Crystal Structure of Yeast Rpn14, a Chaperone of the 19 S Regulatory Particle of the Proteasome*

The ubiquitin-proteasome pathway is a major proteolytic system in eukaryotic cells and regulates various cellular processes. The 26 S proteasome, the central enzyme of this pathway, consists of a proteolytic core particle and two 19 S regulatory particles (RPs) composed of ATPase (Rpt) and non-ATPase (Rpn) subunits. Growing evidence indicates that proteasome assembly is assisted by a variety of chaperones. In particular, it has been reported recently that Nas2, Nas6, Rpn14, and Hsm3 bind specific Rpt subunits, thereby contributing to the formation of 19 S RP. Rpn14 transiently binds to the C-terminal domain of the Rpt6 subunit (Rpt6-C) during maturation of the ATPase ring of 19 S RP. In this study, we determined the crystal structure of yeast Rpn14 at 2.0 Å resolution, which revealed that this chaperone consists of a unique N-terminal domain with unknown function and a C-terminal domain assuming a canonical seven-bladed β-propeller fold. The Rpt6-binding site on Rpn14 was predicted based on structural comparison with the complex formed between Nas6 and Rpt3-C. The top face of Rpn14 exhibits a highly acidic surface area, whereas the putative interacting surface of Rpt6-C is basic. By inspection of structural data along with genetic and biochemical data, we determined the specific residues of Rpn14 and Rpt6 for complementary charge interactions that are required for 19 S RP assembly.

The 26 S proteasome, the major proteolytic machine found in eukaryotic cells, plays a key role in ubiquitin-dependent proteolysis by degrading proteins conjugated to ubiquitin (1,2). This large machine consists of a proteolytically active 20 S core particle (CP) 3 and one or two 19 S regulatory particles (RPs). The 19 S RP is responsible for binding to the ubiquitin-tagged substrate, unfolding, translocating into the 20 S CP, and releasing ubiquitin from the substrate. The 20 S CP is composed of 28 subunits arranged in a cylindrical shape of four heteroheptameric rings, ␣ 1-7 ␤ 1-7 ␤ 1-7 ␣ 1-7 , whereas the 19 S RP is composed of at least 19 different subunits and is divisible into two subcomplexes, the base and the lid. The base is composed of six different homologous AAA ϩ ATPase subunits (Rpt1-Rpt6) and three additional non-ATPase subunits (Rpn1, Rpn2, and Rpn13). The lid is composed of at least nine non-ATPase subunits. The lid-base association is stabilized by Rpn10 (3,4).
Rpn14/PAAF1 was originally identified as a 19 S RP-associated protein that negatively regulates the 26 S proteasome activity in mammalian cells (16,17). Rpn14 transiently binds to Rpt6 during assembly and consequently is not present in the mature 26 S proteasome (16). A major Rpn14-containing assembly intermediate is Nas6-Rpt3-Rpt6-Rpn14 in both yeast and human cells (13-15, 18, 19). A previous genetic study suggested that Rpn14 and Nas6 function cooperatively in the formation of the subassembly and 19 S base (14).
We reported recently the crystal structure of Pba3/Dmp2-Pba4/Dmp1 complexed with the ␣ 5 -subunit of the CP and provided a structural basis for mechanisms underlying CP assembly (10). On the other hand, although the crystal structures of Nas6 and its complex with the C-terminal domain of Rpt3 have been reported (20,21), detailed mechanisms regarding how the 19 S RP is formed remain unclear. For a better understanding of assembly mechanisms, it is necessary to elucidate the three-dimensional structure of the other chaperones, i.e. Rpn14, Hsm3, and Nas2. In this study, we report the crystal structure of yeast Rpn14 and predict a structural model for the Rpn14-Rpt6 interaction. Using yeast genetics and biochemical studies, we show that a specific interaction between Rpt6 and Rpn14 is required for the proper assembly of the 19 S RP.

EXPERIMENTAL PROCEDURES
Plasmid Construction-We amplified the Saccharomyces cerevisiae Rpn14 open reading frame by PCR from yeast genomic DNA of BY4741 and cloned it into the pET28(ϩ) vector (Novagen). Site-directed mutants were generated with the QuikChange site-directed mutagenesis kit (Stratagene) using pET28-Rpn14 as the template. We subcloned wild-type and mutant RPN14 genes into the pGEX6P1 vector (GE Healthcare) and the pYF5 vector (22), which express N-terminally FLAG-tagged Rpn14 under the control of the GAL1 promoter. We also constructed the pET28(ϩ)-Rpt6-C vector, which expresses the C-terminal region (amino acids 320 -405) of Rpt6. Details of the expression plasmids and the primer sequences used for plasmid construction are given in supplemental Tables S1 and S2, respectively.
Protein Expression and Purification-We cultured Escherichia coli BL21(DE3) cells (Novagen) harboring pET28-Rpn14   in LB medium containing kanamycin (50 g/ml) at 37°C. Protein expression was induced by the addition of 0.1 mM isopropyl ␤-D-thiogalactopyranoside at 18°C for 20 h. Cells were lysed in buffer A (20 mM Tris-HCl (pH 7.5) and 150 mM NaCl) by sonication. Cell supernatants were obtained by centrifugation at 15,000 ϫ g for 10 min at 4°C. We purified the recombinant protein by affinity, anion exchange, and gel filtration chromatographies and then concentrated it to 10 mg/ml. The selenomethionine derivative of Rpn14 was expressed in E. coli B834(DE3) (Novagen). The preparation and purification of the selenomethionine derivative of Rpn14 were carried out by the same procedures as described for the wild-type protein.
Crystallization and Data Collection-Crystals of Rpn14 were obtained under the condition of 200 mM magnesium chloride, 50 mM Tris-HCl (pH 8.0), and 25% (w/v) polyethylene glycol 3350 by the hanging-drop vapor diffusion method using Hampton screening kits. We obtained single crystals under an optimized crystallization condition of 8% (w/v) polyethylene glycol 3350 by microseeding at 293 K. Crystals (0.2 ϫ 0.2 ϫ 0.8 mm) were equilibrated in cryoprotectant buffer containing reservoir buffer plus 30% (v/v) glycerol and then cryogenized in a cold nitrogen stream at 100 K. Diffraction data sets were collected at beamline BL44XU of SPring-8 (Hyogo, Japan). These data were indexed and integrated with the program MOSFLM (23) and scaled using the program SCALA (24). Data collection statistics are given in Table 1. Crystals belonged to the space group P6 4 , with cell dimensions of a ϭ 73.3, b ϭ 73.3, and c ϭ 102.4 Å. The value of the Matthews coefficient was 2.13 Å 3 /Da for one molecule in the asymmetric unit, corresponding to a solvent content of 42.4%.
Structure Determination and Refinement-The structure of Rpn14 was determined by the multiwavelength anomalous dispersion (MAD) method using a selenomethionine derivative. The positions of the heavy atoms were obtained using SHELXD (25) and refined using SHARP (26). Initial MAD phases were extended to a higher resolution with the diffraction data collected from a native crystal to 2.0 Å resolution with solvent flattening using PIRATE (24) and DM (27). The initial model was constructed with the program ARP/wARP (28). This program automatically constructs a structural model containing 365 amino acid residues. The remaining parts of the structure were built manually using the program COOT (29). The model was refined at 2.0 Å resolution with the program REFMAC5 (30). The final Rpn14 model contained residues 1-414. There were no residues in disallowed regions of the Ramachandran plot. The final refinement statistics are summarized in Table 1.
Glutathione S-Transferase (GST) Pulldown Assay-GST-Rpn14 or the mutants expressed in Rosetta 2(DE3) cells (Novagen) were prebound to glutathione-Sepharose 4B and mixed with the E. coli lysate that expressed Rpt6-C for 2 h at 4°C. After centrifugation, beads were washed five times with buffer A and subjected to SDS-PAGE.
Native PAGE and Fluorescence Imaging-We analyzed proteasome assembly as described previously (14). In brief, the cell lysate of Rpn1-yEGFP1F-expressing cells was loaded onto 3.5% native polyacrylamide gel. After electrophoresis, the fluorescence of Rpn1-yEGFP1F on the gel was imaged using a Typhoon 9400 scanner (GE Healthcare). Using the same gel, proteasome peptidase activities against succinyl-LLVY-7-amido-4-methylcoumarin (Peptide Inc.) were imaged using an LAS-4000 IR system (Fujifilm).
Western Blotting-Total cell extracts were prepared by the mild alkali method (33). Antibodies used for Western blotting were horseradish peroxidase-conjugated anti-FLAG antibody M2 (Sigma), anti-Rpt6 antibody (a generous gift from Dr. Dan Finley), anti-Nas6 antibody (a generous gift from Dr. Junko Takeuchi), and anti-Rpn14 antibody. In this study, we raised anti-Rpn14 polyclonal antibody against rabbit using a fulllength Rpn14 recombinant protein.

RESULTS AND DISCUSSION
Overall Structure of Rpn14-The crystal structure of S. cerevisiae Rpn14 was determined by the MAD method and refined to 2.0 Å resolution (Table 1). Rpn14 consists of a small N-terminal domain (residues 1-75) and a C-terminal WD40 repeat domain (residues 76 -417) (Fig. 1, A and B). The N-terminal domain is composed of two antiparallel ␤-sheets (S1/S2/S3 and S4/S5/S6) comprising six ␤-strands and a short ␣-helix (H1). The C-terminal WD40 repeats assume a ␤-propeller domain with seven blades, each of which adopts a ␤-sheet conformation composed of four antiparallel ␤-strands. Thus, Rpn14 exhibits a canonical ␤-propeller structure, although a 23-residue loop is inserted between SD1 and SD2. This insertion is not strictly conserved in sequence or in length among organisms (supplemental Fig. S1). The top face of the propeller is defined by loops 2-3 and 4-1, which joins S4 (A-F) and S1 (A-F) of consecutive blades, and the bottom face by loops 1-2 and 3-4. The width of Rpn14 from the top face to the bottom face is ϳ22 Å, and the diameter is ϳ43 Å. The central channel of the ␤-propeller is ϳ7 Å in diameter.
Structural homologs of Rpn14 were searched in the Protein Data Bank using the Dali server (34). Rpn14 shares a significant degree of structural similarity with protein-binding proteins such as the signal transduction regulator RACK1 (35) and the G protein ␤-subunit, G␤ (36), with root mean square deviations for related elements ranging from 2.1 to 3.2 Å. Rpn14 has 16 -19% amino acid identity with these proteins. The core structure of ␤-propellers superimposes well, but the structure of the N-terminal domain of Rpn14 has no significant similarity to proteins with a known structure.
Electrostatic surface potential analysis of Rpn14 showed that the region of blades A, F, and G at the top of the propeller and the N-terminal region exhibit negatively charged surfaces, which are composed of Asp 13 (Fig. 1C). The remaining regions of the molecule have mixed electrostatic potentials. Mapping of the conserved residues on the surface of Rpn14 was performed using a ClustalW (37) multisequence alignment of Rpn14 proteins from eight different species and the ConSurf program (38). The conserved patches of residues are located at the top of the propeller, around blades A and B and the N-terminal region (Fig. 1D), whereas the bottom and perimeter surfaces are noticeably less conserved. The conserved surface at the top face of the molecule is relatively abundant in negatively charged residues, i.e. Asp 13 , Asp 22 , Glu 23 , Glu 27 , Glu 115 , and Asp 157 .
Structural Comparison with the G Protein ␤-Subunit-Rpn14 shows overall sequence and structural similarity to G␤ (19% sequence identity and root mean square deviation of 2.1 Å) ( Fig. 2A). G␤ exhibits a ␤-propeller fold composed of seven four-stranded ␤-sheets and an N-terminal extended helix and interacts with many binding partners. The crystal structure of G␤ complexed with an ␣-helical peptide (Protein Data Bank code 1XHM) has been reported previously (39). Because the C-terminal domain of Rpt6 is predicted to form an ␣-helical structure, one might expect that Rpn14 binds to this domain in a manner similar to the binding of the ␤-propeller of G␤ to the helical peptide exclusively at its top surface. This interaction site is separated into two regions: one bearing acidic residues (Asp 228 , Asn 230 , and Asp 246 ) and one bearing hydrophobic residues (Met 101 and Met 188 ) (Fig. 2B).
These residues are less conserved between G␤ and Rpn14. In Rpn14, the corresponding residues form a negatively charged area at the top face. Therefore, this unique surface property is likely required for binding to Rpt6.
Structural Features and Functional Similarity between Rpn14 and Nas6-Although the tertiary structures of Rpn14 and Nas6 are quite different, Rpn14 contains WD40 motifs, and Nas6 contains ankyrin repeats, and these chaperones bind to the C-terminal regions of specific Rpt subunits (13)(14)(15)19). Nas6 was shown to interact specifically with the C-terminal region of Rpt3, Rpt3-C (residues 348 -428), through surface charge complementarity (Fig. 3, A and B), i.e. one positively and two negatively charged patches in Nas6 and one negatively and two positively charged patches in Rpt3-C (20,21). On the basis of the above results, we hypothesized that the Rpn14-Rpt6 interaction may also occur through the complementary charge interaction, with the acidic surface at the top of Rpn14 accommodating positively charged residues of Rpt6-C. To investigate this possibility, a homology model of Rpt6-C (Fig. 3C) was built from the Rpt3-C structure (Protein Data Bank code 2DZN) using the MOE software package (Molecular Operating Environment, Chemical Computing Group Inc., Montreal, Canada). Structural comparison revealed that the surface charge environments of the interfaces are significantly different between Rpt3-C and Rpt6-C despite their structural similarities. Rpt6-C has three positively charged patches at the surface corresponding to the Nas6-binding site of Rpt3-C. We assumed that these differences account for the specificities of Rpt subunits by the 19 S base-dedicated chaperones.
Effects of Rpt6 Mutations on Complex Formation-To test the hypothesis that surface charges are involved in the interaction, we mutated the residues in the charged patches of Rpt6 and examined Rpn14 binding in vitro. Although the expression of the Rpt6-C protein alone yielded an aggregated form in E. coli, Rpt6-C was soluble when coexpressed with Rpn14. We then mutated basic residues (R333A/ R334A, R344A/R345A, and R373A/ R374A) in Rpt6-C and coexpressed them with Rpn14. As shown in Fig. 4A, the R373A/R374A mutation reduced binding to Rpn14, whereas the R333A/R334A and R344A/R345A mutants had no effect on the interaction. This result suggested that Arg 373 and Arg 374 in Rpt6-C are critical residues for Rpn14 binding. To determine the physiological relevance of the interaction, we constructed a yeast rpt6 mutant strain in which Arg 373 and Arg 374 of the RPT6 gene were mutated to Ala. The rpt6R373A/R374A cells grew normally even when cultured at the high temperature of 37°C (Fig. 4B). Previously, we found that the ⌬nas6 ⌬rpn14 double deletion cells show severe growth defects at high temperatures, whereas cells with single deletions do not show such defects (14). This genetic interaction suggests that the Nas6-Rpt3 and Rpn14-Rpt6 interactions are not indispensable for 19 S RP assembly but lethal when both interactions are absent. If Arg 373 and Arg 374 of Rpt6 are critical residues for Rpn14 binding, genetic interaction between rpt6R373A/R374A and ⌬nas6 should be observed. As anticipated, the rpt6R373A/R374A ⌬nas6 double mutant cells showed a temperature-sensitive phenotype, as did the ⌬nas6 ⌬rpn14 cells (Fig. 4B). The protein levels of Nas6, Rpn14, and Rpt6 were not affected by deletion or mutation (Fig. 4C).
Next, we investigated the impact of the basic residues of Rpt6 on 26 S proteasome assembly. To monitor proteasome assembly, we constructed green fluorescent protein-tagged FIGURE 4. Binding of Rpn14 to the basic residues of the Rpt6 C-terminal region. A, pulldown assay using Rpt6 mutants with Rpn14. His 6 -tagged Rpn14 was coexpressed with untagged Rpt6-C or its mutants in E. coli and pulled down on nickel-nitrilotriacetic acid resin. The resin-bound proteins were eluted by imidazole-containing buffer, and fractions were analyzed by SDS-PAGE, followed by Coomassie Blue staining. B, genetic interaction between the rpt6R373A/R374A mutation and the NAS6 deletion. Ten serial dilutions of the indicated strains were spotted onto a YPD plate and incubated for 2 days at 25 or 37°C. C, expression of Nas6, Rpn14, and Rpt6 in mutant cells. Exponentially growing cells in YPD medium were shifted to 37°C for 4.5 h. Extracts were prepared by the mild alkali method and analyzed by Western blotting with the anti-Nas6, anti-Rpn14, and anti-Rpt6 antibodies. D, native PAGE analysis of the 26 S proteasome assembly in mutant cells. The indicated cells expressing Rpn1-yEGFP1F were cultured as described for C. Extracts were subjected to native PAGE. After electrophoresis, Rpn1-containing complexes or proteasome peptidase activities on the gel were imaged using fluoroimagers. RP2CP and RP1CP indicate singly and doubly capped 26 S proteasomes. Rpn and Rpt subunits are abbreviated as n and t, respectively.
RPN1 strains through which RP subassemblies can be detected directly on native polyacrylamide gel (14). Similar to previous reports, 26 S proteasome levels were decreased in the ⌬nas6 ⌬rpn14 mutant due to a defect in RP formation (Fig. 4D). In the rpt6R373A/R374A single mutant cells, a slight decrease in 26 S proteasomes was observed, presumably because of a defect in the RP-CP interaction rather than RP assembly. Alternatively, severe impairment of RP assembly was observed in the rpt6R373A/R374A ⌬nas6 double mutant cells and in the ⌬nas6 ⌬rpn14 mutant cells. Thus, Arg 373 and Arg 374 in Rpt6 are critical residues for Rpn14 binding in vivo.
Genetic data indicated that the specific residue Asp 157 plays an important role in the function of Rpn14. We confirmed that both Rpn14 mutants D157A and D157K were expressed as well as wild-type Rpn14 in yeast cells (data not shown). Finally, we conducted in vitro pulldown assays using GST-Rpn14 and E. coli lysates expressing Rpt6-C. Compared with wild-type Rpn14, the D157A and D157K mutants resulted in reduced binding to Rpt6-C (Fig. 5C). On inspection of these data, we concluded that Asp 157 of Rpn14 contributes to the interaction with Rpt6 and the function of RP assembly.
In conclusion, we determined the crystal structure of yeast Rpn14 at 2.0Å resolution. The refined structure in conjunction with mutational and genetic data provided further insight into the interaction between Rpn14 and Rpt6. In particular, our findings suggest that the complex formation between Rpn14 and Rpt6 occurs through electrostatic interactions in a manner similar to that between Nas6 and Rpt3. Although mechanistic actions of how the base-dedicated chaperones regulate RP assembly remain unclear, our structural study of the Rpn14-Rpt6 interaction provides a framework for future studies of RP assembly.