Identification, purification, and characterization of a PA700-dependent activator of the proteasome.

The activity of the intracellular protease, the proteasome, is modulated by a number of specific regulatory proteins. One such regulator, PA700, is a 700,000-Da multisubunit protein that activates hydrolytic activities of the proteasome via a mechanism that involves the ATP-dependent formation of a proteasome-PA700 complex. Four subunits of PA700 have been shown previously to be members of a protein family that contains a consensus sequence for ATP binding, and purified PA700 expresses ATPase activity. We report here the identification, purification, and initial characterization of a new modulator of the proteasome. The modulator has no direct effect on the activity of the proteasome, but enhances PA700 activation of the proteasome by up to 8-fold. This activation is associated with the formation of a proteasome/PA700-containing complex that is significantly larger than that formed in its absence. The modulator has a native Mr of approximately 300,000, as determined by gel filtration chromatography, and is composed of three electrophoretically distinct subunits with Mr values of 50,000, 42,000, and 27,000 (p50, p42, and p27, respectively). Amino acid sequence analysis of the subunits shows that p50 and p42 are members of the same ATP-binding protein family found in PA700. The p50 subunit is identical to TBP1, a protein previously reported to interact with human immunodeficiency virus Tat protein (Nelbock, P., Dillion, P. J., Perkins, A., and Rosen, C. A. (1990) Science 248, 1650-1653), while the p42 subunit seems to be a new member of the family. The p27 subunit has no significant sequence similarity to any previously described protein. Both p50 and p42, but not p27, were also identified as components of PA700, increasing the number of ATP-binding protein family members in this complex to six. Thus, p50 and p42 are subunits common to two protein complexes that regulate the proteasome. The PA700-dependent proteasome activator represents a new member of a growing list of proteins that regulate proteasome activity.

The proteasome is a 700,000-Da multicatalytic protease that participates in a number of proteolytically mediated intracellular processes, including the constitutive turnover of many intracellular proteins (1), the rapid elimination of proteins with abnormal structures (2,3), the temporal reduction in levels of critical regulatory proteins for control of the cell cycle and transcription (4 -7), the proteolytic activation of the transcription factor NF-B (8), and the processing of antigens for presentation by class I major histocompatibility complex proteins (9,10). Despite the important role of the proteasome in these various processes, the mechanisms by which its action is controlled remain unclear. Several lines of evidence indicate that proteasome function is controlled by specific regulatory proteins. First, the proteasome can be isolated as part of a larger protein complex (M r Ն 1,500,000) referred to as the "26 S protease" (11). This complex displays catalytic and regulatory properties that differ considerably from those of the purified 20 S proteasome, most likely because of regulatory influences exerted by the non-proteasome components of the complex. Second, individual proteasome regulatory proteins have been identified and purified. One of these proteins, which we call PA700 and which has been independently described in several laboratories, appears to represent the major non-proteasome component of the 26 S protease (10,(12)(13)(14). PA700 is a 700,000-Da multisubunit ATP-dependent proteasome activator. It forms a complex with the proteasome by a mechanism that requires ATP hydrolysis. The proteasome-PA700 complex has physical properties, such as molecular weight, and catalytic properties, such ATP-dependent degradation of ubiquitinated proteins, that are characteristic of the purified 26 S protease. At least four of the ϳ20 electrophoretically distinct subunits of PA700 are homologous to one another and are members of a large protein family that contains a consensus sequence for ATP binding (15)(16)(17). Some of these same proteins have been identified as components of the purified 26 S protease, providing additional strong evidence that the proteasome-PA700 complex is similar, if not identical, to the 26 S protease (13,15,16). One or more of these ATP-binding proteins may be responsible for the function of ATP in proteasome activation. In fact, purified PA700 expresses ATPase activity. Surprisingly, many of these "ATPase" subunits of PA700 have been identified independently as proteins involved in processes with no obvious relationships to proteasome function (17). These findings may be explained by new and unexpected roles for the proteasome or may indicate that a given ATPase protein has multiple cellular functions.
During the course of our continuing characterization of the function of PA700, we have identified a new protein complex that functions as a PA700-dependent activator of the proteasome. This report describes the identification, purification, and initial structural and functional characterization of this protein, which contains two members of the ATPase protein family. Furthermore, the same two proteins are identified here as new subunits of PA700, raising the number of family members in this complex to six.

MATERIALS AND METHODS
Purification and Assay of the Proteasome and PA700 -The 20 S proteasome and PA700 were purified from bovine red blood cells as described previously (12,18). Proteasome activity was measured by the hydrolysis of the synthetic peptide succinyl-Leu-Leu-Val-Tyr 7-amino-4-methylcoumarin, also as described previously (12). The production of free 7-amino-4-methylcoumarin was monitored continuously at 380 nm (excitation) and 460 nm (emission), and initial steady-state rates were assessed. One unit of proteasome activity is defined as a change in 7-amino-4-methylcoumarin concentration of 1.0 nM/min under standard assay conditions. PA700 activity was assessed by measuring the proteasome activity after preincubation with pure PA700. The preincubation contained 45 mM Tris-HCl, pH 8.0, 5.6 mM dithiothreitol, 200 M ATP, and 10 mM MgCl 2 in a final volume of 50 l and was carried out for 45 min at 37°C. This solution was then added to 1.0 ml of substrate solution for the measurement of proteasome activity as described above (12).
PA700-dependent Activator (Modulator) Assay-The identification, purification, and characterization of the PA700-dependent activator (modulator) was carried out with a variation of the PA700 assay. The modulator sample to be tested was mixed with the purified proteasome and PA700 and then preincubated under the same conditions as the normal PA700 assay. Modulator activity is expressed as the increase in the PA700-dependent proteasome activity caused by the modulator. One unit of modulator activity is defined as an increase of 1 unit of PA700 activity.
Purification of a PA700-dependent Proteasome Activator (Modulator)-Bovine red blood cells were collected, washed, and lysed as described previously (12). Fraction II from the soluble lysate was prepared using DEAE-cellulose (DE52, Whatman). Fraction II was dialyzed against Buffer X (20 mM Tris-HCl, pH 7.6, 20 mM NaCl, 1 mM MgCl 2 , 0.1 mM EDTA, 0.5 mM dithiothreitol, and 20% glycerol) for 16 h. After dialysis, solid ammonium sulfate was added to the sample to 38% saturation. The precipitated proteins were collected by centrifugation, resuspended in a large volume of Buffer X containing ammonium sulfate to 38% saturation, and collected again by centrifugation. The resulting pellet was dissolved in Buffer X containing 100 mM NaCl and dialyzed against 4000 ml of the same buffer for 8 h. After dialysis, any undissolved material was removed by centrifugation, and the soluble sample was chromatographed on a 5 ϫ 100-cm column of Sephacryl S-300. The column was equilibrated and eluted with buffer consisting of 50 mM Tris-HCl, pH 7.6, 5 mM ␤-mercaptoethanol, 100 mM NaCl, and 20% glycerol. The column fractions were assayed for PA700 activity and for modulator activity as described above and in the legend to Fig. 1. The fractions containing PA700-dependent activator activity were pooled, dialyzed against column buffer containing 50 mM NaCl, and applied to a 2.5 ϫ 5-cm column of DEAE-Fractogel. The bound proteins were eluted with a 500-ml linear gradient of NaCl (50 -300 mM) in the same buffer. The fractions were assayed for PA700-dependent activator activity as described above and in the legend to Fig. 2. The fractions with this activity were pooled and dialyzed for 16 h against 20 mM potassium phosphate buffer, pH 7.6, 1 mM ␤-mercaptoethanol, and 20% glycerol and then applied to a 2.5-cm diameter column containing 6 g of hydroxylapatite equilibrated in the same buffer. The bound proteins were eluted with a 300-ml linear gradient of phosphate (20 -100 mM). The fractions with peak activity were pooled, dialyzed against Buffer X containing 5% glycerol, and concentrated to 100 -500 g/ml protein. This material was stored at Ϫ70°C, where it was stable for at least 6 months.
Glycerol Density Gradient Centrifugation-Complexes formed by the proteasome with PA700 and PA700/modulator were assessed by glycerol density gradient centrifugation. Linear gradients of 10 -40% glycerol were prepared in 50 mM Tris-HCl, pH 7.6, 1 mM ␤-mercaptoethanol, 200 mM ATP, and 300 mM MgCl 2 in a final volume of 4.55 ml. Samples in a volume of 200 l were applied per tube and were centrifuged at 4°C for 16 h at 30,000 rpm in a Beckman SW 50.1 rotor. 24 samples of 200 l were collected and subjected to assays as described in the text below. Separate tubes contained marker proteins including thyroglobulin and aldolase.
Protein Determinations-Protein was determined by the method of Bradford (19) using reagents purchased from Bio-Rad. Bovine serum albumin was used as a standard.
Isolation of PA700 and Modulator Subunits-Subunits of purified PA700 or the modulator were isolated by reverse-phase chromatogra-phy using an HPLC 1 system from Waters Chromatography (Milford, MA) equipped with a 6 ϫ 150-mm Shodex RS Pak D4-613 column. Chromatography was conducted at a flow rate of 0.75 ml/min and a column temperature of 50°C. Two solvents were employed: solvent A containing 0.06% (v/v) trifluoroacetic acid in water and solvent B containing 0.05% trifluoroacetic acid in 70% acetonitrile, 30% water (v/v/v). The sample was initially injected onto a column equilibrated with a solution composed of 63% solvent A, 36% solvent B. After 10 min of isocratic flow under initial conditions, elution proceeded with a 130-min linear gradient from 36 to 70% solvent B. Eluant absorption was monitored at 214 nm. Peaks were collected manually, dried in a Savant Instruments Speed-Vac concentrator, redissolved in SDS sample buffer, and heated at 95°C for 4 min. Subunits in each pool were then further resolved by SDS-PAGE in 12.5% polyacrylamide gels as described previously (15).
Production and Separation of Peptide Fragments-PA700 or modulator subunits, subjected to the two-dimensional isolation procedure described above, were electroblotted from SDS gels to Immobilon P SQ paper (Millipore Corp., Bedford, MA) by the method of Matsudaira (20). Blots were stained with Coomassie Blue R-250, and the bands were excised individually. Immobilized protein was digested in situ with sequencing-grade trypsin or Lys-C protease (Boehringer Mannheim). Released peptides were purified by reverse-phase chromatography using a Model 130A HPLC system from Applied Biosystems Division of Perkin-Elmer (Foster City, CA) with a 2.1 ϫ 100-mm RP 300 column, also from Applied Biosystems. Separations were conducted at 50 l/min in 0.1% (v/v) trifluoroacetic acid or 0.1% (w/v) ammonium acetate in water. Peptides were eluted with a 100-min linear gradient of 0 -70% acetonitrile.
Amino Acid Sequence Analysis-NH 2 -terminal sequence analysis of Lys-C and of tryptic peptides was performed by automated Edman degradation with a Model 477A sequencer from Applied Biosystems using the manufacturer's standard programing and chemicals.
Antibodies and Immunochemical Analysis-Polyclonal antibodies against human TBP1 were prepared in rabbits against the protein expressed in Escherichia coli as described previously (6). Western blotting was conducted with an ECL Western blotting kit (Amersham Corp.) following the manufacturer's directions.

RESULTS
Identification and Purification of a PA700-dependent Modulator of the Proteasome-As part of a search for proteins that regulate proteasome function, we have been screening cell extracts for proteins that can influence one or more activities of the purified 20 S proteasome. A number of such proteins have now been identified, and they include both activators and inhibitors. Because it seemed reasonable to assume that some proteasome regulatory proteins might act in concert or might themselves be regulated by other proteins, we also have designed assays to test the effects of cell extracts on the proteasome in the presence of previously identified regulators. In the experiments described here, we tested the ability of a red blood cell extract, fractionated by ammonium sulfate precipitation and gel filtration chromatography on Sephacryl S-300, to affect the proteolytic activity of the exogenous 20 S proteasome in the presence of purified PA700, an ATP-dependent proteasome activator (12). When column fractions were added to assays containing the purified proteasome and PA700, two peaks were identified that contained more activity (3-8-fold in various preparations) than that accounted for by the purified proteasome and PA700 alone (Fig. 1). One of these peaks was coincident with the elution position of PA700 endogenous to those fractions. Therefore, this peak probably resulted from the concentration-dependent increase in proteasome activity from increased PA700 in the assay. The second peak of enhanced proteasome activity was identified in fractions with an apparent M r of ϳ300,000. Control assays indicated that this peak did not result from endogenous proteasome activity, which was very low in these column fractions (Ͻ1.5 units/assay) and was not observed unless exogenous proteasome, PA700, and MgATP were present in the assays (see further characterization below).
The fractions containing the PA700-dependent proteasomeactivating activity were pooled and subjected to further purification by ion-exchange chromatography on DEAE-Fractogel as described under "Materials and Methods." The PA700-dependent proteasome-activating activity bound to this resin and was eluted as a single peak at a position corresponding to ϳ100 mM NaCl (Fig. 2). The fractions containing the peak activities were pooled and subjected to hydroxylapatite column chromatography. The PA700-dependent proteasome-activating activity eluted from this resin as a single peak (Fig. 3). SDS-PAGE analysis of the fractions from the hydroxylapatite column showed that three major proteins (denoted with arrows in Fig.  3) had elution profiles indistinguishable from one another and were coincident with PA700-dependent proteasome activation. These proteins had apparent M r values of 50,000, 42,000, and 27,000. Retrospective analysis of the column fractions from the Fractogel ion-exchange chromatography by SDS-PAGE also showed that these three proteins coeluted with one another and with modulator activity (data not shown). The activity from the hydroxylapatite chromatography was subjected to a second Sephacryl S-300 chromatography step. The activity eluted at a position corresponding to its originally estimated M r of 300,000 and was coincident with the three proteins described above (data not shown). Therefore, we conclude that red blood cell extracts contain a PA700-dependent proteasome activator composed of protein subunits with M r values of 50,000, 42,000, and 27,000.
The Modulator Functions as a PA700-dependent Proteasome Activator-The purified PA700-dependent proteasome activator (for simplicity, hereafter termed modulator) was characterized with respect to its effect on proteasome activity. In the absence of PA700, the modulator had no effect on proteasome activity (Fig. 4). However, the modulator stimulated PA700-dependent proteasome activity up to 8-fold. The magnitude of this effect was deceased at very high concentrations of PA700 ( Fig.  4; see "Discussion"). We previously showed that PA700 activation of the proteasome required preincubation of both proteins in the presence of ATP (12). The modulator's effect on proteasome activity required the simultaneous preincubation of all three proteins with ATP. Preincubation of any individual protein or various combinations of any two proteins, followed by the addition of the other(s) immediately prior to the addition of the substrate for assay of proteasome activity, did not result in activated rates of PA700-dependent proteasome activity (data not shown). The modulator did not affect the rate of proteasome activation compared with that promoted by PA700 alone (data FIG. 1. Identification of a PA700-dependent activator (modulator) of the proteasome by gel filtration chromatography. Proteins from Fraction II that precipitated between 0 and 38% saturated ammonium sulfate (see "Materials and Methods") were solubilized and chromatographed on Sephacryl S-300. Column fractions were assayed for PA700 activity (E) and PA700-dependent activator (modulator) activity (q) as described under "Materials and Methods." PA700 activity was assessed using 5 l of column fractions and the purified exogenous 20 S proteasome (0.25 g/assay, 0.4 units). Modulator activity was assessed using 5 l of column fractions and the purified exogenous proteasome (0.25 g/assay) and purified exogenous PA700 (0.64 g/ assay, 7.6 units). Control assays for the endogenous proteasome in column fractions had Ͻ1.5 units/assay (not shown). The column fractions were also subjected to Western blotting with anti-TBP1 (top panel). To show all fractions, results from two different blots are shown. Standards of TBP1 on each blot produced bands of equal intensity.
FIG. 2. Ion-exchange chromatography of the PA700-dependent proteasome activator (modulator). Column fractions from the Sephacryl-S300 column (see Fig. 1) containing modulator activity (fractions 110 -125) were pooled and subjected to ion-exchange chromatography on DEAE-Fractogel as described under "Materials and Methods." Column fractions were assayed for PA700-dependent activator (modulator) activity. Samples (5 l) of the column fractions were assayed in the presence of the purified exogenous proteasome (0.25 g) and PA700 (0.67 g, 8.8 units).
FIG. 3. Hydroxylapatite chromatography of the PA700-dependent proteasome activator (modulator). Column fractions from the DEAE-Fractogel column (see Fig. 2) containing modulator activity (fractions 14 -21) were pooled and subjected to hydroxylapatite chromatography as described under "Materials and Methods." Column fractions were assayed for PA700-dependent activator (modulator) activity. Samples (5 l) of the column fractions were assayed in the presence of the purified exogenous proteasome (0.25 g) and PA700 (0.67 g, 5.3 units). Inset, column fractions were subjected to SDS-PAGE. Three protein bands (p50, p42, and p27), denoted with arrows, had elution profiles similar to one another and modulator activity. In this figure, the p27 protein did not reproduce well, although it was clearly visible on the original gel. not shown).
The Modulator Promotes Formation of an Activated Proteasome-containing Complex-To determine possible mechanisms of action of the modulator, proteasome/PA700-containing complexes formed in the presence and absence of the modulator were isolated by glycerol density gradient centrifugation. As reported previously, the proteasome and PA700 form a complex that is activated with respect to proteasome and that is much larger than either individual protein. Preincubation of all three proteins resulted in a complex that was 2-4-fold more active and significantly larger than the complex formed after preincubation of the proteasome and PA700 alone (Fig. 5).
The Modulator Contains Homologous Members of an ATPbinding Protein Family-To learn about the structural basis of modulator function, we subjected each of its subunit proteins to amino acid sequence analysis. The p50, p42, and p27 proteins were isolated by HPLC and digested with trypsin or Lys-C protease. The resulting peptides were isolated by HPLC, and selected peptides were subjected to automated Edman degradation as described under "Materials and Methods." Sequences were obtained for 12 peptides of the p50 subunit; the sequences contained 211 amino acids. Comparison of these sequences with those in current data bases showed that they exactly matched the sequence of a previously described human protein, TBP1 (human immunodeficiency virus Tat-binding protein) (21) (Fig. 6). TBP1 is a member of a large protein family that contains a consensus sequence for nucleotide binding (see below). Sequences were obtained for five peptides of the p42 subunit, containing 88 amino acids. Comparison of these sequences with proteins in current data bases indicated that p42 had significant sequence similarity to TBP1 (p50) as well as to other members of this protein family ( Fig. 7 and data not shown). It is not possible from these partial data to determine whether p42 is the clear homolog of any one member of the family. Eight tryptic peptides of the p27 subunit of the modulator were isolated and sequenced. 90 amino acids were identified, and these sequences had no significant similarity to those of any protein listed in current data bases (data not shown).
p50 (TBP1) and p42 Are Common Subunits of Two Different Proteasome Regulatory Complexes: PA700 and the Modulator-Four members of a protein family containing a consensus sequence for ATP binding have been identified as components of PA700 or of the 26 S protease, a large proteasome-containing complex that contains PA700 as its major, if not its only, non-proteasome component. These proteins include S4 (22), MSS1 (23, 24), TBP7 (15,25), and p45 (15,26). Although TBP1, a member of this family originally identified as a human immunodeficiency virus Tat-binding protein (21), has not been identified as a component of PA700 by direct sequencing, an antibody prepared against human TBP1 was shown to crossreact with a 50,000-Da subunit of PA700 from rat liver (6). We used this antibody to examine the bovine modulator and PA700 proteins. Fractions from the Sephacryl S-300 column on which PA700 and the modulator were first isolated contained a single immunoreactive band of 50,000 Da. This band was present in two peaks that where coincident with PA700 and the modulator, respectively (Fig. 1). Interestingly, more immunoreactive protein was present in the modulator peak than in the PA700 peak. The same immunoreactive band was observed in the purified PA700 and modulator proteins (Fig. 8). These results indicate that TBP1 (p50) is a subunit of each protein. To provide direct evidence for this conclusion, the PA700 subunit that reacted with the anti-TBP1 antibody was isolated by a twodimensional procedure involving HPLC and SDS-PAGE (15). Its retention time on HPLC was identical to that of the p50 (TBP1) subunit of the modulator (peak 4 of Fig. 9 and data not shown). The isolated PA700 subunit was subjected to digestion by Lys-C protease. The resulting peptides were isolated by HPLC and sequenced by automated Edman degradation as described under "Materials and Methods"; sequences were obtained for six peptides containing a total of 80 amino acids. Each of these sequences was identical to sequences of TBP1, FIG. 4. Effect of the modulator on proteasome activity. The purified modulator was tested for its ability to activate the purified proteasome. The indicated amounts of modulator were preincubated for 45 min with 0.25 g of proteasome in the presence (q) or absence (f) of PA700 (0.64, 2.5, or 5.0 g/assay) and in the presence of 100 M ATP prior to the assay for proteasome activity using succinyl-Leu-Leu-Val-Tyr 7-amino-4-methylcoumarin as a substrate. thereby establishing the identity of the p50 subunit of PA700 as TBP1 and demonstrating that this protein is a subunit of both PA700 and the modulator (Fig. 6). In light of this surprising result, we next determined whether the other two modulator subunits, p42 and p27, might also be components of PA700.
Because we do not yet have antibodies against p42 or p27, we compared their two-dimensional separation patterns with those of PA700 subunits. The p42 subunit of the modulator had a retention time on HPLC that was the same as that of peak 5 of PA700 ( Fig. 9 and data not shown). SDS-PAGE of peak 5 showed that it contained a subunit of a size indistinguishable from that of modulator subunit p42 (Fig. 10). The p42 subunit of PA700 was subjected to digestion by Lys-C protease followed by separation of the resulting peptides by HPLC. Automated Edman degradation of three of these peptides yielded 54 amino acids, and these sequences were identical to three peptides sequenced from p42 of the modulator (Fig. 6). Thus, p42, like p50 (TBP1), is a component of both PA700 and the modulator. The p27 subunit of the modulator did not have a retention time on HPLC similar to any component of PA700 ( Fig. 8 and data  not shown). Furthermore, no PA700 subunit in this molecular weight range had a sequence similar to that of p27 ( Fig. 8 and  data not shown). Therefore, there is no current evidence that p27 is a subunit of PA700. DISCUSSION We have identified a new multisubunit protein complex that regulates proteasome function. Unlike previously identified proteasome regulators, this new protein, referred to here as modulator, does not directly influence proteasome activities, but enhances, by up to 8-fold, the effect of PA700, an ATP-de-FIG. 6. The p50 subunits of the modulator and PA700 are identical to one another and to TBP1. The p50 subunit of the modulator was isolated by HPLC and SDS-PAGE and subjected to amino acid sequencing as described under "Materials and Methods." The sequences of 12 peptides produced by Lys-C digestion were determined and are shown aligned with the complete sequence of human TBP1 (21). The p50 subunit of PA700, identified by its reactivity to an anti-TBP1 antibody, was isolated and subjected to sequencing in the same manner as the modulator subunit. The sequences of five peptides were determined and are shown aligned with TBP1. FIG. 7. The p42 subunits of the modulator and PA700 are identical to one another and are homologous to p50 (TBP1). The p42 subunit of the modulator was isolated by HPLC and SDS-PAGE and subjected to amino acid sequencing as described under "Materials and Methods." The sequences of five peptides produced by Lys-C digestion were obtained and are shown aligned with peptides from p50. A dash denotes an identical amino acid at a given position. The p42 subunit of PA700 (isolated as described under "Materials and Methods") was subjected to amino acid sequencing, and three peptides (denoted by asterisks) had identical sequences to the corresponding peptides from p42 of the modulator.
FIG. 8. The anti-TBP1 antibody cross-reacts with a subunit of both the modulator and PA700. Left panel, PA700 (2 g) and the modulator (0.6 g) were subjected to SDS-PAGE and stained with Coomassie Blue; right panel, PA700 (4 g) and the modulator (1 g) were subjected to immunoblotting with an antibody against TBP1 as described under "Materials and Methods." Protein standards are indicated.
FIG. 9. Identities of PA700 subunits, including six members of the ATP-binding protein family. Purified PA700 was subjected to two-dimensional analysis as described under "Materials and Methods." The individual peaks resolved by HPLC (peaks 1-12) were electrophoresed on a SDS-polyacrylamide gel. Individual subunits whose identities are established, including the six members of the ATP-binding protein family, are denoted with arrows and are as follows. peak 1: upper band, S4 (22), and lower band, p45 (15,26); peak 2: MSS1 (23, 24); peak 3: p40 (Mov34) (7,33,34); peak 4: p50 (TBP1) (21,35); peak 5: upper band, TBP7 (15,25,35), and lower band, p42 (this report); peak 6: p58 (P91A) (15,36); peak 10: p31 (Nin1p) (6); peak 11: upper band, p112 (Sen3p) (15) (GenBank TM accession number L06321), and lower band, p97 (15,16); peak 12: p44 (HUMORF07) (16). pendent proteasome activator. Because the modulator's stimulatory effect magnifies a 50 -200-fold stimulatory effect by PA700, its influence on total proteasome activity is very large. The mechanism by which the modulator exerts its effect is presently unclear. PA700 activation of the proteasome involves the formation of a proteasome-PA700 complex in which PA700 binds to one or both of the proteasome's terminal rings. The finding that the modulator promotes the formation of a larger complex than that which is formed in its absence (Fig. 5) suggests two possible mechanisms for its action, which are not mutually exclusive and do not represent all possible mechanisms. First, the modulator might form a ternary complex with PA700 and the proteasome. Alternatively, the modulator could promote the formation of more complexes in which the proteasome is bound to two, rather than just one, PA700 molecule. Each of these models would account for the larger proteasomecontaining complex caused by the modulator, and structural changes associated with each presumably result in increased proteasome activation.
Two modulator subunits, p50 and p42, are homologous to one another and are members of a large protein family that contains a consensus sequence for ATP binding (17). The p50 subunit is identical to TBP1, previously identified as a human immunodeficiency virus Tat-binding protein (21), while p42 seems to be a new family member. We are currently determining the complete primary structure of p42 to confirm this latter conclusion. In any case, this work demonstrates that p50 and p42 modulator subunits are also subunits of PA700. Four other members of this protein family, S4, MSS1, p45, and TBP7, previously were shown to be subunits of PA700. Thus, the results presented here demonstrate that PA700 contains at least six members of this ATPase family.
The surprising finding that p50 and p42 are common subunits of two distinct proteasome regulators raises obvious questions regarding the origin of the modulator protein, on the one hand, and the basis for the identification of p50 and p42 as PA700 subunits, on the other. For example, could the modulator represent a subcomplex of PA700, derived from the dissociation of PA700? To address this question, we attempted to generate the modulator from purified PA700 by treating purified PA700 (which contains p50 and p42 as major components as judged by staining intensity; Fig. 9) with a variety of chaotropic and other agents (including 38% ammonium sulfate) and then subjecting the protein to gel filtration chromatography or to density gradient centrifugation. The p50 and p42 subunits of PA700 always migrated coincidently with PA700 activity and with the rest of the PA700 subunits, and we never observed formation of the modulator from PA700 in these experiments. 2 The possible origin of the modulator as a dissociated subcom-plex of PA700 also would imply the existence of a complementary PA700 subcomplex devoid of p50 and p42 and suggests that modulator activity might represent a "reconstitution" effect, i.e. that the restoration of critical subunits to a functionally incompetent form of PA700 restores PA700 activity. We attempted to detect such PA700 species in our preparations by fractionating purified PA700 by gel filtration or glycerol density gradient centrifugation and then assaying the fractions for PA700 activity and for the ability to be stimulated by the purified exogenous modulator. In all cases, both of these activities were exactly coincident with one another and with all PA700 subunits (including p50 and p42) detected by SDS-PAGE. Such results would not be expected if the modulator had a selective effect on a subpopulation of PA700 that lacked p50 and p42 because such a population should have a lower molecular weight than native PA700. Thus, we have been unable to provide any evidence that the modulator is derived from dissociation of PA700, although our current analysis cannot rigorously exclude this possibility.
Could the identification of p50 and p42 in PA700 preparations represent contamination by the modulator, either nonspecifically or as the result of a specific interaction between PA700 and the modulator? Nonspecific contamination seems unlikely because PA700 and the modulator were well separated from each other early in the purification (Fig. 1) and differed significantly in their chromatographic behavior on several additional columns. Furthermore, both p50 and p42 represented major PA700 components (Fig. 9). Although we cannot completely exclude the possibility that our PA700 preparations contain complexes formed by the specific interaction of the modulator and PA700, the failure to dissociate the modulator from PA700 preparations (as described above) seems to argue against this possibility. Therefore, these various results support the conclusion that the modulator represents a distinct protein complex that shares two subunits, p50 (TBP1) and p42, with PA700.
The sharing of p50 and p42 between two different proteins suggests that these subunits could have functions common to each of the complexes. Because p50 and p42 are members of the ATP-binding protein family, such a function might involve the role of ATP in proteasome activation, particularly in the assembly of the proteasome-containing complex. We are currently examining this possibility. In any case, another member of the ATP-binding protein family, p45, has been identified in at least two different multiprotein complexes: PA700 (15) and the transcriptional mediator complex that contains Sug1p, the homolog of p45 in yeast (26 -28). It seems possible that other members of this ATP-binding protein family will be shared among different protein complexes.
The regulation of proteasome function by the combined action of two proteins, as shown here, is reminiscent of early reports by Hershko and co-workers (29) showing the requirement for two factors, termed CF1 and CF2, for proteasome activation. Therefore, it is reasonable to question the possible relationship between PA700 and the modulator, on the one hand, and CF1 and CF2, on the other. Direct comparison between these protein pairs is difficult because CF1 and CF2 were not purified. Nevertheless, as discussed previously, we believe that CF1 is structurally and functionally related, but not identical, to PA700 (12). In contrast, there are significant differences between the modulator and CF2. Goldberg and co-workers (30) purified a protein with functional properties indistinguishable from CF2 (i.e. the ability to activate the proteasome in a CF1-dependent manner in a reconstitution system where proteasome activation required both proteins). In the absence of CF1, this CF2 protein functioned as a protea-2 R. J. Proske and G. N. DeMartino, unpublished observations. FIG. 10. Comparison of modulator and PA700 subunits by SDS-PAGE. Subunits from the modulator and PA700 were compared by SDS-PAGE. Lanes 1 and 5, 1.5 g of purified modulator; lanes 2-4, PA700 subunits isolated by HPLC: lane 2, p50 subunit (peak 4 from Fig.  9); lane 3, p42 subunit (peak 5 from Fig. 9); lane 4, peak 9 from Fig. 9 showing no common band with p27 from the modulator. some inhibitor and was judged identical to a previously identified proteasome inhibitor with a subunit size of 40,000 Da (31). We have failed to detect inhibition of the proteasome by the modulator under a variety of assay conditions. 2 Furthermore, the 40,000-Da inhibitor was subsequently shown to have the same structural, functional, and immunological properties as ␦-aminolevulinic-acid dehydratase (32). These various findings clearly distinguish the modulator described here from previously described CF2 proteins and suggest that the modulator may be one of several proteins that regulate proteasome function indirectly through PA700.