Nuclear pores as versatile reference standards for quantitative superresolution microscopy

Abstract

Quantitative fluorescence and superresolution microscopy are often limited by insufficient data quality or artifacts. In this context, it is essential to have biologically relevant control samples to benchmark and optimize the quality of microscopes, labels and imaging conditions. Here, we exploit the stereotypic arrangement of proteins in the nuclear pore complex as in situ reference structures to characterize the performance of a variety of microscopy modalities. We created four genome edited cell lines in which we endogenously labeled the nucleoporin Nup96 with mEGFP, SNAP-tag, HaloTag or the photoconvertible fluorescent protein mMaple. We demonstrate their use (1) as three-dimensional resolution standards for calibration and quality control, (2) to quantify absolute labeling efficiencies and (3) as precise reference standards for molecular counting. These cell lines will enable the broader community to assess the quality of their microscopes and labels, and to perform quantitative, absolute measurements.

Figure 1:. Nup96 cell lines.

(a) Representative confocal x-z and (b) x-y image of the Nup96-GFP cell line. Green: Nup96-GFP, magenta: membranes (DiD). (c) EM density of the nuclear pore complex with C-termini of Nup96 indicated in red. (d) Side view and (e) top view schematic. (f) Widefield, (g) confocal and (h) airy scan images of Nup96-GFP. (i) Raw STED image of Nup96-GFP labeled with an AberriorStar635P-coupled anti-GFP nanobody. Resolution estimates based on Fourier power spectra for f-i can be found in Supplementary Figure 3a. (j) Widefield expansion microscopy image of Nup96-GFP labeled with an Atto488-coupled anti-GFP nanobody. (k) As before, but imaged using structured illumination. Estimates of the expansion factor based on the analysis of the ring diameters can be found in Supplementary Figure 3c. (l) As before, but imaged using SRRF. (m) SMLM image of Nup96-mMaple, (n, o) SMLM of Nup96-SNAP labeled with BG-AF647 in GLOX/MEA. (p, q) Dual-color SMLM image of Nup96-SNAP labeled with BG-AF647 (red) and WGA-CF680 (cyan) in GLOX/MEA. (r, s) Corners of the NPC can be used as a resolution target in x,y (r) and z (s). Resolution estimates based on Fourier Ring Correlation for m-q can be found in Supplementary Figure 3b. Representative images of one (j-l), two (a,b,i), three (p-s), four (f-h,n,o) or six (m) independent experiments are shown. Scale bars 10 μm (b), 1 μm (f-n,p) and 100 nm (o,q,r,s).

Figure 2:. Nuclear pores as calibration reference standards. (a-h) experimental characterization of Nup-96 positions in the NPC.

(a) SMLM image of lower nuclear envelope, (b) circle fit of a single NPC, (c) histogram of fitted radii (R = 53.7 nm ± 2.1 nm, N = 3, n_C = 7, n_NPC = 2536) (d) Equatorial SMLM image of Nup96, (e) a single NPC in a side view. A fit with a double Gaussian returns the ring-distance d and the standard deviation of each ring. (f) Histogram of separation between rings (d = 49.3 ± 5.2 nm, N = 2, n_C = 14, n_NPC = 379). (g) 3D SMLM image of lower nuclear envelope. The localizations are color-coded according to their z-position. (h) x-z reconstructions with z-profiles as indicated. (i) NPCs as calibration reference standard for astigmatic 3D SMLM. Histogram of ring-distances before correction (magenta, d = 42.1 ± 1.1 nm, N = 1, n_C = 3, n_NPC = 1021) and after correcting for depth-induced calibration errors (green, d = 49.8 ± 1.9 nm). (j) Standard deviation of z-profiles from double Gaussian fit result in an upper bound for the experimental localization precision in z of 13.3 ± 1.0 nm (N = 1, n_C = 3, n_NPC = 1021). N denotes the number of biologically independent experiments, n_C the number of imaged cells and n_NPC the number of analyzed NPCs. All values depict weighted mean ± SD, based on n_NPC. Representative images of two (d,e) three (g) or four (a) independent experiments are shown. Scale bars 1 μm (a,d,g), 100 nm (b,e,h). All data on Nup96-SNAP-AF647 in GLOX/MEA.

Figure 3:. Effective labeling efficiencies. (a-d) Workflow.

(a) All NPCs in a cell are automatically segmented. (b) We fit a circle to the localizations and reject localizations outside a ring as background localizations. (c) We rotate the localizations to optimally fit an eightfold-symmetric template and count the number of slices that contain at least one localization. (d) We fit the histogram of the number of corners with a probabilistic model to directly obtain the absolute ELE. The statistical error is estimated by bootstrapping with 20 re-sampled data sets. (e-h) Gallery of NPCs. (e) Nup96-GFP labeled with an anti-GFP nanobody coupled to AF647. (f) Nup96-SNAP labeled with BG-AF647. (g) Nup96-Halo labeled with chloroalkane-AF647. (h) Nup96-mMaple. The numbers indicate the numbers of visible corners the algorithm detected. (i) Effective labeling efficiencies for various cell lines and ligands. Bars denote the mean, error bars the standard deviation and individual data points measurements of a single cell. These data are derived from N biologically independent experiments, n_C imaged cells and n_NPC analyzed NPCs: GFP-NB-Q-AF647: N = 2, n_C = 6, n_NPC = 2913; GFP-NB-Q-CF680: N = 2, n_C = 5, n_NPC = 1805; GFP-NB-X4-AF647: N = 2, n_C = 9, n_NPC = 4303; GFP-NB-X4-CF680: N = 2, n_C = 6, n_NPC = 2011; GFP-NB-S-AF647: N = 2, n_C = 4, n_NPC = 8768; GFP-NB-S-AF647 (2y): N = 2, n_C = 3, n_NPC = 1000; GFP-Antibody: N = 3, n_C = 14, n_NPC = 7380; SNAP-AF647: N = 4, n_C = 11, n_NPC = 5372; Halo-Cy5: N = 5, n_C = 14, n_NPC = 5967; Halo-O2-AF647: N = 2, n_C = 5, n_NPC = 1393; Halo-O4-AF647: N = 2, n_C = 6, n_NPC = 3395; Halo-PAJF549: N = 3, n_C = 17, n_NPC = 4066; mMaple: N = 6, n_C = 16, n_NPC = 8146; mMaple live: N = 3, n_C = 6, n_NPC = 1343; Example images for all labels can be found in Supplementary Figure 9, and imaging conditions are listed in Tables 4 and 5 (Methods). Representative images of two (e,g), four (a,f) or six (h) independent experiments are shown. Scale bars 1 μm (a) and 100 nm (e-h). *labeled in live cells, imaged after fixation. **measured on Nup107-GFP.

Figure 4:. Counting of protein copy numbers in complexes. (a-d) Counting in diffraction limited microscopy.

(a) Confocal image of the reference protein Nup96-GFP with the majority of nuclear pores resolved. (b) Confocal image of the target protein Nup107-GFP imaged with the same microscope settings. (c) Histograms of intensities of local maxima (see Methods) for the reference and target structures together with Gaussian fit to determine the mean intensity values. (d) Mean intensity values for several reference and target cells. These values show a small variation and are similar for reference (〈I_ref〉 = 1552 ± 55 ADU, N = 1, n_C = 8, n_R = 10104) and target complex (〈I_tar〉 = 1603 ± 77 ADU, N = 1, n_C = 6, n_T = 7178). (e-h) Counting with SMLM. (e) Reconstructed superresolution image for reference cell line Nup96-mMaple and (f) for target cell line Nup107-mMaple. NPC structures are automatically segmented to determine the numbers of localizations per NPC. (g) Histogram of number of localizations per NPC for reference and target. The number of Nup107-mMaple proteins per NPC is calculated from the average relative number of localizations. (h) The stoichiometry of Nup107 in the NPC (N_Nup107 = 32.1 ± 2.5, N = 5, n_C = 13, n_T = 1928) shows a high accuracy and low statistical errors of this counting approach. (i-m) Counting in yeast. (i) Mixture of Nup188-mMaple+Abp1-GFP reference cell lines with Nup82-mMaple+Nup188-mMaple target cell lines, which can be distinguished by the GFP signal. (j) Superresolution reconstruction and (k) individual nuclear pores. (l) Histograms of the number of localizations per nuclear pore, arrows indicate the mean (N = 2, n_C = 508, n_NPC = 1190 for Nup188 and n_NPC = 1176 for Nup82+Nup188). (m) Copy number of several yeast nucleoporins per NPC, determined using Nup188 as a reference. These data are derived from: Nup82: N = 2, n_C = 242, n_T = 678, n_R = 686; Nup82+Nup188: N = 2, n_C = 508, n_T = 1176, n_R = 1190; Nup192: N = 2, n_C = 558, n_T = 992, n_R = 916; Nic96C: N = 2, n_C = 304, n_T = 1102, n_R = 1127; Nic96N: N = 2, n_C = 532, n_T = 1078, n_R = 1079; Nic96N+Nup49GFP: N = 2, n_C = 303, n_T = 1137, n_R = 1149; Nup188 (CHX treatment): N = 2, n_C = 521, n_T = 1157, n_R = 1154. N denotes the number of biologically independent experiments, n_C the number of analyzed cells, and n_T/n_R the number of analyzed NPCs for the counting target/reference. Bars denote the mean, error bars the standard deviation and data points individual acquisitions. Shown values depict weighted mean ± SD, based on n_NPC. Representative images of one (b), two (a,i-k), five (f) or six (e) independent experiments are shown. Scale bars 10 μm (i), 1 μm (a,b,e,f,j), 100 nm (k).