A Ubiquitin-Protein Ligase Specific for Type III Protein Substrates*

A previously studied species of ubiquitin-protein ligase contains specific sites for the binding of basic (Type I) and bulky hydrophobic (Type II) NH2-terminal amino acid residues of protein substrates. We now describe another enzyme that ligates ubiquitin specifically to proteins that have NH2-terminal residues other than the above two categories (Type III substrates). The new species of ligase, that we call E3 beta, is separable from the formerly described ligase (termed E3 alpha) by affinity chromatography on protein substrate columns. E3 beta was partially purified from extracts of rabbit reticulocytes and was shown to be required for the breakdown of Type III proteins. Apart from its different substrate specificity, it resembles E3 alpha in some physical properties, in a requirement for ubiquitin carrier protein (E2) for conjugate formation, and in its action to ligate multiple ubiquitin units to the substrate protein. The denatured derivative of bovine pancreatic ribonuclease is a specific substrate for E3 alpha, while that of ribonuclease S-protein is a good substrate for E3 beta. Since S-protein is formed by the removal from ribonuclease of NH2-terminal S-peptide, it is suggested that E3 beta interacts with an NH2-terminal determinant exposed in ribonuclease S-protein.

A previously studied species of ubiquitin-protein ligase contains specific sites for the binding of basic (Type I) and bulky hydrophobic (Type II) NH2-terminal amino acid residues of protein substrates. We now describe another enzyme that ligates ubiquitin specifically to proteins that have NH2-terminal residues other than the above two categories (Type III substrates).
The new species of ligase, that we call Es@, is separable from the formerly described ligase (termed Intracellular protein breakdown is a highly selective process. Some proteins are committed to degradation by ligation to the 76-amino acid polypeptide ubiquitin (for reviews, see Refs. l-3). The rates of degradation of specific proteins are thus greatly influenced by their rates of ligation to ubiquitin. To understand the mechanisms of selectivity of intracellular protein breakdown, it is necessary to identify and characterize the enzymes involved in ubiquitin-protein ligation. A reasonable expectation is that several ubiquitin-protein ligases may exist, each of which recognizes certain structural features of proteins destined for degradation.
We have studied previously a species of ubiquitin-protein ligase, Es, that interacts mainly with certain NH*-terminal amino acid residues of substrate proteins (4)(5)(6). By the use of simple derivatives of amino acids that inhibit the binding of NH2-terminal residues of proteins to Ear three types of protein substrates that bind to distinct sites of EB could be identified. Type I protein substrates have basic NH*-terminal amino acid residues, Type II substrates have bulky hydrophobic NH*-terminal residues, and Type III substrates have NH*terminal residues other than the above two types (5). The two separate binding sites for basic and bulky hydrophobic NH2terminal residues coexist on a single species of Es, as indicated by affinity chromatography on immobilized protein substrate columns (6). The same species of Es also acts on some Type III proteins (6). We have observed, however, that another species of enzyme that ligates ubiquitin preferentially to Type III protein substrates does not bind to Type I or II protein substrate affinity columns (6).
In the present study we have followed up these observations. The new species of ligase (termed E$) resembles the formerly described ligase (called &OI) in some properties and physical characteristics, but is highly specific for certain Type III protein substrates. Reduction and carboxymethylation of RNase A and S-protein were carried out as described by Ferber and Ciechanover (9). Amino acid analysis of acid hydrolysates of these preparations showed essentially complete conversion of cysteine residues to carboxymethyl derivatives.
All proteins were radioiodinated by the chloramine-T procedure, as described (10).

Preparations of Enzymes
Fraction II (a crude ubiquitin-free fraction) was prepared from lysates of rabbit reticulocytes by chromatography on DEAE-cellulose, as described (11 original volume of Fraction II and was incubated (37 "C, 1 h) with an equal volume of H-aLA-Sepharose (12.5 mg of protein/ml of swollen gel) under conditions similar to those described previously for the affinity chromatography of E30i (6). The fraction not adsorbed to H-aLA-Sepharose was concentrated and incubated again with regenerated H-aLA-Sepharose under conditions similar to those described above. The affinity-unadsorbed fraction from the last treatment was collected. In this preparation, around 90% of E3a was removed, as judged by the decrease in the rate of the degradation of iZ51-H-uLA, a Type I substrate.
The affinity-unadsorbed fraction was precipitated with ammonium sulfate (O-38% saturation) under conditions similar to those described for the preparation of Fraction IIA (13). This preparation was concentrated by ultrafiltration to one-fifth of the initial volume of Fraction II and was subjected to gel filtration chromatography, as described under Fig. 2.
Determination of E, Activity The following assays were used for the determination of activities of either E3a or E3/3.

Protein
Breakdown Assay-E3 activity was determined by the stimulation of protein degradation in an ATP/ubiquitin-dependent system reconstituted from purified components under conditions in which E3 was rate-limiting.
The reaction mixture contained in a volume of 50 ~1: 50 mM Tris-HCl (pH 7.6), 3 InM DTT, 5 mM MgCl,, 4 mM ATP, 3 pg of ubiquitin, 1.5 microunits of El, 0.7 microunits of E, (see Ref. 11 for the definition of a unit of activity), 5 ~1 of 26 S protease complex, 0.5 ~1 of Factor X and '251-labeled protein substrate, and Eacu or -Es/3 as indicated in the legends to figures. Following incubation at 37 "C for 60 min. the release of labeled material soluble in 20% trichloroacetic acid was determined as described (11). A parallel incubation was conducted without Es, and Es-dependent activity was calculated by the difference. All results were taken from the range linear with Es concentration.

Ubiquitin Conjugation
Assay by Gel Electrophoresis-Reaction mixtures contained in a volume of 20 ~1: 50 mM Tris-HCl (pH 7.6), 5 mM MgCIP, 2 mM DTT, 2 mM ATP, 1.5 microunits of El, 6.7 microunits of ET, I5 ag of ovalbumin. 3 aa of MeUb. and ""I-labeled nrotein substrate and Esa or E3/3 as specified in the figure legends.'MeUb was used instead of ubiquitin to prevent the formation of very large polyubiquitin derivatives (11,14) that remain at the origin in polyacrylamide gel electrophoresis. Following incubation at 37 "C for 30 min, the samples were separated on 12.5% polyacrylamide sodium dodecyl sulfate gels, dried, and radioautographed.

Quantitative
Ubiquitin Conjugation Assay-This was carried out by a previously described method in which the conjugation of lZ51ubiquitin to protein is followed by adsorption to a mixture of DE52 and CM52 resins (11). For the determination of the activities of E,a and E#, the substrates were Ox-RNase or S-protein (0.2 mg/ml), respectively.

AND DISCUSSION
In a previous study on the affinity purification of Es with immobilized protein substrates, we noted that an enzyme that ligates ubiquitin to some Type III proteins does not bind to either Type I or II protein substrate affinity columns (6). Several enzymes that ligate ubiquitin to proteins are apparently not involved in protein breakdown (15-18). We have therefore first examined whether this ligase is required for the degradation of Type III substrates. In the experiment shown in Fig. 1, a crude extract of reticulocytes was applied repeatedly to Type I and II protein substrate affinity columns, and the activity of the fraction not adsorbed to the affinity columns in the ATP/ubiquitin-dependent degradation of different protein substrates was examined. The '*sI-labeled protein substrates used in this experiment were H-(YLA (a Type I substrate, NH, terminus: Lys) and ribonuclease S-protein (a Type III substrate, NHe terminus: Ser). There was a marked decrease in the activity of the "affinity-unadsorbed" fraction in the breakdown of the Type I substrate '""I-H-oLA, in comparison with the activity of the untreated extract (Fig.  1A). By contrast, there was little, if any, decrease in the activity of the same fraction in the breakdown of the Type "Affinityunadsorbed' fraction was prepared from Fraction II as described under "Experimental Procedures." The activities of the affinity-unadsorbed fraction (0) or of untreated Fraction II (0) on the degradation of different labeled proteins were tested by a procedure similar to that described under "Experimental Procedures" for the protein breakdown assay of E, activity, except that the indicated amounts of the crude fractions were supplemented instead of all purified enzymes. Parallel incubations were conducted in the absence of ATP, and values of ATPindependent degradation were subtracted from all results. A, degradation of '251-H-otLA (0.2 pg, 1.1 X lo5 cpm); B, degradation of i2'I-S-protein (0.5 pg, 7 X lo4 cpm).
III substrate 'Y-S-protein (Fig. 1B). These results are compatible with the interpretation that the species of ligase that does not bind to Type I and II affinity columns is involved in the breakdown of Type III proteins.
The ligase specific for Type III proteins was partially purified from the affinity-unadsorbed fraction by fractionation with ammonium sulfate (see "Experimental Procedures") and gel filtration chromatography (Fig. 2). Its activity was followed in purification by two assays: the ligation of S-protein to ubiquitin and the ATP/ubiquitin-dependent degradation of the same substrate in a proteolytic system reconstituted from purified components, the activity of which depends upon the supplementation of Es (see "Experimental Procedures"). As shown in Fig. 2A, a peak of enzyme activity that stimulated the breakdown of 'Y-S-protein in the reconstituted proteolytic system eluted at an apparent molecular weight of approximately 350,000. There was little ATP-independent proteolytic activity in all column fractions ( Fig. 2A), presumably due to the removal of most of ATP-independent multicatalytic protease by prior fractionation with ammonium sulfate (13, 19). To examine whether the enzyme that stimulates ATPdependent degradation of Y-S-protein is indeed identical to the ligase, fractions of the same gel filtration separation were assayed for the ligation of S-protein to ubiquitin. As shown in Fig. 2B, there was a strong ligase activity that eluted in exact coincidence with the activity that stimulated protein breakdown (center of peak at fractions 32-33 in both assays). We call this new species of ubiquitin-protein ligase E& while the formerly characterized species of E3 (that has binding sites for the NH,-terminal residues of Type I and II protein substrates) is called E3a. E3P resembles E3a in several physical characteristics. Both enzymes have a similar apparent molecular weight of 350,000 in gel filtration chromatography ( Fig. 2   Samples of 5 ~1 of column fractions were assayed for the conjugation of ""I-S-protein (-0.8 kg, 1.3 X lo5 cpm) by the gel electrophoresis assay described under "Experimental Procedures." Prior to incubation, samples of column fractions were treated with ubiquitin-aldehyde (5 pM) for 10 min at 37 "C to inactivate ubiquitin-protein hydrolases (24). Fraction numbers are indicated at the top. Contam., a contamination in the preparation of '251-S-protein.
ilarities in the mode of action of the two EB enzymes include the formation of polyubiquitin derivatives (see below) and the requirement for EP for conjugate formation (data not shown). Es/3 is thus clearly different from several types of E2 that transfer ubiquitin to certain proteins (15,18,20).
The specificities of the action of Escv and E3P in the ligation of ubiquitin to different proteins are compared in Fig. 3. While ERa acts strongly on the Type I and II protein substrates, lysozyme (NH, terminus: Lys) and PLG (NH, terminus: Leu), little if any activity of Ee/3 on these proteins was observed. By contrast, E3@ was much more active in the ligation of ubiquitin to some Type III proteins, such as ribonuclease S-protein and ribonuclease T1 (NH, termini: Ser and Ala, respectively).
The two enzymes are similar in that they both ligate several molecules of ubiquitin to the substrate protein, as indicated by the observation that high molecular weight multiple conjugates of 'Y-S-protein are formed with reductively methylated ubiquitin (Figs. 2B and 3). In addition Es& like EBB, can form polyubiquitin chains linked to the substrate protein. This was examined by the use of ribonuclase T1 that has a single lysine residue (21). With reductively methylated ubiquitin, a single low molecular weight conjugate was formed, while with native ubiquitin EsP promoted the formation of multiple high molecular weight polyubiquitin derivatives of ribonuclease T1 (Fig. 3).
The specificities of the two types of EB for different protein substrates were further examined by the quantitative protein breakdown assay (Table I) for the gel electrophoresis assay. '""I-Labeled proteins (indicated at the top) were supplemented at l-l.8 x 10" cpm and 1 pg, except for ""I-ribonuclease T, that was 0.1 pg. Lanes I, without ES; lanes 2, with 0.5 microunits of &n; lanes 3, with 0.12 microunits of E&. All incubations were carried out in the presence of MeUb (see "Experimental Procedures"), except for those with ""I-ribonuclease T, (lanes 1-3)   The rates of degradation of different "%labeled proteins were determined with the reconstituted proteolytic system (see "Experimental Procedures") in the presence of 1.5 microunits of &a or 0.44 microunits of &/3. A, the degradation of '*'I-Sprotein (0.4 pg, 5.5 X lo4 cpm) was determined in the reconstituted proteolytic system (see "Experimental Procedures") in the presence of E:$ (0.46 microunits) and the indicated amounts of unlabeled rcm-RNase (0) or rem-S-protein (0). B, the degradation of '*'I-Ox-RNase (0.04 fig, lo5 cpm) was determined under similar conditions but in the presence of &ol (0.7 microunits). All results were expressed as the percentage of the values of control incubations (without unlabeled proteirs added) which were (% degradation (h)): 'ZSI-S-protein, 23; ""I-Ox-RNase, 14.9. clease S-protein is a good substrate for ES@. Ribonuclease Sprotein is formed by the removal of a 20-amino acid NH,terminal fragment ("S-peptide") from bovine pancreatic ribonuclease (22). It is possible, therefore, that the exposure of a new NHz-terminal signal renders S-protein to become a good substrate for E$. However, it is also possible that the removal of the NHs-terminal fragment produces a conformational alteration in the protein that is recognized by E$. To distinguish between these possibilities, we have tested the effects of reduced carboxymethylated (rem-) derivatives of RNase and S-protein.
Reduction-carboxymethylation of disulfide bridges converts RNase to a random coil (22), and thus any remaining recognition determinants are solely due to primary structure. In the experiment shown in Fig. 4A, the effects of unlabeled reduced-carboxymethylated derivatives of RNase and S-protein on the degradation of Y-S-protein in inhibited much more efficiently by rem-S-protein than by rem-RNase. This indicates the role of an NH*-terminal primary sequence determinant recognized by E& A control experiment in which the effects of the same reduced carboxymethylated proteins on the degradation of iz51-Ox-RNase by EBB were tested (Fig. 4B) showed that in this case, rem-RNase inhibited more efficiently than rem-S-protein.
It should be noted that higher concentrations of both unfolded RNase derivatives inhibited both E3a and E3/3 (Fig. 4). This might be due to relatively nonspecific binding of denatured proteins to the putative "body sites" (6) of both ligases. The nature of the NH*-terminal signal in Type III proteins that is recognized by EQ@ remains to be elucidated. Varshavsky and co-workers (23) have recently reported that the degradation in reticulocyte lysates of derivatives of P-galactosidase that have NH,-terminal Ser, Ala, or Thr residues is specifically inhibited by dipeptides that have similar amino-terminal residues. These investigators proposed that a distinct "N-end recognizing activity" exists for the degradation of proteins with small and uncharged NHz-terminal amino acid residues. Most of the NH2 termini of the Type III protein substrates of ES/3 used in the present study belong to this category (Table   I). However, we could not detect any inhibition of EQP by homologous dipeptides with the presently used Type III protein substrates (data not shown). Further study is required to define the NHz-terminal signal and other recognition determinants that may exist in Type III proteins.