Distinct 19 S and 20 S Subcomplexes of the 26 S Proteasome and Their Distribution in the Nucleus and the Cytoplasm*

The 26 S proteolytic complex (“26 S proteasome”) is a macromolecular assembly thought to be involved in ATP- and ubiquitin-dependent protein degradation in the cytoplasm of higher eukaryotic cells. This complex is composed of one 20 S cylinder particle (multicatalytic proteinase, 20 S proteasome) and two cap-shaped 19 S particles comprising a set of polypeptides in the Mr range of 35,000-110,000. Here we show that cell supernatant fractions contain both these two subunit complexes as distinct particles as well as assembled to 26 S proteasomes. We have separated and purified all three forms from Xenopus laevis oocytes and have determined their peptidase and protease activities. Using various anti- bodies specific for either a constitutive p52 polypeptide of the 19 S cap complex or for proteins of the 20 S cylin- der particle, we have immunolocalized these complexes in both the cytoplasm and the nucleus of diverse species and cell types. The occurrence of all three forms, the 26 S proteasome, the 20 S cylinder particle, and the 19 S cap complex in the nucleoplasm has also been demonstrated in analyses of isolated giant nuclei from Xenopus oo- cytes. In addition, we show that the 19 S and 20 S subcomplexes can be released from 26 S proteasomes by ATP depletion and that readdition of

which are specifically recognized by the 26 S proteasome. In addition to the control of the lifetime of misfolded and special regulatory proteins, this pathway has also been implicated in the processing of intracellular antigens to T lymphocytes (14). However, activity of the 26 S proteasomes is not restricted to ubiquitinated proteins. Recently, for example, the antizymedependent degradation of the short-lived nonubiquitinated enzyme ornithine decarboxylase has been shown to be catalyzed by the 26 S proteasome (15).
The 26 S protease has first been isolated from the anucleate reticulocytes of rabbits (2) and subsequently also from other vertebrate cells and tissues as a large complex (estimated molecular mass 1.5-2.0 MDa) composed of probably more than 25 polypeptide subunits in the size range from M , = 22,000 to 110,000 (16)(17)(18)(19). The integrity of the complex seems to depend on the presence of ATP because 26 S proteasomes are not detectable in lysates of ATP-depleted reticulocytes but can be reconstituted in the presence of ATP from three "conjugate breakdown factors" (CF)' 1, 2, and 3 (4).
CF3 has been identified as the 20 S cylinder particle (20 S proteasome; Refs. 20 and 211, a M , = 700,000 multicatalytic proteinase complex composed of a series of low molecular weight subunits ( M , = 19,000-36,000) in all eukaryotic cells (for reviews see Refs. 22 and 23) but comprising only two different polypeptides in the archaebacterium Thermoplasma (24). All polypeptides of the 20 S particle are found in purified 26 S proteasomes (251, and electron microscopy clearly shows that the 26 S proteasome is assembled from the cylinder-or barrel-shaped 20 S particle to which two large cap-shaped complexes are bound end-on (17,25,26; for earlier erroneous claims of an absence of 20 S particle proteins in 26 S proteasomes, see Refs. 27 and 28). The essential contribution of the 20 S subcomplex to the structure and function of the 26 S proteasome is also evident from the observation that various yeast mutants of the 20 S proteasome accumulate ubiquitin conjugates (29)(30)(31) and do not degrade model substrates of the ubiquitin system (32,33).
CF2 has been purified from reticulocytes and proposed to be identical with the previously identified 240-kDa inhibitor of the 20 S proteasome which is composed of six identical 40-kDa subunits (34)(35)(36). Chi and Etlinger (35) reported that CF2 incorporated into the 26 S proteasome is ubiquitinated, a modification not seen in the unassembled form. CF1 has not yet been purified. Recently, a hetero-oligomeric protein complex designated as "20 S ball" has been isolated from untreated, i. e. not ATP-depleted, reticulocytes (37), and a very similar particle (in this case termed "p") has been purified from extracts of Drosophila embryos (38). Both components dithiothreitol; NEPHGE, non-equilibrium pH gradient electrophoresis; The abbreviations used are: CF, conjugate breakdown factor; Dm, PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline.

7710
Subcomplexes of the 26 S Proteasome have been reported to associate with 20 S particles to form complexes resembling the native 26 S proteasome. Their relationship to CF1 and CF2 is unknown.
The intracellular distribution of the 26 S proteasome and its subunit complexes has remained unclear. Using highly specific antibodies, we have originally localized the 20 S barrel-shaped particle to both the cytoplasm and the nucleoplasm where it is distributed in a rather disperse form (39). This finding has since been confirmed by various authors in a broad range of tissues (40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50)(51). Obviously due to the lack of suitable antibodies to the other components of the 26 S proteasome, nothing is known about the intracellular location of the 26 S complex, as there is also insufficient information about the co-existence of the three complexes involved, the 26 S proteasome, the barrelshaped 20 S subunit particle, and the cap-like components associated with either end of the 20 S particles ("balls" sensu Rechsteiner; 10, 37).
Using mono-and polyclonal antibodies specific to individual polypeptides of these subunits and cell fractionation techniques, we have determined the localization and the physical state of these three particles. Because of the particular importance of the Xenopus oocyte system in studies of the nucleocytoplasmic distribution of diffusible proteins and particles (for reviews see Refs. 52 to 54), we have used this cell system with particular emphasis. We show that the 26 S proteasome occurs in a dispersed form in both the nucleus and the cytoplasm of a variety of cells, and that the two subcomponents, the 20 S cylinder particle and the newly identified 19 S cap complex, occur as distinct and free forms in both cell compartments, probably existing in an assembly equilibrium with the 26 S proteasome.
Moerschell (Harvard Medical School). Clawed toads of the species Xenopus laeois were obtained from the South African Snake farm (Fish Hoek, Republic of South Africa). Conditions used for culturing X. Zaeuis kidney epithelial cells of line A6 (XLKE-A6), rat kangaroo PtK2 cells, rat vascular smooth muscle-derived fibroblastoidal RVF cells and human liver PLC carcinoma cells have been described (55,56).
Preparation of Cytosolic and Nuclear Extracts from Xenopus 0ocytes"cytosolic extracts (protein concentration -30 mg/ml) were prepared as described (17), snap-frozen, and stored at -70 "C. For preparation of nuclear extracts, nuclei were isolated from stage %VI oocytes (57) following the protocol of Scalenghe et al. (58) with modifications (59). 1 m~ ATP was added to the buffer used to lyse the oocytes and to the sucrose solution in which the nuclei were collected (cf. Ref. 58). Nuclei were homogenized by pipetting with a narrow glass pipette, and the homogenate was clarified by centrifugation for 20 min at 3,500 x g. The resulting low speed supernatant (protein concentration -2.0 mg/ml) was snap-frozen and stored at -70 "C. For some experiments, ooplasms and nuclei were separated manually (60).
Purification of Cytosolic 26 S Proteasome Complexes-1.5 to 2.0 ml of oocyte extract were diluted 1:6 in acetate buffer with ATP (100 m~ potassium acetate/KOH, pH 7.2, 2.5 m~ magnesium acetate, 5 m~ EGTA) and loaded on six 35-ml sucrose gradients (1540% W/V) containing the same buffer. The gradients were centrifuged in a Beckman SW28 rotor (Beckman Instruments) for 21.5 h at 28,000 rpm and 4 "C. 1.2-ml fractions were collected using an Isco density gradient fraction collector model 640 (Isco, Lincoln, NE). Fractions containing 26 S complexes (as judged from peptidase assays or absorbance profiles) were loaded directly onto a Mono Q column (Pharmacia, Uppsala, Sweden) connected to an FPLC system (Pharmacia). The column was developed with a gradient of 0-1.0 M KC1 in "5:l" buffer with ATP (83 m~ KCI, 17 m~ NaCI, 1 m~ MgC12, 1 m~ ATP, 2 m~ DTT, 10 m~ Tris-HC1, pH 7.4). Fractions containing 26 S complexes (eluted with -0.3 M KC1 in "51" buffer) were further purified by recentrifugation through 12-ml sucrose gradients (1540% w/v) in a SW40 rotor (Beckman) for 1618 h at 30,000 rpm and 4 "C. 0.4-ml fractions were collected as above. In preparations used for proteolysis assays, the second sucrose gradient centrifugation step was omitted to obtain higher yields. In these experiments, 26 S complexes from Mono Q fractions were sedimented by centrifugation for 18 h in a SW60 rotor (Beckman) at 33,000 rpm and 4 "C. The supernatant was carefully removed, and the resulting pellet was redissolved in the remaining buffer volume (50-100 pl). The solution was diluted to a protein concentration of -2.0 mg/ml and stored at -70 "C in the presence of glycerol (60% v/v).
Purification of Nuclear 26 S Proteasome Complexes-Nuclear extracts were loaded on 35-ml sucrose gradients (2.0 ml/gradient) which were prepared, centrifuged, and fractionated as above. 26 S fractions were'further purified by affinity chromatography as described (17) using Sepharose columns to which 20 S particle antibodies had been coupled. Protein fractions were passed over -1.0-ml antibody columns twice at room temperature. The columns were subsequently washed 5 times with 1.0 ml of the protein buffer, and bound polypeptides were then eluted with 5 x 1.0 ml of 3.0 M KSCN in phosphate-buffered saline (PBS; 137 m~ NaC1, 3 rn KCI, 7 rn Na2HP04, 1.5 rn KHzP04, pH 7.4). For separation by gel electrophoresis, proteins in the load, flowthrough, wash, and eluate fractions were precipitated with trichloroacetic acid.
Purification of 20 S Cylinder Particles"100,OOO x g supernatant fractions of whole Xenopus ovaries were separated by DEAE-Fractogel chromatography as described (61). Fractions containing 20 S complexes were precipitated with ammonium sulfate (80% saturation) and stored at -70 "C. Before use (e. g. for proteolysis assays), precipitated protein was redissolved in acetate buffer and further purified by centrifugation through 12-ml sucrose gradients (SW40 rotor) as described above.
ATP Depletion of Purified 26 S Proteasome Complexes and of Cytosolic Extract-Purified 26 S complexes from Mono Q fractions (see above) were incubated in the presence of 0.25 UniUpl of grade I apyrase for 15 min at 30 "C and fractionated by sucrose gradient centrifugation as described above (SW40 rotor) but in the absence of ATP. Total cytosolic extract was diluted 1:6 in acetate buffer without ATP, incubated for 30 or 45 min at 30 "C, and fractionated by centrifugation through sucrose gradients (either in the SW28 or SW40 rotor) in the absence of ATP. In control experiments, ATP was added back to the diluted extract after the heat treatment, and the solution was incubated for 15 min at 20 "C, or the 30 "C incubation was done in the presence of an ATPregenerating system (2 m~ ATP, 10 m~ phosphocreatine, and 0.03 unit/pl creatine phosphokinase). In these control experiments, sucrose gradient centrifugation was done in the presence of 2 rn ATP.
Purification of 19 S Cap Complexes-Total cytosolic extract was ATPdepleted by incubation at 30 "C and fractionated by sucrose gradient centrifugation (SW28 rotor). 19-20 S fractions were loaded directly onto a Mono Q column which was developed with a gradient of 0-1.0 M KC1 in "5:l" buffer as above. Fractions containing 19 S complexes (eluted with -0.3 M KC1 in "5:l" buffer) were further purified by affinity chromatography on columns with bound 20 S particle antibodies or by precipitation with ammonium sulfate (38% saturation). All steps were done in the absence of ATP at 4 "C. In some experiments, untreated extracts were used and the first sucrose gradient centrifugation was done in the presence of ATP.
Preparation of Antibodies-Fractions highly enriched in 26 S complexes by Mono Q or affinity chromatography (see above) were mixed with equal volumes of complete ABM2 emulsion (Sebak, Germany). TWO BALB/c mice were each injected subcutaneously with 200 pl of emulsion containing 50 pg of 26 S complex protein. The same amount of protein, emulsified in ABM2, was injected 4 weeks later and again, in PBS, 4 weeks after the second injection. 3 days after the last injection, the spleen cells were fused with the myeloma cell line Ag8.653 at a ratio of 1O:l (62). Hybridoma culture supernatants were tested for antibody production by immunoblotting on total proteins of fractions enriched in 26 S complexes using the blotting apparatus MN28 (Biometra, Gottingen, Germany). Positive cultures were subcloned twice by limited dilution, and ascites fluids were produced as described (63). Immunoglobulin subclasses were determined in an enzyme-linked immunosorbent assay using subclass specific primary antibodies (Medac, Hamburg, Germany) and peroxidase-coupled secondary antibodies (Sigma). p32 subunit of 20 S complexes has been described (17, 39). Mouse The preparation of guinea pig antibodies specifically recognizing a antibodies recognizing a p52 polypeptide of the 19 S subcomplex of 26 s proteasomes were also prepared.

.
_ . soluble antigens during the various incubation steps (cf. Refs. 39, 61, 64, and 65). In all cases, 2 mM MgCI, was added to the PBS solution. Cultured cells grown on coverslips were shortly washed in PBS, fixed with methanol a t -20 "C for 10 min, and dipped four times in ice-cold acetone. After drying, cells were incubated with antibodies (undiluted hybridoma supernatants or sera diluted 1:lOO in PBS) for 10, 15.20, or 120 min. Bound antibodies were visualized by incubation for 30 min with Texas Red-conjugated secondary antibodies (Dianova, Hamburg, Germany) diluted 1:200 with PBS. After washing in PBS, specimens were dipped in ethanol and embedded in Mowiol (Hoechst, Frankfurt, Germany). Alternatively, the cells were fixed with 2% (w/v) formaldehyde (prepared freshly from paraformaldehyde) in PBS for 10 min, incubated for 5 min in 50 mM NH,CI in PBS, and permeabilized by incubation for 10 min in 0.1% (v/v) Triton X-I00 in PBS.
Cryostat sections of frozen tissues were air-dried and either fixed with acetone for 10 min a t -20 "C or with formaldehyde as described above except that PBS was used without MgCI,. After drying, sections were incubated with antibodies as described for cultured cells. Specimens were viewed with epifluorescence illumination (39).
Electron microscopy of protein fractions was done as described (17), except that samples from sucrose gradients were adsorbed to the carbon support film directly without prior vacuum dialysis.
Enzyme solution (3% w/v) and addition of 560 pl of a 5% (w/v) trichloroacetic acid solution. After a 15-min incubation on ice, precipitated proteins were removed by centrifugation, and radioactivity in the supernatant was determined by y-or scintillation counting. ATPase assays were done as described (61). Gel Electrophoresis and Immunoblotting-One-and two-dimensional gel electrophoresis was done according to Thomas and Kornberg (67) and OFarrell et al. (68), respectively. For SDS-PAGE, protein samples were precipitated with 20% (w/v) trichloroacetic acid for 1 h on ice or overnight a t 4 "C, washed with 90% and 100% (v/v) acetone, dried, dissolved in sample buffer, and separated on 15% (w/v) acrylamide gels. 5-20 pg of protein were loaded per lane, and separated polypeptides were stained with Coomassie Blue. For two-dimensional separation, 5

RESULTS
Purification a n d Characterization of 26 S Proteasome Complexes from Xenopus Oocyte Cytosol-26 S proteasome complexes were purified from cytosolic fractions by sucrose gradient centrifugation, Mono Q chromatography, and recentrifugation through sucrose gradients (Fig. l a ) . This Fig.  la, lane 1 ) were obtained when the final sucrose gradient centrifugation was omitted and 26 S complexes were sedimented directly from Mono Q fractions by high speed centrifugation. The protein complexes in the purified fraction and the associated peptidase activity sedimented as a symmetrical absorbance peak to the same position in the final sucrose gradient as the peptidase activity associated with 26 S particles did when total extracts were analyzed on a parallel gradient (data not shown; cf. Fig. 6, a and b). This sedimentation behavior and the observation that all polypeptide spots (ranging from 22 to 110 kDa) seen previously by two-dimensional gel electrophoresis of immunoaffinity-purified 26 S complexes (25) were also present in conventionally purified 26 S proteasomes (Fig. l b ) indicate that neither dissociation nor major loss of proteins occurred during isolation. It should be mentioned, however, that a "shoulder" of peptidase activity sedimenting with -23 S was noticed in the final sucrose gradient centrifugation (see, for example, Fig. 6b) which in electron microscopical analysis contained predominantly "one end-capped" particles while fractions corresponding to 26 S were enriched in the "double-capped" form of the complex (data not shown; for detailed description, see Refs. 17 and 25).
Purified 26 S complexes contained peptidase activity when examined with fluorogenic substrates. This activity was partially inhibited in the presence of 0.01% SDS (to 3650% of control reactions) while the peptidase activity of 20 S particles purified from oocyte cytosol was stimulated 2-to %fold under these conditions (data not shown; cf. Fig. 6). The 26 S complexes degraded ubiquitinated lysozyme in an ATP-dependent way (Fig. 2a, a'), but some proteolysis of unconjugated lysozyme and casein was also observed (Fig. 2, b and c ) . In contrast, degradation of lysozyme by 20 S particles was largely independent from substrate ubiquitination and ATP (Fig. 2a, a' and  b ) , whereas proteolysis of casein by 20 S particles was weakly stimulated in the presence of ATP (Fig. 2c). These results confirm our previous assumption (17) that the 26 S complexes from Xenopus oocytes correspond to the ubiquitin-dependent protease originally isolated from rabbit reticulocytes (2). Purified 26 S complexes also contained a low ATPase activity (17 nmol/mg/ min in Mono Q fractions of the purity shown in Fig. la, lane 1 ).
Imrnunolocalization of 20 S and 26 S Proteasome Complexes-Immunofluorescence microscopy under conditions that minimize the loss of soluble proteins from permeabilized cultured cells showed immunostaining for the 20 S cylinder particle-specific proteins as well as for the p52 polypeptide contained in 26 S complexes in both the nucleus and the cytoplasm of cultured Xenopus kidney cells (Fig. 3, a-b and d-e' 1. Usually, the nuclear immunofluorescence was more intense and, in the case of p52, often appeared somewhat finely granular. The immunostaining in both compartments was rather evenly dispersed and not demonstrably associated with any cytoplasmic or nuclear structure. Only nucleoli and metaphase chromosomes were essentially unstained. Interestingly, resealed nuclear fragments, so-called micronuclei, also showed accumulation of immunostained particles (arrowhead in Fig. 3b).
Practically identical results were obtained in cultured mammalian cells which showed mostly nuclear enrichment of staining (e.g. Fig. 3c) but sometimes also reduced nuclear staining (Fig. 3f). In the various cell types, essentially identical observations were made with the different mono-and polyclonal antibodies to the 20 S and 26 S particle polypeptides tested (see "Experimental Procedures"). Corresponding results, i.e. general cytoplasmic and nucleoplasmic immunostaining, were obtained when frozen tissue sections were examined (data not shown).2 Identification and Characterization of Nuclear Forms of 26 S Proteasome Complexes-20 S cylinder particles are abundant in the amphibian oocyte nucleus (73). 26 S proteasomes, however, have so far only been purified from cytoplasmic fractions of such oocytes (17,25) as of other kinds of cells (2, 16, 18, 19). In our cell fractionation studies of Xenopus oocytes, we noticed that the nuclei ("germinal vesicles") contained unexpectedly large quantities of 26 S complexes. Therefore, we decided to study the nuclear forms of these particles in detail.
We fractionated extracts from oocyte nuclei by sucrose gradient centrifugation and assayed the fractions for peptidase activity and by SDS-PAGE and immunoblotting with mAb26S-17 (Fig. 4). Fractions corresponding to 26 S contained the major peptidase activity present in nuclear extracts (data not shown) which co-sedimented with a set of polypeptides resembling the protein pattern of cytosolic 26 S proteasomes ( Fig. 4a, fractions 18-21 1. One of these polypeptides was identified as p52 (Fig. 4~').
Upon immunoafinity chromatography using antibodies to 20 S cylinder particles, this set of polypeptides was specifically retained (Fig. 4b), indicating that the 20 S cylinder particles (composed of 22-32-kDa polypeptides) are physically associated with the 35-110-kDa polypeptides. In electron microscopical analyses, negatively stained 26 S complexes present in sucrose gradient fractions appeared as double-capped particles identical in size and shape with the previously described purified cytosolic 26 S proteasomes (data not shown; cfi Refs. 17 and 25). Since isolated nuclei from Xenopus oocytes are largely devoid of cytoplasmic contaminations, these results strongly in- The polypeptide pattern of affinity-purified nuclear 26 S complexes as analyzed by two-dimensional gel electrophoresis (Fig. 4.c) was practically identical with that of the 26 S proteasomes isolated from the cytosol (cf. Fig. lb). We noticed, however, that two 50-and 60-kDa polypeptides of nuclear 26 S complexes were resolved into three isoelectric variants (Fig. 4c,  arrowheads) while cytoplasmic complexes contained only two forms of these subunits, which may indicate that nuclear and cytoplasmic 26 S proteasomes are identical in composition but bear different post-translational modifications.
It should be mentioned that the majority of polypeptides of 20 S particles seen in nuclear fractions is incorporated into 26 S complexes (Fig. k ) , in contrast to the situation in cytosolic extracts in which at least 50% are present as "free" 20 S particles (17; see also noblotting experiments with mAb26S-17 we noticed that the p52 polypeptide of 26 S proteasomes does not only occur in nuclear fractions containing 26 S complexes but also in 10 S and 19 S particles of a n as yet unidentified nature (Fig. 4a' 1. This suggests that not only 20 S particles but also other subcomplexes (10 S, 19 S) of the 26 S proteasome can exist in distinct unassembled forms.
Identification a n d Characterization of a 19 S Subcomplex of the 26 S Proteasome-When cytosolic extracts were analyzed by immunoblotting of sucrose gradient fractions with mAb26S-17 (Fig. 5c), the p52 polypeptide was not only detected in 26 S complexes (fractions 18-21) but also found in particles sedimenting with -19 S (fractions 12-15). This indicated the existence of p52 in a "free," i.e. unassembled subcomplex with a sedimentation coefficient similar to that of the 20 S cylinder particles present in the same fractions (Fig. 5, a and b). Clearly, p52 was not associated with 20 S particles as this polypeptide was not retained on immunoaffhity columns that bound 20 S particles (see below). We then examined whether the p52-containing 19 S complex represented a subcomponent of the 26 S proteasome. Since the presence of ATP is required for stability of the 26 S proteasome and ATP depletion results in the dissociation of the 20 S cylinder particle from the 26 S holoproteasome (4), we treated purified 26 S complexes with apyrase to achieve ATP depletioninduced disassembly (Fig. 6). In analyses of apyrase-treated samples by sucrose gradient centrifugation, both peptidase activity and the typical polypeptides of the 26 S complex were not recovered in fractions corresponding to 26 S. Instead, the peptidase activity was found in fractions corresponding to 19-20 S (Fig. 6c) which contained polypeptides of 20 S cylinder particles (22-32 kDa) as well as the group of 35-110-kDa polypeptides characteristic for 26 S proteasomes (data not shown). This indicated that disassembly of 26 S complexes had occurred, resulting in subcomplexes almost co-sedimenting at 19-20 s. This conclusion was supported by the absence of intact 26 S complexes in electron microscopical specimens of the 19-20 S fraction (data not shown) and by the known phenomenon (2) that the peptidase activity present could be stimulated by addition of 0.01% SDS (Fig. 6c), a diagnostic property of"free" 20 S particles not found for 26 S proteasomes (Fig. 6, a+).
We next examined whether the ratio of 19 S and 26 S complexes could also be influenced by altering the ATP concentration in cytosolic extracts. Extracts were incubated for 30 or 45 min at 30 "C, assuming that the presence of endogenous ATPases (e.g. Ref. 61) would lead to ATP depletion. Subsequently, proteins were separated by sucrose gradient centrifugation, and fractions were analyzed for peptidase activity and by immunoblotting with p52 antibodies (Fig. 7). After this treatment, no peptidase activity was recovered in fractions corresponding to 26 S, while the activity of 20 S particle fractions was significantly increased as compared to fractions assayed in control experiments (Fig. 7a). This loss of peptidase activity from 26 S particle fractions correlated with a shift of p52 from 26 S to a position corresponding to 19 S (Fig. 7, b and c), suggesting that disassembly of 26 S proteasomes to 20 S and 19 S particles had occurred. When the 30 "C incubation of extract was done in the presence of an ATP-regenerating system, the peptidase activity of 26 S complexes was unaffected (data not shown), demonstrating the ATP specificity of the observed effects. When ATP was added back after the 30 "C treatment and the extract was incubated for 15 min further at 20 "C, to allow complex formation, the peptidase activity was again recovered in 26 S complexes (Fig. 7d).
We In c, purified 26 S complexes were preincubated with apyrase before gradient centrifugation which was done in the absence of ATP, and peptidase activities were determined as above. Fraction 1 corresponds to the top of the gradient, and the 20 S and 26 S positions are indicated in a . Note that the presence of SDS partially inhibits the peptidase activity of 26 S complexes ( a and b ) but stimulates 20 S particles ( a and c). Note that after apyrase treatment the peptidase activity previously associated with 26 S complexes is now recovered at the position corresponding to -20 S of the gradient. Lanes R contain reference proteins as in and immunoaffhity chromatography (Fig. 8). Alternatively, we purified the 19 S complex directly from untreated oocyte extracts which brought identical results but lower yields (data not shown). The majority of p52, monitored by immunoblotting (Fig. 8, b' and c ' ) , co-purified with the set of polypeptides (35-110 kDa) characteristic for 26 S proteasomes which are not present in 20 S particles (Fig. 8c). Immunoblotting with anti- or after readdition ofATP (filled circles), and fractions were analyzed for peptidase activity. Note that the 30 "C treatment leads to disappearance of peptidase activity in fractions corresponding to 26 S ( a ), but that this activity can be recovered after readdition of ATP (d). Bars in lanes R indicate the position of reference proteins as in Fig. 4. bodies specific for 20 S particle polypeptides gave negative results (Fig. & I ) . Also, no significant peptidase activity was present in the final fractions of purified 19 S complexes in  Fig. 6c. lanes 13-16) were separated by Mono Q chromatography. and fractions were analyzed for peptidase activity (filled squares) and by recording the absorbance at 280 nm (A280, continuous line ). The dotted line shows the percentage of volume of buffer B used for elution (cf. "Experimental Procedures"; lower lecel, 0%; upper leuel, 1000). b and c, Mono Q fractions from the experiment shown in a were analyzed by SDS-PAGE and Coomassie Blue staining ( a ) and by immunoblotting with mAb26S-17 ( b ) . In addition to p52 (arrowhead), an -100-kDa band of unknown identity is recognized. c and c', fractions containing 19 S complexes (as in b, lanes 16-18) were further purified by afinity chromatography using antibody columns ( E ) and in the flow-through fraction ( F ) were analyzed by SDS-PAGE specific for 20 S cylinder particles. Polypeptides in the eluted fraction and Coomassie Blue staining (c) or immunoblotting (c') with a mixture of antibodies directed against 20 S and 26 S complexes. The arrowhead in c' denotes the p32 polypeptide of 20 S cylinder particles (recognized by mAb26S-161), and the black dot marks the p52 subunit of the 19 S complex (recognized by mAb26S-17). Note that subunits of 20 S cylinder particles and 19 S complexes have been completely separated by the affinity chromatography (c and c' ). Lanes R contain reference proteins (or bars indicating their position) as in Fig. 1. contrast to purified 20 S particles (data not shown). The p52containing protein complex could also be separated from 20 S particles by precipitation with 6 3 8 % ammonium sulfate which left the 20 S particles in the soluble fraction (data not shown).
When purified 19 S complexes were analyzed by two-dimensional gel electrophoresis and silver staining, all major polypeptide components of the 26 S proteasome that are not present in 20 S particles could be detected. This suggests that the 26 S proteasome does not contain any other major component than the 20 S cylinder particle and the 19 S complex. We assume that the 19 S complexes isolated in this study correspond to the particles seen in the electron microscope to cap both ends of the 20 S cylinder particle in the 26 S proteasomes (17, 25). Therefore, we refer to them as 19 S cap complexes. DISCUSSION From our results we conclude that both major compartments of the cell, the nucleoplasm and the cytoplasm, contain three distinct but functionally related particles which co-exist in an assembly-disassembly equilibrium. The 26 S holoproteasome and its two subcomplexes, the 20 S cylinder particle (multicatalytic proteinasel20 S proteasome) and the newly identified 19 S cap complex. The specific assembly-disassembly state is apparently influenced by ATP and probably other regulatory factors. The 19 S cap particle purified in this study from X. laevis oocytes seems to be identical with the 20 S ball complex prepared from rabbit reticulocyte lysates by Rechsteiner and colleagues (37) and also to the "p particle" recently described in Drosophila embryo extracts (38) while this work was in progress.
The relationship of the 19 S cap complexes isolated from ATP-containing cellular extracts to the subcomplexes of the 26 S proteasome which were identified in ATP-depleted reticulocytes (conjugate breakdown factors; Ref. 4) is presently unknown. It is conceivable that the 19 S cap complex corresponds to CF1, or it could represent a complex of CF1 and CF2. Further work is required to clarify these relationships, The availability of a series of antibodies specific for components of each the 20 S particle and the 19 S complex has allowed us not only to identify these proteins during cell fractionation but also to determine their intracellular distribution by immunolocalization techniques. Our results show that the 19 S cap complexes as the 20 S cylinder particles and the complete 26 S proteasomes are diffusely spread over the entire cytoplasm and nucleoplasm. This is in agreement with previous reports of a nucleocytoplasmic dispersion of proteins of the 20 S cylinder particle (3941,44,45,48) which has been reported to vary in relation to cell cycle and differentiation processes (42,  43, 46, 47,49, 50, 74, 75). Other reports claiming a restriction of the 20 S particles to the cytoplasm (76, 77) or a specific association with certain substructures in the nucleus (43) or in the cytoplasm (78-80) seems to reflect either preparative losses of these soluble proteins or unspecific "sticky retention" of residual particles on certain structures.2 The occurrence of considerable amounts of 26 S proteasomes and their 19 S and 20 S subcomplexes in the cell nucleus was demonstrable with particular clarity by cell fractionation techniques in the case of the amphibian oocyte nucleus. From these experiments, and also from our immunolocalization studies, we gained the impression that these particles can be enriched in the nuclei of many different, although not of all cells (see also Ref. 81). The concentration of 26 S proteasomes and their subcomplexes in nuclei, together with the reports of an intranuclear occurrence of considerable amounts of ubiquitin (82, 83) and the ubiquitin-activating enzyme El (84, 85) suggest nucleus-specific functions of the 26 S proteasome and its subcomplexes, e.g. in the regulation of the life span of nuclear proteins, including chromatin constituents (for a review on histone ubiquitination, see Ref. 86; for oncoproteins see Refs. 87 and 88). We hope that our findings inXenopus oocyte nuclei will provide a new basis for future experiments elucidating such nuclear functions and the nuclear transport of such large particles (for discussions, see also Refs. 89 to 91) as these giant nuclei are particularly suitable for molecular biological and biochemical studies.
While a series of individual polypeptides of the 20 S cylinder particle have been sequenced and characterized in considerable detail (for reviews, see Refs. 22 and 91), our knowledge ofthe 19 S cap components is still very limited. Dubiel et al. (92,931 have reported that two polypeptides of the 20 S balV19 S cap complex show considerable amino acid sequence homology to a 15s Mg2+-ATPase particle of 600 kDa which is similarly nucleocytoplasmically distributed and represents a hexamer of only one polypeptide subunit of 97 kDa (61, 72). Work on the other subunits and the functional meaning of the ATPase activities is in progress in various laboratories.
Our finding that the 19 S cap complex can be released from 26 S complexes by ATP depletion suggests that in the living cell the 19 S and 20 S particles exist in a n equilibrium between "free" forms and the 26 S proteasome assembly. Future experiments will have to show the turnover characteristics of these particle proteins and whether additional special components are involved in the dynamic regulation of this equilibrium and in possible changes of the proteolytic activities of the 26 S proteasome and its subcomplexes.