The Bipartite Nuclear Localization Sequence of Rpn2 Is Required for Nuclear Import of Proteasomal Base Complexes via Karyopherin αβ and Proteasome Functions*[boxs]

26 S proteasomes fulfill final steps in the ubiquitin-dependent degradation pathway by recognizing and hydrolyzing ubiquitylated proteins. As the 26 S proteasome mainly localizes to the nucleus in yeast, we addressed the question how this 2-MDa multisubunit complex is imported into the nucleus. 26 S proteasomes consist of a 20 S proteolytically active core and 19 S regulatory particles, the latter composed of two subcomplexes, namely the base and lid complexes. We have shown that 20 S core particles are translocated into the nucleus as inactive precursor complexes via the classic karyopherin αβ import pathway. Here, we provide evidence that nuclear import of base and lid complexes also depends on karyopherin αβ. Potential classic nuclear localization sequences (NLSs) of base subunits were analyzed. Rpn2 and Rpt2, a non-ATPase subunit and an ATPase subunit of the base complex, harbor functional NLSs. The Rpt2 NLS deletion yielded wild type localization. However, the deletion of the Rpn2 NLS resulted in improper nuclear proteasome localization and impaired proteasome function. Our data support the model by which nuclear 26 S proteasomes are assembled from subcomplexes imported by karyopherin αβ

unfolded proteins. Proteolysis of ubiquitylated proteins is ATPdependent and requires the association of 20 S core with 19 S cap complexes, which regulate substrate recognition and unfolding. Ubiquitin moieties of the substrate are recognized by at least two subunits, namely Rpt5 and Rpn10 (2)(3)(4). The lid subunit Rpn11 harbors a deubiquitinating metalloprotease activity, which allows ubiquitin peptide recycling before substrate degradation (5,6). The base ATPases possess chaperone functions, which promote substrate translocation into the catalytic chamber of the core particle (7). Apart from key roles of 19 S cap complexes in proteasomal proteolysis, base complexes were proposed to serve separate functions in transcription control (8). Despite the importance of the 19 S cap complex, the functions of most of its subunits remain unknown.
In yeasts 80% of 26 S proteasomes are localized inside the nucleus. Core and cap subunits are present in roughly equal stoichiometry with an abundance of 15,000 -30,000 molecules/ cell, which constitute ϳ1% of the total protein (9).
Immunoelectron microscopy in fission yeast revealed predominant proteasome decorations at the nucleoplasmic side of the NE 1 (10). In our group, we addressed the question how proteasomes are targeted into the nucleus. The NE is gated by nuclear pores. Their dimensions could accommodate assembled 26 S proteasomes. Nevertheless, large cargo proteins do not passively diffuse through nuclear pores but use a signal-mediated transport system. They are carried from the cytoplasm into the nucleus by nuclear transport receptors, called karyopherins or importins. The heterodimer karyopherin ␣␤ was the first described carrier and is referred to as the classic NLS receptor. The ␣ subunit recognizes the cargo NLS, and the ␤ subunit adapts the cargo complex to the nuclear pore and moves it into the nucleus. Classic NLS are characterized by short stretches of basic amino acid residues. The SV40 T antigen NLS was established as a monopartite NLS prototype, and the nucleoplasmin NLS as a bipartite NLS prototype. The basic amino acid residues of a bipartite NLS are spaced by 10 -12 unconserved amino acid residues (11).
With regard to nuclear import of proteasomes, we discovered that 20 S core particles are imported as inactive precursor complexes by karyopherin ␣␤ (12). Karyopherin ␣␤ was confirmed to be associated with core subunits by proteomics organizations that systematically analyzed tandem-affinity-purified protein complexes of yeast (13).
In this work, we investigated nuclear import of 19 S cap complexes. We found that it depends on karyopherin ␣␤ and that Rpn2 confers a crucial NLS for efficient nuclear targeting of base complexes under restrictive conditions.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Plasmids-Saccharomyces cerevisiae strains used in this work are listed in Table I. Standard methods and media were used for yeast growth and manipulation (14). Rpn1 and Rpn11 were C-terminally tagged by GFPHA, CFPHA, or Protein A by homologous recombination of SacIXhoI-cut Rpn1 tag::HIS3::URA3 and Rpn11 tag::HIS3::URA3 into the chromosomal locus, respectively (15). To use LEU2 as an auxotrophic marker in gene replacements, HIS3 and URA3 were excised from the original pBSHU construct as a XbaI-PstI fragment and replaced by the LEU2 gene.
Endogenous Rpn2 was replaced by the HA-tagged version by homologous recombination techniques using SacI-PvuII-cut Rpn2-HA 2 :: HIS3::URA3. Between the last amino acid of Rpn2 and the HA tag a spacer of 15 amino acids (AGSRANSSTLAAVTS) has been inserted. The C-terminal deletion of 161 amino acids of Rpn2 in PW⌬Cn2H has been created by using the same integration cassette as for Rpn2 HA. Immunoblot analysis of PW⌬Cn2H lysates revealed HA-tagged  (Table II) covering the chromosomal RPT2 deletion. pRpt2H is pRS315 (CEN, LEU2) containing the 1.8-kb XhoI-BamHI RPT2 fragment and a 75-bp BamHI-XbaI double HA fragment. Into the unique BamHI site of pRpt2H a fragment encoding GFP without a stop codon was inserted resulting in pRpt2G. p⌬NRpt2H and p⌬NRpt2G were derived from pRpt2H and pRpt2G, respectively. The RPT2 promoter region from Ϫ482 bp to the start codon was PCR-amplified as an XhoI-SacI fragment, the coding region from base pair 129 to the last codon as an SacI-XbaI fragment. Both fragments were ligated into pRS315 obtaining plasmids encoding HA-or GFPHA-tagged ⌬1-43 rpt2.
Localization Studies-For in vivo localization studies cells were grown to mid-log phase in synthetic complete medium, transferred to slides, and immediately examined with a Leica DMR microscope (15). Filters for fluorescein isothiocyanate/GFP (495-nm excitation and 525-nm emission) or CFP (436-nm excitation and 480-emission) were used (17). Photographs were taken with a Hamamatsu C5985-10 charge-coupled device camera using the MetaVue Software (Visitron Systems, Puchheim, Germany). Settings were identical for all images in a given experiment. Quantification of Rpn1-GFP and Rpn11-GFP in srp1-49 and SRP1 cells was done with the MetaVue system by selecting a fixed square of approximately the size of one-fifth of a yeast nucleus in the cytoplasm or nucleoplasm, respectively, and calculating the average pixel intensity per area. A minimum of 20 values from different images was averaged. The ratio of average pixel intensity was obtained for each strain by dividing the cytoplasmic intensity by the nucleoplasmic intensity.
Affinity Purifications-Standard IgG Sepharose affinity chromatography was performed using buffer PB (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml DNase I-EP) Cells were grown to mid log phase in 1 liter of YPD medium at 28°C and disintegrated by passing them trough a French pressure cell. After removal of cell debris (14,000 ϫ g, 20 min), the cell extracts were passed twice over a 1/25 lysate volume of IgG-Sepharose beads (Amersham Bioscience) at room temperature. The resin was washed with 50 bed volumes of buffer PB, and bound proteins were eluted with 0.5 M HOAc/NH 4 OAc (pH 3.4). Sequential IgG-Sepharose affinity chromatography from PWn11HTP lysates was performed following the standard protocol but in the presence of 2 mM ATP and 10 mM MgCl 2 . Proteins were eluted with 4 ml of buffer B (without ATP and MgCl 2 ), 20 S proteasomes were eluted with 4 ml of buffer B containing 300 mM NaCl, and 19 S base complex was eluted with 4 ml of buffer B containing 600 mM NaCl.
NLS-GFP-Strep-tagged constructs were affinity-purified over Strep-Tactin matrices followed by elution with 5 mM desthiobiotin according to the manufacturer's instructions (Institut fü r Bioanalytik GmbH, Göttingen, Germany). To eliminate biotinylated proteins from the extract prior to binding to the matrix, it was precleared for 30 min at 4°C by adding 40 g of avidin (Sigma) per ml of lysate.
Far Western Analysis-NLS-containing proteins were identified by far Western analysis using Protein A-tagged karyopherin ␣ (Srp1) as the binding partner. French press lysates (20 mg/ml protein) of strain CEY1b (20) were prepared in buffer PB with EDTA-free Complete Protease Inhibitor mixture (Roche Applied Science). Proteins were separated by SDS-PAGE, transferred to a polyvinylidene fluoride membrane, and refolded in renaturation buffer (50 mM HEPES, pH 7.3, 100 mM KOAc, 5 mM MgOAc, 0.3% Tween, 0.5% bovine serum albumin) overnight at 4°C. After blocking for 30 min in 5% dried milk in PBST (MPBST), the filter was incubated in 10 ml of MPBST containing a 1:5 dilution of CEY1b lysate to detect NLS-GFPS proteins and a 1:1 dilution of CEY1b lysate to detect 19 S subcomplexes. As a control, filters were incubated in 10 ml of MPBST containing 10 g of Protein A. Incubation for 4 h at 4°C was followed by washing with PBST. The blot was incubated with horseradish peroxidase-coupled rabbit IgG in MPBST for 1 h at 4°C and processed for enhanced chemiluminescence detection. After ECL detection the blot was stained with Amido Black.

RESULTS
In this work, we set out to investigate the nuclear import of base and lid complexes of proteasome regulatory particles. To study in vivo localizations of both subcomplexes, each complex was functionally tagged with fluorescent proteins. Rpn1 and Rpn11 were chosen as reporter subunits of the base and lid complexes, respectively. By exploiting homologous recombination techniques (12) yeast strains were created in which endogenous Rpn1 and Rpn11 were chromosomally replaced by GFPHA-and CFPHA-tagged variants. To compare the localization of the proteasomal base and lid subunits with the localization of proteins marking the ER/NE network, nucleoporin Nup49 and the ER-resident proteasomal substrate Pdr5* were co-expressed as GFP-labeled proteins (21,22). The cyan fluorescence of both proteasomal subunits was predominantly detected in the nucleus and specifically merged with the green fluorescence of Nup49 and Pdr5* at the NE ( Fig. 1A) consistent with previous reports on the localization of Mts4 and Pad1, the fission yeast homologues of budding yeast Rpn1 and Rpn11 (10).
To prove the functionality of the tagged versions of Rpn1 and Rpn11, glycerol gradient ultracentrifugation was performed. The tagged versions of Rpn1 and Rpn11 were detected in fast migrating fractions comprising proteasomes (not shown). Hence, the tagged versions of Rpn1 and Rpn11 were incorporated into their proteasomal subcomplexes. Furthermore, cells were created in which either Rpn1 or Rpn11 were functionally tagged with Protein A, allowing the affinity purification of proteasomal subcomplexes. Lysates of these cells were prepared and applied to standard IgG-Sepharose affinity chromatography. Tightly bound proteins were released from the affinity resin by low pH and subjected to SDS-PAGE followed by Coomassie Blue staining. All proteins were 19 S subunits as identified by fingerprint mass spectroscopy (23). Rpn11-ProA and Rpn1-ProA appeared to be stoichiometrically associated with base and lid subunits (Fig. 1B,  lanes 1 and 2). Endogenous Rpn1 and Rpn11 were not present in the complexes confirming that they were replaced by the Protein A-tagged versions, respectively.
Based on these findings we concluded that our tagged versions of Rpn1 and Rpn11 are functional, thus suited to report on the in vivo localization of base and lid complexes. To get insight into the import pathway of these complexes, GFPHAlabeled Rpn1 and Rpn11 were expressed in a set of nuclear import mutants. Significant delocalization of both reporter proteins from the nucleus to the cytoplasm was observed in mu- S base and lid complexes, respectively. A, direct fluorescence microscopy of S. cerevisiae wild type cells expressing CFPHAlabeled Rpn1 and Rpn11, respectively (strains PWn1C and PWn11C). Either GFP-tagged nucleoporin Nup49 or ERresident protein Pdr5* were co-expressed from centromere-based plasmids. Fluorescent proteins were visualized with fluorescein isothiocyanate/GFP and CFP filter sets. B, proteasomal 19 S complexes were isolated by IgG-Sepharose affinity chromatography from lysates of cells expressing Protein A-tagged Rpn11 (lane 1) and Rpn1 (lane 2) instead of the endogenous subunits. Proteins tightly bound to the affinity resin were eluted by low pH, subjected to SDS-PAGE, and stained with Coomassie Blue. All proteins identified by fingerprint mass spectroscopy were assigned. Ubp6 was confirmed to be associated with 19 S cap complexes (41).
tants affecting the classic NLS-dependent import pathway, especially in the temperature-sensitive srp1-49 mutant. At restrictive temperatures, GFPHA-labeled Rpn1 and Rpn11 ob-viously accumulated inside the cytoplasm of the srp1-49 mutant compared with isogenic wild type cells ( Fig. 2A). By using the MetaVue system we measured the pixel intensities within FIG. 2. srp1-49 cells, which are defective in classic NLS import, accumulate GFPHA-labeled Rpn1 and Rpn11 in the cytoplasm at restrictive temperature. A, Rpn1 and Rpn11 were chromosomally replaced by GFPHA-tagged versions in srp1-49 and isogenic SRP1 cells (strains PWn1G1-49, PWn1GSrp, PWn11G1-49, and PWn11GSrp). Cultures were grown for 3 h at 28°C or 37°C. The in vivo localization of GFP-labeled Rpn1 and Rpn11 was visualized by direct fluorescence microscopy. B, digital images of at least 20 cells per strain were quantified to yield the average GFP pixel intensity (see "Experimental Procedures"). The ratio of the cytoplasmic to the nuclear pixel intensity is indicated as percent value. All values are shown Ϯ S.D. C, exponentially grown wild-type and srp1-49 cells expressing Rpn11-GFPHA instead of the endogenous protein were shifted for 4 h to 37°C. Extracts were prepared and run in parallel on 10 -40% glycerol gradient ultracentrifugation. Fractions were collected from the top to bottom of the gradient and subjected to SDS-PAGE followed by immunoblot analysis using antibodies against HA epitopes, Rpt6, and ␤5, respectively. Matured m␤5 subunits mark the proteasome peak fractions. Immature pro␤5 subunits indicate 13-16 S precursor complexes (fractions 8 -13).
the cytoplasm and nucleoplasm of srp1-49 and wild type cells expressing GFPHA-labeled Rpn1 and Rpn11, respectively. The ratio of cytoplasmic to nuclear fluorescence intensity was calculated. At restrictive temperature, the ratio doubled in the srp1-49 mutant, whereas it did not change in wild type (Fig.  2B). One possibility to explain this phenomenon is that the srp1-49 mutation affects nuclear import of base and lid complexes. To ascertain that increased cytoplasmic fluorescence derived from base and lid complexes but not from non-incorporated subunits, lysates from srp1-49 and wild type cells expressing GFPHA-labeled Rpn1 and Rpn11 were fractionated by glycerol gradient ultracentrifugation (Fig. 2C, only shown for Rpn11-GFPHA). All gradient fractions were probed for Rpn11-GFPHA, the base ATPase Rpt6, and the core subunit ␤5. Rpn11 co-migrated with Rpt6 in fast sedimenting fractions showing that all fusion proteins were incorporated into their proteasomal subcomplexes of srp1-49 as well as of wild type cells ( Fig. 2C; panels labeled with Rpn11-GFPHA and Rpt6). Therefore, it is unlikely that single lid and base subunits accumulated inside the cytoplasm of srp1-49.
Proteolytically active proteasomes peaked in fraction numbers 12-15 as shown by the presence of matured ␤5 (Fig. 2C, panels labeled with pro-␤5 and m-␤5). Compared with wild type, Rpn11-GFPHA and Rpt6 were slightly shifted to slower sedimenting fractions suggesting that lid and base subcomplexes accumulate in the srp1-49 mutant ( Fig. 2C; base and lid subunits starting at fraction 8 in srp1-49 and at fraction 10 in wild type). To verify this assumption, we determined in which glycerol gradient fractions 19 S regulatory complexes and their subcomplexes migrated. 19 S regulatory complexes were isolated by sequential affinity purification using a modified protocol of Saeki and coworkers (24) and subjected to ultracentrifugation. All fractions were analyzed by SDS-PAGE and Coomassie Blue staining. Whole 19 S regulatory complexes were mainly detected in fractions 10 -15 (Fig. S1), whereas 19 S subcomplexes migrated in fractions 8 -11 as will be shown below (see Fig. 4D).
Karyopherin ␣␤-dependent nuclear import is mediated by nuclear localization signals. If 19 S subcomplexes are cargoes of karyopherin ␣␤, their subunits should harbor classic NLS. Analysis of all 19 S subunits revealed that potential NLS are present in Rpn1, Rpn2, Rpt2, Rpt4, Rpn5, Rpn6, Rpn8, Rpn9, and Rpn12 (as highlighted in Fig. 3A). Further basic amino acid residues were found within the C-terminal region of each ATPase subunit, which hardly conformed with classic NLS (25). In this work, we focused on NLS of base subunits, especially of domains conserved from yeast to human. The potential NLS motifs within the N-terminal region of Rpt2, within the C-terminal region of Rpn2 and the putative NLS within the C-terminal region of Rpt1, were chosen for further investigation (as underlined in Fig. 3A).
If these protein domains of the base subunits function as NLSs in vivo, they should direct a non-nuclear protein such as GFPS into the nucleus of living yeast cells. As positive and negative controls we expressed GFPS fused to the monopartite SV40 T antigen wild type NLS and mutant NLS (26), respectively. The SV40 NLS fusion protein clearly localized inside the nucleus; the mutant SV40 NLS did not accumulate inside the nucleus (Fig. 3B, panels 1 and 2). The presumptive NLS within the C-terminal region of Rpt1 fused to GFPS did not efficiently localize to the nucleus (Fig. 3B, panel 3). The potential NLS within the N-terminal region of Rpt2 and within the C-terminal region of Rpn2 directed GFPS into the nucleus with high efficiency, comparable with wild type SV40 NLS (Fig. 3B,  panels 4 and 5).
Next, we searched for direct interactions of karyopherin ␣␤ with our NLS-GFPS fusion proteins. The GFPS fusion proteins were isolated from yeast cell lysates by affinity chromatography on Streptactin matrices and subjected to far Western blots. The blots were probed for classic NLS cargoes by using Protein A-tagged karyopherin ␣␤. Specific reactions between karyopherin ␣␤ and the NLS chimera were detected by horseradish peroxidase-conjugated rabbit IgG and visualized by enhanced chemiluminescence (Fig. 3C, right panel). As positive and negative controls we established wild type and mutant SV40 NLS-GFPS fusion proteins, respectively (lanes 1 and 2, respectively). The putative NLS of Rpt1 did not react with karyopherin ␣␤ (lane 3). However, the NLS of Rpt2 and Rpn2 were recognized by the NLS receptor (lanes 4 and 5) verifying our observations above that the Rpt2 and Rpn2 NLS function in nuclear targeting. To confirm that equal protein amounts were subjected to far Western analysis, the blot was stained with Amido Black (Fig. 3C, left panel). Our results were in line with co-immunoprecipitation experiments using NLS-GFPS chimera as baits.
Equal amounts of GFPS fusion proteins were affinity-purified from cells shown in Fig. 3B, separated by SDS-PAGE, blotted, and probed for karyopherin ␣␤ and GFP as loading control. NLS chimera of Rpt2 and Rpn2 were found to be associated with karyopherin ␣␤ like wild type SV40 NLS (Fig. 3D, lanes 1,  4, and 5). NLS chimera of mutant SV40 NLS and Rpt1 were not recognized by karyopherin ␣␤ (lanes 2 and 3). The next question was whether the N-terminal region of Rpt2 and the C-terminal part of Rpn2 confer crucial NLS to the base complex. Deletion analysis of each NLS should give the answer. Strains were created in which the NLS of Rpt2 and Rpn2 were truncated yielding ⌬NLSrpt2HA and ⌬NLSrpn2HA mutants, respectively. The short HA epitope at the C terminus of each mutant subunit was introduced to mark the protein. The corresponding isogenic wild type strains expressed Rpt2HA and Rpn2HA instead of the endogenous subunits. Mutations with strong impact on proteasome functions cause sensitivity against elevated temperatures and canavanine, an amino acid analogue of arginine. Both conditions induce an increase of aberrantly synthesized proteins, which are preferred proteasome substrates (1). We tested our ⌬NLSrpt2HA and ⌬NLSrpn2HA mutants for those phenotypes generally displayed by proteasomal mutants. The ⌬NLSrpt2HA mutant showed no canavanine sensitivity, whereas the ⌬NLSrpn2HA mutant showed strong canavanine sensitivity compared with the isogenic wild type (Fig.  4A, panels labeled with Ϫcan and ϩcan). To test temperature sensitivity, cells were spotted on complete media and grown at 28°C and 37°C. The ⌬NLSrpt2HA and ⌬NLSrpn2HA mutants were strongly temperature-sensitive (Fig. 4A, panels labeled with 28°C and 37°C). In the case of the ⌬NLSrpt2HA mutant, the corresponding wild type Rpt2HA was already temperature-sensitive suggesting that the Cterminal HA tag interfered with ATPase functions.
Because the ⌬NLSrpn2HA mutant displayed both canavanine-and temperature-sensitivity, we anticipated that the NLS deletion of Rpn2 affects proteasome functions. Thus, the ⌬NLSrpn2HA mutant was assayed for polyubiquitylated protein accumulation. Equal protein amounts were loaded on SDS-PAGE, blotted, and probed for polyubiquitylated proteins. Increased polyubiquitylation was detected in the ⌬NLSrpn2HA mutant (Fig. 4B, lower panel, lane 1) compared with Rpn2HA wild type (lane 2). HA-tagged proteins, Rpt6 and ␤5, were immunodetected (upper panel).
Before we monitored the subcellular localization of ⌬NLSrpt2HA and ⌬NLSrpn2HA, we checked the incorporation of ⌬NLSrpt2HA and ⌬NLSrpn2HA subunits into the base complex, after the cells were grown at restrictive temperatures.
In the case of the ⌬NLSrpn2HA mutant, two activity peaks arose from 26 S proteasomes in the presence of ATP (Fig. 4C,  upper diagram). Compared with the wild type a reduction in proteasome activity of about 50% was determined in the ⌬NLSrpn2HA mutant consistent with our finding that the ⌬NLSrpn2HA mutant is strongly temperature and canavanine sensitive. As a control we measured the proteolytic activity of the core complexes in the absence of ATP. Under this condition the proteolytic activity did not significantly differ in the mutant and wild type (standard deviations of 15%).
Furthermore, we isolated base complexes from ⌬NLSrpt2HA, ⌬NLSrpn2HA, and their wild type complexes.
For this purpose, Protein A-tagged Rpn11 was chromosomally introduced into the respective strains. Cell lysates were passed on IgG-Sepharose beads to isolate 26 S proteasomes via the lid subunit Rpn11-ProA. The base and core complexes were eluted by 600 mM salt, whereas the lid complexes remained bound to the affinity resin. Then, the eluate containing base and core complexes was fractionated by glycerol density gradient ultracentrifugation. In mutant ⌬NLSrpt2HA and wild type Rpt2HA, FIG. 4. Characterization of mutants expressing ⌬NLS rpt2HA and ⌬NLS rpn2HA. A, logarithmically grown wild type and NLS deletion strains were diluted to 1 ϫ 10 7 cells/ml. 4 l of 10-fold serial dilutions were spotted onto YPD plates to test temperature-sensitive growth and onto minimal media plates supplemented with either 0.6 g/ml canavanine or 20 mg/ml arginine to test canavanine-sensitive growth. Minimal plates were incubated at 28°C, and YPD plates were incubated at indicated temperatures for 3 days. B, ⌬NLSrpn2HA mutants accumulate polyubiquitylated proteins. Logarithmically grown cells expressing either ⌬NLSrpn2HA (lane 1) or Rpn2HA (lane 2) were incubated at 37°C for 1 h. Extracts from cell equivalents were subjected to SDS-PAGE followed by immunoblotting using anti-ubiquitin, HA, Rpt6, and ␤5 antibodies. C, ⌬NLSrpn2HA is incorporated into the proteasome. Exponentially grown cells either expressing ⌬NLSrpn2HA and Rpn2HA instead of the endogenous proteins were shifted for 1 h to 37°C (strains PWn2H␤5H and PW⌬Cn2H␤5H). Extracts were prepared in the presence of 150 mM salt and subjected to 10 -40% glycerol gradient ultracentrifugation. Fractions were collected from the top to bottom of the gradient and assayed for peptide cleavage activity using the substrate benzyloxycarbonyl-Leu-Leu-Glu-␤-naphthylamide in the absence and presence of ATP. The specific activity is referred to as the maximum value (upper diagram). Protein samples were subjected to SDS-PAGE followed by immunoblot analysis using antibodies against HA epitopes and Rpt6 (lower panels). D, ⌬NLSrpt2HA is tightly associated with the base complex, whereas ⌬rpn2HA is loosely associated with the base complex. Base complexes from ⌬NLSrpt2HA, ⌬NLSrpn2HA, and their wild types were isolated by affinity chromatography as follows: Protein A-tagged Rpn11 was chromosomally introduced into cells expressing ⌬NLSrpt2HA, ⌬NLSrpn2HA, Rpt2HA, and Rpn2HA, respectively. Lysates were passed on IgG-Sepharose beads to bind 26 S proteasomes via the lid subunit Rpn11-ProA. The base and core complexes were eluted by 600 mM salt, while the lid complexes remained bound to the affinity resin. Then, the eluate containing base and core complexes was fractionated by 10 -40% glycerol density gradient ultracentrifugation. All fractions were analyzed by SDS-PAGE as shown for Rpt2HA and ⌬NLSrpt2HA in the lower panels and for Rpn2HA and ⌬NLSrpn2HA in the upper panels. Molecular weight markers indicate the Coomassie Blue-stained fractions. Immunoblots using HA, Rpt6, and ␣4 antibodies are shown below the Coomassie Blue-stained gels. base complexes migrated in glycerol gradient fractions 8 -11. Their protein patterns resembled each other (Fig. 4D, Coomassie Blue-stained gels shown in the lower panels). Western blot analysis confirmed the tight association of ⌬NLSrpt2HA and Rpt2HA with base complexes (Fig. 4D, immunoblots labeled with HA and Rpt6 as shown below the Coomassie Blue-stained gels). The core complexes migrated in fractions 11-14 as indicated by the presence of core subunit ␣4.
In mutant ⌬NLSrpn2HA and wild type Rpn2HA, base complexes migrated in fractions 8 -11 (Fig. 4D, Coomassie Bluestained gels shown in the upper left and right panels). However, the fractions of mutant base complexes were slightly shifted to slower sedimenting fractions. The protein pattern of ⌬NLSrpn2HA base complexes differed from wild type base complexes. Although Rpn2HA was stably incorporated into the base complex, ⌬NLSrpn2HA dissociated from the base components by 600 mM salt and migrated in slow sedimenting fractions 4 -6 (Fig. 4D, compare Coomassie Blue-stained gels and HA-labeled blots; Rpn2HA in fractions 8 -10, upper right panel; ⌬NLSrpn2HA in fractions 4 -6, upper left panel). As described above, fractions 11-14 comprise core complexes. These results suggested that ⌬NLSrpn2HA is incorporated into base complexes under physiological conditions (Fig. 4C) but not as tightly as the wild type subunit, because ⌬NLSrpn2HA is dissociated from base complexes by 600 mM salt (Fig. 4D).
We concluded that under physiological conditions the tagged versions of ⌬NLSrpt2 and ⌬NLSrpn2 are suited to report on base complex localization. Cells expressing either Rpt2GFPHA or ⌬NLSrpt2GFPHA instead of the endogenous proteins were monitored by direct fluorescence microscopy at 28°C and 37°C. The base complex harboring ⌬NLSrpt2GFPHA localized to the nucleus like wild type base complexes independent of temperature ( Fig. 5A; Rpt2GFPHA, lower panels, and ⌬NLSrpt2GFPHA, upper panels). In addition, we monitored the localization of Rpn1-and Rpn11-GFPHA in mutant ⌬NLSrpt2HA and wild type Rpt2HA cells. Their major localization inside the nucleus remained unchanged even at restrictive temperatures supporting our conclusion that the Rpt2 NLS is not essential for nuclear targeting of base complexes (not shown).
Vital cells expressing GFP-labeled Rpn2 and ⌬NLSrpn2 could not be obtained. Therefore, we examined cells expressing HA-tagged variants of Rpn2 and ⌬NLSrpn2, respectively, by indirect immunofluorescence microscopy. Base complexes containing ⌬NLSrpn2HA instead of the endogenous protein were predominantly nuclear at permissive temperature (Fig. 5B, upper panels, compare HA epitope staining with nuclear DAPI staining). At restrictive temperature, ⌬NLSrpn2HA appeared to be dot-like distributed around the nucleus and in the cytoplasm suggesting that base complexes are no more imported into the nucleus. Wild type Rpn2HA localized to the nucleus at permissive and restrictive temperatures (Fig. 5B, lower panels). As a control we checked the integrity of the NE. For this purpose, we stained nuclear pores with specific antibodies. Regular punctate staining of nuclear pore complexes was found in mutant and wild type cells at permissive and restrictive conditions (right panels as indicated). Furthermore, we tried to localize Rpn1-and Rpn11-GFPHA in mutant ⌬NLSrpn2HA.

FIG. 4-continued
However, neither tagged Rpn1 nor Rpn11 could be expressed in the mutant background. Taken together, our data suggested that the Rpn2 NLS is crucial for proper nuclear localization of the base complex, at least at elevated temperatures. At permissive temperatures, the deletion of the Rpn2 NLS can be compensated by NLS of other base subunits, most likely by the Rpt2 NLS, because a ⌬NLS rpn2 rpt2 double mutant is synthetic lethal.
Finally, we wanted to confirm that base complexes are cargoes of karyopherin ␣␤, whereas base complexes containing the ⌬NLSrpn2HA and ⌬NLSrpt2HA subunits, respectively, are not recognized by karyopherin ␣␤. Wild type and mutant base complexes were purified by using affinity chromatography and stepwise elution of proteasomal subcomplexes with salt as described in Fig. 4D. The base complex fractions were pooled and subjected to far Western blot analysis. Instead of full-length Rpn2HA and Rpt2HA, base complexes isolated from ⌬NLS mutant strains harbored the truncated ⌬NLS subunits, respectively (Figs. 6, A and B, left panels, compare lanes labeled wt  with ⌬). Far Western analysis revealed that karyopherin ␣␤ recognized Rpn2 as cargo protein either tagged with HA ( Fig.   FIG. 5. Intracellular localization of base complexes harboring Rpt2 and Rpn2 with NLS deletions. A, cells expressing ⌬NLSrpt2GFPHA and Rpt2GFPHA (strains PW⌬Nt2G and PWt2G) were grown to mid-log phase at permissive (28°C) and restrictive temperature (37°C) for 3 h and monitored by direct fluorescence microscopy. B, cells expressing ⌬NLSrpn2HA and Rpn2HA (strains PW⌬Cn2H and PWn2H) were grown to mid-log phase at 28°C and 37°C for 1 h. Indirect immunofluorescence microscopy was performed on formaldehyde-fixed cells. ⌬NLSrpn2HA (upper panels) and Rpn2HA (lower panels) were detected by mAb12CA5 and visualized by Cy3labeled donkey anti-mouse IgG. Nuclear pore proteins were detected by mAb414 (right panels) and visualized by Cy3-donkey anti-mouse IgG. Nuclear DNA was stained with DAPI. were purified from yeast (strains PWn2Hn11P and PW⌬C2Hn11P) as described above and subjected to far Western analysis followed by Amido Black staining (left panel). Far Western signals derived from cargo proteins, which were recognized by Protein A-tagged karyopherin ␣␤. Binding reactions were detected by horseradish peroxidase-coupled rabbit IgG and enhanced chemiluminescence (middle panel). Rpn2HA, ⌬NLSrpn2HA, Rpt2, and Rpt6 were assigned. HA-tagged subunits and Rpt6 were identified by specific antibodies (right panels). Proteins marked with an asterisk were cross-reacting Rpn11-ProA fusion proteins, which leaked from the IgG-Sepharose. B, as described above, base complexes from cells expressing Rpt2HA (lane wt; strain PWt2Hn11HTP) and ⌬NLSrpt2HA (lane ⌬; strain PW⌬Nt2Hn11HTP), respectively, were analyzed by Amido Black staining (left panel) and far Western blotting (middle panel). Rpn2, Rpt2HA, ⌬NLSrpt2HA, and Rpt6 were assigned. HA-tagged subunits and Rpt6 were identified by specific antibodies (right panels). C, base complexes interact with karyopherin ␣␤. IgG-Sepharose beads were loaded with Protein Atagged karyopherin ␣␤ (right) or mock treated (left). Wild type base complexes were loaded (lane 1). The beads were washed three times (lanes 2-4). Tightly bound proteins were eluted at low pH (lane 5). Fractions were subjected to SDS-PAGE followed by Coomassie Blue staining (CB, upper panels) and immunoblotting. Protein A-tagged karyopherin ␣ was probed with rabbit IgG-horseradish peroxidase conjugates (WB, middle panels). Base complexes were detected by antibodies against Rpt1 (lower panels).
To complete our studies, we tested direct binding of base complexes to karyopherin ␣␤. Protein A-tagged karyopherin ␣␤ was immobilized to IgG-Sepharose beads to yield karyopherin ␣␤ affinity beads. Wild type base complexes were passed on the affinity column (Fig. 6, right panel, lane 1). As a control equal amounts of base complexes were passed on mock treated beads (left panel, lane 1). After three wash steps (lanes 2-4), proteins bound with high affinity were eluted at low pH (lane 5). All fractions were subjected to SDS-PAGE, stained with Coomassie Blue (upper panels) and analyzed by Western blotting. The blots were probed for karyopherin ␣␤ (middle panels) and base complexes (lower panels). Base complexes were found to be associated with karyopherin ␣␤ allowing the final conclusion that they are cargoes of karyopherin ␣␤.

DISCUSSION
This work provides evidence that 19 S regulatory subcomplexes of 26 S proteasomes are imported into the nucleus via the NLS receptor karyopherin ␣␤. Our localization studies of temperature-sensitive srp1-49 mutants indicated that the classic NLS-dependent pathway is mainly responsible for nuclear import of lid and base components. The basal transport activity of karyopherin ␣ in the srp1-49 mutant suffices to achieve efficient nuclear import of NLS cargoes at permissive temperatures. At restrictive temperatures srp1-49 cells show aberrant nuclear morphologies, defective proteasomal proteolysis, and arrest in mitosis (27)(28)(29). An increased fluorescence intensity of GFP-labeled base and lid subunits was measured inside the cytoplasm. Because base and lid subunits were incorporated into complexes of approximately the size of their subcomplexes, we assume that base and lid complexes accumulate as cargoes of karyopherin ␣␤ in the cytoplasm of the srp1-49 mutant. Just an up-regulation of proteasomal gene expression cannot explain this phenomenon. Newly synthesized subcomplexes should not only arise in the cytoplasm but also in the nucleoplasm, if their nuclear import is independent of karyopherin ␣␤.
Proteasome nuclear trafficking is pivotal for proteasome function and cell cycle progression in yeasts as originally discussed by McDonald and Byers (30), who at first studied the in vivo localization of GFP-labeled base subunits, namely of Rpt4/Pcs1. Nuclear substrates seem to be abundant. Their proteolysis requires proteasomes in the nucleus (30). However, apart from proteolytically active proteasomes, we discovered a considerable amount of inactive precursor complexes in the nucleus. We concluded that nuclear core particles are matured at their place of destination (12,31). To ensure the constitutive presence of proteasomes in the nucleus, it is plausible that the same transport receptor is used for nuclear import of proteasomal subcomplexes. Base and lid complexes should be synergistically imported with precursor complexes and cooperatively assembled to complete nuclear 26 S proteasome biogenesis. Insufficient amounts of proteasomes in the nucleus may be a main reason for the pleiotropic phenotype of the srp1-49 mutant.
Based on genetic evidence Nomura and coworkers (27) already considered that nuclear import of proteasomal components depends on karyopherin ␣. STS1, a gene essential for nuclear segregation and division, and RPN11 are high copy suppressors of the srp1-49 mutation, which reflected a failure of nuclear import of Sts1, Rpn11, and related proteins by karyopherin ␣ (27). Sts1 directly bound to Rpn11 and karyopherin ␣ as judged by the yeast two-hybrid analysis. Neither yeast twohybrid nor in vitro interactions were detected between Rpn11 and karyopherin ␣ leading to the conclusion that karyopherin ␣ fulfills separate functions in proteasomal proteolysis apart from nuclear import of proteasomal components (27). Indeed, isolated lid complexes were not recognized by karyopherin ␣ in far Western blots. 2 Hence, nuclear import of lid components might be promoted by transiently bound adapter proteins, which render signals for karyopherin ␣ recognition. A candidate adapter protein might be Sts1/Cut8, which is essential for nuclear localization of lid complexes (32).
Yin6, a protein regulating cell division and mitotic fidelity via the proteasome, was further discussed to be involved in targeting lid complexes into the nucleus. The deletion of Yin6 resulted in delocalization of lid components to the cytosol, which correlated with proteasome inactivation. One interpretation of these results was that Yin6 accompanies lid complexes into the nucleus, as it contains a presumptive bipartite NLS (33,34).
To substantiate our conclusion, that 19 S base complexes are cargoes of karyopherin ␣␤, we analyzed base subunits for functional NLS. Our experimental design facilitated the comparison between in vitro and in vivo data. Fused to GFP-Streptactin each NLS-like peptide could be tested for its potency in nuclear targeting. Localization studies in living cells, co-precipitation, and far Western blot experiments yielded consistent results. The Rpt2 NLS and the Rpn2 NLS turned out to be functional. Both peptides efficiently targeted GFP into the nucleus and were specifically recognized by karyopherin ␣␤. The Rpn2 NLS fulfills the structural requirements of a bipartite NLS, whereas the Rpt2 NLS consists of two monopartite NLS spaced by 17 amino acid residues, which exceed the common linker length of a bipartite NLS (25).
The tertiary structure environment of the NLS peptides within the protein additionally plays an important role in determining their accessibility for karyopherin ␣. By using far Western blot analysis and solution binding assays we verified that the NLS function in full-length Rpt2, Rpn2, and in base complexes. In the case of Rpt2, the NLS comprises the Nterminal amino acid residues 11-37. From the literature, we knew that the truncation of the N-terminal 85 amino acid residues (of S4, the mammalian homologue of Rpt2) abrogates binding to neighboring ATPase subunits. Thus, the N-terminal region of Rpt2 was proposed to be needed for base complex assembly (35). Our data show that at least the N-terminal 43 amino acid residues of Rpt2 can be deleted without interfering with base complex assembly and nuclear localization. Similar results were obtained by McDonald and coworkers (36), who analyzed a putative NLS in the N-terminal region of Rpt4. Virtually all mutations that affected the N-terminal region of Rpt4 containing the putative NLS resulted in wild type phenotype (36).
In the case of Rpn2, the NLS is flanked by charged amino acid residues, namely the KEKE motif (37), which were disrupted in our ⌬NLSrpn2 mutant. Because C-terminal amino acid residues adjacent to the NLS and KEKE motif were pre-viously shown to be dispensable for proteasome function (38), we were allowed to conclude that the ⌬NLSrpn2 phenotype is caused by the lack of the Rpn2 NLS. Quaternary structure modeling implies that Rpn2 forms a toroid placed on top of the base ATPase ring (39). The hydrophilic region embedding the NLS and KEKE motif protrudes to the protein surface allowing protein interactions such as with karyopherin ␣␤.
In our previous work we already discussed the redundancy of NLS in ␣ subunits of core particles. We found that the deletion of a single NLS could be compensated by the presence of multiple NLS in neighboring subunits (12). With regard to the base complex, the synthetic lethal phenotype of mutants lacking both the Rpt2 and Rpn2 NLS suggested that Rpt2 and Rpn2 contribute the major recognition sites for karyopherin ␣. Moreover, the functions of the base complex NLS might be arranged in a hierarchical order, because the Rpn2 NLS is crucial for proper nuclear localization under restrictive conditions.
Non-classic NLS of proteasomal subunits used for karyopherin ␣ binding might still exist, which escaped our approach. STAT1 homodimers and STAT1-STAT2 heterodimers are examples for unconventional cargoes of karyopherin ␣␤. They harbor nonclassic NLS, which cooperate in STAT dimers and bind to karyopherin ␣ apart from the "major" NLS binding groove (40). Presumptive nonclassic NLS in Rpt6 might explain that Rpt6 was detected as a potential cargo of karyopherin ␣ in far Western blot analysis.
Taken together, we present evidence that base complexes are direct cargoes of karyopherin ␣␤. The bipartite NLS of Rpn2 is essential for nuclear localization and proper proteasome functions at least under restrictive conditions. The depletion of nuclear base complexes in our ⌬NLSrpn2 mutant correlated with reduced 26 S proteasome activity. The temperature sensitivity of the ⌬NLSrpn2HA mutant argues for changes in conformation and dissociation of ⌬NLSrpn2HA from the base complex as the cause for impaired nuclear import. Consistent with the latter explanation is the finding that the binding of ⌬NLSrpn2HA to the base complex is weakened by salt compared with the wild type subunit (Fig. 4D). However, under physiological conditions ⌬NLSrpn2HA was found to be incorporated into the base complex even at restrictive temperatures (Fig. 4C). The question arose as to why removal of the Rpn2 NLS did not block nuclear import at any temperature. Based on our result that Rpn2, Rpt2, and potentially Rpt6 (Fig. 6A) confer NLS to the base complex, we conclude that multiple NLS are present, which might be simultaneously recognized by karyopherin ␣␤ at permissive temperature. At elevated temperatures, conformational changes within the ⌬NLSrpn2 base complexes might weaken Rpt2 binding to karyopherin ␣␤ as the cause for insufficient nuclear import of mutant base complexes.
Finally, the accessibility of base complex NLS to karyopherin ␣␤ should be regulated by conformational changes and structural masking. Future tasks are needed to solve the mechanism by which base complexes escape recognition by karyopherin ␣␤ to remain in the cytoplasm.