Isolation, Characterization, and Partial Purification of a Novel Ubiquitin-Protein Ligase, E 3 TARGETING OF PROTEIN SUBSTRATES VIA MULTIPLE AND DISTINCT RECOGNITION SIGNALS AND CONJUGATING ENZYMES*

Degradation of a protein via the ubiquitin system in- volves two discrete steps, conjugation of ubiquitin to the substrate and degradation of the adduct. Conjugation fol- lows a three-step mechanism. First, ubiquitin is activated by the ubiquitin-activating enzyme, E 1. Following activa- tion, one of several E 2 enzymes (ubiquitin-carrier proteins or ubiquitin-conjugating enzymes, UBCs) transfers ubiquitin from E 1 to the protein substrate that is bound to one of several ubiquitin-protein ligases, E 3s. These en- zymes catalyze the last step in the process, covalent attachment of ubiquitin to the protein substrate. The bind- ing of the substrate to E 3 is specific and implies that E 3s play a major role in recognition and selection of proteins for conjugation and subsequent degradation. So far, only a few ligases have been identified, and it is clear that many more have not been discovered yet. Here, we de-scribe a novel ligase that is involved in the conjugation and degradation of non “N-end rule” protein substrates such as actin, troponin T is ubiquitin- and E2-F1-dependent Degradation of 125 I-labeled actin and troponin T was determined by measuring the radioactivity released into the trichloroacetic acid-solu-ble fraction as described (10). Results shown reflect ATP-dependent proteolysis.

The ubiquitin pathway is involved in the processing and degradation of key regulatory short-lived proteins. Among these are mitotic cyclins and cyclin-dependent kinase inhibitors, the NF-B p105 precursor and the inhibitor IB␣, p53 and the transcriptional activator AP-1, and major histocompatibility complex class I antigens (reviewed in Refs. [1][2][3]. Degradation of a protein via the ubiquitin pathway involves two distinct steps: signaling of the protein by covalent attachment of multiple molecules of ubiquitin and degradation of the targeted protein by the 26 S proteasome complex. Conjugation of ubiquitin involves a three-step mechanism. Following activation of the C-terminal Gly of ubiquitin by E1, one of several E2 enzymes transfers ubiquitin to the substrate that is specifically bound to a member of the ubiquitin-protein ligase family, E3. The ligase catalyzes the last step in the conjugation process, formation of an isopeptide bond between the activated Gly residue of ubiquitin, and an ⑀-NH 2 group of a Lys residue in the substrate or in the previously conjugated ubiquitin moiety. The structure of the ubiquitin system appears to be hierarchial: a single E1 carries out, most probably, activation of ubiquitin required for all modifications. Several major species of E2 enzymes were characterized in mammalian cells, plants, and yeast (4 -6). Yeast UBC2 (RAD6) and mammalian E2-14 kDa are involved in degradation of "N-end rule" substrates (4,5,7,8). UBC4 and UBC5 are involved in the degradation of short-lived and abnormal proteins in yeast (9). UBC8, (E2-15 kDa; Ref. 6) from Arabidopsis thaliana and its rabbit (E2-F1; Ref. 10) and human (UbcH5; Ref. 11) homologs are involved in ubiquitination of certain non N-end rule substrates such as p53 (11)(12)(13), the NF-B precursor, p105 (14), and glyceraldehyde-3-phosphate dehydrogenase (10). All these E2 enzymes act in concert with E3s, and it appears that each E2 enzyme can act with one or more E3 proteins.
Five E3 enzymes have been described so far. (a) Mammalian E3␣ (UBR1 in yeast) recognizes protein substrates via their free basic or bulky hydrophobic N-terminal amino acid residues (N-end rule; reviewed in Refs. 7 and 15). (b) Similarly, E3␤ recognizes N-end rule substrates via their free small uncharged N-terminal residues (16). (c) E6-AP is involved in the recognition of p53 (12,17,18). Recently, a series of unique E6-AP homologous proteins have been identified (19); however, their role as E3 enzymes and the identity of their putative substrates is still obscure. (d) E3-C is involved in cell cycle-dependent conjugation and degradation of mitotic cyclins. The ligase is a component of a large particle (ϳ1,500 kDa), the cyclosome (20) or the anaphase-promoting complex (21), that regulates the programmed destruction of mitotic cyclins containing the "destruction box" motif RAALGNISEN (22). E3-C is inactive during interphase and is activated during M phase, most probably by phosphorylation. The cyclosome contains also CDC16, CDC27 (21,23), CDC23, and CSE1 (24). It is not clear whether E3 is one of these components, or whether it is a distinct component of the complex. (e) Processing of p105, the NF-B transcriptional activator precursor, involves yet a novel ligase (14); however, the enzyme and its mode of action have not been characterized.
An important, yet unresolved, problem involves the hierarchial relationships between the ligases and their cognate substrates. It is unlikely that a single E3 enzyme recognizes a specific, single substrate. Rather, it is conceivable that each ligase recognizes similar, but not necessarily identical, shared motifs in a subset of protein substrates. These motifs can be either primary, or generated post-translationally. Secondary targeting signals can be, for example, cell cycle-or signal transduction-induced phosphorylations.
N-end rule substrates are recognized by a "pair" of defined E2-E3 enzymes, the mammalian E2-14 kDa and its yeast homolog UBC2/RAD6, and E3␣ and its yeast homolog UBR1 (7,15). The mode of recognition of non-N-end rule substrates is more elaborate: the recognition motifs are poorly understood and the conjugating enzymes do not necessarily consist of defined pairs. Recognition of some substrates is mediated by E2 enzymes that are homologous to the yeast UBC4 and UBC5. These enzymes can be either the A. thaliana UBC8 (6,12), the rabbit E2-F1 (10), or the human UbcH5 (11). However, the conjugation reactions are catalyzed by distinct ligases. For example, the ligase that is involved in processing of p105 (14) is clearly different from E6-AP, the p53 conjugating enzyme (12,17). Thus, it appears that E2-F1 or UbcH5 can each act in concert with several species of E3s. For the N-end rule pair, it has been shown that E3␣/UBR1 has a binding site for its cognate E2 enzyme, E2-14 kDa/UBC2-RAD6 (8,25,26). It is not known whether E2-F1 or UbcH5 associates with their different ligases, although the necessity to preserve the activation energy suggests that such interaction does exist. Despite this complexity, it is assumed that a single substrate follows a single degradation pathway: it is recognized via a single recognition motif and by one pair of conjugating enzymes. The repressor MAT␣2 appears to be one exception to this rule. It has two degradation signals (27) and is recognized by four E2 enzymes (28; see "Discussion").
Here we demonstrate that lysozyme is conjugated and degraded by two distinct, however ubiquitin-dependent, pathways. The protein is conjugated by E2-14 kDa-E3␣ and recognized via its basic N-terminal residue, Lys (29). In addition, it is also conjugated and degraded by a non-N-end rule pathway consisting of E2-F1 and a novel, yet unidentified, species of E3 (designated E3L). The recognition motif, that is clearly distinct from the N-terminal Lys, has not been identified.

Preparation of Rabbit Reticulocytes Lysate and Conjugating Enzymes
Reticulocyte-rich blood was induced in rabbits by successive injections of phenylhydrazine, and reticulocyte lysate was prepared as described (30). Lysates were resolved by anion exchange chromatography on DEAE-cellulose into unadsorbed material (Fraction I) and high salt eluate (Fraction II) as described (30). E1, E2-14 kDa, and E3␣ were purified from Fraction II by affinity chromatography over immobilized ubiquitin and immobilized ␤-lactoglobulin as described (31). E2-F1 was purified to homogeneity from Fraction I as described (10). UbcH5 cDNA cloned into pET3a vector was a generous gift from Dr. M. Scheffner, Deutsches Krebsforschungszentrum (DKFZ), Heidelberg, Germany. The cDNA was expressed in BL21 Escherichia coli (11) and purified from the bacterial extracts following induction as following: extract (from 1 liter of cultured cells) was chromatographed on DEAE-anion exchange resin. Fraction I (ϳ75 mg of protein) was concentrated (to ϳ2.5 ml) using Centriprep 10 and resolved on HiLoad Superdex 75 HR, employing an FPLC system, in a buffer containing 20 mM Tris-HCl, pH 7.2, 150 mM NaCl, and 2 mM DTT (buffer A). 2.4-ml fractions were collected and the enzyme was detected via Coomassie staining of aliquots from the various fractions that were resolved by SDS-PAGE. Salt was removed by repeated concentration-dilution cycles using Centricon 10 microconcentrators and a buffer containing 20 mM Tris-HCl, pH 7.2, and 2 mM DTT (buffer B). The UbcH5-containing fractions were concentrated to 100 l and pooled. We obtained ϳ1 mg of Ͼ85% homogeneous protein.

Degradation and Conjugation Assays
125 I-Labeled proteins were prepared by the chloramine T method as described (10). Degradation and conjugation assays were performed essentially as described (13). Briefly, the reaction mixture contained in a final volume of 25 l: crude reticulocyte lysate (10 l; ϳ1 mg of protein) Fraction II (ϳ100 g of protein), or wheat germ extract (ϳ100 g of protein), 40 mM Tris-HCl, pH 7.6, 5 mM MgCl 2 , 2 mM DTT, 5 g of ubiquitin, and 125 I-labeled substrate (ϳ2 ϫ 10 5 cpm). Reaction mixture without ATP contained 0.5 g of hexokinase and 10 mM 2-deoxyglucose. ATP-dependent degradation was monitored in the presence of 0.5 mM ATP, 10 mM phosphocreatine, and 5 g of creatine kinase. In the conjugation assays, 0.5 g of ubiquitin aldehyde (32) and 5 mM ATP␥S substituted for ATP and the ATP-regenerating system. Conjugation assays containing purified components were performed in the presence of: 2 g of E1, 0.5 g of E2-F1 (or UbcH5 when noted), and the indicated amounts of the various preparations of E3L (for the purification procedure of E3L, see below). Degradation reactions were carried out for 2 h at 37°C and conjugation assays for 30 min at the same temperature. Following incubation, reaction mixtures for monitoring conjugation were resolved via SDS-PAGE, and the gels were fluorographed, dried, and exposed to XAR-5 (Kodak) film or to an intensifying screen of a bioimaging analyzer (PhosphorImager; Fuji). Degradation of iodinated proteins was determined by measuring the radioactivity released into the trichloroacetic acid-soluble fraction as described (10). Recombinant MyoD was purified from bacterial extracts as described (33) and tested for conjugation by E1, E2-F1, and E3L as described above. High molec- 1 The abbreviations used are: TEMED, N,N,NЈ,NЈ-tetramethylenediamine; Arg-Ala, arginyl-alanine; ATP␥S, adenosine-5Ј-O-(thiotriphosphate); DTT, dithiothreitol; E1, ubiquitin-activating enzyme; E2, ubiquitin-carrier protein or ubiquitin-conjugating enzyme, UBC; E3, ubiquitin-protein ligase; E6-AP, E6-associated protein; ECL, enhanced chemiluminescence; FPLC, fast protein liquid chromatography; HPV, human papillomavirus; NEM, N-ethylmaleimide; oxRNase A, oxidized ribonuclease A; PAGE, polyacrylamide gel electrophoresis. ular weight ubiquitin-MyoD adducts were identified following separation on SDS-PAGE, transfer into nitrocellulose paper (Western blot), and detection with a rabbit polyclonal, affinity-purified anti-MyoD. Proteins were visualized using the ECL detection system (Amersham).

Determination of the Molecular Mass of E3L
Fraction II was fractionated by (NH 4 ) 2 SO 4 into Fraction IIA (0 -38%; see below). 2 ml of Fraction IIA (20 mg/ml) were loaded onto a HiLoad Superdex 200 HR gel filtration chromatography column, and the proteins were resolved in buffer A using an FPLC system. 2.4-ml fractions were collected and salt was removed by repeated concentration-dilution cycles using Centricon 30 microconcentrators and buffer B. Fractions were concentrated to 100 l and activity was determined as described above, using 2-l aliquots from the various column fractions as a source for E3L. Molecular mass of E3L was determined by separation of a set of molecular weight standard markers in the same column and under the same conditions.

Partial Purification of E3L
Chromatographic separations were carried out at 4°C and dialysis of samples at 0°C. E3L-containing fractions from the different purification steps were subjected to analysis of activity (stimulation of conjugation of 125 I-lysozyme in the presence of E1 and E2-F1 as described above), quantitative protein determination, and Coomassie Blue staining following SDS-PAGE. Enzyme activity was expressed in units. One unit represents incorporation of 1% of radioactivity into conjugates/mg of protein as determined in the linear range of the reaction. Quantification was performed using PhosphorImager.
(NH 4 ) 2 SO 4 Precipitation (Fraction IIA)-Solid ammonium sulfate was added to a final concentration of 38% to 300 ml of crude reticulocyte Fraction II (Fraction II contains ϳ15 mg/ml protein). Following stirring on ice for 30 min, the mixture was centrifuged at 28,000 ϫ g for 15 min. The precipitate was dissolved in buffer B, and the proteins were reprecipitated in 38% ammonium sulfate. The pellet was dissolved in buffer B and dialyzed against the same buffer.
Anion Exchange Chromatography on Resource Q-ϳ1,350 mg of the ammonium sulfate-precipitated proteins (Fraction IIA) were loaded onto a 18 ϫ 85 mm Resource Q column equilibrated in buffer B containing 200 mM KCl (buffer C) and connected to an FPLC system. Unadsorbed proteins were removed with 90 ml of buffer C, and the adsorbed proteins were eluted by a linear gradient of up to 600 mM KCl in buffer C. Following elution, all fractions were dialyzed against buffer B. All the conjugating activity adsorbed to the resin and eluted at 430 -480 mM KCl (fractions 29 -34).
Sepharose 6B-CL Gel Filtration Chromatography-Fractions from the Resource Q column that contained activity of E3L were pooled, concentrated to 5 ml (a protein concentration of ϳ12.5 mg/ml), and loaded onto Sepharose 6B-CL column (25 ϫ 500 mm) equilibrated with buffer A. 5-ml fractions were collected and tested for E3L activity and protein composition.
HIC Methyl Hydrophobic Chromatography-Fractions obtained from the Sepharose column (protein eluted at ϳ550 kDa; ϳ18.5 mg) were pooled, concentrated, and dialyzed against 20 mM NaP i , pH 7.4, 1 M ammonium sulfate, and 2 mM DTT (buffer D). The proteins were loaded onto an Econo-Pac HIC methyl cartridge equilibrated with buffer D. The adsorbed proteins were eluted by a 100-ml linear gradient of 1.0 -0.0 M ammonium sulfate in 20 mM NaP i , pH 7.4, and 2 mM DTT. Following elution, all the fractions were dialyzed against Buffer B.
Hydroxylapatite Chromatography-Fractions 17-20 from the HIC methyl column (eluted at ϳ0.5 M ammonium sulfate), contained all the activity. This partially purified protein (ϳ3.5 mg) was loaded onto an Econo-Pac HTP cartridge equilibrated against 10 mM KP i , pH 7.4, and 2 mM DTT. Elution of the adsorbed protein was carried out by a linear gradient of 10 -600 mM KP i , pH 7.4, containing 2 mM DTT. The ligase activity was detected at ϳ350 mM KP i and coincided with a protein of ϳ270 kDa in the Coomassie staining.
Mono Q Anion Exchange Chromatography-Dialyzed active fractions from the HTP column (ϳ2 mg of protein) were purified further by an additional anion exchange step (Mono Q). The adsorbed proteins were eluted by a 0.2 M-0.7 M KCl gradient in buffer B. A sharp peak of activity was detected at ϳ0.5 M KCl in coincidence with an ϳ270-kDa protein in the Coomassie staining.

Cross-linking of 125 I-Labeled Lysozyme to E3L and E3␣
Cross-linking of 125 I-lysozyme to E3L and E3␣ was carried out essentially as described (29). Briefly, the reaction mixture contained in a volume of 20 l: 25 mM KP i , pH 7.4, 125 I-lysozyme (ϳ1 g, 1 ϫ 10 6 cpm), 10 g of ovalbumin, ϳ3.0 microunits of E3␣ (29), or ϳ500 units of E3L (see above; it should be noted that activity of the two enzymes is expressed in different units). When indicated, 100 g of unlabeled lysozyme was added prior to the addition of the labeled protein. Following incubation at 37°C for 5 min, the samples were chilled on ice, and 2 l of 2 mM freshly dissolved bis(sulfosuccinimidyl)suberate (Pierce) were added. Following an additional incubation for 15 min at 0°C, the reaction was stopped with the addition of 2 l of 1 M ethanolamine, pH 9. Samples were resolved via SDS-PAGE (6%), dried, and autoradiographed as described above.

Determination of Protein
Protein concentration was determined using the Bradford method (34) and bovine serum albumin as a standard.

Effect of Arg-Ala and E2-F1 on the Degradation and Conjugation of 125 I-Labeled Lysozyme and oxRNase A in Crude Reticulocyte Lysate and Crude Reticulocyte Fraction II-Ly-
sozyme is an N-end rule (7) type I (29) substrate. It has a basic (Lys), "destabilizing" amino acid at the N-terminal position. The protein ligase E3␣ recognizes and binds such proteins and, in the presence of E1 and E2-14 kDa enzymes, catalyzes the addition of multiple molecules of ubiquitin to these substrates. It has been shown previously that the dipeptide Arg-Ala is a potent inhibitor of the type I binding site of E3␣ (binding of basic N-terminal residues). Addition of the inhibitor to ubiquitin-supplemented Fraction II completely abolished both conjugation and degradation of type I substrates (29). Inhibition is probably mediated via binding of the basic N-terminal peptide to the E3␣ binding site of the substrates. We noted that addition of the inhibitor to a complete lysate resulted only in partial inhibition of degradation of labeled lysozyme, whereas the inhibition in Fraction II was complete (Fig. 1A). We tested the hypothesis that a factor that is missing in the ubiquitin-supplemented Fraction II, probably E2-F1, is responsible for the Arg-Ala-resistant activity observed in the whole lysate. This E2 probably acts with a novel E3 enzyme, distinct from E3␣. The ligase is inactive in ubiquitin-supplemented Fraction II because its cognate E2 is missing, and the activity observed in this partially resolved system is catalyzed solely by E2-14 kDa and E3␣. Indeed, addition of E2-F1 to Fraction II in the presence of Arg-Ala, restores the inhibited activity, clearly demonstrating that a second pathway, resistant to Arg-Ala, is involved in lysozyme degradation (Fig. 1A). We tested the possibility that other type I substrates follow the two degradation pathways as well. As can be seen in Fig. 1B, degradation of oxRNase A is inhibited completely by Arg-Ala in crude lysate and, therefore, not surprisingly in ubiquitin-supplemented Fraction II to which E2-F1 was added (Fig. 1B). Thus, oxRNase A appears to be degraded by a single proteolytic recognition pathway involving E2-14 kDa and E3␣. To further corroborate these observations and to test the notion that the Arg-Alaresistant proteolysis is due to the presence of a novel conjugating enzyme, we tested the conjugation of the two N-end rule substrates to ubiquitin. Conjugation of lysozyme in complete lysate was only partially inhibited by Arg-Ala ( Fig. 2A, lanes  1-3), whereas it was completely abolished following addition of the dipeptide to ubiquitin-supplemented Fraction II ( Fig. 2A,  lanes 4 -6). Addition of purified E2-F1 to this system restored the inhibited activity ( Fig. 2A, lanes 7). Conjugation of ubiquitin to oxRNase was sensitive to Arg-Ala in complete lysate and, as expected, could not be reconstituted by the addition of E2-F1 to ubiquitin-supplemented Fraction II (Fig. 2, B and C).
A Novel Ubiquitin-Protein Ligase, E3, Is Involved in Conjugation of Lysozyme via N-end Rule-independent Pathway-The E2-F1-dependent, but Arg-Ala-insensitive, degradation of ly-sozyme strongly suggested that a novel, N-end rule-independent, ligase is involved in targeting the protein, in addition to E3␣. In order to identify the ligase involved in the targeting of lysozyme via the E2-F1 pathway, we employed an assay in which Fraction IIA was used as a source for the novel enzyme. Fraction IIA is devoid of E1 and known E2s (10), but contains an activity that, along with E1 and E2-F1 (or recombinant UbcH5, not shown), is necessary for conjugation of lysozyme (Fig. 3, 2-C and 2-D, lanes C). To initially characterize the ligase involved, we resolved Fraction IIA by gel filtration chromatography over Superdex 200 column. Aliquots from the resolved fractions were analyzed for conjugate formation in the presence of E1 and either E2-14 kDa (Fig. 3, 1-A and 1-B) or E2-F1 (Fig. 3, 2-C and 2-D). To demonstrate that the novel ligase is distinct from E3␣, the conjugation reactions were carried out in the absence (A and C) or presence (B and D) of Arg-Ala. As shown in Fig. 3, 1-A, a peak of E2-14 kDa-dependent conjugating activity eluted at an apparent molecular mass of ϳ350 kDa (Fraction 29). This activity is completely abolished by incubating the fractions in the presence of Arg-Ala (Fig. 3,  1-B) and probably represents the protein ligase E3␣. However, a novel and different peak of conjugating activity was detected when the same fractions were monitored in the presence of E2-F1 (Fig. 3, 2). This activity corresponds to an apparent molecular mass of ϳ550 kDa (Fraction 25; Fig. 3, 2-C). In striking contrast to E3␣, the novel ligase is not inhibited by the addition of Arg-Ala (Fig. 3, 2-D). Taken together, these findings clearly indicate that the enzyme that conjugates lysozyme in the presence of E2-F1 (and designated E3L, E3 large) is a novel protein distinct from E3␣ and probably from other known ligases. For example, E6-AP has a native and denatured molec- ular mass of ϳ100 kDa (17,12,14). It is also distinct from the p105 NF-B precursor processing enzyme that has a molecular mass of ϳ320 kDa (14). It should be noted that the E2-14 kDa enzyme acts only with E3␣ within the N-end rule pathway (Fig.  3, 1-A), whereas E2-F1 recognizes E3L but not E3␣ (Fig. 3, 2-C  and 2-D; note specifically lanes B and C).
Purification of the Ubiquitin-Protein Ligase E3L-The initial identification of the ligase E3L prompted us to further purify and characterize the protein. Reticulocyte lysate was used as a source for the enzyme. The purification scheme (Fig. 4) involved a combination of anion exchange (DEAE-cellulose, Pharmacia Resource and Mono Q), gel filtration (Sepharose CL-6B), hydrophobic (Bio-Rad HIC methyl), and hydroxylapatite (Bio-Rad HTP) chromatographies. Resolved fractions from the different columns were assayed for E3L activity based on their ability to conjugate ubiquitin to 125 I-lysozyme in the presence of purified E1 and E2-F1 enzymes. The summary of the purification procedure is described in Table I: overall purification is ϳ215-fold with a recovery of ϳ2.7%.
In parallel, the protein profile of the different fractions was followed by SDS-PAGE and Coomassie Brilliant Blue staining. The peak of activity eluted from all columns co-migrates in SDS-PAGE with a protein of ϳ270 kDa. Figs. 5 and 6 demonstrate the peak of activity recovered from the two last purification steps, hydroxylapatite and Mono Q anion exchange chromatographies, along with the profile of the proteins resolved by electrophoresis. Although the E3 enzyme has not been purified to homogeneity, it is highly likely that the 270-kDa protein is indeed E3L: the activity and the protein band co-migrate (see also below for additional functional assay).
To strengthen further the notion that the 270-kDa band is indeed E3L, we examined its ability, via chemical cross-linking, to recognize and bind labeled lysozyme. As can be seen in Fig. 7, preincubation of E3L (derived from the last purification step) in the presence of labeled lysozyme followed by the addition of the chemical cross-linker results in the formation of a specific, high molecular mass cross-linking product. The molecular mass of the product corresponds to the approximate sum of masses of the 270-kDa E3L subunit and lysozyme (the migration of the cross-linking product was compared with that of unreacted E3L; not shown). As a control, we carried out a similar reaction in the presence of E3␣. It is known that this enzyme binds specifically to some of its substrates, lysozyme and ␤-lactoglobulin for example (29). Indeed, we were able to demonstrate the cross-linking product between this ligase and labeled lysozyme. Addition of excess unlabeled lysozyme prior to the addition of the labeled protein inhibited formation of the labeled cross-linking products, indicating that the association between the enzymes and the substrate is specific.
Characterization of the E3L Ubiquitin-Protein Ligase-Using gel filtration chromatography, we have demonstrated that the E3 protein has an apparent molecular mass of ϳ550 kDa (Fig. 3, 2-C and 2-D; calibration of molecular mass markers is not shown). It is probably a homodimer, as the molecular mass of the SDS-PAGE-resolved protein is ϳ270 kDa. To further characterize the novel E3, we investigated the possibility that, like E6-AP (35), it also contains an active -SH group. As can be seen in Fig. 8 Thus, it appears that E3L, like E6-AP, does not function simply as a "docking" protein (15), but has an enzymatic function by itself: it is involved in the transfer of activated ubiquitin from E2-F1 to the substrate via formation of an intermediate thiol ester with ubiquitin.
Substrate Specificity of E3L-The initial identification and characterization of E3L indicated that the enzyme is distinct from all other known ligases. We have shown that E3L is different from E3␣ (Figs. 1-3) and E3␤. The molecular mass of E3␤ is similar to that of E3␣ (16), whereas E3L is clearly FIG. 4. Purification of E3L. E3L was partially purified via a 7-step purification procedure as described under "Experimental Procedures." larger. Also, E3L does not conjugate ubiquitin to the E3␤ substrates, Protein S for example (not shown). The high molecular mass of E3L shows that this ligase is also different from E6-AP, the p53-conjugating enzyme that has a molecular mass of ϳ100 kDa (17), and from the p105 NF-B precursor conjugating E3 that has a molecular mass of ϳ320 kDa (14). Thus, it is not surprising that E3L does not promote conjugation of other E3␣ substrates such as oxRNase A (Figs. 2 and 3), the E3␤ substrate Protein S (see above; not shown), p53, and p105 (not shown). Three other substrates, all of them muscle proteins, are recognized by the novel ligase. Actin, a 42-kDa N-␣-acetylated protein, was previously shown to be a substrate of the ubiquitin system (36). Troponin T, a 31-kDa protein, is a subunit of the 72-kDa heterotrimeric muscle protein troponin. As can be seen in Table II and Fig. 9, degradation and conjugation of both actin and troponin T are ubiquitin-and E2-F1-dependent. To identify the E3 involved in the conjugation of these two proteins, we monitored the fractions derived from the last three purification steps of E3L (hydrophobic, hydroxylapatite, and Mono Q anion exchange chromatographies) for conjugation of 125 I-actin and 125 I-troponin T. The activity toward these proteins co-migrates with that of E3L as monitored by using 125 Ilysozyme as a substrate. Fig. 10 demonstrates the activity profile of the Mono Q column (compare Fig. 6A to Fig. 10, A and  B). The third muscle protein which is recognized also by E3L is MyoD. MyoD is a myogenic regulator necessary for muscle differentiation in mammals (reviewed in Ref. 37). We have shown that the ubiquitin pathway is involved in MyoD degradation. 2 As depicted in Fig. 11, the enzymes involved in this process are E1, E2-F1, and E3L. E3␣ could not recognize the tract-We tested the notion that, like E6-AP and the p105 NF-B precursor conjugating ligase, E3L is not ubiquitous to all eukaryotes and is not present in wheat germ. Indeed, as shown in Fig. 12, wheat germ extract does not contain E3L activity. Addition of exogenous E3L stimulates the conjugation of lysozyme significantly (Fig. 12, compare lane 3 to lane 2), indicating that the other two conjugating enzymes, E1 and the cognate E2, exist in the extract. Thus, E3L is probably different from the previously described plant E3␥ that also acts with UBC8 (E2-15 kDa; Ref. 6), the plant homolog of rabbit E2-F1 and human UbcH5 (6, 10 -12). The E3L-and N-end rule-independent activity in the wheat germ extract (Fig. 12, compare lane 2 to lane 1; the N-end rule pathway was inhibited by the addition of Arg-Ala) is probably catalyzed by non-N-end rule ligases, E3␥ for example, that, as noted, are different from E3L. It should be noted that addition of E3L also stimulated significantly the conjugation activity monitored in the presence of 125 I-labeled ubiquitin (not shown).

DISCUSSION
It is well established now that recognition of proteins for conjugation and subsequent degradation by the ubiquitin system is carried out by E3 enzymes. Except for the N-end rule enzymes, few other ligases have been identified, and their mode of recognition is still obscure. Because of their crucial role in substrate recognition, it is important to identify additional E3s and study their mode of action.
In this study, we identified, characterized, and partially purified a novel E3 (E3L) enzyme involved in recognition of substrates in an N-end rule-independent mode. It is different from all other known ligases, both structurally and functionally. The enzyme is found, although not exclusively, in muscle extracts, and recognizes the muscle proteins actin, troponin T, and MyoD. Although the ubiquitin system is probably involved in certain pathophysiological processes unique to muscle (38), it is not clear that E3L is indeed a muscle-specific enzyme. It is present in reticulocytes and other mammalian tissues (not shown), but not in wheat germ. A more extensive substrate survey will be necessary in order to better define the specificity of this novel ligase.
An interesting finding relates to the targeting of lysozyme by E3L. Lysozyme is a bona fide N-end rule substrate that has a destabilizing amino acid, Lys, at the N-terminal residue (7). It is clear that recognition of lysozyme by E3L does not traverse the rule, as Arg-Ala does not inhibit the conjugation reaction. Thus, lysozyme is recognized by at least two distinct signals, the N-terminal amino acid residue and a new, yet unidentified, "body" motif. The two motifs are identified by two distinct pairs of conjugating enzymes, E2-14 kDa-E3␣ and E2-F1-E3L. Quantitative analysis reveals that the N-end rule pathway contributes ϳ30% and the rule-independent pathway ϳ70% to the overall proteolysis of the labeled substrate ( Fig. 1 and data not shown). We hypothesized that each of the different pathways recognizes and targets for degradation a distinct part of the molecule. To test this hypothesis, we attempted to identify a processed product of lysozyme following incubation in a proteolytic mixture that contains only the N-end rule pathway enzymes, a ubiquitin-supplemented Fraction II that does not contain E2-F1. However, we were not able to identify any proteolytic intermediate, suggesting that each pathway can degrade the substrate completely to free amino acids. Thus, interpretation of the finding that the activity of the two path- FIG. 8. E3L has an active -SH group. Conjugation of ubiquitin to lysozyme was monitored as described under "Experimental Procedures." All reaction mixtures contained E1, E2-F1, and partially purified E3L (derived from the last purification step, step 7, Mono Q anion exchange chromatography; purification steps are described under "Experimental Procedures" and in Fig. 4). Lane 1, complete reaction mixture; lanes 2-4, same as lane 1, but the E3L was preincubated for 10 min at 25°C in the presence of 10 mM NEM (lane 2), iodoacetamide (lane 3), and p-hydroxy mercuribenzoate (lane 4). Following preincubation, the alkylating agent was neutralized with the addition of 8 mM DTT, and the treated enzyme was added to the reaction mixture. Lane 5, same as lane 2, but DTT was added to E3L prior to the addition of NEM. Lane 6, E2-F1 was treated with NEM following neutralization with DTT. Ori. and D.F. denote origin of gel and dye front, respectively.

TABLE II Degradation of actin and troponin T is ubiquitin-and
E2-F1-dependent Degradation of 125 I-labeled actin and troponin T was determined by measuring the radioactivity released into the trichloroacetic acid-soluble fraction as described (10 ways is additive, is still missing. It should be noted that degradation in the cell-free system is, in most cases, limited, and only a fraction (ϳ20 -30%) of the substrate is degraded. It is possible that the system is gradually inactivated during incubation and what one measures is the sum of proteolytic yields of each of the pathways during the period in which it is active: removal of one pathway decreases the overall proteolysis that can be observed in a system in which the two pathways are active (crude lysate). Lysozyme is not the first substrate that is targeted via two distinct domains and different conjugating enzymes. The MAT␣2 repressor is signalled for degradation by two independent signals, DEG1 and DEG2 (1,27), and is targeted by two distinct pairs of E2 enzymes, UBC4 and UBC5, and UBC6 and UBC7 (28; see introduction). However, the case of lysozyme and E3L appears to be different. First, it is not clear whether DEG2 is at all a ubiquitin degradation signal. DEG1 is probably recognized by a complex between UBC6 and UBC7, whereas the role of UBC4 and UBC5 in the process remains obscure. The two signals in lysozyme are clearly targeted by the ubiquitin system. They are independent and each of them is sufficient to target the protein to complete degradation. Also, the identity of the E3s that are involved in the degradation of the repressor is not known. The existence of dual and independent proteolytic pathways within the ubiquitin system appears, at first glance, to be redundant. Yet, it raises interesting problems related to the evolution of the system and its physiological roles. It may well be that for key regulatory proteins, nature evolved a "safety" mechanism to ensure their prompt removal even under conditions that one pathway is inactivated. It is clear that lysozyme is only a model substrate. The discovery that it has two degradation signals and it shares two proteolytic pathways is the result of fortuity, that however reflects the complexity of selective degradation of specific cellular proteins.

FIG. 11. Conjugation of MyoD is E2-F1-and E3L-dependent.
MyoD was expressed in bacteria and purified as described (33). 1.4 g of MyoD were added to the conjugation assay as described under "Experimental Procedures." Following incubation, reaction mixtures were resolved via SDS-PAGE. Proteins were transferred to nitrocellulose (Western blot) that was incubated in the presence of affinity-purified rabbit polyclonal anti-MyoD antibody. Visualization was carried out using the ECL detection system (Amersham). Notes and molecular mass markers are as described in the legend to Fig. 2.   FIG. 12. E3L is present in reticulocytes but not in wheat germ extract. Conjugation assay was carried out in complete wheat germ extract in the presence or absence of ATP␥S and E3L (40 units) derived from the last purification step (Mono Q anion exchange chromatography) as described under "Experimental Procedures." Arg-Ala (5 mM) was added to all reaction mixtures to inhibit background activity of the N-end rule pathway. Notes and molecular mass markers are as described in the legend to Fig. 2.