Ramirez O, García A, Rojas R, Couve A & S Härtel
Journal of Microscopy, Sep 1;239(3):173-83

The quantification of colocalizing signals in multi-channel fluorescence microscopy images depends on the reliable segmentation of corresponding regions of interest (ROIs), on the selection of appropriate colocalization coefficients, and on a robust statistical criterion to discriminate true from random colocalization. Here, we introduce a confined displacement algorithm (CDA) based on image correlation spectroscopy in combination with Manders Colocalization Coefficients M1ROI and M2ROI to quantify true and random colocalization of a given florescence pattern. We show that existing algorithms based on block scrambling exaggerate the randomization of fluorescent pattern with resulting inappropriately narrow probability density functions (PDFs) and false significance of true colocalization in terms of p values. We further confine our approach to sub-cellular compartments and show that true and random colocalization can be analyzed for model dendrites and for GABAB receptor subunits GABABR1/2 in cultured hippocampal neurons. Together, we demonstrate that the CDA detects true colocalization of specific fluorescence patterns down to sub-cellular levels.

Acknowledgments The authors thank Diego Díaz-Espinoza and Maria Osorio-Reich for the ImageJ plugin, N. Contreras from Area Kreativa for support with the figures and J. Cowan for insightful comments on the paper. Research in SCIAN-Lab (SH) is funded by FONDECYT 1090246, FONDEF D07I1019, and PBCT ACT47 (CONICYT, Chile). AG is funded by D07I1019. OR is funded by FONDECYT 1090246. AC is funded by FONDECYT 1071001. SH and AC are members of NEMO, ICM-P04-068-F and the Advanced Imaging & Bioinformatics Initiative AI�BI (www.aibi.cl).

Download PDF