Efficiency of Importin α/β-Mediated Nuclear Localization Sequence Recognition and Nuclear Import

Little quantitative, kinetic information is available with respect to the process of nuclear import of conventional nuclear localization sequence (NLS)-containing proteins, which initially involves recognition and docking at the nuclear pore by importin α/β. This study compares the binding and nuclear import properties of mouse (m) and yeast (y) importin (IMP) subunits with respect to the NLSs from the SV40 large tumor antigen (T-ag), and theXenopus laevis phosphoprotein N1N2. m- and y-IMPα recognized both NLSs, with y-IMPα exhibiting higher affinity. m-IMPβ greatly enhanced the binding of m-IMPα to the T-ag and N1N2 NLSs, but y-IMPβ did not significantly affect the affinity of y-IMPα for the T-ag NLS. In contrast, y-IMPβ enhanced y-IMPα binding to the NLS of N1N2, but to a lesser extent than the enhancement of m-IMPα binding by m-IMPβ. NLS-dependent nuclear import was reconstituted in vitro using the different importin subunits together with the transport factors Ran and NTF2. Whereas T-ag NLS-mediated nuclear import did not exhibit an absolute requirement for NTF2, N1N2 NLS-mediated transport strictly required NTF2. High levels of NTF2 inhibited nuclear accumulation conferred by both NLSs. We conclude that different NLSs possess distinct nuclear import properties due to differences in recognition by importin and requirements for NTF2.

plasmic RanGDP (19,20), the complex is translocated through the NPC (21) and importin ␣ and the NLS-containing protein released into the nucleus through the action of RanGTP (19). Alternative signal-mediated nuclear import pathways independent of the importin ␣/␤ heterodimer have recently been identified where either importin ␤ (22,23) or importin ␤-related homologs (24 -26) appear to fulfil the role of both importin ␣ and ␤ in binding NLSs and targeting them to the NPC (24,25,27).
Although NLS receptors from different species share structural and functional homology, the recognition of different types of NLS by importin subunits and its importance in the nuclear import process has not been adequately investigated in quantitative terms (28,29). In this study, we compare the NLS binding and transport properties of mouse and yeast importin subunits. We find that although mouse (m-IMP␣) and yeast (y-IMP␣) importin ␣s can bind different types of NLS, y-IMP␣ binds with higher affinity. y-IMP␤ does not significantly affect the affinity of y-IMP␣ for the NLS of the simian virus SV40 large tumor antigen (T-ag) but does enhance binding to the NLS of the Xenopus laevis phosphoprotein N1N2. In contrast, m-IMP␣ requires m-IMP␤ for high affinity binding. Nuclear import of a T-ag-NLS containing fusion protein in vitro was found to be enhanced by NTF2 but did not require it absolutely, whereas import of an N1N2 NLS-containing fusion protein had an absolute requirement for NTF2. Surprisingly, high NTF2 concentrations inhibited nuclear import in all cases. The results indicate that different types of NLS have different requirements with respect to the factors mediating nuclear import, and imply a differential, regulatory role for NTF2 in nuclear transport.
ELISA-based Binding Assay-An ELISA-based binding assay (28,34,35,38) was used to examine the affinity of binding between importin subunits (with and without GST moieties) and NLS-containing proteins. The latter were coated onto 96-well microtiter plates, incubated with increasing concentrations of importin subunits, and bound importin-GST detected using a goat anti-GST primary antibody, an alkaline phosphatase-coupled rabbit anti-goat secondary antibody, and the substrate p-nitrophenyl phosphate (34). In experiments where importin ␣/␤ combinations were tested, importin ␣-GST and thrombin-cleaved GSTfree importin ␤s were used. Absorbance measurements at A 405 nm were performed over a 90-min period using a plate reader (Molecular Devices), with values corrected by subtracting absorbance both at 0 min and in wells incubated without importin (34). Quantitation was also performed in identical fashion for ␤-galactosidase, and these values subtracted from those for the T-ag and N1N2-␤ galactosidase fusion proteins (28,34,35). All fusion proteins were also subjected to a parallel ␤-galactosidase ELISA (34) to correct for any differences in coating efficiencies and enable a true estimate of bound importin (28,34,35,38).
In Vitro Nuclear Transport-Nuclear import kinetics were measured at the single cell level using mechanically perforated HTC rat hepatoma cells in conjunction with confocal laser scanning microscopy as previously (31,34,35,38). Experiments were performed for 40 min at room temperature in a 5 l volume containing 30 mg/ml bovine serum albumin, 2 mM GTP, an ATP regenerating system (0.125 mg/ml creatine kinase, 30 mM creatine phosphate, 2 mM ATP), and transport substrate (0.2 mg/ml 5-iodoacetamidofluorescein-labeled fusion protein), or a control (70-kDa fluorescein isothiocyanate-labeled dextran; Sigma) to assess nuclear integrity. Where indicated, 4 M RanGDP, 1 M m-IMP␤ or y-IMP␤, 1 M m-IMP␣ or y-IMP␣, and NTF2 (0.04 -6 M) were added. Image analysis and curve-fitting were performed as described (34,35,38); the level of accumulation at the nuclear envelope, relative to medium fluorescence, was measured using NIH Image 1.60 in the line plot mode, as previously (32,39).

y-IMP␣ Has Higher NLS Binding Affinity Than m-IMP␣-
We have used an ELISA-based assay to determine the binding affinities of mouse importin subunits for different NLSs (28,34,35,40), with the results revealing relatively low NLS-binding affinity for m-IMP␣ compared with that for the m-IMP␣/␤ heterodimer (28,34). We set out to compare the NLS binding affinity of m-IMP␣, y-IMP␣, m-IMP␣/␤, and y-IMP␣/␤ for proteins containing either the T-ag NLS, or the bipartite NLS of the X. laevis nuclear factor N1N2 (see Table I). y-IMP␣ bound these fusion proteins with significantly (p Ͻ 0.05) higher affinity (about 2-fold) than m-IMP␣ (Figs. 1 and 2; see Table II), the latter only attaining high affinity in the presence of m-IMP␤. The specificity of binding in all cases was indicated by the fact that binding to NLS mutant derivatives was severely reduced (Table II). The T-ag NLS mutant showed negligible binding affinity for IMP␣/␤ from either mouse or yeast, whereas the two N1N2 NLS mutant derivatives showed different importin binding affinity (Fig. 2). N1N2A-␤-Gal, mutated in the proximal arm of the bipartite NLS, showed apparent dissociation constants (K d values) of 48 and 35 nM for y-IMP␣/␤ and m-IMP␣/␤, respectively, whereas N1N2A-␤-Gal, mutated in the distal arm, exhibited negligible binding. Clearly, mouse and yeast IMP␣/␤ are able to bind both types of NLS specifically, with y-IMP␣ binding with higher affinity than m-IMP␣ (Figs. 1 and 2; Table II).  Table II. UD, K d unable to be determined due to low binding.

TABLE I NLS sequences contained in the proteins used in this study
Fusion proteins contain the indicated amino acid sequences fused amino-terminal to E. coli ␤-galactosidase (amino acid 9 -1023) (see "Materials and Methods" and Refs. 30 -32). The single letter amino acid code is used with small letters indicating variation from the wild type sequences (33).

TABLE II Binding parameters of NLS-containing ␤-gatactosidase fusion proteins for yeast and mouse importins as measured using an ELISA-based assay
Sequence details are presented in Table I. Data represent the mean Ϯ S.E. (number in parentheses) for the apparent dissociation constant (K d ) determined as outlined under "Materials and Methods" (28,34).  Table  II.
Interestingly, the affinity of y-IMP␣ after precomplexation to y-IMP␤ for the T-ag NLS was not significantly different from that in its absence ( Fig. 1 and Table II), but was enhanced 2-fold (p Ͻ 0.02) in the case of the N1N2 NLS ( Fig. 2; Table II). This was in contrast to the significantly larger effect of m-IMP␤ on NLS recognition by m-IMP␣ (p Ͻ 0.03), where the presence of m-IMP␤ increased the binding affinity 2-and 4-fold for the T-ag and N1N2 NLSs respectively (Table II). Clearly, y-IMP ␣ is a higher affinity NLS receptor than m-IMP␣, which requires m-IMP␤ to achieve high affinity. That y-IMP␤ increased the binding affinity of m-IMP␣ for the bipartite N1N2 NLS but not for the T-ag NLS implies that different NLSs have different requirements to achieve high affinity recognition by importin.
Nuclear Import Conferred by the T-ag NLS Does not Require NTF2-The ability of m-IMP␣/␤ and y-IMP␣/␤ to mediate nuclear import in a reconstituted in vitro system was compared (Fig. 3). In the presence of RanGDP, y-or m-IMP␣ or ␤ alone could not mediate nuclear import of T-ag-CcN-␤-Gal (data not shown) with the levels of nuclear accumulation similar to those in the absence of importin subunits (Fig. 3, A-C and Table III). This was in contrast to the combination of m-IMP␣/␤ where maximal nuclear accumulation relative to that in the cytoplasm (Fn/c max ) was about 5-fold, half-maximal levels (t 1/2 ) being achieved within 5.3 min. In the presence of NTF2, the nuclear import rate was enhanced significantly (t 1/2 ϭ 2.6 min; p Ͻ 0.05), whereas Fn/c max remained unaffected (Fn/c max of 5.0; see Table III).
In the absence of exogenously added NTF2, y-IMP␣/␤ mediated nuclear import to levels (Fn/c max of 5.8; t 1/2 ϭ 4.9 min) only slightly higher than m-IMP␣/␤, while in the presence of NTF2, both the import rate and the level of accumulation (Fn/c max of 9.4; t 1/2 ϭ 1.8 min) were increased further (p Ͻ 0.05; see Table  III). The specificity of transport in all cases was demonstrated by the fact that the T-ag-Cc-␤-Gal protein, containing a nonfunctional NLS, did not accumulate to any significant extent (Fig. 3, A and C, and Table III). The results thus demonstrate that both m-IMP␣/␤ and y-IMP␣/␤ can mediate nuclear import of an T-ag NLS-containing transport substrate independent of  Table III. NTF2 ( Fig. 3 and Table III).
We also examined the ability of m-IMP␣/␤ and y-IMP␣/␤ to mediate binding to the nuclear envelope in the absence of RanGDP/NTF2 (Fig. 3A, top panel). No targeting of T-ag-CcN-␤-Gal to the nuclear envelope could be observed in the absence of exogenously added importin subunits (results not shown). Although m-IMP␣ or ␤ alone could not mediate binding, as expected, the combination of the two was able to do so with high efficiency (Fig. 3A). The maximal level of accumulation mediated by y-IMP␣/␤ at the nuclear envelope, as determined by image analysis using the line plot mode (39), was 4.6 Ϯ 0.6 times higher than that in the cytoplasm, significantly higher (p Ͻ 0.008) than that effected by m-IMP␣/␤ of 2.4 Ϯ 0.1 (n ϭ 4). This activity, together with its higher affinity compared with m-IMP␣/␤ for the T-ag-NLS (Table II), is presumably the basis of the ability of y-IMP␣/␤ to mediate nuclear import more efficiently than m-IMP␣/␤. Nuclear Import Conferred by the N1N2 NLS Requires NTF2-Nuclear import kinetic measurements were also performed using the bipartite NLS-containing substrate N1N2-␤-Gal ( Fig. 4 and Table III). As for T-ag-CcN-␤-Gal, nuclear accumulation was negligible in the absence of either RanGDP or m-IMP␣/␤. In the presence of m-IMP␣/␤ and RanGDP but in the absence of NTF2, a lack of nuclear accumulation was also evident, although nuclear accumulation was clearly observed in the presence of NTF2, with an Fn/c max of 3.3 and t 1/2 of 7.8 min.
Similar results were observed for y-IMP␣/␤ in the presence of Ran-GDP, no nuclear accumulation being observed in the absence of NTF2 ( Fig. 4; Table III); in its presence y-IMP␣/␤ yielded an Fn/c max of 4.2 (t 1/2 ϭ 4.1 min). Of the two N1N2-NLS mutant derivatives, N1N2A-␤-NLS was excluded from the nucleus, whereas N1N2B-␤-Gal showed reduced nuclear accumulation in the presence of RanGDP and NTF2 (Fn/c max of 2.8, t 1/2 ϭ 13.9 min and Fn/c max of 2.6, t 1/2 ϭ 8.4 min for y-IMP␣/␤ and m-IMP␣/␤, respectively) compared with wild type N1N2-␤-gal. Clearly, although both arms of the bipartite N1N2 NLS are necessary for optimal nuclear import, mutation of the proximal arm has a much greater effect on nuclear import (Fig. 4) through its more severe effect on importin binding (Fig. 2).

Inhibition of Nuclear Import by High Levels of NTF2-Pre-
vious studies have shown that NTF2 is necessary for nuclear import in several vertebrate systems, being able to bind directly to RanGDP, importin ␤ and proteins of the NPC (15,21,(41)(42)(43)(44). Because the T-ag and N1N2 NLSs appeared to have different requirements for NTF2 for nuclear import, we performed transport measurements in the presence of increasing concentrations of NTF2 (Fig. 5). Although NTF2 concentrations up to 0.15-0.3 M enhanced nuclear import of the T-ag NLScontaining fusion protein in the case of both m-IMP␣/␤ and y-IMP␣/␤, higher concentrations inhibited transport strongly (Fig. 5A). Optimal nuclear import efficiency appeared to be obtained with an NTF2 concentration of about 0.08 M.
Similar results were also observed for N1N2 NLS-mediated nuclear import (Fig. 5B), whereby m-IMP␣/␤ or y-IMP␣/␤ mediated import was enhanced in the presence of NTF2 concentrations up to 0.3 and 0.6 M, respectively, with the optimal concentration of NTF2 for N1N2 NLS-mediated nuclear accumulation being about 0.15 M. Nuclear accumulation was inhibited essentially completely by NTF2 concentrations above 0.6 M. NTF2 thus appears to play a regulatory role in nuclear import, its concentration determining whether it has enhancing or inhibitory effects. DISCUSSION This study represents the first quantitative comparison of two conventional NLS binding receptors and two different NLSs, and their ability to mediate nuclear protein import in a system using purified components. Although in vitro systems for nuclear transport have existed since 1990 (4), surprisingly little kinetic analysis has been performed. True understanding of the subtleties of signal-dependent nuclear import requires rigorous quantitation and this study can be seen as a first step in this direction. One of our important findings is that the importin ␣ subunits from mouse and yeast appear to be different in their requirement for importin ␤ to achieve high affinity NLS binding. This is highlighted by the results for the T-ag NLS, where y-IMP␣ and y-IMP␣/␤ have almost identical binding affinities, in stark contrast to m-IMP␣, which requires m-IMP␤ to obtain high affinity binding. In the case of the N1N2 TABLE III Kinetics of nuclear import reconstituted in vitro using purified proteins Results represent the mean Ϯ S.E. from data (see Figs. 3C and 4C) fitted to the function Fn/c(t) ϭ Fn/c max *(1Ϫe Ϫkt ), where Fn/c max is the maximal level of nuclear accumulation, k is the rate constant, and t is time in minutes. NTF2 was used at the optimal concentration for nuclear accumulation (see Fig. 5).
Nuclear Import Parameter
NLS, y-IMP␤ does enhance y-IMP␣ binding, but not to such a great extent as the enhancement effected on m-IMP␣ binding by m-IMP␤. y-IMP␣ thus appears to resemble importin ␣ from the plant Arabidopsis thalania to some extent, the latter being able to bind a variety of NLS types in the absence of importin ␤ (32). There would thus appear to be a gradation through evolution in the properties of importin ␣, the vertebrate form being much more strongly dependent on the additional presence of importin ␤ to achieve high affinity NLS binding.
As shown here, both y-IMP␣/␤ and m-IMP␣/␤ can mediate the nuclear import of proteins carrying the T-ag or N1N2 NLSs in combination with human Ran and NTF2, but the role of NTF2 in nuclear import of T-ag is different from that of N1N2. Nuclear import mediated by the T-ag NLS does not require NTF2 absolutely, whereas nuclear accumulation in the case of the N1N2 NLS is completely dependent on NTF2. Varying the concentration of NTF2 changes nuclear import dramatically, high levels of NTF2 inhibiting nuclear accumulation mediated by both the T-ag and N1N2 NLSs. This is consistent with the observations of Tachibana et al. (45), who found that cytoplasmically injected NTF2 strongly inhibited nuclear import mediated by the T-ag NLS, as well as nuclear export mediated by the nuclear export sequence of the cAMP-dependent protein kinase inhibitor PKI, in a dose-dependent manner. The basis of this may relate to the reported ability of NTF2 to regulate the functional size of the central transporter element of the NPC (46), as well as through its ability to associate with RanGDP, importin ␤, and nucleoporins including p62 (14,15,41,43,44,47), which itself is located close to the central channel of the NPC (48). The present study shows that the nuclear import process is finely balanced, with an individual NLS possessing its own particular properties and requirements with respect to the different nuclear transport components to be translocated to the nucleus with optimal efficiency. In this regard, NTF2 would appear to play a central regulatory role (42,(45)(46)(47)49).
Our mutagenic analysis clearly shows that both arms of the N1N2 bipartite NLS are necessary for high affinity binding by importin ␣/␤ as well as efficient nuclear targeting. Mutagenesis of the proximal arm of the NLS reduces importin binding and nuclear import, whereas mutation of the distal arm completely oblates both activities. Essentially similar results have been obtained for the nucleoplasmin NLS (50), our results thus confirming the bipartite nature of the N1N2 NLS (37). From our results here and elsewhere (28,40), the importin binding and in vitro nuclear import of bipartite NLS-containing proteins appear to be intrinsically less efficient than those of proteins containing the T-ag NLS. Although T-ag may be rather special in terms of being a viral protein that uses the host cell transport system with unusually high efficiency, it remains unclear as to whether as yet unidentified transport factors may be involved in importin-mediated nuclear import of proteins carrying bipartite NLSs under physiological conditions. Investigation of this possibility is currently in progress in this laboratory.
The results presented here indicate a close relationship between the importin binding affinity and rate of nuclear import, where the highest affinity conformation of NLS and NLS receptor leads to the most efficient nuclear accumulation. This is illustrated most poignantly by the two different mutant variants of the N1N2 NLS, which exhibit quite different importin binding activities; the proximal arm mutant (the N1N2B-␤-Gal protein) is recognized by importins with higher affinity, and in consequence is able to be transported to the nucleus much more efficiently than the distal arm mutant (N1N2A-␤-Gal). NLS binding by importins is clearly a crucial step in nuclear import, consistent with the idea that the affinity of NLS recognition by importins is critical in determining the rate and maximal extent of NLS-dependent nuclear import (28,34,35,40).