AbstractCollagen type I is widely used in tissue engineering scaffolds due to its role in structuring the extracellular matrix. Engineering scaffolds which closely mimic the heterogeneous and multiscale architecture of different biological tissues remains a key challenge. The overall goal is to combine bottom-up self assembly with top-down structuring to approximate the complexity found in nature. In this thesis, we focus on the fundamentals of bottom-up self-assembly, i.e. how to tune the micro-architecture of collagen networks by adjusting the physical environment during self-assembly. Although collagen assembly is known to be sensitive to the physical environment, the information on individual effects is scattered throughout the literature. A predictive model is lacking which relates the cumulative effects of assembly parameters to the collagen fibril and network morphology. To investigate the interactive effects of a range of experimental parameters on the kinetics and morphological result of collagen assembly, efficient experimentation and analysis are crucial. To this end, a high-throughput experimental and computational workflow was developed, integrating design of experiments, automated (big) data acquisition and analysis complemented by empirical modelling.
In this thesis, two types of experimental design for the multifactorial study of collagen assembly are explored: a two-level factorial design and a central composite design. Factorial designs are ideal for the efficient study of multiple factors and their interactions, but require linearity of the responses over the studied factor settings. Based on an extensive literature search, a factorial experimental design was implemented to describe the assembly behaviour of collagen at different temperatures, collagen concentrations, pH, sodium phosphate and sodium chloride concentrations. As the morphology of collagen networks did not follow a linear response pattern with respect to assembly parameters, a second central composite design was proposed linking pH, sodium phosphate concentration and sodium chloride concentration to fibril diameter and pore area. This design was augmented with additional data points compared to the factorial design to capture quadratic curvature in the responses.
A large amount of data was, if possible automatically, collected to analyse the system. We performed high-throughput turbidity measurements to monitor the kinetics of assembly, while the resulting collagen networks were characterised with automated scanning electron microscopy. Lag time and rate of assembly, as well as fibril diameter and pore area distributions were successfully extracted from thousands of data points and images by image analysis routines specifically developed for this purpose Based on these metrics, a linear empirical model was developed relating assembly temperature, collagen concentration and sodium chloride concentration to the rate of self-assembly. This model successfully predicted the lateral assembly rates of collagen gels and deepened our understanding of the self-assembly process. Due to time limits for this thesis and limitations to equipment availability, the formulation of quadratic models relating network topology to assembly parameters has not yet been achieved, but we expect to develop these upon fine tuning of the image acquisition and analysis workflow. The predictive models will enable the rational design of scaffolds with well-defined morphologies and properties. Based on these fundamentals, extensions to additive manufacturing can be envisioned to engineer multiscale collagen scaffolds with a complexity that closely resembles the ones found in nature.
|Date of Award||27 May 2020|
|Supervisor||M.P. (Paula) Vena (Supervisor 1), Heiner Friedrich (Supervisor 2), Remco Tuinier (Supervisor 2) & Sandra Hofmann (Supervisor 2)|