Dependence of Proteasome Processing Rate on Substrate Unfolding*

Protein degradation by eukaryotic proteasomes is a multi-step process involving substrate recognition, ATP-dependent unfolding, translocation into the proteolytic core particle, and finally proteolysis. To date, most investigations of proteasome function have focused on the first and the last steps in this process. Here we examine the relationship between the stability of a folded protein domain and its degradation rate. Test proteins were targeted to the proteasome independently of ubiquitination by directly tethering them to the protease. Degradation kinetics were compared for test protein pairs whose stability was altered by either point mutation or ligand binding, but were otherwise identical. In both intact cells and in reactions using purified proteasomes and substrates, increased substrate stability led to an increase in substrate turnover time. The steady-state time for degradation ranged from ∼5 min (dihydrofolate reductase) to 40 min (I27 domain of titin). ATP turnover was 110/min./proteasome, and was not markedly changed by substrate. Proteasomes engage tightly folded substrates in multiple iterative rounds of ATP hydrolysis, a process that can be rate-limiting for degradation.

Protein degradation by eukaryotic proteasomes is a multistep process involving substrate recognition, ATP-dependent unfolding, translocation into the proteolytic core particle, and finally proteolysis. To date, most investigations of proteasome function have focused on the first and the last steps in this process. Here we examine the relationship between the stability of a folded protein domain and its degradation rate. Test proteins were targeted to the proteasome independently of ubiquitination by directly tethering them to the protease. Degradation kinetics were compared for test protein pairs whose stability was altered by either point mutation or ligand binding, but were otherwise identical. In both intact cells and in reactions using purified proteasomes and substrates, increased substrate stability led to an increase in substrate turnover time. The steadystate time for degradation ranged from ϳ5 min (dihydrofolate reductase) to 40 min (I27 domain of titin). ATP turnover was 110/min./proteasome, and was not markedly changed by substrate. Proteasomes engage tightly folded substrates in multiple iterative rounds of ATP hydrolysis, a process that can be ratelimiting for degradation.
To degrade folded proteins, chambered protease complexes unfold substrates in an energy-dependent manner that requires two components (1). The first, a regulatory complex, recognizes substrates and initiates their processing. The second, an associated proteolytic complex, contains sites at which peptide bonds are hydrolyzed. These sites are present in a closed chamber sequestered from the general cellular environment. The proteolytic chamber is accessed through a pore that excludes folded protein domains but accommodates an unfolded or unstructured polypeptide (2,3). Regulatory and catalytic complexes must thus collaborate to degrade native proteins: the regulatory complex actively unfolds substrates containing structured domains and translocates them into the catalytic complex. Substrate unfolding and translocation by bacterial ATP-dependent proteases has been extensively studied. It was found that substrate unfolding can be rate-limiting for degradation and that increased mechanical stability of substrates prolongs degradation (4). Is that also true of eukaryotic protea-somes? Studies from our laboratory (5,6) and by others (7)(8)(9)(10) are consistent with this conclusion, but studies using purified proteasomes and substrates of well-defined structure have been limited and have not determined the kinetic parameters associated with proteasome action.
There are several requirements for rigorously performing such an analysis. Homogeneous substrates must be available that differ in their resistance to unfolding, but are otherwise identical in structure and proteasomal interaction. Most substrates are designated for destruction by the conjugation of ubiquitin chains, which provide a high affinity tag for proteasome association, but some substrates utilize alternate tags to designate degradation (11,12). Capture of ubiquitin chains and their processing can significantly alter substrate processing rates (13). To avoid these processes as a variable that could potentially confound kinetic analysis, we designed substrates that are not dependent on ubiquitin conjugation. We could thus focus effectively and specifically on the following question: Is substrate domain stability a significant determinant of proteasome action? We examined the effects of folding stability on turnover using two folded domains, the I27 module of the titin protein and mammalian dihydrofolate reductase (DHFR). 2 Using site-directed mutation or a tight-binding ligand, we altered the stability of each. In both cases we observed a significant increase in turnover time associated with stabilization.
Protein Purification-Substrates containing an Rpn10 domain were expressed in Escherichia coli BL21 Rosetta strains (Novagen) as N-terminal hexahistidine-tagged proteins. Unlabeled forms were induced when host strains reached A 600 ϳ1 with 0.5 mM IPTG at 30°C overnight (16 -18 h). 35 S-radiolabeled substrates were prepared and purified as follows. Cells were grown to A 600 ϳ0.7 in LB medium containing 30 g/ml kanamycin and 34 g/ml chloramphenicol. After harvesting, cells were resuspended to A 600 ϳ3 in M9 medium supplemented with 0.063% methionine assay medium (Difco #242310), 0.4% glucose and 30 g/ml kanamycin and incubated at 37°C for 1 h. 1 mM IPTG was then added. After 30 min incubation at 30°C, [ 35 S]methionine and [ 35 S]cysteine (as EasyTag protein labeling mix (Perkin Elmer)) were added to 14 Ci/ml as well as 34 nM of unlabeled methionine and incubation continued for 1 h. The cells were then harvested and the pellet frozen at Ϫ80°C. Labeled and unlabeled proteins were purified using Talon metal affinity resin (Clontech) according to the manufacturer's instructions. Radio-labeled proteins were eluted from Talon resin and concentrated on 10,000 molecular weight cutoff Centricon concentrators (Amicon Ultra filter devices). Substrate buffer was exchanged on BioSpin chromatography columns (Bio-Rad) into HRV Cleavage Buffer (25 mM Tris, pH 8.0, 150 mM NaCl, 2 mM DTT, 10% glycerol). After elution, the unlabeled proteins were dialyzed into HRV Cleavage Buffer. Vector-derived N-terminal sequences were removed from all substrate proteins by cleavage with 100 units HRV protease per mg of substrate protein (Turbo 3C Protease, Accelagen) at 4°C for 24 h. HRV protease and residual uncleaved substrate proteins were removed by subsequent incubation with Talon resin.
Proteasome Purification-26 S proteasomes were purified from strains YAH96 or YAH106 in which a 3xFLAG tag has been appended to the C terminus of the Rpn11 subunit of the proteasome. Cells were grown to an A 600 of 3-4 in 1.5 liters of YPD, harvested, washed once with 50 ml of ice-cold water. The cell pellet was then resuspended in an equal volume of Proteasome Buffer (PB: 50 mM Tris pH 7.5, 25 mM NaCl, 5 mM MgCl 2 , 2 mM ATP, 10% glycerol) supplemented with 2ϫ Complete EDTA-free protease inhibitor mixture (Roche Diagnostics) and lysed by five passes through a Microfluidics microfluidizer at a pressure of 100 psI. Lysates were clarified by centrifugation at 20,000 ϫ g at 4°C for 30 min. 26 S proteasomes were affinitypurified by incubation with M2 anti-FLAG affinity gel (Sigma) for 1 h at 4°C. Bound proteasome complexes were washed with 200 column volumes of PB and eluted from the resin overnight at 4°C in an equal volume of PB supplemented with 0.2 mg/ml 3xFLAG peptide (Sigma). The eluates were concentrated on 10,000 molecular weight cuttoff Centricon concentrators. Proteasome composition was determined by denaturing gel electrophoresis followed by Coomassie staining. Complex assembly was assessed by native gel electrophoresis (18). Proteasome molarity was calculated using a molecular weight 2,400 kDa.
In Vitro Degradation Assay-Degradation reactions containing radiolabeled protein substrates were carried out at 30°C by 50 nM 26 S proteasome in a buffer (25 mM HEPES pH 7.5, 100 mM KCl, 20 mM MgCl 2 , and 10% glycerol) containing 2 mM DTT, 5 mM ATP, and an ATP regenerating system (3 mM phospho(enol)pyruvic acid, 2 mM NADH, and 0.25 units of pyruvate kinase/lactate dehydrogenase mix (Sigma)). Substrates were present at the concentrations specified in the text. Degradation was initiated by substrate addition and time-dependent degra-dation was assessed by periodically removing 10 l volumes from the reaction and precipitation in 140 l of 20% TCA for 30 min on ice. After 30 min centrifugation at 18,000 ϫ g, the supernatant, containing the peptide products of proteolysis, was added to a scintillation vial for to quantification of radioactivity. For each experiment, the total radioactivity present in substrate was determined by adding an aliquot of the degradation assay mixture to TCA, transferring the solution to a scintillation vial and counting. Kinetic parameters were determined using the Michaelis-Menten equation by fitting the data to a hyperbolic function, using the SigmaPlot software package. For SDS-PAGE analysis of degradation, aliquots were periodically removed from the degradation assay reaction. Samples were resolved on a 12% SDS-PAGE and the distribution of radiolabeled proteins on the dried gel determined by autoradiography. Signals from autoradiographic films were quantified using TotalLab software (Nonlinear Dynamics).
ATPase Assays-Proteasome ATPase activity was measured in a coupled assay (19,20) in 25 mM HEPES (pH 7.5), 100 mM KCl, 20 mM MgCl 2 , 10% glycerol, 250 mU/ml LDH/pyruvate kinase (Sigma), 7.5 mM phospho(enol)pyruvic acid, 1 mM NADH, 2 mM DTT, and 5 mM ATP. Proteasome concentration was 30 nM and, when present, substrates or other proteins containing an Rpn10 moiety were 1.5 M. ATPase activity was followed by loss of NADH absorbance at 340 nm. ATP consumption was calibrated using an NADH standard curve, performed with each assay. Reactions were carried out in 96-well plates in a Molecular Devices Spectra Max spectrophotometer. Rates were calculated from least squares fits of each time series. Data were collected separately for: 1) each substrate or pseudosubstrate alone or 2) while co-incubating each substrate or pseudo-substrate with proteasomes or 3) for proteasomes alone. Proteasome-specific ATP hydrolysis rates were calculated as (proteasomes ϩ substrate rate) Ϫ (substrate-only rate).

Design and Structure of Substrates Used in These Studies-
Our previous studies of the native ubiquitin-independent proteasome substrate ornithine decarboxylase (ODC) provided the basis for the design of the substrates we generated and utilized. No post-translational modification of ODC is needed for its proteasomal degradation; in particular, ubiquitin is not involved (21). Vertebrate forms of ODC have a small conserved degradation tag (22) (37 amino acids in mouse and human) at its C terminus (cODC). Degradation is strictly unidirectional and begins at the tag (5). Fusing this tag to other proteins promotes their rapid turnover by the eukaryotic proteasome (17), including those of mammals, plants, and fungi.
Proteasomal degradation, whether dependent on (23) or independent of ubiquitin (24), requires two distinct substrate properties: proteasome association and an unstructured element at which degradation begins. Both of these functions are provided by the cODC degron. Mutating a critical cysteine residue (C441) in the mouse cODC sequence prevents proteasome association (25). However, proteins stabilized by a Cys-441 mutation could be restored as proteasome substrates by appending them to the Rpn10 subunit of the proteasome (24). Such fusions bind at the proteasome site which normally accommodates Rpn10. Once tethered in this way these chimeric proteins were degraded only if they also provided an unstructured element of some minimal length. Here we used similarly designed fusions to deliver domains of varying mechanical stability to the proteasome within a uniform context, and thus determined the effect of this parameter on turnover. Fig. 1 shows diagrammatically the composition of the substrates used in these studies.
Titin I27 Degradation in Yeast Cells-The domain we first utilized for analysis of unfolding was the I27 module of the giant titin protein, which is composed of more than two hundred individually folded domains connected by unstructured linkers.
I27 provides several advantages. Its mechanical unfolding has been extensively investigated by biophysical methods and is among the most stable proteins that have been studied by these means (26 -28). It thus presents a formidable challenge to a biochemical unfoldase. Importantly, single residue mutations have been characterized that do not significantly alter the structure of I27 but do reduce the force required for mechanical unfolding. We compared wild-type I27 and a destabilized mutant containing a single residue mutation, V13P. Compared with the wild-type form, I27 V13P requires less pulling force for unfolding (29).
We first asked whether proteasomes could degrade I27 in living cells and whether the V13P mutation altered turnover. Because the ubiquitin independent cODC tag can direct degradation in yeast (17), we appended that degron to the C terminus of I27 and I27 V13P (Fig. 1) and expressed these proteins in Saccharomyces cerevisiae. Cells were incubated with radio-labeled amino acids and pulse-chase analysis used to measure turnover (Fig. 2, A and B). During a 1-h chase period more than 90% of the labeled I27 V13P -cODC was degraded during the first 30 min of the chase, and its turnover appeared to be first order, with a half-life of about 6 min. At the 60 min time point, residual protein was below our detection limit. In marked contrast, the wild-type I27 form showed little degradation during the 60 min chase period; at most, 20 -30% of the labeled I27-cODC present at the beginning of the chase disappeared. We conclude that, as for other proteins, the cODC degron can direct degradation of I27, and that a destabilizing mutation significantly accelerates its turnover.
Although cODC is an effective degron in vivo, it has not in our hands proven reliably effective for use with purified yeast  proteasomes. For biochemical analysis we therefore chose an alternative means to target substrates to yeast proteasomes, which depends on fusion to the Rpn10 protein, as previously described (24) and further developed here. Before exploiting that approach in vitro we tested it in yeast cells, again asking whether these fusions provided an effective delivery vehicle for I27 and whether the wild type and mutant form of the folded domain were degraded differently. R-I27 was expressed in yeast, with or without the V13P mutation. As a control for specificity, these were expressed either with or without an unstructured C-terminal extension (see Fig. 1 for diagram of substrate structures). This consisted of cODC C441S , which has been shown to support proteasomal degradation of tightly folded Rpn10-tethered proteins in yeast (24). Pulse-chase analysis was used, as above, to assess turnover (Fig. 2, C and D). Without a C-terminal extension we observed no degradation of either wild-type I27 or its destabilized V13P mutant form. When the extension was present, both forms of I27 were degraded. However, the rate and extent of degradation of the two was very different, and closely matched the results observed using the cODC degradation tag. Only 20 -30% of labeled substrates containing wild-type I27 were degraded during a 2 h chase. In contrast the I27 V13P substrate that was otherwise identical, but for a single residue mutation, was degraded with a half-life of about 15 min, and none was detected after 2 h. These data show that distinct means of targeting substrates to proteasomes (cODC, a native eukaryotic degron, and fusion to Rpn10, an endogenous mechanism for delivery of ubiquitin-conjugated proteins (30,31)) give concordant results in yeast cells. We therefore determined unfolding kinetics of Rpn10 tethered substrates by yeast proteasomes in vitro.
Titin I27 Degradation by Purified Proteasomes-Because in vitro experiments can be performed with purified components at known concentrations, they make it possible to calculate kinetic parameters, such as absolute turnover rates, rather than relative rates, as in the in vivo experiments just described. We therefore affinity purified yeast proteasomes and generated recombinant substrates in bacteria with the same structure (Rpn10-"folded domain"-extension) as those used with intact yeast cells. Purified proteasomes were analyzed by SDS-PAGE and by native gel electrophoresis (supplemental Fig. S1). These data confirmed that proteasomes had the expected polypeptide composition and that Ͼ95% were doubly capped, with two 19 S regulatory complexes per 20 S proteolytic core complex. SDS-PAGE confirmed that the substrates were homogeneous and had the expected electrophoretic mobility (supplemental Fig. S1).
We initially analyzed a substrate pair containing the I27 domain or its V13P mutant; each produced with or without a C-terminal extension (see Fig. 1). We monitored degradation in reactions with 50 nM 26 S proteasomes and 250 nM radio-labeled Rpn10-tethered proteins, measuring the production of acid soluble radiolabeled peptide products (Fig. 3A). (In initial experiments we compared isogenic Rpn10 and rpn10⌬ proteasomes; the results were similar (Ref. 24 and supplemental Fig.  S2). We subsequently used only wild-type proteasomes). Timedependent progress curves were qualitatively similar to the data observed with the cognate substrates in vivo. No degradation occurred in the absence of a C-terminal extension. Both I27 substrates with extensions were degraded, but the rate and extent of destruction of the destabilized mutant was about 8 -10-fold greater. These results show that relative degradation rates in our in vitro system match those observed in intact cells.
DHFR-We next asked whether these observations and conclusions could be extended from I27 to other proteins. Mammalian dihydrofolate reductase (DHFR) is well-suited for testing the generality of our findings. DHFR, like I27, has a well characterized structure, the stability of which can be experimentally altered. The antimetabolite methotrexate binds tightly to DHFR and stabilizes its structure, a conclusion supported by numerous biochemical (7,(32)(33)(34) as well as biophysical (35) studies. We investigated the degradation of Rpn10tethered DHFR by purified proteasomes (Fig. 3B). Again we compared a series of substrates: an extension was present or absent, and methotrexate was present in or omitted from the reaction. The results, although similar to those seen with I27, were quantitatively different. An extension was not an absolute requirement for degradation, but its presence accelerated degradation. Regardless of the presence or absence of an extension, the stabilizing ligand methotrexate slowed degradation. For both DHFR and I27 tethered substrates, we performed two tests of degradation specificity. First we asked whether the Rpn10 moiety of substrates determines proteasome association by adding a fifty-fold excess of Rpn10-GFP as competitor. (Rpn10-GFP can functionally replace Rpn10, but is not an active substrate because it lacks an unstructured extension (24)). The rate of degradation of substrates was reduced more than 10-fold by Rpn10-GFP (supplemental Fig. S3), demonstrating that Rpn10 binding is required for substrate activity in this system. We also tested whether the proteasome rather than an adventitiously co-purified protease was responsible for substrate degradation by preincubating reactions with the highly specific proteasome inhibitor epoxomicin (36). Epoxomicin (200 M) reduced degradation by more than 95% compared with control reactions without the inhibitor.
Substrate Degradation Kinetics-We performed kinetic analysis using proteasomes and the substrates described above. For all substrates tested, degradation rates were determined using 50 nM proteasomes and reaction conditions that produced linear initial velocities (representative primary data is shown in supplemental Fig. S4). The resultant substrate concentrationversus-velocity data were well-approximated by a hyperbolic function, a result descriptive of Michaelis-Menten kinetics (Fig.  4). Despite their great structural complexity, proteasome activity is approximated quite well by classical Michaelis-Menten kinetics, a property commonly observed with simpler enzymes. Inspection of the kinetic parameters derived from curve fitting (Table 1) reveals that the four substrate conditions yielded broadly similar K D values, indicating that the variable C-terminal domain in these substrates does not significantly perturb the Rpn10-proteasome interaction. In contrast, the V max values of the various substrates were more diverse. The V max values exhibit a 4.8-fold faster turnover rate of the destabilized I27 V13P substrate compared with its wild-type counterpart and a 2.7fold slower turnover rate of the DHFR substrate when associated with a stabilizing ligand.
Proteasome ATPase Activity-We also determined the ATP hydrolysis activity of proteasomes and asked whether substrate processing affected that activity ( Table 2). The rate of ATP turnover was ϳ110/proteasome/min. and varied modestly among several independently prepared lots. The effect on ATP hydrolysis of incubating proteasomes with Rpn10-tethered substrates or with pseudo-substrate controls (those lacking a C-terminal extension) was examined. Reactions were carried out under the same conditions as used for degradation, but the added proteins were present at 1.5 M, 5-10-fold above the substrate K m values. The measured rates therefore reflect ATPase activity in the presence of near-saturating substrate. The apparent stimulation of ATPase activity by substrates and pseudo-substrates was at best modest. The Rpn10 tethered proteins produced ϳ10 -30% stimulation above basal rates ( Table 2; see supplemental Fig. S5 for primary data). The modest stimulation of ATPase activity observed is close to the boundary of experimental error. Furthermore, this is unlikely to be physiologically significant for substrate degradation because of its small magnitude and the lack of systematic correlation of the extent of stimulation with the proper-ties of Rpn10-tethered proteins important for their degradation: domain stability and the requirement for an unstructured extension.
What Limits the Turnover Rate of More Slowly Degraded Substrates?-The kinetic rate parameter can be expressed as the catalytic constant, k cat , or as its reciprocal, the time required at steady state for a proteasome to complete the reaction cycle ( Table 1). The steady-state turnover time of the wild-type I27 substrate is 40 min. What is status of the substrate-proteasome pair during this prolonged period? It is likely that there are multiple iterations of unfolding attempts associated with cycles  of ATP hydrolysis, each with a low probability of success. One important question is whether substrate and proteasome remain continuously associated during this process. To test this, we initiated degradation of radio-labeled R-I27-ext, then after 30 min added a 10-fold excess of the same substrate, but in unlabeled form (Fig. 5, arrow). Before addition of the unlabeled competitor, radiolabeled substrate was present at a concentration of 250 nM, in excess of the K m value for this substrate, which is 145 nM. Hence proteasomes were more than half saturated. Addition of competitor immediately and markedly inhibited the subsequent generation of labeled proteolysis products. In contrast, under the control condition without competitor, production of labeled peptides continued unabated (Fig. 5). This result implies that the substrate must undergo a rapid on/off process. Excess competitor quickly re-equilibrates proteasomes with substrate that is 11-fold less radioactive and the generation of labeled proteolysis products slows commensurately. The rapid inhibitory effect of adding excess unlabeled substrate is inconsistent with stable substrate association and clearance of substrate from proteasomes solely by proteolysis: the mean time for that clearance mechanism exceeds 30 min. Therefore, the Rpn10-tethered substrate undergoes a rapid on/off process that may contribute to the measured duration of steady-state degradation.
Are Stability-dependent Rate Differences an Artifact of Interrupted Degradation?-Lastly, we considered whether non-processive degradation could have artifactually distorted the turnover times reported here. We (6,37) and others (8,38) have shown that a tightly folded protein domain can halt degradation, producing partial degradation products with termini containing short extensions of the folded domain. Such partial degradation products could act as competitive inhibitors. SDS-PAGE and autoradiography was used detect the production from R-I27-ext of terminally trimmed products. Reactions were performed as in Fig. 3A; incubation without addition of proteasomes was done in parallel to control for non-proteasomal degradation. We observed time-and proteasome-dependent disappearance of the substrate band (supplemental Fig. S6). Densitometry showed that less than 20% of the radiolabeled substrate initially present accumulated as intermediates; the rest was degraded to peptide fragments not resolved by SDS-PAGE and autoradiography. Therefore, both R-I27-ext and R-I27 V13P -ext are predominantly degraded processively and non-processive outcomes did not markedly alter the reported kinetic parameters.

DISCUSSION
We found mechanical stability of substrate domains to be an important determinant of degradation rate. Previous studies have shown this to be true for bacterial ATP-dependent proteases (39 -42). The literature supports a similar conclusion for proteasomes, but data are less systematic in approach or the experiments were performed using cellular extracts rather than with purified components. Using purified yeast proteasomes and a recombinant multidomain protein incorporating DHFR, A. Matouschek (10) observed methotrexate-induced stalling, a result consistent with slower unfolding of the ligand-stabilized domain. C. Pickart (43) along with our laboratory further investigated this question using mammalian proteasomes and found methotrexate impaired the degradation of DHFR domains targeted to the protease by either ubiquitin chains or the ODC C-terminal degron. Data in living cells also support the conclusion, that unfolding can be a rate-limiting step in proteasomal degradation (6), but the complex environment of the cell makes it difficult to confidently attribute such results to intrinsic properties of the proteasome.
In the present studies we used Rpn10 to tether substrates. Rpn10 normally functions as one of the integral substrate receptors of the proteasome. A significant fraction of Rpn10 is present free in the cytoplasm (44); it may thus shuttle between a free and bound state to deliver ubiquitin-conjugated proteins to the proteasome (45). Although Rpn10 is an integral protein, we find that proteasomes affinity purified from wild-type strains or from Rpn10 deletion strains have a similar capacity to degrade substrates that depend on Rpn10 for association (Ref. 24 and supplemental Fig. S3). This may reflect a dynamic association of Rpn10 with proteasomes, or its sub-stoichiometric presence in purified proteasomes. Rpn10-tethered proteins additionally require an unstructured C-terminal extension of adequate length but no apparent sequence specificity (24). We used as an extension element in the present experiments the  37-residue cODC degron, with a C441S mutation that renders it ineffective for proteasome association.
Within the Rpn10-"folded domain"-extension sandwich we embedded one of two proteins, the titin I27 protein module or human DHFR. For each of these we could manipulate folding stability, by a destabilizing point mutation (I27 V13P ) or by docking a tight binding antimetabolite ligand to DHFR to stabilize it. The effects of each of these alterations on the mechanical stability of their respective targets have been previously documented and utilized for various experimental purposes. Single molecule force versus extension studies using atomic force microscopy have determined the forces required for unfolding these target molecules. Under specific experimental extension regimes the force required to unfold I27 was reduced by the V13P mutation from 204 to 132 pN (29), whereas methotrexate association increased the unfolding force for DHFR from 27 to 80 pN (35). It should be noted that although such biophysical measurements of pulling force may be expected to correlate with force requirements for unfolding by an ATPase, the two processes differ in the site of pulling and the direction of the deforming force applied to a protein and these differences can alter its unfolding path and the required force (46).
Using the substrates described above, we found that domain stability affected the turnover rate both in vivo and in vitro. These biological and biochemical data therefore mutually support each other. Further, in vivo results whether obtained with cODC, a native degron, or by Rpn10 tethering were concordant, supporting the conclusion that results observed with the latter are not an artifact of exotic substrate design.
Determining kinetic parameters requires the use of a range of substrate concentrations, including concentrations sufficiently high to approach saturation of proteasome active sites. The high affinity of the Rpn10 tag greatly facilitated fulfilling this technical requirement. We found that the K m conferred by the Rpn10 tether is not dissimilar from that provided by ubiquitin chain tagging (43). The proteasome substrate-velocity curves we obtained are fit quite well using the classic Michaelis-Menten equation. However, the proteasome is a multisubunit complex, and several steps are logically required to degrade substrates: Rpn10 binding, association with the unstructured C-terminal moiety, translocation, protein unfolding, and peptide bond hydrolysis. That all these steps are captured by a relatively simple model of catalysis suggests that a single factor dominates binding (presumably Rpn10 association) and degradation (presumably unfolding). Under steady-state conditions, almost all the population of proteasome-associated R-I27-ext was exchangeable (Fig. 5). In apparent contrast to the substrates we employed, ubiquitinated proteasome substrates progress from a readily dissociable ubiquitin chain-dependent form of proteasome interaction to one of higher affinity that is less readily dissociated; progress to this second state requires both an unstructured extension and ATP hydrolysis (47).
To establish whether substrate degradation affects ATP activity, we measured the rate of ATP hydrolysis in the absence and presence of our substrate set. The proteasome hydrolyzes ATP at a rate of ϳ110 per minute (Table 2). This figure is 4-fold higher than the reported activity of purified mammalian proteasomes (48) and similar to that of its 19 S regulatory complex (49). If we assume that a doubly capped proteasome has two ATPase hexamers and that each of its 12 ATPase active sites fires at an equal rate, then each active site on average hydrolyzes more than 300 ATP molecules for each productive I27 unfolding event. It is important to note that wild-type I27 is not typical of substrates that a proteasome might encounter in cells, but is likely to approximate an extreme case of high mechanical stability. The low frequency of productive ATP hydrolysis events, with payoff odds reminiscent of a lottery, is therefore not intrinsically implausible.
We observed no significant change in proteasomal ATPase activity using the Rpn10-linked substrates and pseudo-substrates described here (Table 2). Failure to alter activity is surprising in light of the marked responsiveness of some bacterial ATP-dependent proteases to substrate processing events, e.g. (42). There are several possible explanations of the unchanging ATPase activity we observed. Although the proteasomes were biochemically active, regulatory components may have been lost during purification (50), despite our use of single step affinity purification in low salt buffer. Alternatively, evolution may have co-opted ATPase regulation in the service of ubiquitin processing, a function in part mediated by proteasome-associated proteins. If so, our use of substrates that are not conjugated to ubiquitin may have precluded observing ATPase modulation. Lastly, it is possible that proteasome ATPase activity is simply constitutive.
Perhaps the most revealing information we provide here are turnover times, the average time required for a proteasome to degrade a substrate molecule under saturating conditions at steady state. These ranged from about 5 min (DHFR, no methotrexate) to almost 40 min (wild type I27). How do these values compare with those in the literature? Published data providing kinetic parameters for proteasomes are sparse. From the data in Fig. 3a of Thrower et al. (43) one can infer for Ub5-DHFR and mammalian proteasomes a turnover time of about 20 min. The corresponding figure for R-DHFR-ext was 5 min in our experiments. The same figure cited (43) supports the conclusion that the addition of methotrexate reduces turnover by a factor of about 10, which may be compared with the 2.7-fold effect we observed using that ligand. Given the considerable difference in experimental approach (among these the use of mammalian versus yeast proteasomes), our results can be regarded as in reasonable agreement.
Substrate unfolding and degradation by the bacterial protease ClpXP has been studied using I27-ssrA, wild type, and V13P. Turnover times were respectively 4 min and 0.32 min (42). The corresponding values we obtained using Rpn10-tethered I27 were for wild type 40 min and for I27 V13P 9 min. The substrates used in the present investigations contained Rpn10 in addition to I27 (or DHFR). Rpn10 unfolding could obscure differences dependent on unfolding the I27 moiety. Alternately or additionally, there may be a kinetically significant step that follows translocation; the existence of such a terminal delay has been inferred by comparing ClpXP degradation kinetics in experiments using steady-state solution versus single molecule approaches (51). The presence of such a step could also blur rate differences associated with unfolding. Such a terminal delay might be found if the proteasome enters a prolonged quiescent state after completing its substrate processing cycle (52).
Our data do not resolve the time required for unfolding and translocation but do provide a lower bound on the former. For I27 steady state turnover time is about four times longer for wild type than for V13P. Therefore, at least three quarters of the overall processing time must be required for unfolding I27. For R-DHFR-ext, methotrexate increases steady state turnover time about 2.7-fold, and a like calculation implies that unfolding occupies at least 60% of the overall processing time of that substrate in association with methotrexate. The tethering mechanism used here requires that substrates contain a folded Rpn10 domain; the present experimental design therefore precludes presenting proteasomes with fully unfolded substrates. Our calculations of the fractional time devoted to unfolding therefore represent lower bounds.
Lastly, the approach we used provides a surrogate measure of unfolding and an indirect way to test the limits of proteasome unfolding power. More direct means for assessing these should be developed.