Efficient Tip Tracking And Analysis Of Eb1-Decorated Dynamic Microtubules From In Vitro Time-Lapse Images

A major hurdle in studying microtubule (MT) dynamics is the difficulty of tracking the MT, especially the MT tip. Time-lapse images (e.g., taken by total internal reflection fluorescence microscopy) often have uneven illumination and the pixel intensity varies along MTs, which make it difficult to track the MT contour. In addition, tip tracking is important in cases when MT-associated proteins that target the tip are involved. We introduce a novel multistep computational method to reliably track most of the MTs in the field of view. The algorithm employs a two-zone approach where a circular running window consisting of inner and outer zones is used to locally apply contrast-based principal component analysis. Locations where the aspect ratio between the major and minor principal values is greater than a threshold value are marked by ¯beads.' In this way, a bead is likely to be located on a filament, among which aligned beads that have similar directions of major axes are connected by ¯bonds.' We thereby generate beads-on-chain (BOC) models of MTs. Additional error correction is made by comparing spatial locations and orientations of the BOC models across different time frames. Since only BOC models instead of individual pixels need to be processed, our method is very fast. We demonstrate its utility by analyzing images of MTs decorated with the EB1 protein that track the MT plus end. This allows us to elucidate quantitatively the altered dynamics of the MT by EB1. We expect our method to be broadly applicable to tracking and analyzing dynamic MTs.