A 220-kDa Activator Complex of the 26 S Proteasome in Insects and Humans

The S10b (SUG2) ATPase cDNA has been cloned by reverse transcription-polymerase chain reaction/rapid amplification of cDNA ends from mRNA of intersegmental muscles of the tobacco horn moth (Manduca sexta). The S10b ATPase is a component of the 26 S proteasome, and its concentration and that of its mRNA increase dramatically during development in a manner similar to other ATPases of the 19 S regulator of the 26 S proteasome. The S10b and S6′ (TBP1) ATPases are also present in a complex of ∼220 kDa in intersegmental muscles. The 220-kDa complex markedly activates (2–10-fold) the 26 S proteasome, even when bound to anti-S10b antibodies immobilized on Sepharose, and increases in concentration ∼5-fold like the 26 S proteasome in the intersegmental muscles in preparation for the programmed death of the muscle cells. A similar activator complex is present in human brain and placenta. Free activator complexes cross-activate: the Manduca complex activates rat skeletal muscle 26 S proteasomes, and the placental complex activates Manduca 26 S proteasomes. The placental activator complex contains S10b and S6′, but not p27. This 220-kDa activator complex has been evolutionarily conserved between species from insect to man and may have a fundamental role in proteasome regulation.

Type II programmed neuromuscular cell death is a feature of some abdominal motor neurons and intersegmental muscles (ISM) 1 at eclosion in the tobacco horn moth (Manduca sexta) (2)(3)(4). After emergence, these cells die within 24 -36 h in response to changes in circulating levels of ecdysteroid hormone. Previous studies have shown that eclosion is preceded in ISM by a massive increase in polyubiquitin gene expression (5) and a large increase in ubiquitinylated proteins (6) and in the levels of proteasomes (7,8). In the muscles of larvae at stage 0, before the hormone-dependent changes in gene expression activate the death process, proteasomes appear to be depleted in some regulatory ATPases (e.g. S4 and S7). 2 Marked increases in several ATPases (S4, S6, and S7) then occur, so that by stage 8, just before cell death, the 26 S proteasomes prepared from muscle contain at least four of the regulatory ATPases (S4, S6, S6Ј, and S7). The simplest explanation is that proteasomes are increased in number and equipped with the ATPases needed to degrade the accumulating multi-ubiquitinylated proteins during the programmed elimination of the muscles (8). The first observations that the ubiquitin/26 S proteasome system is involved in programmed cell death in M. sexta (5)(6)(7)(8) have been followed by other data that implicate proteasomes in the decisions that favor either cell death or survival (9 -16).
There are six ATPase molecules found in the 19 S regulator of the 26 S proteasome (1). As an extension of studies on these ATPases in programmed cell death in ISM during eclosion in M. sexta (8), we have cloned a second S10b (SUG2) ATPase and shown that this ATPase is not only associated with 26 S proteasomes, but is also found in a much smaller complex, of ϳ220 kDa, with the S6Ј (TBP1) ATPase. The concentration of the 220-kDa complex increases in ISM during programmed cell death at the same time as that of the 26 S proteasome, suggesting a role for the 220-kDa complex in muscle cell death. A modulator complex containing the S10b and S6Ј ATPases and a p27 protein has been previously described (17). Complexes similar to the Manduca 220-kDa complex are present in human brain and placenta and contain the S10b and S6Ј ATPases, but not p27. The 220-kDa complexes can activate preparations of 26 S proteasomes across species barriers, indicating the evolutionary conservation of these proteasomal activators. Their potential roles in the proteolytic mechanisms of the 26 S proteasome and other possible functions of the S10b, S6Ј, and other ATPases in the cell are discussed.

EXPERIMENTAL PROCEDURES
Insect Culture and Staging-The insects were raised, the stages of pre-ecdysial development recognized, and muscles collected as described previously (8).
Reverse Transcription-PCR Amplification of M. sexta S10b (SUG2) cDNA-Degenerate primers matching the ATPase boxes A and B (18) were used to clone several different ATPase sequences from Manduca genomic DNA. The two primers TAYGGNCCNCCNGGNACNGGNA (corresponding to the protein sequence YGPPGTG) and GGNGCRT-CRATYTCRTCRATRAA (antisense strand for the protein sequence FI-DEIDA) were used at a concentration of 0.5 M in a 50-l PCR containing 50 M dNTPs, 200 ng of genomic DNA, and 2.5 units of Taq DNA polymerase (Roche Molecular Biochemicals, Lewes, United Kingdom). A 200-base pair product was cloned into the EcoRV site of the pSK Ϫ plasmid (Stratagene, Cambridge, UK) and manually sequenced by the dideoxynucleotide method. This sequence information was used to design oligodeoxynucleotides for subsequent reverse transcription-PCR.
PCR/3Ј-RACE was used to obtain the 3Ј-end of a clone that showed good homology to other proteasomal ATPases, but appeared to represent a distinct gene ("AAA (ATPases associated with diverse cellular activities) clone"). RNA (1 g) obtained from stage 7 Manduca ISM (19) was reverse-transcribed (37°C, 2 h) in the following 20-l reaction: 100 units of SuperScript II reverse transcriptase (Life Technologies, Inc., Paisley, UK), 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, 0.125 mM dNTPs, 20 units of RNase inhibitor (Amersham Pharmacia Biotech, St. Albans, UK), and 0.5 g of T 17 adapter primer (GACTCGAGTCGACATCGATTTTTTTTTTTTTTTTT). Next, 2 l of the reaction was used in a 50-l PCR as described above, but with 25 pmol of adapter primer (GACTCGAGTCGACATCGA) and 25 pmol of AAA clone-specific primer (AACTACGCGCGCGACCACCAG) (annealing temperature of 60°C, 35 cycles). A 650-base pair product was cloned into pSK Ϫ as described above and sequenced.
PCR/5Ј-RACE was used to obtain the remaining sequence of the AAA clone transcript. Poly(A) ϩ RNA was isolated from 50 -100 g of total RNA using the PolyATtract kit (Promega, Southampton, UK). Reverse transcription was performed as described above using 20 ng of poly(A) ϩ RNA and 0.5 g of an oligo(dT) primer. Excess primers were removed with an S-400 microspin column (Amersham Pharmacia Biotech), and after concentration, the 3Ј-end of the cDNA was tailed (37°C, 20 min) using 20 units of terminal deoxynucleotidyltransferase (Life Technologies, Inc.) and 100 pmol of dGTP. The tailing reaction was halted by addition of EDTA and NaCl. The tailed cDNA was pelleted using cetyltrimethylammonium bromide, washed with 70% ethanol, resuspended in 10 l of 1 M NaCl, and then repelleted by centrifugation following addition of ethanol and 0.25 g of glycogen. The pellet was resuspended in 20 l of 10 mM Tris-HCl and 1 mM EDTA (pH 8.0), and PCR was performed as with 3Ј-RACE, but with the addition of 25 pmol of C 17 adapter primer (GACTCGAGTCGACATCGACCCCCCCCCCCC-CCCCC) and another AAA clone-specific primer (CAGACTATTATTA-TTATTACAATGC). Following the initial reaction, a second round of PCR was performed with 2 l of the initial reaction product, the adapter primer, and a second AAA clone-specific primer (AAATCTATGAAGT-AATTACAGAT). A 1350-base pair PCR product was ligated into the EcoRV site of the pSK Ϫ plasmid and sequenced.
Finally, the full-length coding sequence of the Manduca homologue of S10b was amplified (annealing temperature of 55°C, 35 cycles) with 2.5 units of the proofreading Pfu DNA polymerase (Stratagene), 500 M dNTPs, and 25 pmol of primers ATGCCTGCCGGACCTTCC and AAATCTATGAAGTAATTACAGAT using reverse-transcribed RNA obtained from stage 7 ISM; a 1200-base pair product was cloned into the EcoRV site of pSK Ϫ and named pSK.S10b.cds.
Preparation of Tissue Extracts-Frozen (Ϫ70°C) ISM (500 -1000 mg) taken from different developmental stages of M. sexta was homogenized in 4 volumes of homogenization buffer (20 mM Tris-HCl, 2 mM ATP, 10 mM MgCl 2 , 1 mM dithiothreitol, and 10% glycerol (pH 7.5)) using a Polytron homogenizer. The homogenate was centrifuged twice at 15,000 ϫ g for 10 min at 4°C, and the supernatant was collected (S2, soluble muscle extract). Human placental extract was prepared as described for ISM. Homogenized human brain was similarly processed except, that one centrifugation at 8000 ϫ g for 20 min was carried out. Supernatants (S2) were taken for gradient fractionation. Fraction II was prepared from human erythrocytes essentially as described (17,20). Proteins bound to DE52 were eluted with buffer containing 0.25 M NaCl and then 0.5 M NaCl, pooled, and used directly for glycerol gradient centrifugation.
Gel Filtration Chromatography-ISM extract (from ϳ200 g of tissue) was filtered through a 0.2-m filter. Soluble proteins (up to 200 l) were applied to a Superose 12 column (Amersham Pharmacia Biotech) and eluted (flow rate of 400 l/min) using homogenization buffer. Thirty-six fractions (200 l) were collected (from 6.8-to 14-ml elution volume).
Antibody Production and Characterization-The Manduca S10b coding sequence was cut from pSK.S10b.cds, using EcoRI and HindIII restriction enzymes, and ligated into pRSET.C (Invitrogen, San Diego, CA). His-tagged S10b fusion protein was prepared by transforming the plasmid construct into Escherichia coli strain BL21(DE3) (Invitrogen) and inducing expression with isopropyl-␤-D-thiogalactopyranoside (0.1 mM) for 3 h at 37°C. Harvested cells were resuspended in 0.2 volumes of 6 M guanidine hydrochloride, 20 mM sodium phosphate, and 500 mM NaCl (pH 7.8) and sonicated. After removal of cell debris by centrifugation, the His-tagged ATPase was purified by chromatography on nickel-charged chelating Sepharose Fast Flow (Amersham Pharmacia Biotech) at room temperature (8). Eluted protein was dialyzed against 10 mM Tris-HCl (pH 8.0) and 0.1% Triton X-100 overnight at 4°C, freeze-dried, checked for purity by SDS-polyacrylamide gel electro-phoresis, and then used for immunization of New Zealand White rabbits. Antiserum to S10b ATPase was tested against the expressed fusion protein by Western analysis of whole cell extract from isopropyl-␤-D-thiogalactopyranoside-treated E. coli BL21(DE3) cells transformed with pSK.S10b.cds.
The anti-S6 antibody was raised against the recombinant Manduca protein expressed from pSMS73c (8). Anti-S6Ј and anti-S10b antibodies was raised against His-tagged human S6Ј and S10b expressed from pET recombinant plasmids (Novagen, Abingdon, UK) as described above. Antisera to Manduca and human S10b and human S6Ј were affinity-purified by binding to purified His-tagged recombinant antigen protein, coupling to CNBr-activated Sepharose (Sigma, Poole, UK), and eluting with 1 M glycine (pH 2.5). Antiserum to a control (irrelevant) peptide (AG93, a fragment of pro-islet amyloid precursor protein) was similarly affinity-purified on immobilized AG93 peptide.
Monoclonal antibodies to human S10b, S6Ј, S7, and S8 ATPases and the 20 S subunits HC2/MCP20 and Z/MCP205 were kindly provided by Dr. Klavs Hendil (August Krogh Institute, University of Copenhagen). Polyclonal antiserum to human recombinant p27 was from Prof. Keiji Tanaka (Tokyo Metropolitan Institute of Medical Science, Bunkyo-ku, Tokyo, Japan).
Western Analyses-Western analyses were performed as described (8).
Immobilization of Proteasomal Activators on Anti-S10b Immunoaffinity Matrix-Affinity-purified Manduca anti-S10b and human anti-S10b (10 ml) and anti-AG93 (equivalent to 2.5 ml of antiserum) antibodies were coupled to CNBr-activated Sepharose 4B (1 ml). Activator complexes from glycerol gradient fractions 3-5 were bound to the mat-FIG. 2. Glycerol gradient analysis of 26 S proteasome catalytic activities and S10b ATPase content at developmental stages 0 and 7. Soluble protein (5 mg) from stage 0 (A) and stage 7 (B) ISM was fractionated by centrifugation through a 10 -40% glycerol gradient. Protein and chymotrypsin activity were determined and immunoassay of M. sexta S10b ATPase in the fractions was carried out as described under "Experimental Procedures." Western analysis results shown in the inset are from 2-min exposures of the transfers. The stage 7 Western analysis in B was exposed for 30 min (C). Ⅺ, protein concentration (mg/ml); छ, chymotrypsin activity (arbitrary units).

FIG. 3.
Western analysis of the S10b, S6, and S6 ATPases after gel filtration of soluble ISM proteins on Superose 12. Soluble proteins (1.7 mg) from stage 0 and 7 ISM were size-fractionated on Superose 12, and 200-l fractions were collected. Proteins from equal volumes of odd-numbered fractions from stage 0 (A and E, 40 l) and stage 7 (B and F, 8 l (S6 and S10b); and D, 24 l, (S6Ј)) were subjected to SDS-polyacrylamide gel electrophoresis and Western analysis. Transfers were immunoprobed with anti-S10b (A and B, primary antibody at 1:500 dilution and secondary antibody at 1:4000 dilution), anti-S6Ј (D, affinity-purified primary antibody at 1:5 dilution and secondary antibody at 1:10,000 dilution), and anti-S6 (E and F, primary antibody at 1:500 dilution and secondary antibody at 1:4000 dilution) antibodies. Bands in the first three fractions represent 26 S proteasomes. Exposures of between 15 and 30 s are presented for the S10b (A and B) and S6 (E and F) blots. C and G are longer exposures (2 min) of transfers shown in B and F, respectively. The S6Ј transfer (D) was exposed to film for 30 min. The elution positions of molecular mass standards (aldolase, 230 kDa; catalase, 160 kDa; and ovalbumin, 44 kDa) are indicated by arrowheads. The approximate molecular masses of the immunoreactive species detected on the blot are indicated. Small amounts of free ATPases are evident together with a truncated form of S10b.
rices by inversion for 1-2 h at room temperature. After repeated washing with homogenization buffer, activator complexes were eluted from the immunoaffinity matrices for Western analyses with 1 M glycine (pH 2.5), and residual bound material was released by boiling in urea/SDS loading buffer (21).

RESULTS
Reverse Transcription-PCR/RACE of the S10b cDNA and Sequence Comparisons-As part of the characterization of all proteasomal regulatory ATPases from Manduca, multiple DNA sequences encoding the region between the two elements of the Walker ATP-binding motifs were amplified and cloned. Sequencing of individual clones revealed five different sequences, some of which bore very close homology to previously identified proteasomal ATPases. One clone that bore less homology to known proteasomal ATPases was chosen for further study, and full-length cDNA sequences were obtained using PCR/RACE. The DNA and predicted protein sequences of the ATPase shown in Fig. 1 are closely homologous to the 19 S regulatory S10b (SUG2) ATPase (1). 3 As is the case for many other members of this subgroup of the AAA superfamily, the S10b ATPase has a putative coiled-coil motif, the Walker A and B nucleotidebinding motifs, and putative DNA helicase motifs. The S10b ATPase mRNA and protein increase ϳ5-fold in concentration in the muscles in preparation for cell death (data not shown). Again, these changes are similar to those observed for Manduca S4, S6, and S7 ATPases during the same period (8). The S10b ATPase is a component of the 26 S proteasome as shown by glycerol gradient centrifugation (Fig. 2); this analysis also demonstrates that the increase in the tissue content of the S10b ATPase is mainly associated with the 26 S proteasome.
The S10b ATPase Is Present in the Cell Independent of the 26 S Proteasome-Overexposure of the enhanced chemiluminescence signals in the Western analysis of the glycerol gradient fractionation of stage 7 ISM (Fig. 2B) shows that the S10b ATPase is not only present in the 26 S proteasome, but is also present in fractions (3-6) containing lower molecular mass complexes (Fig. 2C). The nature of these complexes was further investigated by gel filtration chromatography on Superose 12. Western analysis of the fractions after gel filtration shows that both the S10b and S6Ј ATPases elute as if present in complexes of ϳ220 kDa (Fig. 3, A, B, and D). The relatively weak immunoreactivity of S6Ј is probably due to the fact that the affinitypurified anti-S6Ј antibody was raised against the human S6Ј protein. The S6 ATPase is not enriched in complexes of this size, with the antigen detected in this region probably representing the trailing edge of proteasomal antigen spread by diffusion during chromatography (Fig. 3, E and F). The 220-kDa complex is found in muscles at the beginning (stage 0) and toward the end (stage 7) of the period of preparation for muscle cell death. The volumes of the stage 7 fractions analyzed by Western blotting were five times less than those of stage 0 since the total ATPase content of ISM increases ϳ5-fold during pupal development. The similar intensities of the S10b bands in the 220-kDa complex from stages 0 and 7 (Figs. 2 and 3) therefore demonstrate that, like the proteasomal ATPases (8), the concentration of the complexes containing the S10b ATPase increases ϳ5-fold in the muscle cells during preparation for cell death. The amounts of the ATPases in the 220-kDa complex do not change when soluble extracts are prepared from muscles in the absence of ATP, indicating that the complex containing the S10b ATPase is unlikely to be produced by dissociation of 26 S proteasomes on homogenization (data not shown). The fact that the S6 ATPase is not in the 220-kDa complex (Fig. 3, E and F [3][4][5] or Superose 12 fractionation (material eluting in the size ranged 150 -250 kDa) of Manduca ISM soluble protein were incubated with 26 S proteasome fractions from a glycerol gradient showing peak peptidase activity (fraction 15) or adjacent fractions 14 and 16, with corresponding fractions from rat skeletal muscle, or with purified 20 S proteasomes from human erythrocytes (see "Experimental Procedures"). In Experiment C, 20 S proteasomes were incubated in the absence of ATP with an ϳ3or 6-fold molar excess of purified human placental 11 S (PA28) activator. Proteasomal activities were in the range of 40 -190 units/assay. In one experiment, apyrase was included in the incubation. Activation represents the chymotrypsin activity measured with activator plus proteasomes (20 S or 26 S) minus the sum of activity measured when activator and proteasomes were assayed separately, expressed as a % of the proteasomal activity. Thus, 100% activation represents a 2-fold activation.

220-kDa Activator Complex of the Proteasome
also supports the notion that the complex is not produced by dissociation of ATPases from the 26 S proteasome. A careful analysis of the relative amounts of the S10b ATPase in the 26 S proteasome compared with the 220-kDa activator indicates there is ϳ3-5% of the S10b ATPase in the activator relative to the 26 S proteasome (data not shown).
Fractions Containing the 220-kDa Complex Are Able to Activate the 26 S Proteasome-Glycerol gradient fractions containing the 220-kDa complex from stage 7 (fractions 1-8) were incubated with similarly prepared 26 S proteasomes to investigate whether the 220-kDa complex was able to directly influence the peptidase activity of the proteasome (Fig. 4). Fractions enriched in the 220-kDa complex (fractions 3-6) cause considerable activation of the chymotrypsin activity of the 26 S proteasome in a concentration-dependent manner. The Manduca activator was further characterized to determine whether it is related to the modulator complex isolated from human erythrocytes, which stimulates assembly of 26 S proteasomes from 20 S cores and 19 S regulators (17,22,23). These experiments are summarized in Table I.
The capacity of the Manduca activator complex to increase the chymotrypsin-like activity of fractions enriched in the 26 S proteasome varied between different preparations, from Ͼ10fold (Fig. 4) to ϳ2-fold (Table I). This presumably reflects variation both in the activator complex itself and in the capacity of the 26 S proteasome to respond in different experiments. Activation capacity does not appear to depend on the proportions of 20 S and singly or doubly capped 26 S proteasomes in the preparation since the glycerol gradient fractions showing peak peptidase activity were activated by the 220-kDa complex to a greater extent than leading or trailing edges of the proteasome peak (Table I, Experiment B). If the activator was simply stimulating the assembly of doubly capped 26 S complexes from 19 S and 20 S components (22,23), we would expect the extent of activation to be greatest on the lighter (trailing) side and least on the heavier (leading) side of the peak. The Manduca activator increases 26 S proteasome activity only in the presence of ATP (Table I, Experiment D). Indeed, in contrast to the purified human 11 S (PA28) activator (Table I, Experiment C), the Manduca 220-kDa activator did not activate purified 20 S proteasomes (Table I, Experiments A and D), and thus is not the Manduca equivalent of PA28 (24). However, the Manduca 220-kDa activator was able to activate a 26 S proteasome preparation from rat skeletal muscle, demonstrating that it can activate mammalian proteasomes efficiently (Table I, Experiment D).
The data presented so far have shown that activator activity and the S10b and S6Ј ATPases copurify together. To prove that activator activity is dependent on the S10b ATPase, activator preparations were incubated with affinity-purified polyclonal anti-S10b antibody bound to Sepharose beads. All of the Manduca activator bound to the beads, and the immobilized complex could still activate the Manduca 26 S proteasome chymotrypsin activity (Fig. 5). The activator did not bind to unsubstituted Sepharose. Further purification and characterization of the activator from Manduca ISM were not possible due to the small amounts of tissue available. However, since the activator stimulated mammalian proteasomes, human tissues were analyzed to determine whether a similar human 220-kDa complex exists.
Human Brain, Placenta, and Embryonic Kidney Cells Have Complexes Containing the S10b and S6Ј ATPases-Glycerol gradient fractionation of human brain cortex (Fig. 6) reveals a distribution of 26 S proteasomes similar to insect muscle (Fig.  3). In addition, the S10b and S6Ј ATPases, but not the S7 and S8 ATPases, are found in a smaller complex (fractions 3-7) similar to the 220-kDa complex found in insect muscle (Fig. 2). Similar evidence for a smaller complex containing the S10b and S6Ј ATPases, but not the other ATPases, was found in early gradient fractions prepared from other human tissues: normal and neonatal brain and placenta and in extracts from human embryo kidney cells (data not shown). Small complexes containing the S10b and S6Ј ATPases seem to be widely distributed in human cells as well as in insect muscle.
Properties of the Placental Activator-The human placental activator showed an activation capacity similar to that of the Manduca activator in a dose-dependent manner (Table II, part  B). Interestingly, glycerol gradient fractions containing the placental activator increase the activity of Manduca 26 S proteasomes considerably more than placental 26 S proteasomes (Table II, Experiments A and B). An increase in the activity of purified human 20 S proteasomes sometimes occurs (Table II, Experiment B, but not part C) with fractions containing the placental activator due to the presence of the 11 S activator in human tissues. This does not occur with the insect activator since the 11 S activator does not appear to be present in insect muscle (Table I). Thus, ATPase-containing activator complexes can activate 26 S proteasomes across species barriers: Manduca complexes activate rat skeletal muscle (Table I, Experiment D), and placental complexes activate Manduca 26 S proteasomes (Table II, Experiments A and B). The combined observations indicate that the activator has been evolutionarily conserved between species from insect to man.
The p27 Protein Subunit and Activator Complexes from Nonred Blood Cells-A modulator complex can be purified from bovine red blood cells (17) and contains the S10b and S6Ј ATPases together with a novel p27 subunit that is claimed also to be present in 26 S proteasomes (25). As expected, glycerol gradient fractionation of human red blood cell fraction II (Fig.  7) confirmed that S10b, S6Ј, and p27 (but not the S7 ATPase) were incubated with anti-S10b antibody-Sepharose beads (50 l) for 1 h at room temperature. The Sepharose beads were then sedimented, and the supernatant (unbound material) was collected. The beads were washed in 1.5 ml of 10 mM Tris-HCl and 1 mM EDTA (pH 7.5) for 20 min at room temperature. For the activation assay, the beads were resuspended in homogenization buffer (100 l) and incubated with 26 S proteasomes (1 l) at 37°C for 30 min (Bound). Assays were also performed in parallel with pooled activator fractions (150 l) not incubated with Sepharose (Load) and with unbound material (Unbound) (both with and without proteasomes), with 26 S proteasomes alone (1 l), and with 26 S proteasomes (1 l) added to unsubstituted Sepharose beads (50 l; Bound to Sepharose). All assays were performed in duplicate and averaged. Activation is the chymotrypsin activity measured with fractions plus 26 S proteasomes minus the sum of activity measured when fractions and proteasomes were assayed separately, expressed as a percentage of the 26 S proteasome activity.

220-kDa Activator Complex of the Proteasome
were markedly enriched in early gradient fractions, which is consistent with earlier observations (17). However, very little S10b, S6Ј, and p27 were present in the proteasome fractions, indicating that the high salt, ATP-depleted conditions used for isolation of fraction II (17,20) have disrupted the proteasome and possibly the 19 S regulator. In an analogous fashion, proteasomes prepared from placental tissue frozen at Ϫ70°C for several months appear to be disrupted with the vast majority of S10b and S6Ј in small complexes, with S7 in rather larger complexes, and with very little ATPases in particles resembling 19 S regulators or 26 S proteasomes (Fig. 8, compare with Fig.  6). However, the p27 subunit was not detected in early glycerol gradient fractions containing the S10b and S6Ј ATPases whether material was analyzed from frozen ( Fig. 8) or fresh (data not shown) human placenta. Indeed, p27 could not be detected in placental S2 fractions (Fig. 9). More significantly, when placental activator complexes were immobilized by affinity attachment to anti-S10b antibody-Sepharose (Fig. 9), both S10b and S6Ј were enriched, but neither p27 or the S7 ATPase could be detected in bound material. The activator did not bind to anti-AG93 antibody-Sepharose, an irrelevant immunoaffinity control matrix, proving the specificity of the interaction with anti-S10b antibody-Sepharose (data not shown). Therefore, the p27 protein may not be widely distributed in different human tissues and does not form part of the 220-kDa activator demonstrated in this work.

Proteasomes and Type II Programmed Cell Death-Previous
studies on the programmed elimination of abdominal intersegmental muscles in the tobacco horn moth (M. sexta) have shown that the ubiquitin/26 S proteasome system is dramatically ac-tivated to degrade muscle proteins during the death process (5)(6)(7)(8). The combination of a massive increase in polyubiquitin gene expression, accumulation of ubiquitinylated proteins, and a large increase in levels of the 26 S proteasome (8) provides the degradation machinery necessary for the proteolytic destruction of muscle proteins. The developmental changes in 26 S proteasomes occur only in the abdominal intersegmental muscles, which are destined to die, and not in flight muscles, which are needed in the emerging adult moths for locomotion (8,26). Furthermore, at least one ATPase (S6) is expressed only at high levels in other muscles that undergo developmentally programmed cell death: increased expression of the ATPase occurs at distinct times in different muscles to coincide with the death process (27). The increase in 26 S proteasome components is hormone-regulated and suppressed by the ecdysteroid agonist RH-5849 (26,27).
The 220-kDa Activator and Cell Death-In this study, we have cloned the cDNA for another Manduca ATPase (S10b; Fig.  1) and studied the expression pattern and biochemical properties of the protein. The S10b ATPase is a component of the muscle 26 S proteasome and shows developmental increases in expression in the intersegmental muscles (Fig. 2) similar to other ATPases studied previously (8).
The S10b ATPase is found not only in the 26 S proteasome, but also in other molecular forms in the muscle cells. Western analysis of glycerol gradient fractions shows that smaller amounts of the S10b ATPase are found in fractions containing lower molecular mass species than the 26 S proteasome (Fig.  2). Size fractionation of proteins in soluble muscle extracts on Superose 12 (Fig. 3) shows that both the S10b (Fig. 3, A and B) and S6Ј (Fig. 3D) ATPases are in a protein complex of ϳ220 FIG. 6. Distribution of 26 S proteasome subunits after glycerol gradient fractionation of human brain cortex. Glycerol gradient fractions (30 l) from human Alzheimer's brain cortex were analyzed by SDS-polyacrylamide gel electrophoresis and Western analysis. Transfers were incubated with hybridoma culture supernatants containing monoclonal antibodies to subunits of the 19 S complex at the following dilutions: 1:10, S6Ј ATPase; 1:20, S10b ATPase; 1:5, S8 ATPase; 1:10, S7 ATPase; and 1:5000 and 1:500, monoclonal antibodies MCP20 and MCP205 raised the 20 S ␣-subunit HC2 and 20 S ␤-subunit Z, respectively. Proteasome chymotrypsin activity was concentrated in fractions 9 -17. kDa. The S6 ATPase is not enriched in this complex (Fig. 3, E  and F). The increase in the concentration of the 220-kDa complex in the muscles from the beginning (stage 0) to the end (stage 7) of preparation for muscle death is similar to that for the ATPases of the 26 S proteasome (Figs. 2 and 3) (8). The 220-kDa complex is able to activate Manduca 26 S proteasomes 2-10-fold in an S10b-dependent manner (Figs. 4 and 5 and Table I).
A modulator complex containing both the S10b and S6Ј ATPases plus the p27 protein has been purified from bovine red cells and characterized as an activator of the 26 S proteasome (17). The p27 subunit is also a component of the red cell 26 S proteasome (25). Insects do not have red blood cells since oxygen is supplied by diffusion from the spiracles via the hemolymph, so contamination of ISM extract with a red cell-derived modulator is not possible. The 220-kDa complex from intersegmental muscles is therefore the first 26 S proteasome activator described other than that in red blood cells.
The bovine 300-kDa "modulator" has recently been shown to promote the assembly of 26 S proteasomes from 20 S particles and the PA700 (19 S) regulator: the modulator increases the number of singly and doubly capped complexes (22), although activation by the modulator requires addition of only a single PA700 cap (23). Significantly, the increase in the concentration of the 220-kDa complex during development parallels the increase in the 26 S proteasome content of muscles (Figs. 2 and 3): the change in the 220-kDa complex in Manduca intersegmental muscles may facilitate the assembly of 26 S proteasomes in preparation for muscle cell death. However, ϳ3-5% of the S10b ATPase present in the 26 S proteasome (data not FIG. 7. Identification of the p27 protein subunit in glycerol gradient fractions of Fraction II from human red cells. Human red blood cell lysate Fraction II was fractionated by glycerol gradient centrifugation. छ, chymotrypsin activity (arbitrary units); Ⅺ, protein concentration (mg/ml). Pairs of fractions were pooled and subjected to Western analysis with different antibodies as indicated (see legend to Fig. 6 for details). FII (last lanes) shows a sample of total Fraction II protein, and S10b and p27 (first lanes) show samples of His-tagged recombinant protein run as standards.

TABLE II
Properties of the human placental activator Activator fractions from glycerol gradients of human placental soluble protein were incubated with human placental or Manduca ISM 26 S proteasome fractions from a glycerol gradient showing peak peptidase activity or with purified 20 S proteasomes from human erythrocytes (see "Experimental Procedures"). Activation is defined as in the legend to Table I shown) is found in the 220-kDa complex, and ϳ90 -95% of proteasomes in stage 7 ISM extracts are in 26 S complexes with correspondingly few 20 S particles (8). The data indicate that a complex containing a small fraction of total cellular S10b and S6Ј ATPases can substantially activate (2-10-fold) preparations of Manduca 26 S proteasomes (Figs. 4 and 5 and Table I). Furthermore, the activator does not stimulate smaller (20 S particle-enriched) complexes within a proteasome preparation to a larger extent than larger (doubly capped 26 S particleenriched) complexes (Table I, Experiment B). These observations are inconsistent with an activator mechanism involving reconstitution of depleted 26 S particles (5% of proteasomal particles) with their "missing" complement of S10b and S6Ј ATPases. An alternative interpretation is that the activator is a normal physiological entity and that coordinated synthesis of the activator and 26 S proteasomes is required to permit full proteolytic activity of 26 S proteasomes during ISM cell death. The insect muscle 220-kDa activator is distinct from the 11 S (PA28) complex, which activates 20 S proteasomes in an ATPindependent manner (Table I) The bead-immobilized Manduca activator complex retains the ability to activate the 26 S proteasome (Fig. 5). The gradient fractions enriched in 26 S proteasomes that were used in the experiments (Figs. 2 and 3) will contain only small amounts of 19 S and 20 S particles (8). If the mechanism of activating the 26 S proteasome is similar to that of the bovine modulator (22), then the affinity-immobilized S10b⅐S6Ј ATPase complex would have to bind either the 19 S or 20 S particles in the glycerol gradient fractions in such a manner that their productive association is catalyzed. The extent of activation observed renders this mechanism unlikely. An alternative scenario could be that the immobilized activator is able to bind to either (a) a complete 26 S particle or (b) a 19 S particle that subsequently binds to a 20 S particle. It is possible that an S10b⅐S6Ј-depleted 19 S particle is involved in these interactions, but the concentration of such particles in stage 7 ISM extracts is very low (8). In either case, the interaction with the 220-kDa complex increases proteasomal activity. This mechanism of activation by the immobilized complex is consistent with the fact that the full action of the modulator requires only a singly capped proteasome (23). Clearly, the 220-kDa complex and the modulator are related supramacromolecular structures.
A similar 220-kDa complex, containing the S10b and S6Ј ATPases, but not the S6 or S7 ATPase, has been demonstrated in both insect and human tissues and cells (Figs. 6 and 9 and Table II), and the properties of the complex suggest that it is not an assembly intermediate or catalyst for 26 S proteasome formation (see above). However, a critical question is whether the activator or the modulator results from 26 S proteasome dissociation during tissue disruption and biochemical analyses. Several experimental approaches (e.g. the use of chaotropic salts) were adopted in studies on the bovine modulator to discount this notion (17). Similarly, in this work, omission of ATP from homogenization buffers did not result in the generation of more 220-kDa complexes from Manduca muscle 26 S proteasomes. However, comparison of the bovine modulator FIG. 8. Disruption of proteasomal structures in extracts from human placental tissue frozen for several months. S2 extracts from human placenta that had been frozen for several months were fractionated by glycerol gradient centrifugation. छ, chymotrypsin activity (arbitrary units); Ⅺ, protein concentration (mg/ml). Pairs of fractions were pooled and subjected to Western analysis with different antibodies as indicated (see legend to Fig. 6 for details). S10b, S6Ј, and p27 (first lanes) show samples of His-tagged recombinant protein run as standards.
with the current activator is complicated by differences in isolation conditions. In yeast, ATP depletion and moderate salt concentrations (0.5 M NaCl) (28) cause 26 S complexes to dissociate into 20 S and PA700 (19 S) particles. For these reasons, we have used low salt conditions in the presence of ATP for partial isolation of the 220-kDa complex from insect and human tissues. The bovine modulator was isolated following elution of soluble proteins from DEAE-cellulose in 0.5 M NaCl and precipitation in 38% ammonium sulfate (17,20), which may well encourage dissociation of PA700 particles from 20 S cores and even partial disruption of PA700. Indeed, our analyses of red blood cell extracts isolated under similar conditions up to elution from DEAE-cellulose (Fig. 7) suggest that the majority of the S10b and S6Ј ATPases and some of the S7 ATPase have dissociated from PA700, consistent with earlier data (17). Although we were able to detect large amounts of p27 in the same fractions as S10b and S6Ј in red cell extracts, we could not detect p27 protein in supernatants from human placenta (or from human brain or liver) and certainly not in the 220-kDa complex. This contrasts with the apparent wide tissue expression of human p27 mRNA (25). Although p27 is claimed to be a component of both the modulator and PA700, the amount of this protein does not appear to be stoichiometric in relation to S10b and S6Ј (17,25). These analyses and others relating the p27 content of intact 26 S particles and dissociated PA700 complexes (29) are difficult to interpret in view of the possible effect of isolation conditions on p27 interactions with these different complexes. However, since p27 has not been proved to be essential for modulator function, its absence from our 220-kDa activator may not be of great significance.
It is apparent that when human placenta is deliberately not analyzed immediately, but is frozen for several months, large amounts of proteasomes and PA700 particles dissociate into smaller subcomplexes. These include small S10b⅐S6Ј complexes and larger complexes containing the S7 ATPase (Fig. 8). In fresh tissue, the S7-containing complexes are not observed, and the majority of the S10b and S6Ј ATPases remain in the 26 S proteasomes. The 220-kDa complexes isolated under these latter conditions seem unlikely to be dissociated from placental 26 S proteasomes since the 220-kDa complex is able to activate Manduca 26 S proteasomes much more effectively than 26 S placental proteasomes isolated from the same glycerol gradient as the 220-kDa complex itself (Table II). We have already shown that Manduca stage 7 proteasome preparations comprise (90 -95%) 26 S particles (8). Whatever the origin of the 220-kDa complex, it is clear from its ability to cross-activate 26 S proteasomes of different species that this proteasomal activator has been conserved during evolution. S10b and S6Ј can form heterodimers or larger complexes with each other in vitro, but show much less tendency to interact with other proteasomal ATPases (1). Currently, we do not know whether the 220-kDa complex identified here comprises (a) several molecules of both S10b (44 kDa) and S6Ј (45 kDa) ATPases (24) in the form of, for example, a tetramer or hexamer or (b) one molecule of both these ATPases together with other unidentified subunits.
Recently, significant amounts of S7 ATPase have been found dissociated from the 26 S proteasome in human bladder carcinoma cells (T24 cells) during mitosis, associating instead with a nuclear protein containing leucine-rich heptad repeats (HEC (highly expressed in cancer cells)) that inhibits the degradation of mitotic cyclin B in vitro (30). The HEC protein peaks in the M phase of the cell cycle and may be part of a complex containing the anaphase-promoting complex/cyclosome that targets mitotic cyclins for degradation (31). The exact role for the S7 ATPase in such a complex is not clear. The S8 ATPase also interacts with HEC in a yeast two-hybrid screen (30) and has been found in a heterogeneous range of molecular species in HeLa cell nuclear extracts (32). Numerous other reports have been made of interactions of individual proteasomal ATPases with other proteins or complexes, particularly those involving transcription factors (32)(33)(34)(35)(36)(37). Most of these interactions have been detected by yeast two-hybrid screens, and therefore, it is not known whether the ATPase involved is a single subunit or part of a complex with other ATPases.
The proteasomal ATPases belong to the AAA superfamily of ATPases involved in functions as diverse as membrane docking and fusion, peroxisome biogenesis, mitochondrial biogenesis, and proteasome regulation (38). The common molecular mechanism(s) by which the ATPases carry out these diverse functions are unknown, although protein unfolding and translocation may be common elements in these activities. Recently, one of the AAA ATPases has been shown to have molecular chaperone activity in that the ATPase domain recognizes and binds unfolded polypeptide substrates prior to their translocation to a peptidase domain for degradation (39). Proteasomal ATPases may act in an analogous manner. The potential role of individual ATPases, or species such as the 220-kDa activator complex described here, in substrate recognition for proteasome degradation or other processes remains to be elucidated. However, it is at least conceivable that the activator functions as an independent chaperone that can deliver substrates to the proteasome in a "processed" form, independent of the 19 S ATPases, which thus accelerates proteolytic degradation. FIG. 9. Placental S10b and S6 ATPases are in a complex that is bound to affinity-purified polyclonal anti-S10b antibody on Sepharose beads. Pooled glycerol gradient fractions (fractions 3-5) from placenta (500 l) were incubated with anti-S10b antibody-Sepharose beads (50 l), recovered, and washed as described in the legend to Fig. 5. The activator complex was eluted from washed anti-S10b antibody-Sepharose beads with 1 M glycine (pH 2.5) (3 ϫ 100 l) into 1 M Tris-HCl (pH 8.0) (3 ϫ 100 l). Loaded material and pooled elutions were reduced to 150 l with Pall Filtron Nanosep centrifugal concentrators (10-kDa cutoff). Western analysis was carried out on the placental extract (S2), unbound material, the pooled elutions (elution), and the Sepharose beads boiled in SDS-polyacrylamide gel electrophoresis loading buffer (150 l; beads). In some cases, His-tagged recombinant proteins were run as standards (first lanes).