Table 1. Analysis of different approaches for signal co-localization.

Summary10 nuclei	No threshold			Thresholded
Coloc. Metric	PC	Mander's		PC	Mander's
		Green	Red		Green	Red
Raw files (no filtering)	−0.03646	0.18368	0.05448	−0.05216	0.09689	0.0401
Median Filter 3×3×3	−0.07823	0.11228	0.01549	−0.10285	0.04953	0.01616
Gaussian Filter 0.08 µm	−0.04365	0.5957	0.42345	−0.11055	0.1041	0.07763

A variety of automatic and manual protocols were tested to monitor levels of co-localization in samples generated throughout this study. Confocal series were collected (with sequential imaging in the labeling channels) and data files imported into image analysis software (Imaris suite). Pearson's and Mander's coefficients were used as indicators of the extent of co-localization between different channels (see text). Entire confocal series for 10 different nuclei (like those shown in Fig. 1C) were used to analyze apparent co-localization between the imaging channels using the different conditions identified in the Table, as discussed in the text. It is notably that the different conditions used have only a superficial impact of levels of co-localization, with very weak co-localization seen in all cases. Simple median filtering improves the quality of the images and decreases apparent co-localization relative to the unprocessed images. However, using Gaussian filtering as an alternative dramatically increases the apparent co-localization, by spreading the edges of the labeled structures (this is evident from differences in the respective Mander's coefficients). Thresholding after preliminary processing (filtering) eliminates low-intensity volexs but reduces levels of co-localization only slightly.