The Protein Substrate Binding Site of the Ubiquitin-Protein Ligase ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY 01986 by The American Society of Biological Chemists, Inc.

Vol. 261, No. 26, Issue of September 15,pp. 11992-11999 1986 Printed in irS.A.

The Protein SubstrateBinding Site of the Ubiquitin-Protein Ligase System* (Received for publication, February 3, 1986)

Avram Hershkoz, HannahHellerS, Esther Eytan, andYuval ReissS From the Unit of Biochemistry, Faculty of Medicine, Technion-ZsraelInstitute of Technology, Haifa, Israel and the Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, Pennsylvania 191 11

In order to gain insight into the mechanisms that protein substrates are first ligated to the polypeptide Ub and determine the selectivity of the ubiquitin proteolytic then degraded by a system whichspecifically attacks Ubsite of the ubi- conjugated proteins (for reviews, see Refs. 1-3). pathway, the protein substrate binding quitin-protein ligase system was identified and exam- The question arises of how specific proteins are selected by ined. Previous studies had shown that the ligase system the Ub ligation system to be committed to degradation. The consists of three components: a ubiquitin-activating answer must lie both in specific features of protein structure enzyme ( E , ) ,ubiquitin-carrier protein(EZ), and a third which are recognized and in the mode of action of the Ub enzyme, E3, themode of action of which has not been ligase system which enables itto recognize such protein defined. E3 from rabbit reticulocytes was furtherpurified by a combination of affinity chromatography, structures. In a previous study we found that a free NH2terminal a-NHzgroup of the protein substrate is an important hydrophobic chromatography, and gel filtration proas the sub- structural determinant for its degradation by the Ub system cedures. A 180-kDa protein was identified unit of E3. Two independent methods indicate that E3 (4). In the present investigation, we concentrated on the problem of which component of the Ub-protein ligase system has the protein binding site of the ubiquitin ligase system. These are the chemical cross-linking of IZSI- contains the binding site for the protein substrate and the labeled proteins to the E3 subunit and the functional properties of this site. It appeared reasonable to assume that at least some initial selection of protein structures suitable conversion of enzyme-bound labeled proteins to ubifor Ub ligation would occur at such a binding site. quitinconjugatesin pulse-chaseexperiments.The trapping of E3-bound protein for labeled product for- The Ub-protein ligase system consists of three enzyme mationwas allowedby the slowdissociation of components: a Ub-activating enzyme (El), which has been E3*proteincomplex. thoroughly investigated (5-7); Ub-carrier proteins (E2),which E3, accept activated Ub from El (8, 9); and a third enzyme, E3, The specificity of binding of different proteins to examined by both methods, showed a direct correlation the mode of action of which was not defined except for the with their susceptibility to degradationby the ubiqui- observation that it is required for the final transfer of Ub tin system. Proteins withfree a-NH2 groups, whichare from E2 to amide bond formation with proteins (8). Other good substrates, bind better toE3 than corresponding types of Ub-protein ligase systems have been described, such proteins with blocked NH, termini, whichare not sub- as the E3-independent transfer of Ub from certain species of strates. Oxidation of methionine residues to sulfoxide Ez to some basic proteins (9). Still another Ub conjugation derivatives greatly increases the susceptibility of some system exists in the case of certainproteins that require a correspond- tRNA for degradation and Ub ligation (10, 11).The mode of proteins to ligation with ubiquitin, with ing increase in their bindingE3. to However, a protein derivative which was subjected to both amino group action of tRNA has notyet been elucidated. With regard to the question of the protein binding site of modification and oxidation binds strongly to the enzyme, even though it cannot be ligated to ubiquitin. It the major Ub ligase system involved in protein breakdown, E3partic- the possibilities were that either E2, or E3, or both,are thus seems that the substrate binding ofsite ipates in determining the specificity of proteins that involved in substrate protein binding. In this study we show that E3contains the protein binding site. Some properties of enter theubiquitin pathway of protein degradation. the binding process and its specificity have been examined. EXPERIMENTALPROCEDURES

Intracellular proteinbreakdown is a highly selective process in which specific proteins are degraded at widely divergent rates. At least part of this selective degradation is carried out by the ubiquitin (Ub’) proteolytic pathway. In this pathway,

* Supported by United States Public Health Service Grant AM25614 and a grant from the United States-Israel Binational Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18U.S.C. Section 1734 solely to indicate this fact. $Supported by American Cancer Society Grant BC-414 to Dr. Irwin A. Rose. The abbreviations used are: Ub, ubiquitin; Me-, reductively methylated; Ox-, oxidized with performic acid; MetO-, methionine residues oxidized to sulfoxides; DTT, dithiothreitol; SDS, sodium dodecyl sulfate.

Materials-Cytochrome c from horse heart (Type VI) and from Saccharomyces cereuisiae (TypeVIII), enolase from bakers’ yeast (Type 111), chicken egg white lysozyme (Grade I, 3x crystallized), bovine pancreatic RNase A (Type XI1 A), Ox-RNase, and chicken egg ovalbumin (Grade V) were obtained from Sigma. Crystalline rabbit muscle enolase was from Boehringer Mannheim. Ubiquitin was purified from human erythrocytes as described (12, 13). All of these proteins were found to be essentially homogeneous by SDSpolyacrylamide gel electrophoresis. Phenyl-Sepharose CL-4B was obtained from Sigma, Ultrogel AcA 34 from LKB Instruments, and hydroxylapatite (Bio-Gel HT) from Bio-Rad. Bis(sulfosuccinimidy1)suberate was from Pierce Chemical Co. and hematoporphyrin from Sigma. Molecular mass standards for SDS-polyacrylamide gel electrophoresis (Sigma) were (kDa) myosin, 205; @-galactosidase,116; phosphorylase b, 97; bovine serum albumin, 66; ovalbumin, 45. Protein Modifications-Reductive methylation of Ub and lysozyme was carried out as described previously (4,14). Modification of amino

11992

11993

Protein BindingSite of Ubiquitin Ligase groups was >95% in allcases, as determined with fluorescamine (15). Specific oxidation of methionine residues to methionine sulfoxide was carried out by dye-sensitized photooxidation in acid, essentially as described by Jori et al. (16). Briefly, proteins at a concentration of 2 mM in 70% (v/v) acetic acid were mixed with an equal volume of 2 mM hematoporphyrin in the same solvent. The solution was exposed to illumination by four 100-watt light bulbs placed on four sides of a transparent water bath at a distance of 30 cm. A gentle stream of oxygen was passed through the solution, and thereaction was carried out for 60 min a t 25 "C. Hematoporphyrin was removed by a column of Sephadex G-25 equilibrated with 0.2 N acetic acid. The preparation was dialyzed overnight against water and lyophilized. Oxidation of all methionine residues was essentially complete, as found by the resistance of oxidized proteins to cleavage by cyanogen bromide. Enzyme Purification-E,, Ezrand E3 were purified from extracts of rabbit reticulocytes by a slight modification of the previously described affinity chromatography procedure (8).Fraction I1 was first applied to Ub-Sepharose (approximately a 20 mg of Ub/ml of swollen gel) in the absence of ATP ata ratio of fraction I1 to a column of 1:l (by volume). By this method, Ea free of Ez was obtained in the pH 9 eluate (8).The unadsorbed fraction, which was diluted 1.5-fold relative to fraction 11, was applied again to Ub-Sepharose in the presence of ATP ata ratio of 3:l (by volume), and E,, Ez,and residual E3 were isolated by the sequential elution protocol described earlier (8). In both cases, the pH 9 eluate was brought to 2-3%of the starting volume of fraction 11. El and E2 (low-molecular weight form eluted with DTT) (8)were further purified by gel filtration chromatography as described (8)except that a column (1X 50 cm) of Ultrogel AcA 34 was used. Activities of all three enzymes were determined by the rapid quantitative assay for "'I-Ub conjugation described earlier (8). A unit of activity is defined as the amount of enzyme that converts 1 pmol of Ub to conjugates per min. Eawas further purified by hydrophobic chromatography as follows. A 1-ml column (0.5 X 5 cm) of phenyl-Sepharose was equilibrated with 250 mM potassium phosphate (pH 6.7) containing 1 mM DTT. 1 ml of pH 9 eluate (containing approximately 1.5-2 mg of protein and 330-500 microunits of E3activity) was mixed with 10 ml of the above buffer and applied to the column. The unadsorbed fraction (fraction A) was collected. The column was successively eluted with 10-ml portions of the following solutions, all of which contained 1 mM DTT: 10 mM potassium phosphate, pH 6.7 (fraction B); 5 mM potassium phosphate, pH 6.7 (fraction C); 1 mM Tris-HC1, pH 7.6 (fraction D); and 1 mM Tris-HC1, pH 7.6, containing 50% ethylene glycol (fraction E). All fractions were collected into tubes containing 1 mg of ovalbumin. The fractions were concentrated by centrifuge ultrafiltration with CF-25 Centriflo membrane cones (Amicon Corp.), and buffers were changed by two successive lo-fold dilutionswith 20 mM potassium phosphate, pH 7.4, containing 1 mM DTT, followed by ultrafiltration in the same cone. The final volume of all fractions was approximately 0.3 ml. Cross-linking Conditions-Reaction mixtures contained, in a volume of 20 pl, 25 mM potassium phosphate, pH 7.4, 3.3 microunits of E3 (purified through the phenyl-Sepharose step), approximately 1pg of lZ5I-labeledprotein (1.5-3 X 105 cpm), and 0.5 mg/ml ovalbumin. The addition of ovalbumin was necessary to prevent nonspecific adsorption of 'Z51-labeledproteins. Ovalbumin is not a substrate for Ub ligation (8),and at concentrations up to 2 mg/ml, it did not interfere with the cross-linking of '=I-labeled proteins to Ea. The mixtures were allowed to stay at0 "C for 30 min, following which 2 pl of 0.5 mM bis(sulfosuccinimidy1)suberate (dissolved in ice-cold phosphate buffer immediately prior to use) was added. After an additional 15 min at 0 "C, the reaction was stopped by the addition of 2 pl of 1 M ethanolamine-HCl, pH 9.0. The samples were mixed with electrophoresis sample buffer (containing 2% SDS and 3% 2mercaptoethanol), boiled for 5 min, and resolved by electrophoresis on an 8% SDS-polacrylamide gel (17). The gels were stained, dried, and autoradiographed. "Pulse-Chase" Experiments-Unless otherwise stated, the ''pulse" mixture contained, in a volume of 10 pl, 40 mM Tris-HCl, pH 7.6, 2 mM DTT, 1 mg/ml ovalbumin, 2.0 microunits of E3, and 0.15 pg of '251-lysozyme (approximately 7 X 10' cpm). E3 was treated prior to use with 5 mM iodoacetamide at 25 "C for 10 min to inactivate isopeptidases (see Ref. 8). Followingpreincubation at 37 "C for 5 min, the pulse mixture was rapidly mixed with 10 p1 of "chase" mixture containing 40 mM Tris-HC1, pH 7.6, 2 mM DTT, 2 mM ATP, 10 mM MgClZ,5 pg Me-Ub, 0.7 microunits of E,, 0.5 microunits of E2, and 60 pgof unlabeled lysozyme. Unlabeled lysozyme used for chase experiments was subjected to iodination with unlabeled NaI and chloramine T, under conditions identical to those used for radioio&-

nation. The chase mixture was also preincubated at 37 "C for 5 min, prior to mixing. Rapid mixing was obtained by holding the reaction tube on a Vortex mixer while injecting the chase mixture; by this method, mixing was complete within 1s. Following further incubation at 37 "C for the time periods indicated in the figure legends, the reaction was stopped by the addition of 20 pl electrophoresis sample buffer (containing 2% SDS final concentration). The samples were electrophoresed on a 12.5% SDS-polyacrylamide gel. RESULTS

Purification of E3-We have previously purified E3partially by affinity chromatography of reticulocyte extract on UbSepharose, followed by elution at pH 9 (8).This procedure is not specific for Ea, and other Ub-binding proteins are also eluted at pH9. These include a Ub-COOH-terminal hydrolase (18, 19) and at least three different isopeptidases? As shown in Fig. lA, lane I, the pH 9 eluate of the affinity column contains numerous proteins. One way to identify which of these is a subunit of E3 is to examine the coincidence of protein bands with E3activity upon further purification. We noted previously that E3 has a relatively low affinity for Ub, and underconditions optimized for affinity purification about one-half of E3activity remains in the unadsorbed fraction (8). When the unadsorbed fraction is applied again to Ub-Sepharose, a significant part of E3 activity is bound and can be eluted with pH 9 buffer. Analysis of this preparationby silver staining (Fig. lA, lane 2) shows that atleast six protein bands are reduced or absent, as compared to the first pH 9 eluate. These are presumably proteins which have higher affinity than E3 for Ub-Sepharose andare thus more completely removed by the first passage on the affinity column. Since the samples analyzed contained equal amounts of E3activity, the E3 subunit has to be one of the other proteins, equally present in both preparations. E3 from the pH 9 affinity eluate was further purified by hydrophobic chromatography on phenyl-Sepharose. The enzyme wasstrongly bound to thehydrophobic column and was eluted with a marked reduction of ionic strength (Fig. 1B). Analysis of the protein composition of various column fractions (Fig. l A , lanes A-E) showed that several proteins were eluted at higher ionic strength (fraction B), including a major protein of 100 kDa. Some of these may beisopeptidases, since about 60% of isopeptidase activity is eluted in this fraction (datanot shown). On the other hand, another major protein of 180 kDa coincided with E3 activity, eluting mainly in fraction D. The possibility that the 180 kDa protein is a subunitof E3 was examined by coincidence in chromatography on Ultrogel AcA 34. On gel filtration columns, E, elutes at M, 300,000 (8). As shown in Fig. 2, the 180-kDa major band co-eluted with E3 activity on the gel filtration column (peak center, fractions 27-29). Several minor protein bands were not in coincidence with E3activity (Fig. 2B). Similar coincidence of the 180-kDa protein with Eaactivity was observed in hydroxylapatite chromatography, where the enzyme eluted with 100150 mM phosphate concentrations (data notshown). In none of these procedures was a homogeneous preparation of E3 obtained, but the 180-kDa protein was predominant over other reticulocyte proteins following the gel filtration step (Fig. 2B). Carrier ovalbumin could be removed by a second chromatography on phenyl-Sepharose, which doesnot adsorb ovalbumin (not shown). These findings suggest that the 180kDa protein is the subunit of E3. Cross-LinkingofProtein Substratesto E3Subunit-We next examined the possible binding of substrates to purified E3by a cross-linkage method. The enzyme was preincubated with various lZ5I-labeledproteins and then subjected to theaction

-

E. Leshinsky and A. Hershko, unpublished observations.

Site Binding Protein

11994

2 205"

-

A B C D E

of Ubiquitin Ligase

-Origin

"-

...

5

In

B, 205-

23 2 S 2 7 2 8 7 9 31 33

-0rtgm

____.

" " " "

4

ti6" I

97". 66-

. ..

., 45-%

I n

4

J

-$

r-"-oy

D

I

I

I

20

25

30

35

40

Fraction Number FIG. 2. Chromatography of E3 on Ultrogel AcA 34. 200 pl (23.3 microunits of E, activity) of fractionD of the hydrophobic column (see Fig. 1)was applied toa column (0.7 X 30 cm) of Ultrogel AcA 34 equilibrated with 20 mM Tris-HCI, pH 7.2, 1 mM DTT, and 1 mg/ml ovalbumin. Fractions of 0.2 ml were collected. A , E, activity was determined in samples of 5 pI of columnfractions. R, silver staining of samples of 20 pI of column fractions separated on an 8% SDS-polyacrylamide gel. Fraction numbers are indicated a t the top. On the left side are shown the positions of molecular mass markers (kDa). Arrowhead, position of the 180-kDa protein. Ou, ovalhumin carrier.

a general correlation to their susceptibility to the action of the Ub system. Thus, yeast cytochrome c and lysozyme, which have free NH2 termini and are good substrates, areefficiently cross-linked to E:+On the other hand, equine cytochrome c (which has a blocked a-NH2 group and is not a substrate for Ub conjugation and protein breakdown (4)) is cross-linked to a much lesser extent. Modifications of RNase A which inE crease its susceptibility to conjugation with Ub (see below) also increase the extent at which it is found to be cross-linked to the E3subunit (Fig. 3). Theseresults may indicatethatproteinsthatare good substrates bind tightly to a specific site of E,, whereas poor Fraction substrates bindless tightly, or perhapsin a manner that does FIG. 1. Purification of Es by hydrophobic chromatography. A , silver staining(30) of E, preparationsseparatedon 8% SDS- not allow cross-linking to occur under these conditions. We polyacrylamide gel. Lane I, 0.33 microunits (1.7 pg of protein) of the therefore examined the effect of unlabeled proteins on the first pH 9 eluate of Ub-Sepharose; lane 2, 0.37 microunits (1.4 pg of cross-linkage of "'I-labeled yeast cytochrome c to &. As seen protein) of the second pH 9 eluate (see "Materials and Methods"); in Fig. 4, unlabeled yeast cytochrome c prevented the crosslanes A-E, 1 pl of corresponding fractionsof hydrophobic chromatog- linkage of the labeled substrate, whereas a similar excess of raphy of the second pH 9 eluate (see "Materials and Methods" for equine cytochrome c had much less of an effect. Derivatives details of hydrophobic chromatography and designationof fractions). Numbers (kDa) on the left side indicate the position of molecular of RNase A that aregood substrates (seebelow) also compete well on the cross-linkage of labeled substrate (Fig. 4). The mass markers. Arrows next to lane 2 indicate the position of protein handsthatare decreased or absent in the second pH 9 eluate. only exception observed was in the case of native RNase A, Arrowhead on the right side indicates the position of the 180-kDa which is a poor substrate, yet it partially decreased the crossprotein. 0 0 , ovalbumin carrier. R, E, enzymatic activity in fractions linkage of '"I-labeled yeast cytochrome c. These findings of hydrophobic chromatography. Results are expressed as the persuggest that the cross-linkage of labeled proteins mostly recentage of E, activity applied, which was 420 microunits. flects their binding toEa. Demonstration of a Functional &,.Protein Complex: Conof the water-soluble bifunctional agent bis(su1fosuccin- jugate FormationinPulse-ChaseExperiments-The above 21). A s shown in Fig. 3, bis(su1fo- cross-linkage experiments suggest thatproteinsubstrates imidy1)suberate(20, succinimidy1)suberatecausedconsiderableaggregation and bind to Ea but do not necessarilyindicate that binding is nonspecific cross-linkage of '2sI-labeled proteins in the abfunctional, i.e. that it leads to the conjugation of the protein sence of Es.However, in the presenceof Ea,a prominent high- with Ub. An approach to examining functional enzyme-submolecular weightcross-linkage product of labeled protein strate binding is the "isotope trapping" technique (22, 23). substrates was formed. The molecular mass of the E3-specific When a labeled substrate is first bound to its enzyme and product (approximately 190-200 kDa)fitstheassumption then mixed with an excess of unlabeled substrate together that it consists of a single molecule of '2sI-labeled protein with a second substrate that completes the reaction, part of linked to the 180-kDa E3 subunit. Examination of the cross- labeled enzyme-bound substrate is converted to labeled prodEasubunit revealed uct provided that product formation is faster than substrate linkage of different labeled proteins to the

Protein Binding Site of Ubiquitin Ligase 1

BS3

2

3

4

5

6

1 2 3 4 5 6 71 82 3 4 5 6 7

n n nn-n

c

- + + -+ + -+ + + + + + + +

-+-+-+

-

11995

- "Origin 3

1-205

"-

Origin-

Origin-

-

m

-116 -97

-66

I:[(;. :i ( / c / t ) . Cross-linking of lZ'II-labeledproteins to E:, suhunit. Cross-linking conditions were as described under ">laterials and Methods" except that E, (fraction D of phenyl-Sepharose chromatography,cf. Fig. 1) and bis(sulfosuccinimidy1)suherate were supplemented where indicated by + signs. The following IY5I-labeled proteins were used (numbers a t top): 1 , cytochrome c from S. cereuisae; 2, cytochrome c from horse heart; 3, lysozyme; 4, RNase A; 5,Ox-RNase; 6, MetO-RNase. All labeled proteins were supplemented a t 200,000 cpm (0.81.3 pg). Positions of marker proteins (kDa) are indicated on right the side. Arrowhead,position of &-specific crosslink product. This is not much influenced by the protein, since theM,of all labeled proteins tested is in the range of 12,000-14,500. FIG. 4 (center). Competition of cross-linking of 1z61-labeledyeast cytochrome c by various unlabeled proteins. All incubations contained 0.5 pg of "'I-labeled cytochrome c from S. cereuisae. Other cross-linking conditions were as described under "Materials and Methods." Lane I , without E2; lanes 2-7, with Ea;lanes 3-7, with the addition of 10 pg of the following unlabeled proteins: lane 3, cytochrome c from S. cereuisae; /one 4, cytochrome c from horse heart; lane 5, RNase A; lane 6, Ox-RNase; lane 7, MetO-RNase. Arrowhead, position of E,-specific cross-linking product. FIG. 5 (right). Trapping of 1251-lysozymebound to E3 for the formation of labeled conjugates. Pulsechase experiments were conducted as described under "Materials and Methods," with the following variations: lanes 1-4, preincubation of "'I-lysozyme with E3; lanes 5-8, preincubation of '"I-lysozyme with E , and E*; lane I , chase mixture not supplemented lane 2, chase mixture lacking unlabeled lysozyme; lane 3, unlabeled lysozyme (60 pg) present in t,he pulse phase; lane 4, pulse-chase experiment. Inlanes 5-8, 12sI-lysozymewas preincubated with a mixture of E,, E?, Me-Ub, and MgATP ina composition similar to that of the regular chase mixture but without unlabeled lysozyme. Lane 5 , without further additions;lane 6, E, added in the chase phase but withoutunlabeled lysozyme; lane 7, unlabeled lysozyme (60 p g ) added in the pulse incubation and then E, in the chase incubation; lane 8, E, and unlabeled lysozyme (60 p g ) added in the chase incubation. All chase incubations were for 2 min. Contarn, contaminations in the preparationof "'I-lysozyme; Lys, '2sI-lysozyme.

dissociation. In the experiment shownin Fig. 5 , '2sI-lysozyme Kinetics of Conjugate Formation andEs.Substrate Dissocia(at a high specific radioactivity) was preincubated with ER tion-The time course of the formation of "'I-lysozyme-Me(pulse), and thena 400-fold excess of unlabeled lysozyme was Ub conjugates in the chase phase of the pulse-chase experiadded together with E,, E2, Me-Ub, and MgATP, and incu- ment is shown inFig. 6 and the quantitation of the experiment bation was continued for a short period of time (chase). Me- in Fig. 7. It may be seenthatthe low-molecular weight Ub was used instead of native Ub to prevent the formation of conjugates 1-3 are formed in the first 10-30 s. Thereafter, polyubiquitin chains (14), which would disperse any trapped radioactivity in bands 1-2 declines, whereas that in bands4radioactivity to more numerous products. Incorporation of 7 increases after a short initiallag. This suggests a precursor":'I-lysozyme radioactivity into conjugates was observedin product relationship between low- and high-molecular weight the pulse-chase experiment (Fig. 5 , lane 4 ) in the regions conjugates, though some low-molecular weight conjugatesperexpected from the control in which no unlabeled lysozyme sist even after a prolonged chase incubation. I t should be increase in radioactivityof high-molecwas added (lane 2). Anothercontrol showed that isotope further noted that the dilution by unlabeled lysozyme was sufficient to prevent the ular weight conjugates greatly exceeds the decrease in bands significant formation of labeled conjugates when unlabeled 1-2. The total amount of label in allconjugates increased substrate was added prior to ERin the pulse phase (lane 3 ) . until 2 min and approached a final level after 5-10 min of When '"I-lysozyme was preincubated with E2 and El and chase incubation (Fig. 7). In the experimentalprocedure emthen Es was added at the chase phase, no significant formation ployed (see "Materials and Methods"),mixing with excess of labeled conjugates was observed (Fig. 5 , lane 8).A parallel unlabeled lysozyme was complete within 1 s of its addition. accumulation of label in conjugates control to which unlabeled lysozyme was not added and ER Therefore, the continuing was present only in the chase (lane 6) showed less conjugate must originate from an &-bound pool of '2511-lysozyme. This formation than in the case of preincubation of Ea with implies that thedissociation of ER.12sI-lysozymecomplex has lysozyme (lane 2). This is apparently due to the slow associ- to be slow relative to the rateof conjugate formation. ation of":'I-lysozyme with ER (see below). Thesefindings These assumptionswere tested by the determinationof the provide evidence for a functional enzyme-substrate complex rate of dissociation of E,~.lZsI-lysozyme complex (Fig. 8). 9 Lysozyme was preincubated with En, following which an exand indicate that ER,but not E2, has the protein substrate binding site of the Ub ligation system. cess of unlabeled lysozyme was added forvarious periods prior

Protein Binding Site of Ubiquitin Ligase

11996

I

2

3

4

5

6

7

Origin-

-7

- 4-6

Contorn

20

-3 -2 -I

-

10

I

,

1

1

2

3

1

4

5

Time(minutes 1 FIG. 8. Kinetics of dissociation of E3.'251-lysozyme

complex. Experimental conditions were as described under "Materials LYS"( and Methods," except that unlabeled lysozyme (60 p g ) was added a t various time intervalsbefore the additionof the chasemixture, which lacked unlabeled lysozyme. Following all additions, mixing was comFIG. 6. T i m e c o u r s e of conjugate formation in the pulsechase experiment.. Reaction conditions were as described under plete within 1 s. The zero time of the experiment was a sample in which unlabeled lysozyme was added togetherwith the chase mixture "Materials and Met.hods," except that the amount of '""Ilysozyme cases, incubation was continwas increased 2-fold. The complete chase incubation was stopped (regular pulse-chase incubation). In all of El/& mixture. Following gel after the following time periods: lane I , zero time (chase mixture not ued for 5 min after the addition added); lane 2, 10 s; lane 3, 30 s; lane 4, 1 min; lane 5, 2 min; lane 6, electrophoresis and radioautography, radioactivity in all conjugates was quantitated as described in Fig. 7. The results are expressed as 5 min; lane 7, 10 min. Numbers in the right margin indicate the different conjugates of ""I-lysozyme with Me-Ub (14) in order of the percentage of radioactivity in conjugates at time 0, which was increasing molecular size. Contam, contaminations in the preparation 9500 cpm. Time denotes the interval between the additions of unlabeled lysozyme and of E1/E2 mixture. of ""I-lysozyme; Lys, "sl-lysozyme. I

I ,

1

2

3

4

5

6

I

I

7

8

1

9

2

,

1

0

Time (minutes)

FIG. 7. Quantitation of kinetics of formation of different conjugates in the pulse-chase experiment. The different bands of '"'I-lysozyme-Me-Ub conjugates shown in Fig. 6 were excised, and their radioactivity was estimated by gamma counting. Radioactivity of corresponding regions of the zero time sample was subtractedfrom all samples, and the resultswere expressed as the percentageof "'Ilysozyme radioactivity supplemented. Numbers in the figure indicate the correspondingly numbered conjugate bands inFig. 6. Total, radioactivity in all conjugate bands 1-7.

to supplementation with a mixture of E , , EP, Me-Ub, and MgATP. The amount of labeled conjugates formed is a minimal estimate of the amount of '2'II-lysozyme bound to Ea a t the time of addition of the E I / E 2mixture. It may be seen that dissociation of "'I-lysozyme. ERcomplex is slow and has at least two components. About one-half of &-bound "'I-lysozyme dissociated with a half-time of about 10 s, whereas the

4

6

8

10

Time (minutes)

FIG.9. T i m e c o u r s e of binding of '261-lysozyme to E3. Experimental conditions were as described in "Materials and Methods" for the pulse-chase incubation except that'2'II-lysozyme was omitted from the pulse mixture. Following preincubation a t 37 "C for 5 min, "51-lysozyme was added, and incubation was continued for various time periods before the addition of the chase mixture. Chase incubation was for 5 min in all cases. Radioactivity incorporated into all conjugates was quantitated as described for Fig. 7. The results are expressed as the percentageof 12sI-lysozymeconverted to conjugates. Time denotes the intervalbetween the additionsof 12sI-lysozymeand the chase mixture.

-

rest was released more slowly, tH 2.7 min. This may be due either to heterogeneity of binding sites on Es or to a heterogeneity of '2'I-lysozyme molecules (see "Discussion"). The time course of the binding of '*'I-lysozyme to ERwas examined in the experiment shown in Fig. 9. "'I-Lysozyme was incubated with Es for various periods before the supple-

Protein Binding Site of Ubiquitin Ligme

11997

mentation of the chase mixture, and the amount of labeled 1 2 3 4 5 6 -Origin conjugates formedwas estimated.Theassociation of T lysozyme with Es is also slow; half-maximal binding was a t around 1 min, and binding was complete only after 5 min. Specificity of Binding of Protein Substrates to Es-Our next question was to whatdegree is the specificity of the Ub ligation system determined by the properties of the protein binding site of Ea. For this purpose, it was desirable to compare the affinity of various proteins toEs with their susceptibility to Ub ligation and degradation. We previously found that a free ru-NH, group of proteins is required for their degradation by the Ub system (4). However, not all proteins with free NH, termini are good substrates, and it is of interest to define which specific alterations render these proteinsmore susceptible toUb ligation. It was previously noted (8)that RNaseA (free (U-NH, group, poor substrate) is converted to a good substrate by performic acid oxidation (Ox-RNase). Oxidation of RNase with performic acid produces drastic alterations including cleavage of disulfide bonds, oxidationof methionine residues to methionine sulfone, and conversion of protein FIG. 11. Conjugation of oxidized derivatives of RNase with structure toa random coil (24). We now find that thespecific Me-Ub. Reaction mixtures contained in a volume of 20 pl: 40 mM oxidation of methionine residues to methionine sulfoxide (by Tris-HCI, pH 7.6, 1 mM DTT,5 m M MgCI,, 2 mM ATP, 3 pg Me-Ub, photooxidation a t low pH) converts RNase to aneven better 0.5 microunits of E,, 0.4 microunits of 172, 2.9 microunits of En, and 1 proteins. Following substrate. The rate of ATP, Ub-dependent degradation of pg (approximately 300,000 cpm) of 12511-labeled incubation at 37 "C for 1 h, the reaction products were separated on MetO-RNase is more than 2-fold faster than that of Ox- a 12.5% SDS-polyacrylamide gel. Lunes 1-2, '"I-RNase A; lanes 3-4, RNase (Fig. lo), and formation of high-molecular weight Ub 12sII-Ox-RNase;lanes 5-6, 12sII"etO-RNase; lanes I, 3, and 5,without conjugates of theformerderivative was much more pro- El, E2, and En; lanes 2, 4, and 6, with E,, E?, and En. nounced (Fig. 11). It is interesting to note that the rate of digestion of MetO-RNase by trypsin or chymotrypsin was 4- ently recognized by the Ub ligation system. 5-fold lessthan thatof Ox-RNase (data not shown), indicating Theseproteinsubstrates were used to characterize the a lesser degree of denaturation. Rather, some specific altera- specificity of the binding site of Ea. As shown in Figs. 3 and tion produced by oxidation of methionine residues is appar- 4, cross-linkage with Ea of various derivatives of RNase and cytochromes from different species was in correlation with their susceptibility to the action of the Ub system. We further examinedthesecorrelations, using the functional isotope trapping assay. When Es is incubated with 12sII-lysozyme and an unlabeled protein in the pulse phase of the pulse-chase experiment, decreased formation of '2'II-lysozyme-MeUbconjugates is due to competition on the E, binding site. Competition at other sitesof the ligase system (such as on E,-MeUb) is already maximalat thechase phase, due to the presence of the large excess of unlabeled lysozyme. The results of such competition experiments are summarized in Table I. Yeast cytochrome c andyeast enolase(free NH, termini, good substrates for protein breakdown) effectively competed on the binding 12sI-lysozymeto Es, whereas equine cytochromec and rabbit muscle enolase (N"-acetylated, not substrates) did not (Table I, experiment1).Oxidation of methionine residues of lysozyme and RNase greatly increased their ability to inhibit Es (Table I, experiment 2). It should be noted, however, that native RNase alsocompeted partially even though it a is poor substrate. This resembles a similar effect of this protein on Ea-substrate cross-linkage (Fig. 4). That the effects of all proteins tested were indeed due to competition for the E, binding site rather than the competence of the chasereaction was indicated by the observation that they had no significant influence when added at the chase phase of the pulse-chase FIG. 10. Influence of oxidation of methionine residues of incubation (data not shown). From these results, it might be concluded that atleast two R N a s e on i t s d e g r a d a t i o n b y t h eUb system. Reaction mixtures contained in a volume of 200 pl: 50 mM Tris-HCI, pH 7.6,1mM DTT, structural features of proteins are recognized by the binding 5 mM MgCI,, 4 mM ATP, 12 pg U b , 40 pl fraction I1 from reticulocytes site of E3: a free a-NH2 group and oxidation of methionine (approximately 1 mg of protein), and 4 pg (4-6 X lo5 cpm) of "'Iresidues. The question arises of whether a free (u-NH, group labeled proteins indicated in the figure. Incubation was a t 37 "C. At is obligatory for substrate binding. When amino groups of various times, aliquotsof 50 pl were withdrawn and treated with1 ml lysozyme were blocked by reductive methylation, its efficiency of 15% trichloroacetic acid. Release of acid-soluble radioactivity was determined as describedpreviously (8). Parallelincubations were to compete on the binding of 12sI-lysozymewas greatly decarried out without ATP, and ATP-dependent degradation was cal- creased (Table I, experiment 3). However, a derivative which culated by the difference. Above 90% of ATP-dependent degradation was first methylated and thenoxidized (MetO, Me-lysozyme) of all three substrateswas also Ub-dependent. was a strong inhibitor of "'I-lysozyme binding even though

11998

Protein BindingSite of Ubiquitin Ligase TABLE I

as its subunit, as shown by its coincidence with E3activity in hydrophobic chromatography (Fig. l), second affinity chromatography (Fig. lA), gel filtration (Fig. 2), and hydroxylapatite chromatography (not shown). However, the subunitcomposition of E, is not clear. The native molecular size of E3 from rabbit reticulocytes had been previously estimated by gel filtration chromatography at around 300 kDa (8). The enzyme may be composed of two identical 180-kDa subunits or of one 180-kDa subunit with some lower-molecular weight subunits. It also cannot be ruled out at present that the enzyme is composed of only one 180-kDa protein and that its Protein added to pulse 1251-Lysozyme trapped higher apparent nativesize is due to its tightassociationwith endogenous reticulocyte protein substrates. In any case, the Clg % control 180-kDa protein contains the substrate binding site as shown Experiment 1 by cross-linking experiments (Figs. 3 and 4 ) . Cytochrome c, S. cereuisae 5 36.0 Cytochrome c, horse heart 5 95.0 Examination of the binding of different proteins t o E3 58.2 30 Enolase, bakers' yeast indicates that thespecificity of the binding sitemay account Enolase, rabbit muscle 30 104.0 for a part of the selectivity of the Ubligation system. Proteins with free a-NH2groups bind better to E3than proteins with Experiment 2 blocked a-NH2groups, and proteins with oxidized methionine Lysozyme 77.0 1.5 residues bind more tightly than their native counterparts. MetO-lysozyme 1.5 27.6 This correlates with the susceptibility of the above proteins 3 59.3 RNase A 3 48.4 Ox-RNase to Ub conjugation and protein breakdown. However, in con3 18.7 MetO-RNase trast to the absolute requirement fora free a-NH2 group for Ub-dependentdegradation (4), a methylatedand oxidized Experiment 3 derivative of lysozyme strongly competes on the binding site 10 15.3 Lysozyme (Table I). In addition, native RNase A (free a-NH2, but bad 10 70.5 Me-Lysozyme substrate) inhibits significantly the binding of labeled sub10 0 MetO, Me-lysozyme possessing strates (Fig. 4 and TableI). It appears that proteins only part of the structural requirements for Ub ligation may its amino groups were blocked. It is concluded that proteins bind toE3 in an abortivecomplex. Further selectivity may be without a free a-NH2group, or all amino groups, can bind to exerted at the stages of Ub conjugation, including ligation E3 provided thattheycontainanotheralteration which with the first Ub and the successive addition of multiple Ub strongly increases their affinity to the protein binding site. molecules. The observation that oxidation of methionine residues of DISCUSSION RNase greatly increases its susceptibility to the Ub system In this study we employed twoindependent methods, chem- (Figs. 10 and 11) raises the question whether a similar alteration may be a physiological signal for protein degradation. ical cross-linkage and functional isotope trapping, identify to and characterize the protein substrate binding site of the Oxidative damageof methionine residues in proteinsprobably ubiquitin-protein ligase system. Both methods indicated that occurs in cells, and an enzyme system was described which Ea has the protein binding site, and both yielded essentially specifically reduces methionine sulfoxide residues in proteins similarresults with regardtothepreferentialbinding of (26). The oxidation of asingle histidine residue has been implicated in the inactivation and breakdown of glutamine proteins that aregood substrates for the Ubsystem. effect of the The trapping of E3-bound '251-lysozymefor conjugate for- synthetase (27, 28). In the present study, the mation was allowed by the slow dissociation of E3.'251-lyso- oxidation of all methionine residues in a model protein was zyme complex. The biphasic dissociation kinetics(Fig. 8) may examined, and further studies with more physiological subindicate two types of binding sites on E3. Alternatively, it is strates are required to assess thepossible significance of this possible that the population of lZ5I-labeledlysozyme molecules observation. In addition to the protein binding site reported in this is heterogeneous. The latter possibility seems morelikely, since radioiodination with the chloramine-T procedure causes study, it appearsreasonable to assume thatE3 contains some ligation. The the oxidation of some methionine residues to the sulfoxide other sitesnecessary for its action in Ub-protein derivatives (25). It is possible, therefore, that themore slowly observation that E3 can be isolated by affinity chromatogradissociating component is due toa more extensively oxidized phy on Ub-Sepharose (8) indicates that this enzyme has Ub portion of 1251-lysozymemolecules, which have higher affinity binding site(s). E3 has to interactwith E,-Ub for Ub-protein for E3. It should be noted that unlabeled lysozyme used for ligation to occur. In addition, it is possible that some Ubchase experiments was subjected to a similar iodination pro- protein conjugates arealso tightly boundt o E3. Since ligation with multiple molecules of Ub isobserved in spiteof substrate cedure. is either In view of the tight bindingof '251-lysozymeto E3,it seems excess, ithas beensuggested thatthereaction reasonable to assume that some endogenous reticulocyte pro- processive or has a strong preference for proteins conjugated tein substrates are also tightly bound to this enzyme. This to Ub (29). In the present investigation, the continuing utilimay account for the observation that even the most highly zation of intermediates in the presence of high amounts of purified preparations of EBare heavily contaminated by en- unlabeled lysozyme (Fig. 6) suggests thatboththebound protein and its conjugates with Ub have very slow rates of dogenous proteinsubstrates for Ub conjugation (datanot shown). The apparently slow binding of '251-lysozymeto E3 dissociation. The affinity of E3 to Ub might be due to the (Fig. 9) might be due to the slow dissociation of endogenous presence of specific sites for the abovecomplexeswhich contain Ubmoieties. Further investigationis needed to define protein substrates,being replaced by the labeled protein. Though E3 has not been purified tohomogeneity, not much other binding sitesof EB,involved inother stagesof the action doubt remains about the identification of the 180-kDa protein of the Ub-protein ligase systems. Competition on bindingof '25Z-lysozyme toE3 by natiue and modified proteins Experimental conditions were as described under "Materials and Methods" for the pulse-chase experiment, except that the duration of the pulse incubation was 10 min. Unlabeled proteins were added before the pulse incubation at the amounts specified in the table. Radioactivity incorporated into all '251-lysozyme-Me-Ubconjugates was quantitated as described in Fig. 7. Results are expressed as the percentage of parallel controlincubations to which no unlabeled protein was added in the pulse phase. These were 1.6-1.9% of I2'Ilysozyme converted to conjugates in the different experiments.

Binding Protein

Site of Ubiquitin Ligase

Acknowledgments-We thank Dr. Irwin A. Rose for discussions, suggestions, and much help. We acknowledge the assistance of Sarah ~ l at the i initial ~ ~stages of this work, We also thank Clara Segal and Judith Hershko for excellent technical assistance.

11999

13. Ciechanover, A., Elias, S., Heller, H., and Hershko, A. (1982) J. Bwl. Chem. 257,2537-2542 14. Hershko, A., and Heller, H. (1985) Biochem. Biophys. Res. Commun. 128, 1079-1086 15. Bohlen. P.. Stein. S.. Dairman. W.. and Udenfriend. S. (1973) Arch.'Biochem. Biophys. 155; 213-220 16. Jori, G., Galiazzo, G., Tamburro, A. M., and Scoffone, E. (1970) J. Bwl. Chem. 245,3375-3383 17. Laemmli, U.K. (1970) Nature 227, 680-685 18. Rose, 1. A., and Warms, J. V. B. (1983) Biochemistry 22, 42344237 19. Pickart, C. M., and Rose, I. A. (1985) J. Biol. Chem. 260, 79037910 20. Staros, J. V. (1982) Biochemistry 21,3950-3955 21. Giedroc, D. P., Puett, D., Ling, N., and Staros, J. V. (1983) J . Biol. Chem. 288.16-19 22. Rose, I. A., O'Connell, E. L., and Litwin, S. (1974) J . Biol. Chem. 249,5163-5168 23. Rose. I. A. (1980) Methods Enzvmol. 64. 47-59 24. Richards, F . M.,' and Wyckoff: H. W. 11971) in The Enzymes (Boyer, P. D., ed), Vol. 4, pp. 647-806, Academic Press, OrI

REFERENCES 1. Hershko, A,, and Ciechanover, A. (1982) Annu. Reu.Biochem. 5 1,335-364 2. Hershko, A. (1983) Cell 34, 11-12 3. Hershko, A., and Ciechanover, A. (1986) Progr. Nucleic Acid Res.

Mol. Biol., in press 4. Hershko, A., Heller, H., Eytan, E., Kaklij, G., and Rose, I. A. (1984) Proc. Natl. Acad. Sci. U. S. A. 81, 7021-7025 5. Ciechanover, A., Heller, H., Katz-Etzion, R., and Hershko, A. (1981) Proc. Natl. Acad. Sei. U. S. A. 78, 761-765 6. Haas, A. L., Warms, J. V. B., Hershko, A., and Rose, I. A. (1982) J. Biol. Chem. 257,2543-2548 7. Haas, A.L., and Rose, I. A. (1982) J. Bwl. Chem 257, 1032910337 8. Hershko, A., Heller, H., Elias, S., and Ciechanover, A. (1983) J. Biol. Chem. 258, 8206-8214 9. Pickart, C. M., and Rose, I. A. (1985) J. Biol. Chem. 260, 15731581 10. Ciechanover, A., Wolin, S. L., Steitz, T. A., and Lodish, H. F. (1985) Proc. Natl. Acad. Sei. U. S. A. 82, 1341-1345 11. Ferber, S., and Ciechanover, A. (1986) J.Biol. Chem. 261,31283134 12. Ciechanover, A., Elias, S., Heller, H., Ferber, S., and Hershko, A. (1980) J. Biol. Chem. 255, 7525-7528

lando, FL

.

.

25. Schechter, Y., Burstein, Y., and Patchornik, A. (1975) Biochemistry 14,4497-4502 26. Brot, N.,and Weissbach, H. (1982) Trends Biochem. Sci. 7, 137139 27. Fucci, L., Oliver, C. N., Coon, M. J., and Stadtman, E. R. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 1521-1525 28. Levine, R. L. (1983) J. Biol. Chem. 258, 11823-11827 29. Hershko, A., Ciechanover, A., Heller, H., Haas, A. L., and Rose, I. A. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 1783-1786 30. Wray, W., Boulikas, T., Wray, V. P., and Hancock, R. (1981) Anal. Bwchem. 118, 197-203