Characterization of an Unusual Importin α Binding Motif in the Borna Disease Virus p10 Protein That Directs Nuclear Import*

Nuclear import of many cellular and viral proteins is mediated by short nuclear localization signals (NLS) that are recognized by intracellular receptor proteins belonging to the importin/karyopherin α and β families. The primary structure of NLS is not well defined, but most contain at least three basic amino acids and harbor the relative consensus sequence K(K/R)X(K/R). We have studied the nuclear import of the Borna disease virus p10 protein that lacks a canonical oligobasic NLS. It is shown that the p10 protein exhibits all characteristics of an actively transported molecule in digitonin-permeabilized cells. Import activity was found to reside in the 20 N-terminal p10 amino acids that are devoid of an NLS consensus motif. Unexpectedly, p10-dependent import was blocked by a peptide inhibitor of importin α-dependent nuclear translocation, and the transport activity of the p10 N-terminal domain was shown to correlate with the ability to bind to importin α. These findings suggest that nuclear import of the Borna disease virus p10 protein occurs through a nonconventional karyophilic signal and highlight that the cellular importin α NLS receptor proteins can recognize nuclear targeting signals that substantially deviate from the consensus sequence.

The transport of many cellular and viral macromolecules into the nucleus is mediated by soluble cytosolic receptor proteins that recognize and bind nuclear localization signals (NLS) 1 present on karyophilic proteins (for recent reviews on nuclear import of proteins see Refs. [1][2][3][4]. NLSs are characterized by short, single, or bipartite stretches of basic amino acids (5)(6)(7). The prototypes of such topogenic sequences were identified as 126 PKKKRKV 132 in the T antigen of the simian virus 40 (SV40) and as 155 KRPAATKKAGQAKKKK 170 in the cellular nucleoplasmin. Despite their oligobasic character, there is con-siderable variation in the primary sequences of these "classical" NLSs, and the context in which these signals are embedded is also important for their function (8 -10). Therefore, only a relative consensus motif has been formulated as "K(K/R)X-(K/R)" for monopartite NLSs since they usually contain at least three basic amino acids (9).
The nuclear import of proteins containing classical NLSs is mediated by binding to the cytosolic 60-kDa NLS receptor protein importin ␣ (also called karyopherin ␣ among other names, see Refs. [11][12][13] from which six isoforms have been identified in humans (14). Importin ␣ heterodimerizes through its N-terminal domain with the 97-kDa protein importin ␤ (also known as karyopherin ␤) whereby its affinity for the NLS is increased (15)(16)(17)(18). The receptor-cargo complex first binds through importin ␤ to components of the nuclear pore complex (NPC) and is then transported through the nuclear envelope (19,20). Mechanistically, the translocation process has been proposed to be driven either by sequential interactions of increasing affinity of importin ␤ with different nucleoporins along the central channel of the NPC or, alternatively, by the hypothetical ability of transport receptors to penetrate a hydrophobic meshwork composed by nucleoporins (21,22). Nuclear import is completed by binding of the Ras-related nuclear protein in its GTP-bound form to importin ␤ which results in the disassembly of the receptor cargo complex in the nucleus (16,23).
Several recent studies (summarized in Ref. 1) have shown that nuclear import of various proteins is mediated via an importin ␣-independent mechanism. The human immunodeficiency virus type 1 Rev and Tat proteins exemplify factors that interact directly with importin ␤ via a highly basic importin ␤ binding domain which then mediates import without the necessity of the NLS receptor subunit (24,25). Moreover, a karyophilic signal has been described for import of heterogeneous nuclear ribonucleoprotein (hnRNP) A1 that is imported through interaction with the importin ␤ homologue transportin (26). The recognition site for transportin in hnRNP A1 is the glycine-rich 38-amino acid M9 domain (27,28). Interestingly, a given nuclear protein may not depend on a single transport receptor as suggested by the finding that import of some ribosomal proteins is mediated by binding to either importin ␤ or the importin ␤-like receptors transportin, RanBP5, or RanBP7 through very basic ␤-like import receptor binding domains (29). Finally, other nuclear proteins have been identified that bear nuclear targeting sequences that do not resemble any of the signals mentioned above (30,31). It is currently not known if this indicates the existence of alternative import receptors that remain to be discovered or if this reflects a high variability of the sequences that can be recognized by the characterized importins (32).
The p10 protein is the most recently identified distinct gene product of the Borna Disease Virus (BDV) that is unique among the non-segmented negative strand RNA viruses (order Mononegavirales) of animals and man, because it replicates in the nuclear compartment (33,34). The precise function of this 10-kDa protein in the viral replication cycle is currently unknown. However, we have recently demonstrated (35,36) that the p10 protein interacts with and is localized in the nuclear compartment of infected cells together with the viral N and P proteins that are expected to be essential components of the viral replicative ribonucleoprotein (RNP) complex. Conceivable roles of the p10 protein that require its nuclear import may therefore involve regulatory functions during viral RNA synthesis, splicing, and/or the intracellular trafficking of viral RNP. Because the primary sequence of the p10 protein lacks a recognizable nuclear targeting signal, it was proposed previously (37)(38)(39) that the p10 protein enters the nucleus only in the presence of the viral karyophilic N or P proteins.
However, here we demonstrate that the p10 protein exhibits all the characteristics of an actively transported karyophilic protein in digitonin-permeabilized cells, which indicates that it carries an autonomous nuclear targeting signal. Import activity was found to reside within the 20 N-terminal p10 amino acids ( 1 MSSDLRLTLLELVRRLNGNG 20 ) and, surprisingly, to correlate with the ability of this domain to bind directly to the NLS receptor protein importin ␣. Di-alanine scanning identified several basic, hydrophobic, and polar amino acids within this domain as critical for this interaction. These results suggest that import of p10 occurs via the importin ␣/␤-mediated pathway despite the unusual targeting signal.

EXPERIMENTAL PROCEDURES
Maintenance of Tissue Culture Cells-A human oligodendrocyte cell line ("Oligo") and human Huh-7 cells were maintained and passaged in Dulbecco's modified Eagle's tissue culture medium (DMEM) containing 10% fetal calf serum (FCS) and 2 mM L-glutamine.
Intracellular Localization of Expressed BDV p10 Protein-Solitary p10 protein was expressed by transfection of 5 g of pcDNA-p10 in oligodendrocyte cells using LipofectAMINE 2000 reagent (Invitrogen). The transfected cells were seeded on glass coverslips in DMEM, 10% FCS and were incubated at 37°C and 5% CO 2 . Cells were processed for analysis by confocal laser scanning microscopy after 24 h through fixation in 2.5% methanol-free formaldehyde (Polysciences Inc., Warrington, PA), and permeabilization of cells in 0.2% Triton X-100 was done as described (42). Cells were stained with rat anti-GST-p10 serum (36) diluted in PBS, 3% BSA (1:400). The cells were washed and incubated with FITC-conjugated donkey anti-rat IgG (Jackson Immu-noResearch Laboratories Inc., West Grove, PA). Subsequently, the samples were mounted in MOWIOL 4-88 and analyzed by using a 63ϫ/1.2 objective and a Zeiss LSM 510 confocal imaging system.
Preparation of Import Substrates-The p10 wild type and mutant proteins were expressed as glutathione S-transferase (GST) fusion proteins in E. coli and purified by affinity chromatography on glutathioneagarose (36). The fusion proteins were eluted from the resin by incuba-tion with elution buffer (30 mM glutathione, 100 mM Tris-HCl, pH 8.0, 2% n-octyl glucoside) and dialyzed against Na 2 CO 3 . For direct fluorescence detection, the GST proteins were reacted with FluoroLink Cy3 reagent (Amersham Biosciences) and were subsequently purified by gel chromatography on a PD10 column (Amersham Biosciences AB) and concentrated in a Centricon 10 (Amicon Inc., Beverly, MA) ultrafiltration device. Preparation of FITC-labeled BSA-conjugated to SV40 T antigen (Tag) NLS-derived peptide has been described elsewhere (43).
In Vitro Transport Assays-Cells were grown in DMEM, 7% FCS on collagen-(Sigma) or Cell-Tak (Becton Dickinson)-coated 12-mm coverslips in 24-well dishes at 37°C and 5% CO 2 . The cells were washed with DMEM and permeabilized by treatment with 80 g/ml digitonin (Sigma) in DMEM for 10 min at 37°C. They were washed three times for 10 min at 4°C in washing buffer (transport buffer, 2 mM magnesium acetate, 20 mM Hepes, pH 7.3, 110 mM potassium acetate, 1 mM EGTA, 5 mM sodium acetate, 1 mM dithiothreitol containing 1% BSA and 10% goat serum), followed by a 10-min incubation at 37°C in washing buffer. Preincubation of permeabilized cells with 50 g/ml wheat germ agglutinin (Roche Molecular Biochemicals) in transport buffer was performed during this step. Permeabilized cells were incubated with similar amounts of fluorescently labeled protein (all showing approximately the same protein:fluorophore ratio) in 20 l of transport buffer containing 24 mg/ml rabbit reticulocyte lysate (Promega, Madison, WI), 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 g/ml pepstatin (all from Sigma) for 20 min at 37°C. For competition experiments, peptides derived from the SV40 T antigen (PKKKRKVED) or the hnRNP A1 M9 domain (YNNQSSNFGPMK) were added to final concentrations up to 1 mM. In these experiments, the FITC-BSA SV40 Tag-NLS and FITC-BSA M9 conjugates were used in a final concentration of 10 g/ml as internal controls. If required, an ATP-generating system (1 mM ATP, 5 mM creatine phosphate, and 20 units/ml creatine phosphokinase (Sigma)) or an ATP/GTP-depleting system (7 mM glucose, 1 units/ml hexokinase (Sigma)) was added (44). Cells were fixed for microscopic analysis on the coverslips using 3% paraformaldehyde in PBS and permeabilization in PBS, 0.1% Triton X-100. After two washings in PBS, cells were exposed to an anti-NPC antibody (1:500; Babco, Richmond, CA) diluted in PBS, 10% goat serum, 1% BSA. Coverslips were washed in PBS and incubated with affinity-purified FITC-conjugated goat anti-rabbit IgG (Dianova, Hamburg, Germany) diluted in PBS, 10% goat serum, 1% BSA. Cells were washed and embedded in MOWIOL as described above. Confocal laser scanning microscopy was performed on a Leica DM IRBE microscope. Analysis of Cy3-labeled GST fusion proteins and peptide conjugates was done by using the TRITC and FITC fittings at a pinhole size of 1.
In Vitro Binding Assays-The BDV p10 protein or mutant derivatives thereof were expressed from pGEX-p10 plasmids as GST fusion proteins in E. coli BL26. Synthesis of GST fusion proteins was induced by addition of 0.5 mM isopropyl-␤-D-galactopyranoside. Bacterial cell lysates containing GST fusion proteins were adsorbed to glutathione-Sepharose (Amersham Biosciences AB) according to the protocol supplied by the manufacturer, and contaminating proteins were removed by three washes with PBS. Full-length and truncated (amino acids 65-529) human importin ␣1 (also called hSRP1␣) were synthesized and labeled with [ 35 S]methionine in coupled 100 l of TNT transcription/ translation reactions (Promega) programmed with pRSET-hSRP1␣ and pKW313, respectively (kind gifts of K. Weis, University of California, San Francisco). Aliquots of the translation reactions or 1 g of purified recombinant importin ␣1 (a kind gift of D. Görlich, University Heidelberg, Germany) were incubated overnight at 4°C with 20 l of coated glutathione-Sepharose beads in 1 ml of binding buffer (50 mM Hepes-NaOH, pH 8.0, 100 mM NaCl, 0.05% Nonidet P-40, 0.1% BSA). The beads were washed three times with binding buffer, and the precipitated proteins were separated by SDS-gel electrophoresis and visualized by autoradiography. Recombinant importin ␣1 was detected by immunoblotting using affinity-purified rabbit importin ␣-specific antibodies (a kind gift of M. Köhler, Max-Delbrü ck Centrum, Berlin, Germany) and secondary horseradish peroxidase-conjugated swine anti-rabbit IgG antibody (Dako Diagnostika, Hamburg, Germany). Quantification of precipitated proteins was performed with a Fujifilm FLA-2000 Bioimaging System and the Image Gauge version 3.0 software (Fuji Photo Film Co, Ltd.).
Co-immunoprecipitation Analysis-BDV p10-Myc protein was expressed in oligodendrocyte cells by transfection with 4 g of pcDNA-p10-Myc plasmid per 35-mm dish using LipofectAMINE 2000 reagent. Cytoplasmic extracts of transfected and non-transfected cells were prepared after 24 h by incubation in hypotonic buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 0.5 mM Pefabloc (Roche Molecular Biochemicals)) for 15 min on ice followed by 10 passages through a syringe attached to a 26-gauge needle. The nuclei in the homogenate were pelleted by a 5-min centrifugation at 5,000 rpm and 4°C in a tabletop centrifuge. The supernatant was cleared by another centrifugation at 14,000 rpm and was adjusted to 100 mM NaCl and 100 g/ml BSA for use in subsequent experiments. Extracts equivalent to 5 ϫ 10 6 transfected and non-transfected cells, respectively, were incubated with the Myc tag-specific monoclonal antibody 9E10 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at 0°C. Immune complexes were collected by incubation with 20 l of protein G-agarose beads (Roche Molecular Biochemicals). The beads were washed three times with binding buffer, and the precipitated proteins were analyzed by immunoblotting using primary anti-importin ␣1specific rabbit antibodies as described above.

RESULTS
The BDV p10 Protein Enters the Nucleus in the Absence of Other Viral Gene Products-We have demonstrated previously that the BDV p10 protein localizes to the nucleus in persistently BDV-infected cells (35). Based on the absence of a recognizable NLS and its interaction with the karyophilic viral P protein, nuclear entry of p10 has been suggested previously to rely on other viral proteins (38,45). However, when expressed in the absence of other viral gene products, we detected the p10 protein both in the nucleus and the cytoplasm (Fig. 1). Thus, it was demonstrated that the p10 protein can enter the nucleus without the requirement for other virus-encoded proteins. However, this approach did not distinguish whether nuclear entry of p10 was facilitated by other factors or if it occurred by diffusion because its molecular weight is below the 40 -50-kDa exclusion limit of the nuclear pore (46).
The N-terminal Domain of the p10 Protein Mediates Active Nuclear Import in Permeabilized Cells-To determine whether the BDV p10 protein contains a signal(s) that mediates nuclear import, we studied its karyophilic properties in digitonin-permeabilized mammalian cells, an established system for studying nuclear transport of proteins (44). The p10 protein was expressed as a GST fusion protein in E. coli and was purified by affinity chromatography to near-homogeneity. For direct fluorescence detection, we covalently coupled GST proteins to a Cyanine dye 3 (Cy3) fluorophore. Fluorescently Cy3-labeled p10 fusion protein was incubated with the permeabilized cells at 37°C in the presence of cytosolic extract and an ATP energy source. The cells were subsequently stained with an anti-nuclear pore complex antibody (43) and examined by confocal microscopy (Fig. 2). The p10 fusion protein accumulated in the nucleus within 20 min (Fig. 2, left panel, column GST-p10 wt). This nuclear accumulation was specific because GST alone was not appreciably imported under identical conditions (Fig. 2, column GST). Import of GST-p10 did not occur in the absence of cytosolic factors or under concomitant incubation with an ATP-depleting system or at 0°C (Fig. 2, right panels, columns Ϫcytosol, ϪATP, and ϩATP, 0°C). In addition, nuclear accumulation of p10 was inhibited after preincubation of the permeabilized cells with the lectin wheat germ agglutinin (Fig. 2, column ϩATPϩWGA), which is known to interfere with transport through the NPC (47). These findings imply that nuclear migration of p10 depends on soluble cytosolic factors and the presence of energy for recycling of these factors. They therefore rule out a merely diffusion-driven mechanism of combined nuclear entry and retention and suggest that nuclear migration of the p10 protein occurs by an active transport mechanism through the nuclear pores.
To determine the karyophilic signal within p10, we prepared Cy3-labeled GST fusion proteins containing BDV amino acids 1-20, 21-87, and 41-87. It was found that p10 amino acids 1-20 could direct nuclear import, whereas fusion proteins carrying amino acids 21-87 or 41-87 did not (Fig. 2, columns GST-p10 21-87, GST-p10 1-20, and data not shown). Thus, the p10 protein harbors a karyophilic signal within its 20 N-terminal amino acids ( 1 MSSDLRLTLLELVRRLNGNG 20 ). The activity of this domain in nuclear transport was unexpected, because the corresponding sequence does not contain a consensus motif of mono-or bipartite oligobasic NLSs nor does it resemble other known nuclear targeting signals like the M9 sequence. This suggested that the p10 protein either interacts with known transport receptors through an unusual signal or that import of p10 is mediated by a novel cytosolic factor(s) that recognizes a previously unknown karyophilic signal.
Nuclear Import of the p10 Protein Can Be Competed by a Peptide Derived from the SV40 T Antigen-NLS-We wished to determine whether import of the p10 protein follows an established or a previously unrecognized pathway. The best characterized nuclear import pathway of proteins is mediated via the importin ␣/␤ heterodimer. Because transport via this route is a saturable process, we used a competition approach to examine if import of p10 depended on the same factors as of proteins carrying oligobasic NLS. These experiments revealed that transport of both recombinant p10 protein and a conjugate of bovine serum albumin with a peptide derived from the classical SV40 Tag-NLS was concomitantly inhibited after preincubation of the cytosolic factors with SV40 Tag-NLS peptide up to 1 mM (Fig. 3). In contrast, the transport of neither substrate was inhibited when 1 mM of a peptide derived from the M9 domain of hnRNP A1 protein was added to the extract (Fig. 3). This indicated that nuclear transport of the p10 protein is mediated by the importin ␣ and ␤ proteins despite the low similarity of the p10 signal with oligobasic NLSs.
The N-terminal Domain of the p10 Protein Interacts Directly with Importin ␣-If import of the p10 protein occurred through the importin ␣-dependent pathway, it was expected to bind to the cellular NLS receptor protein. To examine this prediction, we immobilized GST-p10 fusion protein on glutathione-agarose beads and analyzed for binding to importin ␣1 (also termed Rch1 and hSRP1␣) that was synthesized and labeled by translation in the presence of [ 35 S]methionine in vitro. Analysis of precipitated proteins by gel electrophoresis showed that the p10 protein interacted with importin ␣ (Fig. 4A, lanes GST and  p10 WT). The specificity was demonstrated by the low background levels of importin ␣ precipitation when GST was used. Binding of importin ␣ to p10 occurred at a similar level as with the human cytomegalovirus UL56 protein that was recently shown to interact with importin ␣ through an oligobasic NLS (Fig. 4A, lane UL56c (40)). When truncated p10 proteins were tested, we found that p10 amino acids 1-20 precipitated importin ␣1 from solution, which parallels its karyophilic behavior in the import assays (Fig. 4A, lane p10 1-20). In contrast, the residual portion of p10 (amino acids 21-87) failed to interact (Fig. 4A, lane p10 21-87). Identical results were obtained when we used for precipitation a truncated importin ␣ that because of a deletion of amino acids 1-65 was unable to bind to importin ␤ that is present in the reticulocyte lysates (Fig. 4B) (15,18). Thus, the previously demonstrated activity of the N-terminal p10 region to mediate nuclear import correlated with the ability to interact with the NLS receptor protein importin ␣.
The detection of a specific p10-importin ␣ interaction by GST pull-down analysis raised the question whether such complexes also form in cells. Thus, we examined cytoplasmic extracts from transfected cells that expressed a Myc-tagged p10 protein by co-immunoprecipitation analysis. Fig. 4C shows that importin ␣ was specifically co-precipitated with a tag-specific antibody from extract of p10-Myc-expressing cells but not of non-transfected cells. This result strongly suggests that the p10 protein associates with the cellular NLS receptor protein in vitro and in vivo. The previous experiments did not distinguish if p10 binds to importin ␣ directly or if this interaction is facilitated by a third factor. However, in a further pull-down experiment it was shown also that affinity-purified recombinant importin ␣1 was specifically precipitated by GST-p10 (Fig. 4D). This demonstrates that binding of p10 to importin ␣ is direct and does not depend on any other protein present in reticulocyte or cytoplasmic lysates that were used in the previous experiments. The correlation that transport activity and the ability to bind importin ␣ both reside in the N-terminal p10 domain provided strong evidence that import of the BDV p10 protein is mediated by the importin ␣/␤-dependent pathway.
Mutational Analysis of the p10 N-terminal Domain Correlates Transport Activity with Binding to Importin ␣-The precipitation assays suggested that p10 amino acids 1-20 contain  -p10 wt 1-87, GST-p10 1-20, and GST-p10 21-87) were purified from bacterial lysates and fluorescently labeled with Cy3. Cells were permeabilized with digitonin and incubated with the labeled proteins in the presence of cytosolic extract and ATP at 37°C (columns to the left, cytosol ϩ ATP). The cells were processed for microscopic analysis, and the nuclear pore complexes were stained with a primary NPC-specific antibody and secondary FITC-conjugated goat anti-rabbit IgGs. Import of Cy3-labeled proteins was analyzed by confocal microscopy using the TRITC channel signal (upper row, Cy3). The NPCs in the same cells were detected using the FITC fitting (medium row, NPC). Computer-generated overlays of each of the two channels are shown in the bottom row (merge). In control reactions, nuclear import of wild type GST-p10 was assessed after preincubation of the cytosolic fraction with wheat germ agglutinin (ϩATPϩWGA) or after incubation at 0°C (ϩATP, 0°C) or when incubated at 37°C in the absence of either cytosolic factors (Ϫcytosol) or in the presence of an ATP-depleting system (ϪATP) as indicated below (columns to the right).
FIG. 3. Nuclear import of GST-p10 is inhibited by a peptide derived from the oligobasic SV40 Tag NLS. Cells were permeabilized by treatment with digitonin and incubated at 37°C with Cy3-labeled GST-p10 (shown in red), an FITC-labeled conjugate of BSA and Tag-NLS peptide (shown in green), and energy-regenerating components and cytosolic factors (column positive control). For competition of import, reactions were complemented with unconjugated SV40 Tag NLS peptide at concentrations between 0.125 and 1 mM or with 1 mM M9 peptide as indicated on the top. In a control reaction, the energy-regenerating components were replaced by an ATP/GTP-consuming system (column negative control). The nuclear pore complexes were stained by a monoclonal antibody and a Cy5-conjugated secondary IgG (shown in blue). The analysis of the cells by confocal laser scanning microscopy is shown as overlays of the TRITC (fields in the lower row, "Cy3-GST-p10") or FITC (fields in the upper row, "SV40Tag NLS-FITC-BSA") channel signals, respectively, with the Cy5 signals (shown in blue).
an unusual importin ␣-binding motif. To characterize this signal in more detail, we prepared eight recombinant p10 derivatives, in each of which two amino acids between positions 3 and 20 were replaced by alanines. The quantitative analysis of these mutants in pull-down assays demonstrated that six of the eight di-alanine alterations significantly reduced binding to importin ␣ (Fig. 5A). Mutation of the Arg-14 and Arg-15 completely abolished binding, whereas the p10 proteins with mutations at positions 6/7, 8/9, 12/13, and 18/19 were less than 10% active as the wild type. The A16/A17 mutant showed a moderate activity of importin ␣ binding of about 20%. These results suggest that p10 binds to importin ␣ through a polypeptide stretch containing amino acids 6 -19. The two basic arginine residues at position 14 and 15 appear to be essential for the recognition of importin ␣, but other hydrophobic and polar amino acids up to eight positions adjacent to these basic residues also contribute critically to this binding.
By having identified specific amino acids in p10 that are important for recognition of importin ␣, we asked how mutation of these residues would affect p10-dependent transport. Initial immunofluorescence stainings of transiently transfected cells consistently revealed a partial but incomplete reduction of nuclear localization of p10 di-alanine mutant proteins in com-parison to the wild type (data not shown). This suggested that a fraction of the small mutant proteins can enter the nucleus in transfected cells by diffusion through the NPC or was passively included into the karyoplasms during cell division. However, a clearer picture emerged when we tested the p10 di-alanine mutant A6/A7 representing a highly reduced importin ␣-binding protein and p10 A10/A11 as a mutant with wild type-like activity for transport in permeabilized cells. As expected, the nuclear transport of p10 wild type and A10/A11 mutant proteins was indistinguishable (Fig. 5B). In contrast, the p10 A6/A7 mutant was not imported, although the cells were fully capable of supporting nuclear import of an BSA-M9-peptide conjugate in parallel (Fig. 5B). Thus, these results show that nuclear transport of the BDV p10 protein and binding to importin ␣ are intimately linked and establish the p10 N-terminal domain as a non-canonical karyophilic signal that functions through the importin ␣/␤-dependent import pathway. DISCUSSION The nuclear membrane segregates distinct functions within the eukaryotic cell, requiring sophisticated transport processes to ensure the presence of the correct enzymes in the compartment where they act. This observation is not only true for cellular proteins but also for proteins of those viruses that have to make use of the cellular transport machinery because they replicate inside the nucleus of non-dividing cells. Thus, BDV as such a virus had to develop a strategy that allows the genome and the viral proteins involved in replication and transcription to enter the nucleus via the nuclear pore complex.
The best characterized so-called classical nuclear import pathway involves the activity of importin (or karyopherin) ␣. Six human isoforms have been identified that, although they differ in their amino acid sequences up to about 50%, appear capable of mediating transport of proteins with oligobasic consensus-type NLSs (14). There is some evidence that isoforms can discriminate between particular NLSs, because it has been shown that proteins like RCC1 and STAT1 are selectively bound and transported by various importin ␣ isoforms (14,48). Thus, the variability among oligobasic NLSs in mammalian cells may in part be explained by the existence of several receptor proteins that are specialized for subgroups of signals. Little information is currently available about the range of sequences that can be utilized by each importin ␣ isoform, because only few discrete signals have been analyzed in detail for interaction with different importin ␣ proteins.
Due to missing consensus motifs of karyophilic signals, nuclear import of the BDV p10 protein was suggested previously (37,38,45) to depend on interaction with the viral P or N proteins that carry oligobasic NLSs. However, analysis by confocal laser scanning microscopy showed that the p10 protein can localize to the nucleus independently of other viral gene products and might therefore carry a karyophilic signal of its own. Others (37,38) have previously localized solitary p10 predominantly in the cytoplasm by conventional immunofluorescence techniques and may have underestimated the amounts of nuclear p10 protein.
Due to its low molecular mass of about 9.5 kDa that is below the 40 -50-kDa limit for diffusion through the NPCs (46), p10 could in principle passively enter the nucleus. However, analysis in digitonin-permeabilized cells revealed that translocation of p10 into the karyoplasm depended on the addition of cytosolic factors, and an ATP/GTP energy source required incubation at elevated temperature and was inhibited by the lectin wheat germ agglutinin. This shows that nuclear translocation of the p10 protein occurs via an active receptor-mediated and energy-dependent transport mechanism through the NPC as has been shown for classical import substrates such as The upper panel shows the detection of importin ␣ in the precipitate by immunoblot analysis; T, total lysate representing 2% of that used for immunoprecipitation. The position of the IgG heavy chain that was stained by the secondary antibody is indicated to the right. In the lower panel the p10-Myc protein was detected by immunoblotting in the lysates using a Myc tag-specific antibody. D, glutathione-agarose beads coated with GST or GST-p10 fusion protein were reacted with affinity-purified recombinant importin ␣1. Precipitated proteins were analyzed by SDS-gel electrophoresis and immunoblotting using rabbit antibodies specific for importin ␣1. T, 2.5% of the total amount used for precipitation.
the SV40 Tag (reviewed in Ref. 2). Although nuclear import of p10 would theoretically not require an active mechanism in vivo, these results are in accordance with the observation that small proteins below the exclusion limit of the nuclear envelope are usually actively directed into the nucleus, as exemplified by histone H1 (49). This does not exclude that in BDV-infected cells some p10 molecules may also be transported in association with other viral karyophilic proteins. In digitonin-treated cells, the presence of the viral P protein did not enhance the rate or level of p10 nuclear accumulation. 2 Due to the lack of a BDV reverse genetic system, it is currently difficult to assess the contribution of other virus proteins in p10 import in the context of a viral infection.
The analysis of truncated p10 mutants revealed that the 20 N-terminal amino acids are necessary and sufficient to confer transport activity. Significantly, this region is the most conserved part of the otherwise relatively variable sequences of the p10 proteins from different BDV strains, suggesting that it has an important function (50). This sequence contains only three dispersed basic arginine residues at positions 6, 14, and 15 and therefore does not match the consensus of classical NLSs. However, several lines of evidence strongly suggest that nuclear import of p10 is mediated by the importin ␣-dependent pathway. First, import of p10 and oligobasic NLS was concomitantly reduced by a specific peptide inhibitor of this pathway. Second, in cell extract we could demonstrate the presence of an immunoprecipitable complex that contains both importin ␣ and the p10 protein. Third, the p10 protein specifically precipitated importin ␣ from solution which did not require its importin ␤ binding domain or any other cytosolic factor. Fourth, the ability to bind importin ␣ and the import signal both coincide in the N-terminal domain suggesting that these are not independent phenomena but that binding causes transport.
How can the p10 protein interact with importin ␣ although its transport signal shows only a low degree of resemblance to conventional oligobasic NLS motifs? The concomitant inhibition of p10-and SV40Tag-NLS-import by increasing NLS-peptide concentrations suggests that both substrates bind to the same or overlapping sites within importin ␣. Recent crystallographic studies of yeast and one mammalian importin ␣ complexed with NLS peptides identified a major cluster of six binding pockets termed P 1 to P 6 in the N-terminal region and at the C terminus a minor cluster composed of P 1 Ј to P 4 Ј each of which accommodates single NLS amino acids (51)(52)(53). Monopartite NLS peptides can apparently associate in an extended conformation with either region, whereas a bipartite NLS interacts with both clusters simultaneously by contacting the major site with its oligobasic core sequence and the minor site with its basic upstream amino acids. Lys-128 and Lys-129 of the SV40Tag NLS which had been identified as most critical for transport function were found to bind to the pockets P 2 -P 3 and P 1 Ј-P 2 Ј, which a have a strong preference for basic residues (51,53). Thus, if p10 and the SV40 Tag bind in the same region, it might be speculated that the p10 residues Arg-14 and Arg-15 that were most sensitive to mutation are held in these pockets (Fig. 6). Accordingly, p10 amino acids 13-18 (VRRLNG) would interact with P 1 -P 6 in the major binding site. Such a scenario would explain why replacing the four amino acids downstream of the two arginines in the p10 signal reduced the binding to importin ␣ by at least 5-fold. Furthermore, p10 amino acids Arg-6 to Leu-9 also contribute substantially to the interaction, because mutations at these positions reduced binding to importin ␣ to less than 10% and inactivated nuclear transport. Such a stabilization by amino acids that would be expected to interact outside the NLS binding pockets is apparently not required for oligobasic monopartite signals. In the case of the p10 signal, these additional interactions may support the presence of an arginine at P2, which was not well tolerated at this position in other NLSs (54). Thus, a stretch of at least 13 amino acids spanning positions 6 -19 seems to be involved in the binding of p10 to the NLS receptor protein. To our knowledge, the only other characterized example of an importin ␣-interacting motif containing just two clustered basic residues is located at the N terminus of the influenza A virus nucleoprotein NP (55) ( 6 TKRSXXXM 13 for importin ␣1 and 3 SXGTKRSYXXM 13 for importin ␣5). However, at present it is unclear if and how these two viral targeting signals are related to each other, because there is no further identity between those sequences. 2 M. Kann, unpublished observations. FIG. 5. Mutational analysis of the p10 N-terminal domain for interaction with importin ␣1 and nuclear transport. A, GST-p10 fusion proteins with di-alanine changes at p10 positions 4 -19 were expressed in E. coli. Equal amounts of GST, GST-p10 wild type, and mutant fusion proteins were adsorbed to glutathione-agarose beads and used to precipitate 35 S-labeled importin ␣1 from solution as described in the legend to Fig.  4A. The introduced mutations for each p10 protein are indicated on the top of the lanes. T, 10% of the total amount used for precipitation. Each mutant was tested in at least three independent experiments, and a representative gel is shown. B, GST, GST-p10 wild type, and the two corresponding p10 A6A7 and A10A11 mutant proteins were purified from bacterial lysates and fluorescently labeled with Cy3. Each substrate was mixed with a FITC-labeled BSA-M9 peptide conjugate and analyzed for import into permeabilized cells. Nuclear import of the indicated proteins was monitored by confocal analysis of cells using the TRITC channel to detect Cy3-labeled proteins (fields in the lower row) and the FITC setting for the BSA-M9 conjugate (fields in the upper row) that served as an internal control.
We have observed that the p10 protein contains a karyophilic signal sequence but does not completely localize to the nucleus in transfected cells. This suggests another level of regulation for its nuclear transport that may be explained by two mutually unexclusive reasons. First, the p10 protein may carry a cytoplasmic retention signal that competes with the activity of the nuclear transport signal. This hypothesis is supported by the fibrillous appearance of solitary cytoplasmic p10 which suggests that it is retained by binding to localized factors. Second, there are several known examples where nuclear transport of a protein is regulated by phosphate modifications (56). In fact, the p10 protein has been reported to become phosphorylated (57). However, it remains to be determined which specific amino acids are modified and whether this influences the nuclear transport of p10.
In summary, the data presented in this study lead to the identification of an unconventional karyophilic signal sequence in the BDV p10 protein that does not match a minimal consensus for classical NLSs, but appears to function through interaction with importin ␣. These findings highlight that importin ␣ can utilize karyophilic signals that deviate considerably from the consensus sequence. They also indicate that the prediction of a karyophilic signal in a protein is even more complicated than previously thought.
FIG. 6. Comparison of BDV p10 amino acids 6 -19 with oligobasic NLSs. The NLSs of the SV40 T antigen and of the c-Myc protein are aligned and positioned to corresponding binding pockets P1 to P6 in yeast and mammalian importin ␣ that were recently determined by crystallographic analyses (51,53). The p10 amino acids 6 -19 are presented below in a suggested alignment (see text). The p10 positions where mutations decreased the binding to importin ␣ to less than 10% of the wild type are indicated in boldface type.