Cohesiveness tunes assembly and morphology of FG nucleoporin domain meshworks - Implications for nuclear pore permeability

Abstract

Nuclear pore complexes control the exchange of macromolecules between the cytoplasm and the nucleus. A selective permeability barrier that arises from a supramolecular assembly of intrinsically unfolded nucleoporin domains rich in phenylalanine-glycine dipeptides (FG domains) fills the nuclear pore. There is increasing evidence that selective transport requires cohesive FG domain interactions. To understand the functional roles of cohesive interactions, we studied monolayers of end-grafted FG domains as a bottom-up nanoscale model system of the permeability barrier. Based on detailed physicochemical analysis of the model films and comparison of the data with polymer theory, we propose that cohesiveness is tuned to promote rapid assembly of the permeability barrier and to generate a stable and compact pore-filling meshwork with a small mesh size. Our results highlight the functional importance of weak interactions, typically a few kBT per chain, and contribute important information to understand the mechanism of size-selective transport.

Figure 1

Supramolecular assembly of FG domains, and predictions by polymer theory. (a, left) Schematic cross section of the yeast NPC perpendicular to its axis. FG domains are anchored to the NPC wall at high density. The flexible, intrinsically disordered chains explore the NPC channel, interpenetrate, and may form cross-links (red circles) to form the permeability barrier. (a, right) Monolayers of SLB-grafted FG domains are used as model systems of the permeability barrier, to study the impact of cohesive interactions on the organization and dynamics of FG domain assemblies at the nm scale. (b) Schematic phase diagram, summarizing simple theoretical predictions for films of flexible, regular, end-grafted polymers as a function of the Flory interaction parameter χ (which, as we propose, mirrors overall cohesiveness in our irregular FG domains) and grafting density. The brush and disrupted film phases are of particular interest for this study. Boundaries between phases are drawn qualitatively. The interpenetration of chains in the brush film (inset) gives rise to the correlation length ξ, a measure of the average mesh size. To see this figure in color, go online.

Figure 2

Kinetics of FG domain film assembly and maximal grafting density depend strongly on FG domain type. (a) Film formation was monitored by ellipsometry. At 0 min, 0.9 μM FG domains were added to SLBs containing 10 mol-% bis-NTA functionalized lipids. All FG domains were incubated under identical stirring conditions. Initial adsorption (inset) is similar for all FG domains and limited by mass transport (the gray shaded area represents theoretically estimated mass transport rates, see Supporting Material for details). At higher coverage, adsorption rates differ significantly between FG domains. (b) Schematic free energy profiles for the binding of FG domain molecules to an FG-domain-covered SLB. The barrier due to entropically unfavorable partitioning of new molecules into the existing FG domain film (black curve) is lowered through weak, attractive interchain interactions (purple dashed curve). To see this figure in color, go online.

Figure 3

The strength of inter-FG repeat interactions affects the thickness and concentration of FG domain films. (a) AFM indentation assays (schematically described in the inset) on FG domain films with a grafting density of 5.4, 5.1, and 4.8 pmol/cm² for Nup98-glyco, Nsp1-WT, and Nsp1-FILV→S, respectively, were carried out to estimate the film thickness. The thickness was determined by the distance between the onset of repulsive forces (arrowheads) and the hard-wall compression limit (d = 0). Control curves on SLB-covered silica before and after the indentation assays were taken to validate that the interaction with the probe remained short-ranged. (b) Film thicknesses determined by AFM indentations assays, and independently through viscoelastic modeling of QCM-D data (Fig. S5) and optical modeling of SE data on identically prepared FG domain films. The stronger the cohesive interactions the thinner and denser the film. To see this figure in color, go online.

Figure 4

Cohesive interactions lead to stiffer FG domain films. A parametric plot of ΔD/−Δf vs. −Δf, monitored by QCM-D during film formation, provides an estimate of the evolution of the elastic compliance (inverse of stiffness) of FG domain monolayers with coverage. Differences in the mechanical properties of the FG domain films can be clearly discriminated by this plot. To evaluate the contribution of phenylalanines to the stiffness of Nsp1-WT meshworks, we included additionally an Nsp1 construct in which exclusively phenylalanines were mutated to serines (Nsp1-F→S). Nsp1-F→S and Nsp1-FILV→S exhibited very similar curves, indicating that the F→S mutation is sufficient for the loss of cohesive interactions compared to Nsp1-WT. To see this figure in color, go online.

Figure 5

Grafting density and cohesive interactions influence the morphology of FG domain assemblies. Images by AFM of different FG repeat domain films. The left column shows low magnification images (2.5 × 1.25 μm²). The right column shows images at higher magnification (1 × 0.5 μm²), which were either obtained by digital zoom (from zones encased by orange solid lines in the left column) or by imaging at a higher resolution. Nup98-glyco films at a grafting density of (a) 12 pmol/cm² and (b) 9 pmol/cm² show a homogeneous surface with a distinct and stable small-scale morphology. (c) At 5.4 pmol/cm² shallow depressions of typically 100 nm width and a few nm in depth appear. (d) At 4.0 pmol/cm² the film becomes highly heterogeneous, with holes of several 100 nm in width that are likely to fully traverse the film. (e) Nsp1-WT films at 6.6 pmol/cm² appear homogeneous with some apparent roughness that could not be imaged stably. (f) Control image of a pure SLB. Color bar: false color coding of relative heights; scale bars: 200 nm; insets show height profiles of selected scan lines (white dashed lines).

Figure 6

Lateral mobility of FG domain films. Measurements by fluorescence recovery after photobleaching (FRAP). (a) Nup98-glyco, even at a relatively low surface coverage (5.4 pmol/cm²), did not show any significant recovery of the bleached central spot within 4 h. (b) In contrast, a slow yet significant recovery was observed for Nsp1-WT even at close-to-maximal surface density, demonstrating that the FG domains are laterally mobile within the film. (c) Also Nsp1-FILV→S showed recovery, which was quicker and almost complete within 30 min. Scale bar: 30 μm. To see this figure in color, go online.

Figure 7

Quantification of the energy for FG domain film compaction from binding rates. Activation energies E_a were determined from the FG domain binding rates dΓ/dt in Fig. 2 as a function of grafting density Γ through dΓ/dt = cA exp(E_a/k_BT), where c = 0.9 μM is the concentration of FG domains in the bulk solution. Following the theoretical considerations for the formation of brushes by Ligoure and Leibler (43), we set $A=6{\pi }^2/2^{1/3}\times k_{\text{B}}T{M}_{\text{w}}/\left({\eta }_0{N}_{\text{A}}{\rho }_{\text{P}}{a}^5{N}^2\right)\approx 4.13\text{mol}\text{m}{\text{s}}^{-1}{\text{g}}^{-1}\times M_{\text{w}}/N^2$. η₀ = 0.9 mPa s is the solution viscosity, N_A = 6.02 × 10⁻²³ mol⁻¹ is Avogadro’s number, ρ_P = 1.36 g/cm³ is the protein partial specific volume (64), a = 0.76 nm is the statistical (Kuhn) segment length (or twice the contour length of an amino acid), N is the number of segments (or half the number of amino acids) per FG domain chain, and M_w is the FG domain molecular weight. As long as binding is limited by the permeation of an incoming molecule through the FG domain film (solid lines), the activation energy is equivalent to the free energy increase associated with partitioning into and stretching of the chain in the film. In the limits of low and high coverage (dashed lines), mass transport limitations and saturation of the Ni²⁺-NTA binding sites on the SLB, respectively, impose an upper limit on the experimentally accessible activation energies. The differences between the activation energies for Nsp1-FILV→S on one hand, and Nsp1-WT or Nup98-glyco on the other, correspond to the energies of film compaction due to cohesive interactions. At 5 pmol/cm², it amounts to 3 k_BT for Nsp1-WT and $\ge$ 7 k_BT for Nup98-glyco. To see this figure in color, go online.

Figure 8

Impact of cohesive interactions on the supramolecular FG domain assembly in the NPC topology. Scheme of a cross section along the channel axes, based on predictions from polymer theory for flexible, grafted polymers. (a) If the interchain interactions are only weakly attractive or repulsive, an extended brush forms, and chains entangle into a meshwork with a characteristic mesh size. (b) With moderate inter-FG-domain interactions, the meshwork becomes denser; the characteristic mesh size and the size-exclusion limit decrease. (c) Strong cohesive interactions generate a discontinuous and leaky meshwork that fails to fill the entire channel. Tuning the cohesiveness of FG domains hence provides a robust strategy to optimize the size-selectivity of the nuclear pore permeability barrier. To see this figure in color, go online.