The Proteasome Activator 11 S Regulator or PA28

The proteasome 11 S regulator (REG) consists of two homologous subunits, REGα and REGβ. Each subunit is capable of activating the proteasome, and when combined, they form superactive REGα/REGβ complexes. We have previously shown that a highly conserved loop in the REGα crystal structure is critical for proteasome activation. We now show that hetero-oligomers formed from REGα activation loop mutants and wild-type REGβ or vice versa are partially active. By contrast, hetero-oligomers bearing mutations in the activation loops of REGα and REGβ subunits are inactive, demonstrating that both α and β subunits contribute to proteasome activation. We have also characterized REG proteins with mutations near or at their C termini. Partially active REGα(Y249C) and REGα(M247V) and an inactive REGα subunit bearing five additional C-terminal amino acids formed active hetero-oligomers with REGβ. REGβ subunits lacking 1, 2, or 9 C-terminal amino acids did not bind or activate the proteasome, but each of these mutants formed partially active hetero-oligomers with the monomer REGα(N50Y). However, hetero-oligomers formed from REG subunits lacking the last 14 amino acids were unable to bind the proteasome. Thus, C-terminal regions of both α and β subunits are required for hetero-oligomers to bind the proteasome.

The proteasome is a large proteolytic enzyme found in all three kingdoms, the archeae, prokaryotes, and eukaryotes (1)(2)(3)(4)(5)(6)(7). In higher eukaryotes, the proteasome is believed to play an important role in a variety of cellular processes, including cell cycle progression (8,9), control of gene expression (10), and antigen presentation (11)(12)(13). The proteasome from the archaebacterium Thermoplasma acidophilum is formed from 14 identical ␣ subunits and 14 identical ␤ subunits. To date, 17 distinct subunits have been identified in proteasomes from higher eukaryotes. All can be classified into ␣ or ␤ families based on their homology to the ␣ or ␤ subunits of the Thermoplasma proteasome (13). The quaternary structures of yeast and Thermoplasma proteasomes are quite similar, as revealed by electron microscopy (14 -16) and confirmed by x-ray crystallography (17,18). The enzymes are composed of four stacked rings with seven subunits in each ring. The two inner rings are formed from ␤ subunits, and the two outer rings consist of ␣ subunits. The proteolytic active sites are located in an inner chamber of the proteasome, with the N-terminal threonines on the ␤ sub-units acting as nucleophiles (17)(18)(19). These active sites are not readily accessible from the cytosol, because only two 13-Å pores are present in the ␣ rings of the archaebacterial proteasome, and even these small pores are absent in the crystallized yeast proteasome (17,18). Therefore, some mechanisms must exist to promote substrate access to the proteasome's active sites.
Several proteins have been found to bind the proteasome (20). Of these, the 19 S regulatory complex (21)(22)(23)(24) and 11 S regulator (REG) 1 , or PA28 (25,26), are best characterized. The 19 S regulatory complex consists of at least 15 different subunits (27). It binds to the proteasome in a nucleotide dependent reaction to form the 26 S protease, which degrades intact proteins, including those marked by polyubiquitin chains (28). Association of the 11 S REG with the proteasome does not require energy and increases the ability of the proteasome to cleave fluorogenic peptide substrates in vitro (25,26). The 11 S REG from human red blood cells are ring-shaped complexes composed of two homologous subunits, REG␣ and REG␤ (29 -31). Another REG protein, REG␥, which is homologous to REG␣ and REG␤ in sequence and forms homo-oligomers, also activates the proteasome. Recombinant REG␣ forms a heptamer, as shown by x-ray crystallography (32) and equilibrium sedimentation studies (33), whereas REG␤ is monomeric under many conditions (34). REG␣, by itself, is capable of activating the proteasome, but to a lesser extent than REG␣/REG␤ hetero-oligomers (34). We have shown that amino acids Arg 141 to Gly 149 in REG␣ are critical for proteasome activation, and this region forms a loop in the REG␣ crystal structure (35). Thus, we call this region the activation loop. There are two reports that REG␤ is inactive (36,37), and it has recently been proposed that REG␤ functions only to modulate the activity of REG␣ (37). However, several publications from our laboratory have shown that REG␤, by itself, activates the proteasome (34,35,38). Moreover, a single-site mutation in the presumed activation region of REG␤ produces an inactive protein, REG␤(N135Y), that is still capable of binding the proteasome (35).
Although we know that REG␤ activates the proteasome, our previous studies did not directly address whether REG␤ contributes to proteasome activation in REG␣/REG␤ hetero-oligomers. Consequently, we have now measured proteasome activation by REG␣/REG␤ hetero-oligomers formed from wildtype REG␣, wild-type REG␤, and ␣ or ␤ subunits bearing mutations in the activation loops. In contrast to the conclusions reached by DeMartino and colleagues that REG␤ serves only to modulate REG␣ activity (37), we have found that in the 11 S REG both ␣ and ␤ subunits contribute to proteasome activation. We also provide evidence that C-terminal regions of the three homologs, REG␣, REG␤, and REG␥, play an important role in proteasome binding.
Generation of REG␣ and REG␤ Mutants with Altered Activities-All of the single-site REG␣ mutants used in the study were isolated by screening randomly mutagenized REG␣ cDNA libraries as described (35). REG␣241␥8, a REG␣ variant bearing the eight C-terminal amino acids of REG␥, was produced by the Kunkel site-directed mutagenesis method. This method was also used to obtain the REG␤ mutants described in these studies. A detailed description on how to make a REG␤ mutant and REG␣241␥8 using this method can be found in an article by Zhang et al. (38).
Expression and Purification of REG␣, REG␤, and Their Mutants-Expression of REG␣ and REG␤ subunits was performed essentially as described (35,38). Briefly, bacteria containing plasmids with the desired mutation were grown at 30°C and induced for 2 h with either 0.6 mM isopropyl-␤-thiogalactopyranoside when the A 600 was between 0.2 and 0.3 or 0.8 mM isopropyl-␤-thiogalactopyranoside when the A 600 was between 0.4 and 0.5. For most REG␣ mutants, sufficient protein could be purified from 1 liter of LB (10 g of tryptone, 5 g of yeast extract, 5 g of NaCl/liter, pH 7.5) culture for the biochemical studies described below. Wild-type and mutant REGs were purified as described (34). Each REG␣ mutant was at least 85% pure as estimated by Coomassie Blue staining of overloaded 15% SDS-polyacrylamide gels. The concentration of REG␣ or a REG␣ mutant protein was determined by measuring the absorbance at 280 nm or by using the Pierce Coomassie Blue protein assay. The extinction coefficient of each REG␣ variant was calculated based on its amino acid composition according to the method described by Gill and Von Hippel (39). The concentrations of those selected REG␣s determined by both methods agreed well with each other within the experimental errors. The concentration of REG␤ or REG␤ mutants was determined only by the Pierce Coomassie Blue protein assay.
REG␤ Filter Binding Assay-This assay was performed essentially the same as described (34).
Gel Filtration Chromatography-A Superdex 200 column (Hiload 26/60, Amersham Pharmacia Biotech) was used to purify all of the proteins used in these studies and to determine whether a REG␣ mutant interacts with REG␤. To determine whether a given REG␣ mutant binds REG␤, equal amounts of the REG␣ mutant and REG␤ were mixed and incubated overnight at 4°C. The mixture was then loaded to the Superdex 200 column and eluted with TSD (10 mM Tris, 10 mM NaCl, 25 mM KCl, 1.1 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, pH 7.2), 175 mM KCl. Selected fractions were then analyzed on 15% SDS gels and/or assayed to determine whether active REG␣/REG␤ hetero-oligomers formed. To purify REG␣/REG␤ hetero-oligomers, we applied the same incubation procedure described above, and proteins from fractions containing REG␣/REG␤ hetero-oligomers were pooled, concentrated, and dialyzed against 0.5 ϫ TSD before use.
Proteasome Activation by REGs or Their Mutants-Purified human red blood cell proteasomes were used for all of the activity assays. Typically, 170 ng of proteasomes in 50 l of 10 mM Tris, pH 7.5, were incubated for 10 min at 37°C with REGs or mutants at the concentrations indicated in the figure legends. Reactions were started by adding 50 l of 200 M fluorogenic peptide LLVY-MCA. After additional incubation, as indicated in the figure legends, the reactions were quenched with 200 l of cold ethanol, and the fluorescence was measured as described (34).
Proteasome Binding Assay-The proteasome binding assay was performed essentially as described (34,35). Briefly, REG␣, REG␤, and their variants (e.g. mutants and REG␣/REG␤ hetero-oligomers) were incubated with proteasomes tethered to an enzyme-linked immunosorbent assay plate. After washing away unbound proteins, proteins that bound the proteasome were eluted with a high-salt buffer (20 mM Tris, pH 7.5, 0.5 M NaCl), transferred to a nitrocellulose membrane, and detected with either REG␣-or REG␤-specific antiserum as described (34,35).
Analysis of REG␤-to-REG␣ Ratio in Hetero-oligomers Using HPLC-The apparent molar ratio of REG␣ and REG␤ subunits in heterooligomers was determined by HPLC as described (38).

RESULTS
Proteasome Activation by Hetero-oligomers Formed from Mutant REG␣ Subunits-We previously characterized 36 singlesite REG␣ mutants. Most are monomers that are unable to bind or activate the proteasome (35). Because 11 S REG consists of both REG␣ and REG␤ subunits (29 -31), we wanted to determine whether each REG␣ mutant forms hetero-oligomers with REG␤, and if formed, whether they are able to activate the proteasome. Thirty-four REG␣ mutant proteins were dot-blotted onto a nitrocellulose filter and probed with [ 35 S]methionine-labeled REG␤. Twenty-five of the inactive REG␣ mutants bound to REG␤, three bound REG␤ less tightly than wild-type REG␣, and six did not bind REG␤ (Table I). Each of the six REG␣ mutants that failed to bind REG␤ in the filter assay, and REG␣(D205V), a weak REG␤ binder, were mixed with an equal amount of REG␤ and incubated overnight at 4°C. These mixtures were then analyzed on a Superdex 200 column. Whereas REG␣(R181C) (Fig. 1, B and D, and Table I) and REG␣(D205V) ( Table I) failed to form active oligomers, the other five REG␣ mutants did (for example, see Fig. 1, A and C; also see Table I). As an indirect method to assay for heterooligomer formation, we measured proteasome stimulation by each inactive REG␣ mutant after mixing with wild-type REG␤. Such an assay is possible because at low concentrations (Ͻ10 g/ml) REG␤ barely stimulates the peptidase activity of the proteasome (34). Twenty six of 36 mtREG␣/REG␤ combinations were as active as hetero-oligomers formed from wild-type subunits; 9 were partially active compared with REG␣/REG␤ but considerably more active than wild-type REG␤ alone. Only one combination, REG␣(R181C)/REG␤, did not stimulate the proteasome more than the REG␤ control (Table I). Seven mutants that bound REG␤ on filters and showed enhanced proteasome activation when mixed with REG␤ were further analyzed using gel filtration chromatography. Each formed active heterooligomers with REG␤ (Table I). Taken together, these analyses lead to the conclusion that in our collection of REG␣ mutants, all variants, except REG␣(R181C) and possibly REG␣(D205V), form hetero-oligomers with REG␤. Moreover, in every case the resulting hetero-oligomers activate the proteasome.
REG␣ and REG␤ Subunits Contribute to Proteasome Activation in REG␣/REG␤ Hetero-oligomers-Of the nine REG␣ mutants that formed partially active mtREG␣/REG␤ hetero-oligomers (Table I), seven are altered at REG␣ residues critical for proteasome activation (35). The seven mutants include five in the region Arg 141 -Gly 149 that forms the activation loop (32,35), and the other two are mutations at Pro 240 . Because we have previously shown that all of these mutants formed inactive hetero-oligomers with REG␤(N135Y), a REG␤ subunit mutated in the activation loop (35), these experiments strongly suggest that in the 11 S REG, wild-type REG␤ activation loops contribute to proteasome activation. As shown in Table I, hetero-oligomers bearing REG␣ subunits with altered activation loops activated the proteasome to differing extents. For example, REG␣(R141W)/REG␤ retained ϳ30% of the activity of REG␣/REG␤ hetero-oligomers, whereas REG␣(N146S)/REG␤ was 80% active. Therefore, depending on the specific mutation in the REG␣ activation loop, proteasome activation by heterooligomers is differentially affected. Still, the extent of activation is close to 50% for most combinations.
In the 11 S REG, ␣ and ␤ subunits are present in about equal amounts (36); thus these results are consistent with the idea that REG␤ subunits activate the proteasome. Alternatively, one could argue that REG␤ subunits function only to alter the conformation of the mutated REG␣ activation loop. By this hypothesis, loss of proteasome activation by hetero-oligomers formed from REG␤(N135Y) and REG␣ activation loop mutants would result from failure of REG␤(N135Y) to "rescue" the conformation of altered REG␣ activation loops. To distinguish among these possibilities, we converted Gly 139 , a residue in the activation loop of REG␤, to Glu. We also mutated Asn 53 and Ser 165 , two residues located outside the activation loop of REG␤, to Tyr and Phe, respectively. These three REG␤ variants correspond to REG␣(G149E), REG␣(N50Y), and REG␣-(S175F), respectively (the corresponding amino acids between REG␣ and REG␤ are defined according to the sequence alignment in Ref. 34). All of these REG␤ variants were inactive (data not shown). Subsequently we measured proteasome activation by a series of hetero-oligomers formed from "reciprocal" combinations of mutant subunits.
We first determined proteasome activation by hetero-oligomers containing wild-type REG␣ and REG␤(S165F) or REG␤ and REG␣(S175F). A fixed amount of REG␣ was mixed with increasing amounts of REG␤(S165F) or vice versa and incubated overnight to allow formation of hetero-oligomers, and their effects on proteasome activation were measured. Maximal proteasome activation by REG␣/REG␤(S165F) and REG␣-(S175F)/REG␤ hetero-oligomers were virtually identical ( Fig.  2A) and equivalent to the stimulation reached by mixing wildtype REG␣ and wild-type REG␤ subunits (data not shown). Thus, converting Ser residues, located outside the activation loops of both REG␣ and REG␤ subunits, to Phe does not affect proteasome activation by REG␣/REG␤ hetero-oligomers. We then measured proteasome activation by fixing the amount of REG␣ and increasing REG␤(N135Y) or by fixing the level of REG␤ and increasing REG␣(N146Y). REG␣(N146Y) and wildtype REG␤ hetero-oligomers stimulated the proteasome to a much greater extent than hetero-oligomers formed from wildtype REG␣ and REG␤(N135Y) (Fig. 2B). Moreover, the maximal proteasome activation by hetero-oligomers REG␣/REG␤-(N135Y) or REG␣(N146Y)/REG␤ is lower than that of wild-type REG␣/REG␤ hetero-oligomers. Clearly, converting the Asn to Tyr in the activation loops of REG␣ and REG␤ affects proteasome activation by hetero-oligomers. Moreover, in this combination the Asn to Tyr mutation in the REG␤ activation loop reduces proteasome activation more than does the corresponding mutation in the REG␣ activation loop. A direct proteasome binding assay revealed that proteasome binding by REG␣/ REG␤(N135Y) and REG␣(N146Y)/REG␤ hetero-oligomers was quite similar (Fig. 2, C and D). Thus, the differences in proteasome activation by REG␣/REG␤ hetero-oligomers with Asn to Tyr mutations in either REG␣ or REG␤ activation loops are probably not attributable to their proteasome binding affinities.
In the experiment shown in Fig. 2B, it seems that the contribution of REG␤ to proteasome activation in REG␣/REG␤ hetero-oligomers is greater than that of REG␣. However, the differential contribution of REG␣ and REG␤ subunits in this reciprocal combination might be specific for these pairs. This is, indeed, the case, as shown by results from other reciprocal hetero-oligomers. For example, REG␣/REG␤(G139E) was as active as REG␣(G149E)/REG␤, and both were ϳ50% as active as hetero-oligomers formed from wild-type REG␣ or REG␤ subunits (Table II). Examination of the data in Table II shows that mutations in the activation loops of REG␣ or REG␤ produce REG␣/REG␤ hetero-oligomers that vary in their ability to activate the proteasome. However, the data in Table II provide a consistent picture of proteasome activation by hetero-oligomers. When both subunits contain mutations in their activation loops, the resulting hetero-oligomer is inactive. When either REG␣ or REG␤ subunits contain an activation loop substitution, the hetero-oligomers are partially active. Mixing REG␣ or REG␤ variants bearing mutations outside the activation loops results in fully active hetero-oligomers. These results strongly support the idea that the activation loops of both REG␣ and REG␤ contribute to proteasome activation by hetero-oligomers ( Fig. 3; see "Discussion").  (35). Note that residues critical for proteasome activation are bold face, and REG␣(R181C) that cannot form heptamers with REG␤ is in italics. Second column, ability of REG␣ mutants to bind REG␤ determined by a REG␤ filter binding assay. Each mutant protein (2 g) was dot-blotted on a nitrocellulose membrane in triplicate. After blocking the membrane with 5% nonfat dry milk in Tris-buffered saline Tween 20, the membrane was probed with [ 35 S]methionine-labeled REG␤ and subjected to PhosphorImager analysis. ϩϩ, REG␣ mutant that bound REG␤ as tightly as did wild-type REG␣; ϩ, those that bound less well than wild-type REG␣; Ϫ, those that failed to exhibit any binding. Third column, ability of REG␣ mutants to bind REG␤ tested by gel filtration chromatography. Each of the mutants that failed to bind REG␤ in the REG␤ filter binding assay was mixed with an equal amount of REG␤ (final concentration, 0.3 mg/ml each), incubated overnight at 4°C, and subjected to a gel filtration analysis as described under "Experimental Procedures." ϩ, REG␣ mutant that formed a hetero-oligomer with REG␤; Ϫ, a REG␣ mutant that did not form a hetero-oligomer with REG␤; ND, not determined. Fourth column, proteasome activation by mtREG␣/REG␤ hetero-oligomers. 1 g of a mutant REG␣ was mixed with 1 g of wild-type REG␤ and 170 ng of proteasome in 50 l of 10 mM Tris buffer, pH 7.5. After a 10-min incubation, the reaction was initiated by adding 50 l of 200 M LLVY-MCA. The reaction was quenched after 10 min, and the fluorescence was measured as described (34). ϩϩ, mtREG␣/REG␤ hetero-oligomers that exhibited Ͼ90% of the activities of the wild type REG␣/REG␤ hetero-oligomers; ϩ, those that exhibited Ͻ90% activation compared with REG␣/REG␤ but higher activation than REG␤ (1 g) control. The numbers in parentheses are the activities of mtREG␣/REG␤ heterooligomers expressed as a percentage of the activities of wild-type REG␣/ REG␤ hetero-oligomers. Only one mutant exhibited a similar level of proteasome activation compared with REG␤ alone (thus designated as Ϫ).

REG␣ variants
Binding to REG␤ Hetero-oligomer activity (mtREG␣/REG␤) Filter assay Gel filtration assay The C-terminal Regions of Both REG␣ and REG␤ Subunits Are Required for the 11 S REG to Bind the Proteasome-Based on deletion or mutation of REG␣ Tyr 249 , Song et al. (37) have proposed that the C-terminal residue is important for proteasome activation. In our screen for inactive REG␣ subunits, we obtained four REG␣ mutants at or near the C terminus. The four mutants include two single-site variants, REG␣(Y249C) and REG␣(M247V), one variant lacking the last 14 amino acids (REG␣⌬14), and one variant extended by five additional amino acids (REG␣ϩ5). We have purified and characterized these four proteins (Table III). REG␣(Y249C), REG␣(M247V), and REG␣ϩ5 formed heptamers as judged by gel filtration chromatography, whereas REG␣⌬14 partitioned between monomers (55%) and heptamers (45%). Proteasome activation assays showed that REG␣(Y249C) and REG␣(M247V) were partially active compared with wild-type REG␣ (Table III). Although both REG␣⌬14 and REG␣ϩ5 were inactive, REG␣ϩ5 bound the proteasome. By contrast, REG␣⌬14 binding to the proteasome was not detected (Table III). All four REG␣ variants formed hetero-oligomers with wild-type REG␤, and those containing REG␣(Y249C), REG␣(M247V), or REG␣ϩ5 were fully active. However, REG␣⌬14/REG␤ hetero-oligomers did not bind or activate the proteasome (Table III), thereby demonstrating that some portion of the last 14 amino acids of REG␣ is necessary for association of REG␣/REG␤ hetero-oligomers with the proteasome.
To determine whether the C terminus of REG␤ is also needed for proteasome binding, we produced recombinant REG␤ subunits lacking 1, 2, 9, and 14 C-terminal amino acids and characterized the four mutant proteins, designated REG␤⌬1, REG␤⌬2, REG␤⌬9, and REG␤⌬14. All four truncated subunits behaved as monomers after gel filtration chromatography (data not shown). None of the four REG␤ variants acti-vated the proteasome (Fig. 4B), nor did any of them bind the proteasome under conditions in which wild-type REG␤ did (data not shown). Therefore, the C-terminal Tyr of REG␤ is required for this subunit to bind the proteasome. Whether C-terminal portions of REG␤ are involved in hetero-oligomerization was determined by mixing each mutant with an equal amount of REG␣(N50Y), an inactive monomeric REG␣ mutant. These mixtures were then subjected to gel filtration analysis, and it was found that each REG␤ C-terminal deletion variant formed hetero-oligomers with REG␣(N50Y) (data not shown). In addition, we determined the ratio of REG␤ subunit to REG␣(N50Y) in each hetero-oligomer by HPLC (see "Experimental Procedures"). The REG␤-to-REG␣ molar ratios in all four hetero-oligomers ranged from 1.2 to 1.4, equivalent to the ratio of hetero-oligomers formed from wild-type REG␤ and REG␣(N50Y) (data not shown). From these results we conclude that deleting up to 14 amino acids from REG␤ did not affect its ability to associate with REG␣(N50Y), and the resulting heterooligomers contained a fixed ratio of REG␣ and REG␤ subunits.
A REG␣ Variant with the REG␥ C Terminus Binds the Proteasome Tighter than Does Wild-type REG␣-REG␣ and REG␥ have ϳ30% sequence similarity and form homo-oligomers that activate the proteasome. However, REG␥ binds the proteasome tighter than REG␣ (34). Moreover, they stimulate peptidase activities of the proteasome differentially. REG␥ preferentially activates hydrolysis of fluorogenic peptides with positive charged residues at the P1 position, whereas REG␣ activates the proteasome to cleave after a variety of residues at the P1 position (34). To further address the function of the REG C

FIG. 2. Corresponding Asn to Tyr mutations in the activation loops of REG␣ and REG␤ differentially affect proteasome activation.
A, proteasome stimulation by hetero-oligomers formed from REG␣ with increasing amounts of REG␤(S165F) (squares) or by hetero-oligomers formed from REG␤ with increasing amounts of REG␣(S175F) (diamonds). 0.75 g of wild-type REG␣ or REG␤ was incubated with the indicated amount of REG␤(S165F) or REG␣(S175F) and 170 ng of proteasomes in a volume of 50 l for 15 min at 37°C. Then 50 l of 200 M LLVY-MCA were added to start the reaction. After an additional 10 min, the reaction was quenched and fluorescence was measured. Because REG␣/REG␤ hetero-oligomers bind the proteasome much tighter than either REG␣ or REG␤ alone, the observed proteasome activation by a mixture of REG␣/REG␤ hetero-oligomers and REG␣ or REG␤ predominately results from hetero-oligomers (34). B, proteasome stimulation by REG␣ with increasing amounts of REG␤(N135Y) (squares) and REG␤ with increasing amount of REG␣(N146Y) (diamonds). The experiment was performed as described in Fig. 2A. Note that the decreased proteasome activation by hetero-oligomers in the presence of excess amounts of REG␣(N146Y) or REG␤(N135Y) could result from either the inhibitory effect of these two mutants (38) or the formation of REG␣/REG␤ hetero-oligomers containing more inactive subunit REG␣(N146Y) or REG␤(N135Y). C and D, Proteasome binding by REG␣(N146Y)/REG␤ and REG␣/REG␤(N135Y) probed with anti-REG␣-specific antiserum (C) and with anti-REG␤-specific serum (D). Equal amounts of REG␣ and REG␤(N135Y) or REG␣(N146Y) and REG␤ were mixed, incubated at 4°C overnight, and used to perform direct proteasome binding assay as described under "Experimental Procedures." Hetero-oligomer concentrations used to incubate the proteasomes are indicated at the left. Note that results presented in C and D are from two independent experiments. Control, signals obtained when 30 g/ml hetero-oligomers were incubated with wells coated with antibodies but in the absence of proteasomes. terminus, we produced a REG␣-REG␥ chimera, REG␣241␥8, in which the last eight amino acids of REG␣ are substituted by those found in REG␥. The REG␣241␥8 chimera bound the proteasome much tighter than did wild-type REG␣ (Fig. 7). However, the proteasome cleavage specificity activated by REG␣241␥8 was identical to REG␣ and distinct from that seen with REG␥ (data not shown). Therefore, the C-terminal region of REG␥ significantly increases proteasome binding by REG␣ when it replaces the corresponding region of REG␣ but has little, if any, effect on proteasome activation by REG␣.

TABLE III
Biochemical properties of REG␣ C-terminal mutants REG␣ mutants or REG␣ as indicated at the top of each column were purified to homogeneity. The biochemical properties of each REG␣ mutant were characterized as described (35). First row, proteasome activation by each mutant or wild-type REG␣. The activity is expressed as a percentage of wild-type REG␣ activity. Ϫ, mutants that did not activate the proteasome under the experimental conditions. Second row, the oligomerization state of each mutant is expressed as heptamer percentage of the total species (sum of monomer and heptamer) as revealed by a Superdex 200 column. Third row, proteasome binding of each mutant. Ϫ, a mutant that did not exhibit proteasome binding at either 60 or 20 g/ml concentration; ϩ, a mutant that showed binding at 60 but not at 20 g/ml; ϩϩ, REG␣ mutants that exhibited binding at both 60 and 20 g/ml. Fourth row, proteasome activation by mtREG␣/ REG␤ hetero-oligomers. Each REG␣ mutant was mixed with REG␤ and incubated at 4°C overnight, and the resulting hetero-oligomers were assayed for proteasome activation. Hetero-oligomer REG␣⌬14/REG␤ was purified by gel filtration chromatography before the proteasome activation and proteasome binding assays. Ϫ, a hetero-oligomer that did not exhibit any activity; ϩϩϩ, those that were as active as wild-type REG␣/REG␤ hetero-oligomers. Fifth row, proteasome binding by mtREG␣/REG␤ hetero-oligomers. Ϫ, a hetero-oligomer that did not exhibit binding at either 60 or 20 g/l; ϩϩϩ, hetero-oligomers that bound the proteasome tighter than wild-type (wt) REG␣ alone; ND, not determined.  REG␣ and REG␤ variants Equal amounts of REG␣ and REG␤ subunits were mixed and incubated overnight at 4°C. The resulting hetero-oligomers were either purified by gel filtration chromatography and then used to perform proteasome activation assays or used directly for proteasome activation assays as described under "Experimental Procedures." For each activation assay, 2 g of hetero-oligomer protein and 170 ng of proteasome were used. The activity was expressed as a percentage of the activity of hetero-oligomers formed from wild-type (WT) REG␣ and wild-type (WT) REG␤. REG 3. Model of proteasome activation by 11 S regulator. A, the 11 S regulator consists of REG␣ and REG␤ subunits and is a heptamer. 2 The activation region of REG␣ forms a loop at the presumed proteasome binding surface, as revealed by the crystal structure of REG␣ (32). Thus, we hypothesize that this region in REG␣ and the corresponding region in REG␤ form loops in the REG␣/REG␤ hetero-oligomers and are located at the proteasome binding surface. Each of these loops in either wild-type REG␣ or REG␤ is represented by a solid line. B, when the REG␣ subunit is mutated outside its activation loop (dashed ovals), the resulting hetero-oligomer is as active as wild-type REG␣/REG␤ heterooligomer (Table I). Note that we also studied two single-site mutations outside the activation loop of REG␤ subunits, and they did not affect proteasome activation by REG␣/REG␤ hetero-oligomers either (Table  II). C, when the activation loops of REG␣ subunits are altered (dashed line), the resulting mtREG␣/REG␤ hetero-oligomers retain 30 -80% of the activity of wild-type REG␣/REG␤ (Table I). D, when the activation loops of REG␤ are altered (dashed line), REG␣/mtREG␤ is ϳ20 -50% active (Table II). E, when the activation loops of both REG␣ and REG␤ are altered, the resulting REG␣/REG␤ hetero-oligomer does not activate the proteasome (Table II and Ref. 35).

Formation of mtREG␣/REG␤ Hetero-oligomers Results in
Proteasome Activation by REG␣ Mutants-We have shown by three different approaches that most monomeric REG␣ mutants are capable of forming hetero-oligomers with REG␤. When mixed with REG␤, 35 of 36 REG␣ mutants activated the proteasome, indicating that each of the 35 mutants associates with REG␤ and presumably forms hetero-oligomers. The only REG␣ variant that did not exhibit proteasome activation higher than REG␤ alone is REG␣(R181C), a variant that also did not bind REG␤ by filter or gel filtration assays (Table I and Fig. 1). Because the other REG␣ variants bound REG␤ on filters or formed hetero-oligomers with REG␤ as demonstrated by gel filtration chromatography, we conclude that virtually all of the REG␣ variants isolated in our screen are capable of forming hetero-oligomers with REG␤. By formation of REG␣/ REG␤ hetero-oligomers, the REG␣ variants become highly active. We attribute the fact that the monomeric REG␣ mutants cannot activate the proteasome by themselves to their failure to form heptamers, even in the presence of proteasomes.
Because the REG␣(D205V)/REG␤ mixture activates the proteasome (Table I), the two subunits probably form hetero-oligomers. However, any REG␣(D205V)/REG␤ hetero-oligomers that formed did not survive elution from a gel filtration column (Table I). This may reflect different buffer conditions used for the activity assay and for the sizing column, or it may result from dilution during gel filtration. Nonetheless, the fact that virtually all of the monomeric REG␣ mutants formed active hetero-oligomers with REG␤ indicates that REG␣/REG␤ hetero-oligomers are more stable than REG␣ heptamers. Singlesite mutations in REG␣ that compromise formation of homoheptamers are generally not sufficient to prevent formation of REG␣/REG␤ hetero-oligomers. In this regard, we have observed that REG␣/REG␤ hetero-oligomers at 100 g/ml are stable when eluted with buffers containing 200 mM KCl from a Superdex 200 gel filtration column, whereas under these conditions, 50% of REG␣ heptamers dissociate to monomers.
Both REG␣ and REG␤ Subunits Contribute to Proteasome Activation by Hetero-oligomers-In the crystal structure of the REG␣ heptamer, amino acids Arg 141 -Gly 149 form a loop in each REG␣ subunit (32). This region is critical for proteasome activation (35). Therefore, we call this region and the corresponding region in REG␤ (Lys 131 -Gly 139 ) activation loops. In Table I, we highlighted residues in the activation loop of REG␣ and Pro 240 , a residue that contacts the REG␣ activation loop (32,35). Proteasome activation by REG␣/REG␤ hetero-oligomers formed from REG␣ variants bearing mutations in the activation loop ranged from 30 to 80% of that seen with REG␣/REG␤ hetero-oligomers formed from wild-type REG␣ (Table I), indi-

FIG. 5. Properties of REG␣/REG␤ hetero-oligomers formed from REG␣(N50Y) and REG␤ or REG␤ C-terminal mutants. A,
proteasome activation by REG␣/REG␤ hetero-oligomers. REG␣(N50Y), a monomeric REG␣ mutant, was incubated with an equal amount of REG␤ or each of the REG␤ mutants. Each of the resulting heterooligomers was purified using a Superdex 200 column and used to perform proteasome activation assays as described in Fig. 4B. B, proteasome binding by REG␣/REG␤ hetero-oligomers. The purified heterooligomers were used to performed proteasome binding assay as described under "Experimental Procedures." Numbers at the left are hetero-oligomer concentrations containing wild-type REG␤ and REG␣(N50Y), whereas numbers at the right are hetero-oligomer concentrations containing REG␤ C-terminal mutants and REG␣(N50Y). FIG. 6. Properties of REG␣/REG␤ hetero-oligomers formed from REG␣(P240A) and REG␤ or REG␤ mutants. A, proteasome activation by the REG␣/REG␤ hetero-oligomers. Each of these hetero-oligomers was purified and used to perform the activity assays as described in Fig. 4B. B, proteasome binding of the hetero-oligomers. The experiment was performed as described under "Experimental Procedures." The heterooligomer concentrations used to incubate with the tethered proteasome are indicated at the right. Control, defined as described in Fig. 2 cating that in hetero-oligomers the activation loop of REG␣ clearly affects proteasome activation. Song et al. (37) and Kuehn and Dahlmann (36) report that REG␤ is inactive, but it will generate a superactive complex when mixed with REG␣. Based on these results, Song et al. (37) proposed that REG␤ functions only to modulate the activity of REG␣. However, the results presented here and several other considerations indicate that the activation loop of REG␤ also contributes to proteasome activation. First, the REG␣ and REG␤ activation loops are virtually identical in sequence (35), suggesting that these loops have similar functions. Mutational studies support this hypothesis, because converting a critical Asn to Tyr in the activation loops of both REG␣ and REG␤ produces proteins that bind the proteasome but do not activate it (35). Second, three studies by our laboratory have shown that REG␤, by itself, is able to bind and activate the proteasome (34,35,38). Consequently, it seems unlikely that the REG␤ activation loop would not function in hetero-oligomers. Last, mutations in the activation loop of REG␤, like those in the activation loop of REG␣, reduce proteasome activation by REG␣/REG␤ heterooligomers, whereas mutations outside the activation loop of REG␤ have little effect on proteasome activation by REG␣/ REG␤ hetero-oligomers (Tables I and II and Fig. 2). All of these findings indicate that the REG␤ activation loop also contributes to proteasome activation by the 11 S REG.
When the activation loop of either REG␣ or REG␤ was mutated, the resulting REG␣/REG␤ hetero-oligomer was partially active (Tables I and II and Fig. 3). By contrast, when the activation loops of both ␣ and ␤ subunits were altered by mutation, the resulting hetero-oligomer was completely inactive (Table II and Fig. 3). Therefore, we propose that the activation loops of both REG␣ and REG␤ contribute to proteasome activation. This predicts that the activation regions of REG␣ and REG␤ subunits are located on the same surface and adopt similar conformations in the REG␣/REG␤ hetero-oligomers. Because REG␣ and REG␤ share ϳ50% similarity in sequence, it is reasonable to assume that REG␣ and REG␤ subunits have similar folds. This inference is supported by mutational studies on both REG␣ and REG␤. For instance, deletion of the last 14 amino acids from either REG␣ or REG␤ results in REG␣/REG␤ hetero-oligomers unable to bind the proteasome, suggesting that both REG␣ and REG␤ C-terminal regions are located on the proteasome binding surface (Table III and Fig. 5). Moreover, deletion of the REG␣ homolog-specific insert (28 amino acids), amino acid sequences unique to each REG homolog, has no effect on the proteasome binding or activation by REG␣/ REG␤ hetero-oligomers, whereas deletion of the REG␤ insert (15 amino acids) only causes a modest impairment in proteasome binding or activation (38). Thus, it appears that the REG␣ and REG␤ inserts are located on the "upper" surface of the REG␣/REG␤ hetero-oligomer, whereas the activation loops are on the "lower" surface, as defined by the crystal structure of the REG␣ heptamer (32). In view of all these considerations, we conclude that both the REG␣ and REG␤ activation loops contribute to proteasome activation.
Attempts to quantify the relative contribution to proteasome activation by REG␣ and REG␤ subunits were complicated by the finding that specific mutations in the activation loop of either REG␣ or REG␤ affected proteasome activation differently ( Fig. 2 and Table II). This could be attributable to subtle, but differing, conformational changes induced by the specific mutations. The average activation by hetero-oligomers with altered REG␣ activation loops is ϳ60% of that of wild-type hetero-oligomers, whereas the reciprocal hetero-oligomers with altered REG␤ activation loops are ϳ40% active (Table II). These averages are close to the relative contributions expected on the assumption that REG␣/REG␤ hetero-oligomers are heptamers composed of three REG␣ plus four REG␤ subunits, and that each subunit contributes equally to proteasome activation. These "averages" are interesting because mass spectrometry shows that hetero-oligomers formed from recombinant REG␣ (N50Y) and REG␤ subunits are heptamers containing three ␣ subunits and four ␤ subunits. 2 However, we do not know whether hetero-oligomers formed from other REG␣ or REG␤ variants indicated in the Tables I and II are a mixture of hetero-oligomers with different ␤/␣ ratios. Nor are we certain that different pairs of REG␣/REG␤ hetero-oligomers have the same proteasome binding affinities. Because of these uncertainties, the exact quantitative contributions by ␣ and ␤ subunits to proteasome activation remain undetermined. Nonetheless, our results provide good evidence that in REG␣/REG␤ hetero-oligomers REG␤ contributes similarly to proteasome activation as does REG␣.
The C Termini of REG␣ and REG␤ Are Necessary for Proteasome Binding-To study the function of REG␣ residues near the C terminus, we characterized four REG␣ mutants isolated from the random mutagenesis screen. REG␣(Y249C), REG␣(M247V), and REG␣ϩ5 formed fully active REG␣/REG␤ FIG. 7. Exchanging REG␣ C-terminal 8 amino acids with those of REG␥ increases proteasome binding by the REG␣ variant. A, diagram shows REG␣ and REG␣241␥8. The last 8 amino acids of REG␣ and REG␣241␥8 are shown in one-letter code. The REG␣⅐REG␥ chimera, REG␣241␥8, was produced using site-directed mutagenesis to exchange the REG␣ C-terminal 8 amino acids with those from REG␥ as described under "Experimental Procedures." The remaining sequences of REG␣ and REG␣241␥8, represented by the rectangles, are identical. B, proteasome binding by REG␣ and REG␣241␥8. The proteasome binding assay was performed as described under "Experimental Procedures," and the concentrations of REG␣ or REG␣241␥8 used to incubate with the tethered proteasome are indicated to the left. Control, signals obtained when 30 g/ml REG␣ or REG␣241␥8 were incubated with wells coated with antibodies, but in the absence of proteasomes. C, REG␣ antiserum binds REG␣ and REG␣241␥8 with similar affinities. REG␣ or REG␣241␥8 proteins with a 2-fold dilution were dot-blotted on a nitrocellulose membrane and probed with anti-REG␣ serum. The numbers at the right are the amounts of REG␣ or REG␣241␥8 on the membrane. hetero-oligomers with REG␤, indicating that these three mutations in the C-terminal region of REG␣ do not severely alter the ability of hetero-oligomers to bind or activate the proteasome. Gel filtration analysis showed that REG␣⌬14 formed hetero-oligomers with REG␤, but these hetero-oligomers were unable to bind the proteasome (Table I). Thus, some portion of the last 14 amino acids of REG␣ is necessary for association of 11 S REG with the proteasome.
Song et al. (37) reported that conversion of Tyr 249 of rat REG␣ to Ser results in partial loss of its activity, although this mutant forms a fully active hetero-oligomer with rat REG␤. This finding agrees with results presented here. Song et al. (37) also reported that deletion of REG␣ Tyr 249 yields completely inactive REG␣/REG␤ hetero-oligomers, and they concluded that this amino acid is important for proteasome activation. However, they did not distinguish whether deletion of Tyr 249 affects proteasome activation per se or binding by REG␣/REG␤ hetero-oligomers to the proteasome. In their previous studies, DeMartino and colleagues showed that treatment of PA28 (11 S REG) with a carboxyl peptidase destroys its ability to bind the proteasome, and they concluded that the REG␣ C terminus is important for proteasome binding (40). This is in full agreement with our finding that deletion of 14 amino acids from the REG␣ C terminus results in total loss of proteasome binding by REG␣/REG␤ hetero-oligomers.
Deletion of 1, 2, 9, or 14 amino acids from the REG␤ C terminus produced inactive REG␤ unable to bind or activate the proteasome (Fig. 4). Moreover, deletion of even one or two amino acids from the REG␤ C terminus significantly reduced proteasome binding by REG␣/REG␤ hetero-oligomers (Fig. 5B) but did not impair proteasome activation to such a great extent (Fig. 5A). Song et al. (37) reported that deletion of one amino acid from the C-terminal REG␣ totally inactivated REG␣/ REG␤ hetero-oligomers. Because deletion of one or two amino acids from the REG␤ C terminus only partially reduced proteasome binding and activation by REG␣/REG␤ hetero-oligomers, it would appear that the REG␣ C terminus contributes more to proteasome binding than does the REG␤ C terminus. Alternatively, it is possible that the partially active REG␣-(N50Y)/REG␤⌬1 and REG␣(N50Y)/REG␤⌬2 hetero-oligomers observed by us would be inactive under the assay conditions used by Song et al. (37), because we assay proteasome activation using 10-fold higher REG concentrations.
Although our experiments cannot rule out the possibility that the REG␣ and REG␤ C termini contribute directly to proteasome activation, three observations suggest that REG activation loops and C termini perform different functions. First, three REG␣ C-terminal mutants, REG␣(Y249C), REG␣-(M247V), and REG␣ϩ5, formed fully active hetero-oligomers with REG␤, whereas each of the five REG␣ variants with changes in the REG␣ activation loop only formed partially active hetero-oligomers (Tables I and III). Second, REG␤ subunits lacking either the last one or two amino acids formed partially active hetero-oligomers with REG␣(P240A), and these hetero-oligomers bound the proteasome more weakly than hetero-oligomers formed from REG␣(P240A) and wild-type REG␤. By contrast, REG␤(N135Y) formed totally inactive hetero-oligomers with REG␣(P240A), and these hetero-oligomers bound the proteasome as tightly as REG␣(P240A)/REG␤ (Fig. 6). Third, a REG␣ chimera bearing the last 8 amino acids of REG␥ bound the proteasome more tightly than did wild type REG␣ (Fig. 7). Nonetheless, the peptide activation pattern of the chimera was identical to that of REG␣ but distinct from that of REG␥. This indicates that the C-terminal region of REG␥ contributes mostly to proteasome binding but does not influence the proteasome activation pattern.
In summary, we have demonstrated that the activation loops of both REG␣ and REG␤ subunits contribute to proteasome activation by the 11 S REG. Attempts to determine the relative contribution by REG␣ and REG␤ subunits to proteasome activation are complicated by the fact that specific mutations in the activation loops of either REG␣ or REG␤ subunits affect proteasome activation by REG␣/REG␤ hetero-oligomers differently. Still, our results clearly show that REG␤ contributes to proteasome activation in a manner similar to REG␣. Furthermore, we demonstrate that residues near the C termini of both REG␣ and REG␤ are necessary for the association of REG␣/REG␤ heterooligomers with the proteasome. Whether C-terminal sequences in REG␣ and REG␤ contribute to proteasome activation per se is still an open question, but the evidence presented in this work suggests that the answer is no. Future structural studies on the 11 S REG-proteasome complex should shed light on this matter.