|Home | About | Journals | Submit | Contact Us | Français|
Eukaryotic cells target proteins for degradation by the 26S proteasome by attaching a ubiquitin chain. Using a rapid assay, we analyzed the initial binding of ubiquitinated proteins to purified 26S particles as an isolated process at 4°C. Subunits Rpn10 and Rpn13 contribute equally to the high affinity binding of ubiquitin chains, but in their absence ubiquitin conjugates bind to another site with 4-fold lower affinity. Conjugate binding is stimulated 2-4 fold by binding of ATP or the nonhydrolyzable analog, ATPγS (but not ADP) to the 19S ATPases. Following this initial, reversible association, ubiquitin conjugates at 37°C become more tightly bound through a step that requires ATP hydrolysis and a loosely folded domain on the protein, but appears independent of ubiquitin. Unfolded or loosely folded polypeptides can inhibit this tighter binding. This commitment step precedes substrate deubiquitination and allows for selection of ubiquitinated proteins capable of being unfolded and efficiently degraded.
Eukaryotic cells direct proteins for degradation by the 26S proteasome by covalent attachment of a chain of ubiquitin (Ub) molecules to lysine residues on the protein (Ikeda and Dikic, 2008). The cell's many Ub ligases (E3s) bind specific protein substrates and catalyze the formation of a chain of Ub molecules by linking Ub molecules to the substrate and then to one of the seven lysine residues on the preceding Ub (Ikeda and Dikic, 2008). In vivo, Ub chains composed of different types of linkages target proteins to different intracellular fates. Generally, K48-linked Ub chains promote degradation by the proteasome, while K63-linked chains are involved in DNA repair and signal transduction or target proteins to the lysosome (Ikeda and Dikic, 2008). However, in vitro, both types of chains support degradation of proteins by 26S proteasomes (Hofmann and Pickart, 2001; Kim et al., 2007). A Ub chain of 4 or more Ub molecules is sufficient to promote degradation by purified 26S proteasomes (Thrower et al., 2000). This complex is composed of the 20S proteolytic particle and one or two 19S regulatory particles, which bind the polyubiquitinated proteins.
Protein degradation by the 26S complex is an ATP-dependent process (Goldberg 2003) and the base of the 19S complex contains six homologous ATPase subunits (Rpt1-6) which form a ring that interacts with the 20S particle. The degradation of globular proteins requires their unfolding by these ATPase subunits, which then translocate the protein into the 20S proteasome where it is rapidly cleaved to small peptides. A key function of the ATPases is to open the gated entry channel in the outer ring of the 20S through which the unfolded substrate enters the particle (Smith et al., 2007). The 19S complex also contains three deubiquitinating activities that catalyze disassembly of the Ub chain to individual Ub molecules, which are then reutilized. To insure efficient proteolysis, these various steps must be carefully integrated. It is now clear that the binding of ubiquitinated proteins to the 26S stimulates gate opening in the 20S (Bech-Otschir et al., 2009; Li and Demartino, 2009; Peth et al., 2009). This step facilitates substrate entry and occurs when the Ub chain interacts with one of the proteasomal deubiquitinating enzymes (DUB), Ubp6/Usp14 (Peth et al., 2009). Thus, disassembly of the Ub chain is linked to activation of proteolysis. However, the precise order of events and how the various 26S functions (e.g. unfolding and deubiquitination) are coordinated during the degradation of the substrate remains unclear.
It is presently unknown how different types of ubiquitinated proteins associate with the 26S proteasome, and whether this binding step necessarily commits the substrate to degradation. Genetic studies indicated that the 19S particle contains multiple subunits that can bind Ub chains, especially Rpn10/S5a, which binds Ub conjugates by its Ub interacting motif (UIM) domain, and Rpn13, which interacts with Ub through its pleckstrin homology domain (Deveraux et al., 1994; Elsasser et al., 2004; Husnjak et al., 2008; Schreiner et al., 2008). Rpn10 and 13 have recently been proposed to be in sufficiently close proximity to interact with the same Ub chain and to comprise one binding site (Zhang et al., 2009). In addition, cross-linking studies have suggested that the ATPase subunit, Rpt5, may also bind polyUb conjugates (Lam et al., 2002). A major complication is that cells also contain several Ubl-Uba proteins (i.e. Rad23) that can function as shuttling factors and facilitate the binding of certain ubiquitinated proteins to the proteasome (Elsasser et al., 2004). It is unclear if these different binding sites function independently or sequentially or how the initial binding is linked to disassembly of the Ub chain, substrate unfolding, and translocation.
Recent findings suggest that the ubiquitination of a substrate is not sufficient for its degradation by the 26S, and that efficient proteolysis also requires an unstructured domain in the substrate (Prakash et al., 2009; Takeuchi et al., 2007; Zhao et al., 2009). Possibly, this unstructured domain allows interaction with the proteasomal ATPases to initiate the unfolding or translocation process. The 19S ATPases evolved from the proteasome-regulatory ATPase complex in archaea, PAN (Zwickl et al., 1999), which binds unfolded protein in an ATP-stimulated manner (Benaroudj et al., 2003). Like the PAN-20S complex, the 26S proteasome also degrades unstructured non-ubiquitinated proteins (e.g. casein) in an ATP-dependent fashion (Cascio et al., 2002). Therefore, we investigated whether the 19S complex also interacts with ubiquitinated substrates in a similar nucleotide-dependent manner.
After binding, a ubiquitinated substrate can either disassociate, be deubiquitinated and released, or be degraded, and the factors influencing these decisions are completely unclear. Purified 26S often fail to degrade various polyubiquitinated substrates (Bech-Otschir et al., 2009; Peth et al., 2009; Prakash et al., 2009), which suggests an editing step, in which certain ubiquitinated proteins are spared from destruction, or proteins that cannot be degraded are released.
The present studies were undertaken to clarify how the proteasome processes ubiquitinated proteins by analyzing their initial binding to the 26S as an isolated event, independently of deubiquitination, unfolding, and degradation. Prior studies of conjugate binding involved prolonged assays during which multiple events may occur (Elsasser et al., 2004). Our approach allowed us to define the cofactor requirements for conjugate binding and to identify the contributions of individual 19S subunits to this process. In addition, we show that conjugate binding occurs in two steps. The initial high-affinity association depends only on the Ub chain, is stimulated by ATP binding, but is also readily reversible. Then a temperature-dependent step that requires ATP hydrolysis and a loose domain on the ubiquitinated protein leads to tighter binding of the substrate. This step precedes the removal and disassembly of the Ub chain and seems to serve an editing function that distinguishes proteins capable of complete hydrolysis.
We developed a simple quantitative assay to measure the initial binding of polyubiquitinated proteins to 26S proteasomes as an isolated event, distinct from the multiple subsequent steps that lead to their degradation. This assay enabled us to define the substrate specificity, reversibility and cofactor requirements of the binding process, independent of deubiquitination, unfolding, and proteolysis.
To obtain chemical amounts of polyUb conjugates immobilized on a GSH-resin, we incubated a GST-linked E3 ligase (typically either E6AP, which forms K48 chains, or Nedd4, which forms K63 chains (Kim et al., 2007)), with E1, E2, Ub and ATP to allow autoubiquitination. These reagents were removed by washing. The immobilized Ub conjugates were then incubated with 26S proteasomes, which had been affinity purified from rabbit muscle or yeast (Besche et al., 2009). To prevent deubiquitination, unfolding and proteolysis, the 26S proteasomes and matrix-bound ubiquitinated proteins were incubated at 4°C. After 30 minutes at 4°C, we washed the column to remove the unbound proteasomes, and the bound 26S particles were assayed at 37°C by measuring the peptidase activity associated with the resin in the presence of ATP. However, a similar extent of binding was also observed during incubations at 37°C (Fig. S1B).
Initial experiments confirmed that assaying peptidase activity against Suc-LLVY-amc reflects the amount of 26S particles bound to the Ub conjugates. This activity correlated very tightly with the amount of 20S proteasome subunits (α3) bound to the polyUb conjugates as determined by Western Blot (Fig. S1A). One potential complication with this approach was the recent finding that binding of polyUb conjugates to the 26S stimulates the peptide hydrolysis by causing gate opening in the 20S (Bech-Otschir et al., 2009; Li and Demartino, 2009; Peth et al., 2009). Since proteasome activation might potentially complicate our activity-based assays, we also compared the binding of wildtype (wt) and open gated mutant (α3ΔN) yeast 26S particles, in which conjugate binding cannot stimulate peptide hydrolysis (Peth et al., 2009). Using the peptidase assay, the binding of the wt and the open-gated proteasomes were indistinguishable (Fig. 1A). Also, by both assays, the extent of proteasome binding was proportional to the 26S concentration in the reaction. Thus, assaying peptidase activity is a valid, easy and quantitative measure of the amount of 26S bound.
Under our typical conditions, approximately 20-25% of the 26S particles present (10 nM) bound to the immobilized Poly-Ub-E6AP (30 nM) and no significant binding of proteasomes to the E6AP column was seen in the absence of ubiquitination. Large changes in proteasome mobility on native PAGE were observed previously after Ub conjugate binding to proteasomes, which are most likely caused by binding of two proteasome particles to the ubiquitinated CDC34 dimer (Elsasser et al., 2004). Under our conditions, no such changes after the initial binding of ubiquitinated E6AP were observed. This binding process was very similar using mammalian or yeast 26S proteasomes with several different autoubiquitinated E3s (i.e. Poly-Ub-E6AP or Poly-Ub-Nedd4), which form conjugates that vary widely in length (Peth et al., 2009), as well as with Ub5-DHFR, a homogenous synthetic conjugate containing a Ub4 chain. We also observed proteasome binding to linear Ub chains (Fig. S1C, ,4F)4F) which also support degradation (Prakash et al., 2009), suggesting these types of Ub chains can bind to the proteasome.
Although the main signal in vivo for degradation by the 26S proteasome is the attachment of a K48 chain; however, K63 chains can also support rapid degradation by purified 26S proteasomes (Kim et al., 2007).. Therefore, we compared 26S binding to K63 chains (Nedd4) and K48 chains (E6AP). Both types of chains support binding of 26S particles (Fig. 1B) in accord with previous suggestions (Hofmann and Pickart, 2001; Kim et al., 2007). To determine whether K63- and K48 chains bind to the same or different sites on the 26S, we tested if unanchored K48- or K63 Ub4 chains can compete with the GST-E6AP-K48 Ub chain for binding to the 26S. As shown in Fig. 1B, adding Ub4 composed of K63 or K48 linkages reduced proteasome binding to a similar extent. Virtually identical inhibition was observed using GST-Nedd4-K63 polyUb with free K48 or K63 Ub4 (not shown). From these data a K1/2 of approximately 32±1 nM was calculated for each which is very similar to the affinity of the 26S for Ub5-DHFR binding measured previously (35 nM) (Thrower et al., 2000), and to the concentrations of conjugates that activate gate-opening (Bech-Otschir et al., 2009; Peth et al., 2009).
Thus, despite their very different structures, K48 and K63 chains bind to the proteasome very similarly, and as shown previously, unanchored chains containing four Ub molecules bind to the 26S (Thrower et al., 2000). However, only 50% of proteasome binding to ubiquitinated E6AP could be inhibited by high concentrations of K48- or K63 Ub4 (even when the two were added together). Therefore, half of the E6AP Ub conjugates must have associated with proteasomes with higher affinity than Ub4. To test this conclusion, we studied binding to the shorter homogenous Ub conjugate Ub5-DHFR, and with 300 nM Ub4, which reduced 26S binding to poly-Ub-E6AP by only 50%, we observed a complete inhibition of proteasome binding (Fig. 1C). These findings indicate that the initial binding affinity is through the Ub chain and strongly suggests that an increased length of the Ub chain results in a higher affinity as reported previously (Thrower et al., 2000).
Recent reports have implicated Rpn10 and Rpn13 as binding sites on the 26S proteasome for Ub chains (Husnjak et al., 2008). However, the physiological importance of each is uncertain, since their relative affinities for Ub chains and di-Ub differ markedly and were determined using isolated Rpn10 and Rpn13, whose structures may differ in intact proteasomes. To investigate the contributions of each to the binding of polyubiquitinated proteins, we incubated a set amount of polyUb conjugates with increasing concentrations of 26S proteasomes purified from wt, rpn10ΔUIM, rpn13KKD or rpn10ΔUIM/rpn13KKD yeast strains and determined their affinities and maximal binding capacities. These mutations specifically inactivate the Ub binding domains of these subunits and should not alter the structures of the complex (Husnjak et al., 2008). As expected, the wt 26S particles showed the highest affinity for the polyUb column (Fig. 2A). Both the Rpn10ΔUIM and Rpn13KKD mutant proteasomes showed a two-fold reduction in affinity, while the Rpn10ΔUIM/Rpn13KKD double mutant particles showed about a 4-fold decrease in affinity for polyUb conjugates (which corresponds to an affinity between 100-200nM based on Fig. 1D and and2A).2A). Thus, Rpn10 and Rpn13 contribute approximately equally to the high affinity binding site on wt particles. It is noteworthy that although these mutations similarly reduce affinity for Ub chains, they did not affect the total amount of proteasomes that could bind. Furthermore the lower affinity binding site that was evident in the absence of Rpn10 and Rpn13 at high 26S concentrations enables binding to Ub conjugates to a similar maximal extent as do the high affinity sites.
The subunit responsible for this lower affinity site could be one of the three DUBs or the ATPase subunit Rpt5 (Lam et al., 2002). To determine if the DUBs in mammalian 26S proteasomes contribute to this initial binding of polyUb conjugates, we blocked the active sites of UchL5 and Usp14 by pretreatment with Ub-Vinylsulfone, which covalently attaches a Ub moiety to the active site, or inactivated Rpn11 with ortho-phenanthroline, a Zn2+ chelating agent. Treatment of 26S particles with either did not affect binding to Ub conjugates (Fig. S2A). Thus the 19S DUBs do not appear to be critical for the high or lower affinity binding.
The six 19S ATPase subunits serve multiple functions in protein degradation. While ATP hydrolysis is necessary for protein unfolding, binding of ATP and nonhydrolysable analogs (e.g. ATPγS), but not ADP, stimulate gate opening and entry of peptides or unfolded proteins into the 20S (Peth et al., 2009; Smith et al., 2005). To learn whether nucleotide binding or hydrolysis also influences proteasomal affinity for Ub conjugates, we performed assays in the presence of no nucleotide, ADP, ATP or ATPγS (Fig. 2B). Although ADP reduced binding or had no effect, ATP and ATPγS both enhanced 26S binding to the conjugates. (Rabbit proteasomes, unlike those from yeast, do not dissociate rapidly in the absence of nucleotides (not shown)). Thus, binding of Ub conjugates seems maximal when the 19S ATPases are in the ATP-bound conformation, and this process does not require nucleotide hydrolysis.
To confirm that this stimulation of conjugate binding involves nucleotide association with the ATPases, we compared 26S particles from a wt and Rpt5K228S mutant strain. This mutation prevents ATP binding to the ATPase subunit Rpt5 (Rubin et al., 1998) and resulted in a significant decrease in polyUb conjugate binding to the proteasomes (Fig. 2C, S2B).
To determine if ATP influences the function of a single Ub-binding subunit or all of them, we assayed binding to polyUb conjugates of 26S particles purified from the rpn10ΔUIM, rpn13KKD and rpn10ΔUIM/rpn13KKD mutant strains in the presence of ADP, ATP and ATPγS. Proteasomes with nonfunctional Rpn10 and Rpn13 both show a 2-3 fold increase in binding in the presence of ATP and ATPγS (Fig. 2D). Interestingly, this stimulation by ATPγS was consistently larger (5-fold) in the Rpn10/Rpn13 double mutant particles. Thus, ATP binding most likely alters the conformations of Rpn10 and Rpn13 as well as the unidentified lower affinity site so as to promote their capacity to bind to Ub conjugates and that hydrolysis of ATP to ADP must reduce the affinity of each binding site for Ub chains.
At some stage after associating with the proteasome, a ubiquitinated protein presumably must become more tightly bound to prevent its dissociation and to ensure complete deubiquitination and degradation. Such a commitment step may also involve an editing step with release of some Ub conjugates (e.g. ones that may resist degradation) and tighter binding of the ubiquitinated proteins appropriate for destruction. To learn if such a transition to tighter binding occurs, we first examined under what conditions the 26S bound at 4°C can be washed off the column. With increasing salt concentrations (Fig. 3A), all the proteasomes bound to the Ub conjugates could be released (i.e. by 300 mM NaCl or KCl). At these high salt conditions, the Rpn10-Rpn13 conformation may not allow association with the Ub chains. Thus, the association of Ub conjugates with the 26S may involve ionic interactions with the 26S (in addition to requiring hydrophobic domains on the Ub chain (Beal et al., 1996)). Proteasomes bound at 4°C could also be released from the conjugates by adding high concentrations of the UIM domain of S5a (Fig. 3F) (Besche et al., 2009). Thus, this initial binding process is readily reversible and no tighter binding (“commitment step”) was observed at 4°C. We therefore pre-bound 26S particles to the Ub conjugates at 4°C, removed the unbound fraction of 26S particles and then incubated in the presence of ATP for an additional 30 min at 37°C (see Fig. 3B). After this treatment, most of the bound 26S proteasomes could no longer be washed off the Ub chain with the high salt buffer or with the UIM domain (Fig. 3F). By contrast, if the pre-bound proteasomes were maintained at 4°C for the additional 30 minutes, they could be completely dissociated (Fig. 3C). The maximal transition from the initial to the tighter binding occurred approximately 20 minutes after the shift from 4°C to 37°C (Fig. 3D). Interestingly the time to maximal tighter binding resembled the time to the maximal stimulation of 26S gate opening by Ub conjugates, which reflects the time when the Ub chain interacts with the 19S DUB Usp14/Ubp6 (see below). These findings indicate a temperature-dependent step that leads to very tight association, and that seems to require ATP hydrolysis. Unlike addition of ATP, ADP never allowed the transition to the tighter binding (Fig. 3E). Furthermore, ATPγS failed to promote this transition. This step thus differs from 20S gate-opening (Smith et al., 2005) or the stimulation of Ub conjugate binding (Fig. 2D) where ATPγS caused a larger stimulation than ATP. (In some experiments, ATPγS caused a small stimulation, which is probably due to its ability to spare the low amounts of ATP used to stabilize the 26S). Thus, ATP hydrolysis and the 19S ATPase subunits appear essential for this transition to tighter binding at 37°C.
Since many Ub conjugates appear to be delivered to the proteasome by Ubl/Uba “shuttling factors” (e.g. Rad23)(Elsasser et al., 2004), we tested if mammalian hHR23A can increase the initial or the subsequent tighter binding of Ub conjugates to 26S particles. Addition of hHR23A at 4°C stimulated the maximal amount of proteasome binding to ubiquitinated E6AP with a K1/2 of approximately 16 nM (Data not shown). This binding was also salt sensitive but after 30 min at 37°C the association was salt-resistant (Fig. S3A). Thus, the direct binding of Ub conjugates and the binding mediated by hHR23A appear to eventually merge into one pathway leading to substrate degradation.
In addition to ubiquitination, the efficient degradation of a substrate by the 26S requires an unfolded or easily unfolded domain in the protein (Prakash et al., 2009; Zhao et al., 2009). We therefore tested if the structure of the ubiquitinated protein is important for its initial binding or its subsequent transition to tighter binding. We used in this assay immobilized Ub5-DHFR, because in the presence of the inhibitor methotrexate (MTA) DHFR assumes a tightly folded conformation which prevents degradation by the proteasome (Johnston et al., 1995; Prakash et al., 2009). Using methotrexate, we could test whether assuming the stable conformation affects the association of 26S with the ubiquitinated DHFR. The cofactors required for binding of proteasomes to its short Ub chain at 4°C was very similar to those for ubiquitinated E6AP and Nedd4. Importantly, this binding was not affected by the presence of MTA (Fig. 4A). Thus, the 26S initially binds to Ub chains apparently independently of the folding of the attached protein.
We then studied the possible influence of DHFR's conformation on the tight (i.e. salt-resistant) binding seen at 37°C in the presence of ATP. The presence of MTA during this incubation reduced the fraction of proteasomes that became tightly bound to the Ub5-DHFR (Fig. 4B). Identical results were obtained when binding was assayed using Western Blots to exclude the possibility that the MTA affected the activity assay (Fig. 4C). Thus, the transition to the tightly bound state is impaired if the protein lacks easily unfolded domains.
It is noteworthy that at this time there was no significant degradation of the Ub5-DHFR, and the Ub chain was still intact (Fig. 4C). In analogous experiments with poly-Ub-E6AP, we observed that poly-Ub-E6AP became tightly bound at this time with only minimal deubiquitination (not shown). Thus, this ”commitment step” precedes the removal of the Ub chain and significant translocation of the substrates into the 20S. Accordingly, inhibition of the 19S associated DUBs with Ub-vinylsulfone (which inhibits Usp14 and Uch37) (Fig. S3B) or ortho-phenanthroline (which inhibits Rpn11)(Fig. S3C) does not impair the transition to tighter binding.
Even in the absence of ubiquitination, the 26S also degrades unfolded or structureless polypeptides, such as casein, which most likely binds to the 19S ATPases, since their degradation is ATP-stimulated, and homologous AAA-ATPases (e.g. PAN from archea and ClpA, HslU or Lon from E. coli) bind such unfolded proteins (Benaroudj et al., 2003). To determine if the loosely folded domain required for tight binding of DHFR involves similar interactions with the 26S, we studied the effects of high concentrations of casein on this association. Although casein addition did not reduce the initial binding to poly-Ub-E6AP at 4°C, it markedly decreased the amount of 26S particles that became tightly bound at 37°C (Fig. 4D). In addition, this tighter binding step could be inhibited by high concentrations of nonubiquitinated DHFR, which did not influence the binding at 4°C. However, after a preincubation with methotrexate, DHFR could no longer inhibit the transition of the Ub conjugate to tighter binding (Fig. 4E). Thus, high concentrations of even a loosely folded region of a protein can efficiently inhibit the tighter binding, though it cannot do so when in a tightly folded conformation. In the transition to stronger binding, a domain of the ubiquitinated substrate apparently binds to the same site as an unfolded polypeptide, which is most likely on the 19S ATPase subunits.
To further test if the tighter binding leads to proteolysis, we analyzed the binding of the 26S to ubiquitinated Barstar (Ub4-Barstar) with and without an unstructured region (Ub4-USR2-Barstar), because Matouschek and coworkers showed that only ubiquitinated Barstar containing the unstructured region can be degraded by the 26S (Prakash et al., 2009). Although the initial binding of the two species was indistinguishable, only the conjugate containing the unstructured region could become tightly bound (Fig. 4F). Thus, the requirement for an unstructured or easily unfolded domain for tight binding to the 26S can account for the similar requirements for proteasomal degradation.
This tightly bound state differs in an additional way from the initial binding, which is solely dependent on the Ub chain since it can be blocked by free Ub chains (Fig. 1C) or the UIM domain of S5a (Fig. 3F). The tighter binding step appears to be independent of the Ub chain. Following conjugate binding at 4°C the transition to tight binding could not be blocked by high concentrations of the UIM domain of S5a (Fig. 3F), which removes the great majority of Ub conjugates that associated with newly isolated 26S particles (Besche et al., 2009).
The present studies provide a number of insights concerning the association of ubiquitinated substrates with the proteasome and the subsequent steps leading to their degradation. These findings were made possible by the development of a rapid, highly specific and quantitative method to assay the initial binding of ubiquitinated proteins to the 26S. A key difference from previous assays is that we use low temperatures (4°C) to prevent further proteasome-associated reactions (e.g. unfolding, deubiquitination). It is noteworthy that the purified 26S binds K48 and K63 Ub chains with similar high affinities (K1/2 ~30nM), even though in vivo a K63 chain do not target proteins to the proteasome (Newton et al., 2008). The present findings, however, are in accord with prior reports that attachment of K63 chains to a substrate can support its efficient degradation by pure proteasomes (Hofmann and Pickart, 2001; Kim et al., 2007). Therefore, in cells, some unknown factors must prevent this high affinity association of K63 chains with the 26S.
Our findings also extend the prior conclusion that the 19S particle contains two primary “receptor” proteins for Ub conjugates, Rpn10 and Rpn13 (Husnjak et al., 2008). As shown here, Rpn10 and Rpn13 contribute equally to the high affinity binding site for the Ub chain (K1/2 was reduced 2-fold upon inactivation of either). Accordingly, recent NMR findings suggested that isolated Rpn10 and Rpn13 molecules can bind simultaneously to the same Ub chain (Zhang et al., 2009). Our studies also demonstrated another conjugate binding site with 4-fold lower affinity in the proteasome. The existence of such an additional site had been proposed to explain why the deletion of rpn10, rpn13, rad23, dsk2 and ddi1 is not lethal in S. cerevisiae (Husnjak et al., 2008). Though dispensible, the high affinity site is probably important for the efficient binding of ubiquitinated substrates, which then may be translocated to the lower affinity site leading to degradation.
An important finding is that ATP binding to the 19S ATPases stimulates the association of ubiquitinated proteins with both high affinity and low affinity sites. With Rpn10, Rpn13 and the low affinity site, binding was maximal in the ATP bound conformation, as shown with ATPγS, and minimal with ADP bound. These effects of nucleotides closely resemble findings on the role of the ATPases in stimulating gate-opening in the 20S particle to allow substrate entry (Smith et al., 2005). This step is also activated maximally with ATPγS and is not supported by ADP. Thus, ATP binding affects multiple aspects of the 19S structure and not just the conformations of the six ATPase subunits. Presumably, the ability of ATP to promote conjugate binding as well as gate-opening allows the coordination of these processes to promote efficient proteolysis.
Although Rpn10 and 13 lack nucleotide binding sequences, Rpn10 has been shown to directly interact with the ATPase subunit, Rpt3. Since both Rpn10 and 13 appear to form a single binding site, Rpn13 is most likely also located in close proximity to the ATPase ring (Davy et al., 2001; Nickell et al., 2009; Zhang et al., 2009), and thus structural changes in the ATPase subunits upon ATP binding and hydrolysis probably alter the conformations of Rpn10 and 13. Interestingly, the magnitude of the stimulation of conjugate binding by ATP and ATPγS was consistently greater in cells lacking Rpn10 or Rpn13 (especially in the double mutant) than in the wt 26S particles. Perhaps the remaining low affinity binding site in Rpn10 mutants is the ATPase subunit, Rpt5, which has been reported to crosslink to Ub chains (Lam et al., 2002), and nucleotide binding to Rpt5 directly affects its conformation.
Presently, it is only possible to measure conjugate binding in the two extreme states when either ADP or ATPγS is bound. However, both these states are transitory ones and during protein degradation, bound ATP is rapidly hydrolyzed to ADP, which then exchanges with new ATP molecules. Consequently, in vivo the 19S particle must be repeatedly cycling between ATP and ADP-bound states, which clearly differ in affinity for Ub conjugates. Possibly, these ATP-driven dynamic changes in affinity for Ub chains might facilitate its release and diffusion from the initial Rpn10/Rpn13 binding site to the lower affinity site and eventually to the DUBs. Furthermore, because ATP and ADP probably bind to the six different ATPase subunits in a asynchronous cyclic fashion, these ATP-driven changes in conjugate affinity may affect the different Ub binding sites asynchronously (e.g. if they reduce the affinity of the Rpn10/Rpn13 site while enhancing the affinity of the second site, it could provide a mechanism for efficient chain diffusion between sites).
These studies have uncovered a key step in the handling of ubiquitinated proteins by the proteasome. After the initial easily reversible binding, the association of the conjugate with the 26S becomes tighter (i.e. salt-resistant). This step appears to commit the substrate to degradation and probably reflects a quality control mechanism in which Ub conjugates that cannot be partially unfolded or translocated are released. The step to tighter binding requires the presence of a loose domain on the substrate. Such proteins as well as ones that are inherently structureless, partially misfolded or damaged postsynthetically should be efficiently hydrolyzed. By contrast, proteins containing only tightly folded globular domains that would resist unfolding and translocation into the 20S (Prakash et al., 2009) are probably important to release from the 26S, because their continued association could prevent the degradation of other substrates.
These observations nicely account for the findings by Matouschek, Coffino and colleagues that efficient degradation by the 26S (e.g. Ub5-DHFR or Ub4-Barstar) requires an unstructured site in the protein (Prakash et al., 2009; Takeuchi et al., 2007). In vivo, the proteasome also does not degrade ubiquitinated DHFR that is stabilized through association with methotrexate (Johnston et al., 1995). The ligand-stabilized Ub5-DHFR initially associates with the 26S through its Ub chain as tightly as the ligand-free, degradable form, but it can't proceed to the more tightly bound state. Similarly Ub4-Barstar without an unfolded domain could not be tightly bound or degraded by the 26S. Thus, the ability to undergo this transition correlates with susceptibility to proteolysis.
Although a loosely folded domain on a substrate is critical for the tighter association and rapid degradation, not every protein substrate contains such a domain. In fact, the presence of such features on regulatory proteins may have evolved to allow their rapid clearance, while their absence may be a feature of long-lived polypeptides. Recent studies indicate that the p97/VCP ATPase complex can act on such tightly folded ubiquitinated proteins and facilitate their degradation by the 26S (Beskow et al., 2009). In this role p97 with its associated cofactors may function to expose domains that allow their tighter binding to the proteasome. In any case, highly stable ubiquitinated proteins are probably continually being released from the 26S in vivo with Ub chains attached for further destabilization by the p97 complex or deubiquitination in the cytosol.
It is noteworthy that after the transition to the tightly bound state, the Ub chain on the substrate is still largely intact, but it does not appear to be essential for the tighter binding which, unlike the initial binding, is not blocked by an excess of the UIM domains. Furthermore, inactivating the 19S-associated DUBs does not inhibit this transition (Fig. S3B, C). Therefore, conjugate disassembly must occur at a later step in the degradation process. Recently, we showed that the binding of Ub conjugates to Usp14/Ubp6 activates gate-opening in the 20S and thus facilitates polypeptide degradation (Peth et al. 2009). When conjugates were initially bound at 4°C and then switched to 37°C, the time lag until tight binding and gate-opening reached their maximum were very similar (about 20 - 25 minutes)(Fig, 3D). Presumably, during this period, there is an ATP-dependent structural rearrangement in the 26S (or the Ub conjugate) allowing the 26S to hold the substrate more tightly and to avoid its release when the Ub chain is removed. Because this tighter binding precedes deubiquitination, it may even be a prerequisite for possible editing of the Ub chain by DUBs (e.g. Usp14) or E3s (e.g. Hul5), which have been shown to regulate substrate degradation (Crosas et al., 2006). The existence of this further editing step would also imply that this commitment step does not necessarily lead to proteolysis. Such a commitment step that follows initial binding of a Ub conjugate and precedes its deubiquitination has been hypothesized previously (Verma et al., 2002). The present findings confirm this key biochemical transition and directly link it to the requirement of loosely folded domains in the substrate for degradation.
It is very likely that during the transition to the committed state, the easily unfolded domains bind directly to the ATPases, since the addition of casein or ligand-free DHFR inhibited this step. The 19S ATPase ring, like its evolutionary precursor, the PAN ATPase, has an intrinsic affinity for unfolded proteins, such as casein, which can be degraded in an ATP-stimulated process without ubiquitination (Tanaka et al., 1983). Unlike the initial binding of conjugates, which is supported better by ATPγS than by ATP, the transition to tighter binding requires ATP and does not occur at low temperatures. Thus, it seems most likely that this step involves an association of the easily unfolded domain with the ATPases and probably a partial ATPase-driven translocation of the polypeptide into the ATPase ring.
In summary, our findings have dissociated a specific sequence of events occurring on the 26S during the degradation of a ubiquitinated protein. 1) The Ub chain first binds to a site formed by Rpn10/Rpn13, when these proteins are in their high affinity conformation, which requires ATP binding to the ATPases. 2) Subsequently, a loosely folded domain of the polypeptide interacts with a site probably on the ATPase subunits, which has an affinity for unstructured proteins. ATP-hydrolysis then leads to a tighter association with the 26S and further commits the substrate on the path for degradation. 3) During this process, a transfer of the Ub conjugate to a lower affinity site seems likely, leading in turn to interaction with the DUBs and perhaps chain editing. The association of the Ub chain with the Usp14 active site leads to either shortening or removal of the chain, and simultaneously enhances gate opening in the 20S, facilitating polypeptide entry. 5) Concomitantly, the ATPase subunits unfold and translocate the protein through the fully opened gate into the 20S particle for processive degradation.
26S particles were purified from rabbit muscle in the presence of 150 mM NaCl or from yeast strains without addition of NaCl as described before (Besche et al., 2009; Peth et al., 2009). Ub conjugates were obtained after auto-ubiquitination reactions of purified GST-tagged E3 ligases (Kim et al., 2007). After auto-ubiquitination, the Ub conjugates were washed five times with 50 mM Tris/HCl (pH 7.5), 100 mM NaCl, 10% Glycerol, and 1 mM DTT. The Ub5-DHFR was a kind gift from Millennium Pharmaceuticals (Cambridge).
10 nM of 26S proteasomes were incubated with poly-Ub conjugates (30 nM) in the presence of 25 mM Hepes-KOH, pH 8, 125 mM Potassium Acetate, 1 mM DTT, 2.5 mM MgCl2, 0.1 mg/ml BSA, 0.05% Triton X-100 and 1mM ATP. The binding reaction takes place while rotating for 30 minutes at 4°C, unbound 26S particles were removed by subsequent washing with 400 μl of binding buffer (without BSA) and 1 ml of 50 mM Tris/HCl pH 7.5, 10 mM MgCl2, 1 mM DTT and 1 mM ATP.
For the tighter binding of 26S particles to poly-Ub conjugates, 26S proteasomes were pre-bound at 4°C as described above and then washed once with 400 μl of 1x binding buffer. At this stage, 26S particles can be removed by washing with 1 ml of 50 mM Tris/HCl, pH 7.5, 10 mM MgCl2, 300 mM NaCl, 1 mM DTT and 1 mM ATP. After a 30-minute incubation at 37°C, bound 26S particles are resistant towards washing with the high salt washing buffer.
Immobilized Ub5-DHFR were generated after binding of the biotin-tagged Ub5-DHFR to a Streptavidin-resin overnight at 4°C. Unbound Ub5-DHFR was removed by washing five times with 50 mM Tris/HCl (pH 7.5), 100 mM NaCl, 10% Glycerol, and 1 mM DTT. Binding of 26S particles to Ub5-DHFR was measured as described for heterogeneous poly-Ub conjugates..
Activity of 26S particles was measured with 50 μM of the fluorogenic substrate LLVY-amc (Bachem, Switzerland) in 50 mM Tris/HCl, pH 7.5, 10 mM MgCl2,40 mM NaCl, 1 mM ATP and 1 mM DTT. Cleavage of the peptide was monitored at λex 380 nm; λem 460 nm at 37°C for mammalian or 30°C for yeast 26S proteasomes. The rates of proteasomal activities were calculated 30-60 minutes after the start of the reaction and expressed as U/min (AU).
The yeast strains sub62 (WT), sub545 (α3-myc) sub544 (α3ΔN), sDL66 (rpn11-proA) and DY65 (rpt5K228S) were kindly provided by Dan Finley (Harvard Medical School) as were the rpn10ΔUIM, rpn13KKD and rpn10ΔUIM/rpn13KKD strains, which were transformed to generate strains with Rpn11-ProA tagged 26S particles (Knop et al., 1999).
The authors are grateful to Mary Dethavong for her expert assistance and to Dan Finley and Suzanne Elsasser for valuable advice and assistance in generating yeast strains. Ubiquitinated Barstar constructs were kindly provided by Andreas Matouschek. The authors also thank Henrike Besche, James Nathan and Eugene Drokhlyansky for helpful input on this manuscript. The studies were supported by a grants from the NIGMS and Multiple Myeloma Foundation to A. L. G. and a fellowship to A.P. from the Deutsche Forschungsgemeinschaft.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.