Nuclear Pore

For IF staining of PBRM1, the fresh bull semen spread onto polylysine-coated slides was air dried, fixed using 4% paraformaldehyde for 5 min, washed three times in PBS, permeated in 0.2% Triton X-100 for 15 min, and blocked using 5% BSA for 1 h at room temperature. After washing with PBS three times, the samples were incubated with primary antibodies diluted with PBS (including anti-rabbit PBRM1 [1:100] (AB196022, Abcam, Cambridge, UK) or mAb 414 antibody (1:100, 902,097, BioLegend, San Diego, USA)) at 4°C overnight. Then, samples were incubated with secondary antibodies, including Alexa Flour® 488-conjugated goat anti-rabbit IgG (1:150; ZF-0511, ZSGB-BIO, Beijing, China) for 1.5 h at 37°C. For co-localization analysis of the nuclear pore complex (NPC) and PBRM1, the mAb 414 antibody was visualized by using a TRITC-conjugated secondary antibody, and PBRM1 labelling was visualized by using a FITC-conjugated secondary antibody. The samples were incubated in DAPI staining solution (C1005, Beyotime, China) for 10 min at room temperature and washed three times with PBS. The samples were added with an anti-fluorescence quencher and placed in a cover glass, and the edge was sealed with nail oil. Immunofluorescence staining was imaged at 200× magnification using an inverted fluorescent microscope (IX73, OLYMPUS, Japan).

Each cell line was manually annotated by a group of at least three experts using an ontology of 45 annotation terms for different organelles or cell structures, on the basis of consensus of the ≳250 cells imaged per cell line. This extended our previous description of the characteristic appearance of over 30 different organelles and structures⁸³ . This hierarchical ontology has specific (for example, ‘nucleoplasm’ or ‘axoneme’) and more general terms localizations (for example, ‘nucleus’ or ‘flagellum’). The more general term was used if a localization was ambiguous. If localizing to multiple organelles then all relevant terms were used (that is, additive annotation) (Supplementary Table 1).
We used a further ontology to describe any additional structure within each organelle (for example, ‘patchy’, ‘weak’ or ‘points’)⁸³ . In some cases, these were used for lower-confidence annotations (for example, nucleus (points) rather than nuclear pores). The ‘weak’ modifier was reserved for localizations with signal comparable to background auto-fluorescence (Supplementary Table 2).
Manual annotation of weak signals were supplemented by automated mNG fluorescence signal intensity, measured from all cells from all images of all cell lines. For reference^{4 (link)}, independent samples of the parental cell line were grown and prepared for microscopy identically to tagged cell lines. Individual cells were identified, oriented and cropped from the images automatically using intensity thresholding of the phase contrast image after a series of unsharp and background subtraction filters to generate cell masks, as previously described using ImageJ v1.52a (refs. ^{84 ,85 (link)}) and mean mNG signal intensity (sensitive to overall signal) and 99th percentile mNG signal intensity (sensitive to small bright structures) calculated. Auto-fluorescence tended to occur in the mitochondrion, cytoplasm and/or endocytic system. Therefore, any mitochondrion, cytoplasm or endocytic system annotation where both mean and 99th percentile green signal intensity were below the parental cell line were automatically given the ‘weak’ modifier. mNG images are displayed mapping black to the median signal outside of cells and mapping white to 4,500 or the maximum pixel value, whichever is higher.
Cell lines were non-clonal, as necessitated by the high throughput. From previously determined transfection efficiency^{76 (link)}, we estimate that populations were typically derived from 5–20 clones. In some cell lines this leads to heterogeneity, and in these cases organelle annotations were given a modifier of the approximate proportion of the population with the signal.
For all downstream analyses, a protein was listed as a component of an organelle if it was annotated as localizing to that organelle or cell structure, any substructure of that organelle, not annotated ‘weak’ and occurring in at least ~10% of the population. For some analyses, a simplified set of localizations are used. In these cases, proteins were listed as localizing to the nearest parent term in the simplified list.
This database of microscopy data and human annotations can be viewed and downloaded at http://tryptag.org with the annotations and an example image viewable and searchable at http://tritrypdb.org.