PIF - Progressive Idealization and Filtering
Automated Single Subunit Counting
PIF is software written in Matlab to automatically analyze single subunit counting data obtained with a TIRF microscope. For any details, please refer to McGuire et al. (2012) J. Biol Chem. 287 (43):35912-21.
Subunit composition of membrane protein assembly can be precisely evaluated using single-molecule fluorescence. The protein of interest fused to a fluorescent protein (e.g. GFP) is expressed in mammalian cells at sufficiently low level for single-molecule imaging. Fluorescence from these cells is recorded and the movies are automatically analyzed by the PIF algorithm. It discriminates individual spots between relevant proteins and contaminants and analyzes the associated fluorescence intensity trace to determine the number of step-wise photobleaching events, which occur upon photochemical destruction of the fluorescent protein. The automated analysis of many fluorescence traces lead to a photobleaching step frequency distribution (histogram), from which the protein composition can be readily determined.
Background
Communication between mammalian cells is mediated by a diverse array of plasma membrane bound proteins which are receptive to local changes in signaling molecules. This form of cellular communication is employed throughout human development; from the migration of cells in foetal tissue to the rapid electrical signaling in the adult brain. Interestingly, the role of these proteins is not fixed throughout the mammalian life cycle but is essentially dynamic. In many cases, adaptation is achieved by altering the subunit types which comprise an individual protein complex to fine-tune their responsiveness. Given this, a number of experimental methods have been developed to elucidate the precise subunit composition (or stoichiometry) of membrane proteins.
One of the most successful strategies in determining subunit composition has been the use of single-molecule fluorescence,
in particularly single subunit counting (Ulbrich
PIF (“Progressive Idealization and Filtering”) was designed as a fully-automated algorithm that determines the step number distributions, from which the subunit composition can be derived. Automation allows analysis of large datasets derived from mammalian cells and prevents any user-bias in selection or interpretation of the data. PIF first uses a set of selection rules to discriminate between relevant spots on the cell surface from “contaminant” spots. Fluorescence intensity traces from relevant spots are then subjected to a series of filtering steps to remove background fluorescence in preparation for step detection. The step detection algorithm is the most critical part of the process and functions by iteratively reducing the trace until only regions of constant intensity interrupted by the photobleaching steps remain (red traces). In this way, the step detection algorithm remains accurate even for traces with low signal to noise ratio, as is typical of fluorescence data obtained from mammalian cells. The accuracy of the step detection algorithm was initially evaluated using simulated traces purposefully designed to exhibit the characteristics observed in experimental data. Fluorophore blinking, variations in step amplitudes, noise and number of bleaching steps were all considered. PIF was able to detect each photobleaching step with an accuracy of more than 90%, and the distribution always indicated the correct step number (1-6 steps).