Ubiquitin Carboxyl-terminal Hydrolase Acts on Ubiquitin Carboxyl-terminal Amides*

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society of Biological Chemists, Inc. Vol. 260, No. 13, Issue of July 5, pp, , 1985 Printed in U.S.A. Ubiquitin Carboxyl-terminal Hydrolase Acts on Ubiquitin Carboxyl-terminal Amides* (Received for publication, February 19, 1985) Cecile M. Pickart and Irwin A. Rose From the Institute for Cancer Research, Fox Chase Cancer Center, Philodelphia, Pennsylvania 191 I1 Ubiquitin carboxyl-terminal hydrolase (formerly known as ubiquitin carboxyl-terminal esterase), from rabbit reticulocytes, has been shown to hydrolyze thiol esters formed between the ubiquitin carboxyl terminus and small thiols (e.g. glutathione), as well as free ubiquitin adenylate (Rose, I. A., and Warms, J. V. B. (1983) Biochemistry 22, ). We now show that this enzyme hydrolyzes amide derivatives of the ubiquitin carboxyl terminus, including those of lysine (6-amino), glycine methyl ester, and spermidine. It also (Reticulocytes possess multiple E2s, only one of which funchydrolyzes ubiquitin COOH-terminal hydroxamic tions in breakdown of serum albumin; Ref. 5.) Conjugation of acid, but is inactivated under the conditions for assay- ubiquitin to t-amino groups of target protein lysines is known ing ubiquitin-hydroxylamine adduct hydrolysis. to occur (2); conjugation to the target protein NH, terminus Amide adducts formed between ubiquitin and c-amino may also occur (although it has not been reported), since a groups of protein lysine residues are much poorer subfree target protein NH, terminus is required for ubiquitinstrates than is the ubiquitin amide of the t-amino group of free lysine. The enzyme is thus a general hydrolase dependent proteolysis (6). The target protein portion of these that recognizes the ubiquitin moiety, but is highly seconjugates is degraded by other enzymes in a ubiquitin-spelective for small ubiquitin derivatives. Itprobably cific and ATP-dependent fashion (7). functions to regenerate ubiquitin from adventitiously Ubiquitin acts catalytically in the degradative pathway (2, formed ubiquitin amides and thiol esters. It also has 8) and must therefore be regenerated. Enzymes that catalyze the correct specificity to function in regenerating ubiq- ubiquitin release from large (multiply ubiquitinated) conjuuitin from small ubiquitin peptides that are probable gates are present in reticulocyte extracts (2, 7-9) and may end products of ubiquitin-dependent proteolysis. A simple, large-scale preparation of the enzyme from human erythrocytes is described. Reticulocytes and other eukaryotic cells possess an ATPdependent proteolytic pathway in which the small polypeptide ubiquitin serves as a cofactor (1). During ubiquitin-dependent protein breakdown, ubiquitin becomes covalently attached to target protein substrates, forming conjugates in which the linkage is through the ubiquitin carboxyl terminus and has the stability expected of an amide (2). Conjugate formation is a three-step process (3) in which ubiquitin is successively activated to a thiol ester of ubiquitin-activating enzyme (E1;l Ref. 4), then transferred to a thiol group of a small ubiquitin (Ub) carrier protein (E2), and finally transferred to a protein * This work was supported by American Cancer Society Grant BC- 414 (I. A. R.), National Institutes of Health Fellowship GM (C. M. P. and I. A. R.), CA and RR (I. C. R.), and also supported by an appropriation from the Commonwealth of Pennsylvania. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: El, ubiquitin activating enzyme; E2, ubiquitin carrier protein; E3, ubiquitin-protein ligase; DTT, dithiothreitol; Lys-ubiquitin, ubiquitin amide of the c-amino group of lysine; GME-ubiquitin, ubiquitin amide of the a-amino group of glycine methyl ester; Dm-ubiquitin, ubiquitin thiol ester of DTT; PPase, inorganic pyrophosphatase; SDS, sodium dodecyl sulfate; AMP-Ub, Ub adenylate amino group by a ligating enzyme (E3) (Equations 1-3). M%+ El + 2ATP + 2Ub 7 + AMP + 2PPi (1) El-S-Ub + E2 E2-S-Ub + El (2) E3 E2-S-Ub + protein-nhz Ub-NH-protein + E2 (3) function in ubiquitin regeneration during protein breakdown. (These isopeptidases could also reverse ubiquitination that serves a more general purpose of protein modification, by analogy with the case of the nuclear ubiquitin-h2a conjugate (9, lo).) However, the cell must also be able to regenerate ubiquitin from small ubiquitin peptide or amino acid adducts (e.g. Lys-ubiquitin) that are probable end products of ubiquitin-dependent proteolysis. Small ubiquitin amides will also be formed adventitiously as a result of ubiquitin transfer to small amines (in particular, polyamines) catalyzed by E2-ubiquitin thiol esters (5). Rose and Warms suggested (11) that small ubiquitin amides might be substrates for ubiquitin carboxyl-terminal esterase, an abundant enzyme from reticulocytes that hydrolyzes ubi- quitin esters of small thiols. The observation that simple primary amines act as acceptors of ubiquitin from E2-ubiquitin thiol esters (5) has now made a test of this suggestion possible. Our finding that ubiquitin amides are substrates, reported here, has prompted us to change the name of the enzyme from esterase to hydrolase. It is referred to as ubiquitin carboxyl-terminal hydrolase throughout this paper. MATERIALS AND METHODS Most commercially available reagents and enzymes were obtained from previously described sources (5). Hydroxylamine hydrochloride (Aldrich) was crystallized twice from ethanol/water (7/1). The prep- arations of [Y-~*P]ATP, ubiquitin, %I-ubiquitin, ubiquitin-sepharose affinity resin (7 mgof Ub/ml of resin), and fraction I1 (soluble reticulocyte lysate components retained on DEAE-cellulose) have been described (3, 5). %I-ubiquitin conjugates of native lysozyme, prepared as previously described (7), were a gift of Dr. Avram Hershko of the Technion, Haifa, Israel.

2 7904 Specificity of Ubiquitin Carboxyl-terminal Hydrolase Preparation of El and E2s2 E l (AMP/PPi eluate) and E2 (DTT eluate) were prepared, and their concentrations determined (by PPi burst), as described previously (5). Purified E2 protein 3 was obtained by Sephacryl 200 fractionation of the DTTeluate (5). Preparations of E l and E2 were freed of traces of ubiquitin-carboxyl-terminalhydrolase by passage through a 0.3-ml ubiquitin-sepharose column in the absence of ATP. Under these conditions, E l and E2 break through the column (3), while hydrolase binds (11).The amountsof E l and E2 obtained from Fraction I1 by affinity chromatography correspond to concentrations of 1 and 2 nmol, respectively, per 12 ml of packed reticulocytes. /C-kC" 116 -' Preparation of '26Z-UbiquitinConjugates of Yeast Cytochrome c These were prepared by lsi-ubiquitin transfer catalyzed by the ubiquitin thiol ester form of E2 protein 3, as described in Fig. 9 of Ref. 5 (reaction volume scaled up 8-fold). After 100 min of reaction, KC1 (30 mm) was added, and the mixture was applied to a 0.2-ml carboxymethyl-cellulose column equilibrated with 50 mm Na acetate, ph 7, containing 30 mm KCI. The column was washed with 1 ml of the same buffer; the breakthrough fraction contained 90% of the unconjugated ubiquitin. Cytochrome c-ubiquitin was eluted with 2.5 ml of 50 mm Na acetate, ph 7, containing 0.2 M KCI. This eluate was dialyzed and concentrated (Amicon YM-5 membrane), and characterized by SDS-gel electrophoresis and autoradiography. 12'I-ubiquitin was distributedamong the species free ubiquitin (8%), apparent mono-ubiquitin conjugate of cytochrome c (84%), andapparent triubiquitin conjugate (8%) (see Fig. 9 of Ref. 5). Recovery of conjugated ubiquitin was >go%. Conjugated ubiquitin was estimated to be 0.2 p ~based, on radioactivity; the preparation also contained free cytochrome c (estimated -0.17mM). The cytochrome c conjugates are expected to be lysine adducts, based on 1)the presence of 16 lysine residues and NHZ-terminal threonine in yeast cytochrome c (12), and 2) the previous observation that E2-ubiquitin thiol esters transfer ubiquitin only to amino groups on primary carbons (5). Small-scale Preparation of Ubiquitin Carboxyl-terminalHydrolase This enzyme was purified to 295% homogeneity (judged by SDSgel electrophoresis and silver staining) by a procedure based on its known affinity for ubiquitin andits low molecular weight (11). Fraction I1 (100 mg total protein) was made 0.5mM in EDTA and passed through a 1-ml ubiquitin-sepharose column equilibrated with 50 mm Tris.HCI (10% base), 0.1 mm EDTA, and 0.2 mm dithiothreitol, ph 7.2 (standard buffer). After washing with 5 ml each of 1 M KC1 (in standard buffer) and standardbuffer, remaining ubiquitinbound proteins were eluted with 10 ml of 50 mm Tris.HC1 (90% base), 0.1 mm EDTA, and 10 mm dithiothreitol (ph 9 eluate). (The column was run a t room temperature, but fractions were collected on ice.) The ph 9 eluate was dialyzed against standard buffer and concentrated to about 0.1ml (Amicon cone, CF 25). The major proteins of the ph 9 eluate had subunit molecular weights of 98,000 and 29,000 (not shown). These proteins were resolved on a Sephacryl 200 column (1 X 48 cm; standard buffer, 5 "C), using as starting material the combined ph 9 eluates from 400 mg offraction I1 protein. Fractions of0.7 ml were collected and aliquots were examined for protein composition by SDS-gel electrophoresis. Appropriate fractions were pooled and concentrated, yielding preparations enriched in the 98-kDa protein and in the29-kda protein (Fig. 1,lunes 1 and 2, respectively). The 29-kDa protein was judged to be carboxylterminal hydrolase on the basis of its molecular weight and high specific activity in DTT-ubiquitinhydrolysis, 20 units/mg. (One unit of enzyme hydrolyzes 1 pmol of DTT-ubiquitin/min a t p M DTT-ubiquitin; Ref. 11.) This activity is 20-fold greater than that reported previously ( l l ), presumably reflecting the greater purity of the present preparation. One pass through a ubiquitin-sepharose column as described removed 80-90% of the hydrolase activity from fraction 11; recovery of activity in the concentrated Sephacryl 200 peak, relative to the startingph 9 eluate, was about 20%. The enzyme was quite stable to storage a t -60 "C but tended to lose activity on repeated thawing and freezing. For simplicity, the term "E2" is used to refer to the mixture of ubiquitin carrier proteins obtained in the DTT eluate ( 5 ) (unless otherwise indicated). The thiol ester forms of a t least four of the five E2s catalyze ubiquitin transfer to small amines (5). 44 -I 30- -,... - v FIG. 1. SDS-polyacrylamide gel(silver stained) of concentrated Sephacryl-200 column fractions. Lune I, high-molecular weight fraction (3.2 pg ofprotein); lane 2, carboxyl-terminal hydrolase (30-kDa fraction, 1.6 pg of protein). Molecular masses (kda) of standard proteins run on the same gel are shown a t the left. Large-scale Preparation of Carboxyl-terminalHydrolase from Human Erythrocytes Buffers-Compositions were as follows: buffer 1, phosphate-buffered saline (0.154 M NaCI, 25 mm potassium phosphate, ph 7.4); buffer 2, 3 mm potassium phosphate, 1 mm DTT, ph 7.0; buffer 2A, buffer 2 with 20 mm KCI; buffer 3, 20mM Tris.HCI (10% base), 0.5 M KCI, 1 mm DTT, ph 7.2 (25 "C); buffer 4,50 mm TriseHCI (90% base), 0.1 mm EDTA, 5 mm DTT, ph 9 (25 "C); standard buffer ("Materials and Methods"). The following steps, except affinity chromatography, were performed a t 4 "C. Hemolysate-Outdated human blood was kindly provided by the American Red Cross. The packed cells (400 ml) were washed once with 5 volumes buffer 1and lysed by addition of 3 volumes of 1 mm DTT. After 15 min, the lysate was centrifuged a t 9000 X g for 40 min. The top 75% of the material in each bottle was taken (hemolysate, 930 ml). DE52 Adsorption and Elution-To the hemolysate was added 47 ml of a 50% (v/v) suspension of DE-52 previously equilibrated with buffer 2, and the resulting suspension was stirred for 1 h. After an additional hour, the suspension was filtered through sintered glass. The cellulose was washed with buffer 2A (500 ml) to remove residual bound hemoglobin, and transferred to a beaker containing 100 mlof buffer 3 for elution of the enzyme. The suspension was stirred for 1 h before filtering through two pieces of Whatman No. 1 paper (Buchner funnel).the volume of the filtrate (DE52 KC1 eluate) was 93 ml. Trichloroacetic Acid Precipitation-Solid DTT was added to the KC1 eluate to provide an additional 1 mm concentration. Trichloroacetic acid solution (50% w/v) was added, with stirring, to a final concentration of 5%.Protein was pelleted by centrifugation a t 12,000 X g for 6 min. The pellet was suspended in 25 ml of buffer 4, using a

3 glass rod, and the ph adjusted to -7 with 1 N NaOH. After homogenization with a Dounce, residual insoluble protein was removed by centrifugation (as above) and discarded. Affinity Chromatography-The supernatant from the trichloroacetic acid precipitation step was warmed to room temperature and applied to a 5.5-ml ubiquitin-sepharose column previously equilibrated with standard buffer. The breakthrough was saved and assayed for hydrolase activity. The column was washed successively with (a) 8 ml of standard buffer, (b) 30 ml of standard buffer containing 0.5 M KCl, and (c) 17 ml of standard buffer. Residual ubiquitin-bound proteins were then eluted with 30 ml of buffer 4; this fraction was collected, on ice, into 3 ml of 1 M Tris.HC1 (10% base, ph 7.2). It was concentrated to -0.7 ml using an Amicon ultrafiltration cone (CF 25), diluted 10-fold with standard buffer, re-concentrated to -1.5 ml, and stored at -60 "C. DTT-Ubiquitin Hydrolysis Assay The procedure was a slight modification of that described previously (PPi production in excess of ubiquitin, in the presence of El and dithiothreitol; Ref. 11). The assay contained, in a volume of 32 pl (ph 7.3, 37 T): 50 mm Tris.HC1 (20% base), 5 mm MgC12, 4-5 pm [y-32p]atp(-350 cpm/pmol), 50 mm KC1,20 mm dithiothreitol, 0.1 mm EDTA, -0.6 units/ml PPase, 4 pmol of ubiquitin (0.12 pm), and carboxyl-terminal hydrolase as indicated. The reaction was initiated with El, -0.7 pmol (-20 nm); 10-pl aliquots were withdrawn as a function of time and quenched by addition to 240 pl of a solution containing 0.89 N HzSOl and 0.67 mm Pi. [s2p]pi was extracted as the phosphomolybdate complex into 2-butanol (13) and quantitated by liquid scintillation counting in a toluene/ethanol/liquifluor (New England Nuclear) (2/1/0.085) mixture. Production of PPi was linear with time after the initial burst; when necessary, rates obtained in the presence of hydrolase were corrected by subtracting a small blank rate obtained with El alone (caused by traces of hydrolase in some El preparations). Assays for Hydrolysis of Lys-Ubiquitin, GME-Ubquitin, and Spermidine- Ubiquitin These assays also depended on in situ generation of substrate, for which E2 was required in addition to El. The concentrations of buffer, MgC12, [32P]ATP, EDTA, and PPase were the same as for DTT-ubiquitin hydrolysis; the reaction mixtures (32 pl) also contained amine (50 mm L-lysine/HCl, 50 mm glycine methyl ester/hc1, or 16.7 mm spermidine/hcl), 0.2mM dithiothreitol, 7.2 pmol of ubiquitin (0.22 pm), -0.7 pmol of El, and 3-6 pmol of E2 ( pm). The procedure was the same as for DTT-ubiquitin hydrolysis. Steady-state Assay for Ubiquitin-Hydroxyiumine Adduct Hydrolysis A ubiquitin-hydroxylamine adduct was generated by the reaction of hydroxylamine (95% base) with the El-ubiquitin (and, in some cases, E2-ubiquitin) thiol esters. In the case of El, hydroxylamine reacts with the ubiquitin thiol ester and not with bound ubiquitin adenylate, as shown by abolition of adduct formation after treatment of El with iodoacetamide (this treatment prevents thiol ester, but not adenylate, formation (14)). The second-order rate constant for reaction of total hydroxylamine with El-ubiquitin, determined from the rate of hydroxylamine-dependent [32P]PPi production in the presence of excess ubiquitin, was -25 M" min" (ph 7.3, 37 'C, 0.1 M C1- and ionic strength 0.1). Rate constants for individual E2- ubiquitin thiol esters were not determined; however, an E2 mixture that reacted with total lysine with a second-order rate constant of 6.4 M" min" reacted with total hydroxylamine with a rate constant of -100 M" min" (conditions as for El-ubiquitin). The assay for ubiquitin-hydroxylamine adduct hydrolysis contained hydroxylamine, El, and E2 as indicated, KC1 to give 0.1 M C1-, 0.1 mm EDTA, and 0.2 mm dithiothreitol; other conditions and the procedure were the same as for Lys-ubiquitin hydrolysis. Assay of Cytochrome c- Ubiquitin Hydrolysis Reaction mixtures contained, in 12 pl (ph 7.3,37 T): 50 mm Tris. HCl (20% base), 0.3 mm dithiothreitol, 0.14 mg/ml ovalbumin, and 0.16 pm cytochrome c-'261-ubiquitin (4000 cpm/pmol, above). Reactions were initiated by adding hydrolase (20-40 microunits, -3-6 nm) and quenched after min with 12 pl of SDS sample buffer (5). After SDS-gel electrophoresis and autoradiography of the dried gel, free '%I-ubiquitin was quantitated by counting the excised band in an Abbott Autologic y-counter. Specificity of Ubiquitin Carboxyl-terminal Hydrolase 7905 Assay of Ubiquitin Small amounts (400 pmol) of ubiquitin could be accurately determined by incubating samples in the presence of El under the conditions of the standard DTT-ubiquitin hydrolysis assay (except that hydrolase was rigorously excluded). Under these conditions, ubiquitin is essentially quantitatively converted to DTT-ubiquitin, with stoichiometric production of [32P]PPi. The kinetics of ["PIPPi production are independent of the concentration of dithiothreitol (20-50 mm) and show a very rapid burst and little or no subsequent rate (as expected, since ubiquitin cannot be regenerated from DTT-ubiquitin). Results obtained with samples were compared to a linear standard curve constructed using known amounts of ubiquitin. In constructing the standard curve, it was noted that the ubiquitin concentration obtained using this assay was only 40% of that calculated from the 280 nm absorbance of the ubiquitin stock solution. (A similar discrepancy was noted in catalytic transfer of ubiquitin to amines.) Since the ubiquitin was homogeneous by SDS-gel electrophoresis and isoelectric focusing, we assume that 60% of the ubiquitin as isolated lacks its COOH-terminal glycylglycine. Removal of this dipeptide by cellular proteases (to produce ubiquitin-74) is known to be a problem in ubiquitin purification (15). However, the presence of a significant concentration ubiquitin-74 in hydrolase assays is not expected to affect any results, since ubiquitin-74 binds at least 10 times more poorly to hydrolase than does ubiquitin-76 (11). Ubiquitin concentrations reported in this paper are those obtained from the assay using El (i.e. ubiquitin-76 concentration). Synthesis, Characterization, and Hydrolysis of Ubiquitin Hydroxamic Acid Conditions were (ph 7.3, 37 T): 0.1 M Tris.HC1 (20% base), 10 mmmgc12, 10 pm ["PIATP, 0.2 mm EDTA, 0.4 mm dithiothreitol, 1.5 units/ml PPase, 1.4 p~ ubiquitin, 2.5 mg/ml bovine serum albumin, 50 mm hydroxylamine (95% base), 62 nm El, and 117 nm E2. After 60 min, assay of [3zP]PPindicated that all ubiquitin had reacted with hydroxylamine. An equal volume of 1 M hydroxylamine (90% base) was then added together with 0.4 mm iodoacetamide. (This additional incubation was to ensure complete conversion of ubiquitin from O-acyl adduct to N-acyl hydroxamate.) After 120 min more, trichloroacetic acid (10% final) was added; the precipitate was collected by centrifugation, washed twice with 2% trichloroacetic acid, and resuspended in an appropriate volume of buffer containing 0.1 M Tris.HC1(20% base), 0.1 mm EDTA, and 0.2 mm dithiothreitol. This ubiquitin hydroxamic acid solution was assayed for ubiquitin (above), and was found to contain no detectable ubiquitin (relative to a control that lacked ATP in the initial incubation). Incubation of the ubiquitin hydroxamic acid at ph $13 and 37 "C for 5 min did not result in appearance of detectable ubiquitin (the same incubation had no effect on the -ATP control); in contrast, the O-acyl hydroxylamine adduct of benzoic acid is instantanebusly hydrolyzed by exposure to 10 mm NaOH at room temperature (16). These results indicate that the ubiquitin hydroxamic acid preparation contained no detectable 0- acyl hydroxylamine adduct. The ability of the hydrolase to cleave ubiquitin (N-acyl) hydroxamic acid was tested in 25-p1 incubations containing 20 mm Tris. HCl and 25 pmol of ubiquitin hydroxamic acid (ph 7.3,37 "C). After 30 min, hydrolase was inactivated by treatment with iodoacetamide (1.8 mm final) for 20 min; dithiothreitol (20 mm final) was added to quench iodoacetamide, and the amount of free ubiquitin product was assayed by reaction with El (above). Electrophoretic Methods SDS-gel electrophoresis (acrylamide/bisacrylamide = 30/0.6) using 12.5% acrylamide gels was done by the discontinuous slab procedure of Laemmli (17). The procedure for silver staining was that of Wray et al. (18). Autoradiography of dried gels was as described (5). RESULTS Hydrolase Acts on the Ubiquitin Amide of Lysine-Hydrolysis of Lys-ubiquitin was assayed by a modification of the ubiquitin "recycling" assay used previously for DTT-ubiquitin ("Materials and Methods;" Ref. 11). EZ-ubiquitin thiol esters catalyze ubiquitin transfer to the -amino group of free lysine (5); incubation of a small amount of ubiquitin together with El, E2, lysine, and Mg-[y3'P]ATP results in conversion of

4 7906 Specificity of Ubiquitin Carboxyl-terminal Hydrolase all the ubiquitin to Lys-ubiquitin with a stoichiometric (5) burst AMP + kj LJ El A\TP SCHEME 1 (Scheme I), concomitant L RNH2 I hydrolase of PPi production (Fig. 2, crosses). (PPi is measured as [ PI Pi by including inorganic PPase in the incubation.) According to Scheme 1, production of PPi in excess of the amount of ubiquitin present depends on ubiquitin regeneration by hydrolysis of Lys-ubiquitin. If all steps in Scheme 1 are rapid relative to the ubiquitin-regeneration step, the rate of PPi production in the steady state will be essentially equal to the rate of Lys-ubiquitin hydrolysis. Addition of ubiquitin carboxyl-terminal hydrolase under these conditions of low ubiquitin concentration stimulates steady-state PPi production (Fig. 2, squares). For small amounts of hydrolase, the increase in steady-state rate is proportional to the amount of hydrolase added (Fig. 2, squares and triangles). These results suggest that the hydrolase acts on Lys-ubiquitin. (The rate seen in the absence of added hydrolase, crosses, is probably caused by traces of hydrolase in the El and/or E2 preparations.) In the presence of a large amount of hydrolase, the rate of PPi production is the same as the rate of Lys-ubiquitin formation seen at high ubiquitin concentration (Fig. 2, filled versus open circles), indicating that under these conditions (filled circles) Lys-ubiquitin hydrolysis no longer contributes to rate limitation. The observed hydrolysis of Lys-ubiquitin (Fig. 2) is an activity of ubiquitin carboxyl-terminal hydrolase, and not of a contaminating am&se, based on the following arguments. 1) The hydrolase preparation is 295% homogeneous by the /d -a /.Lu 2 0 ; 30 a Z g P 0.7pJ none I I I I I minutes FIG. 2. Hydrolysis of Lys-ubiquitin by ubiquitin carboxylterminal hydrolase, ph 7.3 and 37 C. Reactions were carried out as described under Materials and Methods (Lys-ubiquitin hydrolysis assay) with 31 nm El and 188 nm E2, except that the incubations contained ubiquitin, lysine, and carboxyl-terminal hydrolase as follows: 0, 65 pmol of ubiquitin (2.0 PM), no lysine (50 mm KCl), no hydrolase; 0, 65 pmol of ubiquitin, 50 mm lysine, no hydrolase; X, 7.2 pmol of ubiquitin (0.22 FM), 50 mm lysine, no hydrolase; W, 7.2 pmol of ubiquitin, 50 mm lysine, 0.7 microunits of hydrolase (0.038 nm); A, 7.2 pmol of ubiquitin, 50 mm lysine, 1.5 microunits of hydrolase (0.081 nm); 0, 7.2 pmol of ubiquitin, 50 mm lysine, 17.5 microunits of hydrolase (0.94 nm). Stimulation of steadystate PPi formation by hydrolase required the presence of lysine (not shown). criterion of SDS-polyacrylamide gel electrophoresis (Fig. 1) and has a high specific activity (20 units/mg) in DTT-ubiquitin hydrolysis ( Materials and Methods ). DTT-ubiquitin is the standard thiol ester substrate for the hydrolase (11). 2) Hydrolytic activities (Iz,.,/K,,,) toward Lys-ubiquitin and DTT-ubiquitin are quantitatively similar (below). 3) Hydrolytic activities toward these two substrates are (a) stable to brief exposure of the enzyme to trichloroacetic acid, (b) sensitive to iodoacetamide treatment, and (c) protected by ubiquitin against iodoacetamide inactivation (Table I); the last three properties were reported previously for ubiquitin carboxyl-terminal hydrolase (11). Hydrohse Acts on Ubiquitin Amides of Other Small Am&s-Data summarized in Table II show that the hydrolase acts on the ubiquitin amides of glycine methyl ester (GME-ubiquitin), and spermidine (spermidine-ubiquitin); the rates are similar to that obtained with the same concentration of Lys-ubiquitin. Very limited steady-state kinetic data suggest that the conditions of the standard assay of DTT-ubiquitin (0.11 KM) are those of k,,,/k,,,, while those of the assay of Lys-ubiquitin (0.22 PM) are closer to kcat. (For these experiments, substrate concentration was varied by varying the ubiquitin concentration in the assay; not shown.) For DTTubiquitin, the hydrolase rate increases approximately linearly with increasing substrate concentration in the range TABLE Loss of activity in hydrolysis of DTT-ubiquitin and Lys-ubiquitin Treatment Trichloroacetic acid Iodoacetamided (ph 7.3,37 C, 0.5 mm iodoacetamide x 20 min) Iodoacetamide pm ubiquitin (as above) I Hydrolytic activity, % control DTT-ubiauitin Lvs-ubiauitid Standard assay as described under Materials and Methods. Control activity was 17.4 microunits/pl enzyme. * Standard assay ( Materials and Methods ); control activity: 6.4 microunits/rl enzyme. Enzyme (8.7 microunits), in a volume of 50 ~1 (containing 34 rg of bovine serum albumin as carrier), was precipitated by addition of 100 ~1 of cold 16% trichloroacetic acid. The pellet was collected by centrifugation at 5 C, washed once with 300 ~1 of cold distilled water, and suspended in 50 ~1 of buffer containing 100 mm Tris.HCl, ph 7.6,0.1 mm EDTA, and 0.2 mm dithiothreitol; aliquots were assayed for activity. denzyme (17.4 microunits) was treated with iodoacetamide in a volume of 10 ~1 (containing bovine serum albumin, 1 mg/ml). The reaction was quenched with 5 mm DTT. TABLE Hydrolysis of ubiquitin derivatives by carboxyl-terminal hydrolase, ph 7.3 (37 C) Reactions carried out with appropriate amounts of e&erase as described under Materials and Methods (32 ~1 reaction volume). Substrate concentrations were 0.11 pm (DTT-ubiquitin) or 0.22 /.IM (amides). Derivative II Rate ~mol/min/mg DTT-ubiquitin 20 Lys-ubiquitin 7.4 GME-ubiquitin 5 Spermidine-ubiquitin 5 Q Glycine methyl ester, 50 mm, inhibited DTT-ubiquitin hydrolysis by 35%; the rate shown for GME-ubiquitin has been corrected on the assumption that 50 mm glycine methyl ester also inhibits GMEubiquitin hydrolysis by 35%. The basis for this inhibition is unclear at this time.

5 Specificity of Ubiquitin Carboxyl-terminal Hydrolase PM; the slope of the line gives a value for kat/k,,, of -10 M s-l. In the case of Lys-ubiquitin, the hydrolase rate increases only slightly (20%) on increasing substrate concentration from 0.23 to 0.46 PM. The data are consistent with values of -6 s for kat and 26 X lo7 M s-l for kat/km. Thus, the values of kat/k,,, for thiol ester (DTT-ubiquitin) and amide (Lys-ubiquitin) substrates are close to diffusioncontrolled and probably represent the rate of association of enzyme and substrate. Hydrolase Has a Low Activity toward Ubiquitin Amides of Proteins-Enzyme activity toward a ubiquitin conjugate of yeast cytochrome c was also examined. (The ubiquitin in this adduct is almost certainly connected uia an e-amino group of a cytochrome lysine residue; Materials and Methods. ) Slow production of free ubiquitin, with a rate of2.2 x pmol/ min, was observed in the presence of 0.16 PM cytochrome c- ubiquitin and 21 microunits of hydrolase (not shown; assay described under Materials and Methods ). Results obtained with ubiquitin conjugates of native lysozyme were quantitatively similar. Hydrolysis of Lys-ubiquitin under similar conditions of substrate concentration would occur with a rate of 3.9 pmol/min. These results show directly that the hydrolase is highly selective for ubiquitin amides of small amines. It is possible that the observed activity toward ubiquitin-protein conjugates reflects a low level (-0.05%) of contamination by an isopeptidase. Hydrolase Acts on Ubiquitin Hydroxamic Acid-Hydroxylamine reacts with El- and E2-ubiquitin thiol esters ( Materials and Methods ). In many but not all cases (16), the reaction of hydroxylamine with acyl compounds occurs in two steps: an initially formed, unstable O-acyl adduct reacts with a second molecule of hydroxylamine to form the stable (Nacyl) hydroxamic acid (16, 19). Typically, the second step is much slower than the first (19). While the initial product of the reaction of hydroxylamine with the enzyme-ubiquitin thiol esters has not been chemically characterized, the product resulting from prolonged incubation with hydroxylamine is probably the N-acyl hydroxamic acid, as shown by the lack of a detectable free ubiquitin COOH terminus even after an incubation in alkali that should hydrolyze any O-acyl adduct present ( Materials and Methods ). Incubation of this alkali- stable COOH-terminal ubiquitin derivative (1 p ~ with ) hydrolase (0.43 or 0.86 nm) for 30 min resulted in regeneration of the ubiquitin COOH terminus to the extent of -95%, as assayed by restoration of its ability to react with El (not shown). This result suggests that the enzyme can hydrolyze ubiquitin (N-acyl) hydroxamic acid. Inactivation during Hydroxylamine Adduct Cleavage-An attempt was made to quantitate activity toward ubiquitinhydroxylamine adduct(s) with a steady-state assay similar to those used for the other ubiquitin derivatives. As shown in Fig. 3 (filled squares versus open circles), a large amount of hydrolase allows extensive recycling of ubiquitin in the presence of 10 mm hydroxylamine (together with El, E2, and MgATP). The observed rate of PPi production is equal to the rate of hydroxylamine adduct formation seen in the presence of a high concentration of ubiquitin and no hydrolase (Fig. 3, open squares), indicating that adduct cleavage does not con- tribute to rate limitation when a large amount of hydrolase is present. The ability of a large amount of hydrolase to recycle ubiquitin very rapidly (e.g. Fig. 3, filled squares) was independent of hydroxylamine concentration in the range 1-50 mm (not shown). This result suggests that if the reaction of hydroxylamine with enzyme-ubiquitin thiol esters proceeds through an O-acyl adduct, that adduct is a substrate for the hydrolase. (At very low hydroxylamine concentration, the minutes FIG. 3. Hydrolase allows ubiquitin recycling in the presence of hydroxylamine, but loss of activity occurs. Assays were carried out as described under Materials and Methods, with 62 nm El and 94 nm E2. Low concentrations of ubiquitin (0.22 p ~ 7.2, pmol) were supplemented with: 50 mm hydroxylamine alone (0)(this result is insensitive to hydroxylamine concentration, not shown), 50 mm hydroxylamine and 7 microunits of (0.38 nm) hydrolase (+), or 10 mm hydroxylamine and 470 microunits (25 nm) hydrolase (m). The latter rate (m) is identical to that observed at 10 nm hydroxyl- amine in the absence of hydrolase, but in the presence of a high concentration of ubiquitin (1.5 p ~ 47, pmol) (0). The result for conditions of no hydroxylamine, no hydrolase, and high ubiquitin concentration is essentially identical to that shown by the open circhs (not shown). The reason for production of excess PPi at high ubiquitin concentration (0) is under investigation. minutes L FIG. 4. Basis of inactivation in the presence of hydroxylamine. Assays were carried out as described under Materials and Methods, with 20 mm hydroxylamine, 188 nm El, 0.19 pm (6 pmol) ubiquitin, and no further additions (0); 690 microunits of (37 nm) hydrolase (0); or 13.8 microunits of (0.74 nm) hydrolase (O), which was supplemented with either 6 pmol of ubiquitin (V) or 13.8 microunits of fresh hydrolase (A) at the time shown by the arrow. reaction of hydroxylamine with an O-acyl adduct should be very slow (19).) Attempts to assay ubiquitin-hydroxylamine adduct hydrolysis in the presence of small amounts of hydrolase resulted in apparent inactivation of the enzyme (Fig. 3, diamonds; Fig. 4, squares). Failure to restore activity by adding fresh ubiquitin (Fig. 4, inverted triangles) shows that the rate decrease does not reflect time-dependent accumulation of an unhydrolyzable form of ubiquitin. Such a result would not be expected, since the hydrolase cleaves (N-acyl) ubiquitin hydroxamic acid. The small burst observed on addition of fresh ubiquitin corresponds to transfer of this ubiquitin to hydroxylamine; observation of the burst excludes the trivial explanation of El inactivation for the rate decrease shown in Fig. 4 (squares). Addition of fresh hydrolase after turnover has ceased brings about a further round of turnovers (Fig. 4, triangles). Thus the rate decrease (Fig. 3, diamonds; Fig. 4, squares) is actually caused by hydrolase inactivation. The mechanism of this inactivation is currently being investigated; inactivation does

6 ~ ~~ 7908 Ubiquitin Specificity of Carboxyl-terminal Hydrolase TABLE I11 Large-scale preparation of ubquitin carboxyl-terminal hydrolase from erythrocytes Procedure described in detail under Materials and Methods. Total Hydrolase % RecoveIy step Unita/mg [protein] Activityb Hydrolase In step Overall ml mg/ml milliunitslml total units Hemolysateb DE-52 KC1 eluate Resuspended trichloroacetic acid pellet Concentrated affinity column 7590 eluate 1.2d Affinity column breakthrough a Standard assay of DTT-ubiquitin hydrolysis as described under Materials and Methods, except KC1 was not added. From 400-ml erythrocytes (source: outdated human blood). E Estimated from absorbance at 280 nm. By the method of Lowry et al (20). not occur on incubation of hydrolase with hydroxylamine alone (50 mm, not shown). Preparation of Hydrolase from Erythrocytes-A relatively simple preparation of hydrolase from outdated human blood, which is available in large quantity, is described under Materials and Methods. In addition to its ubiquitin affinity property, the purification scheme takes advantage of the enzyme s acidic PI (relative to hemoglobin) and acid stability. A volume of erythrocytes that was comfortably processed in a day yielded -2 mg of protein containing -1 mg of hydrolase (Table 111; calculated using a specific activity of 20 units/mg). Most of the remaining contaminating proteins are of higher molecular weight than hydrolase (not shown) and could be removed by adding a molecular weight fractionation step. DISCUSSION ubiquitin thiol ester in which the El moiety is denatured (11). These results suggest that the enzyme is quite selective for adducts in which the size of the moiety at the ubiquitin COOH terminus is small. This size specificity distinguishes the hydrolase from the less well characterized isopeptidases that act on large ubiquitin-protein conjugates (2, 7-9). Both the hydrolase and isopeptidases are inactivated by iodoacetamide (Table I; 7,9,11). It is not known whether the isopeptidases can act on small ubiquitin amides, or on other COOH-terminal ubiquitin adducts. Matsumoto et al. (22) have recently reported that ubiquitin catalyzes the hydrolysis of p-nitrophenyl acetate. We have attempted to repeat these experiments, with the expectation that the reported activity might represent formation of a COOH-terminal acetate ester of ubiquitin, followed by hydrolysis catalyzed by carboxyl-terminal hydrolase contaminating the ubiquitin preparation. However, we were unable to repeat their observation that ubiquitin reacts with p-nitrophenyl acetate in a burst; nor did we find any evidence for Specificity of the Enzyme-Initial studies with ubiquitin carboxyl-terminal hydrolase showed that this enzyme acts on small ubiquitin thiol esters and free ubiquitin adenylate (11). The present work extends the known specificity of the enzyme catalysis of p-nitrophenyl acetate hydrolysis by ubiquitin. to include more thermodynamically stable (small) COOH- Possibly the ubiquitin preparation used by these authors was terminal ubiquitin adducts, including the ubiquitin amides of contaminated with a nonspecific esterase. lysine (e-amino), glycine methyl ester, and spermidine, as well The reaction of El with ubiquitin requires that the COOHas ubiquitin hydroxamic acid. The observed activity of the terminal glycylglycine of ubiquitin (residues 75 and 76) be enzyme toward the ubiquitin-hydroxylamine adduct formed intact (15). The observation of a constant rate in the ubiquitin at very low hydroxylamine concentration suggests that the recycling assay of hydrolase (e.g., for 30 min as shown in Fig. enzyme can also hydrolyze an 0-acyl hydroxylamine adduct 2, solid sqwres) shows that the product of the hydrolase ( Results ). However, it remains to be shown experimentally reaction is free ubiquitin with glycines 75 and 76 intact. For that 0-acyl ubiquitin hydroxylamine is actually the predomi- the case of the glycine methyl ester adduct of ubiquitin, nant species being hydrolyzed under these conditions. The faithful formation of this ubiquitin-76 product requires that finding that the enzyme hydrolyzes amides as well as thiol the hydrolase be able to distinguish between the Gly 75-Gly esters prompted us to change its name from esterase (11) 76 and Gly76-Gly 77 (methyl ester) junctions as sites of to hydrolase (Introduction). cleavage. The terminal GlyGly residues of ubiquitin may be The broad specificity of the hydrolase with respect to the important in its binding to the hydrolase, as indicated by the nature of the chemical linkage at the ubiquitin COOH ter- observation that ubiquitin lacking its COOH-terminal GlyGly minus is reminiscent of the broad specificity of proteases such inhibits DTT-ubiquitin hydrolysis at least 10 times more as chymotrypsin. Chymotrypsin hydrolyses esters, amides, weakly than does intact ubiquitin-76 (11). However, simple and hydroxamates having the correct structure to bind at the glycylglycine (1 mm) does not bind to the hydrolase (11). enzyme s active site (21). Ubiquitin carboxyl-terminal hydro- Inactivation of hydrolase in the presence of hydroxylamine lase acts on substrates of the same chemical classes; the (Figs. 3 and 4) does not occur in the absence of ubiquitin ubiquitin moiety of the substrate is responsible for specificity, ( Results ) and is therefore likely to be caused by interaction as shown by inhibition of DTT-ubiquitin hydrolysis by free of the enzyme with a ubiquitin-hydroxylamine adduct, or by ubiquitin (11). reaction of hydroxylamine with an activated intermediate in In contrast to it similar activities toward amides of several turnover. Hydroxamic acid substrate derivatives inhibit a different small amines (Table 11), the hydrolase has a much number of proteases, including elastase (231, thermolysin (241, lower activity toward ubiquitin amides (conjugates) of the and aminopeptidases (25). However, all of these are metalsmall proteins, cytochrome c and lysozyme (-3500-fold less loenzymes, and inhibition results from coordination of an activity at similar substrate concentrations; Results ). It was essential metal ion, usually zinc, by the hydroxamate subshown previously that hydrolase does not cleave an El- strate analog (which is usually not cleaved). The stability of

7 ubiquitin carboxyl-terminal hydrolyase during isolation and storage in EDTA-containing buffers ( Materials and Methods ), and to treatment with strong acid (Table I; Ref. ll), argue against but do not exclude the possibility of a tightly bound metal ion that is required for catalysis. Studies on its mechanism of inactivation may shed light on the mechanism of the enzyme in normal catalysis. This reaction may also have practical value as a way to specifically inactivate hydrolase present in other enzyme preparations. Biological Role-The question of the role of the hydrolase in uiuo remains incompletely resolved. It almost certainly functions to rescue ubiquitin from various adducts that may result from adventitious reactions. A calculation based on the glutathione concentration in erythrocytes (-3 mm; Ref. 26), the approximate second-order rate constant for reaction of dithiothreitol with the El-ubiquitin thiol ester (380 M min ; Ref. 14), and the approximate concentration of El in rabbit reticulocytes, 80 nm, suggests that if all El were in the ubiquitin thiol ester form and were able to react with glutathione, all ubiquitin (-15 IM) in the cell would be converted to the glutathione thiol ester in less than 1 min. Similarly, if all E2 (-160 nm) were in the ubiquitin thiol ester form and were available to react with intracellular polyamines (1-2 mm; Ref. 27), all ubiquitin would be converted to polyamine amide adducts in a matter of minutes (Kp-18 M min ; Ref. 5). These calculations are very approximate but serve to show that the large amount of hydrolase in rabbit reticulocytes, -30 milliunits/ml packed cells (11), is probably necessary to prevent accumulation of adventitious adducts of ubiquitin. In the absence of an enzyme with a broad specificity like that of the hydrolase, these reactions would result in entry of ubiquitin into metabolically inactive pools. Based on its activity toward Lys-ubiquitin, the enzyme may be able to release ubiquitin from small lysine isopeptides that are expected to result when ubiquitin-protein conjugates are degraded (2). Studies of the specificity of the hydrolase have been limited, however, by the specificities of the enzymes used for synthesis of ubiquitin derivatives; at present, the only a- amino ubiquitin amide available is that of glycine (methyl ester). The finding that the hydrolase acts on the glycine methyl ester derivative suggests that other amino acidubiquitin amides may be substrates as well. This question is important in considering a role for the hydrolase in regenerating ubiquitin from possible ubiquitin conjugates of a-amino groups of proteins (6), once the size of such conjugates have been decreased by the ubiquitin- and ATP-dependent protease system. Pending experimental demonstration of a dependence of protein breakdown on hydrolase activity, the hypothesis that most or all ubiquitin might be regenerated during breakdown by the action of isopeptidases on relatively large ubiquitinprotein adducts cannot be excluded. Because conjugate formation appears to be rate-limiting in protein breakdown (28), no kinetic evidence is available concerning the nature of the steps in ubiquitin-dependent proteolysis that follow ubiquitinprotein conjugate formation, except that conjugate degradation is ATP-dependent (7). Thus, we do not know how long ubiquitin must be retained on fragments of the original target protein molecule in order to ensure complete degradation of those fragments. We have been unable thus far to address experimentally the question of whether protein breakdown is hydrolase-dependent because of difficulty in totally depleting reconstituted lysates of hydrolase. This problem may perhaps be overcome in the future by treatment with hydroxylamine under appropriate conditions. Ozkaynak et al. (29) have recently reported thathe Specificity of Ubiquitin Carboxyl-terminal Hydrolase 7909 ubiquitin gene in yeast consists of six ubiquitin-coding repeats in a head to tail arrangement. The sixth (COOH-terminal) repeat is distinct from the others in that it encodes a ubiquitin having an extra residue, asparagine, at the COOH terminus (i.e. COOH-terminal sequence -Gly 75-Gly 76-Asn 77). This gene arrangement raises the possibility that the ubiquitin monomer arises from processing of a poly-ubiquitin precursor protein. The author suggest that ubiquitin carboxyl-terminal hydrolase might function in such processing, whichwould presumably involve removal of the COOH-terminal Asn res- idue, and cleavage at the internal Gly-Met junctions (not necessarily in this order). Hydrolysis of the ubiquitin COOHterminal amide of Asn may be an activity of the hydrolase, based on observed activity toward the ubiquitin amide of glycine methyl ester. However, a role for hydrolase in cleavage at internal Gly-Met junctions seems inconsistent with results, discussed above, which suggest that the enzyme is highly specific for small (non-protein) ubiquitin amide adducts. Further studies on this enzyme should be facilitated by a simpler purification scheme (Table 111) using red cells, which are more easily obtained in large quantity than are reticulocytes. ATP-dependent proteolysis, a significant part of which is ubiquitin-dependent (30), decreases to negligible levels during maturation of reticulocytes to red cells (28, 31). It is not known which specific components of the ubiquitin system are lost during this developmental process. Speiser and Etlinger (31) reported that addition of a crude reticulocyte ubiquitin fraction to erythrocyte extracts restored ATP-dependent proteolysis to nearly the levels seen in reticulocyte extracts. Stimulation was tentatively attributed to ubiquitin itself or to an associated factor in the preparation. Added ubiquitin might stimulate if all red cell ubiquitin were tied up in smallmolecule adducts as a result of a maturational loss of hydrolase activity. The finding that human red cell hydrolase activity, -120 milliunits/ml packed cells, is higher than that reported in (rabbit) reticulocytes, -30 milliunits/ml packed cells (ll), argues against this explanation, which would also appear to be inconsistent with the observation that human red cells are an excellent source of ubiquitin (15). The observed stimulation thus seems most likely to be caused by some other factor. REFERENCES 1. Ciechanover, A., Finley, D., and Varshavsky, A. (1984) J. Cell. Bwchem. 24, Hershko, A., Ciechanover, A., Heller, H., Haas, A. L., and Rose, I. A. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, Hershko, A., Heller, H., Elias, S., and Ciechanover, A. (1983) J. Bwl. Chem. 268, Haas, A. L., Warms, J. V. B., Hershko, A., and Rose, I. A. (1982) J. Bwl. Chem. 267, Pickart, C. M., and Rose, I. A. (1985) J. Biol. Chem. 260, Hershko, A., Heller, H., Eytan, E., Kaklij, G., and Rose, I. A. (1984) Proc. Natl. Acad. Sci. U. S. A. 81, Hershko, A., Leshinsky, E., Ganoth, D., and Heller, H. (1984) Proc. Natl. Acad. Sci. U. S. A. 81, a. Haas, A. L., and Rose, I. A. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, Matsui, S.-I., Sandberg, A., Negoro, S., Seon, B. K., and Goldstein, G. (1982) Proc. Natl. Acad. Sci. U. s. A. 79, Busch, H., and Goldknopf, I. L. (1981) Mol. Cell. Biochem. 840, Rose, I. A., and Warms, J. V. B. (1983) Biochemistry 22, Dayhoff, M. 0. (ed) (1969) in Ath of Protein Sequence and Structure, Vol. 4, p. D-24, National Biomedical Research Foundation, Silver Spring 13. Berenblum, I., and Chain, E. (1938) Biochem. J. 32,

8 7910 Specificity of Ubiquitin Carboxyl-terminal Hydrolase 14. Haas, A. L., Warms, J. V. B., and Rose, I. A. (1983) Biochemistry 22, Wilkinson. K. D., and Audhva. -. T. K. (1981)., J. Biol. Chem ' 16. Jencks, W. P. (1958) J. Am. Chem. Soc. 80, Laemmli, U. K. (1970) Nature 227, Wray, W., Boulikas, T., Wray, V. P., and Hancock, R. (1981) Anal. Biochem. 118, Jencks, W. P. (1958) J. Am. Chem. SOC. 80,4& Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, Foster, R. J., and Niemann, C. (1955) J. Am. Chem. SOC. 77, Matsumoto, H., Taniguchi, N., and Deutsch, H. F. (1984) Arch. Biochem. Biophys. 234, Nishino, N., and Powers, J. C. (1980) J. Biol. Chem. 255, Holmes, M.A., and Matthews, B. W. (1981) Biochemistry 20, Wilke, S. H., and Prescott, J. M. (1983) J. Biol Chem. 258, Griffith, 0. W. (1981) J. Biol. Chem. 256, Cohen, S. S. (1971) in Introduction to the Polyamines, p. 29, Prentice-Hall Inc., Englewood Cliffs, NJ 28. Hershko, A., and Ciechanover, A. (1982) Annu. Rev. Bwchem. 51, Ozkaynak, E., Finley, D., and Varshavsky, A. (1984) Nature 312, Tanaka, IC., Waxman, L., and Goldberg, A. (1983) J. CeU Bwl. 96, Speiser, S., and Etlinger, J. D. (1982) J. Biol. Chem. 257,