A designer FG-Nup that reconstitutes the selective transport barrier of the nuclear pore complex

Nuclear Pore Complexes (NPCs) regulate bidirectional transport between the nucleus and the cytoplasm. Intrinsically disordered FG-Nups line the NPC lumen and form a selective barrier, where transport of most proteins is inhibited whereas specific transporter proteins freely pass. The mechanism underlying selective transport through the NPC is still debated. Here, we reconstitute the selective behaviour of the NPC bottom-up by introducing a rationally designed artificial FG-Nup that mimics natural Nups. Using QCM-D, we measure selective binding of the artificial FG-Nup brushes to the transport receptor Kap95 over cytosolic proteins such as BSA. Solid-state nanopores with the artificial FG-Nups lining their inner walls support fast translocation of Kap95 while blocking BSA, thus demonstrating selectivity. Coarse-grained molecular dynamics simulations highlight the formation of a selective meshwork with densities comparable to native NPCs. Our findings show that simple design rules can recapitulate the selective behaviour of native FG-Nups and demonstrate that no specific spacer sequence nor a spatial segregation of different FG-motif types are needed to create selective NPCs.


Overview of the Supporting information:
Supplementary Note 1 -Amino acid sequences of all proteins used in this work (NupX, Nsp1, Nsp1-S, Kap95, BSA). Supplementary Table 1

Supplementary Figure 1 -NupX and Kap95 SDS-PAGE.
SDS-PAGE gel shows the band for the 32.5 kDa NupX molecule (left, thick bend running at ~37 kDa) and the 95kDa Kap95 molecule (right) after purification. This experiment was reproduced more than three times yielding similar results.

Supplementary Figure 2 -NupX coating of gold surfaces under different concentrations.
Real-time monitoring of NupX-coating of gold-coated quartz chips under concentrations of 100 nM (a), 1 μM (b), and 2 μM (c) at constant flow-rate (20 μL/min), using QCM-D. The slight increase in frequency at the end of the incubation represents the washing step, which induces a subsequent release of nonspecifically bound NupX molecules. The NupX solution included also 1 mM of TCEP in order to reduce the cysteines, which was present during the coating step.

Supplementary Figure 3 -SPR measurements of protein-and MUTEG-functionalized chips.
Angular reflectivity spectra of SPR measurements for dried samples in air. a Data for APTES (blue crosses), APTES + Sulfo-SMCC (red asterisks), and APTES + Sulfo-SMCC + NupX (black triangles) on silicondioxide-coated sensors. b Data for MUTEG (380 Da) on gold sensors grafted using 1 mM concentration. c Data for 1 µM Nsp1 (blue crosses), 1 µM Nsp1-S (red asterisks), 2 µM (black triangles) and 60 nM NupX (pink diamonds) on gold sensors. ΔθSPR denotes the angular shift in millidegrees of the resonance angle. Solid lines show Fresnel model fits (which are offset in the y-direction to match the measurement reflectivity minima), which are used to determine the thickness of each adlayer.

Supplementary Figure 4 -Passivation of NupX-covered Au-surfaces using MUTEG.
Real-time monitoring of 1 mM MUTEG binding to gold chips that were pre-functionalized with NupX at NupX concentrations of 100 nM (a), 1 μM (b), and 2 μM (c) at constant flowrate (20 μL/min), using QCM-D. Note that final dissipation shifts were all close to zero, while (negative) frequency shifts were found in the range of ~10-14 Hz, indicating the formation of a thin monolayer. The MUTEG solution included also 10 mM of TCEP in order to reduce the thiols, which was present during the coating step (same for Supplementary Fig. 5a).
Real-time monitoring of Kap95 interacting with a MUTEG-passivated gold surface. a, Frequency shift over time upon flushing of 1 mM of 1-mercapto-11-undecyltetra(ethyleneglycol) (MUTEG) onto a gold surface. From an independent measurement using SPR ( Supplementary Fig. 3b) where the same incubation time and concentration were used, the MUTEG coating of a bare gold surface yielded a dense monolayer with a 0.59 ± 0.01 nm average grafting distance. b, Frequency shift over time upon flushing of 730 nM of Kap95 to a MUTEG-passivated gold layer. Only minor interactions (< 2 Hz) were detected.

Supplementary Figure 6 -Kap95 dissociation from NupX by 0.2 M NaOH.
Real-time monitoring of Kap95 interacting with a NupX-coated surface with subsequent dissociation upon flushing of 0.2 M NaOH.

Supplementary Figure 8 -Testing the stability of NupX coatings against milliQ and 50% pure ethanol.
Real-time monitoring of milliQ and 50% pure ethanol rinsing of gold surfaces that were precoated with NupX and subsequently passivated with 1 mM MUTEG, at different NupX concentrations of 100 nM (a,d, respectively), 1 μM (b,e, respectively), and 2 μM (c,f, respectively), at constant flow-rate (20 μL/min), using QCM-D. Note that final frequency and dissipation shifts are all close to zero, indicating that our protein and MUTEG layers are stably bound to the gold surface.
Supplementary Figure 11 -Coarse-grained simulations of cargo adsorption in a NupX brush with a 5.7 nm grafting distance.
NupX brush system, simulated for a grafting distance of 5.7 nm. a,b Snapshots of the simulations (see also: Figure 2h). c, Top panel: Time and laterally averaged protein density distributions for the NupX brushes and for the two different types of FG motifs present inside the NupX brushes. The density profiles of the GLFG and FG motifs within the NupX brush are multiplied by 5 for clarity of display. Dark central lines and light shades indicate the mean and standard deviation in density profiles, respectively. These measures were obtained by averaging over the density profiles of trajectory windows 50 ns in length (N=60). A highdensity region (up to a maximum of 300 mg/mL) forms near the attachment sites of the NupX to the surface ( = 0 to 2 nm). Further away from the scaffold, the protein density remains at a constant value of ~ 170 mg/mL up to a distance of ~8 nm, after which it decays. FG and GLFG motifs predominantly localize near the transitioning point (8 nm). Bottom panel: Freeenergy profiles (PMF-curves) of the center of mass of the model Kap95 and inert particle along the z-coordinate, where = 0 coincides with the substrate. The PMF-curves originate from a z-value where the distance between the center of mass of either particle and the substrate approaches one half of the former's diameter. The difference in sign between the PMF-curves of both particles indicates a strong preferential absorption of the model Kap95 to NupX brushes and a repulsive interaction of the inert particle. Compared to the brush with a lower grafting distance (see Figs. 2h,i in the main text), the repulsion and adsorption are less strong, which is due to the decrease in the density of the NupX brush and FG/GLFG-motifs, respectively (see c top panel).

Supplementary Figure 12 -Model Kap95 particle.
The computational model of the Kap95 particle used in this work, rendered using VMD 2 . The Kap95 (and the inert particle) consists of sterically repulsive beads (i.e., only repulsive excluded volume interactions with its surroundings, here shown in dark grey) arranged in a geodesic shell. In the case of the Kap95 particle a strip of 10 hydrophobic binding sites (brownorange) is placed on the surface, and the net charge (-43e) is distributed equally over all the surface beads. Binding sites are modeled as Phenylalanine beads and are spaced 1.3 nm apart on an arc. Phe-beads are the most hydrophobic particle type in our 1-BPA model. The diameter of Kap95 is equal to two times the Stokes radius (see Supplementary Table 1).

Supplementary Figure 13 -Conductance decrease upon NupX-coating of solid-state nanopores.
NupX-coating of solid-state nanopores. a, I-V characteristics for a bare (red) and NupX-coated (blue) 30 nm pore. b, Ionic conductance of differently sized nanopores over a range of 10-60 nm is plotted vs diameter for bare (red) and coated (blue) pores.

Supplementary Figure 14 -Current Power Spectral Density before and after NupXcoating.
Power Spectral Density of the ionic current noise before (yellow curve) and after (purple) NupX-coating for a 30 nm pore. Sensitivity of the average pore and access region protein densities against a change in grafting distance dg. a, Upon increase resp. decrease of the grafting distance with approx. 10% (6.0/5.0 nm as compared to 5.5 nm used for SiN in the main text), we observe a corresponding decrease (-17%) and increase (+22%) in the maximal pore region density (light blue and dark blue curves, resp.). The maximal access region density (red) is only sensitive to a decrease in grafting distance, where a 10% decrease in grafting distance yields a large density increase of +60% due to NupX proteins being expelled from the pore region (light red). b, Axi-radial density distributions of 30 nm NupX-lined nanopores upon an approximate 10% decrease (left panel) or increase (right panel) in grafting distance. The high-density region towards the central axis of the nanopore increases in density and extends over a larger range in the z-direction when the grafting distance is decreased, whereas the opposite occurs for an increase in grafting distance. The overall morphology of the NupX-meshwork in the nanopore remains consistent under an increase of approximately 10% in grafting distance. Figure 19 -Density distribution of NupX in a 30 nm pore in the presence of Kap95 molecules.

Supplementary
Axi-radial density map of the protein density distribution in a 30 nm NupX-lined nanopore that interacts with Kap95 particles. The density distribution shifts towards the interface between the pore and access regions, rather than focus centrally in the pore, which is the case when no Kap95 particles are present (as shown in Figure 4b).