Colocalization analysis between RNPs and fluorescent reporters

Analysis was carried out as described in Gáspár et al. (2017) and Heber et al. (2019). Briefly, images for colocalization analysis were deconvolved using Huygens Essential (https://svi.nl) and segmented using a custom ImageJ (Schneider et al., 2012) plug-in. Nearest-neighbor pairs between RNPs and fluorescent reporters were identified, and their position and signal intensities were extracted from the nurse cell and oocyte compartments, excluding any nuclear areas in the field of view. Quantification of mRNA copy number per RNP and normalization of fluorescent reporter signal was carried out by fitting multiple Gaussian functions to the corresponding signal intensity distributions taken from the nurse cells using the mixtools package in R (Benaglia et al., 2009; R Core Team, 2014). The µ value of Gaussian fit that described the largest portion of the distribution in the nurse cells (almost the lowest value of all fitted µ values) was taken as the signal intensity of one unit (for RNPs the intensity of a signal mRNA molecule). These unit values were used to normalize raw signal intensities. RNPs were clustered based on this normalized intensity under the following rule: (2ⁱ:2ⁱ⁺¹), i ∊ [0:8], i.e., 1, 2:3, 4:7, 8:15, etc. The observed nearest-neighbor colocalization frequencies were computed for each cluster and were compared to the expected colocalization frequencies (governed by the object-densities determined in randomized object localizations). The reported colocalization frequencies were corrected by subtracting the expected values from the observed values, which could result in negative frequencies, i.e., depletion of a molecule from oskar mRNPs. To estimate the 95% confidence intervals in each cluster, RNPs were randomly grouped together in groups of 50, then mean and 95% confidence interval of corrected colocalization values calculated. True difference from zero was established by one sample t tests (normality of the distribution was assumed due to random sampling). Similarly, the mean, normalized intensity of colocalizing fluorescent reporter molecules was calculated for each cluster. Correlation between RNA content and the normalized mean fluorescent reporter intensity was tested and compared using least-squares means analysis in R (Lenth, 2016). We typically analyzed 5,000–40,000 oskar RNPs (100–800 groups) and 2,000–4,000 stau RNPs (40–80 groups) imaged in 4–6 egg chambers collected from 3–4 females per condition.