact 3D bioprinting's future application and inspires researchers to further realize how computational methods might be utilized to advance in-vitro 3D bioprinting research.