Fig. 1: Autonomous targeting of HIV-1 capsid species to NPCs.

a, Representative negative-stain electron micrographs of HIV-1 capsid preparations used in this study. b, HeLa cells were grown in multiwell slides and permeabilized with 30 µg ml⁻¹ of digitonin to perforate their plasma membranes^6,46, releasing soluble transport factors and allowing entry of the indicated fluorescent species: EGFP (3 µM), a monomeric CA–EGFP fusion (200 nM) or 40 nm capsid spheres (about 2.8 nM with 15% of the CA protomers being fused to a C-terminal GFP). After 30 min of incubation at room temperature, confocal laser scans were taken directly through the live samples. Note that the assembled capsid spheres bound very efficiently to NPCs and colocalized with the NPC marker (an Alexa647-labelled anti-Nup133 nanobody⁴⁷). The monomeric CA–EGFP fusion gave no discernible NPC signal. Scan settings were adjusted individually. c, Targeting of 40 nm capsid spheres to NPCs was performed as in b but higher resolution images were taken by Airy scans on a Zeiss LSM880 microscope. d, Experiments were performed as in b but included control incubations with 3 µM RanGTP which was locked in its GTP state by a Q69LΔC double mutation. RanGTP displaces cargo from importins, which normally happens inside nuclei. The mutant, however, triggers this dissociation prematurely in the cytoplasm. The RanGTP resistance of capsid binding to NPCs indicates importin-independent targeting. By contrast, the NPC targeting of the IBB–GFP·importin β complex (0.2 µM) was completely abolished by the Ran mutant. Scan settings were identical for corresponding ±RanGTP pairs. Experiments were repeated independently with identical outcomes (b, n = 7; c,d: n = 3). Scale bars, 100 nm (a), 10 μm (b,d), 1 μm (c).