Supplementary Figure 3: Signal processing and CLING’s specificity. | Nature Structural & Molecular Biology

Supplementary Figure 3: Signal processing and CLING’s specificity.

From: Spatiotemporal allele organization by allele-specific CRISPR live-cell imaging (SNP-CLING)

Supplementary Figure 3

Signal processing of all acquired SNP-CLING images in (a) mESCs cells, in (b) mEFs, and CLING images in (c) RPE-1 cells. Arrowheads depict specific and non-specific CLING foci. The images collected on the LSM880+Airyscan have increased sensitivity (4-8 x) due to the 32 channel gallium arsenide phosphide photomultiplier tube (GaAsP-PMT) area detector that collects a pinhole-plane image at every scan position (Huff 2015). This increased sensitivity results in detection of both specific and background fluorochrome-derived signals in each image, which can be differentiated in post-acquisition image processing (described below and in methods). The number, sizes and brightness of non-specific background signals vary depending on the expression of the transfected fluorescent proteins and the imaging conditions (laser power, see Figure S2). The number of visible signals corresponding with the known number of alleles in a given cell line were obvious in the live view. For post-acquisition image analysis, z-stack planes were first merged by performing a maximum intensity projection (MIP). The ‘best fit’ analysis option was used to preliminarily adjust intensity thresholds. X-Y dimensions at this stage are shown in the top row. In sequential manual adjustment steps, signal from background haze, random fluorescent protein accumulation, and/or potential off-target dCas9 binding events were removed (second and third row). Signals present in final processed images (bottom row) correspond to expected karyotypes: (a) a single signal for the sole Firre locus and a single signal for the allele-specifically labeled Ypel4 locus were detected in male mESCs; (b) a single signal for the allele-specifically labeled Firre locus was detected in female mEFs; (c) two signals for either the CISTR-ACT or the FIRRE loci were detected in diploid RPE-1 cells (Darrow, Huntley et al. 2016). To investigate the difference in signal intensities between specific and non-specific signals in all cell types, we calculated the relative signal intensities by generating the median intensity of all detected signals in a given nucleus. The median intensity was used to normalize all detected signals. The intensities as ratios are represented as bar graphs below the corresponding image and show that the specific signals were the brightest. Of note, specific signal intensities were at least two-fold higher than non-specific signals. In validation experiments without the transfection of sgRNAs, no bright nuclear foci were detected (see Figure S2g). (d, e) We also measured the three-dimensional sizes (x, y, z dimensions) of all foci detected by the ‘best fit’ analysis. The signals that corresponded with processed specific signals were the biggest signals, as can be seen in represented image panel (d). We further quantified the difference in signal size with measurements from at least 30 nuclei (e). The specific ‘biggest’ signals were present in 12.7±1.98 z-stack planes (each 0.17 µm), in comparison to the smaller background signals that were detected in 5.3±1.5 planes. In all instances, the brightest and biggest signals correspond to the known number of alleles in RPE-1, mESCs and mEFs. (f) To further address CLING’s specificity, we quantified signals of CISTR-ACT, XIST, and SOX9 in human RPE-1 cells in the presence and absence of dCas9. As expected, dCas9’s presence dramatically increased the occurrence of specific signals. On average, 30 % of transfected cells showed the expected diploid status (two signals), whilst some cells showed only one signal, and ˜30 % of transfected cells showed either no or many signals. (g) In SNP-CLING, we determined a specific separation between the parental alleles in 83 % of the imaged nuclei in hybrid 129S1/CAST MEFs. The remaining 17 % of cells showed either no or multiple signals of the second allele, or cross-labeling of one of the two alleles. Of the 83 % of cells, 12 % showed two foci, indicating mitotic stage G2. (h) Haploid signals of sgRNAs targeting either the 129S1 (75 %) or CAST (90 %) allele determined SNP-CLING’s specificity in hybrid 129S1/CAST MEFs. (i) The co-localization frequency (78 %) of targeting CISTR-ACT with orthologous MS2 and PP7 sgRNA pools was highly specific in RPE-1 cells. (j) In ˜6 of 10 cells, a clear separation between XIST and TSIX was achieved (no co-localization) in RPE-1 cells. The heterochromatin formation of the inactive X chromosome and different chromatin compaction rates may also influence the spatial distances between two given loci.

Back to article page