Quantification of Endomembrane Phenotypes Using Chemical Genetics and Image Informatics
- Author(s): Ung, Nolan Michael
- Advisor(s): Raikhel, Natasha V
- et al.
Recent advances in computer vision and image analysis have enabled biological screens of large volumes. Such high-throughput assays increase the likelihood and scope of discovery ultimately leading to functional analysis of a gene or protein. To increase the efficiency of high-throughput screens, computational tools are essential expedite image analysis, models to make sense of the extracted data, and biological assays to characterize the mutation or small molecule. Chemical genomics, the use of small molecules to inactivate proteins, is an advantageous approach when studying a conserved process such as endomembrane trafficking. Endomembrane trafficking spans kingdoms and is important in defense, development, stress response and other vital physiological process. We have taken numerous approaches to study the quantitative behavior of the dynamic endomembrane system to accelerate discovery and better understand these complex phenomena. First, We identified six small molecules altering endomembrane trafficking in tobacco pollen; we characterized their effect on trafficking dynamics using video tracking facilitated by commercially available software giving us insight into intrinsic quantitative properties of the endomembrane system. Next, we wanted to create an automated tool to enable automatic phenotypic screening in Arabidopsis. EndoQuant is an automated computational tool for automatic sub cellular phenotypic analysis. Once this data was collected we developed a model to predict the biological being disrupted based on the cellular phenotype of a fluorescent marker. Using Gaussian Mixture Model, we were able to successfully predict that a subset of small molecules was disrupting endocytic recycling, leading to better experimental design and faster discovery of bioactive molecules. One particular biologically active molecule in Arabidopsis drastically reduced the root length of seedlings. This was found to be a result of cellulose deposition, most likely due to the miss localization of the cellulose synthesis machinery in response to disrupted trafficking membrane. We have analyzed real time membrane dynamics, automated phenotypic analysis, created a predictive model to link a phenotype with a biological process and characterized a small molecule disrupting the transport of a vital protein complex. These tools and methodologies will augment and accelerate the discovery process and our understanding of endomembrane trafficking in plant cells.