Cytomics and location proteomics: Automated interpretation of subcellular patterns in fluorescence microscope images

In this column, I briey reect on the manner in whichautomated analysis of the subcellular distribution of pro-teins (location proteomics) is relevant to the eld ofcytomics. There are many denitions of cytomics that varyslightly in emphasis. Fundamentally, however, it is the sys-tematic, comprehensive study of at least one cytome,where a cytome is the collection of cell states exhibitedby a tissue or organism. We dene a cell state as a uniquecombination of all observable cell behaviors or pheno-types. Different cell types represent different cell states,of course, but the same cell type can exist in more thanone state (e.g., activated and quiescent). Clearly, a cell’sstate is inuenced by and reected in the set of proteinsthat it expresses.However, simply knowing how much of a protein isexpressed is not sufcient to understanding its contribu-tion to the cell state. It is particularly important to alsoknow its subcellular location because changes in proteinsubcellular location can cause dramatic effects on cellbehavior. Perhaps the most thoroughly studied example ofthis phenomenon is the changes in protein location asso-ciated with apoptosis (1). Changes in location within acell type may also cause or result from disease, as illus-trated by the suspected involvement of the Wnt pathwayand b-catenin in a number of cancers (2).Based on the success of the various genome projects, thefeasibility and desirability of undertaking projects to study asingle aspect of gene or protein structure or function hasbecome accepted. Many such projects have been initiated,including projects to determine or predict all protein struc-tures and to measure gene and protein expression levels inmany cell types and under many conditions. However, sub-cellular location has received less attention than many otheraspects of gene and protein behavior. The major exceptionis in yeast, in which almost all proteins have been assignedto a set of major subcellular structures (3,4) using fusion ofcDNAs with the coding sequence of uorescent proteinssuch as the green uorescent protein. For example, Huhet al (4) used green uorescent protein tagging of cDNAsand visual examination to assign proteins to 12 categories:cellperiphery,bud,budneck,cytoskeleton,microtubule,cytoplasm, nucleus, mitochondrion, endoplasmic reticulum,vacuole, vacuolar membrane, and punctate. They then usedcolocalization with red uorescent protein markers todivide the cytoskeleton class into two classes, actin cytoske-leton and spindle pole, and to add nine new categories:nucleolus, nuclear periphery, golgi apparatus, three types oftransport vesicles, endosome, peroxisome, and lipid parti-cle. In all, 4,156 proteins were assigned to these 22 cate-gories in their study.Pilot projects in mammalian cells have also been de-scribed. For example, Simpson et al. (5) used cDNA tag-ging to localize approximately 100 proteins in a humancell line, and Jarvik et al. (6) used a clever genomic-taggingapproach (termed CD-tagging) to localize a similar num-ber of proteins in mouse 3T3 cells. As with the yeast stud-ies, analysis was restricted to assignment of proteins toone of a limited number of major locations.These results, although useful and illustrative, do notprovide location information with sufcient resolution tobe useful for understanding and modeling cell behavior.The limited resolution also applies to systems that havebeen designed for predicting subcellular location from pro-tein sequence. Further, there is an implicit assumption inmany prediction schemes or curated protein databases thatproteinshaveasinglelocationregardlessofcelltypeorcondition. In contrast, location is not necessarily the samebetween different cell types, as illustrated by the differ-