A systems biology approach to understanding Autosomal Dominant Polycystic Kidney Disease

The purpose of the thesis was to study the pathogenesis of Autosomal Dominant Polycystic Kidney Disease, the most common genetic disease affecting the kidney, using novel bioinformatics and computational approaches combined with experimentation. A new multi-stage framework for the analysis of time-series microarray data which identifies a set of possibly relevant genes and then builds a dynamic model for their regulatory network was produced during this work. The framework combines statistical filtering, Support Vector Machines, clustering and system identification in order to achieve these goals. As a practical application, it was employed to analyse two published microarray datasets derived from genetically modified Pkd1 mice. A defined set of genes was obtained from this analysis which provided good discrimination for the measurements coming from healthy and diseased animals. Also, it was noted that some genes previously linked to the disease and others related to cancer pathogenesis were identified. A potential model for their interactions was also derived. In the second part of the project, time-lapse microscopy combined with mathematical modelling was used to study human normal and disease kidney tubular cells in both low-density migration and wound closure assays. It was found that disease cells migrated more slowly than normal cells due to a reduction in their velocity and diffusion coefficient. Of interest, the somatostatin analogue, octreotide, partially restored cell migration in disease cells primarily by increasing cell velocity. Disease cells also showed a reduced capacity to close a wound in a monolayer and this was associated with randomisation of the directionality of movement. Using textural analysis, it was noted that cell tightness appears lower in disease cells during cell migration after wounding suggesting a reduction in cell-cell adhesion in these cells.