Nucleotidase activities of the 26 S proteasome and its regulatory complex.

The 26 S proteasome can be assembled from the multicatalytic protease (or 20 S proteasome) and a large multisubunit regulatory complex in an ATP-dependent reaction. The 26 S proteasome and its regulatory complex were purified from rabbit reticulocytes for characterization of their nucleotidase properties. Both particles hydrolyze ATP, CTP, GTP, and UTP to the corresponding nucleoside diphosphate and inorganic phosphate. The Km values for hydrolysis of specific nucleotides by the 26 S proteasome are 15 μM for ATP and CTP, 50 μM for GTP, and 100 μM for UTP; Km values for nucleotide hydrolysis by the regulatory complex are 2-4-fold higher for each nucleotide. Several ATPase inhibitors (erythro-9-[3-(2-hydroxynonyl)]adenine, oligomycin, ouabain, and thapsigargin) had no effect on ATP hydrolysis by either complex whereas known inhibitors of proteolysis by the 26 S enzyme (hemin, N-ethylmaleimide (NEM), and vanadate) significantly reduced ATP hydrolysis by both particles. Hydrolysis of all nucleotides was inhibited by 5 mM NEM but only GTP and UTP hydrolysis was significantly reduced at 50 μM NEM. The 15 μM Km for ATP hydrolysis by the 26 S proteasome is virtually identical to the observed Km of 12 μM ATP for Ub-conjugate degradation. Although nucleotide hydrolysis is required for protein degradation by the 26 S proteasome, nucleotide hydrolysis and peptide bond cleavage are not strictly coupled. Substrate specificity constants (kcat/Km) are similar for hydrolysis of each nucleotide, yet GTP and UTP barely supported Ub-conjugate degradation. Further evidence that nucleotide hydrolysis is not coupled to peptide bond cleavage was obtained using N-acetyl-leucyl-leucyl-norleucinal (LLnL). This compound inhibited peptide hydrolysis by the multicatalytic protease and Ub-conjugate degradation by the 26 S proteasome, but it had little effect on ATP or UTP hydrolysis by the 26 S enzyme.

The multicatalytic protease (MCP) 1 or 20 S proteasome is a major protease found in the nucleus and cytosol of eukaryotic cells (reviewed in Refs. [1][2][3][4][5]. The structural prototype is an archeabacterial enzyme from Thermoplasma which consists of four stacked heptameric rings forming a cylinder with three internal chambers; the enzyme active sites are located in the central chamber well removed from the particle's surface (6). Eukaryotic MCP forms the proteolytic core of an even larger 26 S proteasome, identified by its ability to degrade ubiquitin (Ub)-lysozyme conjugates in an ATP-dependent reaction (7). Initial studies on assembly of the 26 S proteasome identified three factors (CF1, CF2, and CF3) required for degradation of Ub-conjugates (8); CF3 was subsequently shown to be the MCP (9,10). Besides the low molecular weight MCP subunits (20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30), the 26 S proteasome contains 15 or more additional proteins ranging in molecular mass from 25 to 110 kDa (11). These higher molecular weight subunits, named S1-S15 in order of decreasing molecular weight (12), can be purified together as a ϳ20 S regulatory complex (13). Electron microscopy of the 26 S proteasome shows these regulatory complexes capping one or both ends of the MCP cylinder (14).
Fractionation of rabbit reticulocyte lysate led to the isolation of a particle that assembles with MCP to form the 26 S proteasome (13). Equivalent particles were subsequently identified from several sources: the particle from Drosophila embryos (15), PA700 from bovine blood cells (16), the 19 S cap complex from Xenopus oocytes (17), and the regulatory complex from human erythrocytes (18). Originally called "the ball" (13), then the ATPase complex (19), we now call this particle the regulatory complex (RC) because it contains subunits that recognize proteolytic substrates in addition to subunits belonging to an ATPase family (20). In vivo studies indicate that a preformed MCP combines with a preformed RC to produce the 26 S proteasome in murine lymphoma cells (21). Thus, these regulatory complexes are not artifacts of purification.
The first demonstration of a relevant ATPase activity in the 26 S proteasome was provided by Armon et al. (22), who observed that formation of the 26 S proteasome from CF1, CF2, and CF3 generated nucleotidase activity. ATPase activity in the purified 26 S proteasome (23,24) and in PA700 (25) has also been reported. Moreover, available sequence information from individual 26 S subunits indicates that several subunits belong to a family of putative ATPases (20). In this report, we have characterized the nucleotidase activities of the RC and the 26 S proteasome with regards to kinetic constants, inhibitors, effect of MCP, and we have evaluated the role of nucleotide hydrolysis in the degradation of ubiquitin-lysozyme conjugates.
Protein Purification-The proteins were purified from rabbit reticulocytes as described previously (13). Briefly, reticulocytes were produced in rabbits by phenylhydrazine injection (26), the blood was harvested by cardiac puncture, the red cells were washed, and then lysed in 1 mM DTT. The lysate was fractionated on a TSK-DEAE-650 S column. Peak fractions (Fig. 2, Ref. 13) were pooled, then pelleted to concentrate protein and finally sedimented on 10 -40% glycerol gradients.
Peptidase Assays-Peptide hydrolysis assays with MCP and the 26 S proteasome were previously described (13,19). Briefly, 100 M sLLVY-MCA peptide was incubated with proteasome in 30 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, and with or without 2 mM ATP, in a 100-l volume for 15 min at 37°C, then quenched with ethanol and the fluorescence read at 380/440 nm. For LLnL inhibition studies, 2 g of glycerol gradient purified 26 S proteasome was preincubated with LLnL for 15 min on ice prior to addition of peptide substrate.
ATPase Assays-Nucleotide hydrolysis was measured using colorimetric detection of phosphate by Malachite Green (27). ATP hydrolysis by 5 g of RC or 5 g of 26 S proteasome was linear for 2 h at 37°C. The standard ATPase assay included 2 g of protein (26 S proteasome or RC) and 200 M ATP (in 20 mM HEPES, pH 7.2, 5 mM MgCl 2 , 1 mM DTT, 100 l volume) incubated for 60 min at 37°C; 900 l of 0.034% Malachite Green, 1.1% ammonium molybdate, 1 M HCl, and 0.04% Triton X-100 was added to stop the reaction. After 10 min at room temperature, absorbance was measured at 645 nm. The ATPase assay was later modified by adding sodium citrate (3.4% final concentration) to stabilize the color reaction (28), and the absorbance at 660 nm converted to phosphate produced using a 660-nm standard curve. To show that ATP was hydrolyzed to ADP, early ATPase experiments used [␣-35 S]ATP as substrate and the products were separated on polyethyleneimine-cellulose plates in 1 M formic acid, 1 M lithium chloride (1:1) (29).
For the inhibitor experiments, a 2-g aliquot of either the 26 S proteasome or RC was preincubated with the specified compound for 10 min at room temperature, ATP was then added to 200 M, and the reaction incubated for 60 min at 37°C. The inhibitors EHNA (30,31), NEM, sodium orthovanadate, and oligomycin (32), and thapsigargin (33) were dissolved in H 2 O, whereas ATA (34), hemin (7), and ouabain were dissolved in dimethyl sulfoxide. The levels of dimethyl sulfoxide added to the assay did not affect ATPase activity.
For N-acetyl-leucyl-leucyl-norleucinal (LLnL) inhibition studies, 2 g of glycerol gradient-purified 26 S proteasome was preincubated with LLnL (0 -260 M) on ice for 15 min. The substrate sLLVY-MCA was added to 100 M concentration and incubated with the LLnL/protein mixture for 15 min at 37°C (100 l volume) then the reaction was quenched with 200 l of ethanol and fluorescence read directly (excitation 380 nm/emission 440 nm). For inhibition of nucleotide hydrolysis by LLnL, ATP or UTP was added to 500 M final concentration and incubated with the LLnL/protein mixture for 1 h at 37°C, then the phosphate product was determined by the Malachite Green assay.
Ubiquitin-conjugate Degradation-Ubiquitin (Ub)-125 I-lysozyme conjugates were prepared and degraded as described (35). The 26 S proteasome (13 g of rabbit reticulocyte DEAE-purified protein) was preincubated with nucleotide for 2 min at 37°C (100 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT, 50 l volume) to bring the reaction up to temperature. Then, 50 l of 125 I-lysozyme Ub-conjugate substrate (ϳ4800 cpm) was added and the 37°C incubation continued for 10 min. The reactions were stopped by addition of 800 l of 1% bovine serum albumin and 100 l of 100% trichloroacetic acid on ice, and the precipitated proteins were pelleted by 10 min centrifugation in a microcentrifuge. The acid-soluble radioactivity was quantitated with a ␥ scintillation counter. The acid-soluble counts from a reaction without 26 S proteasome were subtracted, and then the percentage of Ub-conjugate degradation was calculated.
26 S Assembly Assays-Typically, 2 g of rabbit MCP and 2 g of rabbit RC were incubated in 40 l of 20 mM HEPES pH 7.2, 2 mM ATP, 5 mM MgCl 2 , 1 mM DTT for 30 min at 37°C (19). Alternatively, a 30-min incubation on ice was used as indicated to minimize nucleotide hydrolysis. The assembled proteins were electrophoresed through nondenaturing gels, assayed for peptidase activity with substrate overlays, and stained for protein by Coomassie Blue or Bio-Rad silver stain.
Kinetic Analysis-A standard curve of phosphate (0 -20 nmol) versus absorbance 645 nm was used to determine the phosphate produced. The Cleland nonlinear regression analysis program (36) was used to statistically analyze the data for determination of K m and V max . Molecular weights of 700,000 daltons for RC and 1,400,000 daltons for the 26 S proteasome (assuming 1 RC at 700 kDa and 1 MCP at 700 kDa make the 26 S proteasome at 1,400 kDa) were used to calculate turnover numbers.

Enzymatic Activities of Isolated 26 S Proteasome and RC-
The ATP-dependent 26 S proteasome and its RC were purified from rabbit reticulocytes and identified by their sedimentation characteristics, their migration through nondenaturing gels, their subunit composition, and their peptidase or ATPase activities (Fig. 1). The 26 S proteasome hydrolyzes peptides and ATP. Both MCP and RC exhibit an apparent sedimentation coefficient of ϳ20 S, but MCP only hydrolyzes peptides (not shown) and the RC only hydrolyzes nucleotides (Fig. 1A). A third ATPase activity sedimenting slower than the RC was detected in these glycerol gradients. It co-purified with a 90-kDa protein that constituted about 75% of the protein in fractions 18 to 20 shown in Fig. 1; amino acid sequencing identified this protein as hsp90. Because this chaperonin reportedly has ATPase activity (37), the hsp90-containing fractions were used for comparison of nucleotidase activities in several experiments presented below.
FIG. 1. Physical and enzymatic characteristics of the 26 S proteasome and the RC. DEAE fractions containing 26 S proteasome and RC were pooled together and sedimented on a 10 -40% glycerol gradient. Each gradient was fractionated from the bottom into 1-ml aliquots, from which 10-l aliquots were assayed for sLLVY-MCA hydrolysis (dashed line). A 10-l aliquot from each fraction was also assayed for ATPase activity (solid line). Aliquots (40 l) of the peak fractions were electrophoresed through nondenaturing gels, analyzed by peptide substrate overlay (right, inset), and stained with Coomassie (left, inset). The migration of native complexes of 26 S proteasome, RC, and hsp90 are designated by arrowheads. The peak of 26 S proteasome is centered on fraction 8, RC peaks in fraction 10, and hsp90 is most abundant in fraction 18. The 26 S proteasome and RC used in the rest of the experiments were each pooled separately from DEAE chromatography fractions and sedimented on separate glycerol gradients to minimize the overlap seen in the glycerol gradient shown above.

Altered Sedimentation of ATPase by Combination with the
Multicatalytic Proteasome-To demonstrate that the ATPase activity sedimenting at ϳ20 S in Fig. 1 was associated with the RC and not with some other protein complex, such as hexamers of p97 (38), we sedimented MCP and RC either alone or in combination on glycerol gradients and then assayed for peptidase and ATPase activity. Combining MCP and RC to form the 26 S proteasome caused both the peptidase activity and the ATPase activity to sediment deeper in the gradient (compare Fig. 2, A and B). Nondenaturing gel electrophoresis confirmed that both MCP and RC, which alone sedimented to fractions 13-15, were found in fractions 10 -13 when mixed prior to centrifugation (Fig. 2C). This shift to a faster sedimenting species was expected since RC and MCP combine to form the 26 S proteasome. However, in three separate experiments the assembled 26 S proteasome sedimented slightly slower than the purified 26 S enzyme. Although the reassembled and the endogenous 26 S proteases were indistinguishable by subunit composition, nondenaturing gel electrophoresis, or substrate specificity, the slightly altered sedimentation indicates that the assembled 26 S proteasome is either not as stable as the endogenous 26 S proteasome or that it is missing some components. Despite this difference, the ability of MCP to speed sedimentation of both ATPase activity and RC subunits is evidence that the ATPase activities in fractions 10 -13 of Fig. 2 are conferred by subunits of the regulatory complex. This conclusion was further supported by comparisons presented below of the NTPase properties of RC and 26 S proteasome.
Kinetic Analysis of NTPase Activities-Hydrolysis of [␣-35 S]ATP by the 26 S proteasome and the RC produced ADP and P i as assayed by thin layer chromatography (not shown, see "Experimental Procedures"). More extensive experiments using a spectrophotometric assay showed that both the RC and the 26 S proteasome possess nucleotidase activity capable of hydrolyzing ATP, CTP, GTP, and UTP. Graphs of velocity versus substrate concentration indicate apparent Michaelis-Menten kinetics for nucleotide hydrolysis by the RC and the 26 S proteasome. Kinetic constants for nucleotide hydrolysis by the two particles are comparable, although the 26 S proteasome consistently exhibited lower K m values for nucleotide hydrolysis than the RC did (e.g. ϳ15 M versus ϳ30 M for ATP, see Table I). Turnover numbers were similar for the 26 S proteasome (27 ATPs hydrolyzed/min/26 S complex) and RC (23 ATPs/min/RC). Specificity constants (k cat /K m ) are similar for all NTPs, but they hint that ATP and CTP are better substrates than GTP and UTP. In contrast to the broad nucleotide specificity of the RC and 26 S proteasome, the hsp90-containing fraction hydrolyzed ATP and dATP (K m values of ϳ40 and 30 M, respectively), but there was no detectable hydrolysis of CTP, GTP, or UTP.
MCP Alters the K m for NTP Hydrolysis by the RC But Not by the 26 S Proteasome-Since the K m values for nucleotide hydrolysis by the 26 S proteasome were significantly lower than those of the RC, we expected that addition of MCP to the regulatory complex would produce the 26 S proteasome and result in lower K m values for nucleotide hydrolysis. For this reason, we assayed the ATPase activity of RC in the presence or absence of MCP (Fig. 3). ATPase activity of the assembled RC⅐MCP complex exhibited a K m of 19 M as compared to 34 M by the RC alone; in fact the K m decreased almost to that of the purified 26 S proteasome (14 M). The decrease in the K m for nucleotide hydrolysis upon addition of MCP to the RC was confirmed using GTP as well. No significant changes in ATPase characteristics of the 26 S proteasome were seen upon addition of MCP (14 M for 26 S proteasome alone and 18 M for the 26 S proteasome and MCP). These results were expected since the 26 S complex was already assembled prior to MCP addition. The MCP-dependent changes in the RC's nucleotidase activities provide further evidence that the nucleotidases are part of the regulatory complex and that their catalytic properties can be affected by association with MCP.
Nucleotide Dependence for Ub-conjugate Degradation-The K m values for ATP and CTP hydrolysis are lower than those for GTP and UTP hydrolysis with both particles (Table I). To determine whether these differences are relevant to proteolysis by the 26 S proteasome, degradation of Ub-125 I-lysozyme conjugates was assayed at various concentrations of ATP, CTP, GTP, and UTP (Fig. 4). Approximately 12 M ATP produced half-maximal Ub-conjugate degradation by the 26 S proteasome, a value almost identical to the 14 M K m for ATP hydrolysis by the 26 S proteasome (Table I). CTP supported Ubconjugate degradation but the observed K m of approximately 100 M was 6-fold higher than the K m for CTP hydrolysis. Although Ub-conjugate degradation was measurable in the presence of GTP and UTP, it was too low to determine reliable K m values. Therefore we performed an alternative experiment in which we measured Ub-conjugate degradation in the presence of 1 mM nucleotides. Relative rates for Ub-conjugate degradation were ATP (100%), CTP (52%), GTP (13%), and UTP (7%) (averaged from three experiments, data not shown). The NTPs which best support Ub-conjugate degradation are those with the lowest K m for hydrolysis by the 26 S proteasome (ATP, CTP Ͼ Ͼ GTP, UTP).
Inhibition of NTPase Activities-Because inhibitors can be useful for characterizing and classifying enzymes, we assayed the effects of known ATPase inhibitors on the NTPase activities of the RC or the 26 S proteasome. From the data presented in Table II, it is clear that compounds specific for several known ATPases had little effect on ATP hydrolysis by RC or the 26 S proteasome. For example, thapsigargin inhibits Ca-AT-Pases in the endoplasmic reticulum (33), EHNA blocks dynein-ATPase activity (30,31), oligomycin and ouabain inhibit Na/Kpump ATPases, but none significantly inhibited the ATPase activities of the 26 S proteasome or its regulatory complex. By contrast, hemin, aurintricarboxylic acid, NEM, and vanadate, which all have a strong inhibitory effect on ATP-dependent proteolysis by the 26 S proteasome (11), significantly reduced ATPase activity in both complexes. For both RC and the 26 S proteasome, ATP hydrolysis was highest in the presence of Mg 2ϩ or Mn 2ϩ , and inhibited to various extents by Ca 2ϩ , Zn 2ϩ , or EDTA. ATP hydrolysis by the hsp90-containing fraction was also highest with Mg 2ϩ and was inhibited by EDTA, but unlike ATP hydrolysis by RC and the 26 S proteasome, it was not inhibited by Ca 2ϩ or Zn 2ϩ .
NEM is a potent inhibitor of proteolysis by the 26 S proteasome, and it also greatly decreased ATP hydrolysis by the RC and the 26 S proteasome (Table II). To determine whether NEM uniformly inhibited hydrolysis of all NTPs, we examined the hydrolysis of ATP, CTP, GTP, and UTP by the RC incubated with various levels of NEM. Hydrolysis of all NTPs was markedly inhibited at 5 mM NEM (Fig. 5). Interestingly, at 50 M NEM only GTP and UTP hydrolysis was significantly inhibited. ATP and CTP can each support degradation of Ubconjugates by the 26 S proteasome whereas GTP and UTP do so to a much lesser extent (Fig. 4). Our findings that GTP/UTP hydrolysis was especially sensitive to NEM raised the possibility that exposure of the 26 S proteasome to low levels of NEM  (36). The results presented here are averaged from several independent experiments (n ϭ 6 -15). The numbers of ␥-phosphate bonds hydrolyzed per minute per particle (turnover number or k cat ) were calculated using 1 RC ϭ 700,000 daltons, and 1 26 S proteasome ϭ 1,400,000 daltons, assuming 1 MCP at 700,000 daltons plus 1 RC at 700,000 daltons produce the 26 S proteasome (see Ref. 21 for a discussion of MCP and RC levels in RMA cells). The specificity constant was calculated as k cat /K m . The K m for dATP hydrolysis by RC is 45 M and by the 26 S proteasome is 20 M. The hsp90-containing fraction hydrolyzed ATP (K m ϭ 39 M) and dATP (K m ϭ 27 M) but did not appreciably hydrolyze CTP, GTP, or UTP.   (13 g) was preincubated with nucleotide for 2 min at 37°C to bring the reaction to temperature, and Ub-125 I-lysozyme conjugates were added. The 37°C incubation continued for 10 min before being quenched by precipitation with trichloroacetic acid (35). Trichloroacetic acid-soluble counts represent Ub-conjugate degradation; the trichloroacetic acidsoluble counts in "no protease" control reactions were subtracted from the 26 S proteasome reactions prior to data analysis. The % of Ubconjugates degraded in the presence of each nucleotide is shown. The K m for ATP and dATP (not shown) were both estimated at ϳ12 M and the K m for CTP was estimated at ϳ100 M; rates of degradation with GTP and UTP were too low to yield a reliable K m estimate. could produce an enzyme unable to degrade Ub-conjugates using GTP or UTP as an energy source, but it might still be able to proteolyze Ub-conjugates in the presence of ATP or CTP. At 250 M NEM, which severely inhibited only GTP and UTP hydrolysis, both Ub-conjugate degradation and fluorogenic peptide cleavage were also inhibited, so we could not determine whether NEM inhibition of conjugate degradation resulted from impaired proteases, ATPases, or both.
Inhibition of Protease Active Sites Does Not Affect Nucleotide Hydrolysis by the 26 S Enzyme-A series of experiments indicated that protein or peptide substrates (e.g. sLLVY-MCA, casein, Ub-conjugates, and multi-Ub chains) had little effect on the ATPase activity of the 26 S proteasome (data not shown). From these observations and the fact that the RC alone exhibits significant NTPase activity, we assumed that the ATPase and peptidase activities are not strictly coupled in the purified 26 S proteasome. To verify this assumption, we inhibited peptide bond hydrolysis by using LLnL and determined the effect on nucleotide hydrolysis. Preincubation of the 26 S proteasome with increasing amounts of LLnL led to a dramatic inhibition of sLLVY-MCA peptide hydrolysis, but these levels of LLnL had little effect on ATP or UTP hydrolysis (Fig. 6). Since nucleotide hydrolysis by the 26 S proteasome continued unabated after some active sites, at least, were poisoned, it seems very likely that the nucleotidase activity of the 26 S proteasome does not require concomitant peptide bond hydrolysis. DISCUSSION ATPase activity was previously reported in the assembled 26 S proteasome from rabbit reticulocytes (22), in the purified 26 S proteasome from human kidney (23) and rat liver (24), and in the regulatory complex from bovine red cells (25). We have further characterized the nucleotidase activities in these two particles with respect to nucleotide specificity, kinetic constants, and their role in ubiquitin-conjugate degradation. We present considerable evidence that the measured nucleotidases reside in the 26 S proteasome and its regulatory complex, not in some other cellular component. Furthermore, the nucleotidase activities of the 26 S proteasome and its regulatory complex are similar but not identical. Both the RC and 26 S proteasome exhibit broad substrate specificity, hydrolyzing ATP, CTP, GTP, and UTP to the corresponding NDP ϩ P i (Table I). They do not hydrolyze NDP or NMP. Also, both particles require FIG. 5. Nucleotidase inhibition by NEM. RC (2 g) was preincubated with NEM (0.05, 0.5, 1, 5 mM) in the absence of nucleotides or DTT for 15 min on ice. Then DTT and the specific nucleotide were added (to 1 and 0.5 mM, respectively), and the sample was incubated for 1 h at 37°C. Nucleotide hydrolysis was determined by measuring the amount of phosphate produced with Malachite Green reagent (27, 28). The nucleotidase activity in the presence of NEM is expressed as % activity relative to hydrolysis in the absence of inhibitor. Inhibition of 50% of GTP and UTP hydrolysis occurs around 35 M NEM while inhibition of 50% of ATP, dATP, and CTP hydrolysis occurs around 2.5 mM NEM.
FIG. 6. Selective inhibition of 26 S peptidase by LLnL. The specified amount of LLnL (or "calpain inhibitor I") was preincubated with 2 g of the 26 S proteasome for 15 min on ice, then substrate (either peptide or nucleotide) was added to initiate the reaction. The peptide substrate, sLLVY-MCA, was incubated with the protein for 15 min at 37°C, followed by quantitation of fluorescent product at 380/440 nm. Nucleotides, either ATP or UTP, were incubated with the protein for 1 h at 37°C, then the amount of phosphate product was quantitated with the Malachite Green assay. The % activity remaining is an average of duplicate sets of data and is determined by comparison to the activity in the absence of LLnL. divalent cations for activity, with strong preference for Mn 2ϩ or Mg 2ϩ . Inhibitors of Ca-ATPases, Na/K-ATPases, or dyneintype ATPases had little effect on either the RC or 26 S ATPase activities (Table II). The 26 S proteasome inhibitors hemin, ATA, and NEM, also inhibit the ATPase activities of both the RC and the 26 S proteasome, and both complexes exhibit the same pattern of inhibition by NEM. That is, GTP and UTP hydrolysis is more sensitive to NEM than is hydrolysis of ATP or CTP (Fig. 5). Finally, addition of the proteasome shifts the nucleotidase activities physically on glycerol density gradients (Fig. 2) and kinetically by lowering the K m for hydrolysis of ATP or GTP (Fig. 3). These findings demonstrate that the nucleotidase activities described here can be attributed to the 26 S proteasome and its regulatory complex, and that the RC provides the 26 S proteasome with its energy dependent properties.
Several 26 S proteasome subunits form the S4 subfamily of a novel group of nucleotidases called the AAA family (ATPases Associated to a variety of cellular Activities) (39). These proteins have one or two diagnostic GPPGXGKT nucleotide binding motifs, are sensitive to NEM, and have a variety of cellular functions. Although sequences of 14 of the approximately 16 subunits of the regulatory complex are known, only six are likely to be ATPases, as indicated by their sequence similarity to members of the AAA family. These putative ATPases are almost surely responsible for the nucleotidase activities reported here. This assumption is supported by comparing the nucleotidases of the 26 S proteasome and its RC to those of two prokaryotic ATP-dependent proteases that have similarity to the AAA family.
In prokaryotes, 70 -80% of the energy-dependent protein degradation is provided by two large multisubunit proteases, Clp and Lon (40). Clp protease consists of two types of subunits: the ClpP subunits contribute the proteolytic activity and the ClpA subunits contribute the ATPase activity. Only ATP or dATP are hydrolyzed and are able to support casein degradation by Clp. The K m for ATP in both reactions is the same, around 150 -210 M (41,42). The ClpAP protease is composed of two heptameric rings of ClpP subunits (43) with a hexameric ring of ClpA subunits on one or both ends of the cylinder (44), similar to the barbell or mushroom structure of the eukaryotic MCP cylinder capped at one or both ends with the regulatory complex. The assembly of the two subcomplexes, the ATPase activity (ClpA) and the proteolytic activity (ClpP), into a larger ATP-dependent protease (ClpAP) is highly reminiscent of coassembly of the RC and MCP to form the eukaryotic ATP-dependent 26 S proteasome. Biochemically, however, the nucleotidase activities of the 26 S proteasome more closely resemble the prokaryotic Lon protease, which is composed of identical subunits that provide both the ATPase and the protease activities. The NTPase of Lon has reported K m values for ATP of 27 M (45) and 200 M (46), whereas we find the rabbit 26 S proteasome's K m for ATP hydrolysis to be 15 M (Table I), between previous estimates of ϳ1 M (22) and 100 -250 M (24). The K m values for ATP required to support casein degradation by Lon are probably the same as those for ATP hydrolysis, yet the variability in reports, even those from the same laboratory, clouds any firm conclusion. For instance, the K m for ATP needed for casein degradation by Lon has been reported as 3.  (49). All of these investigators found, however, that ATP supported casein degradation better than the other nucleotides, and the order of nucleotide preference was consistently ATP Ͼ CTP Ͼ UTP Ͼ GTP. Therefore, both Lon protease and the 26 S proteasome have broad nucleotide specificity, utilizing ATP, CTP, GTP, and UTP, but not ADP or AMP. The similarity in nucleotide pref-erence between the eukaryotic 26 S proteasome and prokaryotic Lon protease suggests that these enzymes may use common reaction mechanisms. In this regard, the presumed nucleotidase domain of Lon contains amino acid sequences GPPGXGKT and DEID which are hallmarks of the S4 family of ATPases (20,50). Thus, while the eukaryotic 26 S proteasome has sequence, structural, and biochemical features in common with both of these prokaryotic proteases, its nucleotidase properties are most similar to those of Lon.
Initial evidence suggesting that nucleotidase activity is not coupled to proteolysis in the 26 S proteasome is provided by the data in Table I and Fig. 4. The regulatory complex by itself exhibits levels of nucleotidase activity comparable to the 26 S proteasome. While each NTP is readily hydrolyzed by the 26 S proteasome, they do not equally support the degradation of Ub-conjugates. For instance, GTP and UTP are hydrolyzed with higher turnover numbers than ATP or CTP, yet the latter two nucleotides are much superior energy sources for proteolysis. In addition, the aldehyde LLnL inhibits peptide hydrolysis by MCP (51,52), and casein and Ub-conjugate degradation by the 26 S proteasome (53), yet it has relatively little effect on ATP and UTP hydrolysis by the 26 S proteasome (Fig. 6). Uncoupling between ATPase and protease active sites was convincingly demonstrated by site-directed mutagenesis of Lon. A mutation in the nucleotide binding domain (K362A) substantially inactivated both the ATPase and the protease but did not affect the peptidase activity (46). Conversely, mutation of the critical serine in the protease active site (S679A) inactivated the protease but left the ATPase unaffected (54). Although it has been suggested that 2 ATPs are hydrolyzed per peptide bond cleaved by Lon (55), the site-directed mutations show that nucleotide hydrolysis is not necessarily coupled to peptide hydrolysis in Lon protease, i.e. phosphate bonds can be cleaved without concomitant cleavage of peptide bonds.
We previously demonstrated that each of the NTPs could support Ub-conjugate degradation by the 26 S proteasome (7). We have extended these results by examining the degradation of Ub-conjugates at lower NTP concentrations (2.5-1000 M, Fig. 4). ATP and CTP were clearly the best energy sources for degradation of Ub-conjugates while GTP and UTP supported conjugate degradation to a lesser extent than we previously reported (7). This may be due to our use of more dilute 26 S protease in the present Ub-conjugate degradation assays. It is well established that only ATP and CTP can promote assembly (16,22) whereas all four nucleotides can support conjugate degradation (7,22). Consequently, in dilute assay mixtures the two competing reactions of Ub-conjugate degradation by the 26 S proteasome and dissociation of the 26 S enzyme to the regulatory complex and 20 S proteasome will favor dissociation. Once dissociated, only ATP and CTP can support reassembly, thus magnifying the difference between ATP/CTP and GTP/ UTP in their relative ability to support Ub-conjugate degradation. In any event, there should be little doubt that all four nucleotides can support conjugate degradation. In competition experiments with [␥-32 P]ATP and the 26 S proteasome, all the NTPs were substrates for hydrolysis in the order ATP Ͼ CTP Ͼ GTP Ͼ UTP, the same nucleotide dependence seen for Ubconjugate degradation (22). Release of Ub from Ub-H2A by ubiquitin hydrolase in the 26 S proteasome also is highest with ATP and follows the heirarchy ATP Ͼ CTP Ͼ GTP (56). The consistency of nucleotide dependence (ATP Ͼ CTP Ͼ Ͼ GTP Ͼ UTP) for each of these 26 S-associated activities (nucleotide hydrolysis, Ub-conjugate degradation, and Ub release) support the idea that all of these activities are inherent to the 26 S proteasome complex, and perhaps reflect the enzymatic activities of the S4-like ATPases.
Hydrolysis of ATP and CTP by the 26 S proteasome differs from hydrolysis of GTP and UTP as illustrated by NEM inhibition studies (Fig. 5). We attempted to determine whether selectively inhibiting GTP/UTP hydrolysis with low levels of NEM would affect proteolysis, but found that both peptide hydrolysis and GTP/UTP hydrolysis were inhibited, making it impossible to draw any conclusions on the role of GTP and UTP hydrolysis in proteolysis. At any rate, the different NEM sensitivities for GTP and UTP hydrolysis versus ATP and CTP hydrolysis, in combination with the large K m differences between ATP/CTP and GTP/UTP hydrolysis, suggests that some members of the S4 subfamily may prefer GTP and UTP more than others. In this regard it may be significant that 26 S proteasome ATPase subunit S6 has a cysteine in the P loop (57), which could make this subunit more sensitive to the sulfhydryl reagent NEM.
In Novikoff rat hepatoma cells and in mouse 3T3 cells, respectively, intracellular nucleotide concentrations are 3.2-4.4 mM for ATP, and 0.4 -1.5 mM for the other nucleotides (58,59). Although these values may not represent the concentrations of free, available NTPs in the cell, the reported NTP levels are all at or above the K m values for hydrolysis by the 26 S proteasome, suggesting that the enzyme's nucleotidases may be operating close to their maximum velocity in vivo. The data in Table I clearly show that the nucleotidase activity can be uncoupled from the peptidase activity of the 26 S proteasome in vitro. Because it would be wasteful for the 26 S proteasome to hydrolyze nucleotides without concomitant protein degradation, mechanisms for coordinating the two activities must exist. In fact, the first description of the ATPase associated with the 26 S proteasome reported that ATPase activity was not detected in any of the three conjugate-degrading factors alone, and that ATPase activity only appeared once all three factors were combined to form the 26 S proteasome (22). Considering that the regulatory complex has full nucleotidase activity, this earlier result suggests that NTPase inhibitors are present in less purified preparations of CF1, CF2, and CF3. Presumably within the cell, the NTPase activities of the 26 S proteasome and the RC are suppressed in the absence of proteolytic substrates.
Our characterization of the nucleotidase activities in these multisubunit complexes provides a frame of reference for future studies with individually expressed ATPases of the 26 S proteasome. It is possible that all of the ATPase subunits will have characteristics similar to those we describe for the multisubunit complexes. Alternatively, these studies may reflect an "average" of several distinct nucleotidases all active in this large complex. So far, the Saccharomyces cerevisiae homolog of human S4 (Yhs4p) is the only individual 26 S ATPase subunit to be characterized biochemically (60). Yhs4p's broad substrate specificity and low K m for ATP (5 M) are comparable to the nucleotidase properties determined here for the regulatory complex. Possible roles for these nucleotidases include chaperonin-type functions such as substrate recognition, binding, and unfolding, or the movement of polypeptide chains into the proteasome central cavity. A 26 S proteasome subunit (S5a) that binds Ub-conjugate substrates has already been identified and its cDNA has been cloned and expressed (61,62). Other non-ATPase subunits whose cDNAs have been cloned (e.g. S1 (63), S2 (64), S5b (65), S12 (66), S14 (67)) must still be assigned functions. Continued studies to define the role of individual subunits within this proteolytic organelle will no doubt increase our understanding of intracellular proteolysis.