Fig. 3: Import and export trajectories.

Imp α–JF549 trajectories were overlaid on a composite NPC image (from Fig. 1g,h). The blue curved lines approximate the nuclear envelope. See Supplementary Table 4 for data summary and Extended Data Figs. 7 and 8 for additional trajectories. a,b, Representative import track observed from the cytoplasm and the side (see Supplementary Video 1). c,d, Representative export track observed from the cytoplasm and the side (see Supplementary Video 2). e–h, Import trajectories (n = 13 for e,f). The polar plot in g (20 tracks, |z| ≤ 25 nm) reveals that the translocation paths are peripheral and largely confined to one of the eight lobes of the NPC scaffold, which is supported by the distribution of angular differences within the channel (h; Gaussian width = 19°). i–l, Export trajectories (n = 17 for i,j). The polar plot in k (20 tracks, |z| ≤ 25 nm) reveals that the export paths are also peripheral and largely confined to one of the eight lobes of the NPC scaffold, which is supported by the distribution of angular differences within the channel (l; Gaussian width = 17°). m, Abortive import trajectories (n = 32). n, Abortive export trajectories (n = 35). o,p, Localization error and diffusional behaviour from jump step and R²/t analysis of all 225 tracks that interacted with an NPC. The simplest simulation model (blue) assuming diffusive movement of a single species (D = 0.049 µm² s⁻¹; σ_x = 4.45 nm = 0.93σ_y = 1.55σ_z) does not simultaneously fit both representations of the experimental data (black) well. A three-species model (red; 7%, 56% and 37% of total, respectively) in which only species 3 undergoes diffusive movement (D = 0.055 µm² s⁻¹) fits the data much better (σ_x = 4.1 nm = 0.93σ_y = 1.55σ_z for species 1, and σ_x = 8.2 nm = 0.93σ_y = 1.55σ_z for species 2 and 3). Although species 1 and 2 do not have any prescribed diffusive movement, the large error for species 2 may subsume confined movements (see Extended Data Fig. 5j,k for additional models).

Source Data