Karyopherin enrichment at the nuclear pore complex attenuates Ran permeability

Nucleocytoplasmic transport (NCT) describes the selective exchange of macromolecules between the nucleus and cytoplasm in eukaryotes (Görlich and Kutay, 1999). This is mediated by conduits of 50–60 nm diameter within the nuclear envelope, known as nuclear pore complexes (NPCs) (Eibauer et al., 2015; Kim et al., 2018; von Appen et al., 2015). Given their considerable size, NPCs are permeable to passive molecules below ∼40 kDa, whereas larger non-specific macromolecules are generally withheld (Popken et al., 2015; Timney et al., 2016). Meanwhile, three main groups of protein are selectively trafficked across the NPC central channel to sustain NCT. These are transport receptors known as karyopherins (Kaps), signal-specific cargos and the Ran GTPase that harmonizes the process.

A priori exclusive NPC access is reserved for Kaps (Kimura and Imamoto, 2014; Tran et al., 2007). These include importins that deliver cargos bearing nuclear localization signals (NLS) (Boulikas, 1994; Cokol et al., 2000) into the nucleus, and exportins that usher cargos containing nuclear export signals (NES) (Xu et al., 2012) out of it. Selective Kap transport is underpinned by multivalent interactions with numerous phenylalanine-glycine (FG)-repeat-rich, intrinsically disordered nucleoporins (FG Nups) that line the NPC channel (Sakiyama et al., 2016). For instance, the classical 97 kDa import receptor karyopherin subunit β1 (KPNB1, hereafter referred to as Kapβ1) (Cingolani et al., 1999) engages up to ten FG repeats (Bayliss et al., 2000, 2002; Bednenko et al., 2003; Isgro and Schulten, 2005). Otherwise, the FG Nups are considered to adopt barrier-like characteristics, such as polymer brushes (Lim et al., 2007; Rout et al., 2000), gel-like meshworks (Frey and Görlich, 2007; Labokha et al., 2013) or variations of these (Yamada et al., 2010). Still, Kap-cargo complexes are considerably larger than the non-specific size cut-off. Furthermore, Kapβ1 recruits adaptor proteins from the karyopherin subunit α family (KPNA, hereafter referred to as Kapα) that bind directly to a diverse repertoire of NLS-cargos such as transcription factors (Pumroy and Cingolani, 2015; Xu and Massagué, 2004). Hence, our understanding of how NPCs reconcile physical size exclusion with biochemical selectivity to mediate NCT remains incomplete.

One peculiarity concerns Ran (Melchior et al., 1993; Moore and Blobel, 1993), which controls the site of cargo release, accumulation and recycling of Kaps to underpin NCT directionality across the nuclear envelope (Weis, 2003). This is sustained by the interconversion of its two nucleotide-bound forms, RanGTP and RanGDP, which are localized to the nucleus and cytoplasm, respectively (Görlich et al., 1996). With a molecular mass of 25 kDa, Ran is below the NPC size limit for non-specific molecules. Also, neither RanGDP nor RanGTP interact with the FG repeats (Rexach and Blobel, 1995). Yet, the concentration of RanGTP is estimated to be at least 200 times higher in the nucleus than in the cytoplasm (Görlich et al., 2003; Kalab et al., 2002; Smith et al., 2002). Thus, how an uncontrolled mixing of RanGTP and RanGDP is prevented at NPCs remains unknown. Importantly, a disruption in the Ran gradient results in the loss of NCT directionality (Nachury and Weis, 1999) and has been linked to apoptosis (Wong et al., 2009), hyperosmotic stress (Kelley and Paschal, 2007) and disease (Eftekharzadeh et al., 2018).

In the nucleus, RanGTP binds Kapβ1 to disassemble NLS-cargo–Kapα–Kapβ1 complexes (Chi et al., 1996; Görlich et al., 1996; Rexach and Blobel, 1995). This serves to facilitate the nuclear retention of NLS-cargos whose return to the cytoplasm is hindered in the absence of FG Nup binding. On the other hand, RanGTP–Kapβ1 retains its interactions with the FG Nups to return through NPCs (Kapinos et al., 2017). At the cytoplasmic periphery, RanGTP is hydrolyzed to RanGDP by RanGTPase-activating protein 1 (RanGAP1) together with the Ran-binding proteins RanBP1 and RanBP2 (Lounsbury and Macara, 1997; Vetter et al., 1999). This frees Kapβ1, which is then able to seek out the next NLS-cargo. Still, Ran seems to accumulate at NPCs (Abu-Arish et al., 2009; Smith et al., 2002; Wong et al., 2009; Yang and Musser, 2006), suggesting that it does not freely diffuse through the NPC like other non-specific molecules of similar size (Timney et al., 2016). RanGDP then recruits a dedicated import factor, i.e. nuclear transport factor 2 (NUTF2, hereafter referred to as NTF2) (Ribbeck et al., 1998; Smith et al., 1998), which returns RanGDP to the nucleus. Upon re-entry, the chromatin-bound enzyme regulator of chromosome condensation 1 (RCC1, also referred to as Ran guanine nucleotide exchange factor or RanGEF) (Klebe et al., 1995b; Renault et al., 2001) recharges RanGDP to RanGTP to complete the cycle. In this manner, NCT cargo delivery and the recycling of Kaps are regulated by RanGAP1 and RanGEF as well as the controlled exchange of RanGTP and RanGDP across NPCs (Abu-Arish et al., 2009; Izaurralde et al., 1997; Kalab et al., 2006, 2002).

More recently, a steady-state enrichment of Kapβ1 was uncovered in FG Nup layers (Kapinos et al., 2014; Schoch et al., 2012; Vovk et al., 2016; Wagner et al., 2015; Zahn et al., 2016) and in NPCs (Görlich et al., 1995; Kapinos et al., 2017; Lowe et al., 2015). We also found that depleting this Kapβ1 pool abolishes NPC barrier function against large non-specific cargos (see Kapinos et al., 2017), which suggests that Kaps serve as bona fide constituents of the NPC barrier mechanism. This motivated our present study, in which we show that NPC-bound Kapβ1 selectively retains RanGTP and RanGDP at the NPC but not passive molecules of similar size (e.g. GFP). This is due to binding interactions with Kapβ1 at the pore, which are stronger for RanGTP and weaker for RanGDP. In comparison, NPCs that lack Kaps show unrestricted Ran movement, i.e. a Ran ‘leak’. Therefore, RanGTP outflow depends on the hydrolysis of RanGTP to RanGDP, whereas RanGDP import requires NTF2. These results explain how Kaps might serve to maintain the Ran gradient by regulating the movement of RanGTP/GDP through NPCs.

A pool of endogenous Kapβ1 and Kapα (endoKapβ1 and endoKapα, respectively; hereafter collectively referred to as endoKaps) is generally retained at the nuclear envelope after cell permeabilization with digitonin (Kapinos et al., 2017). Hence, we asked whether endoKaps can impede the movement of exogenous RanGDP (exoRanGDP) through NPCs (Fig. 1A). For comparison, we incubated permeabilized cells in Ran mix to deplete the pool of endoKaps from the NPC (Kapinos et al., 2017), which resulted in a reduction of endoKapβ1 (∼50%) and endoKapα (∼80%) within the nucleus and nuclear envelope, respectively (Fig. 1B,C). Then, we incubated both samples in 5 µM exoRanGDP for 1 h until equilibration was reached within the nucleus and its exterior. This was followed by a short wash with PBS (3×5 min) to investigate the extent of exoRanGDP nuclear retention in the presence and absence of endoKaps. In comparison to the permeabilized cell control (and under the same conditions), exoRanGDP was reduced by ∼50% in both the nucleus and nuclear envelope when endoKaps were depleted (Fig. 1B,C). Meanwhile, to ensure the efficacy of our assay, we confirmed that nuclear retention of exoRanGDP following its incubation is considerably larger than its remainder after the washing step (Fig. S1A–C). Taken together, our results suggest that the presence of endoKaps in NPCs impedes the outflow of exoRanGDP from the nucleus.

On the basis of physiological estimates, we next filled endoKap-reduced nuclei with 5 µM exoRanGDP (Görlich et al., 2003) followed by 10 µM exogenous Kapβ1 (exoKapβ1) (Eisele et al., 2010) that re-populated the NPCs. These nuclei were compared against samples re-populated with exoKapα–Kapβ1 (20 µM:10 µM) (Fig. 2A). Unexpectedly, exoRanGDP retention was the lowest in nuclei harboring exoKapα–Kapβ1, i.e. 25% less than in control samples that lacked exoKaps (Fig. 2B,C). In comparison, nuclei that contained exoKapβ1 alone showed 20% more exoRanGDP retention than control cells. This signified that exoKapα–Kapβ1 does not impede exoRanGDP outflow at the NPC as effectively as exoKapβ1 alone. Nevertheless, we did obtain exoRanGDP fluorescence at the nuclear envelope of control cells, which suggested residual binding with either endogenous transport receptors or other NPC components (Görlich et al., 1996; Partridge and Schwartz, 2009; Schrader et al., 2008). Following these observations, we rationalized that NPCs might effectively impede exoRanGDP movement based on its biochemical interactions with exoKapβ1.

Subsequently, binding affinity measurements between exoKapβ1 and either exoRanGDP or a GDP-bound, non-hydrolyzable Ran mutant comprising a Gln69 to Leu point mutation (RanQ69L-GDP) resulted in K_d values of 0.5±0.07 µM or 1.3±0.15 µM, respectively (Fig. S1D), consistent with literature values (Forwood et al., 2008; Lonhienne et al., 2009). In comparison, binding between Kapα and Kapβ1 is stronger (K_d=0.2 µM) (Bednenko et al., 2003; Catimel et al., 2001; Kapinos et al., 2017). For this reason, exoRanGDP is unable to outcompete exoKapα for exoKapβ1. Indeed, the nuclear and nuclear envelope exoRanGDP signals are at least ∼30% higher for exoKapβ1 than exoKapα·Kapβ1 (Fig. 2B,C). Therefore, we reasoned that exoKapβ1 in the NPC might serve to restrict exoRanGDP outflow. By contrast, a lack of binding of exoRanGDP to exoKapα–Kapβ1 results in a higher outflow of exoRanGDP from the nucleus.

To validate the latter, we used GFP (∼26 kDa), which is similar in size compared with Ran but does not bind to FG Nups or Kaps and, therefore should not be retained at the nuclear envelope. Indeed, neither exoKapβ1 nor exoKapα–Kapβ1 could stem the outflow of GFP from the nucleus (Fig. S1E,F), which is consistent with in vivo observations of small passive cargos of equivalent size (Abu-Arish et al., 2009; Timney et al., 2016). This is a key finding because it shows that the NPC size exclusion limit for non-specific cargos does not preclude Kap occupancy at the pore.

RanGDP is converted by RanGEF into RanGTP (k_cat=3.5 s⁻¹, see Fig. S2A) (Klebe et al., 1995a), which binds Kapβ1 in the nucleus to displace Kapα during NCT. Given that the binding of RanGTP to Kapβ1 is significantly stronger (K_d=0.035 µM) (Bednenko et al., 2003; Hahn and Schlenstedt, 2011; Kapinos et al., 2017) than RanGDP to Kapβ1, we wondered how their efflux would differ. Upon verifying that RanGEF, RanGAP1 and RanBP2 were present following permeabilization (Fig. S2B,C), we again entrapped exoRanGDP with 10 µM exoKapβ1 or exoKapα–Kapβ1 and added an energy-regenerating mixture comprising 2 mM GTP, 0.1 mM ATP, 4 mM creatine phosphate and 20 U/ml creatine kinase (Ribbeck et al., 1998) to enable RanGEF activity (Fig. 3A). Although exoRanGDP was withheld by exoKapβ1 because they bound to each other (Fig. 2), we observed a dramatic 50% reduction of its fluorescence inside the nucleus and at the nuclear envelope when energy mix was added, indicating that exoRanGDP was converted to exoRanGTP (Fig. 3B–D). Indeed, the same was true in terms of the nuclear signal when exoKapα–Kapβ1 was used, except that the signal at the nuclear envelope was slightly higher. Still, because both exoRanGDP–Kapβ1 and exoRanGTP–Kapβ1 bind FG Nups, we questioned whether exoRanGTP efflux was promoted by RanGAP1, which hydrolyses GTP at a rate of k_cat=2.1 s⁻¹ (Klebe et al., 1995a).

To validate our hypothesis, we used RanQ69L-GDP, which can be converted into its GTP-bound form by RanGEF (Fig. S2A) but cannot be hydrolyzed by RanGAP1 (Klebe et al., 1995a). Similar to exoRanGDP (Fig. 2), exoKapβ1 impedes the outflow of RanQ69L-GDP leading to signal increase in the nuclear envelope and nucleus (Fig. S3) due to their binding (Fig. S1D). Also, RanQ69L-GDP leaked out non-specifically in the presence of exoKapα–Kapβ1 compared to control. Yet, when energy mix was present, the nuclear signal for RanQ69L-GTP was only slightly reduced by 10% (Fig. S4) compared to that of exoRanGTP (∼50%; Fig. 3D). Hence, in the absence of hydrolysis, RanQ69L-GTP remains bound to Kapβ1 residing at NPCs and does not depart from the nucleus.

NTF2 binds to RanGDP with a K_d=75–240 nM (Chaillan-Huntington et al., 2000), and regulates the re-import of RanGDP to the nucleus (Ribbeck et al., 1998) based on interactions between NTF2 and the FG Nups (Wagner et al., 2015). We therefore asked if NTF2 can return exoRanGDP to the nucleus after exoRanGTP hydrolysis in the presence of exoKapβ1 at the NPCs (Fig. 4A). As before (Fig. 3), exoRanGDP was significantly reduced when energy mix was supplied, due to its conversion to exoRanGTP (by RanGEF) followed by hydrolysis back to exoRanGDP at the NPC (by RanGAP1). In marked contrast, the exoRanGDP signal in the nucleus increased almost 200% over control when 4 µM exogenous NTF2 (exoNTF2) was included with the energy mix (Fig. 4B,C). Thus, NTF2 replenishes exoRanGDP in the nucleus to close the Ran cycle following exoRanGTP hydrolysis.

We have tested specific conditions related to NCT to study how the mixing of RanGDP and RanGTP is minimized at NPCs. A key finding is that an enrichment of Kapβ1 within NPCs attenuates the permeability of Ran but does not impede passive molecules of comparable size, e.g. GFP. By contrast, Ran freely diffuses through the NPC when Kapβ1 is absent. This is consistent with simulations that have shown that the Ran gradient is sensitive to changes in NPC permeability and that a retention mechanism may be required to establish the steep Ran gradient (Becskei and Mattaj, 2003; Görlich et al., 2003; Kopito and Elbaum, 2009). This may also explain observations of Ran accumulation at NPCs (Abu-Arish et al., 2009; Smith et al., 2002; Wong et al., 2009; Yang and Musser, 2006).

These findings further underscore the role of Kapβ1 as an essential constituent of the NPC barrier mechanism i.e. Kap-centric control (Kapinos et al., 2017, 2014; Lim et al., 2015) (Fig. 4D). Clearly, the size of Ran is below the non-specific size limit (Paine et al., 1975; Popken et al., 2015; Timney et al., 2016) and does not bind to FG repeats. Hence, the FG Nups alone are insufficient to prevent a mixing of RanGTP and RanGDP. Rather, the binding of RanGTP and RanGDP to Kapβ1 at the NPC would minimize their mixing. Nevertheless, this depends on the distribution of Kapβ1 complexes that co-exist in the NPC at steady-state, which remains poorly defined. RanGDP might interact with Kapβ1 (and not Kapα–Kapβ1) long enough to be converted into RanGTP by RanGEF in the nucleoplasm. Subsequently, RanGTP binds with stand-alone Kapβ1 or Kapα–Kapβ1 to form RanGTP–Kapβ1. Following GTP hydrolysis by RanGAP1, RanGDP then departs from the NPC into the cytoplasm, being the more weakly bound component of the complex. Incoming Kapα might further promote RanGDP release by binding Kapβ1. For these reasons, it is compelling that RanGDP requires NTF2 to be expeditiously shuttled back through NPCs enriched with Kapβ1 (Wagner et al., 2015).

But why would it be crucial to prevent a mixing of the two nucleotide-bound forms of Ran? On a mechanistic level, the mixing of RanGTP and RanGDP reduces the Ran gradient. This can dramatically alter the cellular distribution of Kaps, as shown for exportin-t (Kuersten et al., 2002) leading to cargo mislocalization (Wong et al., 2009) and, potentially, in extreme cases, an inversion of NCT (Nachury and Weis, 1999). Nevertheless, the sensitivity of cells to a mixing of RanGTP and RanGDP is unclear, and several questions remain. How much of this mixing is tolerable in cells? Is the enzymatic activity of RanGEF and RanGAP1 sufficient to maintain the Ran gradient in vivo? How do cells recover from changes to the Ran gradient, such as in response to hyperosmotic stress (Kelley and Paschal, 2007)? Might the mixing of RanGTP and RanGDP be cause or consequence of defects in Kap transport efficiency, cargo directionality and accumulation? Indeed, other small essential cargos (<40 kDa), such as histones (Mühlhäusser et al., 2001) and ribosomal proteins (Jakel and Görlich, 1998), utilize dedicated Kaps for nuclear import. Hence, we postulate that Kap enrichment at the NPC selectively restricts the uncontrolled mixing of small essential proteins across the nuclear envelope.

All exogenous proteins, such as human Kapβ1, Kapα, wild-type Ran (RanWT) and RanQ69L were cloned, expressed and purified as described previously (Kapinos et al., 2017, 2014; Schoch et al., 2012; Wagner et al., 2015). Kapβ1 was eluted in 10 mM TrisHCl pH 7.5, 100 mM NaCl, 1 mM DTT-containing buffer and concentrated to 15–20 μM. A full-length human Kapα construct (pCMVTNT-T7-KPNA2; Addgene plasmid #26678) was cloned using EcoRI/BamII restriction enzymes into the pQE70 vector with a His₆ tag at its C-terminus and a short linker (-GSRSHHHHHH) that does not affect the complex formation of this protein with Kapβ1. Human Kapα was purified and eluted in 20 mM TrisHCl pH 7.5, 100 mM NaCl, 2 mM DTT, 10% glycerol. A plasmid (pQE32) with a full-length human RanQ69L construct was a gift from Ulrike Kutay (ETH Zurich, Zurich, Switzerland) (Kutay et al., 1997). Using site-directed mutagenesis, RanWT was derived from this clone using the following primers: 5′-GTATGGGACACAGCCGGCCAGGAGAAA TTCGGTGGACTG-3′ and 5′-CAGTCCACCGAATTTCTCCTGGCC GGCTGTGTCCCATAC-3′.

RanWT or RanQ69L were purified using a Ni-NTA column (Roche) over an imidazole gradient (10–500 mM) and then dialyzed into a 10 mM HEPES buffer pH 7.2 with 100 mM NaCl as described previously (Kapinos et al., 2017). Purified proteins were incubated for 30 min at 4°C with 10 mM EDTA. Then 25 mM MgCl₂ was added together with 1 mM GTP or GDP that ensured its binding to nucleotide-free Ran. Finally, GTP or GDP-loaded RanWT or RanQ69L proteins were dialyzed into PBS buffer, pH 7.2 (Invitrogen, Lifesciences), in the presence of 1 mM MgCl₂ and purified using size-exclusion column (Superdex 200 HiLoad 16/60; GE). Finally, these proteins were concentrated to 35-45 µM in PBS containing 1 mM MgCl₂.

The full-length rat NTF2 was cloned, expressed and purified as before (Wagner et al., 2015). Typical stock concentration of NTF2 was ∼250–300 μM in PBS (Invitrogen, Lifesciences). RanGEF construct was obtained from GenScript in pUC57 vector and re-cloned into pPEP-TEV plasmid using XbaI/BamHI restriction enzymes for the further expression. It was purified as described before for Kapβ1 (Kapinos et al., 2014; Schoch et al., 2012). The protein quality was verified using 12% SDS PAGE. All purified recombinant proteins were shock-frozen and stored at −80°C.

Recombinant proteins were labeled with fluorescent dyes (1:5 ratio) for 2 h at room temperature in light-protected vials. AlexaFluor 647 maleimide (Invitrogen, Lifesciences) was used to label RanWT or RanQ69L. AlexaFluor 488 maleimide (Invitrogen, Lifesciences) was used to label Kapβ1. Atto 550 maleimide (Sigma-Aldrich) was used to label Kapα. PD MiniTrap G-25 sample preparation spin columns (GE Healthcare, Lifesciences) were used to remove the excessive dye. The degree of labelling (DOL) was calculated following Nanodrop UV-Vis spectrometry to measure the respective dye and protein absorptions.

HeLa cells (ATCC^® CCL-2™; authenticated and confirmed to be contamination-free on 17.4.2019) were washed with PBS and then permeabilized with 40 µg/ml digitonin (5 min incubation time). After permeabilization, cells were washed with PBS three times for 5 min. Cells were then incubated with Ran mix (2 mM GTP, 0.1 mM ATP, 4 mM creatine phosphate, 20 U/ml creatine kinase, 5 µM RanGDP, 4 µM NTF2 and 1 mM DTT) for 1 h, followed by three washes with PBS for 5 min each. Upon this, 5 µM RanGDP-AlexaFluor 647 (DOL=1) or 5 µM RanQ69L-GDP–AlexaFluor 647 was added to cells for 1 h at room temperature. Depending on the assay, some samples underwent a triple washing step in PBS for 5 min each. Following this, permeabilized cells were immediately fixed with 4% formalin, stained with DAPI, mounted on the sample glass using Vectashield (Vector Labs) and imaged. For re-population assays of exogenous Kap, cells were not washed after exoRanGDP incubation but were directly incubated with exoKapβ1, exoKapα–Kapβ1 or PBS for 1 h at room temperature. Subsequently, the cells were washed and fixed except when energy mix (2 mM GTP, 0.1 mM ATP, 4 mM creatine phosphate, 20 U/ml creatine kinase) (Lowe et al., 2015) was added for 1 h, or PBS as a control. After a triple washing step in PBS for 5 min each the cells were also immediately fixed with 4% formalin and mounted on the microscope slide using Vectashield medium. In the NTF2 assay, a similar procedure was followed but with the addition of NTF2 into the energy mix. Endogenous proteins were detected using the following primary antibodies (all Abcam): Kapβ1 (ab2811, 1:200), Kapα (ab6036, 1:200), RanGAP1 (ab2081, 1:500), RanGEF (ab54600, 1:200) and RanBP2 (ab64276, 1:2000).

Fluorescence images were obtained at room temperature using an LSM700 upright confocal microscope with an oil-immersed 63×/1.4 NA PLAN APO objective and multialkali photomultiplier (PMT) detector type (Zeiss). Quantification of fluorescence intensity was performed using CellProfiler software (Kamentsky et al., 2011). DAPI staining was used to define a region of interest (ROI). To define the nuclear envelope, the DAPI ROI was reduced by five pixels and simultaneously expanded by five pixels, i.e. pixel size=0.04 μm×0.04 µm), yielding the ‘nuclear envelope ROI’; the ‘nucleus ROI’, therefore, is the area within the reduced ROI. Both regions were then used to quantify the mean fluorescence intensity of exoRan at the nuclear envelope and within the nucleus. In each case, when calculating the fraction of exoRan, the Ran fluorescence intensity was normalized to the signal obtained from the PBS-treated sample (control); the number of analyzed cells is specified in the figure legends.

RanGDP or RanQ69L-GDP binding to Kapβ1 was verified by microscale thermophoresis (MST). Increasing concentrations of RanGDP or RanQ69L-GDP were added to Kapβ1 (100 nM) labeled with maleimide conjugated to Alexa Fluor 488 (Kapβ1-488) (DOL=2). The final test solutions contained Kapβ1-488 (50 nM) and Ran at varying concentrations (0.00031 µM, 0.000625 µM, 0.00125 µM, 0.0025 µM, 0.005 µM, 0.01 µM, 0.019 µM, 0.039 µM, 0.078 µM, 0.156 µM, 0.3125 µM, 0.625 µM, 1.25 µM, 2.5 µM, 5 µM or 10 µM). These solutions were loaded into glass capillaries (Monolith NT.115 capillary, standard treatment MO-K002, NanoTemper) and the change of normalized fluorescence (‰ F_norm) was measured (60% laser power, 100% LED power, 30 s laser on/10 s laser off) using Nanotemper Monolith NT.115 (NanoTemper).

A Synergy H1 Hybrid Multi-Mode Monochromator Fluorescence Microplate Reader (BioTek) was used to measure the enzymatic activity of RanGEF to exchange RanGDP to RanGTP (Klebe et al., 1995a). 1 µl of 3 µM RanGEF (final concentration: 30 nM) was added to 100 µl of 5 µM RanGDP (WT or Q69 L mutant) and 200 µM N-methylanthraniloyl-tagged GTP (Mant-GTP; Sigma-Aldrich) in PBS at 25°C. As control, we repeated the experiments without RanGEF in order to monitor the non-catalyzed rate of the GDP-GTP exchange. The fluorescent signal was monitored at two wavelengths: 335 nm (excitation at 292 nm; internal tryptophan fluorescence) and 450 nm (excitation at 370 nm; Mant nucleotide fluorescence). Fluorescence change was normalized to control experiments carried out without RanGEF.

We thank Mikel Ghelfi for technical assistance.