Atomic force microscopy reveals structural variability amongst nuclear pore complexes

(A) AFM images and image analysis. The top row shows the NPC of interest (asterisk) near other NPCs with the usual pore radius of ∼44 nm. The second row shows the cropped NPCs from the smallest to largest. The third row shows height profiles of rotationally symmetrized pores (blue line). NPC radii were determined from the radial positions of the peaks (red crosses) in the height profiles, corresponding to the maximum heights of the NPC scaffolds. Minimum and maximum radii shown here are consistent with expectations for NPCs consisting of 7 (expected radius: [7/8] × 44 = 39 nm) and 9 subunits (expected radius: [9/8] × 44 = 50 nm), respectively. (B) Distribution of NPC radii (n = 583). The peak of the distribution lies between 42.5 and 45.0 nm, but some NPCs are very small with a radius of less than 39 nm and others are very large with a radius greater than 50 nm. Scale bars: 100 nm. Colour scales (A): 85 nm (top row, first image), 80 nm top row, other images), 70 nm (second row, left image), and 60 nm (second row, other images).

The top row shows the same NPC images as displayed in Fig 1B–G, built up from scan lines recorded from left to right (“trace”). The second row shows images based on the recording of the right-to-left line scans (“retrace”) during image acquisition, in near-perfect consistency with the trace images of the same pores. The third row shows the height profiles through the centre of each NPC (white dashed lines) from both the trace (blue) and retrace (red) images. The slight horizontal shift between them can be attributed to scanner hysteresis; because of this shift, no attempt was made to average trace and retrace images. The bottom row shows a schematic of the possible FG-Nup conformations (blue lines are FG-Nups at cytoplasmic periphery; red circle representative of a possible cargo molecule). Scale bar for all images (see F, retrace): 100 nm. Colour scale: 60 nm.

Two images of the same area of the NE, offset in time by 17 min. Scale bar: 200 nm. Colour scale: 70 nm.

(A–C) The top row reprints the image shown in Fig 1H, whereas the bottom shows the height profile of the dotted white line on the image above. All filament widths are recorded as ∼8–10 nm in diameter. These values are expected to be larger than the real filament widths because of tip convolution effects (in which a larger tip radius will result in broader features in the resultant image). Therefore, they are consistent with the expected ∼8 nm diameter of a nuclear actin filament (Turgay et al, 2017). Scale bars: 300 nm. Colour scales: 100 nm.

(A) The stiffness heatmap (top) and rotationally averaged effective Young's moduli (E_eff; bottom) determined from Force Volume mode data acquisition on cytoplasmic NPC surfaces (n = 56). Consistent with previous results (Bestembayeva et al, 2015), the heatmap shows an increased stiffness at the scaffold structure and at centre of the transport channel, as compared with the rest of the NPC structure and surrounding NE. The dashed black line is the point of initial contact between AFM tip and NE (hereby termed the true height), whereas the top of the grey fill is the rotationally averaged height of the NPCs as measured by the maximum indentation. The rotationally averaged E_eff (bottom) mirrors the pattern seen in the stiffness heatmap, with an increased elastic response seen in the central channel and at the cytoplasmic ring structure. Grey circles are the radially averaged E_eff values from individual NPCs, and the shaded blue area is the standard deviation. (B) Stiffness heatmap and E_eff of cytoplasmic NPC surfaces as measured using PeakForce QNM mode (PeakForce frequency: 2 kHz) give qualitatively similar results as Force Volume (A), with the largest stiffness and elastic responses recorded at the scaffold structure and in the centre of the NPC channel (n = 145). (C) Force Volume mode applied to the nucleoplasmic side of the NPCs. The stiffness and elastic response at the centre of the NPC diminishes (when compared with the cytoplasmic face of the NPC) as the AFM tip interacts with the moveable nuclear basket, as compared with the relatively stiff and elastic transport barrier (n = 49). (D) This is qualitatively reproduced using PeakForce QNM mode (2 kHz). The true height measurement displays a small peak in the centre of the NPC, as the nuclear basket protrudes from the scaffold structure (n = 19). Each experiment (A–D) was conducted on one NE. F_max (±10% confidence interval): 371 pN (A), 397 pN (B), 480 pN (C), and 300 pN (D).

(A) Two images from the same sample showing the cytoplasmic side of the NE (top row), with the effective elastic moduli (E_eff)—determined from Hertz fits to individual force curves captured during imaging (bottom row; force curve frequency: 2 kHz). (B) As (A), for the nucleoplasmic side of the membrane, highlighting the NPC scaffolds and the lamina network as local enhancements in height and E_eff, and NPC baskets as local increases in height and as reductions in E_eff. (C) Cropped pores from (A), highlighting the variability in transport channel E_eff values for NPCs with similar topographies. Force at maximum indentation (F_max, ±10% confidence interval): 397 pN (A) and 300 pN (B). Scale bar for all images (see A and C, top right panels): 100 nm. Colour scales: 70 nm and 6 MPa (A and C); 75 nm and 3 MPa (B).

Two images of the same 400-nm-wide area of cytoplasmic side of the NE, with concomitant E_eff heatmaps, recorded one after the other. The images show no significant differences in the height (top) and E_eff (bottom) maps. F_max = 397 pN ± 10%. Scale bar: 100 nm. Colour scale: 65 nm (top row) and 2.5 MPa (bottom row).

(A–D) Images from a video sequence (see Video 1) of the cytoplasmic side of the NE, in which, step-by-step, many of the proteins (and chemical energy) required for the classical import cycle of NLS proteins are added to the system. (A) Cytoplasmic side of NE. (B) After addition of the Ran mix and energy mix. (C) hsNTF2 (0.7 μM) is added to the sample. (D) hsImpβ (1 μM) is added and all NPCs fill with protein. (E–H) The rotationally averaged height profiles of the cross-correlation averaged NPCs from the images displayed in (A–D), respectively, showing a filling of the pore lumen and some increase in the pore rim height upon incubation with hsImpβ. (I) Nucleoplasmic side of the membrane before (top) and after (bottom) addition of hsImpβ (1 μM). (J, K) Cropped pores from the image sequence (A–D) showing changes as a function of time. Scale bars: 600 nm (A–D), 300 nm (I), and 100 nm (J and K). Colour scales: 150 nm (A–D) and 80 nm (I–K).

(A) Cytoplasmic side of the NE before (left) and after (right) addition of exogenous Impβ (imaged by Tapping Mode AFM). After incubation with Impβ (0.2 μM; 25 min), followed by extensive washing with buffer to leave ∼0.1 nM of exogenous protein in the buffer, all pores are seen to contain protein in their lumen. This is also seen in the phase (bottom row). (B) Same as for (A), except the Impβ concentration is increased to 2 μM (leaving a concentration of ∼1 nM after washing). Scale bar for all images: 300 nm (see height images after Impβ incubation). Colour scales: 200 nm (A, height images), 150 nm (B, height images), and -3°:3° (A and B, phase images).

(A) Stiffness heatmap of the cytoplasmic NPC surface before (left; n = 51) and 5 min after (right; n = 24) addition of hsNTF2 (3 μM). Before addition of hsNTF2, stiffness is seen at the cytoplasmic ring structure, and in the central transport barrier (see also Fig S5)—and, after addition of hsNTF2, the same pattern is observed. (B) Cytoplasmic NPC surface before (n = 66) and 5 min after (n = 67) addition of hsImpβ (1 μM), showing filling and homogenization of the central channel. (C) Cytoplasmic NPC surface before (n = 26) and 5 min after (n = 41) addition of hsImpα (1 μM) and hsImpβ (0.8 μM), showing a similar effect. (D) Nucleoplasmic NPC surface before (n = 34) and 5 min after (n = 34) addition of hsImpβ (0.9 μM). Without hsImpβ, the NPC shows a soft centre because of the presence of the flexible nuclear basket (see also Fig S5). hsImpβ increases the height profile in the centre of the NPCs and homogenizes the stiffness across the central channel. Each experiment (control and with NTRs) was conducted on one NE. F_max (±10% confidence interval): 350 pN (A), 283 pN (B), 300 pN (C), and 397 pN (D).

(A) In the absence of transport receptors, the FG-Nups can readily alternate between morphologies with enhanced density at the wall (left column) and at the centre (middle column) of the NPC channel. Added transport receptors lead to a homogenization of the FG-Nups across the pore lumen, provided that their binding avidity to the FG-Nups is strong enough (as for hsImpβ). Smaller transport receptors, such as NTF2, translocate the transport barrier very quickly without significantly rearranging the FG-Nups. (B) Images of NPCs consistent with the different proposed conformations. Scale bars: 100 nm. Colour scales: 70 nm (left and middle image) and 100 nm (right image).

Recombinant proteins were analyzed by SDS–PAGE and Coomassie staining.

(A) A PeakForce QNM (2 kHz) force curve recorded in the centre of an NPC (cytoplasmic side). (B) The force curve is segmented into many partitions (initially 20, shown by the dotted grey lines). The standard deviation of each segment is determined, and the smallest is taken as a measure of the baseline noise. This number is multiplied by a factor (3 in this study) and saved as a threshold value. The difference in the mean between neighbouring partitions is calculated and compared with this threshold: if the difference is less than the threshold, it is assigned a 0; if greater than the threshold, it is assigned a 1. If this produces a series of 0s and 1s with only one transition, the force curve is cropped from the end of the baseline to the first segment after the transition point (green dotted line is the transition point and blue line is the segment of force curve for cropping). If there is more than one transition, the number of segments is reduced by one, and the process restarts. The rationale behind this procedure is that it yields a force curve over an indentation that is so small that the increase can be approximated by a linear function. (C) A piecewise linear function is fitted to the cropped force curve using a least squares regression. (D) The knot in the piecewise function is defined as the contact point (z₀; red cross).