Statistical evaluation of colocalization patterns in immunogold labeling experiments

The ultrastructural localization of various antigens in a cell using antibo dies conjugated to gold particles is a powerful instrument in biological re search. However, statistical or stereological tools for testing the observe d patterns for significant clustering or colocalization are missing. The pa per presents a method for the quantitative analysis of single or multiple i mmunogold labeling patterns using interpoint distances and tests the method using experimental data, The clustering or colocalization of gold particle s was detected using various characteristics of the distribution of distanc es between them. Pair correlation and cross-correlation functions were used for exploratory analysis; second order reduced K (or cross-K) functions we re used for testing the statistical significance of observed events. Confid ence intervals of function values were estimated by Monte Carlo simulations of the Poisson process for independent particles, and results were visuali zed in histograms. Furthermore, a suitability of K functions modified by ce nsoring or weighting was tested. The reliability of the method was assessed by evaluating the labeling patterns of nascent DNA and several nuclear pro teins with known functions in replication foci of HeLa cells. The results d emonstrate that the method is a powerful tool in biological investigations for testing the statistical significance of observed clustering or colocali zation patterns in immunogold labeling experiments. (C) 2000 Academic Press .