Mammalian embryo quality is often assessed by morphology, the timing of mitotic events, and aneuploidy. These parameters are more commonly investigated in non-equid mammalian species, therefore, the parameters which indicate embryo quality in the horse remains to be fully elucidated. The overall objective of this dissertation was to optimize the in vitro production (IVP) of equine embryos by understanding the consequences of errors that can arise during pre-mitotic and post-mitotic events. This was accomplished in 3 chapters. Chapter 1 is an introduction to the dissertation. Chapter 2: Review of relevant literature provides a review of the literature pertaining to early mammalian embryo development, knowledge gaps, and implications for equine embryo quality assessment. The specific aims of the experiments conducted in Chapter 3: Pronuclear formation and cytoplasmic extrusion and Chapter 4: Aneuploidy in the early cleavage stage equine embryo were to determine the pre- and post-mitotic events in the zygote using fluorescence microscopy, and the exact percentage of aneuploid embryos in the cleavage stage equine embryo. Currently, the general timing of pronuclear (PN) apposition and fusion is unknown, and this information is critical to bridge the gap in understanding the transition from the PN phase to first cleavage. We found that in vitro PN formation is delayed by more than 6 hours compared to previous observations of PN in the in vivo equine model. The sperm head is still intact at 12 hours post intracytoplasmic sperm injection (ICSI). Additionally, cytoplasmic extrusion (CE) is a pre-cleavage event unique to the horse, and previous studies have shown that CE is a critical event in equine embryo development and occurs as it is associated with developmental success. While CE has also been described as a fragmentation event, there was little previous evidence that the extruded material contains DNA. Using fluorescence microscopy, we confirmed that no DNA was present in the extruded CE material and at this moment, the zygote was at the metaphase of mitosis. In Chapter 4, euploid and aneuploid blastomeres were determined by single-cell sequencing for the first time in the horse. To accomplish this, embryos were removed from culture between days 2 and 3 of embryo culture following ICSI. Because embryonic endpoints were unknown, a prediction model using machine learning using time-lapse data from Chapter 3 was used to determine if the embryos would have failed or survived to blastocyst stage and whether the early developmental stages were predictive of aneuploidy. According to the results from the sequencing data and prediction model, the blastocyst prediction model could not distinguish embryos with a higher percentage of euploid blastomeres in the horse. The percentage of aneuploid embryos was alarmingly high (81.25%), which could be explained by the frequent multipolar divisions observed during the early cleavage divisions. The findings from this work inform future research efforts to improve IVP practices in the horse with potential applications for improving embryo outcome predictions, selection methods as well as pregnancy outcomes.

Batteries are a building block of modern technology. They are especially key in our transition away from fossil fuels because they present possible solutions to two key hurdles: (1) transportation electrification and (2) renewable energy storage. The overarching topic of my graduate studies has been control and estimation problems relating to batteries. There are a wealth of problems in this space that are growing in importance as the demand for batteries increases. In this dissertation, I will focus on my work pertaining to battery pack thermal management.

A key challenge in the optimal use of batteries is that their performance is temperature dependent. A large body of work has quantified the effect that temperature has on Lithium ion batteries. Further work has looked at using this information to better model and control single battery cells accounting for temperature. From here, an interesting direction is to address the emerging problems that arise when we assemble many battery cells into battery packs. One particular challenge is that battery cells in packs perform and age differently due to cell-to-cell variations in temperature across the pack. This variation can be attributed to variations in cell heat generation, such as differences in current and electrical connectivity, as well as differences in battery thermal management system (BTMS) designs.

To effectively design thermal management systems for battery packs, it is essential that we use models to gain fast, low-cost insights on our designs. Modeling of thermal management systems in the literature is often done with computational fluid dynamic (CFD) models that simulate Naiver-Stokes and heat transfer partial differential equations (PDEs) simultaneously to gain high fidelity understanding of battery pack temperature distribution. Chapter 2 proposes a lower order 1D PDE model for battery packs that are fluid cooled. The proposed model is a system of two PDEs where one represents diffusion of heat in the pack, one represents advection of heat in the fluid cooling medium, and they are coupled via a first order heat transfer term. The proposed model allows for faster simulation while also allowing for interesting theoretical analysis of these systems. Additionally, Chapter 2 presents a numerical scheme for simulating this PDE that is demonstrated to be conservative in physical quantities.

The second technical section of this work addresses the problem of temperature non-uniformity in battery packs. To solve this problem, many previous works have used CFD models to design and test proposed thermal management designs. The common finding in the literature is that for better uniformity you must encourage more heat transfer at the part of the pack that gets the hottest. However, due to the complexity of the CFD models that are used, it is hard to use these models to derive generalized results. Chapter 3 takes the reduced order model presented in Chapter 2 and uses it to analytically derive an expression for the optimal thermal resistance that drives temperature to be uniform throughout the pack. This design is shown to homogenize both the pack temperature and cell aging. Further, its performance is compared to other designs via a simulation study.

Retroviral replicating vectors (RRV) are capable of tumor-selective replication and stable integration into the cancer cell genome, and are currently under investigation for prodrug activator gene therapy of recurrent high grade glioma in phase II/III clinical trials. In preclinical studies evaluating RRV gene therapy in intracranial glioma models, we showed effective long-term tumor control in human glioma xenograft models as long as prodrug administration was continued. In contrast, tumor eradication was achieved even after cessation of prodrug administration in immunocompetent syngeneic glioma models. Accordingly, my research has focused on investigating the immunological mechanisms involved in RRV gene therapy, and evaluating combined viro-immunotherapy approaches to improve RRV-mediated tumor transduction and therapeutic efficacy, as well as to confer a dominant foreign antigen to be targeted by adoptively transferred T cells.

RRV replication may be delayed at lower inoculation doses, if initial distribution is confined to the area surrounding an intratumoral injection site. More rapid tumor transduction may be achieved if the inoculum is distributed more evenly within the tumor. Alloreactive cytotoxic T lymphocytes (alloCTL) are migratory cells trained to react to major histocompatibility complex (MHC) that are selectively displayed on gliomas. Combining alloCTL and RRV gene therapy as individual therapies showed added benefit compared to either modality alone. We further tested the use of alloCTL as vector producer cells to deliver marker and suicide genes to tumors after their transduction with RRV (alloCTL/RRV). When compared against either therapy alone in a subcutaneous glioma model, the greatest benefit was achieved by alloCTL/RRV. This suggests that alloCTL/RRV can serve as anti-tumor effector cells targeting endogenous antigens, as well as motile ‘carriers’ that facilitate spread of RRV to tumor cells.

Aside from MHC, there are few well-characterized / immunogenic tumor-associated antigens with defined epitopes in glioblastoma. The efficient spread and permanent genomic integration of RRV can be utilized to induce stable expression of viral antigens that may represent useful immunotherapeutic targets. We developed an RRV displaying the lymphocytic choriomeningitis virus glycoprotein gp33-43 (gp33) epitope fused to the viral env gene. Target cell-specific cytotoxicity by transgenic lymphocytes engineered with a T cell receptor recognizing the gp33 epitope (P14 cells) was observed in vitro and in vivo. P14 adoptive transfer in immunocompetent hosts increased survival. Thus, the use of RRV to deliver highly immunogenic viral antigens to glioma could enable more effective immunotherapy for this disease.