Computational protocol: Spatiotemporal dynamics of HSV genome nuclear entry and compaction state transitions using bioorthogonal chemistry and super-resolution microscopy

Similar protocols

Protocol publication

[…] Super-resolution imaging was performed on Elyra PS1 system (Carl Zeiss) with an Apochromat 63x 1.4 NA oil objective lens, 488nm and 561nm excitation lasers and images were captured on a sCMOS PCO Edge camera. The camera pixel size is 6.5 μm and with 63x objective and additional 1.6x tube lens, this corresponds to 64 nm in the object plane. For analysis of infected cells, image stacks (2 μm) were acquired in Frame Fast mode (single multiband cube) with a z-step of 110 nm and 15 raw images (five phases, three angles) per plane. Raw data was then computationally reconstructed using the ZEN software to obtain a super-resolution 3D image stack with a pixel size of 32 nm in xy and 105 nm in z. The SIMCheck ImageJ/Fiji plugin [] was used to perform quality control on both raw and reconstructed data and to estimate lateral (x-y) resolution (approximately 120 nm) and axial (z) resolution (approximately 300 nm). Images from the different colour channels were registered in ZEN with alignment parameters obtained from calibration measurements with either virus capsids simultaneously labelled in both red and green channels or with TetraSpeck calibration beads 0.1 μm diameter (Thermofisher). 2D Gaussian fitting was done using the PALM analysis function in Zen with 30 pixel image window or ‘GaussFit on spot’ plugin in ImageJ. The Gaussian 1/ sqrt (e) radii were converted to full width at half maximum (FWHM) values by multiplying with 2x sqrt (2*ln(2)).For analysis of HSVEdC capsid and genome dimensions by immunofluorescence and cycloaddition reactions, the expected dimensions can be estimated by a convolution of the SIM resolution (120 nm and 138 nm for 488 nm and 561 nm excitations respectively) with the known sizes of capsid diameter (125 nm) and genome space (100 nm) []. In case of the capsid by immunofluorescence, based on previous analyses [] we estimated and additional 35 nm for the primary/secondary antibody bringing the estimated dimensions of the capsid to 160 nm. The convolutions of the SIM resolution and these sizes results in an estimated size of 200 nm and 170 nm for the capsid and genome respectively. Our measured average size is about 26% smaller than the estimate size. This was in line with measurements of calibration measurements with standard fluorescent beads of different sizes where the FWHM sizes were found to be 22% smaller than expected.Volume analysis was performed the object analyser module of the Huygens image processing suite (SVI, The Netherlands). The image is segmented into defined objects by the seed-threshold level adjustment, and connection process. Introduction of a watershed increases segmentation reliability further. Detected objects are automatically labelled and submitted to a continuous Iso Surface renderer. The segmented image is shown as a coloured iso-surface image. Object statistics are reported for each object, including geometrical data and spatial location. A simulated fluorescence process (SFP) computing algorithm allows visualization of the 3D data and production of the rendered image as an animation. Using this method we estimated volume and sphericity for genomic foci from visions on coverslips versus after entry to the nucleus. [...] Spatial clustering analysis of EdC labelled genomes was carried out using the Spatial Statistics 2D/3D ImageJ plugin []. The plugin analyses the overall distribution of inter-point distances including any local clusters and calculates whether there is evidence for a non-random distribution in the population of cells. A binary mask of each nucleus is generated together with a mask of the genomes using the ‘Find Maxima’ function. The plugin calculates for every nucleus, the distances between every point and its nearest neighbour and generates a cumulative distribution function (CDF) of those distances (the G-function). To compute this function, first the average CDF of a completely random distribution is estimated over a set (500 iterations) of randomly generated point patterns, specific to each reference structure (nucleus) and the number of points (genomes) in that structure. Second, the expected variation of CDFs around their average is estimated using a second set of randomly generated binomial point patterns. The relative position of the observed CDF for the actual test set within this range of variation is used to assign a p-value to the observed pattern, termed the ‘Spatial Distribution Index’ (SDI). Point patterns that tend to clustering have an SDI closer to 0 while patterns tending to even spacing have an SDI close to 1. The CDFs of SDIs of two different populations are compared using the Kolmogorov-Smirnoff (KS) test, which is non-parametric and distribution free. A p-value for the difference between the two populations is calculated, as well as the D statistic which is the largest deviation between the two CDFs.An independent clustering analysis was performed by calculating the Ripley function (K) using the BioImage Analysis platform ICY (http://icy.bioimageanalysis.org) as described []. Nuclei were segmented using the DAPI signal as above and a binary mask created. The ICY Spot Detector plugin identifies the EdC genomes contained within the nuclear mask. These were used to calculate the K function using the Spatial Analysis plugin []. In this approach regions of interest (circles) with increasing radii are drawn around every detected spot and other spots located within the circles are identified in the overall search area (the nucleus). The K function is then based on the number of spots that are closer than the radius, calculated for each increasing radius. The function is used to report the statistical significance of whether a distribution of points is random or clustered by comparing obtained values with critical quantiles under a completely random distribution. The amplitude of the K-function can then be compared to corresponding low and high quantiles (0.01 and 0.99 here). When K is higher than the high quantile for a certain radius, the foci are significantly organised in clusters. Conversely, when K is lower than the low quantile, the foci are dispersed. […]

Pipeline specifications