Multiple Forms of the 20 S Multicatalytic and the 26 S Ubiquitin/ATP-dependent Proteases from Rabbit Reticulocyte Lysate*

We have used native gel electrophoresis followed by fluorogenic peptide overlay to identify multiple forms of rabbit reticulocyte multicatalytic protease (MCP) or 20 S protease, and two forms of rabbit 26 S ubiquitinl ATP-dependent protease. An abundant, fast-migrating 20 S complex (20 SF) possesses modest ability to hy- drolyze the fluorogenic peptide succinyl-Leu-Leu-Val-Tyr-4-methyl-coumaryl-7-amide. In contrast, two mi- nor, slower migrating species cleave the peptide at high rates. A unique 30-kDa polypeptide is associated with one of the active MCPs, and a 160-kDa subunit is associated with the other. Two electrophoretically distinct 26 S proteases can also be isolated from rabbit reticulocyte lysate. The faster migrating form, 26 SF, is more resistant to in-activation by ATP depletion. Despite the differential response to nucleotides and the distinctive electrophoretic mobilities of 26 SF and 26 Ss, we have not identified any subunit differences between the two en- zymes. In addition to active 26 S proteases, we have discovered and purified a proteolytically inactive par- ticle that contains subunits characteristic of the 26 S protease (e.g. molecular masses between 30 and 110 kDa). Incubation of this protein complex with purified MCP and ATP results in the formation of the 26 S proteases. conjugates. Spectrofluorometric assays consisted of 100 p~ fluorogenic peptide in 30 mM Tris-HCI, pH 7.8, 5 mM MgCI,, 10 mM KC1, 0.5 mM dithiothreitol, with or without 2 mM ATP or ATP-regenerating system. The reaction (100 pl final volume) was initiated by adding enzyme and incubating at 37 "C for 15 min prior to quenching with 200 pl of ethanol. Fluorescence was measured on a Perkin-Elmer fluorometer using an excitation wavelength of 380 nm and an emission wavelength of 440 nm. The following fluorogenic peptides were used Suc-Leu-Leu-Val-Tyr-MCA, Suc-Ala-Ala-Phe-MCA, and Pro-Phe-Arg-MCA. Electrophoresis-All samples were analyzed on mini gels using a Mini-Protean gel apparatus (Bio-Rad). Nondenaturing gels consisted of 2.5% stacking and 4.5% resolving gels cast in 90 mM Tris, pH 8.3, 1.6 mM borate, and 0.08 mM EDTA. Samples were electrophoresed for 800 V-h in a cold room. SDS-PAGE consisted of 10% resolving and 4% stacking gels in 25 mM Tris, pH 8.5, 200 mM glycine, 0.05% sodium dodecyl sulfate. Samples were run for 150 V-h at room temperature. After electrophoresis and/or substrate overlay (see be-low), proteins were visualized by staining in 0.2% Coomassie Brilliant Blue in 22.5% methanol, 7.5% acetic acid. Alternatively, gels were stained by the Bio-Rad silver stain method. Substrate Overlays-Protease activity was detected in nondenatur- ing gels by overlaying the gels with 30 mM Tris-HCI, pH 7.8, 5 mM MgC12, 10 mM KC1, 0.5 mM DTT, 2 mM ATP, 200 pM fluorogenic peptide, and incubating the gels at 37 "C for 30-60 min. The SDS-PAGE revealed that several MCP subunits were fluoresceinated, although to different extents.


Multiple Forms of the 20 S Multicatalytic and the 26 S Ubiquitin/ATPdependent Proteases from Rabbit Reticulocyte Lysate*
(Received for publication, March 17, 1992)

Laura Hoffman, Greg Pratt, and Martin RechsteinerS
From the Department of Biochemistry, University of Utah School of Medicine, Salt Lake City, Utah 84132 We have used native gel electrophoresis followed by fluorogenic peptide overlay to identify multiple forms of rabbit reticulocyte multicatalytic protease (MCP) or 20 S protease, and two forms of rabbit 26 S ubiquitinl ATP-dependent protease. An abundant, fast-migrating 20 S complex (20 SF) possesses modest ability to hydrolyze the fluorogenic peptide succinyl-Leu-Leu-Val-Tyr-4-methyl-coumaryl-7-amide. In contrast, two minor, slower migrating species cleave the peptide at high rates. A unique 30-kDa polypeptide is associated with one of the active MCPs, and a 160-kDa subunit is associated with the other.
Two electrophoretically distinct 26 S proteases can also be isolated from rabbit reticulocyte lysate. The faster migrating form, 2 6 SF, is more resistant to inactivation by ATP depletion. Despite the differential response to nucleotides and the distinctive electrophoretic mobilities of 26 SF and 26 Ss, we have not identified any subunit differences between the two enzymes. In addition to active 26 S proteases, we have discovered and purified a proteolytically inactive particle that contains subunits characteristic of the 26 S protease (e.g. molecular masses between 30 and 110 kDa). Incubation of this protein complex with purified MCP and ATP results in the formation of the 2 6 S proteases.
The cytosolic and nuclear compartments of eucaryotic cells contain a variety of proteases (1). These include members from four classes of proteolytic enzymes. Cysteine proteinases are represented by the calpains (2); insulinase provides an example of a cytoplasmic metalloprotease (3). Among the cytoplasmic serine proteases are tripeptidyl peptidase (4) and proline endopeptidase (5), and cathepsin E, an acid protease, is apparently found in red cell cytoplasm (6). The extent to which each of these enzymes contributes to overall intracellular proteolysis is presently unknown.
Eucaryotic cells also contain two large proteolytic complexes whose catalytic mechanism has been a matter for some debate. The multicatalytic protease (MCP)' is a large, 20 S * These studies were supported by National Institutes of Health Grant GM37009. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ sLLVY, succinyl-Leu-Leu-Val-Tyr; AMP-PNP, 5'-adenylyl-P,y-1midodiphosphate. particle composed of 24 or so subunits ranging in molecular mass from approximately 20 to 32 kDa (for review, see Refs. 7 and 8). The subunits are stacked in four layers to produce a cylinder-shaped structure measuring 11 x 16 nm. The particle appears to have a hollow core (9).
Subunits with electrophoretic mobilities characteristic of MCP were also found associated with 10 or more polypeptides ranging from 34 to 110 kDa in size (10). This larger protease complex sediments at 26 S and is capable of degrading ubiquitin-lysozyme conjugates in an ATP-stimulated reaction. Based on similarity in subunit composition, we proposed that MCP subunits were common to both 20 S and 26 S complexes; we also suggested that the higher molecular mass chains (34-110 kDa) confer ATP dependence and ubiquitin recognition on the 26 S protease. Support for this idea has been obtained in several studies that show formation of the 26 S complex when MCP and crude protein fractions are incubated in the presence of ATP (11-13). Both large protease complexes are thought to play major roles in intracellular proteolysis, although firm evidence for this supposition is lacking.
As part of our continuing effort to characterize the 26 S protease, we conducted analyses that involve native gel electrophoresis followed by fluorogenic peptide overlays to localize protease activity. In the course of these studies, we observed that protease activity was not proportional to the distribution of MCP protein. This led to the discovery of electrophoretically distinct 20 S complexes with differing specific activities.
Here we describe differences in subunit composition between the multiple forms of MCP, and we demonstrate that one highly active form of 20 S can be generated by a novel protein complex isolated from human red cells or rabbit reticulocytes. Moreover, we resolved two electrophoretically distinct species of 26 S protease and have examined the subunit composition of the two enzymes. Finally, we identify a single protein complex that associates with MCP subunits to form active 26 S proteases.
Purification of High Molecular Mass Proteases-Reticulocytosis was induced by phenylhydrazine injection following schedules and dosages prescribed by Lingrel (14). Blood was collected by cardiac puncture, and washed red cells (>go% reticulocytes) were lysed in 1.6 volumes of 1 mM DTT. After centrifuging the lysate at 100,000 X g for 90 min, glycerol was added to 20% (v/v), and the lysate was loaded onto a TSK-DEAE column equilibrated in 10 mM Tris-HC1, pH 7, 1 mM DTT, 20% glycerol, pH 7 (TDG). Approximately 3 ml of lysate was added per ml of ion exchange resin. The column was washed overnight with 240 ml of 10 mM Tris-HCI, pH 7,25 mM KCl, 10 mM NaC1, 1.1 mM MgC12, 0.1 mM EDTA, 1 mM DTT, 20% glycerol, p H 7 (TSDG), then washed with 60 ml of TSDG containing 0.1 M KC1, and finally developed with a 0.1-0.25 M KC1 gradient in TSDG. The multicatalytic proteases elute between 175 and 195 mM KC1; the 26 S enzymes are found over the range 180-220 mM KC1 (see Fig. 2).
Appropriate fractions were combined, and the high molecular mass proteases were pelleted by centrifuging a t 100,000 X g for 17 h. The pellets were resuspended in the bottom 0.5 ml of solution in the tube, diluted to 5% glycerol, and layered onto 10-40% glycerol gradients.
After centrifuging for 22 h a t 25,000 rpm in an SW 28 rotor a t 4 "C, the 28-ml gradients were fractionated into 2-ml aliquots from the bottom. Gradient-purified proteases were concentrated by spinning through Centricon-30 microconcentrators.
Preparation of Modified Fraction 11-For certain experiments a modified fraction I1 (15) was used. This was prepared by first washing lysate-loaded TSK-DEAE with 160 ml of 0.15 M KC1 in TDG and then eluting with 0.275 M KC1 in TSDG. The modified fraction I1 was then taken to 38% ammonium sulfate, and the precipitated proteins were collected by centrifugation and dialyzed against 2 liters of TSDG to produce mII38.
Protease Assays-The 20 S protease was assayed using fluorogenic peptides; the 26 S protease was measured using fluorogenic peptides as well as ubiquitin-lysozyme conjugates. Spectrofluorometric assays consisted of 100 p~ fluorogenic peptide in 30 mM Tris-HCI, pH 7.8, 5 mM MgCI,, 10 mM KC1, 0.5 mM dithiothreitol, with or without 2 mM ATP or ATP-regenerating system. The reaction (100 pl final volume) was initiated by adding enzyme and incubating a t 37 "C for 15 min prior to quenching with 200 pl of ethanol. Fluorescence was measured on a Perkin-Elmer fluorometer using an excitation wavelength of 380 nm and an emission wavelength of 440 nm. The following fluorogenic peptides were used Suc-Leu-Leu-Val-Tyr-MCA, Suc-Ala-Ala-Phe-MCA, and Pro-Phe-Arg-MCA.
Electrophoresis-All samples were analyzed on mini gels using a Mini-Protean gel apparatus (Bio-Rad). Nondenaturing gels consisted of 2.5% stacking and 4.5% resolving gels cast in 90 mM Tris, pH 8.3, 1.6 mM borate, and 0.08 mM EDTA. Samples were electrophoresed for 800 V-h in a cold room. SDS-PAGE consisted of 10% resolving and 4% stacking gels in 25 mM Tris, pH 8.5, 200 mM glycine, 0.05% sodium dodecyl sulfate. Samples were run for 150 V-h a t room temperature. After electrophoresis and/or substrate overlay (see below), proteins were visualized by staining in 0.2% Coomassie Brilliant Blue in 22.5% methanol, 7.5% acetic acid. Alternatively, gels were stained by the Bio-Rad silver stain method.
Substrate Overlays-Protease activity was detected in nondenaturing gels by overlaying the gels with 30 mM Tris-HCI, pH 7.8, 5 mM MgC12, 10 mM KC1, 0.5 mM DTT, 2 mM ATP, 200 p M fluorogenic peptide, and incubating the gels at 37 "C for 30-60 min. The fluorescent gels were transilluminated by a UV light and photographed with a Polaroid camera.
Immunodetection of MCP Subunits-MCP was purified to homogeneity from human red blood cells using chromatography on TSK-DEAE, ammonium sulfate precipitation, gel filtration, and glycerol gradient centrifugation (10). Rabbit antibodies to human MCP were prepared by the method of Vaitukaitis (16). Human MCP (100 pg) was emulsified in Freund's adjuvant and injected subcutaneously a t monthly intervals. Sera were collected after the fifth injection. Rabbit IgGs were prepared by precipitation with 40% ammonium sulfate and chromatography on TSK-DEAE. Immunoglobulins were collected in the breakthrough 10 mM sodium phosphate, pH 7.5.
Proteins were transferred from SDS-PAGE or nondenaturing gels onto nitrocellulose using the Bio-Rad Trans-Blot apparatus. The nitrocellulose was blocked with 5% nonfat dry milk in 50 mM Tris, p H 7.0,0.9% NaCl (Tris-buffered saline) prior to incubation in Trisbuffered saline and milk containing 16 pg/ml anti-20 S IgG. The secondary antibody used for detection was "'1-goat anti-rabbit IgG (2 X 10" cpm/ml). Nitrocellulose blots were exposed to Kodak XAR film for autoradiographic localization of the radioiodinated secondary antibody.
Fluoresceination of Rabbit MCP-Rabbit reticulocyte MCP, purified as described above for human MCP, was reacted with fluorescein succinimidyl ester a t a molar ratio of 50 fluorescein succinimidyl ester molecules per MCP. Unconjugated fluorescein succinimidyl ester was removed by Sephadex G-25 chromatography. Assuming full recovery of MCP and unquenched fluorescence yield from fluorescein bound to protein, we calculate that about 4 fluoresceins (F) were conjugated to each MCP particle. SDS-PAGE revealed that several MCP subunits were fluoresceinated, although to different extents.

Identification of Protein Complexes Containing Subunits of
the 20 S and 26 S Proteases-In 1987 we showed (10) that the 20 S and 26 S protease complexes can be separated on native acrylamide gels and subsequently localized by overlaying the gel with the fluorogenic peptide substrate Suc-Leu-Leu-Val-Tyr-4-methyl-coumaryl-7-amide (sLLVY-MCA).
These initial studies, in which samples were typically electrophoresed for 200 V-h on large gels, revealed two distinct 26 S activities. However, a single band of sLLVY-MCA hydrolytic activity was present in the 20 S region of the gel, and it appeared to comigrate with 20 S protein. More recently, preparations of 20 S and 26 S proteases were electrophoresed for longer periods on minigels, and as shown in Fig. 1, it became apparent that cleavage of sLLVY-MCA does not coincide with the major 20 S complex. After separating glycerol gradient-enriched 20 S protease on native gels, enzymatic activity was visualized by peptide overlay (Fig. 1, panel A ) . The  were similarly analyzed on native gels, the two bands of enzyme activity roughly match the protein present in 26 S complexes ( Fig. 1, panel B). However, a slightly faster migrating protein complex is also present. Although it does not cleave sLLVY-MCA, it will be seen that this component contains subunits found in the 26 S protease. For easier description of the experiments, we call the inactive complex the "ball," based on a "ball and cylinder" model of the 26 S protease (17). Likewise, we distinguish 20 S and 26 S enzymes as fast or slow based on their electrophoretic properties. Chromatographic Properties of 20 S and 26 S Complexes-To characterize further the various species described in Fig.   1, we chromatographed reticulocyte lysate on TSK-DEAE and quantitated sLLVY-MCA cleavage in the column fractions (Fig. 2, top panel). It can be seen from the peptide overlays presented in the middle panel of mass from 25 to 110 kDa, can assemble with MCP to form the 26 S complex.

The Subunit Compositions of Fast and Slow Forms of 20
S-Because they contain 20 S protease virtually free of 26 S species, DEAE-gradient fractions 145-151 were combined to form pool I. The 20 S enzyme was pelleted by overnight centrifugation at 100,000 x g, redissolved in 0.5 ml, and sedimented on a 10-4096 glycerol gradient. DEAE-gradient fractions 152-166 were treated similarly to produce glycerol gradient-purified 20 S protease, termed pool 11. Samples from the two 20 S preparations were then analyzed on 4.5% native gels. As expected, both slow and fast electrophoretic forms were present in each pool, and sLLVY-MCA cleavage was largely confined to the slower migrating species (Fig. 3, panel  A ) . However, when the various bands of 20 S protease were excised from the native gel and analyzed by SDS-PAGE, the slow forms in pools I and I1 were found to differ in subunit compositions (Fig. 3, panel €3).  Despite close inspection of many gels, we have been unable to detect any consistent differences in the subunit patterns of the two 26 S enzymes. The 26 SS species often generates an SDS-PAGE pattern that contains more high molecular mass polypeptides (>130 kDa), but the differences between 26 SF and 26 Ss have proved quantitative rather than qualitative.
Moreover, none of the larger proteins is present at levels approaching the llO-kDa/lOO-kDa chains. Since the latter are thought to be present at one copy each in the 26 S protease (lo), it seems unlikely that a high molecular mass component is solely responsible for the observed mobility difference.
Subunit patterns for the two 26 S enzymes are also remarkably similar to that of the ball component. Aligning lanes 1-3 in Fig. 6 with the profiles from either slow or fast 26 S, one can match virtually every subunit with molecular masses greater than 30 kDa. The ball component would seem to be missing only MCP subunits, a fact confirmed by the Western blots in Fig. 8.  After incubation, the proteases were separated on native gels, and their activities were assayed by peptide overlay. Two representative experiments, presented in Fig. 7, demonstrate that the activity of 26 SS decreases to a greater extent than 26 SF when ATP is omitted or substituted by CTP, GTP or UTP, but both 26 SS and 26 SF remain active in the presence of ATP or the nonhydrolyzable analog AMP-PNP. EDTA appeared to have little effect on 26 S activity. The addition of apyrase results in the loss of most 26 S activity and the disappearance or altered migration of the remaining protein complexes.
Western Blot Analysis of 26 S, 20 S, and Ball Complexes Using Anti-MCP Antibodies-The ball component .contains polypeptides with sizes typical of MCP subunits (e.g. 25-31 kDa), raising the possibility that inactive MCP subunits are present in the complex. This was tested by using anti-20 S antibodies to probe Western blots of the various protein assemblies. The nondenaturing gels in the upper three panels of Fig. 8  shown that in the presence of ATP, the 26 S protease can be assembled from three components termed CF1, CF2, and CF3. Although both Hershko's group (11) and Driscoll and Goldberg (12) have provided evidence that CF3 is the multicatalytic protease, some investigators do not believe that MCP is part of the larger protease (19). Thus, we have used a different approach to confirm this assignment. Purified rabbit MCP was labeled with fluorescein and mixed with increasing amounts of protein precipitated from fraction I1 by 38% ammonium sulfate. It is evident from the gels in Fig. 9 that generation of 26 S activity is paralleled by association of fluoresceinated MCP subunits with the 26 S complex. We consider this result to be a direct demonstration that MCP subunits are incorporated into the ubiquitin/ATP-dependent 26 S protease.
Assembly of the 26 S Protease from MCP and the Ball Components-When high molecular mass complexes in DEAE samples 167-185 (Fig. 2) were centrifuged on glycerol gradients, the ball component sedimented slower than the 26 S protease. The observed separation provided us with a protein complex that exhibited no protease activity and lacked MCP subunits but was comprised of polypeptides characteristic of the ubiquitin/ATP-dependent protease (see Fig. 6, panel A). Because the molecular masses of CF1 and CF2 have been estimated a t 600 and 250 kDa, respectively (18), and because the ball component sediments as though it is -700 kDa, we hypothesized that CF1 and CF2 preassemble t o form the ball. If so, the multicatalytic protease and the ball component should combine to produce active 26 S. This was tested by incubating the two protein complexes in the presence of ATP followed by analysis of the reaction mixture on native gels. The peptide overlays and Coomassie patterns in Fig. 10 confirm that 26 S proteases can be assembled from two slower sedimenting particles, MCP and the ball.

DISCUSSION
The multicatalytic protease has been implicated in various reactions ranging from energy-independent degradation of oxidized proteins (20)(21)(22)(23) to ATP-stimulated degradation of ubiquitin-lysozyme conjugates (10) and antigen presentation by major histocompatibility complex class I receptors (24)(25)(26)(27)(28)(29). If the enzyme participates in such diverse processes, one might expect it to interact with other cellular components that would provide specificity for substrate selection. This expectation is supported by the experiments presented above. We have identified five distinct forms of MCP; four of these (20 SsRo,20 Ss160, and 26 Ss/26 SF) are, in fact, associated with one or more additional polypeptides. Moreover, these extra chains can markedly affect catalytic activity or substrate selection by MCP.
The fast electrophoretic form, 20 SF, is the simplest MCP, at least in terms of subunit composition. On one-dimensional SDS-PAGE, rabbit reticulocyte MCP can be resolved into 8- 10 bands which vary in size from 22 to 32 kDa (see Fig. 3). We consider 20 S F to have the "traditional" MCP subunit composition reported by numerous investigators (for review, see Refs. 7 and 8). It should be noted that 20 SF cleaves a variety of fluorogenic peptides, but it does so with very low apparent specific activity (Fig. 4). Of the two slower migrating forms, 20 SsBo also contains subunits with molecular masses below 35 kDa. However, the 32-kDa subunit in 20 SF is underrepresented, and a novel 30-kDa subunit is present. As shown in Fig. 5 (10). Thus, the association of MCP subunits with ball components confers important substrate selection characteristics on the smaller protease.
Our assertion that MCP combines with the ball component to form the 26 S enzyme conflicts directly with a recent paper by Kloetzel and his colleagues (19). Based on their inability to detect MCP subunits in the larger complex, they concluded that the 26 S enzyme and MCP are unrelated proteases. They admit that their negative result might be explained by insufficiently sensitive reagents. In Fig. 8 we show that antibodies to human MCP do, indeed, react with 26 S proteases separated on native gels. Moreover, we provide direct evidence that fluorescein-labeled MCP subunits are incorporated into the 26 S enzyme in an ATP-dependent reaction (Fig. 9). When these results are coupled with those of Hershko's group (ll), Driscoll and Goldberg (12) and Orino et al. (13), the evidence that MCP and ball combine to produce 26 S proteases would seem overwhelming. Thus, we attribute the contrary view to insufficiently sensitive analyses, and we consider the ball and cylinder or "mushroom" model for the 26 S protease strengthened significantly by the results presented here.
We have shown two mechanisms for activating sLLVY-MCA cleavage by MCP. In the first, 20 S is activated by 30-kDa subunits present in the human red cell regulator fraction (Fig. 5). In the second, sLLVY-MCA hydrolysis is markedly enhanced when MCP is mixed with the ball component and assembled into 26 S proteases (Fig. 10). Interestingly, 30-kDa subunits are also present in the ball component. Although we do not yet know if the same 30-kDa proteins exist in both ball and regulator, this is clearly an important question for future study.
The present paper has focused mainly on structural features of the 20 S and 26 S proteases, such as subunit composition, electrophoretic mobility, and sedimentation characteristics. However, a functional aspect of MCP deserves some comment. Based on differential sensitivity of Leu-Leu-Glu-pnitroanilide hydrolysis to detergents or protease inhibitors, Orlowski et al. (40) and Djaballah and Rivett (43) have described two classes of peptidylglutamyl peptide hydrolyzing sites in the multicatalytic protease. The two active sites are apparently thought to reside within the same particle. Although our identification of closely related, yet distinct, MCPs does not speak directly to the issue of the multicatalytic nature of individual complexes, they suggest an additional possibility for "multiple" sites that hydrolyze a particular peptide. Namely, Leu-Leu-Glu-p-nitroanilide active sites with distinct catalytic properties are present in separate protease complexes. We hypothesize that within cells relatively inactive 20 S F particles may be activated by assembly with various other components to produce distinct proteolytic complexes. Future kinetic studies on the known species of MCP (e.g. 20 &SO, 20 Ss160, and 20 SF) should extend our understanding of intracellular proteolysis.