Location Proteomics, Tom Macura

This is an introduction to the biology and informatics techniques used in Location Proteomics.

---

Location Proteomics

The goal of 'location proteomics' is to describe the sub-cellular distribution for each protein in a given cell. It is hoped that knowing the dynamic properties of protein sub-cellular distribution under different environmental conditions will provide information about protein function.

Protein distributions are studied using fluorescence microscopy. An important property strongly correlated with image quality is the signal-to-noise ratio (SNR) of the fluorescence. SNR is constrained due to: (1) 'quantization errors' in the camera and (2) 'out-of-focus fluorescent light.'

Quantization errors are typically addressed by purchasing more expensive, more accurate detectors. One technique to eliminate out-of-focus light is 'de-convolution' which is computationally expensive and requires an accurate model of the point-spread function for a particular microscope. 'Confocal laser scanning microscopy' is an alternative method. Confocal microscopes collect fluorescence from individual small regions of the specimen that are illuminated by a laser scanning beam. Out-of-focus light is removed by placing a pinhole on the light collection path.

In preparation to fluorescence microscopy, the cell's proteins must be made to fluoresce. One approach, 'immunofluorescence', employs monoclonal antibodies that specifically bind to a target protein. This has several disadvantages, e.g. cells must be fixed and permeabilized (i.e. dead) before antibodies can enter. 'Gene tagging' can be used to introduce a DNA sequence encoding a fluorescence protein such as GFP into a target gene. In this way, the tagged cells and their progeny intrinsically fluoresce.

Bob Murphy, for a given cell type, attempts to use random gene tagging coupled with high-throughput fluorescence microscopy to generate images depicting the sub-cellular location patterns of most expressed proteins. In the past, analysis of distributions could only be done by visual examination. Bob Murphy attributes the primary reason for the absence of prior systematic, large-scale efforts to determine sub-cellular location for all proteins to the difficulty of automatically and quantitatively describing sub-cellular location in cells with varying sizes and shapes. These "limitations" can be overcome using automated pattern recognition methods.