Conformational remodeling of proteasomal substrates by PA700, the 19 S regulatory complex of the 26 S proteasome.

PA700, the 19 S regulatory complex of the 26 S proteasome, plays a central role in the recognition and efficient degradation of misfolded proteins. PA700 promotes degradation by recruiting proteasomal substrates utilizing polyubiquitin chains and chaperone-like binding activities and by opening the access to the core of the 20 S proteasome to promote degradation. Here we provide evidence that PA700 in addition to binding misfolded protein substrates also acts to remodel their conformation prior to proteolysis. Scrambled RNase A (scRNase A), a misfolded protein, only slowly refolds spontaneously into an active form because of the rate-limiting unfolding of misfolded disulfide isomers. Notably, PA700 accelerates the rate of reactivation of scRNase A, consistent with its ability to increase the exposure of these disulfide bonds to the solvent. In this regard, PA700 also exposes otherwise buried sites to digestion by exogenous chymotrypsin in a polyubiquitinated enzymatically active substrate, pentaubiquitinated dihydrofolate reductase, Ub(5)DHFR. The dihydrofolate reductase ligand methotrexate counters the ability of PA700 to promote digestion by chymotrypsin. Together, these results indicate that in addition to increasing substrate affinity and opening the access channel to the catalytic sites, PA700 activates proteasomal degradation by remodeling the conformation of protein substrates.

The 20 S proteasome consists of two seven-member rings of ␤-subunits containing the active sites flanked by two sevenmember rings of noncatalytic ␣-subunits (6,7). Either end of the 20 S proteasome can interact with two proteasomal regulatory complexes, PA700 (also called the 19 S regulatory complex) and PA28 (also called 11 S regulatory complex). PA700 contains ATPase sites, polyubiquitin, and misfolded protein binding sites. Together with the 20 S core, PA700 forms the 26 S proteasome responsible for the degradation of large polyubiquitinated and/or damaged proteins (8,9). The PA28⅐20 S proteasome complex degrades smaller polypeptides, producing peptides for presentation as class I major histocompatibility complex antigens (9 -11).
The multifunctional PA700 complex can be disassociated into two subcomplexes, the so-called "lid" and "base" (12). Although the function of most PA700 subunits is obscure, several activities have been defined by genetic and biochemical studies. First, the six AAA 1 ATPase subunits of PA700 base compose a ring, which associates with both ends of the 20 S proteasome when the 26 S proteasome is formed, thereby opening the narrow pore of the 20 S core (13,14). Second, the lid complex contains polyubiquitin chain binding activity. The S5a subunit of PA700 specifically binds polyubiquitin-tagged proteins in vitro (15). However in vivo studies in Saccharomyces cerevisiae and plants show that the ubiquitin chain binding site in S5a is not required for the degradation of all polyubiquitinated proteins (16,17), suggesting that additional sites are involved. Third, an isopeptidase activity (18,19) is present in the lid. A 37-kDa subunit, P37, catalyzes the cleavage of the polyubiquitin chains into ubiquitin monomers, which are then presumably recycled, whereas the targeted substrates are degraded. Fourth, a misfolded protein binding activity has been identified in the base. Both mammalian PA700 (20) and yeast proteasome regulatory complex (21) interact with several misfolding proteins and repress their aggregation. The base alone, which contains the six AAA ATPases and two other proteins, can mediate this activity. Finally, PA700 is presumed to be an unfoldase (22,23). The constricted annulus of even the dilated 20 S proteasome is thought to permit access of only denatured proteins to the catalytic sites within the chamber. By analogy, the AAA ATPase regulators of the prokaryotic Clp protease family, ClpA and ClpX, have been demonstrated to unfold the substrate prior to degradation (24 -26). An unfoldase activity was also demonstrated for the recently described archaebacterial proteasome-activating nucleotidase in addition to its antiaggregation and refolding activities (27). To test the hypothesis that proteasomal substrates are unfolded and/or remodeled by PA700, a misfolded protein, scrambled bovine pancreatic ribonuclease A (scRNase A) and a polyubiquitinated substrate pentaubiquitin-tagged dihydrofolate reductase (Ub 5 DHFR) was selected as model substrates. Our results suggest that PA700 promotes the exposure of otherwise buried sites in these two substrates, highlighting additional parallels between the 26 S proteasome and the Clp proteases.

EXPERIMENTAL PROCEDURES
Materials-PA700 and latent 20 S proteasome were purified from bovine red blood cells as described previously (8,28). Recombinant PA28␣, a potent proteasomal activator of peptide hydrolysis, was purified by standard methods (29). Hsc70 was purified from bovine brain as reported previously (30). Ub 5 DHFR was a kind gift from Dr. C. M. Pickart (Johns Hopkins University). Purified Escherichia coli GroEL and GroES were a gift from Dr. D. Chuang (University of Texas Southwestern Medical Center). scRNase A was made as described previously (31). Protein disulfide isomerase (PDI) was purified from rat liver (32). DsbA was purchased from Stress Gene. Polyribonucleic acid was from Calbiochem. Purified PA700, 20 S proteasome, and PA28␣ concentration were calculated based on their molecular masses of 700, 700, and 210 kDa, respectively. Purified scRNase A, PDI, and DsbA concentrations were determined spectrophotometrically using ⑀ 280 ϭ 9300, 43,000, 11,900 cm Ϫ1 M Ϫ1 , respectively.
Reactivation of scRNase A-scRNase A stock solution was prepared in 0.1% acetic acid. PA700, PDI, and other proteins were preincubated in refolding buffer A (20 mM Tris-HCl, 20 mM NaCl, 0.3 mM EDTA, 0.65 mM DTT, pH 7.6) for 10 min at 30°C, and then scRNase A was added to a final concentration of 6 M. At each time point, 30 l of the solution was removed, and the RNase activity was measured spectrophotometrically at room temperature with a cytidine 2Ј,3Ј-cyclic monophosphate (cCMP) substrate (33). The reaction mixture contained 1.2 M incubated scRNase A, 4 mM cCMP in 20 mM Tris-HCl, 20 mM NaCl buffer, pH 7.35. The hydrolysis of cCMP, reflecting the reactivation of RNase A, was monitored continuously as an increase in absorbance at 296 nm. The noncatalyzed reactivation of scRNase A under the same conditions was subtracted as background. The activity of an equivalent amount of native RNase A in this assay was taken as 100%.
Alternatively, scRNase A activation was determined using RNA as a substrate. 3 g of polymerized nucleic acid was incubated with 50 nM scRNase A in the presence or absence of 60 nM PDI, PA700, and other proteins in refolding buffer A for 15 min at 30°C. The reaction was stopped by adding 5% perchloric acid. The degree of reactivation was monitored by separating RNA on a 2% agarose gel and visualizing the remaining RNA with ethidium bromide fluorescence. In glycerol gradient co-sediment experiments, 15 mg of PA700 was resolved on a 2-ml 10 -35% glycerol gradient and centrifuged at 55,000 ϫ g for 3 h. The fractions were collected (100 l) and assayed (10 l) for their ability to promote the reactivation of 73 nM scRNase A using 2 g of RNA substrate. The fractions were also assessed (20 l) for their ability to stimulate the proteolytic activity of the 20 S proteasome using a suc-Leu-Leu-Val-Tyr 7-amino-4-methylcoumarin fluorogenic peptide substrate as described previously (35). The modification of PA700 and PDI by N-ethylmaleimide (NEM) was performed according to the protocol "conjugation with thiol-reactive probes" offered by Molecular Probes. Approximately, 3 M PDI or PA700 was incubated with 1 mM NEM in 20 mM Tris-HCl, pH 7.5, 20 mM NaCl, I mM EDTA, 10% (v/v) glycerol, and the mixture was kept at 4°C for 6 h under dark. Excess NEM was removed by dialysis extensively at 4°C against the same reaction buffer.
Oxidative Refolding NRCSQGSC (dansyl-K) N-peptide-The peptide was synthesized at the Howard Hughes Medical Institute facility at the University of Texas Southwestern Medical Center using a Rainin Symphony multiplex peptide synthesizer using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry and purified by high performance liquid chromatography (Ͼ95%). The identity was confirmed by matrix-assisted laser desorption ionization mass spectral analysis. The peptide concentration was calculated based on the molecular mass of 1329 Da. The redox folding of the peptide (50 M) was carried out in McIlvaine buffer (0.2 M dibasic sodium phosphate, 0.1 M citric acid, 5 mM reduced glutathione, 1 mM oxidized glutathione, pH 6.5) with appropriate concentrations of DsbA, PA700, and BSA as shown in Fig. 3B (34). Reduced and oxidized peptides were separated by reverse phase high performance liquid chromatography on a 0.46 ϫ 150-cm ultrasphere C18 column (Beckman). The peptides were eluted (1 ml/min) using a 1%/min gradient of 90% buffer A (0.1% trifluoroacetic acid:H 2 O), 10% buffer B (0.085% trifluoroacetic acid:acetonitrile) to 70% buffer A, 30% buffer B. The peptide was detected using the absorbance of the dansyl-lysine group at 340 nm. The oxidized and reduced peptides were eluted at 19% and 20% buffer B, respectively and were well resolved (data not shown).
Degradation of scRNase A by 20 S and 26 S Proteasome-26 S proteasome was assembled in vitro in refolding buffer A as described previously (35) from 20 S proteasome (50 nM) and PA700 (200 nM). The proteasome-specific inhibitor ␤-lactone (100 M) was added after 15 min of assembly. scRNase A stock solution (2.5 g) was added to 30 l of either 20 S or 26 S proteasome. To access the influence of DTT on degradation, 2.5 g of scRNase A was incubated in buffer A with or without DTT for 10 min, and then 75 nM 20 S proteasome was added to start the reaction. To detect the influence of PA700 on the degradation rate, 2.0 g of scRNase A in buffer A was preincubated for 30 or 150 min with or without 0.75 M PA700, and then 75 nM 20 S proteasome was added to the substrate for another hour. All of the reactions were carried out at 30°C. 5ϫ SDS sample buffer was added to stop the reactions. The samples were heated at 95°C for 5 min prior to 10% SDS-PAGE and visualization by Coomassie Blue staining.

PA700-dependent Reactivation of a Misfolded Protein, Scrambled RNase A-
The native bovine pancreatic RNase A structure required for enzymatic activity contains a specific set of disulfide bonds, Cys 26 -Cys 84 , Cys 40 -Cys 95 , Cys 58 -Cys 105 , and Cys 65 -Cys 72 . Any mismatched cysteinyl disulfide results in misfolding, RNase A with one or more inappropriate disulfides is classically denoted as scrambled RNase A (36). The reactivation of this misfolded protein is slow unless the disulfide exchange reaction is enzymatically catalyzed, or the exposure of these mismatched bonds to the solvents is increased (33,37). Both eukaryotic PDI and prokaryotic DsbA, a periplasmic thiol disulfide oxidoreductase, catalyze the formation and/or isomerization of disulfide bonds in vitro (33,37). Isomerization by PDI utilizes the reduced form of its Cys-X-X-Cys active sites to attack disulfide bonds of misfolded proteins and directly promote their exchange (38). To assess whether PA700 could also promote the reactivation of scRNase A, the formation of native enzymatically active RNase was assayed by the RNase-dependent digestion of RNA. Fig. 1A shows that like PDI and DsbA, PA700 can promote the reactivation of scRNase A. At a 1:1.2 molar ratio of scRNase A to enzymes, their relative scRNase A reactivation abilities are PDI Ͼ PA700 Ͼ DsbA. As shown in Fig. 1B, neither of the other proteins such as PA28␣ nor BSA has any obvious ability to accelerate the reactivation of scRNase A. This indicates that the ability of reactivation of scRNase A is not a common property of proteasomal activators like PA28 ␣ or a "sticky" protein, BSA. The data in Fig. 1C demonstrate that the peak of the reactivation of scRNase A activity co-sediments with PA700 protein (data not shown) and proteasome activation activity on a glycerol density gradient (10 -35%) centrifugation. This result argues that the catalyzed reactivation of scRNase A is because of PA700 and not the contamination of a trace amount of known disulfide isomerase.
The kinetics of scRNase A reactivation were assessed in the presence of PDI, DsbA, PA700, PA28␣, and BSA by detecting the RNase A-dependent hydrolysis of cCMP (33). The spontaneous reactivation of scRNase A among different preparations of scRNase A during the 3-h incubation in buffer A varied between 9 and 16%. The reactivation observed in the presence of PA700 is consistently above this background. PDI-dependent reactivation reached a maximum of ϳ40% by 1.5 h, consistent with previous reports (39). PA700 and DsbA, respectively, produced 12 and 6% reactivation above the spontaneous level after the 3-h incubation. PA28␣ and BSA did not catalyze significant reactivation over the same time course (Fig. 2A). The data in Fig. 2B show that PA700 exhibits an initial rate of reactivation of scRNase A linearly proportional to its concentration, suggesting a first-order process with a k cat ϳ0.08% reactivation/ h/nM PA700. The results demonstrate that PA700 accelerates the reactivation of scRNase A, a misfolded protein. PA700 contains six AAA ATPases (14). To assess whether ATP binding or hydrolysis could play a role in the reactivation ability, we carried out the RNA digestion assay for PA700 in the presence of ATP, but no obvious inhibition or promotion effect was observed.
Mechanism of Reactivation PA700 Does Not Share the Same Isomerization Mechanism as PDI-PDI utilizes reduced Cys residues in the Cys-X-X-Cys active sites to catalyze the oxidoreduction steps. NEM, a sulfydryl-specific modifying reagent, blocks the catalytic cysteines in PDI, which results in the loss of activity (Fig. 3A). NEM treatment also effectively inhibits the ATPase activity of PA700 (14) but not its chaperone-like activity (21). In this regard, the NEM-modified reactivation activity of PA700 is unaltered (Fig. 3A), consistent with the lack of ATP dependence in our RNA digestion assay. The results suggest that PA700 does not require ATPase activity for reactivation and that PA700 does not utilize exposed catalytic cysteines for accelerating the exchange of the misfolded disulfide pairs in scRNase A.
The thiol disulfide exchange reaction can be executed by PDI and DsbA on a wide range of substrates including proteins, peptides, and low molecular weight thiols and disulfides (31). Both PDI and DsbA catalyze the formation of an intramolecular disulfide bond in a small largely unstructured decapeptide, NRCSQGSCWN (34). To assess whether PA700 could also promote the formation of a disulfide bond, we synthesized an analogue (NRCSQGSC (dansyl-K)N) of this peptide, replacing the tryptophan residue with a dansyl-modified lysine. As shown in Fig. 3B, the catalytic amounts of DsbA catalyzed the oxidation of the peptide, whereas PA700 and BSA did not promote the formation of the disulfide bond, consistent with their inability to directly catalyze the oxidation step and a lack of an effect on already exposed bonds. Together, these results strongly suggest that PA700 does not directly catalyze the oxidoreduction step of the reactivation reaction, but rather the mechanism for promoting reactivation here is probably the remodeling of scRNase A by PA700 to increase and/or prolong the exposure of the substrate disulfides to the solvent.
PA700 Exposes Buried Chymotryptic Sites in a Polyubiquitinated Substrate-To further assess whether PA700 can expose otherwise inaccessible sites, Ub 5 DHFR was examined as a substrate. Ub 5 DHFR is degraded by the 26 S proteasome (22). The rate of degradation suggests that in light of the high affinity binding of Ub 5 DHFR (K m ϭ 35 nM) by PA700, substrate unfolding may be the rate-limiting step (22). To directly test whether PA700 could remodel this substrate, we examined the ability of PA700 to present otherwise buried sites in Ub 5 DHFR to chymotrypsin in trans. As shown in Fig. 4, at the present concentration of chymotrypsin, Ub 5 DHFR is resistant to chymotryptic digestion (lane 1). The isopeptidase activity (18) of PA700 catalyzes removal of ubiquitin from the Ub 5 DHFR sub- strate (lane 4). After eliminating the contribution because of the deubiquitination of Ub 5 DHFR, the degradation of Ub 5 DHFR by chymotrypsin was accelerated ϳ5-fold by PA700 (lane 5). No obvious acceleration was mediated by the lid subcomplex, which contains the polyubiquitination chain binding subunit S5a and the isopeptidase activity. Significantly, PA700 does not promote the chymotryptic digestion of nonubquitinated DHFR (data not shown). Moreover, Fig. 4 shows that the chymotryptic degradation was inhibited by saturating concen-trations of methotrexate (MTX), a ligand that stabilizes the folded conformation of DHFR (40,41). The ability of two chaperones to promote the exposure of chymotryptic sites in Ub 5 DHFR was also assessed. Hsc70, a chaperone known to be involved in substrate presentation to the proteasome (42), has no effect on the chymotryptic digestion of Ub 5 DHFR in the presence or absence of Mg-ATP (data not shown), suggesting that simple differential binding of the substrate cannot account for the further increased exposure of the chymotryptic sites. By contrast, the GroEL⅐GroES complex, a chaperonin with known unfoldase activity (43,44), also increased the exposure of the sites in Ub 5 DHFR (data not shown). These results are consistent with a PA700-dependent increase in exposure of otherwise occluded or buried sites and suggest that PA700 alone is unable to efficiently act on hyperstable conformations such as MTXstabilized Ub 5 DHFR.
Effect of PA700 on Proteasomal Degradation of scRNase A-The degradation of scRNase A is stimulated when PA700 associates with the 20 S proteasome to form the 26 S proteasome. 20 S proteasome degrades nonubiquitinated proteins, such as mildly oxidized proteins (45)(46)(47)(48), and the cyclin-dependent kinase inhibitor p21 WAF1/CIP1 (49). The molecular basis for the recognition of these substrates by the 20 S proteasome is not yet well understood, although it has been suggested that hydrophobic interactions may play a role (50 -52). We assessed the possibility that scRNase A is also a substrate for the proteasome. As shown in Fig. 5, A and C, the latent bovine red cell 20 S proteasome is capable of degrading nonubiquitinated scRNase A. This activity is enhanced when PA700 associates with the 20 S proteasome to form the 26 S proteasome (Fig. 5A). The degradation of the misfolded protein was inhibited by the addition of ␤-lactone, a specific proteasome inhibitor. In contrast, as shown in Fig. 5B, native RNase A is resistant to degradation by either the 20 S or 26 S proteasome, consistent with earlier reports (48,53).
The scrambled disulfides in scRNase A can be reduced under mild reducing conditions (0.2-1.0 mM DTT) (51). Therefore, we assessed whether the 20 S proteasomal degradation of scRNase A requires the disruption of the mismatched disulfide bridges by DTT. This is not the case as DTT in the reaction buffer actually retarded the degradation (data not shown). This effect is probably the result of accelerated spontaneous reactivation of scRNase A into the proteasomal resistant native RNase A. Thus, the degradation of scRNase A by the latent 20 S proteasome does not require the reduction of the scrambled disulfides by DTT.
Proteasomal Degradation of scRNase A Is Inhibited by Pretreatment with PA700 -PA700 catalyzes the reactivation of scRNase A to native RNase A. Because native RNase is resistant to 20 S proteasomal degradation, it was reasonable to expect that PA700 could rescue misfolded scRNase A from proteasomal degradation by promoting its reactivation and refolding. As shown in Fig. 5D, the preincubation of PA700 with scRNase A and DTT prior to the addition of the 20 S proteasome protects this substrate from degradation. Protection correlates well with the degree of reactivation ( Fig. 2A), suggesting that it is the gain of native structure during the preincubation that is protective rather than the formation of some more proteasome-resistant structure (i.e. aggregates). These results raise the possibilities that PA700 preforms multiple roles in vitro, promoting the degradation of a misfolded protein when it is associated with the 20 S proteasome and reactivating the misfolded substrate when it is not bound to the 20 S proteasome. Both abilities would require the remodeling of the substrates as shown here.

DISCUSSION
In this work, we have shown that PA700, the 19 S regulatory complex of the 26 S proteasome, catalyzes the reactivation of scRNase A and promotes exposure of buried chymotryptic sites in a folded functionally active proteasomal substrate, Ub 5 DHFR. This ability is probably a critical component of a concerted action of PA700 by which it binds the substrate using ubiquitin binding and chaperone-like sites, opens the annulus of the 20 S proteasome, and alters the conformation of the substrate itself. By contrast to the thiol oxidoreductases, PA700 promotes isomerization of the misfolded disulfides by increasing their exposure and does not exhibit the ability to promote the formation of disulfides. Not surprisingly, this reactivation activity is not observed for the proteasomal regulatory complex PA28␣ (Figs. 1 and 2). PA700 also accelerates the exposure of buried chymotryptic sites in the DHFR moiety of Ub 5 DHFR. Notably, this effect is unlikely to be simply attributed to preferential binding and stabilization of an open conformation, because Hsc70 does not exhibit this activity. By contrast, GroEL or GroEL⅐GroES complex, a chaperone molecule with known unfoldase activity (43,44), exhibits a similar ability to accelerate chymotryptic digestion of Ub 5 DHFR. This property of both PA700 and GroEL⅐GroES is inhibited by liganddependent stabilization of DHFR by MTX. Taken together, these data suggest that PA700 remodels the conformation of proteasomal substrates in addition to its well established activities, such as, polyubiquitin binding (15), ubiquitin isopepti-dase (18,19), 20 S proteasome activation (13,14), and the recently characterized "chaperone-like" activity (20,21).
The molecular chaperone-like activity of PA700 requires interaction site(s) that recognize misfolded proteins in the base subcomplex (20,21). Thus, this nonnative state recognition site may reside in the ring formed by the six AAA ATPases. In addition, a proteasome-activating nucleotidase from archaebacteria (27,54), homologous to these six AAA ATPases in PA700, mediates anti-aggregation, refolding, and unfoldase activities to promote proteasomal degradation. In analogy to cellular chaperones such as the GroEL⅐GroES complex and Hsc70, a hydrophobic interaction site(s) on PA700 would be expected to bind the misfolded proteins including scRNase A. The observation that Hsc70 is ineffective in remodeling these substrates suggests that a more extensive hydrophobic surface, such as that found on GroEL (55,56), may be required. Such a hydrophobic surface on PA700 has been detected by 8-anilino-1naphthalene-sulfonic acid binding (data not shown). Interaction with this site could stabilize and trap the more open forms of the misfolded substrates in preparation for translocation into the annulus of 20 S and subsequent proteolysis. Consistent with this model, the ability to promote the reactivation of scRNase A reported here is ATPase-independent. These open states are visited only rarely in MTX-stabilized Ub 5 DHFR and thus they can not be efficiently trapped by PA700 (as shown in Fig. 4). In this regard, the accelerated reactivation of scRNase A catalyzed by PA700 could be explained by a prolonged expo- sure of mismatched disulfides to solvent while bound to this site. Again, the stabilization of these remodeled open states of the substrate could promote their translocation through the narrow channel of the 20 S core particle to the catalytic sites.
The preferential translocation of open conformers to the proteolytic sites probably accounts for the ATPase dependence of the proteolytic process. This model predicts that substrate conformational remodeling by PA700 is coupled to translocation. As such, a more efficient remodeling of proteasome substrates (active unfolding) would require an intact translocation reaction in addition to stabilization of the open conformers. In prokaryotes, the regulatory complexes of Clp protease, ClpA and ClpX, have been shown to unfold green fluorescence protein, a very stable protein (24 -26). The unfolding process is facilitated when these complexes are associated with ClpP to form the proteolytic complexes ClpAP and ClpXP (25,26). Lee et al. (57) suggested that both the Clp protease and the proteasome catalyze unfolding by processively unraveling their substrates from the attachment point of the degradation signals and that the ability of a protein to be degraded depends on the stability of the local structure adjacent to the degradation signal. In these cases, degradation requires the associated regulatory complexes for both Clp protease and proteasome to recognize the substrates. However, as we reported here, the latent 20 S proteasome alone degrades scRNase A but not native RNase A. This observation suggests that the substrate itself may regulate the gating of the latent 20 S proteasome to promote degradation.
The ability of PA700 to either promote proteasomal degradation when forming the 26 S proteasome complex (Fig. 5A) or the refolding of the misfolded protein into a proteasome-resistant form in the absence of 20 S (Fig. 5D) raises an interesting question. Does PA700 catalyze the refolding of misfolded proteins in vivo? To date, there exists no firm evidence for PA700 sans 20 S in a cell other than during the assembly of 26 S. Therefore, The ability of PA700 to promote refolding and suppress aggregation in vitro may simply reflect partial reactions of its normal function in proteasomal activation. However, it is interesting to consider the alternative possibility that the balance between degradation and refolding could be adjusted depending upon the conditions facing the cell by controlling the association of PA700 with 20 S proteasome.