Plasticity of an ultrafast interaction between nucleoporins and nuclear transport receptors

Abstract

The mechanisms by which intrinsically disordered proteins engage in rapid and highly selective binding is a subject of considerable interest and represents a central paradigm to nuclear pore complex (NPC) function, where nuclear transport receptors (NTRs) move through the NPC by binding disordered phenylalanine-glycine-rich nucleoporins (FG-Nups). Combining single-molecule fluorescence, molecular simulations, and nuclear magnetic resonance, we show that a rapidly fluctuating FG-Nup populates an ensemble of conformations that are prone to bind NTRs with near diffusion-limited on rates, as shown by stopped-flow kinetic measurements. This is achieved using multiple, minimalistic, low-affinity binding motifs that are in rapid exchange when engaging with the NTR, allowing the FG-Nup to maintain an unexpectedly high plasticity in its bound state. We propose that these exceptional physical characteristics enable a rapid and specific transport mechanism in the physiological context, a notion supported by single molecule in-cell assays on intact NPCs.

Graphical abstract

Figure 1

Conformation of Nup153FG^PxFG (A) Scheme of Nup153FG constructs. (B) Residual dipolar couplings (RDCs) of Nup153FG^PxFG aligned in phages. Experimentally obtained RDCs (gray bars) were compared with RDCs calculated from the ASTEROIDS ensemble obtained on the basis of experimental chemical shifts (red line). Dashed lines represent positions of FG-repeats and F1374. Color code as in (A). (C) The same conformational ensemble was used to calculate a small angle X-ray scattering (SAXS) curve using CRYSOL (red line). The back calculated scattering curve is in good agreement with measured SAXS data under similar experimental conditions (black dots) (Mercadante et al., 2015). (D) Distribution of the radius of gyration (R_G) from five equivalent ASTEROIDS selections. The three conformations displayed on top represent the most compact, the least compact, and one of the most prevalent conformations in the ensemble.

Figure 2

Nup153FG^PxFG⋅Importinβ Interaction Analyzed by smFRET (A) FRET efficiency (E_FRET) versus fluorescence lifetime (τ) histograms of Nup153FG^PxFG in the presence and absence of Importinβ. The dotted line visualizes the center position of the FRET peak. The dashed (diagonal) lines show the static E_FRET relationship, on which a distribution would lie in the absence of fast dynamics. (B) Fluorescence lifetimes (τ) of the double labeled population accumulated from single molecule data in the absence (black) and presence (green) of Importinβ. Offset from a single exponential lifetime (dashed gray curve and arrow) is a strong indicator of protein dynamics. (C) Fluorescence correlation spectroscopy (FCS) traces retrieved from measurements of Nup153FG^PxFG (black dots) reflect a slower translational motion in the presence of Importinβ (green dots).

Figure 3

Nup153FG^PxFG⋅Importinβ Interaction by NMR Spectroscopy (A) ¹H-¹⁵N HSQC spectrum of Nup153FG^PxFG (red) overlayed with a spectrum of Nup153FG^PxFG in the presence of Importinβ (green, Nup to NTR molar ratio of 1.14, at a Nup concentration of 240 μM). (B) The intensity ratio of the bound and unbound form of Nup153FG^PxFG was plotted under the same conditions as in (A). (C) ¹⁵N R₂ relaxation rates at 25°C and a ¹H frequency of 600 MHz were measured at different concentrations of Importinβ (gray bars are without Importinβ; black, light green and dark green at Importinβ/Nup153FG^PxFG molar ratios of 0.17, 0.33, and 0.72 at the constant Nup153FG^PxFG concentration of 250 μM). (D) ¹⁵N R₂ of Nup153AG^{PxAG, F1374} in the absence (gray) and in the presence of Importinβ (red) overlayed with the rates for Nup153FG^PxFG in the presence of Importinβ under the same conditions (green). (E) For all F in the Nup153FG^PxFG sequence, ¹⁵N R₂ values were plotted against Importinβ concentration and fitted with a linear slope. The same analysis was performed for F1374 in Nup153AG^{PxAG, F1374} and compared to the same F in Nup153FG^PxFG (compare red to green slope). R2 with errors greater than 20% were excluded from the analysis. (F) Local K_d values were calculated from the slopes obtained in Figure S4. Gray bars correspond to K_d values obtained from Nup153FG^PxFG, the red bar shows the local K_d of Nup153AG^{PxAG, F1374} binding to Importinβ. Error bars show SD.

Figure 4

Binding of Nup153FG^PxFG to Importinβ^N (A–C) Contact area between (A) Nup153FG^PxFG and Importinβ^N and (B) diffusion coefficients D as a function of time for the 4 binding events out of 10 simulations (gray/black: prior to binding; different colors: after binding; black/red curves refer to the cartoon in (C) sampled using CHARMM22^∗ force field. (C) Snapshots collected along one of the recorded MD trajectories showing the binding between Nup153FG^PxFG (red cartoon) and Importinβ^N (gray surface). The binding sites on Importinβ^N and Nup153FG^PxFG FG-repeats are colored in orange and cyan respectively. (D) Nup153FG^PxFG radius of gyration (R_G) as a function of end-to-end distance (R_E) for the unbound (black) and bound (green) ensembles of Nup153FG^PxFG obtained from the simulations performed using CHARMM22^∗. See Figure S5 for data using the AMBER force field.

Figure 5

Association Kinetics for Nup153FG with Importinβ (A) Stopped-flow fluorescence anisotropy was used to monitor the binding of Importinβ (Impβ) at different concentrations to Nup153FG-Cy3B. A selection of anisotropy (r) traces against time is shown for Nup153FG alone (purple) and for the binding of Importinβ WT (black) and Importinβ^DA (red). (B) The observed rates (k_{obs,ultrafast}) from association experiments were plotted against the different Importinβ concentrations, the data were linearly fitted to obtain the association rate constants (k_on,ultrafast). (C) Apparent K_d,app values under the different experimental conditions. (D) k_on obtained from association experiments of Nup153FG and Importinβ at different ionic strengths fitted with a Debye-Hückel-like approximation to calculate the basal rate constant at infinite ionic strength. (E) Summary of the k_on values obtained from BD (dark bars) and FSF measurements (light bars) (Table S2D). Error bars show SD.

Figure 6

Nuclear Transport Assays of Importinβ and Importinβ^DA (A and B) DAPI staining shown in blue, and green fluorescent cargo (NLS-MBP-eGFP) in permeabilized HeLa cells incubated with either Importinβ (A) or Importinβ^DA (B) (scale bar 50 μm). After 45 min, cargo accumulation is higher in the nucleus in (A). (C) Single molecule trajectories of fluorescently labeled Importinβ were acquired in the equatorial plane of the nucleus exploiting an inclined (Hilo) illumination. (D) Representative image of acquired single molecule trajectories of Importinβ-Alexa488 (red lines) overlaid with the ensemble image of Importinβ-Alexa647 (in green, scale bar 1μm) used to identify the nuclear envelope position (blue line). Single particle tracks of the fluorescently labeled NTR (cyan lines) crossing the nuclear envelope were analyzed to yield the characteristic barrier crossing time. (E) The crossing time distributions reported for Importinβ (blue bars) and Importinβ^DA (red bars) are very fast.

Figure 7

Binding Modes of IDPs to Folded Proteins Schematic representation of various models describing the binding of an IDP to its folded partner. In an induced-fit mechanism the IDP partially or completely folds upon interacting with its partner, potentially showing an intermediate encounter complex as in the fly-casting mechanism (Shoemaker et al., 2000). In a conformational selection mechanism, the folded protein selects one (or several) conformation(s) of the IDP that best fits its binding pocket. These models suggest a shift in the IDP’s conformational ensemble. For the Nup⋅NTR complex we observed formation of an “archetypal” fuzzy and multivalent complex, a binding mode that on a global scale does not require major energy or time investment for the Nup to transit from its free to the bound conformation. Note that multiple NTRs can bind one Nup and vice versa.

Figure S1

Conformational Behavior of Nup153FG^PxFG/ Nup153FG^PxFG,F1374, Related to Figure 1 (A) Assignment of the ¹H-¹⁵N HSQC spectrum of Nup153FG^PxFG. Amino acid numbers refer to the sequence as in Uniprot P49790 (see sequences in Supplemental Experimental Procedures). Note that residues 1391 and 1392 refer to Cys and Ala for cloning and fluorescence labeling purposes rather than serines. (B) Characterization of Nup153AG^{PxAG, F1374}. Nup153AG^{PxAG, F1374 1}H-¹⁵N HSQC spectrum (red) overlaid with a spectrum of Nup153FG^PxFG (blue). The only F and neighboring G in the remaining FG-motif, as well as all amino acids with sufficient distances to mutated residues display very similar chemical shifts in the AG compared to the FG variant of the protein. (C–E) Atomic resolution ensemble description of Nup153FG^PxFG based on experimental NMR data. (C) Ramachandran maps for all Nup153FG^PxFG residues as obtained from the ASTEROIDS selection. FG-repeats are color-coded as in Figure 1 (A). (D) Local conformational sampling of Nup153FG^PxFG showing the population of four regions of Ramachandran space (βS: β sheet; βP: Polyproline; αR and αL: right and left handed alpha helices). Gray bars result from the ASTEROIDS ensemble selected using chemical shifts. Black lines illustrate the conformational sampling of the initial statistical coil ensemble. (E) Nup153FG^PxFG secondary chemical shifts obtained from experiments (gray bars) and ASTEROIDS selection from a Flexible Meccano ensemble (red), based on the experimental chemical shifts. Black lines represent the initial ensemble, prior to selection. (F) Local dynamic behavior of Nup153FG^PxFG and Nup153AG^{PxAG, F1374}. ¹⁵N R₂, ¹⁵N R₁, and {¹H}-¹⁵N steady-state heteronuclear Overhauser effects of Nup153AG^{PxAG, F1374} (gray bars) at a ¹H frequency of 600 MHz and 25°C compared to Nup153FG^PxFG (blue lines), showing very similar fast (ps-ns) timescale dynamics, characteristic of intrinsically disordered proteins.

Figure S2

Structure and Dynamics of Nup153FG, PxFG-rich region, in Presence and Absence of Importinβ and Controls, Related to Figure 2 (A) Schemes of (601 aa full FG domain) Nup153FG A1391TAG S1312C FRET mutant (probing the PxFG-rich region) and Nup153FG^PxFG. Numbers indicate the selected amino acid range with respect to the full-length Nup153FG (UniProt: P49790). Donor (green, Alexa488) and acceptor (red, Alexa594) were attached on the genetically encoded unnatural amino acid acetyl-phenylalanine (AcF) and a cysteine, respectively. The 2D histograms show burst integrated τ versus E_FRET histograms of a double-labeled Nup153FG. The plots are color-coded for frequency of occurrence. Top and right histograms are projections along the τ and E_FRET axis respectively. Changes in ratio between the so called donor-only peak (arising from molecules that contain no active acceptor) and the FRET peak typically originate from photophysical effects and different labeling efficiencies between the different sample preparation of full-length Nup153FG and Nup153FG^PxFG (see also Figure S3 for additional effects contributing to differences in observed number of events). Also shown are fluorescence correlation spectroscopy (FCS, G(τ)) traces retrieved from the same measurements (black curves). Those show slower translational motion in the presence of Importinβ (green curves). (B) Analog to (A), but data shown for an all F to A mutant termed Nup153AG, which binds Importinβ less well than the WT. (C) Burst integrated fluorescence lifetime (τ) versus FRET efficiency (E_FRET) histograms of a double-labeled Nup153FG in the presence (green) and absence (gray) of Importinβ are overlaid for comparison. Dashed lines represent the static FRET lines and red arrows indicate deviations due to donor-acceptor dynamics. (D) Fluorescence lifetimes (τ) of the double-labeled population accumulated from single molecule data in the absence (black) and presence (green) of Importinβ. The single exponential lifetime fit (dashed curve) does not adequately fit the fluorescence lifetime decay (red arrow), which is a strong indicator for protein dynamics. (E and F) Measurements of donor-acceptor dynamics timescales are non-trivial and have in the past been extracted from a combination of nanosecond resolved FCS (nsFCS) and smFRET for a few small IDPs (Soranno et al., 2012). We measured donor-acceptor dynamics of FRET labeled Nup153FG and Nup153FG^PxFG by splitting polarized light of the donor light channel onto three detectors. Parallel polarized light (||) was detected on two detectors (each recording 50% of the total intensity in the || channel), perpendicular light (⊥) on one detector. These experiments were performed with a higher protein concentration of 5 nM. || light was correlated with || light (red curves) and with ⊥ light (blue curves). The marked difference of red and blue curves in the case of Nup153FG (left plot) indicated that—due to the large size of Nup153FG compared to most previously studied proteins—fluorescence anisotropy significantly hampers analysis of the correlation curves. Small contributions from donor-acceptor fluctuations can therefore not easily be extracted. A similar observation was previously made for bigger protein complexes in accordance with our interpretation (Hillger et al., 2008). Nup153FG^PxFG (E, right plot), however, a much smaller protein, does not pose these difficulties and correlation times can therefore be analyzed: The curves in which || light was correlated with ⊥ light were first fit with a simple diffusion model (down to ∼0.3 μs) in which the triplet component of the donor was extracted. The correlation curves were then normalized and plotted with a linear scale up to 500 ns (F, dots in left panel). These curves were then fit (solid lines) with the following model: G(t) = (1-α∙exp(t/τα))^∗(1+β∙exp(t/ τ_β))(1+γ∙exp(t/τ_γ)), with component α the antibunching, component β the FRET dynamics and γ the triplet component, where τ_γ was fixed as extracted from diffusion fits. The right panel in (F) displays a statistical analysis (median 89 ns and 65 ns; 25^th quantiles 31 ns and 23 ns, 75^th quantile 768 ns and 112 ns without (black) and with (green) Importinβ, respectively) of the obtained fit values (note that amplitudes are so small that the fitting error is rather large). Correlations were performed with SymPhoTime (Picoquant, Berlin, Germany).

Figure S3

Conformation of Different Nucleoporin Regions in the Presence and Absence of Different NTRs Probed by smFRET, Related to Figure 2 This figure shows a cartoon of the labeling site in the respective Nup (Nup153FG or yNup49FG) on the left, as well as the resulting 2D histograms (τ versus E_FRET) in the presence of the respective NTR (depicted on top: Importinβ [orange], TRN1 [purple], CRM1 [blue] and NTF2 [green]). In the smFRET experiments, the Nup was used at pM concentration, and the NTR at 1 μM. Also shown are FCS experiments, were the Nups was used at 10 nM concentration. In summary, FCS detects that overall Nups are binding competent to all tested NTRs, while in the smFRET measurements no substantial conformational changes were observed in agreement with the main conclusion of this work. Note that only in the yNup49FG experiments, NTR concentration was increased to 10 μM, due to an overall lower binding affinity. However, high background in the CRM1 sample preparation did not permit high resolution single molecule FRET measurements anymore at such high concentration, and was thus excluded from this one analysis. Note also that the absolute values for FRET, depend among other factors mainly on the amino spacing of the dyes, which were different for the different constructs. Furthermore, a burst search algorithm was used to select signals from noise with a total photon count >100. In such an analysis it is inherent that the absolute values (intensities, i.e., number of events) for smFRET peaks differ depending on e.g., diffusion time (different for different sized complexes) different background (due to slight contamination from different preparations), sample sticking issues, etc.

Figure S4

Interaction of Nup153FG^PxFG with NTRs Related to Figure 3 (A) Titration of Nup153FG^PxFG with different transport receptors. ¹H-¹⁵N HSQC spectra of Nup153FG^PxFG in the presence of different concentrations of Importinβ (left row), TRN1 (middle row) and NTF2 (right row) were overlayed. Importinβ was added at a molar excess of 0.17 (light blue), 0.33 (dark blue) and 0.72 (magenta), TRN1 at a molar excess of 0.17 (light blue), 0.33 (dark blue), and 0.61 (magenta), and NTF2 at a molar excess of 0.33 (light blue), 1 (dark blue), 2 (magenta), and 4 (orange). Nup153FG^PxFG was kept at a concentration of about 250 μM throughout the titration. The spectrum of the unbound Nup153FG^PxFG is displayed in red. Chemical shift changes are indicated by black arrows. Spectral differences are clearly localized to the same amino acids in all cases. For NTF2 it was experimentally possible to access much higher molar ratios, and the molecular weight of the partner is much smaller than Importinβ and TRN1, explaining why larger shifts are observed. (B) Relaxation of Nup153FG^PxFG at different NTR mixtures. ¹⁵N R₂ of Nup153FG^PxFG in the presence of different concentrations of NTRs (concentrations as described in A; colored in the order of increasing NTR concentration as light blue, dark blue, magenta, and orange). ¹⁵N R₂ of Nup153FG^PxFG alone are displayed as gray bars. (C) Residue specific K_d values. Residue specific K_d values were extracted from the R₂ rates as described in the Supplemental Experimental Procedures. The rotational correlation time of Nup153FG^PxFG bound to TRN1 was assumed to be the same as when bound to Importinβ, an assumption that is justified by the similar architecture of the two transport receptors and their similar molecular weight. Since we do not have an estimate for the rotational correlation time of Nup153FG^PxFG bound to NTF2, we extracted the residue specific K_d values from the absolute chemical shift changes from a titration of Nup153FG^PxFG with NTF2 to near saturation. (D–F) Binding of Nup153AG^{PxAG, F1374} to Importinβ. (D) Illustration of chemical shift changes upon interaction of ¹⁵N labeled Nup153AG^{PxAG, F1374} with Importinβ. ¹H-¹⁵N HSQC of Nup153AG^{PxAG, F1374} (red) overlayed with a spectrum of Nup153AG^{PxAG, F1374} in the presence of Importinβ (green, Nup at 250 μM, Importinβ at 190 μM). (E) ¹⁵N R₂ relaxation rates of Nup153FG^PxAG were measured at different concentrations of Importinβ (gray bars are without Importinβ; dark green, blue, and light green at Importinβ/Nup153AG^{PxAG, F1374} molar ratios of 0.38, 0.6, and 0.76 at a constant Nup153AG^{PxAG, F1374} concentration of 250 μM). Dashed line indicates the position of F1374. (F) Associated ¹⁵N R₁ relaxation rates (see E). (G) Relaxation dispersion of Nup153FG^PxFG when interacting with Importinβ. CPMG relaxation dispersion measurements were performed with Nup153FG^PxFG bound to Importinβ at a stoichiometry ratio of 0.72 (Importinβ/nucleoporin; Nup concentration was 250 μM). The effective R₂ (R₂^eff) at different refocusing frequencies (υ_cpmg) is plotted for two representative Fs in the Nup153FG^PxFG sequence. (H) Estimation of local dissociation constants. ¹⁵N R₂ values plotted against Importinβ concentration (molar ratios of 0.17, 0.133, and 0.72 at the constant Nup153FG^PxFG concentration of 250 μM) and fitted with a slope from which FG specific K_d values were extracted under the assumption that the local rotation time (τ_c) of Nup153FG^PxFG bound to Importinβ equals τ_c of Importinβ itself. (I) Intensity ratio of the bound and unbound form of Nup153FG^PxFG at a Nup concentration of 250 μM and Importinβ at 180 μM as obtained from two ¹H-¹⁵N HSQC spectra. (J) Characterization of the structure and dynamics of Nup153FG^PxFG in complex with Importinβ. ¹⁵N R₁ of Nup153FG^PxFG at different concentrations of Importinβ (gray bars - without Importinβ; light blue, dark blue, magenta at Importinβ/Nup153FG^PxFG molar ratios of 0.17, 0.133, and 0.72 at the constant Nup153FG^PxFG concentration of 250 μM, as in B). (K) Chemical shift titration of Nup153FG^PxFG with NTF2. Residue specific K_d values derived from the absolute chemical shift changes from a titration of Nup153FG^PxFG with NTF2. The data were fit with a simple binding model under the assumption of excess NTF2 (see Supplemental Experimental Procedures). Shown are titration curves for two representative Fs. (L) Chemical shift titration of Nup153FG^PxFG with NTF2. Absolute chemical shift changes mapped for Nup153FG^PxFG in the presence of 4-fold excess of NTF2. (M) Cα and CO secondary chemical shifts obtained from Nup153FG^PxFG alone (gray bars) and in the presence of 24-fold excess of NTF2 at a Nup153FG^PxFG concentration of 80 μM (red curve), demonstrating that local conformational sampling is conserved upon interaction. Note that prolines are excluded from this plot. Error bars show SD.

Figure S5

Computational Analysis of Nup153FG^PxFG and Nup153FG^FxFG Dynamics and of Their Binding to Importinβ^N, Related to Figure 4 (A) Nup153FG^PxFG conformations (means of the cluster) that have been selected from MD simulations in the absence of Importinβ^N using AMBER99-sb^∗-ILDN (left) and CHARMM22^∗ (right). FG-repeats are shown by their van der Walls (vdW) radius. (B) Count for each cluster of the simulated conformations of Nup153FG^PxFG as simulated with AMBER99-sb^∗-ILDN (left) CHARMM22^∗ (right). Stars indicate the clusters corresponding to the conformations represented in (A). (C and D) Average solvent exposure of Nup153FG^PxFG (C) and Nup153FG^FxFG (D) FG-repeats estimated from molecular dynamics trajectories collected from molecular dynamics simulations performed using AMBER99-sb^∗-ILDN (C - left panel) and CHARMM22^∗ (D, left) force fields. The average degree of exposure has been calculated as the ratio between the mean SASA of the FG-repeat within the equilibrated Nup153FG^PxFG/FxFG ensembles and of an FG-repeat not forming any tertiary contacts. (E and H) Contact areas between Nup153FG^PxFG (E) or Nup153FG^FxFG (H) and Importinβ^N as a function of time for the binding events observed in the MD simulations performed using AMBER99-sb^∗-ILDN force field. (F and I) Diffusion coefficients D, of Nup153FG^PxFG (F) and Nup153FG^FxFG (I) as a function of time are reported for the same replicas shown in (E) and (H). The drop in the diffusion coefficient is coincident with the increase in contact area upon the binding of the Nup153FG^PxFG and Nup153FG^FxFG to Importinβ^N. (G and J) Percentage of binding as a function of the binding time observed in the MD runs for the simulations reporting the binding of Nup153FG^PxFG (G) or Nup153FG^FxFG (J) to Importinβ^N. The fit of (solid line for Nup153FG^PxFG - Importinβ^N and for Nup153FG^FxFG - Importinβ^N association) such a cumulative distribution of binding events gives a rough estimate of the k_{on, MD} of approximately 10¹⁰ M⁻¹s⁻¹. We note that due to the low sampling and known lower friction of water and other force field issues, (Mercadante et al., 2015) the results form MD rather define an upper limit. (K) Nup153FG^PxFG radius of gyration (R_G) as a function of end-to-end distance (R_E) for the unbound (black) and bound (green) ensembles of Nup153FG^PxFG obtained from the simulations performed using the AMBER99-sb^∗-ILDN force field. (L) Count of specific contacts has been performed for the simulations of Nup153FG^PxFG and Nup153AG^PxAG (in which all F residues were mutated to A) binding to Importinβ^N. Blue and red bars indicate the five different simulations performed in each case using the AMBER99-sb^∗-ILDN force field. In the case of the Nup153AG^PxAG, two out of five simulations did not yield any specific contact between the partners. Specific contacts have been considered as any contact occurring within a cutoff distance of 0.6 nm between the FG/AG-repeats of Nup153FG/AG^PxFG/PxAG and any binding site on Importinβ^N.

Figure S6

MD Analysis of the Nup153FG^PxFG⋅Importinβ^N Complex, Related to Figure 4 (A) Comparison of the Nup153FG^PxFG-Importinβ^N complexes observed in X-ray crystallographic studies (PDB: 1F59) (left), MD simulations performed using the CHARMM22^∗ (center panel) and the AMBER99-sb^∗-ILDN (right) force fields. The surface of Importinβ^N is shown in gray whereas Nup153FG^PxFG is shown in red. FG-repeats are shown in cyan. (B) Interaction sites of Nup153FG^PXFG and Importinβ^N. “^∗” and “ˆ” symbols account for the contacts observed in the CHARMM22^∗ and AMBER99-sb^∗-ILDN simulations respectively. Grey cylinders represent the α helices along the structure of Importinβ^N. (C) Distance between the center of mass (COM) of Nup153FG^PxFG and Importinβ^N as a function of the observed time of binding. The upper triangles show simulations reporting successful binding events whereas red crosses show simulations in which binding between the partners has not been observed during the simulated time.

Figure S7

Association Stopped-Flow Experiments of Nup153FG and Nup153FG^PxFG to NTRs, Related to Figure 5 (A) Association experiment of Nup153FG^PxFG with Importinβ WT, a selection of anisotropy traces is shown. At the tested conditions no kinetic changes were detected upon binding. (B–F) The observed rates from association experiments of Nup153FG to Importinβ WT (black) and double mutant (red) from the major phase (k_{obs,ultrafast}) (B) and the minor phase (k_obs,fast) (C) are plotted against the different Importinβ concentrations. The amplitudes A1_ultrafast and A2_fast correspond to the amplitudes from the fit associated to k_{obs,ultrafast} and k_obs,fast respectively (D). Bars indicating A1_ultrafast and A2_fast are shown for each measurement at the different concentrations of Importinβ WT and Importinβ^DA mutant. Nup153FG and Importinβ WT were also tested at different ionic strength conditions, the observed rates of the ultrafast component (k_{obs,ultrafast}) (E) and of the fast component (k_obs,fast) (F) obtained from different ionic strengths: 0.05 M (dark green), 0.08 M (light green) 0.1 M (blue), 0.2 M (orange),0.3 M (pink), 0.45 M (purple), 0.65 (green khaki), 0.75 M (brown) and 1 M (cyan) are plotted against Importinβ concentration, numbers in brackets correspond to independent replicate measurements. As discussed in detail in (Milles and Lemke, 2014) anisotropy changes in such a disordered system depend on a variety of parameters, including degree of segmental motion and multivalency. As we have several degrees of multivalency (several FG repeats in the Nup, several binding sites in the NTR, and the ability to form higher order complexes than 1:1) the exact origin of the slower component cannot be assessed. In line with related studies a likely origin could be additional binding of NTRs (Wagner et al., 2015). As such FSF measurements report on overall formation of Nup153FG⋅Importinβ complex i.e., one or more F binding. The fastest k_on measured defines even a lower limit for the first F binding. However, we note that both observed reaction rates (major and minor phase) are remarkably fast, and all our conclusions are also valid if one considers only the minor phase. (G and H) Association rates obtained from Brownian Dynamics simulations. The curves report the estimated k_on as a function of the contact distance between the binding partners when a different number of independent contacts (between 1 to 4) are formed between Importinβ^N or Importinβ^N,DA and Nup153^PxFG. The red curves report the cases in which two reaction criteria are satisfied. The condition in which two reaction criteria are satisfied at a distance of 0.7 nm has been used to extrapolate the value of estimated association rate. Pink shaded rectangles show the range of k_on when two contacts between 0.65 and 0.75 nm are formed. The simulations were performed considering Nup153FG^PxFG interacting with Importinβ^N (G) or Importinβ^N,DA (H) and considering all interactions contributing to the binding or in the absence of electrostatic interactions or apolar desolvation. (I and J) Association kinetics for Nup153FG with different NTRs probed by stopped-flow. (I) Here we show the observed rates (k_{obs,ultrafast} and k_obs,fast from association experiments with Importinβ (orange), TRN1 (purple) and NTF2 (dark green) to Nup153FG Cy3B labeled at the positions 883C, probing an FxFG enriched region. The obtained association constants (k_on,ultrafast, k_on,fast) from the slope of the linear fit are indicated on the plots. (J) Analog to Figure 5, we shown here in addition the observed rates (k_{obs,ultrafast} and k_obs,fast from stopped-flow association experiments with TRN1 (purple) to Nup153FG Cy3B labeled at the positions 1391C (probing a PxFG enriched region). The obtained association constants (k_on) from the linear fit are indicated on the plots and summarized in Table S2C. We note that the observed rates are at the limit of what is detectable with advanced stopped-flow instrumentation. The signal to noise in the observed anisotropy r depends on a variety of parameters, the labeling site, the multivalency, the binding strength, the degree of segmental motion etc. as detailed in (Milles and Lemke, 2014). In line with the observed lower affinity of the used transport receptors for yNup49 (data not shown), signal to noise was not sufficient to perform high resolution stopped-flow measurements of this Nup. Perfectly in line with our results reported in Figure 5, for all those different NTRs and different FG regions we observed remarkably fast rates consistently with a k_on,ultrafast > 5·10⁸ M⁻¹s⁻¹. All data are summarized in Table S2C.