Molecular organization of a high molecular weight multi-protease complex from rat liver

J. MoZ.Biol. (1988) 203, 985-996

Molecular Organization of a High Molecular Weight Multi-protease Complex from Rat Liver Keiji Tanaka?, Tetsuro Yoshimura, Akira Ichihara Institute for Enzyme Research University of Tokushima, Tokushima 770, Japan

Atsushi Ikai, Masaaki Nishigai Department of Biophysics and Biochemistry T;miversity of Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan

Yukio Morimoto, Mamoru Sato, Nobuo Tanaka, Yukiteru Katsube Keiichi Kameyama and Toshio Takagi Institute for Protein Research Osaka University, Suita, Osaka 565, Japan (Received 3 November 1987, and in revised form 26 January 1988) A latent multifunctional protease with a molecular weight of 722,000 to 760,000 purified from rat liver cytosol has been reported. This paper reports on the structure and subunit composition of the enzyme. Electron microscopy showed that the enzyme was a ring-shaped particle of 160( + 7) A diameter and 1lO( + 10) A height with a small hole of 10 to 30 A diameter (1 A = O-1 nm). Small-angle X-ray scattering analysis indicated that the enzyme had a prolate ellipsoidal structure with an ellipsoid cavity in the center. The maximum dimension of the enzyme was estimated to be 210 A from a pair-distance distribution function. The radius of gyration obtained from a Guinier plot and the Stokes radius based on the ellipsoidal model were 66 A and 76 A, respectively. On two-dimensional gel electrophoresis, the purified enzyme separated into 13 to 15 characteristic components with molecular weights of 22,000 to 33,000 and isoelectric points of 4 to 9. These multiple components were not artifacts produced by limited proteolysis during purification of the enzyme, because the cell-free translation products in a reticulocyte lysate with poly(A)mRNA of rat liver consisted of multiple components of similar sizes, and because peptide mapping analyses with lysylendopeptidase and V8 protease demonstrated clear differences in the primary structures of these components. The 13 main components were isolated from the purified enzyme by reverse-phase high performance liquid chromatography and shown to be non-identical. A model of the enzyme is proposed on the basis of these observations and previous physicochemical studies. Interestingly, the morphology of this protease is similar to that of the 16 to 22 S ring-shaped particles found in a variety of eukaryotic organisms. The structural similarity between this multi-protease complex and various reported subcellular particles is discussed.

1. Introduction

Wilk & Orlowski. 1980: Rivett. .1985: Rav & Harris. 1985; Dahlmann’et aZ.1 1985; Tanaka etdaZ., 1986u: 1988u,b; McGuire & DeMartino, 1986; Ishiura et al., 1986; Hough et al., 1987). Since this enzyme is the main neutral protease in the cytosol of these cells, it may be responsible for extra-lysosomal proteolysis, which is thought to cause selective removal of abnormal proteins from cells (Rivett, 1985; Tanaka

Recently, various mammalian cells have been reported to contain a protease of unusually large size (DeMartino & Goldberg, 1979; Rose et al., 1979; t Author to whom all correspondence should be addressed. 985 0022%2836/88/200985-12

$03.00/O

0

1988 Academic

Press Limited

986

K. Tanaka

et al., 1986a). Previously we showed by quantitative enzyme immunoassay and immunohistochemical methods that this enzyme is localized mainly in the cytosol of all rat cells and tissues examined (Tanaka et al., 1986a). This ubiquitous distribution of the enzyme in rats seems to imply the importance of the enzyme in proteolysis. We also reported a procedure for purification of the enzyme from rat liver, and the enzymological and physicochemical properties of the purified enzyme (Tanaka et al., 1986a,b). An interesting characteristic of the protease is that it is present in a latent state in the cells. This must be important for protecting normal cellular proteins from degradation. There is presumably some mechanism for activating the latent enzyme in viva. In vitro, it can be activated in various ways which may induce its conformational change. Another interesting character of the enzyme is the presence of multiple catalytic sites for various proteins or peptides. The enzyme causes endoproteolytic hydrolysis of various proteins, such as casein and denatured bovine serum albumin, and also degrades two types of synthetic substrates: basic and hydrophobic oligopeptides, which are substrates for trypsin-like and chymotrypsin-like enzymes, respectively. Previously we demonstrated that these activities are manifested at different active sites in the enzyme (Tanaka et al., 1986a). Multiple catalytic functions have also been reported for the enzymes of bovine pituitary gland (Wilk & Orlowski, 1983), rat skeletal muscle (Dahlmann et al., 1985) and rabbit erythrocytes (Tanaka et al., 1988a). These multiple enzymatic functions in a single protein are probably advantageous for rapid degradation of proteins in cells, since multi-enzyme complexes are known to catalyze sequential reactions very efficiently (Ginsburg & Stadtman, 1970). The most outstanding physical character of this protease is its high molecular weight of approximately 750,000 (Tanaka et aZ., 1986b). This size is much larger than that’ of any other protease so far reported, and large enough to distinguish the enzyme from other well-known proteases (Barrett, 1977). Why is this protease so large? The reason is unknown, but may be related to some of the specific functions of the enzyme described above. For understanding the interesting characters of this structure and molecular enzyme, its subunit organization must be determined. Here, we provide evidence that the enzyme is an unusual complex consisting of multiple distinct subunits and propose a model of its gross structure, based on data obtained by small-angle X-ray scattering analysis and electron microscopy. This large protease was found by electron microscopy to be a ring-shaped particle, which has been described (Tanaka et al., 1986b). Similar morphology of the enzyme from rat skeletal muscle has been reported (Kopp et al., 1986). This morphology is of interest in connection with many reports of the presence of ring-shaped particles of similar structure in a wide variety of eukaryotic cells, such as Drosophila

et al

(Schuldt

& Kloetzel, 1985; Arrigo et al., 1985), laevis, (Kleinschmidt et al., 1983; Castano et al., 1986), and various mammals (Narayan & Rounds, 1973; Smulson, 1974; Harris & Naeem? 1981; Malech & Marchesi, 1981; Domae et al., 1982; Martins de Sa et aE., 1986). These particles have been found independently as molecules with different functions such as ribonucleoprotein particles named prosomes (Martins de Sa et aE., 1986), 19 S particles containing small heat-shock proteins (Schuldt & Kloetzel, 1985), and particles of pretRNA 5’ processing nuclease (Castano et al., 1986). In this paper, the relationship between this protease and these particles is also discussed. Xenopus

2. Materials and Methods (a) Materials The materials used were as follows: lysylendopeptidase (Wako Pure Chemical Industries, Osaka), Staphylococcus aurem Y8 protease (Miles), a Cosmosil 5C4-300 column (Nakarai Chemicals, Kyoto), a TSK-Gel phenyl SPW-RP column (Toyo Soda Co., Tokyo), pl marker proteins (Oriental Yeast Co., Tokyo), molecular weight marker

proteins

(Pharmacia

Fine

Chemicals, Uppsala),

a

reticulocyte lysate translation kit, and 14C-labeled molecular weight marker proteins (Radiochemical Centre. Amersham). (b) PuriJication of high molecular

weight protease

The enzyme was purified to homogeneity from rat liver as described (Tanaka, et al., 1986a). The protein concentration of the purified protease was calculated from the absorbance at 280 nm, assuming E:Tm = 9.61 (Tanaka et aE., 19866). (c) Electron ,microscopy About 100 pg of purified enzyme was applied to a TSK G4OOOSW column (Toyo Soda, Tokyo) connected to an

HPLCt

system (BIP-I

pump and UVIDEC

1OOV

monitor, Jasco, Tokyo) and eluted with 100 mMammonium acetate buffer (pH 7.4). The elution was monitored as absorbance at 280 nm. Fractions in t,he major peak containing monomeric forms of the enzyme were collected for electron microscopy, while minor peaks, representing oiigomeric forms, were discarded. The enzyme concentration was then adjusted to about 50 pg/ml with buffer. The enzyme was negatively stained on a glow-discharge carbon-coated Formvar membrane with 3% (w/v) uranyl acetate (pH 4.6). The negatively stained sample was examined by electron microscopy in a Hitachi H7000 electron microscope at a direct magnification of 50,000 x For estimation of its height, the enzyme was unidirectionally shadowed with platinum vapor at an angle of 5.7” in a vacuum chamber of an Eiko FD-3 freeze-fracture device at 10-l Torr (1 Torr = 133.322 Pa). The actual shadowing angle was estimated from the ratio of the shadow length to that of latex spheres of 880 A diameter polyethylene (1 A = O*lnm). t Abbreviations used: HPLC, high performance liquid chromatography; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis.

Gross

(d) Small-angle

X-ray

scattering

Structure

of Mammalian

analysis

(i) X-ray experiments The X-ray source was a spot of O-4 mm x 8 mm on the copper anode of a Philips fine-focus X-ray tube which was operated with a Rigaku Denki D9C X-ray generator run at 40 kV/30 mA. Line focusing was used to obtain intense scattered X-rays. The nickel-filtered (10 pm thickness) X-rays were collimated with a slit of 0.3 mm x 10 mm, and then reflected and focussed by a nickel-coated glass mirror through 2 limiting slits. Scattered X-rays were recorded on a l-dimensional position-sensitive proportional counter (delay-line type). The sample-to-counter distance was 307 mm. The scattered X-rays were accumulated for 5000 s in the range of scattering angles of 2.6 x 10m3 to 8.2 x lo-* rad. For each measurement, 8 successive scattering profiles were recorded, and no detectable change was observed. Therefore, the profiles were summed to increase the accuracy of the scattering intensity data. All X-ray experiments were performed at 5.0( kO.1) “C, controlled by circulating water in the sample holder. (ii) Data reduction and analysis The scattering intensities recorded on both sides of the primary beam were averaged at equivalent points after subtracting background intensities. The center of the primary beam (zero-angle) was precisely searched so as to realize the best coincidence of the scattering intensities at equivalent points on the 2 sides of the zero-angle. The coincidences expressed by a reliability factor, R,,, (Blundell &I Johnson, 1976), were all good. The were then intensities, scattering independent I(4 obtained by deconvolution (desmearing) for the beam height effect by the method of Glatter (1982). The scattering parameter, s, is defined by 471sin e/n, where 20 is t,he scattering angle, and 1 the wavelength of X-rays (I = 1.5418 A). The radius of gyration, R,, was obtained from a Guinier plot (Guinier & Fournet, 1955) of the scattering intensities at lower scattering angles by a least-squares method. The pair-distance distribution function, P(r) (Glatter, 1982) was also calculated by Fourier-transform of the scattering intensity data to obtain real-space information of a particle such as the maximum particledimension, D,,,,.. The D,,,,, value was defined as the distance T at P(r) = 0. The interparticle interference effect was estimated from the scattering intensity data at a smaller-angle region (Guinier plot region). Measurements were made on protein solutions ranging in concentration from 5.0 to 22.5 mg/ml. Modest linear increase in R, with decrease of protein concentration was observed until a concentration of 10.0 mg/ml, but no significant increase was observed below this concentration, indicating that the interference effect was negligible at this concentration. Therefore, subsequent analyses were performed using scattering intensities measured on protein solution at a concentration of 10.0 mg/ml. (e) Electrophoresis

SDS-PAGE was carried out by the method of Laemmli (1970) in 15% (w/v) slab gels containing 0.1 y0 (w/v) SDS. Two dimensional PAGE was carried out by the method of O’Farrell (1975) with a slight modification: the purified enzyme was separated in the 1st dimension by discisoelectric focusing on a gel containing carrier Ampholines producing a pH gradient of 3 to 10 in the presence of 8 M-

Large Protease

987

urea, and run from anode to cathode; for separation in the 2nd dimension, materials were subjected to SDSPAGE in 15% gel after treatment with 2% SDS. Proteins were detected with Coomassie brilliant blue. The proteins used as molecular weight markers were: phosphorylase b (94,000); bovine serum albumin (67,000); aldolase (43,000); carbonic anhydrase (30,000); soybean trypsin inhibitor (20,100); and a-lactalbumin (14,400). Various acetylated cytochromes c with p1 values of 4.1, 4.9, 6.4 and 8.4 were used as p1 markers. (f) Cell-free

protein

synthesis

Cell-free protein synthesis was carried out with a rabbit reticulocyte lysate kit by the method of Pelham & Jackson (1976). Poly(A)-mRNA was prepared from rat liver by the method of Brawerman (1976). Assays were performed in a reaction mixture containing 25 ~1 of reticulocyte lysate, 0.9 pg of poly(A)-mRNA, 100 PCi of L-[3H]leucine, 140 mM-potassium acetate and 1.5 mMmagnesium acetate in a total volume of 35 ~1. After incubation for 90 min at 3O”C, the reaction was terminated by addition of 0.9 ml of ice-cold 10 mi%-Trisbuffer (pH 7.4) containing 2 mM-EDTA, 0.1 y0 SDS, 0.1 y0 (v/v) Triton X-100 and 1 mg of unlabeled leucine/ ml, and then the mixture was centrifuged at 130,000 g for 60 min. Immunoprecipitation of synthesized protease was carried out as follows. The 130,000 g supernatant was first treated with 10 pg of non-immunized immunoglobulin G and 10 pl of S. aureus cells to remove proteins with non-specific adsorption. The supernatant obtained by centrifugation was mixed with 10 pg of anti-protease immunoglobulin G (Tanaka et al., 1986a) for 1 h and then with 10 ~1 of S. aureus cells. The resulting precipitate was washed extensively and then subjected to SDS-PAGE. After electrophoresis, the gels were treated with Enlightling (New England Nuclear), dried and subjected to fluorography at - 70°C with DuPont Cronex film and a sensitizing screen. (g) Reverse-phase

HPLC

analysis

The purified enzyme, dialyzed against 0.05% (v/v) trifluoroacetic acid, was applied directly to a Cosmosil 5C4-300 column (10 mm x 250 mm) equilibrated with 0.05 y0 trifluoroacetic acid. Gradient elution was performed for 4 h with 45% to 60% (v/v) acetonitrile containing 0.05% trifluoroacetic acid in a Waters model 141 HPLC system at a flow rate of 1 ml/min, and fractions of 1 ml were collected. The 10 main fractions were collected and concentrated by vacuum centrifugation. Then the samples were rechromatographed for 1 h on a TSK-Gel phenyl-BPW RP reversed phase column (4.6 mm x 75 mm) by the same procedure as described above, except that a gradient of 40% to 50% acetonitrile was used. (h) Peptide mapping Peptide mapping of the purified enzyme was carried out by the 2-dimensional electrophoretic method of Bordier & Crettol-JPrvinen (1979). The multiple components of the enzyme were separated by SDS-PAGE (12.5% (w/v) gel) in the 1st dimension. After electrophoresis, the whole gel was equilibrated in stacking gel buffer and transferred to a second slab gel (15% gel). A solution of V8 protease (50 pg) or lysylendopeptidase (50 pg) was overlaid on the gel and peptides produced by partial proteolysis of the components were separated by SDS-PAGE in the 2nd dimension.

988

K. Tanaka et al.

For mapping of the isolated components, each sample was digested with lysylendopeptidase and then analyzed by the standard method of SDS-PAGE.

3. Results (a) Electron microscopic analysis We have reported preliminary results of electron microscopic analysis of the enzyme (Tanaka et al., 19866). More detailed data were obtained in this study. Figure 1 shows a photograph, at low magnification, of the enzyme negatively stained with many1 acetate. Enzyme molecules appear as ring-shaped particles with a uniform diameter of 160(+7) A (average + standard deviation; n = 108). The actual range of diameters was 140 A to 170 A. At higher magnification (Fig. 2), the size and relative location of the central hole was seen to vary. The diameter of the hole was usually 10 A to 20 A, and at most 30 A, though in some cases the hole was deformed and appeared as a slit on the surface (Fig. 2, second row). The hole was usually centrally located in the molecule (Fig. 2, top row), but sometimes it was not (Fig. 2, third row). Electron micrographs did not show whether the hole was a channel that passed through the molecule,

or a cavity

in the molecule.

We think

that

the hole was either a channel (with at least two accesses) or a dip (with a single access). (Our preliminary electron microscopic work on a similar liver showed ring-shaped enzyme from human molecules with a larger central hole that was definitely a channel.) But even if it was only a cavity within the molecule, it must have access to the solvent, since it was easily stained with the staining solution. A small fraction of the particles in Figure 1 were rectangular with four or five vertical stripes (Fig. 2, bottom row). Other investigators (Kopp et al., 1986), who observed similar structures by electron microscopy, claimed that they were side views of cylindrical molecules whose end-on views were ring-shaped, like those in Figure 1. But we think that the rectangular particles were dimers of the enzyme which were produced during staining with uranyl acetate, because the enzyme tended to aggregate in the presence of heavy metal ions. For example,

when

the enzyme

was stained

with

2%

(w/v) phosphotungstate at pH 7-0, the specimen observed by electron microscopy was extensively aggregated (data not shown). The enzyme formed a visible precipitate when mixed with 2% phosphotungstate.

Other

heavy

metal ions such as cadmium

and chromium also produced similar but much longer aggregates with stripes (data not shown). The height of the enzyme molecule was estimated from the photographs of platinum-shadowed

Figure 1. Electron micrograph of the high molecular weight protease from rat liver. The enzyme was negatively stained with 3% many1 acetate at pH 4.5 after being freed of residual oligomers by chromatography on a TSK G4OOOSW gel permeation column. The bar represents 1000 A. Most of the particles in the Figure are ring-shaped with a small hole in the center, but a few have a slit rather than a hole and a few are rectangular rather than spherical.

Gross Structure

of Mammalian

Large Protease

989

Figure 2. Electron micrographs of individual molecules at higher magnification. The top row shows the commonest “normal” type with a ring-like appearance. The second and third rows show molecules in which the hole is deformed or displaced from the center. The bottom row shows rectangular particles, which we consider to be dimers formed during staining with uranyl acetate. More than 80% of the molecules were of the “normal” type. The bar represents 200 A.

specimens. From the calibrated shadowing angle of 5.7” we estimated the height to be 110( + 10) A, assuming the thickness of the platinum coat to be about 10 t’o 20 A. The estimated height of the molecule supports our opinion that the rectangular forms of particles were not side views of the enzyme. (b) Small-angle

X-ray

scattering

analysis

Small-angle X-ray scattering was used to determine the molecular architecture of the enzyme in solution. Guinier plots of the scattering intensities exhibited excellent linearity in lower scattering angles, indicating monodispersion of an X-ray scatter in solution. The values of the radius of gyration, R,, and the maximum dimension, D,,,,,, of the enzyme were estimated to be 659( kO.4) A The D,,,,, value of the and 210( +5) A, respectively. enzyme is longer than that determined by electron microscopy because of hydration of the enzyme in aqueous solution. Electron micrographs indicate a hole in the center of the enzyme. Based upon these findings, simulation analysis was undertaken in reciprocal space to provide a further experimental

basis for elucidating the enzyme structure. The theoretical scattering profiles of an ellipsoid and an ellipsoidal cylinder with different axial ratios and degrees of hollowness (Hiragi & Ihara, 1981) were compared with the observed scattering intensities. Of the many kinds of models examined, the prolate ellipsoidal structure with an ellipsoidal cavity depicted in Figure 3 showed the best coincidence with the experimental data (Fig. 4). The molecular dimensions of the whole molecule and cavity were 57.5 A x 57.5 A x 115 A and 34.5 A x 23.0 A x 46.0 A, respectively. The R, and D,,,,, values of the model were 65-6 A and 230 A, respectively, which were comparable to the observed values (R, = 65*9( kO.4) A and D,,,,, = 210( f5) A). The Stokes radius, based on the ellipsoidal model, was 75.6 A. The molecular weight was empirically estimated as 750,000 from the model (Kumosinski & Passen, 1982). The structural model obtained from the simulation was supported by analysis in real space. As shown in Figure 5, the P(r) function calculated for this model was also in good agreement with the experimental data. Thus, the structural model deduced from X-ray analysis was consistent with the experimental data.

990

K. Tanaka et al.

Figure 3. Model of a high molecular weight multi-protease complex deduced by small-angle X-ray scattering analysis. The cavity in the center of the enzyme is shown by dotted lines. The dimensions of the model are shown in angstrom units, numbers in smaller type being those of the cavity in the center of the enzyme.

(c) Biochemical analysis The subunit composition of the enzyme complex was examined by biochemical analyses. On onedimensional gel electrophoresis under denaturing conditions, the purified enzyme was separated into multiple components with molecular weights of 22,996 to 33,900 (Fig. 6, left). Since these multiple components were similar in size, their number could not be determined accurately by SDS-PAGE only. Therefore, we examined them by two-dimensional electrophoretic analysis by the method of O’Farrell (1975). When the purified protease was subjected to isoelectric focusing after dissociation in 8 M-urea, the multiple components were distributed in a very broad range of p1, 4 to 9 (Fig. 6, top), suggesting that their amino acid compositions were different. SDS-PAGE in the second On subsequent dimension, about 15 separate spots were obtained (Fig. 6, bottom). Thus, two-dimensional electro-

phoretic analysis gave clear evidence that this large protease was composed of multiple components. It was, however, uncertain from the results whether these multiple components were produced by limited proteolysis during purification of the enzyme or whether they were true subunits of the enzyme. To distinguish between these two possibilities, we examined the homologies of these multiple components by the two-dimensional electrophoretic peptide mapping method of Bordier & CrettolJaarvinen (1979). This technique is useful for peptide mapping of heterogeneous proteins, because limited proteolysis can be performed in the stacking gel during electrophoresis in the second dimension. As shown in Figure 7, when the multiple components separated by SDS-PAGE in the first dimension were digested with lysylendopeptidase (a) or V8 protease (b), their hydrolytic patterns

0

;: Q

-2.4

-3.c 0 Log (SXR,)

Figure 4. Observed scattering profile (0) of the enzyme compared with the theoretical scattering function of the model shown in Fig. 3 (--).

r(a) Figure 5. Pair-distance distribution function, P(r), calculated from the observed scattering profile (-) compared with that for the model shown in Pig. 3 (----).

Gross Structure of Mammalian

Large Protease

991

PI- marker

4.9 4.1 I I

6.4 I

14Figure 6. Pattern on Z-dimensional polyacrylamide gel electrophoresis of high molecular weight protease. Twodimensional PAGE was performed as described in Materials and Methods. The l-dimensional patterns of the purified enzyme on isoelectric focusing and SDS-PAGE are shown at the top and on the left, respectively. Samples of 50 pg of enzyme were used. Proteins were stained with Coomassie brilliant blue.

apparently distinct, suggesting that these components differ in primary structure. These result’s support the idea that the multiple components of the protease are not products of limited proteolysis. For confirmation of this idea, we examined the translational products of mRNA encoding the protease subunits. The poly(A)mRNA isolated from total RNA of rat liver was translated in a reticulocyte lysate system, and the protease components synthesized were isolated by immunoprecipitation and analyzed by SDS-PAGE. Figure 8 is a fluorogram of the translated products labeled with [3H]leucine after reaction with antiprotease immunoglobulin G. Multiple bands were observed in a region corresponding to a molecular

14-

Figure 8. SDS-PAGE of high molecular weight of rat protease synthesized in vitro with poly(A)-mRNA liver. The procedures for cell-free protein synthesis in a reticulocyte lysate and immunological detection of protease in translation products are described in Materials and Methods. Bars on the left show positions of molecular weight markers ( x 10”).

were

weight of 30,000, which was closely similar to the size of protease components, strongly suggesting that these multiple components were intact subunits of the enzyme. The intensities of individual components in the fluorogram were somewhat different from their staining intensities with dye. This may be due to differences in their leucine contents, the precursors used for in-vitro synthesis or their staining intensities with dye. Another possibility is that the newly synthesized components could not be assembled into a multimerit complex and so the polyclonal antibody

W’

94 67;

430

ZL 8

30-

20-

14-

Lysylendopeptldose (a

I

V8-

protease (bl

Figure 7. Two-dimensional electrophoretic patterns of the products of high molecular weight protease digested with lysylendopeptidese and V8 protease. The experimental procedure is described in Materials and Methods. The upper panel shows the SDS-PAGE pattern of the purified enzyme in the 1st dimension, which was from left (cathode) t’o right (anode). The lower panel shows the SDS-PAGE patterns after proteolysis with lysylendopeptidase (a) and 9. aureus V8 protease (b) in the 2nd dimension. Samples of 50 pg of enzyme were used. Proteins were stained with Coomassie brilliant blue.

992

K. Tanaka et al.

I

10

I 90

120

I53 Retention

Cl

(bl

c2

c3

c4

210

1.90

c5

240

300

time (min)

C6

c7

cs

c9

Cl0

94 --**r 67 --r,

14 -“I,

Figure 9. Separation of multiple components from the purified enzyme by HPLC and the electrophoretic pattern of the components in the 10 main peaks. Chromatography on a Cosmosil 5C4-300 reversed phase column (a) and electrophoresis of the individual components separated by HPLC (b) were carried out as described in Materials and Methods. Ten fractions, named Cl to ClO, were collected for subsequent analyses. Portions (05% of total pooled fractions) of the 10 fractions were lyophilized and then subjected to SDS-PAGE. Cl to Cl0 correspond to components 1 to 10 in the upper chromatogram. Proteins were stained with Coomassie brilliant blue. Acetonitrile concentration (-.-.-); absorbance at 280 nm (----) and 215 nm (-).

against the native enzyme might not have reacted equally with all the free components. Since the multiple components of the enzyme were not products of limited proteolysis, the results shown in Figure 7 indicate that the enzyme has non-identical subunits. To obtain more direct evidence for the non-identity of the subunits, we next attempted to isolate these components by reverse-phase HPLC. For this, the purified enzyme was applied directly to a Cosmosil 5C4-300 column equilibrated with 0.05 o/o trifluroacetic acid, and bound proteins were eluted with a linear gradient of 45% to 60°% acetonitrile. As shown in Figure 9(a), when the elution profile was monitored at 280 nm and 215 nm, ten major components were found to be separated reproducibly, and these were named component 1 (Cl) to component 10 (ClO), in order of their elution. The 280 nm/215 nm adsorption ratio of the components differed, suggesting differences in their amino acid compositions. Figure 9(b) shows that on SDS-PAGE, Cl, 2, 3, 4, 5, 8 and 9 each gave only one band, whereas C6, 7

and 10 gave two bands of components of different sizes. When samples of components 6, 7 and 10 were subjected to HPLC again on a different type of reverse-phase column, TSK Gel phenyl-5PW RP, under the conditions described in Materials and Methods, they each gave two clearly separated components (data not shown). The isolated components of the enzyme were subjected to peptide mapping (Fig. 10). The ten major cornponents eluted from a Cosmosil 5C4-300 column were each treated with lysylendopeptidase and the resulting limited digestion products were analyzed by SDS-PAGE. The electrophoretic patterns of the various

components

were

significantly

different,

although some common bands were observed. This finding also indicates that these components are non-identical polypeptides. Larger materials were observed on SDS-PAGE of C5 and C7, since these samples were not readily soluble after lyophilization. In this experiment, fractions 6, 7 and 10 were used as mixtures of two components, and peptide mapping of the individual components in each

Gross Structure of Mammalian

Cl

c5

c4

c3

$2

993

Large Protease C6

c7

CB _;-,

cs _“I _” . ,,-..

Cl0 __ .

9467-

14-

Figure 10. Peptide mapping of the individual components of high molecular weight protease after treatment with lysylendopeptidase. Samples (0.5% of total pooled fractions) obtained by HPLC were lyophilized and incubated with 100 ng of lysylendopeptidase in 50 mM-Tris. HCl (pH 9.0) containing 4 M-urea at 37°C for 2 h. After incubation, proteins were precipitated by addition of acetone and collected by centrifugation. The resulting precipitates were subjected to SDS-PAGE and materials were located by silver staining.

fraction indicated their heterogeneity also (data not shown). These results strongly indicate that the high molecular weight protease is an unusual enzyme complex consisting of multiple non-identical subunits.

4. Discussion (a) Molecular size and shape of the protease complex In the present study the gross structure of a high molecular weight multi-protease complex of rat liver was analyzed by electron microscopic and

small-angle X-ray scattering techniques, and the results are summarized in Table 1. Electron microscopic analyses showed that this protease complex was an almost symmetrical ring-shaped particle with

a small

hole in the center.

The diameter

and

height of the particle were 160( 17) A and 1lO( + 10) A, respectively, and the diameter of the hole was about 10 to 30 A. Similar morphology of the enzyme from rat skeletal muscle has been reported by Kopp et al. (1986). On the other hand, a prolate-ellipsoidal model with

a similar

molecule fitted

prolate-ellipsoidal

the data obtained

cavity

in

the

by small-angle

X-ray scattering. The molecular dimensions of the whole molecule and the cavity were 57.5 A x 57.5 A x 115 A and 34.5 A x 23.0 b x 46.0 A, respectively. The P(r) function based on this model was also in excellent agreement with the experimental data (Fig. 5). The molecular volume of the enzyme was estimated to be 1,440,OOOA3, and the molecular weight was calculated to be 750,000, using the empirical relation of the molecular weight with the molecular volume (Kumosinski & Passen, 1982), which was almost equal to the value determined by the methods of sedimentation equilibrium, low-angle laser light-scattering, and sedimentation velocity and diffusion (Tanaka et al., 19866). The radius of gyration of the protein was determined to be 65.9 A, which provided a Stokes radius of 75.6 A according to the Stokes equation extended to ellipsoids. This value was also compatible with a value of 85 A obtained from the quasi-elastic light-scattering in consideration of hydration around the molecule (Tanaka et al., 1986b). Although the molecular shape observed by electron micrography was somewhat different from that estimated from the data obtained by smallangle X-ray scattering, the molecular dimensions determined by the two methods were very similar,

K. Tanaka

994

et al.

Table 1 Summary i$pecies Whole molecule

of size and shape of the high molecular

of parameters Parameter

Value

Dimensions Molecular

weight

Radius of gyration Stokes radius Cavity

Dimensions

160( + 7) A (diameter) 1 lO( + 10) A (height) 57.5 A x 57.5 A x 115 A 750,000 743,OOOt 722,ooot 760,OOOt 65.9 A 75.6 A 85.0 “it lo-30 A (diameter) 34.5 A x 23.0 A x 46.0 A

weight multi-protease complex Method Electron

microscopy

Small-angle X-ray scattering Small-angle X-ray scattering Sedimentation equilibrium Sedimentation velocity and diffusion Low-angle laser light scattering Small-angle X-ray scattering Small-angle X-ray scattering Quasie-elastic light scattering Electron microscopy Small-angle X-ray scattering

t Data from Tanaka et al. (1986b).

and in particular, the existence of a hole or cavity was clearly indicated by both. The protein molecule may have been somewhat deformed during preparation of samples for electron microscopy, in which case part of the cavity was presumably accessible to solvent, resulting in penetration of the stain into the cavity, and in an apparently “ring-shaped” molecule. Based on these considerations, we propose the following model for the protease complex. The enzyme complex is a somewhat ellipsoidal molecule with diameters of about 100 A and 200 A and a height of about 100 A. An ellipsoidal cavity is present in the center of the complex, and is connected with the surface of the molecule by at least two appertures. The 13 to 15 types of subunits must be arranged around the cavity. Recently Kopp et al. (1986) proposed a cylindershaped model composed of four rings or discs for the same multi-catalytic protease from rat skeletal muscle. Their electron microscopic data are similar to ours, but from repeated and detailed analyses of the protease negatively stained with phosphotungstate we think the cylinder-like shape seen in electron micrographs is an artifact. Moreover, the cylinder-shaped model composed of two or more rings or disks could not be simulated at all from the data obtained by small-angle X-ray scattering. It is thus quite likely that the protease complex is a “hollow, egg-shaped” molecule formed by 13 to 15 types of subunits. We propose that this enzyme complex should be which means miniparticle named “proteasome”, (-some) with protease activity. (b) Multiplicity

of the subunit

composition

Previously (Tanaka et al., 1986b), we reported the physiochemical properties of this high molecular weight protease and demonstrated that the purified enzyme was homogeneous and had an unusually large molecular weight of 750,000. In this work, we obtained the following evidence that the enzyme is a multi-subunit complex consisting of at least 13 to 15 non-identical polypeptides. First, the products of cell-free translation showed similar multiplicity on

reaction with anti-protease immunoglobulin G that the mRNAs encode (Fig. 81, indicating multiple components of similar sizes to those of the purified enzyme. Second, we observed heterogeneity in the number and nature of the peptides of these components by peptide mapping analysis after limited proteolysis (Figs 7 and 10). These findings are compatible with the idea that the high molecular weight protease is a complex of multiple non-identical subunits. In the present work we separated 13 components of the purified protease. These may not have been all the components of the enzyme, because some may have remained bound so tightly to the column that they were not eluted with the acetonitrile gradient. Furthermore, other minor some components may be involved in the complex, because several faint spots were also detected on analysis by two-dimensional PAGE (Fig. 6). However, the number of components separated by two-dimensional PAGE was in fair accord with the number isolated by HPLC, suggesting that the 13 components isolated are the main subunits of the enzyme. Since the intensities of staining of the various individual components on an electrophoretic gel and their amounts eluted by HPLC varied greatly, the enzyme must consist of multiple copies of these components, although the exact number of copies of each component is still unknown. The idea of the presence of multiple copies is supported by the fact that the sum of the molecular weights of all the isolated components, calculated by comparison of their mobilities on SDS-PAGE with those of standard proteins, was 340,000, which is only half the value of the native enzyme. So far only the pattern on SDS-PAGE has been used to show the subunit multiplicity of the enzyme (Wilk & Orlowski, 1980; Rivet& 1985; Ray & Harris, 1985; Dahlmann et al., 1985; Tanaka et al., 1986a, 1988b). This is the first report of purification and partial characterization of the multiple components of high molecular weight protease. The molecular assembly of this multi-component complex is of interest. A disulfide bond does not

Gross Structure of Mammalian seem to participate in the subunit interaction, because these multiple bands on SDS-PAGE were not affected by the presence of a sulfhydryl reagent or 2-mercaptoethanol such as dithiothreitol (Tanaka et al., 1986a). An interesting finding was that the subunits were of similar sizes, but had strikingly different isoelectric points, which ranged from p1 4 to 9 (Fig. 6). The enzyme seems to be a symmetrical molecule with a small cavity in the center, and the similarity in subunit size may be important for construction of a symmetrical complex. On the other hand, the heterogeneity in the net charges of the subunits is presumably related to assembly of the multi-subunit complex: interaction charge may be complementary important for the molecular organization of the enzyme. Further studies are necessary to elucidate the molecular organization of the components in the enzyme molecule. (c) Functional signijkance of properties of the high molecular weight multi-protease complex Previous studies by us (Tanaka et al., 1986a, 198%~) and others (Wilk & Orlowski, 1983; Dahlmann et al., 1985) demonstrated that this high molecular weight protease has three catalytic sites, which are distinguishable by their substrate specificities and sensitivities to inhibitors. These active sites are presumably on distinct polypeptides, because the molecular weights of the subunits are similar to those of typical proteases such as trypsin, chymotrypsin and papain, which each consists of a single polypeptide. In addition, we showed that many, but not all the subunits, were labeled with [3H]diisopropyl fluorophosphate (Tanaka et al., 1986a), which is a specific inhibitor of enzymes with a seryl residue in their active site. Thus, some of the polypeptides in this complex have active sites containing a serine residue(s), and so this large enzyme is apparently a multi-protease complex. This multi-catalytic protease may be well organized for rapid degradation of various proteins; rapid degradation must be important physiologically, because intracellular proteolysis is so fast that no hydrolytic intermediates are detectable in cells. Another interesting feature of this protease is that it is present in a latent form in cells. Its latency may be essential to protect cells from autolytic damage, and is presumably related to regulation of intracellular protein breakdown. Probably recognition of proteins for degradation is coupled with activation of the latent protease, and this activation is the initiation step of intracellular protein degradation. However, the intracellular mechanism controlling the enzyme activity is unknown, although the latent enzyme can be activated in vitro by various treatments that induce conformational change of the enzyme (Tanaka et al., 1986a, 1988a). Thus, the two properties, latency and multiple catalytic actions, of this unusual large protease may be important’ for its physiological function.

Large Protease

995

(d) Structural similarity of ring-shaped 20 S particles widely distributed in a variety of eukaryotic cells In the present and previous work, we showed that the high molecular weight protease from rat liver is an unusually large molecule with a sedimentation coefficient of 20 S and a molecular weight of 750,000, and that the molecule has a cavity giving a ring-shaped appearance by electron microscopy, and is composed of approximately 13 to 15 distinct polypeptides with similar sizes of 22,000 to 32,000 but different pT values of 4 to 9. Some of these properties of this enzyme, and similar types of protease from different tissues, have already been reported (Tanaka et al., 1986a,b, 1988a,b; Wilk & Orlowski, 1980; Rivett, 1985; Ray & Harris, 1985; Dahlmann et al., 1985; McGuire & DeMartino, 1986; Ishiura et al., 1986). In the past 15 years, there have been many reports of the presence of subcellular particles with similar structures to that of the high molecular weight protease in a variety of eukaryotic organisms ranging from Drosophila to man: for example, there are reports of miniature ring-shaped particles released from malignant cultured cells into the medium (Narayan & Rounds, 1973); of welldefined ring-shaped and rectangular particles in the post-ribosomal cytosol of HeLa cells (Smulson, 1974); of cylindrical particles with a hole named “cylindrin” in the plasma membrane (Harris & Naeem, 1981) and cytosol (Malech & Marchesi, 1981) of human erythrocytes; of donut-shaped “miniparticles” in nuclei of human and rat cells (Domae et al., 1982); and of 22 S cylindrical particles in the nucleus and cytoplasm of Xenopus laevis (Kleinschmidt et al., 1983). All these particles were found by electron microscopy, but their functions were not examined. In contrast, Martins de Sa et al. (1986) purified ring-shaped particles with a complex polypeptide composition, and demonstrated that they were ribonucleoprotein particles named “prosomes”. Prosomes are probably associated with cytoplasmic mRNAs and play a regulatory role in the translation process of protein synthesis (Schmid et al., 1984). Prosomes have been found in a variety of cells, including mouse, duck and HeLa cells (Martins de Sa et al., 1986) and also sea urchin cells (Akhayat et al., 1987). Interestingly, in Drosophila, small heat shock proteins of 22,000 to 28,000 are associated with 19 S ring-type particles (Schuldt & Kloetzel, 1985) and these 19 S particles are thought to be identical to prosomes (Arrigo et al., 1985). Recently, pretRNA 5’ processing nuclease has also been purified from X. Zaevis and HeLa cells as hollow, cylindrical particles consisting of at least 14 polypeptides with molecular weights of about 20,000 to 32,000 (CBstano et al., 1986). In addition, a similar type of multi-subunit complex has been found in macrophages as particles co-precipitated with murine histocompatibility complex antibody (Monaco & McDevitt, 1982). Thus, a number of eukaryotic cells

996 contain

K. Tanaka

large

particles

that

resemble

the

high

molecular weight multi-protease complex in size, shape and subunit multiplicity. However, it is unknown whether all these particles are identical, because they have been isolated from different cells and tissues and in different ways. Moreover, since the biological functions of most of these particles are unknown, it is very hard to determine their relationship

to each other.

More

detailed

such as analyses of their gene structure, for resolution of this problem.

studies,

are required

This work was supported in part by grants from the Ministry of Education, Science and Culture of Japan and the Foundation for Applied Enzymes, Osaka.

References Akhayat, O., Grossi de Sa, F. & Infante, A. A. (1987). Proc. Nat. Acad. Sci., U.S.A. 84, 1595-1599. Arrigo, A.-P., Darlix, J.-L., Khandjian, E. W., Simon, M. & Spahr, P.-F. (1985). EMBO. J. 4, 399-406. Barrett, A. J. (1977). Editor of Proteinases in Mammalian North-Holland Pub. Co., Cells and Tissues, Amsterdam. Blundell, T. L. & Johnson, L. pu’. (1976). Protein Crystallography, pp. 318319, Academic Press, London. Bordier, C. & Crettol-Jarvinen, A. (1979). J. Biol. Chem. 254, 2565-2567. Brawerman, G. (1976). Methods Enzymol. 30, 605-761. CBstano, J. G., Ornberg, R., Koster, J. G., Tobian, J. A. & Zasloff, M. (1986). CeZZ,46, 3777387. Dahlmann, B., Kuehn, L., Rutschmann, M. & Reinauer, H. (1985). Biochem. J. 228, 161-170. De Martino, G. N. & Goldberg, A. L. (1979). J. Biol. Chem. 254, 3712-3715. Domae, N., Harmon, F. R., Busch, R. K., Spohn, W., Subrahmanyam, C. S. & Busch, H. (1982). Life Sci. 30, 469-477. Ginsburg, A & Stadtman, E. R. (1970). Annu. Rev. Biochem. 39, 429-472. Glatter, 0. (1982). In Small Angle X-ray Scattering (Glatter, 0. & Kratky, 0.: eds), pp. 1677196, Academic Press, New York. Guinier, A. & Fournet, G. (1955) In Small-Angle Scattering of X-Rays, p. 126, John Wiley and Sons, New York. Harris, J. R. & Naeem, I. (1981). Biochim. Biophys. Acta, 670, 285-290. Edited

et al. Hiragi, Y. & Ihara, S. (1981). Acta Crystallogr. sect. A, 37. 378-382. Hough, R., Prott, G. & Rechsteiner, M. (1987). J. Biol. Chem. 262, 8303-8313. Ishiura, S., Yamamoto, T.. Nojima, M. & Sugita, H. (1986). Biochim. Biophys. Acta, 882, 305-310. Kleinschmidt, J. A., Hugle, B., Grund, C. & Franke, W. UT. (1983). Eur. J. Cell Biol. 32, 1433156. Kopp, F., Steiner, R., Dahlmann, B.? Kuehn, L. & Reinauer, H. (1986). Biochim. Biophys. Acta, 872, 2533260. Kumosinski, T. F. & Passen, H. (1982). Arch. Biochem. Biophys. 219, 89-100. Laemmli, U. K. (1970). Nature (London), 270, 6809685. Malech, H. L. & Marchesi, V. T. (1981). Biochim. Biophys. Acta, 670, 385-392. Martins de Sa, C., Grossi de Sa, M.-F., Akhayat, O., Broders, F., Scherrer, K., Horsch, A. & Schmid. H.-P. (1986). J. Mol. Chem. 187, 479-493. McGuire, M. J. & DeMartino, G. N, (1986). Biochim. Biophys. Acta, 873, 279-289. Monaco, ,J. J. & McDevitt, H. 0. (1982). Proc. Nat. Acad. sci. U.S.A 79 > 3OOl3005. Namyan, K. S. & Rounds, D. E. (1973). Nature New Biol. 243. 1466150. O’Farrell, P. H. (1975). J. Biol. Chem. 250, 4007-4021. Pelham. H. R. B. & Jackson, R. A. (1976). Eur. J. Biochem. 67, 247-256. Ray-. K. & Harris, H. (1985). Proc. ‘Vat. Acad. Sci., U.S.A. 82, 7545-7549. Rivrtt, A. J. (1985). J. Biol. Chem. 260, 1260612606. Rose, I. A., Warms, J. V. B. & Hershko. A. (1979). J. Biol. Chewt. 254, 81358138. Schmid, H. P.. Akhayat, O., Martins de Sa, C., Puvion, F., Koehler, K. & Scherrer, K. (1984). EMBO J. 3, 29-34. Schuldt. C. & Kloetzel, P.-M. (1985). Develop. Hiol. 110, 65574. Smulson, M. (1974). Exp. Cell Res. 87, 2533258. Tanaka, K., Ii, K., Ichihara, A., Waxman, L. & Goldberg, A. L. (1986a). J. BioE. Chem. 261. 1519715203. Tanaka, K., Yoshimura, T., Ichihara, A., Kameyama, K. & Takagi, T. (19866). ,I. Biol. Chem. 261. 1520415207. Tanaka, K.. Waxman, L., Tawa, N. & Goldberg, A. L. (1988a). J. Biol. Chem. in the press. Tanaka. K., Waxman, I,., Boches, F. S. & Goldberg, A. 1~. (1988b). J. Viol. Chem. in the press. Wilk, S. & Orlowski, M. (1980). J. Neurochem. 35. 1172m 1182. Wilk, S. & Orlowski, M. (1983). J. Neurochem. 40, 842.. 849.

by A Persht