Department of Mechanical Engineering 2018 Fall Seminar Series: Class 530.803
An unbiased classification approach reveals multiple fusion categories of VAMP2-mediated exocytosis
Presented by Professor Stephanie Gupton
Department of Cell Biology and Physiology, School of Medicine
University of North Carolina at Chapel Hill
Exocytosis is a fundamental process that secretes cargo into the extracellular space and potentially inserts lipids and proteins into the plasma membrane. A pH-sensitive variant of GFP (pHluorin) fused to the luminal end of a vesicle-SNARE protein, such as VAMP2, provides a fluorescent intensity readout of the fusion pore opening during exocytosis and subsequent fate of the v-SNARE. We previously reported an automated analysis platform that identifies such exocytic events, and records several parameters of their fusion, such as the frequency, half-life fluorescence decay, and spatio-temporal distribution. This led us to hypothesize there were discrete types of exocytic events based on the behavior of the v-SNARE after fusion pore opening. Here, we introduce a novel machine-learning method to automatically classify TIRF images of VAMP2-pHluorin exocytic events in developing murine cortical neurons. We used multiple classifiers, including hierarchical agglomerative clustering, dynamic time warping, and feature selection followed by dimensionality reduction to categorize exocytic events in an unsupervised way. Using a majority-rule committee of 28 indices, run with each of the classifiers individually, we determined the most likely number of classes. The committee always selected four discrete classes of fusion, with each classifier similarly grouping exocytic events. This was surprising, as classically, two modes of fusion are recognized. During full-vesicle-fusion (FVF), the fusion pore dilates after opening and the vesicle collapses into the membrane. During kiss-and-run (KNR) exocytosis a transient fusion pore closes after cargo secretion. All four classes were tetanus sensitive, indicating they were bona fide VAMP2-mediated fusion events, but exhibited unique fluorescence intensity profiles after fusion pore opening. To determine if any of these differences were pH sensitive, we buffered the media with HEPES. This increased the half-life of VAMP2-pHluorin fluorescence in two classes, indicating vesicle re-acidification, consistent with KNR fusion. One KNR class exhibited an immediate decay of fluorescence after fusion pore opening, whereas the other demonstrated a delay in the onset of decay, consistent with the fusion pore remaining open. The two classes that were HEPES-insensitive exhibited fluorescence decay consistent with VAMP2-pHluorin diffusing in the plasma membrane, consistent with FVF. One of these classes also exhibited immediate fluorescence decay, whereas the other class demonstrated a delay in the onset of fluorescence decay. Simultaneous imaging of VAMP2-phluorin with VAMP2-TagRFP to reveal the fate of the vesicle further supports the existence of the four classes. This validated novel unsupervised classification of exocytic modes now allows us to answer important previously unapproachable biological questions. We will provide examples of manipulations that alter the proportion of events and their distribution in neurons.