Experimental and Computational Approaches to Optimizing Bovine Gamete Cryopreservation

Abstract

Cryopreservation uses freezing to suspend the metabolic activity of biological specimens for increased longevity of biotic materials like gametes. Cryopreservation has been known to affect both the functionality (performance of activities) and the viability (survival) of biological specimens during the freezing and thawing process due to four types of damage: (1) thermal, (2) ice, (3) osmotic stress, and (4) cytotoxic. Cryoprotective agents (CPAs) are known to reduce thermal and ice damage but cause osmotic stress and cytotoxic damage. Osmotic stress occurs when the addition of CPAs causes a rapid expulsion of water from the cell as the extracellular environment has become hypertonic. Cytotoxic damage occurs when a cell is exposed for too long to CPAs that may be damaging to the cell at high temperatures, but aid in preservation at low temperatures. The purpose of my project is to minimize osmotic stress in bovine embryos and cytotoxic damage in bovine sperm caused by CPAs using novel algorithmically guided techniques. To minimize osmotic stress in bovine embryos, I aim to facilitate the equilibration of embryos with cryoprotective agents isochorically (constant volume). Isochoric cryoprotectant equilibration, requires a feedback control system that in our case will use real-time image analysis developed in this thesis to estimate current embryo volume and then adjusts the concentration of CPAs being administered to the system. I implemented a colour-based image analysis software that was able to process images of bovine embryos as they were exposed to CPAs at a sub-second rate. The sub-second processing rates include cell volume estimates that are comparable to manual cell volume estimates. To minimize cytotoxic damage in bovine sperm, I optimized cryopreservation media (CPM) composition to maximize post-thaw motility. The composition of CPM can contain many ingredients that have the potential to interact and are infeasible to test only empirically. Here, I combined empirical experiments, data-driven optimization algorithms, and machine learning to optimize the composition of CPM. I used differential evolution and Gaussian process regression to optimize CPM composition that are on par with commercial media after 9 iterations. During the optimization process I determined that Gaussian process regression model was superior to artificial neural networks when predicting post-thaw motility for a given CPM composition. By optimizing these cryopreservation processes, cellular damage can be reduced, improving functionality and viability of gametes used in assisted reproductive technology that can be applied across animal husbandry and biomedical fields.