Location proteomics: building subcellular location trees from high-resolution 3D fluorescence microscope images of randomly tagged proteins

Abstract

The overall object of proteomics is to characterize all of the proteins expressed in a given cell type. With the rapid development of random gene tagging technology and high resolution fluorescence microscopy, it has become possible to generate libraries of digital images depicting the location patterns of most proteins in any given cell type. While the subcellular location of a protein is important to its function, no established methods exist for the systematic description, comparison or organization of protein location patterns. We have previously described classification methods that accurately recognize all major subcellular location patterns in both 2D and 3D images, as well as methods for rigorous statistical comparison of such patterns. We describe here the application of the numerical features from the previous work to images obtained by random tagging of proteins. Spinning disk confocal microscopy was used to collect images depicting the location patterns of 46 NIH 3T3 cell clones expressing proteins randomly tagged with a fluorescent protein. A set of 42 numerical features describing both image texture and object morphology were calculated and used to build subcellular location trees that group the tagged proteins by similarity of location pattern.